CORROSION INDUCED FAILURES OF PRESTRESSING STEEL ... · CORROSION INDUCED FAILURES OF PRESTRESSING...

Corrosion induced failures of prestressing steel

CORROSION INDUCED FAILURES OF PRESTRESSING STEEL

KORROSIONSBEDINGTE VERSAGENSMECHANISMEN BEI SPANNSTAHL

RUPTURES D'ARMATURE DE PRECONTRAINTE INDUITES PAR CORROSION

Ulf Nürnberger

SUMMARY

Rarely in prestressed concrete structures occurring fractures of prestressing

steel in prestressed concrete structure can, as a rule, be attributed to corrosion

induced influences. The mechanism of these failures often is not well under-

stood. In this connection it is difficult to establish the necessary recommendation

not only for design and execution but also for building materials and prestress-

ing systems in order to avoid future problems. This paper gives a survey about

corrosion induced failure mechanisms of prestressing steels with a particular

emphasis on post-tensioning tendons.

Depending on the prevailing corrosion situation and the load conditions as

well as the prestressing steel properties the following possibilities of fracture

must be distinguished:

• Brittle fracture due to exceeding the residual load capacity. Brittle fracture is

particularly promoted by local corrosion attack and hydrogen embrittlement.

• Fracture as a result of hydrogen induced stress-corrosion cracking.

• Fracture as a result of fatigue and corrosion influences, distinguishing be-

tween corrosion fatigue cracking and fretting corrosion/fretting fatigue.

ZUSAMMENFASSUNG

Die gelegentlich an den im Spannbetonbau verwendeten Spannstählen auf-

tretenden Brüche sind im Regelfall auf korrosionsbedingte Einflüsse zurückzu-

führen. Die Versagensmechanismen werden häufig nicht ausreichend verstan-

den. Deshalb ist es schwierig, die notwendigen Empfehlungen nicht nur für Pla-

nung und Ausführung sondern auch für die Auswahl der Baustoffe und Vor-

spannsysteme zu geben, um zukünftige Probleme auszuschließen. Der Beitrag

Otto-Graf-Journal Vol. 13, 2002 9

U. NÜRNBERGER

stellt in einem Überblick die korrosionsbedingten Versagensmechanismen von

Spannstählen, mit Schwerpunkt der Probleme bei nachträglich vorgespannten

Zuggliedern, dar.

In Abhängigkeit sowohl von der vorherrschenden Korrosionssituation und

den Belastungsverhältnissen als auch den Spannstahleigenschaften müssen die

folgenden Brucharten unterschieden werden:

• Sprödbruch durch Überschreiten der Resttragfähigkeit. Das Auftreten eines

Sprödbruches wird unterstützt durch einen lokalen Korrosionsangriff und ei-

ne Wasserstoffversprödung.

• Bruch infolge wasserstoffinduzierter Spannungsrisskorrosion.

• Brüche als Folge von Ermüdung und Korrosionseinflüssen. Hierbei ist zu

unterscheiden zwischen Schwingungsrisskorrosion und Reibkorrosi-

on/Reibermüdung.

RESUME

Les ruptures occasionnelles des armatures de précontrainte peuvent en gé-

néral être attribués à l'influence de la corrosion. Le mécanisme de ces ruptures

n'est souvent pas bien compris. Il est par conséquent difficile d'émettre des re-

commandations, non seulement pour la conception et l'exécution, mais égale-

ment pour le choix des matériaux et des systèmes de précontrainte. Cet article

donne un aperçu sur les mécanismes de ruptures induites par corrosion des

armatures de précontrainte, en particulier sur les armatures précontraintes par

post-tension.

En fonction des conditions corrosives de l'environnement, des conditions

de chargement et des propriétés de l'armature précontrainte, on distingue les ty-

pes de rupture suivants:

• rupture fragile due au dépassement de la capacité résiduelle de charge. La

rupture fragile est favorisée par la corrosion locale et la fragilisation par hy-

drogène.

• rupture par corrosion sous contrainte induite par l'hydrogène.

• rupture en raison des influences combinées de fatigue et corrosion. On dis-

tingue la fatigue sous corrosion et la corrosion par friction/fatigue par fric-

tion.

KEYWORDS: prestressed concrete, corrosion, failures, steel

10


1. INTRODUCTION

Most of the prestressed concrete structures built in the last 50 years in ac-

cordance with the rules for good design, detailing and practice of execution have

demonstrated an excellent durability [1]. Analyses of occasional problems con-

firm that instances of serious failures are rare considering the volume of

prestressing steels that has been in use worldwide.

Major issues which strongly influence the level of durability actually

achieved are insufficient design (poor construction), incorrect execution of

planned design (poor workmanship), unsuitable mineral building materials, un-

suitable post-tensioning system components, including the prestressing steel [1-

3]. Insufficient design and incorrect work execution will mean that the necessary

corrosion control is not guaranteed from the beginning in all areas or that as a

result of natural influences (i. e. carbonation, chloride ingress) it will get lost

soon within the time frame of the originally anticipated life time. Unsuitable ma-

terials or inappropriate substances in materials will further corrosion and/or

stress corrosion cracking. Sensitive prestressing steels cannot withstand even

inevitable building-site influences or will fail while in use.

Most corrosion defects are caused by water which seeps through zones of

porous concrete and vulnerable areas such as leaking seals, joints, anchorages or

cracks, and which flows through the network of ducts which have been grouted

to a greater or lesser extent. The major threat is corrosion due to chlorides. The

source of chlorides can be either de-icing salts or seawater.

Rarely occurring fractures of prestressing steel and failures of prestressed

concrete structure can, as a rule, be attributed to corrosion induced cracking. The

mechanism of these failures often is not well understood. In this connection it is

difficult to establish the necessary recommendation not only for design and exe-

cution but also for building materials and prestressing systems in order to avoid

future problems.

This paper gives a survey about corrosion induced failure mechanisms of

prestressing steels with a particular emphasis on post-tensioning tendons.


U. NÜRNBERGER

2. FRACTURE MECHANISMS OF PRESTRESSING STEEL

The types of corrosion occurring at times as well as their specific manifes-

tation must be regarded as an essential influencing factor on the behaviour of the

prestressing steels under unforeseen or inappropriate service conditions. The

exclusive determination that corrosion was involved is not enough for a critical

case study and for future damage prevention.


well as the prestressing steel properties the following possibilities of fracturing


• Brittle fracture due to exceeding the residual load capacity. Brittle fracture is

particularly promoted by:

− local corrosion attack (pitting and wide pitting corrosion),

− hydrogen embrittlement.

• Fracture as a result of stress corrosion cracking, where we distinguish be-

tween

− anodic stress corrosion cracking and

− hydrogen induced stress-corrosion cracking.


tween

− corrosion fatigue cracking and

− fretting corrosion/fretting fatigue.

In the following such events will be described in more detail, also with re-

gard to prestressed concrete construction.

2.1 Brittle fracture

Brittle fracture may occur in high-strength steels after swift tensile stress.

This is the case in prestressing steels when there is a fracture under loads until

reaching the permissible pre-strain as a result of these influences:

− stress concentration in local notches (e. g. wide corrosion pit),

− high stressing speed and low temperature,

− an embrittlement of the steel structure after hydrogen adsorption

(hydrogen embrittlement).

12


Influence of corrosion

Mainly uniform general corrosion (e. g. after a prolonged weathering on a

building site) does not have any major impact on the load bearing capacity. Not

until, due to corrosion, an underrun of the required residual cross section has

taken place than a prestressing steel fracture may occur after exceeding the re-

sidual load bearing capacity. Such events may happen once prestressing steels in

ungrouted tendon ducts are exposed over a long period of time to water and

oxygen via untight anchorages or construction joints.

If, however, the prestressing steel incurs a local corrosion attack in the

form of pitting or wide pitting corrosion, the load bearing capacity may get lost

at an early stages due to brittle fracture. The following effects are capable of

triggering such attacks in prestressing steel:

The presence of aggressive water in the not yet injected ducts of post ten-

sioning tendons which result from bleeding of the concrete during the erection

of the construction. Already in the not grouted and not prestressed condition the

steel may suffer from strong pitting or wide pitting corrosion and the load bear-

ing capacity can be reduced considerably.

Bleeding is a separation of fresh concrete, where the solid content sinks

down and the displaced water rises or penetrates in the inner hollows. In the

bleeding water significantly high contents of sulphates and increased quantities

of chlorides may be accumulated (Table 1) by leaching of the construction mate-

rials cement, aggregates and water. The high amounts of potassium-sulphates

result from the gypsum in the cement. The watery phase of fresh concrete pene-

trates into the ducts through the anchorages, couplings and defects in the sheet

and accumulates at the deepest points. Because of an access of air the alkaline

water carbonates quickly. As early as in the non-grouted and non-prestressed

condition the steel can suffer from strong pitting. Bleeding water attack may

within a few weeks lead to pitting depths of up to 1 mm.

Table 1: Analysis of bleeding water

sulphate 1.90 - 5.20 g/l chloride 0.13 - 0.18 g/l calcium 0.06 - 0.09 g/l sodium 0.18 - 0.37 g/l potassium 3.60 - 7.30 g/l pH-value 10 - 13


U. NÜRNBERGER

The access of chloride containing waters, e.g. above untight anchorages or

joints, in a non-grouted tendon duct may lead to damaging local corrosion attack

in prestressing steel during the life time and after years of use. Comparable at-

tacks must be expected once chloride salts penetrate to the tendon through a

concrete cover of inferior thickness and impermeability.

The performance characteristics of corroded prestressing steels can be de-

termined in tensile, fatigue and stress corrosion tests (Fig. 1). Such tests to estab-

lish the residual load bearing capacity will, for instance, be carried out while in-

specting older buildings, after damaged prestressing steel samples had been

drawn. This might help to gain the knowledge for necessary repair.

High strength prestressing steels show a far more sensitive reaction to cor-

rosion attack than reinforcing steels, and this increasingly in the sequence tensile

test - fatigue test – stress corrosion test [4]. In case of uneven local corrosion a

corrosion depth of 0.6 mm may suffice for breaking a cold deformed wire under

tension of 70 % of the specified tendon strength of about 1800 N/mm2 (Fig. 1,

tensile test).

At pitting depth of above 0.2 mm cold drawn wires may show fatigue lim-

its (fatigue limits for stress cycles of N = 2 · 106) of 100 N/mm2 and less (Fig. 1,

fatigue test). Like-new smooth surfaced steels normally show a fatigue limit of

more than 400 N/mm2.

In all the performance characteristics of prestressing steels local corrosion

attack has the most detrimental effect on the behaviour to hydrogen induced cor-

rosion cracking. In a test developed by FIP the prestressing steel is immersed

under tension into an ammonium thiocyanate solution. A minimum and average

time of exposure before failure is specified. For cold drawn wire and strand

these values are in the order of 1.5, respectively 5 hours. In this example these

life times are underrun at corrosion depths of > 0.2 mm (Fig. 1, stress corrosion

test).

14


fatigue test

0

100

200

300

400

500

0,0 0,2 0,4 0,6 0,8 1,0

depth of uneven local corrosionin mm

fatig

ue

lo

ad

ing

in

N/m

m²

for

N=

2*1

06

tensile test

0

400

800

1200

1600

2000

0,0 0,2 0,4 0,6 0,8 1,0


tensile

str

ength

Rm

in

N/m

m²

0

2

4

6

8

10

elo

ngation a

t fr

actu

re A

10 in

%

Rm

A10

stress corrosion test

0

2

4

6

8

10

0,0 0,2 0,4 0,6 0,8 1,0


life

tim

e in

h

FIP - test:20% NH4SCN

50° C

Fig. 1: Properties of cold-deformed prestressing steel wires St 1570/1770, dS = 5 mm, in relation to depth of uneven local corrosion [Nürnberger] (scattering of 90% of the test results)

Effect of hydrogen (hydrogen embrittlement)

In a specific corrosion situation prestressing steel corrosion may release

hydrogen which is then absorbed by the prestressing steel, which, if prestressed

at the same time, will allow hydrogen induced stress corrosion cracking with

crack initiation and crack propagation (chapter 2.2). Also if the prestressing steel

is free of any tensile stresses (not prestressed), hydrogen can be absorbed in the

event of corrosion. The steel will not crack, but depending on the quantity of

hydrogen absorbed and the specific hydrogen sensitivity the prestressing steel

may become brittle. This may have an adverse effect on the mechanical charac-

teristics [5], more so on the deformation properties than on the tensile strength.


U. NÜRNBERGER

0

1 0

2 0

3 0

4 0

5 0

0 1 0 2 0 3 0 4 0 5 0 6 0

t im e o f im m e r s io n in te s t in g m e d iu ma c c o r d in g F IP ( 2 0 % N H 4 S C N , 5 0 ° C ) in h

reduction in a

rea Z

in %

- 1 0 0

7 5

2 5 0

4 2 5

6 0 0

7 7 5

9 5 0

1 1 2 5

1 3 0 0

1 4 7 5

1 6 5 0

1 8 2 5

2 0 0 0

tensile

str

ength

Rm

in N

/mm

²

Z

R m

1 5 02 01 05 2

1 9 0 0

1 5 0 0

1 6 0 0

1 7 0 0

1 8 0 0

Fig. 2 Tensile strength (Rm) and reduction in area (Z) of cold deformed prestressing steel wires (4 steel melts) after charging with hydrogen [5]

Prestressing steel fractures as a result of corrosion-caused hydrogen embrit-

tlement may occur, for instance, when prestressing to a high stress level or

shortly after the prestressing, after the steel had been absorbing high quantities

of hydrogen in an enduring unfavourable corrosion situation. If properly and

swiftly processed, such damages, indeed, should not occur.

2.2 Fractures because of stress-corrosion cracking Stress-corrosion cracking is understood to mean crack formation and crack

propagation in a material under the effect of mechanical tensile stresses and of

an aqueous corrosion medium.

16


Anodic stress-corrosion cracking

In the presence of nitrate-containing non-alkaline electrolytes (pH-value

< 9) unalloyed and low-alloy steels may suffer an anodic stress-corrosion crack-

ing. Crack formation and crack propagation are due to a selective metal dissolu-

tion (e. g. along grain boundaries of the steel structure) with a simultaneous ef-

fect of high mechanical tensile stresses [6] on condition that there is special ten-

dency of the steels to passivate in nitrate-containing aqueous solutions.

In the prestressed concrete construction the media-related pre-conditions,

e.g. in the fertilizer storage and in stable ceilings, can be assumed as a fact. In

stables brickwork, salpetre Ca (NO3)2 may be formed by urea. In the presence of

moisture the nitrates may diffuse into the concrete and may cause stress-

corrosion cracking in the case of pretensioned concrete components affecting the

tension wires if the concrete cover is carbonated due to an inferior quality of the

concrete [6].

A specific nitrate sensitivity of the steels is always a pre-condition for an

anodic stress-corrosion cracking. Low-carbon concrete steels are very suscepti-

ble to nitrate induced stress-corrosion cracking. The prestressing steels currently

in use, however, are highly resistant to this type of corrosion.

Hydrogen induced stress corrosion cracking [6,7]

Fractures of prestressing steel as a rule can be referred to hydrogen induced

stress corrosion cracking (H-SCC). It may happen during the erection of the

construction or during later use. The following conditions are necessary:

• a sensitive material or state,

• a sufficient tension load,

• at least a slight corrosion attack.

The risk of fractures due to hydrogen induced stress corrosion cracking

therefore results from the joint action of very prestressing steel properties and

environmental parameters. What is needed is the presence of hydrogen which

comes into being under certain corrosion conditions in neutral and particularly

in acid aqueous media through the cathodic partial reaction of the corrosion.


U. NÜRNBERGER

Table 2: Chemical reactions of corrosion

anodic iron dissolution | Fe sFe 2+ + 2e-

cathodic reactions if pH > 7

~ ½ O2 + H2O + 2e- s2OH-

if pH < 7

¡ H+ + e- s Had (hydrogen discharge)

if potential is low

¢ H2O + e- s Had + OH- (water decomposition)

rivalry reaction with regard to ¡ and ¢ £ 2 Had s H2 (recombination)

is prevented in the presence of promotors

⁄ 2 Had + ½ O2 s H2 O

if oxygen is present

During the corrosion process hydrogen atoms have be set free and get ab-

sorbed by the steel. In sensitive steels the hydrogen under the effect of mechani-

cal stresses can create precracks in critical structural areas such as grain bounda-

ries. These cracks may grow and result in material fracture.

Special conditions have to exist to activate the formation of adsorbable hy-

drogen. To understand the correlations between procedure on site and develop-

ment of damage, the chemical reactions of corrosion should be considered (Ta-

ble 2). Harmful hydrogen can arise only

• if the steel surface is in an active state or depassivated (this is expressed

by reaction 1),

• if the cathodic reaction of corrosion is discharging hydrogen (this is de-

scribed by reaction 3) or water decomposition (this is described by reac-

tion 4),

• if the adsorbable atomic hydrogen is not changed into the molecular state

(see reaction 5).

18


A reduction of oxygen access may support evolution of adsorbable atomic

hydrogen (then reaction 6 is hindered). Therefore at the surface of corroding

steel the amount of adsorbable hydrogen atoms rises

• with increasing hydrogen concentration (reaction 3 or 4 is accelerated),

• in the presence of so-called promotors (reactions 5 is hindered),

• in an electrolyte impoverished in oxygen (reaction 6 is hindered).

From the practical point of view one can say that hydrogen assisted dam-

ages are only possible

• in acid media or if the steel surface is polarized to low potentials (e. g. if

the prestressing steel has contact with zinc or galvanized steel),

• in the presence of promotors such as sulphides, thiocyanate or compounds

of arsenic or selenium,

• and under crevice conditions, because the electrolyte in the crevice is poor

in oxygen.

Fig. 3: Pitting induced stress corrosion cracking

In concrete structures the attacking medium is mostly alkaline and acid

media are limited to exceptions. Nevertheless, in natural environments the pit-

ting induced H-SCC can take place (Fig. 3). Pitting induced H-SCC means crack

initiation within a corrosion pit. In the corrosion pits the pH-value falls down

because of hydrolysis of the Fe2+-ions. Pitting or spots of local corrosion can be

explained by differential aeration or concentration cells. Especially effective is


U. NÜRNBERGER

the attack of condensation water or salt enriched aqueous solution (bleed water,

chapter 2.1), when erecting the constructions.

In prestressed construction chloride contamination supports a local corro-

sion attack. In the case of sensitive prestressing steel all but minimal contents of

hydrogen can lead to irreversible damages. Then a minimal local corrosion at-

tack without visible corrosion products on the steel surface may lead to steel

fracture.

In prestressed concrete structures all types of uneven local corrosion should

be prevented to exclude failures because of hydrogen assisted cracking.

The preconditions for "classical" stress-corrosion cracking are most readily

to be found in prestressed concrete construction, i. e. crack formation and

propagation under purely static stress. By prestressing the stress amplitudes of

the structure caused e. g. by wind and traffic are kept low. Nevertheless, the oc-

currence of pulsating loads or service-related strain changes of the steels will

raise the crack corrosion risk since it will favour hydrogen induced "non-

classical" stress-corrosion cracking [6]. Plastic flow in steel favours an absorp-

tion of atomic hydrogen.

0

25

50

75

100

125

150

175

200

number of stress reversal

str

ess a

mp

litu

de

2 σ

A in

N/m

m2

agent:

RT without failure

1g/l NH4SCN

105

108

107

1062 864

5 22221111555222111552211 5555

lifetime in hours

2 8642 864

solutionaqueous

withwithout

air

KCll/g5,0

SOKl/g5 42

Fig. 4: SCC-behaviour of prestressing steel St 1420/1570 (German standard) ∅ 12.2 mm without and with dynamic stress of low amplitude

20


Fig. 4 [8] compares the behaviour of a quenched and tempered prestressing

steel (from a case of damage) sensitive to hydrogen in a stress-corrosion crack-

ing test with and without superimposed fatigue loading of low amplitude (30 –

80 N/mm2). The aqueous test solution contains 5 g/l SO42-,0.5 g/l Cl¯ without

and alternatively with 1 g/l SCN¯ as a promotor for a hydrogen absorption. The

stress-corrosion cracking test under static stress was realized at 80 % of the ten-

sile strength. This stress corresponds to the constant maximum stress in the ten-

sile fatigues test. Fig 4 represents the stress cycle number as a function of the

amplitude, in the course of which also the life time, calculated over the fre-

quency (f = 5s-1), is applied. The stress corrosion test results without superim-

posed fatigue loading are applied at a range of stress of 0 N/mm2. The hydrogen

insensitive steel failed in the "static" test within a test period of 5000 hours in

the promotor-containing solution but did not fail in the promotor-free solution. If

a fatigue test of low amplitude is superimposed, the lifetime in the promotor-

containing solution will more and more decrease with rising amplitude. In the

wave stress it is striking that fractures also occur on steels in the promotor-free

solution.

It was found that in cold deformed prestressing steels the influence of a su-

perimposed fatigue loading on the hydrogen induced stress-corrosion cracking is

revealing itself weaker. These tests lead to the conclusion that already fatigue

loadings of low amplitude or elongations caused by changes in utilization tend

to significantly jeopardize the susceptibility of prestressing steels to stress-

corrosion cracking.

2.3 Fractures because of fatigue and corrosion

Prestressing steels can only be subject to a noticeable steel stress in dy-

namically strained reinforced concrete structures if there is concrete in a cracked

state. The stress amplitudes of prestressing steel due to acting high dynamic

loads ( e. g. a high traffic load of a bridge) may then amount to > 200 N/mm2 in

the crack region. In the uncracked state the steels will show ranges of stress of

clearly less than 100 N/mm2.

Cracks in concrete may occur in partially prestressed structures. Since such

cracks tend to open and to close in a superimposed fatigue stress the following

facts must be considered:


U. NÜRNBERGER

Corrosion fatigue cracking

If corrosion promoting aqueous media penetrate through the concrete crack

to the dynamically stressed tendon, corrosion fatigue cracking is possible al-

though this type of corrosion has not been observed in prestressing steel con-

struction so far. Corrosion fatigue cracking [6] manifests itself in that a metallic

material under dynamic stress in a reactive corrosion medium (water, salt solu-

tion) will show a much more unfavourable fatigue behaviour than under fatigue

loading in air. This can be explained by characteristic interactions of metal

physical and corrosive processes which favour initial precrack formation and

propagation. As opposed to the stress-corrosion cracking the corrosion fatigue

cracking does not require a specifically acting corrosion medium.

In case of post-tensioning tendons the duct made of thin steel sheets does

not offer a lasting corrosion protection and may even suffer fatigue fractures un-

der dynamic stress [9].

A decrease of the fatigue limit by corrosion is the more distinct the higher

the strength of the steel and the more aggressive an attacking medium are.

Hence the high strength prestressing steels, when e. g. simultaneously attacked

by an aqueous chloride-containing medium, may show a very unfavourable fa-

tigue behaviour.

In traffic carrying bridge structures only the low-frequent stresses lead to

high stress amplitudes. This results in additional unfavourable conditions with

regard to corrosion fatigue cracking: with a falling frequency the influence cor-

rosion will increase and the fatigue limit will consequently drop.

For a cold drawn prestressing steel wire Fig. 5 shows a decrease of the cor-

rosion fatigue limit in the sequence air-water-chloride solution. For frequencies

of 0.5s-1 the fatigue limit for stress cycles of 107 is below 100 N/mm2.

The problem of corrosion fatigue cracking of cracked components can be

remedied by sufficient concrete cover and limiting the crack width. This is the

way of keeping pollutants away from the prestressing steel surface.

Fretting corrosion / fretting fatigue

In the vicinity of concrete cracks due to fatigue loading displacements be-

tween the tendon and the injection mortar or the steel duct respectively will oc-

cur in a cracked component. In bended tendons a high radial pressure acts at the

22


same time on the fretting prestressing steel surface. If air or oxygen advance to

the fretting location through the concrete crack a fretting corrosion is favoured

[6,10]. Fretting corrosion is described as damaging a metal surface similar to

wear as a result of oscillating friction under radial pressure with a partner. In the

presence of oxygen oxidation of the reactive surface will take place.

In fatigue loaded steels and under fretting corrosion stress at the same time

the fatigue behaviour is under a very unfavourable influence due to fretting fa-

tigue [10]. This is attributable to structural disintegration and the occurrence of

additional tensile strengths in the fretting area. In concrete embedded tendons,

subjected to a relative movement and a radial pressure in the concrete crack be-

tween prestressing steel and duct or injection mortar respectively, tolerable fa-

tigue limits of about 150 N/mm2 for cycles to fracture of 2 x 106 were found

[9,11].

In prestressed concrete constructions also the anchorages of the tendons,

due to fretting corrosion influences, show a fatigue limit which is reduced com-

pared with the free length [12]. Under dynamic stress of the anchored tendon the

fatigue limit, depending on the type of anchorage, is reduced to values between

80 and 150 N/mm2. For this reason, anchorages will always be positioned in ar-

eas of least stress changes. In the fatigue experiment the prestressing steels al-

ways fracture in the force transmitting area, i. e. at the beginning of the anchor-

age. Here, the fatigue limit is reduced due to the presence of shifting between

the prestressing steel and the anchor body and the high radial pressures at the

same time.

In prestressed concrete bridges, however, particularly the coupling joints

proved to be problematic. If such joints crack as a result of imposed stresses

(e.g. due to non uniform sun heating and low amount of reinforcement which

crosses the coupling joint) the tendon couplings will suffer major stress fatigue

cycles from the traffic load which also led to prestressing steel fractures owing

to the stress-sensitive couplings [2,11].


U. NÜRNBERGER

Fig. 5: Fatigue behaviour under pulsating tensile stresses of cold drawn prestressing steel wires (Rm ≈ 1750 N/mm2) in air and corrosion- promoting aqueous solutions (Nürn-berger)

3. CONCLUSION


well as the prestressing steel properties the following possibilities of fracturing


• Brittle fracture due to exceeding the residual load capacity. Brittle fracture

is particularly promoted by local corrosion attack and hydrogen embrit-

tlement.

• Fracture as a result of hydrogen induced stress-corrosion cracking.


tween corrosion fatigue cracking and fretting corrosion/fretting fatigue.

REFERENCES

[1] Durability of post-tensioning tendons. Proceedings of workshop held in

Gent University on 15 - 16 November 2001. fib technical report, bulletin

15

[2] Nürnberger, U.: Analyse und Auswertung von Schadensfällen bei Spann-

stählen. Forschung, Straßenbau und Straßenverkehrstechnik 308 (1980) 1 –

195

24


[3] Nürnberger, U.: Influence of material and processing on stress corrosion

cracking of prestressing steel (case studies). Publ. of fib commission 9.5, to

be published

[4] Neubert B., Nürnberger, U.: Erkennen von Spannverfahrensschädigung –

Untersuchung der statischen und dynamischen Kenngrößen in Abhängig-

keit von Rostgrad. Bericht II.6-13675 der FMPA Baden-Württemberg, Ot-

to-Graf-Institut, Stuttgart 31.01.1983

[5] Nürnberger, U., Beul, W.: Entwicklung einfacher und reproduzierbarer

Prüfverfahren für die Empfindlichkeit von Spannstählen gegenüber Span-

nungsrisskorrosion. Bericht 34-14071 der FMPA Baden-Württemberg, Ot-

to-Graf-Institut, Stuttgart 01.03.1996

[6] Nürnberger, U.: Korrosion und Korrosionsschutz im Bauwesen. Bauverlag,

Wiesbaden 1995

[7] Grimme, D., Isecke, B., Nürnberger, U., Riecke, E. M., Uhlig, G.: Span-

nungsrisskorrosion in Spannbetonbauwerken. Verlag Stahleisen mbH, Düs-

seldorf 1983

[8] Nürnberger, U., Beul, W.: Wasserstoffinduzierte Spannungsrisskorrosion

von zugschwellbeanspruchten Spannstählen, S. 302 – 309; in "Bewehrte

Betonbauteile unter Betriebsbedingungen". Wiley-VCH Verlag

[9] Cordes, H.: Dauerhaftigkeit von Spanngliedern unter zyklischen Beanspru-

chungen. Sachstandsbericht. Schriftenreihe Deutscher Ausschuß für Stahl-

beton 370 (1986)

[10] Patzak, M.: Die Bedeutung der Reibkorrosion für nichtruhende Veranke-

rungen und Verbindungen metallischer Bauteile des konstruktiven Ingeni-

eurbaus. Dissertation Universität Stuttgart, 1979

[11] König, G., Maurer, R., Zichner, T.: Spannbeton-Bewährung im Brücken-

bau. Springer Verlag Berlin-Heidelberg-New York-London-Paris-Tokyo,

1986

[12] Rehm, G., Nürnberger, U., Patzak, M.: Keil- und Klemmverankerungen für

dynamisch beanspruchte Zugglieder aus hochfesten Stählen. Bauingenieur

52 (1977) 287 – 298


U. NÜRNBERGER

26

Load bearing behaviour of fastenings with concrete screws

LOAD BEARING BEHAVIOUR OF FASTENINGS WITH CONCRETE SCREWS

TRAGVERHALTEN VON BEFESTIGUNGEN MIT SCHRAUBDÜBELN

COMPORTEMENT SOUS CHARGE DES ANCRAGES AVEC VIS D'ANCRAGE

Jürgen H. R. Küenzlen and Rolf Eligehausen

SUMMARY

Concrete screws are a relatively new fastening system. Their main

advantage compared to traditional post-installed fastening systems is a quick and

easy installation. A hole is drilled into the concrete and threads are cut in the

concrete by the screw as it is installed.

Concrete screws transfer tensile loads into the base material by mechanical

interlock of the threads. Due to their load-bearing mechanism, concrete screws

with a technical approval of the DIBt can be used for fastenings in cracked and

non-cracked concrete.

The typical failure mechanism for concrete screws is concrete-cone failure.

With increasing embedment depth the ratio of the depth of the concrete failure

cone to the embedment depth decreases. The failure load of concrete screws

with continuous threads along the entire embedment depth increases

proportionally to hef1,5 (hef = effective embedment depth), but it is about 20 %

smaller than the failure load of expansion and undercut anchors with the same

embedment depth.

In order for concrete screws to function properly, the threads cut into the

wall of the drilled hole must not be damaged during the installation. This

requirement is achieved by using the embedment depth defined in the Technical

Approvals.


J. H. R. KÜENZLEN, R. ELIGEHAUSEN

ZUSAMMENFASSUNG

Schraubdübel sind ein relativ neues Befestigungssystem. Ihr großer Vorteil

liegt in der einfachen und schnellen Montage. Es wird ein Loch in den Beton

gebohrt, in das der Schraubdübel beim Setzen ein Gewinde schneidet.

Schraubdübel werden in Durchsteckmontage gesetzt.

Schraubdübel übertragen eine angreifende Zuglast über mechanische

Verzahnung der Gewindeflanken, die in die Bohrlochwand einschneiden, in den

Untergrund. Aufgrund ihres Tragmechanismus sind bauaufsichtlich zugelassene

Schraubdübel für Befestigungen im ungerissenen und gerissenen Beton

geeignet.

Das Versagen erfolgt durch Betonausbruch, wobei mit zunehmender

Verankerungstiefe das Verhältnis von Tiefe des Ausbruchkegels zu

Verankerungstiefe abnimmt. Die Bruchlast steigt bei Schraubdübeln mit einem

über die gesamte Verankerungstiefe durchgehende Gewinde proportional zu

hef1,5 an (hef = Verankerungstiefe), jedoch ist sie unter sonst gleichen

Verhältnissen ca. 20 % niedriger als die Betonausbruchlast von Spreiz- und

Hinterschnittdübeln.

Damit Schraubdübel ordnungsgemäß funktionieren, dürfen die in den

Beton geschnittenen Gewindegänge nicht während der Montage beschädigt

werden. Diese Bedingung wird bei Einhaltung der in den bauaufsichtlichen

Zulassungen festgelegten Verankerungstiefe eingehalten.

RESUME

Les vis d'ancrage sont un système de ancrage relativement nouveau. Leur

principal avantage est une installation rapide et facile. Un trou est foré dans le

béton et les spires sont taraudées dans le béton par la vis lors de sa mise en

place. Les vis d'ancrage transfèrent les charges de tension dans le béton par le

couplage mécanique des spires. En raison de leur mécanisme porteur, les vis

d'ancrage avec un agrément technique du DIBt peuvent être utilisées pour des

ancrages dans le béton fissuré et non-fissuré. Le mécanisme de rupture pour les

vis d'ancrage est la rupture par cône de béton. Une augmentation de la

profondeur d'encrage est accompagnée d'une diminution du rapport de la

profondeur du cône de béton à la profondeur d'encrage. La charge de rupture des

vis d'ancrage à filetage continu sur toute la profondeur d'ancrage augmente

proportionnellement à hef1,5 (hef = profondeur d'ancrage effective), elle est

28


néanmoins environ 20 % inférieure à la charge de rupture des chevilles à

expansion et des chevilles à verrouillage de forme avec la même profondeur

d'ancrage. Afin que les vis d'ancrage puissent fonctionner correctement, les

filetages taraudés dans le béton ne doivent pas être endommagés pendant

l'installation. Ceci est réalisé si l'on respecte la profondeur d'ancrage définie

dans l'agrément technique.

KEYWORDS: concrete screw, shearing-off of threads, mechanical interlock

1. INTRODUCTION





In Germany there are currently three different types of concrete screws

from three manufacturers approved by the DIBt for fastenings with single

anchors and groups in cracked and non-cracked concrete [1,2,3]. Further

technical approvals exist for suspended ceilings and other comparable static

systems.

During the technical approval process a large number of tests were

conducted at the Institute of Construction Materials at the University of

Stuttgart. Furthermore, the load bearing behaviour of concrete screws was

systematically investigated through experimental and numerical studies within

the scope of a research project. Important results of research reports [5, 6, 7, 8]

are presented below.

2. CONCRETE SCREWS WITH TECHNICAL APPROVAL OF THE DIBT

Figure 1 shows three concrete screws with a technical approval by the

DIBt. The screws are intended for a drill hole diameter of d0 = 10mm and are

made of galvanised steel. They differ principally in steel strength, core diameter

and thread geometry. Two of the concrete screws have small steel teeth at the

end of the screw for cutting the threads into the concrete. The third concrete

screw has alternating high and low screw threads. Grooves are cut into the

concrete by the specially formed high screw threads.



Figure 1: Concrete screws (d0 = 10mm) with a technical approval of the DIBt

Concrete screws made of galvanised steel intended for a drill hole diameter

of d0 = 5mm and d0 = 6mm are approved for suspended ceilings. Concrete

screws with a drill bit diameter of d0 = 8mm and d0 =10mm have a technical

approval for the fastenings of statically determined and undetermined supported

components in cracked and non-cracked concrete. Fastenings with single

anchors and groups are allowed.

Technical approvals also exist for concrete screws made of stainless steel

with drill bit diameters of d0 = 6mm to d0 = 10mm. To aid in the cutting of

threads into the concrete, one concrete screw has an end made of galvanised

steel. This end cannot be added to the embedment depth. Another concrete

screw has small cutting pins made of carbon steel in the first turns to cut the

threads into the concrete.

While concrete screws made of galvanised steel are only allowed for use in

dry environments, the concrete screws made of stainless steel can be used

outdoors, in industrial environments and near the sea.

Concrete screws made of galvanised steel are cold-rolled and subsequently

tempered and heat-treated. Residual stress and incipient cracks in the steel can

result from this process. To insure flawless products, special tests must be

carried out during manufacturing within the scope of the internal quality control.

Concrete screws made of galvanised steel, which are produced according to

requirements for the technical approvals, have an indefinite lifespan in dry

environments. If concrete screws made of galvanised steel are used in

environments with a high corrosion risk (e.g. outdoors), a brittle failure can

occur as a consequence of stress corrosion cracking. The time until failure

cannot be predicted. In these cases concrete screws made of stainless steel (or

other types of fastenings) must be used.

30


In the following section results of tests with the concrete screws type 1 to

type 3 are presented. It is pointed out that the numbering of the concrete screw

types is not the same as shown in Figure 1 or in the cited references.

3. LOAD BEARING BEHAVIOUR OF CONCRETE SCREWS

During installation, concrete screws cut a thread into the wall of the drilled

hole (Figure 2). Therefore, tensile loads are transferred into the base material by

diagonal struts, i.e. mechanical interlock (Figure 3a). The load transfer

mechanism is similar to that of deformed reinforcing bars cast into concrete

(Figure 3b) because the flanks of the screw thread function in a similar manner

as the ribs of reinforcing bars. However, the laws for deformed reinforcing bars

are only partially valid for concrete screws. One reason for this is that damage

due to small outbreaks in the threads cut into the wall of the drilled hole can

occur, which reduce the area for the mechanical interlock. Additionally, the core

diameter of the concrete screw is smaller than the drill hole diameter to allow for

easier installation. Consequently, the lateral restraint of the concrete is lost in the

region of the highly loaded concrete consoles. To achieve sufficient load transfer

into the concrete, the „relative rib area“ of concrete screws, which corresponds

roughly to the ratio between the depth and the spacing of the threads cut into the

wall of the drilled hole, is much larger than that of commercially available

deformed reinforcing bars.

Figure 2: Concrete screw and a thread cut into the wall of the drilled hole [9]



a)

b)

Figure 3: Transmission of tension load into concrete

a) Concrete screw

b) Cast-in-place deformed reinforcing bar

4. INSTALLATION OF CONCRETE SCREWS

Concrete screws are normally screwed into the concrete using an electric-

screw-gun. In technical approvals the power class [2,3] or the type of electric-

screw-gun [1] is specified. The threads cut into the concrete must not be

destroyed during installation. Limiting the applied torque can do this. Concrete

screws can also be screwed in with a torque wrench. It cannot be excluded that

concrete screws should not be screwed in using a commercial screw-wrench,

because the torque necessary for tightening up after the screw head reaches the

attachment can range between wide limits and therefore the threads cut into the

concrete might be destroyed.

The necessary installation torque for cutting the threads into the concrete

should be small in order to achieve an easy installation. Moreover, the resistance

against shearing-off of the threads should be as high as possible, so that the

threads cut into the concrete are not destroyed while tightening up the concrete

screws.

Figure 4 shows the measured torques while screwing in a concrete screw

(drill bit diameter d0 = 8mm) dependent on the swing angle. The failure

happened by shearing-off of the threads. The anchorage material consisted of

fine-grained concrete (maximum aggregate size 8mm) of the strength class B25.

32


0 250 500 750 1000 1250 1500 1750 2000

Drehwinkel [Grad]

0

25

50

75

100

125

Dre

hm

om

en

t [N

m]

Figure 4: Typical relationship between torque moment and swing angle (Concrete B25,

grading curve BC 8, d0= 8mm, Failure mode: Shearing-off of the thread [10]

Before the screw head reached the attachment, the necessary installation

torque varied only slightly. If the concrete contains coarser aggregates, torque

peaks can occur if a thread is cut into a big piece of aggregate.

After the screw head reaches the attachment, the torque on the concrete

screw rises sharply to the peak value TD. Subsequently, the shearing-off of the

threads begins and the torque decreases rapidly to zero. The damage to the

concrete threads after overtightening the concrete screw is shown in Figure 5.

Figure 5a shows the threads after the screw head reaches the attachment

(installation torque TE). For comparison, the threads cut into the concrete by the

concrete screw (d0 = 10mm) at the remaining torques of TRest ~ 0,75TD,m and

TRest ~ 0,19TD,m after reaching the peak value TD,m are shown in Figure 5b and

Figure 5c, respectively.

a) b) c)

Figure 5: Threads cut into the wall of the drilled hole, concrete screw type 2 [10]

a) Tinst = TE

b) TRest = 100Nm (~0,75 TD,m)

c) TRest = 25Nm (~0,19 TD,m)



4.1 Installation with torque wrench

Figure 6 shows the maximum measured installation torques (TE) of two

concrete screws (d0 = 10mm) with a technical approval when the screw reaches

the attachment. The embedment depth was chosen as hnom = 50mm to achieve

the failure mode of shearing-off of the threads when the screw was further

tightened after coming in contact with the attachment. Figure 7 shows the

measured failure torques TU. The type of concrete screw, the cube strength (くw ~

20N/mm² and くw ~ 70N/mm²), the grading curve of the natural round aggregates

from the Rhine valley (grading curve BC8 (maximum aggregate size 8mm) and

grading curve AB 32 (maximum aggregate size 32mm) according to DIN 1045

[17]) and as well as the drill bit diameter were varied.

According to Figure 6 the installation torque TE is not significantly

influenced by the concrete strength. TE increases, however, with an increasing

aggregate size. A substantial influencing factor is the drill bit diameter, since

there is an increase of the depth of the threads cut into the concrete if the drill bit

diameter is reduced. Furthermore, the type of concrete screw significantly

influences the installation torque.

0

10

20

30

40

50

60

70

80

TE [N

m]

Type 1

Type 2

BC 8

dcut = 10,40mm

fcc = 20N/mm²

BC 8

dcut = 10,06mm

fcc = 20N/mm²

AB 32

dcut = 10,42mm

fcc = 70N/mm²

AB 32

dcut = 10,08mm

fcc = 70N/mm²

Figure 6: Influences on the installation torque moment TE [9]

Upon further tightening after the screw head reached the attachment,

concrete screw type 1 failed in all tests by twisting-off of the screw head (steel

failure). Consequently the variance of the failure torques is small (Figure 7). On

the other hand, concrete screw type 2 failed by shearing-off of the threads cut into

the concrete except in the tests in high strength concrete くw ~ 70 N/mm² with

maximum aggregate size (grading curve AB 32) and a tight drill hole. The failure

34


torques in case of shearing-off of the threads are barely affected by the concrete

strength and the composition of the concrete. However, they increase as was the

case for the installation torques, with decrease of the drill bit diameter.

The different failure modes of concrete screw type 1 and type 2 can mainly

be attributed to the fact that the steel strength of concrete screw type 2 is higher

than the steel strength of type 1. For that reason concrete screw type 2 needs a

larger embedment depth than type 1 to reach the failure mode of steel failure.

0

50

100

150

200

250

TU [N

m]

steel failure, Type 1 steel failure, Type 2shearing off of thread, Type 2

BC 8

dcut = 10,40mm

fcc = 20N/mm²

BC 8

dcut = 10,06mm

fcc = 20N/mm²

AB 32

dcut = 10,42mm

fcc = 70N/mm²

AB 32

dcut = 10,08mm

fcc = 70N/mm²

Figure 7: Influences on the failure torque moment TU [9]

By increasing the embedment depth the installation torque increases only

slightly because the threads are mainly cut into the concrete by the flanks of the

screw thread at the head of the screw.

On the other hand, the failure torque in the case of shearing-off of the

threads cut in the concrete increases with increasing embedment depth (Figure

8), because more threads have to be sheared off. The embedment depth required

by the technical approvals with hnom œ 70mm is significantly larger than the

embedment depth used in the tests shown in Figure 7. This ensures that the

failure mode steel failure occurs and not the failure mode shearing-off of the

threads (Figure 8) if the concrete screw is overtightened during installation.



0

50

100

150

200

250

300

0 10 20 30 40 50 60 70 80

hnom [mm]

Failu

re M

om

en

t [N

m]

steel failure

concrete failure

setting depth according to Technical Approval

fcc ˜ 30N/mm²

grading curve BC 8

Figure 8: Influence of the embedment depth on the failure torque moments [11]

4.2 Installation with electric-screw-gun

While the installation of a concrete screw with a torque wrench or a screw-

wrench requires more than 30 seconds for the screw head to reach the

attachment, installation with a high-performance electric-screw-gun requires

only one to two seconds. For this reason, in practice concrete screws are usually

screwed-in with an electric-screw-gun. In the setting tests electric-screw-guns

with a maximum moment higher than the steel failure torque moment of the

concrete screws were used. Nevertheless, the concrete screws failed by shearing-

off of the threads after the screw head reached the attachment. The time between

reaching the attachment and shearing-off of the threads tK increases with

increasing embedment depth (Figure 9).

0

3

6

9

12

15

40 45 50 55 60 65 70 75

hnom [-]

t K [

sec]

test stopped

Figure 9: Influence of the embedment depth on the time until shearing-off of the threads cut

into the wall of the drilled hole (d0 = 10mm, grading curve BC 8, くw = 26N/mm², dcut =

10,41mm, electric-screw-gun 1)

36


The time between reaching the attachment and shearing-off of the threads

is little affected by the concrete strength and the composition of the concrete

(Figure 10). It is affected significantly by the drill bit diameter, the type of

concrete screw and the type of electric-screw-gun used for installation (Figure

11).

0

2

4

6

8

10

12

14

t K [

sec]

shearing-off of the threads

BC 8

くw = 20N/mm²

dcut = 10,41mm

AB 32

くw= 26N/mm²

dcut = 10,43mm

Figure 10: Influence of composition of concrete on the time until failure tK

(d0 = 10mm, hnom = 50mm)

0

10

20

30

40

50

tK [

se

c]

electric-screw-gun 1 electric-screw-gun 2

Type 1 Type 2 Type 1 Type 2

Figure 11: Influence of electric-screw-gun and type of concrete screw on the time until

failure (d0=10mm, くw = 20N/mm², grading curve BC 8, hnom = 50mm, dcut = 10,41mm) [9]

In practice it may occur that concrete screws are unscrewed after the screw

head reaches the attachment (e.g. for easier installation of a group). Therefore,

the influence of unscrewing concrete screws on the time until failure tK was

investigated. Screws without unscrewing were tested for comparison. Figure 12

shows the test results. If concrete screws are unscrewed one complete turn with

a screw-wrench after the screw head reaches the attachment and then screwed in



again with an electric-screw-gun, the minimum time until failure tK decreases in

comparison with concrete screws that were not unscrewed. If the unscrewing of

the concrete screw takes place with an electric-screw-gun, the time until failure

tK decreases significantly because it is not possible to unscrew concrete screws

in a controlled manner with an electric-screw-gun.

0

4

8

12

16

tK [

sec]

installation with

electric-screw-gun

installation /unscrewing /

installation with electric-

screw-gun

installation with electric-screw-

gun, unscrewing with torque

wrench, installation with electric-

screw-gun

Figure 12: Influence of unscrewing of concrete screws on the time tK until failure (d0 = 10mm,

fcc = 26N/mm², grading curve BC8, dcut = 10,44mm, hnom = 60mm, electric-screw-gun 1)

4.3 Remaining load-carrying capacity

To investigate the influence of the installation torque, i. e. the over-

tightening of the concrete screw, on the pull-out failure load, the concrete screws

were installed until the screw head reached the attachment (T = TE), prestressed

with T ~ 0,9 TD,m or until the torque moment fell to a preset value T = TRest after

reaching the maximum torque. Afterwards the concrete screws were pulled out.

Figure 13 shows the measured failure loads depending on the installation torque.

If the torque of the concrete screw is stopped immediately after reaching the

maximum torque, the measured pull-out failure loads are in the same range like

in the tests with concrete screws that were prestressed with T = TE or with

T ~ 0,9 TD,m. Furthermore, the load-displacement behaviour does not differ

significantly (Figure 14). If the concrete screws are turned further, the failure

load falls rapidly, because the threads cutting into the wall of the drilled hole are

destroyed (cp. Figure 5). Furthermore, the load-displacement behaviour is less

favourable. The behaviour shown in Figure 13 and Figure 14 also applies to

other types of concrete screws if they are seated with an embedment depth at

which shearing-off of the threads is possible.

38


0

2

4

6

8

10

12

14

16

00,20,40,60,811,21,4

T/TU,m [-]

Nu [

kN

]

T = TE

before shearing

0,9

TU,m = 135Nm

after reaching TU,m

Figure 13: Influence of torque before and after reaching the failure torque on the pull-out

load (d0 = 10mm, hnom = 50mm, dcut = 10,42 mm, fcc = 30N/mm²)

0 2 4 6 8 10

s [mm]

0

2

4

6

8

10

12

14

16

Nu

[kN

]

T = 0.19xTD,m

T = 0,75xTD,m

T = TE

T = 0,19 TD,m

T = 0,75 TD,m

Figure 14: Influence of the torque moment before and after reaching the

failure torque on the load-displacement curves

4.4 Required embedment depth

In practice it cannot be excluded that concrete screws are further tightened

after the screw head reaches the attachment, e. g. if the electric-screw-gun is not

stopped immediately or if the attachment should be tightened against the surface

of the concrete slab with a standard screw-wrench. Unscrewing of the concrete

screws and screwing them in again can also occur. This may damage the threads

cut into the wall of the drilled hole, if the embedment depth is not deep enough

because in practice it is normally not possible to stop the installation after

reaching the maximum torque TD. This has been shown by experiences in

practice. A check of concrete screws (d0 = 6mm) that were seated with a small

embedment depth showed that shearing-off of the threads during the installation

had occurred with about 15% of the screws.



To avoid damage of the threads cut into the concrete, the embedment depth

of the concrete screws with a technical approval of the DIBt was defined such

that steel failure and not shearing-off of the threads will occur during installation

with a standard screw wrench (Figure 8). At this embedment depth a long period

of time is needed to shear-off of the threads using an electric-screw-gun. It is

assumed that in practice a time period as long as this is not applied.

Concrete screws with a larger core diameter than the concrete screws with

a technical approval have very high torque moments in the case of steel failure.

Therefore, it makes no sense to evaluate the minimum embedment depth of

these concrete screws since that steel failure occurs. Presently a new concept for

concrete screws with d0 > 10mm is being developed to avoid the damage of the

threads cut into the concrete during the installation.

5. LOAD BEARING BEHAVIOUR OF CONCRETE SCREWS

5.1 Load-displacement behaviour and failure mode

Figure 15 shows typical load-displacement curves measured in pull-out

tests under tension load in cracked (〉w = 0,3mm) and non-cracked concrete.

They increase steeply and lie close together. The failure modes were pull-out

and concrete cone failure.

0 0.5 1 1.5 2 2.5 3

s [mm]

0

5

10

15

20

25

30

Nu

[k

N]

non-cracked concrete

cracked concrete 〉w = 0,3mm

Figure 15: Typical load-displacement curves for concrete screws in cracked and non-

cracked concrete (hef = 65mm, dcut = 10,25mm, fcc = 30N/mm², grading curve AB 16) [11]

Concrete screws with a small embedment depth fail through a concrete

failure cone that starts at the first bearing thread at the tip of the concrete screw

(Figure 16a). If the embedment depth increases, only the concrete at the surface

breaks out and the remaining portion of the screw is pulled out (Figure 16b).

40


The observed failure modes differ from the failure mode of expansion anchors

and undercut anchors. These anchors transfer the load into the concrete near the

end of the embedment depth and the concrete cone failure begins near the end of

the anchor. On the other hand, concrete screws discharge the load over the entire

embedment depth into the concrete.

The failure mode shown in Figure 16 is similar to that of bonded anchors

but the failure load of bonded anchors increases nearly linearly with increasing

embedment depth (hef) [12]. Whereas the failure load of concrete screws

increases by hef1,5 (see section 5.2.1). Therefore, the failure of concrete screws is

due to exceedence of the concrete tension strength in the failure cone and not to

pullout as for bonded anchors.

a)

b)

Figure 16: Typical concrete failure cones [9]

a) hnom = 50mm

b) hnom = 90mm

5.2 Failure Loads

To clarify the influence of different parameters on the failure loads of

concrete screws, pull-out tests in concrete slabs with a cube strength of about くw

~ 30N/mm² were performed. The concrete slabs were produced from concrete

with a grading curve AB16 (aggregates with maximum size 16mm) according to

DIN 1045 [17]. Natural round aggregates from the Rhine valley were used. For

drilling of the holes, drill bits with medium bit diameter according to [4] were

used. The measured failure loads were normalized by くw0,5 to くw = 30N/mm²

because the failure is caused by exceedence of the concrete tension strength.



Influence of the embedment depth

Figure 17 shows the measured failure loads of concrete screws produced by

manufacturer 1 for various embedment depths hef. The investigated parameter is

the drill hole diameter d0. The effective embedment depth was determined

according to equation (1).

hef = hnom – 0,5*h – hS (1)

with:

hnom = length between end of concrete screw and concrete surface

h = threaded length of concrete screw

hS = length of screw without thread

Equation (1) considers that load discharge starts with a transfer from the

top of the concrete screw that is dependent on the kind of thread of the concrete

screw. It enables a better comparison of the test results of concrete screws from

different manufacturers, i. e. with different kind of threads.

According to Figure 17 the failure loads of concrete screws increase

proportionally to hef1,5. That relation also applies to expansion and undercut

anchors failing by concrete cone failure. Figure 17 applies to concrete screws

with threads over the complete embedment depth. If concrete screws only have

threads over part of the embedment depth, the failure load will not increase after

reaching a certain embedment depth, because the failure mode changes to pull-

out failure (shearing-off of the concrete between the screw flanks). This is

similar to the behaviour of torque-controlled expansion anchors, where the

failure mode changes with increasing embedment depth from concrete cone

failure to pull-through failure [14].

0

10

20

30

40

50

60

70

0 20 40 60 80 100 120

hef [mm]

Nu [

kN

]

do = 8mm

do = 10mm

do = 12mm

do = 14mm

do = 18mm

Nu = g*hef1,5

くw = 30N/mm²

Nu = Nu,Versuch*(30/くw)0,5

Figure 17: Influence of embedment depth on failure load

42


Influence of concrete screw Diameter

For identification of concrete screws, the drill hole diameter is used

because concrete screws from different manufacturers intended for the same

drill hole diameter differ in their core and outside diameters. Figure 18 shows

failure loads for various drill hole diameters. Parameter is the embedment depth

hef. For a comparison at the same effective embedment depth the measured

failure loads were normalized by hef1,5. The straight lines in Figure 18 show the

trends of the test results. One can see that the failure loads decrease slightly with

increasing drill hole diameter at lower embedment depth and the failure loads

are independent of the drill hole diameter at larger embedment depth. However,

in all cases the influence of the drill hole diameter on the failure load is not

significant.

0

20

40

60

80

Nu [

kN

]

hef = 105mm

hef = 85mm

hef = 65mm

hef = 45mm

8 1210 1814

d0 [mm]

くw = 30N/mm²


Figure 18: Influence of concrete screw size on failure load

Influence of the concrete screw type

Figure 19 shows the failure loads of concrete screws with a drill bit

diameter d0 = 10mm to various embedment depths. The investigated parameter

is the type of concrete screw. The figure shows that the failure loads of different

concrete screws differ a little under similar conditions. That can be attributed to

the different threads. The different load bearing performances of the concrete

screws were considered when determining the characteristic resistance for

concrete cone failure in the technical approvals of the DIBt.



0

10

20

30

40

50

60

0 10 20 30 40 50 60 70 80 90

hef [mm]

Nu [

kN

]

Typ 1

Typ 2

くw = 30N/mm²


Nu = g*hef1,5

Figure 19: Influence of the type of concrete screw (d0 = 10mm) on the failure loads

Influence of Screw Spacing

To investigate the influence of the screw spacing on the failure loads, groups

with four concrete screws in concrete with the concrete strength near くw ~

30N/mm² were tested. The screw spacing was varied. At small spacing the

groups failed by a combined concrete cone failure (Figure 20a). At a spacing of

s = 2 hnom a changeover to several failure cones was observed (Figure 20b).

a)

b)

Figure 20: Concrete failure cone of square groups with concrete screws

a) s = 1 hnom and

b) s = 2 hnom (d0 = 10mm, hnom = 70mm)

While the screw spacing does not significantly influence the stiffness at the

beginning of the tests, the failure loads and the displacement at failure load

increase with increasing screw spacing (Figure 21). Figure 22 shows the failure

loads of square groups based on the average failure load of a single concrete

screw as a function of the relationship between spacing and effective

embedment depth.

The failure loads of groups increase with increasing screw spacing, but

they did not reach the fourfold value valid of a single anchor at a larger spacing.

The reason for this is not yet known. 44


0 0.25 0.5 0.75

Verschiebung [mm]

0

15

30

45

60

75

90

Nu [kN

]

s = 3 hnom

s = 1 hnom

s = 3 hnom

s = 1 hnom

s = 3 hnom

s = 1 hnom

s = 3 hnom

s = 1 hnom

s = 3 hnom

s = 1 hnom

s = 3 hnom

s = 1 hnom

s = 3 hnom

s = 1 hnom

s = 3 hnom

s = 1 hnom

s = 3 hnom

s = 1 hnom

displacement [mm]

Figure 21: Typical load-displacement curves of groups of concrete screws

(d0 = 10mm, hnom = 50mm)

0

1

2

3

4

5

0,0 0,5 1,0 1,5 2,0 2,5 3,0 3,5 4,0

s/hef [-]

Nu/N

0u [

-]

N0u = medium failure load of a single concrete screw

くw = 30N/mm²

Nu = Nu,test*(30/くw)0,5

ef Ncr,0

Nc,

Nc, h3s for A

A=

Figure 22: Failure loads of square groups of concrete screws based on the average failure

load of a single concrete screw (d0 = 10mm)

Influence of cracks in concrete

The results shown so far apply for non-cracked concrete. In structural

members of reinforced concrete one can assume that cracks in the concrete

appear. If a concrete screw is anchored in a crack, the undercut area of the

thread flanks is reduced in comparison to non-cracked concrete. Furthermore,

the axially symmetric state of stress around the screw is disturbed by the crack.

These effects cause that the stiffness of the fastening and the failure loads in

comparison to non-cracked concrete are reduced (Figure 15). The decrease of

the failure load averages about 30 % at a crack width of 0,3mm. This reduction

is on the same order of magnitude as that for expansion or undercut anchors.



6. CALCULATION OF THE AVERAGE FAILURE LOAD OF SINGLE ANCHORS

Figure 23 shows the failure loads of different types of concrete screws with

varying outside diameters as a function of the embedment depth. For

comparison, the bearing capacity after Equation (2) is shown. The equation

describes the average concrete cone failure load of expansion and undercut

anchors [13].

w

1,5

ef

0

u.c *h*13,5 β=N (2)

with

くw = concrete cube compressive strength (200mm)

hef = effective embedment depth

The failure loads of concrete screws are below the values predicted by

equation (2). This can be attributed to the different failure modes (Chapter 5.1).

If the influence of the concrete screw type and the diameter is neglected, the

measured failure loads can be described with adequate accuracy by Equation (3).

w

1,5

ef

0

u *h*10,5 β=N (3)

with

hef = effective embedment depth after Equation (1)

The values Nu,test/Nu,calculation are normally distributed around average value

of 1,0 with a coefficient of variation v ~ 15% (Figure 24). According to the test

results the failure loads of concrete screws are about 20% lower than the failure

loads of expansion and undercut anchors. Equation (3) does apply to fastenings

in non-cracked concrete. For fastenings in cracked concrete Equation (3) has to

be multiplied with the factor 0,7.

46


0

20

40

60

80

0 20 40 60 80 100 120

hef [mm]

Nu [

kN

]

くw = 30N/mm²

Nu = Nu,Versuch*(30/くw )0,5

1,5ef

h*w

*13,5N 0cu, β=

1,5ef

h**10,5N w0u β=

Figure 23: Maximum pull-out loads of concrete screws in non-cracked concrete as a function

of the embedment depth hef and comparison with prediction by CC-method for concrete cone

failure

0

5

10

15

20

25

30

0,05 0,25 0,45 0,65 0,85 1,05 1,25 1,45 1,65

Nu (test) / N0u (calculation) [-]

nu

mb

er

[-]

n = 158

x = 0,98

v = 15%

Figure 24: Histogram of the quotient of measured and calculated concrete failure load by

tension tests with concrete screws

The failure load Nu of groups with concrete screws evaded centrically can

be calculated according to the CC-Method (Equation (4))

0

u

0

u0

Nc,

Nc,

u N*nN*A

A≤=N (4)

with

0Nc,A = area of concrete cone of an individual anchor with large spacing

and edge distance at the concrete surface, idealizing the

concrete cone as a pyramid with height equal to hef and a base

length equal to scr,N



Ac,N = actual area of concrete cone of the anchorage at the concrete

surface. It is limited by overlapping concrete cones of adjoining

anchors (s ø scr,N) as well as by edges of the concrete member (c

ø ccr,N). Examples for the calculation of Ac,N are given in [14,

15]

N = number of anchors of the group

For expansion and undercut anchors the critical anchor spacing is scr,N =

3hef ([14, 15]). The failure load of concrete screws at the same embedment depth

is lower than that of expansion and undercut anchors. However, Figure 22 shows

that the test results can be described approximately with scr,N = 3 hef.

7. DESIGN OF FASTENINGS WITH CONCRETE SCREWS THAT MEET TECHNICAL APPROVALS

In references [1] to [3] the design of fastenings with concrete screws takes

place according to design method A in [16], which is based on the CC-Method.

The characteristic values necessary for the design of fastenings with concrete

screws with d0 = 10mm under tension load are assembled in Table 1. The high

characteristic resistance NRk,s at steel failure cannot be exploited because it is

higher than the characteristic resistance NRk,p at pullout. The values NRk,p were

determined from the tests for the technical approvals. The behaviour of the

fastening in cracks with opening and closing crack widths was considered as

well. The design at the failure mode “concrete cone failure” takes place

according to the CC-Method for expansion and undercut anchors which is

described in detail in [14, 15].

For consideration of the lower load capacity of concrete screws in

comparison to expansion and undercut anchors in Equation (5) a reduced

embedment depth hef,cal, in comparison to equation (1), is used to calculate the

characteristic resistance against concrete cone failure for a single concrete

screw. The embedment depths used are stated in Table 1.

wWN

1,5

calef,

0

cu, **h*7,0 ψβ=N (5)

with

くWN = nominal value of the cube strength after DIN 1045 [17]

hef,cal = nominal effective embedment depth (Table 1)

ねW = 1,0 for fastenings in cracked concrete

= 1,4 for fastenings in non-cracked concrete 48


Table 1: Characteristic values for the resistances under tension load of concrete screws

(d0 = 10mm) with a Technical Approval of the DIBt

Type of concrete screw [1] [2] [3]

Drill hole diameter d0 [mm] 10 10 10

Embedment depth hnom [mm] 70 75 85

Steel failure

Characteristic resistance NRk,s [kN] 54,1 75,4 58

Pull-out failure

Characteristic resistance in non-cracked concrete B 25

NRk,p [kN] 12,0 16,0 20,0

Characteristic resistance in cracked concrete B 2

NRk,p [kN] 7,5 12,0 12,0

Concrete cone failure

Nominal effective embedment depth

hef,cal [mm] 50 50 60

Characteristic screw spacing scr,N [mm] 150 150 180

Characteristic edge distance ccr,N [mm] 75 75 90

8. SUMMARY





Concrete screws transfer tensile loads into the base material by mechanical

interlock of the threads. Due to their load-bearing mechanism, concrete screws

with a technical approval of the DIBt can be used for fastenings in cracked and

non-cracked concrete.

The typical failure mechanism for concrete screws is concrete-cone failure.

With increasing embedment depth the ratio of the depth of the concrete failure

cone to the embedment depth decreases. The failure load of concrete screws

with continuous threads along the entire embedment depth increases

proportionally to hef1,5 (hef = effective embedment depth), but it is about 20 %



smaller than the failure load of expansion and undercut anchors with the same

embedment depth.

In order for concrete screws to function properly, the threads cut into the

wall of the drilled hole must not be damaged during the installation. This

requirement is achieved by using the embedment depth defined in the Technical

Approvals.

9. ACKNOWLEDGMENT

The primary funding for this research was provided by the Adolf Würth

GmbH & Co. KG. The support of this manufacturer is very much appreciated.

Special thanks are also accorded to Beate Vladika and Matthew Hoehler who

spent many hours in improving the English.

10. REFERENCES

[1] [Deutsches Institut für Bautechnik] Allgemeine Bauaufsichtliche

Zulassung Z-21.1-1712 für Hilti Schraubanker HUS-H zur Verankerung

im gerissenen und ungerissenen Beton, Berlin, 2001


Zulassung Z-21.1-1549 für HECO-MULTI-MONTI-Schraubanker MMS

zur Verankerung im gerissenen und ungerissenen Beton, Berlin, 2001


Zulassung Z-21.1-1624 für Toge Betonschraube TSM zur Verankerung

im gerissenen und ungerissenen Beton, Berlin, 2001

[4] European Organisation for Technical Approvals (EOTA): Leitlinie für die

europäisch-technische Zulassung von Metalldübeln zur Verankerung in

Beton. Deutsches Institut für Bautechnik, 28. Jahrgang, Sonderheft Nr. 16,

Berlin, Dezember 1997

[5] Küenzlen, J. H. R.; Eligehausen, R.: Setz- und Ausziehversuche in

ungerissenem Beton mit Schraubdübeln. Bericht Nr. AF01/01-E00202/1,

Institut für Werkstoffe im Bauwesen, Universität Stuttgart, 2001, nicht

veröffentlicht

[6] Küenzlen, J. H. R.; Eligehausen, R.: Tragverhalten von Schraubdübeln in

niederfestem Beton. Bericht Nr. W8/1-01/1, Institut für Werkstoffe im

Bauwesen, Universität Stuttgart, 2001, nicht veröffentlicht

50


[7] Küenzlen, J. H. R.; Eligehausen, R.: Tragverhalten von Schraubdübeln in

niederfestem Beton. Bericht Nr. W8/3-01/3, Institut für Werkstoffe im

Bauwesen, Universität Stuttgart, 2001, nicht veröffentlicht

[8] Küenzlen, J. H. R.; Eligehausen, R.: Einfluss verschiedener Parameter auf

die Höchstlasten von Schraubdübeln, Institut für Werkstoffe im

Bauwesen, Universität Stuttgart, 2001, Bericht in Vorbereitung

[9] Küenzlen, J. H. R.; Sippel, T. M.: Behaviour and Design of Fastenings

with Concrete Screws. In: RILEM Proceedings PRO 21 „Symposium on

Connections between Steel and Concrete“, Cachan Cedex, 2001, S. 919-

929.

[10] Küenzlen, J. H. R.: Drehmomentversuche mit Schraubdübeln in

ungerissenem Beton. Jahresbericht 2000/2001, Institut für Werkstoffe im

Bauwesen, Universität Stuttgart, 2001

[11] Eligehausen, R.; Hofacker, I. N.; Spieth, H. A.; Küenzlen, J. H. R.: Neue

Entwicklungen in der Befestigungstechnik, Tagungsband, IBK-Bau-

Fachtagung 263: Dübel und Befestigungstechnik, 2000, S. 2.1-2.14,

[12] Meszaros, J.,: Tragverhalten von Einzelverbunddübeln unter zentrischer

Kurzzeitbelastung. Dissertation, Universität Stuttgart, 2001

[13] Eligehausen, R.; Fuchs, W.; Mayer, B.: Tragverhalten von

Dübelbefestigungen bei Zugbeanspruchung. Beton + Fertigteil-Technik

1987, Heft 12, S. 826-832 und 1988 Heft 1, S. 29-35.

[14] Eligehausen, R.; Mallée, R.: Befestigungstechnik im Beton- und

Mauerwerkbau. Ernst & und Sohn, Berlin, 2000.

[15] Fuchs, W.; Eligehausen, R.: Das CC-Verfahren zur Berechnung der

Betonausbruchlast von Verankerungen. Beton- und Stahlbetonbau, 1995,

Heft 1, S. 6-9, Heft 2, S. 38-44, Heft 3, S. 73-76.

[16] Deutsches Institut für Bautechnik: Bemessungsverfahren für Dübel zur

Verankerung in Beton (Anhang zum Zulassungsbescheid). Berlin, 1993

[17] DIN 1045, Beton und Stahlbeton, Bemessung und Ausführung, Ausgabe

1978



52

Prestressed hollow-core concrete slabs – Problems and possibilities in fastening techniques

PRESTRESSED HOLLOW-CORE CONCRETE SLABS – PROBLEMS AND POSSIBILITIES IN FASTENING TECHNIQUES

SPANNBETON-HOHLDECKENPLATTEN – PROBLEME UND MÖG-LICHKEITEN IN DER BEFESTIGUNGSTECHNIK

DALLES ALVEOLAIRES EN BÉTON PRÉCONTRAINT – PROBLÈ-MES ET POSSIBILITÉS DES TECHNIQUES D'ANCRAGE

Clemens Lutz

SUMMARY

In the present, prestressed hollow-cored concrete slabs are tendentiously

used as ceiling systems. Therefore, fastening techniques with regard to these

slabs gain in an increasing importance. In this article, advantages of these mem-

bers and problems by application of anchors are described, and different struc-

tural responses between several types of anchors are experimentally determined.

Accordingly, it seems to be essential to choose or to adapt suitable anchors for

ceiling systems.

ZUSAMMENFASSUNG

Spannbeton-Hohlplatten finden als Deckensystem eine immer breitere

Verwendung und einen immer größeren Anwendungsbereich. Somit gewinnt

auch eine korrekte Befestigung in diesen Platten zunehmend an Bedeutung. In

diesem Artikel wird auf die Vorteile der Platten, aber auch auf die Problematik,

die bei der Montage von Dübeln entstehen, eingegangen. Ferner wird gezeigt,

dass es große qualitative Unterschiede bezüglich der Tauglichkeit verschiedener

Befestigungssysteme gibt, weshalb eine sorgfältige Auswahl bzw. Anpassung

prinzipiell geeigneter Dübel stattfinden muss.

RESUME

Actuellement, les dalles alvéolaires en béton précontraint sont utilisées de

plus en plus fréquemment. Par conséquent, les ancrages appropriés gagnent

d'importance. Dans cet article, nous traitons les avantages de ces dalles et les

problèmes reliés à l'utilisation de chevilles. De plus, nous montrons que les dif-


C. LUTZ

férents systèmes révèlent de grandes différences qualitatives, et qu'il est par

conséquent essentiel de choisir et d'adapter des ancrages adéquats.

KEYWORDS: prestressed hollow-core concrete slab, ceiling, anchorageable thick-

ness, anchor

1. ADVANTAGES AND PROBLEMS

Prestressed hollow-cored concrete slabs made of high-strength concrete are

prefabricated concrete members with large hollow proportions. In practice, they

are interconnected after assembly by joint grouting compound. In comparison

with conventional concrete members, this type of concrete plates has a lot of

economical advantages, especially in saving material, energy and in reducing

weight of transportation. Outstanding features are quality control, schedule time

and costs. Additionally, formworks which are used to produce in-situ concrete

are saved in application of these slabs. In the present, this ceiling system is in-

creasingly used in industrial buildings, office buildings and also in domestic ar-

chitecture. Figure 1 shows cross sections of two types of prestressed hollow-

cored concrete slabs (with different minimal anchorageable material thickness:

25 mm and 30 mm).

1

4

4

Figure 1: Cross sections of two types of prestressed hollow-cored concrete slabs (1: cavity, 2: prestressed wire, 3: steel, 4: minimal anchorageable material

thickness dmat, here: 25 mm and 30 mm)

54


In spite of above mentioned advantages, the application of anchors in

prestressed hollow-cored concrete slabs is not satisfied, particularly in case of

thin members. The worst case is that anchors are fastened in near of position A

(see fig. 2). The distance between the opposite of casting side and the hollows is

the smallest. This distance is defined as minimal anchorageable material thick-

ness dmat (in German: Spiegeldicke). For some types of slabs the value dmat is

very small. This small value of thickness is relevant for load carrying capacity of

anchor systems. Tables 1 and 2 show experimentally measured thickness dmat of

slabs with a minimal anchorageable thickness of 30 mm and 25 mm respec-

tively. All values in table 1 are above 30 mm. Some values in table 2 are only

just 25 mm. For slabs with dmat = 25 mm there is no sufficient reserve in com-

parison to slabs with dmat= 30 mm! Furthermore, crashing of concrete closed to

the hollows often occurs during drilling. Consequently, the minimal anchorage-

able material thickness and also the effective anchorage depth for anchors are

reduced (see fig. 3) and load carrying capacities of ceiling systems are nega-

tively influenced. Therefore, it is necessary to determine whether all types of

fasteners are suitable to be used in prestressed hollow-cored concrete slabs. Ad-

ditionally, it is prohibited to install an anchor in near of a strand of wire because

of interests of safety (zone C in fig 2).

zone B

zone C zone C

zone A

zone B

Figure 2: Sectors of a prestressed hollow-cored concrete slab. Zone A: minimal anchorageable material thickness; Zone B: anchorageable sector; Zone C: prohibited sector for fastenings because of interests of security (prestressed concrete wire)


C. LUTZ

Table 1: measured values dmat (slab with a minimal anchorageable thickness of 30 mm)

39,6 42,3

43,1 44,2

42,3 45,0

40,9 43,8

dmat [mm] (measured mini-

mal anchorageable material

thickness)

37,8 39,2

Range [mm] 38 (>30) to 45

Average [mm] 41,8

Variation coeff. [%] 5,70

Table 2: measured values dmat (slab with a minimal anchorageable thickness of 25 mm)

25,1 26,7

26,7 28,5

27,6 28,5

26,8 27,5

26,2 28,6

dmat [mm] (measured mini-

mal anchorageable material

thickness)

25,0 26,1

Range [mm] 25 to 29

Average [mm] 26,9

Variation coeff. [%] 4,60

dmat dmat, eff

Figure 3: The minimal anchorageable material thickness dmat after drilling is reduced (:=dmat, eff < dmat).

56


2. POSSIBILITIES AND TEST RESULTS

In general, there are three types of possibilities to fasten installation pipes,

suspendic and acoustics ceilings, lighting appliances, safety precaution systems

and beams (see fig. 4). In the following, fastenings with different types of an-

chors according to possibility (1) will be studied in detail. As results, these an-

chors applied in prestressed hollow-cored concrete slabs show their quite differ-

ent suitability.

There are already several types of special fasteners on the market, which

have approvals for using in hollow-cored concrete slabs. Objective of this work

is to investigate suitability and quality of other types of anchors in these mem-

bers. Therefore, pull-out tests were carried out in the uncracked concrete zone of

these slabs with a minimal anchorageable material thickness dmat of 25 mm and

30 mm. Herein, different types of anchors – concrete screws, injection anchors,

suspendic ceiling fasteners, deformation-controlled expansion anchors and

torque-controlled expansion anchors – were used. The sizes of anchors chosen

for these experiments were between M6 and M10.

(1) only anchors

(2) post-installed bonded rebar connections; concrete suspension

(3) construction, fastening through the slab

Figure 4: Three types of possibilities to fasten installation pipes, suspendic and acoustics ceilings, lighting appliances, safety precaution and beams


C. LUTZ

For each test, one borehole was produced with the help of a hammer drill.

Position of the borehole was chosen in such a way that the thickness of concrete

corresponds to the minimal anchorageable thickness dmat. Depth of the borehole

is equal to this minimal thickness (position A, fig. 2). Typical crashing of con-

crete closed to the hollows was often observed after drilling. Consequently, the

effective anchorage depth was reduced. After installation of the anchor system

the fastener was subjected to concentric tension up to failure. For concrete

screws, setting tests with concrete screws were also carried out additionally [1].

Figure 5 outlines an equipment for pull-out tests where one load cell, two

LVDTs and a steel support frame are used. Figure 6 shows a pull-out cone of a

concrete screw. Pull-out test results for different types of anchors are repre-

sented in following figures. Figure 7 shows measured load carrying capacities of

different types of anchors used in concrete slabs with a minimal anchorageable

material thickness of 30 mm. All test results are given in relation to the failure

load of concrete screws, type 1 (which is chosen as reference anchor).

Figure 5: Pull-out tests in a prestressed hollow-cored concrete slab

58


Figure 6: Pull-out cone of a concrete screw

From figure 7, it can be seen that the highest load carrying capacity was ob-

tained in torque-controlled expansion anchors (column 7). The average failure

load was 93% higher in comparison to anchor, type 1 (reference anchor). Rela-

tive loading carrying capacities of other anchors are summarized in the follow-

ing:

- concrete screws, type 2 (column 2) +64%

- deformation-controlled expansion anchors (column 4) +30%

- injection anchors (column 6) +12%

- special fasteners for hollow-cored concrete slabs (column 5) ±0

- concrete screws, type 1 (column 1) ±0

- suspendic ceiling fasteners (column 3) −26%


100

164

74

130

100112

193

0

50

100

150

200

250

1 2 3 4 5 6 7

Dübelart

Nu

,m [

%]

Types of anchors

Re

lative

va

lue

s o

f a

ve

rag

ed

fa

ilure

lo

ad

Nu.m

[%

]

C. LUTZ

Pull-out test results for different types of anchors (minimal anchorageable mate-rial thickness: 30 mm). All test results are given in relation to the failure load of a concrete screw, type 1 [1].

Figure 7:

1: concrete screws, type 1 2: concrete screws, type 2 3: suspendic ceiling fasteners 4: deformation-controlled expansion anchors

5: special fasteners for hollow-cored concrete slabs 6: injection anchors 7: torque-controlled expansion anchors

100

193

120

289

0

50

100

150

200

250

300

350

1 2 3 4

Dübelart

Nu

,m [

%]

Re

lative

va

lue

s o

f a

ve

rag

ed

fa

ilure

lo

ad

Nu.m

[%

]

Types of anchors

Pull-out test results for different types of anchors (minimal anchorageable mate-rial thickness: 25 mm). All test results are given in relation to the failure load of a concrete screw, type 1.

Figure 8:

1: concrete screws, type 1 2: deformation-controlled expansion anchors

3: suspendic ceiling fasteners 4: special fasteners for hollow-cored concrete slabs

60


Figure 8 shows experimental results of different types of anchors used in a

thin hollow-cored concrete slab with a minimal anchorageable material thick-

ness of 25 mm. All mean values of load carrying capacities are represented in

relation to the averaged failure load of anchor, type 1. Relative loading carrying

capacities of other types of anchors are summarized as follows:

- special fasteners for hollow-cored concrete slabs (column 4)

+189%

- deformation-controlled expansion anchors (column 2) +93%

- suspendic ceiling fasteners (column 3) +20%

- concrete screws, type 1 (column 1) ±0

For most of structural designs the averaged load carrying capacity is not

alone the value which characterizes the material properties. Displacements and

statistic values are also important factors. In figure 9 load-displacement-

diagrams on the left and right are compared (diagram a and b): Failure load and

statistic values according to the load carrying capacities of these both types of

anchors are almost the same (see also table 3 for statistic values), whereas the

displacements and statistic values according to the displacements are quite dif-

ferent. Therefore, it may be questioned, which type of anchor is more suitable

for hollow-cored concrete slabs. Anchors of type 1 behave more brittle, anchors

of type 2 behave more ductile. In this case, displacement at the permissible load

is essential. Type 1 seems to behave more positive than type 2. In figure 9 c.)

and d.) anchors of type 3 reach higher load carrying capacities on average in

comparison with type 4, but they have also higher displacements. It is harmful if

displacements are to high and come outside of linear area (see fig. 10).


C. LUTZ

displacement

load

displacement

load

a.) type 1

b.) type 2

displacement

load

displacement

loa

d

c.) type 3 d.) type 4

0 0 sx sx

Nx Nx

0 0

sy sy

Ny Ny

0 0

0 0

Figure 9: Load-displacement-curves for different types of anchors. Diagrams a.) and b.): Difference in displacement with almost the same failure load.

Diagrams c.) and d.): Difference in failure load and displacement

Table 3: Statistic values of two types of anchors in a prestressed hollow-cored concrete slab with dmat= 30 mm (see figure 9)

a.) Type 1 b.) Type 2 Variation coefficient at failure load Nu,m [%] 17 17

Displacement at 0,5 Nu,m [%] 100 560

Variation coefficient for displacement at 0,5 Nu,m [%] 11 25

62


Obviously, the load carrying capacity is not only the aspect characterizing

anchors. In principle, three groups of anchors could be distinguished according

to their load-displacement-behaviours (see fig. 11):

- Group 1: anchors with little displacements, i.e. the tested injection anchors.

- Group 2: anchors with larger displacements and high load carrying capaci-

ties, i. e. one of the tested torque-controlled expansion anchors.

- Group 3: anchors with larger displacements and reduced load carrying ca-

pacities.

0 0.5 1 1.5 2

Displacement s [mm]

Loa

d N

Linear-elastic

area

Linear-elastic area

Figure 10: Experimental load-displacement-curves and simplified linear curves for two types of anchors in a prestressed hollow-cored concrete slab (here: with dmat= 30 mm).


C. LUTZ

Group 1

Group 2

Group 3

Displacement s [mm]

Figure 11: Load-displacement-curves of anchors in prestressed hollow-cored concrete slabs (here: with dmat= 30 mm) can be distinguished in three groups. There are serious differences

in carrying load capacities, but also in displacements

Lo

ad

N [

kN

]

According to the above mentioned observations, it can be concluded that

anchors of group 1 in connection with serious statistic values in according to

load carrying capacities and displacements are most suitable to apply in

prestressed hollow-cored concrete slabs. Though, further tests have to be done

(sustained load, fatigue tests and so on).

REFERENCES

[1] LUTZ, C.: Anchors in prestressed hollow-cored concrete slabs. IWB

Activities 3 (2001)

[2] LUTZ, C.: Nachträgliche Befestigungen in Spannbeton-Hohlplatten. IWB

Mitteilungen, Jahresbericht 2000-2001

64

Pore-size determination from penetration tests on concrete with n-decane

PORE-SIZE DETERMINATION FROM PENETRATION TESTS ON CONCRETE WITH N-DECANE

PORENGRÖSSENBESTIMMUNG AUS N-DECAN-EINDRING-VERSUCHEN IN BETON

DETERMINATION DES PORES DU A LA PENETRATION DE N-DECANE EN BETON

Hans W. Reinhardt, Arno Pfingstner

SUMMARY

Absorption and infiltration tests on concrete mixes have been carried out

with n-decane. The test results show a good agreement with theoretical predic-

tions. The results indicate that the main parameter on the penetration is the wa-

ter-cement ratio. Pore sizes which are reached are different for the absorption

test and the infiltration test.

ZUSAMMENFASSUNG

Absorptions- und Infiltrationsversuche wurden an verschiedenen Betonen

mit n-Decan durchgeführt. Die Ergebnisse haben eine gute Übereinstimmung

mit theoretischen Vorhersagen gezeigt. Die Ergebnisse lassen den Schluss zu,

dass der Wasserzementwert die maßgebliche Größe für das Eindringverhalten

ist. Porengrößen, die erreicht wurden, sind bei Absorptions- und Infiltrationsver-

suchen verschieden.

RESUME

Des essais d'absorption et d'infiltration ont été réalisés sur des bétons de

différentes compositions avec du n-décane. Les résultats montrent une bonne

concordance avec des prévisions théoriques. Les résultats indiquent que le pa-

ramètre principal pour la pénétration dans le béton est le rapport eau-ciment. Les

tailles des pores atteintes sont différentes pour les essais d'absorption et les es-

sais d'infiltration.

KEYWORDS: Concrete, n-decane, penetration, absorption, infiltration, high-

performance concrete


H.W. REINHARDT, A. PFINGSTNER

MOTIVE

Since several years, tests have been carried out on the penetration behav-

iour of organic fluids in concrete [1-5]. Two types of tests are being carried out:

the capillary suction test and the infiltration test with a certain hydraulic head.

The test results show that the capillary suction test satisfies usually technical

requirements for the application of the material properties in assessing the be-

haviour of real structures. However, from a scientific point of view both tests

reveal more than the material property only. Comparing the test results of both

test methods one can calculate the pore radius of capillary pores in concrete.

This will be shown in the following.

EXPERIMENTAL SET UP

For suction tests, specimens were placed into the test fluid. The samples

rested on glass rods to allow free access of the testing fluid to the inflow surface.

The fluid level was approx. 10 mm above the lower end of the specimen. Pene-

tration occurred by capillary forces acting against gravity.

The experimental set-up for infiltration tests described in DAfStb guide-

line [6] was slightly modified. Preliminary tests had proven that the pressure

head of 40 (+/- 5) cm specified there was too small to obtain measurable differ-

ences to capillary suction tests for high performance concretes. The required ex-

ternal pressure was estimated by calculation from the pore radius distributions of

comparable concretes and fixed to 0,2 bar (20 kN/m2) for all infiltration tests.

This pressure was produced by a nitrogen bottle connected to the funnels on the

samples by tubes.

66


100 mm

150 m

m

Concrete cylinder

Glass funnel

Connecting tube,

Testing fluid

to nitrogen bottle

N2 - 0,2 bar (20 kN/m2) pressure

100 mm

Concrete cylinder

Epoxy resin

coating

appro

x.

10 m

m

Testing fluid

a) b)

Fig. 1. Experimental set-up: a) suction test, b) infiltration test

Details on specimen preparation, storage and curing are given in [7].

THEORY

Modelling the pores as a single tube, capillary suction is governed by the

square root of time law acc. to Eq. (1)

1 2

02

/cos r

eσ Θ

η

Ã= ÄÅ Ö

0tÔÕ (1)

with e0 = penetration depth of capillary suction test, Θ = contact angle,

r = pore radius, η = dynamic viscosity, t0 = test duration.

If an external pressure pa is applied the capillary pressure is increased by pa

and Eq. (1) reads



1 222

4

/

p a

cos re p

r

σ Θ

η

ÃÃ Ô= +ÄÄ Õ

Å ÖÅ ÖptÔÕ (2)

with ep = penetration test of infiltration test, tp = test duration. When Eqs.

(1) and (2) are divided by each other one gets

2

0

0

21

p

p a

e t cosr

e t p

σ ΘÃ ÔÃ ÔÄ Õ= −Ä ÕÄ ÕÅ ÖÅ Ö

(3)

Eq. (3) is an explicit equation for the effective pore radius. The effective

pore radius is a fictitious pore size which is related to the single size pore model,

i. e. concrete consists only of parallel pores of this size. Of course this model is a

simplistic one since pores of many sizes take part in the capillary suction which

depends on the pore size distribution of concrete. Studying the effective pore

radius one should carry out experiments of short and long duration, during

which readings are taken. However, in the following experiments only meas-

urements after 72 hours are taken and the results explain the way of evaluation.

Eq. (3) can be written in a different way when the penetration coefficient

B = e t -1/2 has already been evaluated. Eq. (3) becomes

2

0

21

p

a

B cosr

B p


(4)

with the index 0 referring to the suction test and p to the infiltration test.

A similar relation can be derived if the sorptivity S is used instead of the

penetration coefficient B. S is linked to B via the porosity ε

S B ε= ⋅ (5)

with ε = const. Eq. (4) becomes

2

0

21

p

a

S cosr

S p


.

68


CONCRETES USED

The composition of the concretes used is given in Table 1. Table 2 shows

some relevant properties. The air content has been measured in the fresh state.

Table 1. Composition of concretes

Concr. Aggregates

[kg/m3]

Grading Cement

[kg/m3]

Type of ce-

ment

Wadded

[kg/m3]

SF (solid)

[kg/m3]

RE

[kg/m3]

SP

[kg/m3]

Wtot./(C+SF)

-

MR 1905 AB 16 320 CEM I 32,5 R 160 0 0 2.50 0.50

M1 1822 AB 16 338 CEM I 32,5 R 186 0 0 0.80 0.55

M2 1535 AB 2 467 CEM I 32,5 R 257 0 0 0 0.55

M3 1882 AB 32 309 CEM I 32,5 R 170 0 0 0 0.55

M4 1895 U 16 309 CEM I 32,5 R 170 0 0 0 0.55

M5 1677 C 16 405 CEM I 32,5 R 223 0 0 0.50 0.55

M6 1769 AB 16 485 CEM I 42,5 R 150 0 0 9.30 0.32

M7 1762 AB 16 465 CEM I 42,5 R 126 20 0 7.00 0.31

M8 1755 AB 16 445 CEM I 42,5 R 109 40 2.56 10.80 0.32

M9 1748 AB 16 425 CEM I 42,5 R 89 60 2.56 12.00 0.32

M10 1441 AB 2 615 CEM I 42,5 R 153 55 3.59 11.08 0.32

M11 1813 AB 16 338 CEM III/B 186 0 0 0 0.55

M11 1522 AB 2 467 CEM III/B 257 0 0 0 0.55

SF: silica fume RE: retarder SP: plasticizer C: cement W: water

Table 2. Some properties of the concretes tested, mean of three tests

Properties of fresh concrete Compressive strength after 28 days

workability 1),

flow

density of

fresh concrete

air con-

tent

density of hard.

concrete 2)

Compressive strength

smallest value mean value

Mix

[cm] [kg/dm3] [%] [kg/dm3] [N/mm2] [N/mm2]

MR/1 41.8 2.34 1.8 2.35 52.3 53.8

MR/2 44.8 2.33 1.2 2.36 51.5 53.2

M1/1 46.5 - - 2.33 41.8 44.0

M1/2 46.5 2.35 1.5 2.33 45.4 46.2

M2 43.5 2.16 3.6 2.19 41.8 43.1

M3/1 44.5 2.39 0.9 2.37 42.5 43.6

M3/2 46.5 2.39 0.4 2.35 42.2 43.0

M4 46.5 2.40 0.3 2.38 47.5 49.0

M5 43.8 2.28 1.8 2.18 38.2 38.8

M6 43.0 2.40 1.75 2.39 75.2 77.2

M7 48.3 2.37 1.4 2.41 80.9 84.3

M8/1 42.0 2.37 1.5 2.40 88.0 90.5

M8/2 44.8 2.37 0.7 2.40 85.1 86.2

M9 44.0 2.38 1.6 2.38 85.4 88.9

M10 44.5 2.22 2.8 2.24 77.2 78.9

M11/1 47.5 2.36 0.55 2.35 40.9 43.0

M11/2 46.5 2.35 0.44 2.34 45.5 46.0

M12 51.0 2.36 1.4 2.19 33.8 35.6 1) workability: average diameter of the spread concrete determined by the German flow table test 2) determined on 100 mm cubes



The density is the dry density after 28 days. The compressive strength has

been measured on 100 mm cubes (150 mm for M3, due to the maximum aggre-

gate size of 32 mm) after 1 day kept in the mould, 6 days in the moist room and

21 days in the constant climate room at 20°C and 65% RH.

RESULTS

The results show typically the absorbed amount of liquid as function of

time up to 72 hours generated in the capillary suction test and in the infiltration

test. Fig. 2 to 4 show on the left the results of the capillary suction test and on

the right of the infiltration test. The test results can be presented as a straight line

of the absorbed amount vs. square root of time.

There are similar plots for the penetration depth which has been measured

by visual observation at the epoxy resin covered surface of the specimens. The

results are also shown in Fig. 5 to 7.

0

2

4

6

8

10

12

14

16

0 1 2 3 4 5 6 7 8

Square root of time [h1

9/2]

MR1-SDMR3-SDM1/1-SDM1/3-SDM2-SDM3/1-SDM4/2-SDM5-SDMR1-SCM1/1-SC

Absorbed volume [l/m2]

0

2

4

6

8

10

12

14

16

0 1 2 3 4 5 6 7 8


9/2]

MR1-ID

MR3-ID

M1/1-ID

M1/3-ID

M2-ID

M3/1-ID

M4/2-ID

M5-SD

Infiltrated volume [l/m2]

Fig. 2. Absorbed volume (left) and infiltrated volume (right) as function of square root of time: Portland cement concretes with normal strength

70


0

1

2

3

4

5

6

7

0 1 2 3 4 5 6 7 8


9/2]

M6-SD

M7-SD

M8/2-SD

M8/6-SD

M9-SD

M10-SD

M8/2-SC


0

1

2

3

4

5

6

7

0 1 2 3 4 5 6 7 8


9/2]

M6-ID

M7-SD

M8/2-ID

M8/6-ID

M9-ID

M10-ID


Fig. 3. Absorbed volume (left) and infiltrated volume (right) as function of square root of time: Portland cement concretes with high strength

0

2

4

6

8

10

12

14

16

18

0 1 2 3 4 5 6 7 8


9/2]

M11/1-SD

M11/3-SD

M12-SD


0

2

4

6

8

10

12

14

16

18

0 1 2 3 4 5 6 7 8


9/2]

M11/1-ID

M11/3-ID

M12-ID


Fig. 4. Absorbed volume (left) and infiltrated volume (right) as function of square root of time: Blast furnace slag cement concretes

The properties of n-decane are given in Table 3.

Table 3. Physical values of n-decane at 20°C

Fluid Formula Density Surface tension Dynamic vis-

cosity

Ratio

( / ) 0.5

[kg/dm3] [mN/m] [mN.s/m2] [m0.5/s0.5]

n-decane C10H22 0.73 23.9 0.88 5.21

With those values the results have been evaluated and are presented in Ta-

ble 4.



Table 4. Pore parameters calculated from test results with n-decane

Sorptivity

l m-2 h-1/2

penetration coefficient

mm h-1/2

r, from B

µm

cos θ =

r, from S

µm

cos θ =

Concrete

So Sp Bo Bp 1 2/π 1 2/π

MR 0.750 0.931 10.5 12.1 0.79 0.50 1.29 0.82

M1 1.054 1.342 12.2 14.7 1.10 0.70 1.48 0.94

M2 1.491 1.971 14.1 17.2 1.16 0.74 1.79 1.14

M3 1.001 1.111 12.5 13.7 0.47 0.30 0.55 0.35

M4 1.006 1.162 13.3 15.2 0.71 0.45 0.80 0.51

M5 1.378 1.567 13.7 15.2 0.58 0.37 0.70 0.45

M6 0.661 0.757 10.3 11.0 0.38 0.24 0.74 0.47

M7 0.552 0.626 9.2 10.0 0.44 0.28 0.68 0.43

M8 0.422 0.471 7.2 8.0 0.56 0.36 0.59 0.38

M9 0.378 0.474 7.4 9.0 1.09 0.69 1.37 0.87

M10 0.659 0.756 8.1 8.9 0.47 0.30 0.76 0.48

M11 1.169 1.366 13.5 15.2 0.63 0.40 0.87 0.55

M12 1.307 2.245 11.9 18.5 3.37 2.15 4.66 2.97

0

1020

30

4050

60

70

80

90

100

110

120

130

0 1 2 3 4 5 6 7 8 9

Square root of time [h1/2]

MR1-SDMR3-SDM1/1-SDM1/3-SDM2-SDM3/1-SDM4/2-SDM5-SDMR1-SC

Absorption depth [mm]

0

1020

30

4050

60

70

80

90

100

110

120

130

0 1 2 3 4 5 6 7 8 9


MR1-ID

MR3-ID

M1/1-ID

M1/3-ID

M2-ID

M4/2-ID

M5-ID

Infiltration depth [mm]

Fig. 5. Absorption depth (left) and infiltration depth (right): Portland cement con-cretes with normal strength

0

10

20

30

40

50

60

70

80

90

100

0 1 2 3 4 5 6 7 8 9


M6-SD

M7-ID

M8/2-SD

M8/6-SD

M9-SD

M10-SD


0

10

20

30

40

50

60

70

80

90

100

0 1 2 3 4 5 6 7 8 9


M6-ID

M7-ID

M8/2-ID

M8/6-ID

M9-ID


Fig. 6. Absorption depth (left) and infiltration depth (right): Portland cement con-cretes with high strength

72


0

20

40

60

80

100

120

0 1 2 3 4 5 6 7 8


9/2]

M11/1-SD

M11/3-SD

M11/1-SC


0

20

40

60

80

100

120

0 1 2 3 4 5 6 7 8 9


M11/1-ID

M11/3-ID

M12-ID


Fig. 7. Absorption depth (left) and infiltration depth (right): Blast furnace slag cement concretes

DISCUSSION

The sorption and infiltration tests show in Fig. 2 and 7 an almost perfect

straight line in the square root of time plot. A second general feature is that the

infiltration results are mostly close. In the sorption results, i.e. the pressure of 20

kPa is not important.

Concrete mixes M1 to M5 are made with a water-cement ratio of 0.55 but

with variations in the grading curve. Fig. 2 shows that suction proceeds the fast-

est with a maximum grain size of two millimetre and a high cement content of

467 kg/m3 (M2). The same is also true for the infiltration test. Also the mix with

fine grading C16 and a cement content of 465 kg/m3 is fast in suction but not so

fast in infiltration. The mixes M1, M3 and M4 vary less because the cement con-

tent is rather similar and also grading curves are similar. The concrete mix MR

shows the lowest suction and infiltration rates because the water-cement ratio is

only 0.50.

Fig. 3 contains the results of the high performance concrete with water-

cement ratios of 0.32 and typically a high cement content. Except M10 which

has a maximum grain size of 2 mm the others have all 16 mm maximum grain

size. There is however a variation in silica fume content. M6 and M10 have the

fastest absorption and infiltration. the reason for that is that there is either no sil-

ica fume used (M6) or the cement content is very high with 615 kg/m3 (M10).

One should notice that the vertical scale of Fig. 3 is less than half of Fig. 2. All

other high performance concrete mixes show smaller absorption and infiltration

qualities.



A blast furnace slag cement has been used in the mixes of Fig. 4. The sorp-

tive tests led to results which were similar to those with Portland cement and a

water-cement ratio of 0.55 (Fig. 2). The infiltration tests on M12 which has a

maximum grain size of 2 mm is different from the others since the infiltration

rate is rather high. A similar result has been obtained in Fig. 2 with Portland ce-

ment.

The absorption depth and the infiltration depth are rather similar as can be

seen from Figs. 5 to 7. The absolute results of MR, M1 to M5 and M11 and M12

are almost the same i. e. the influence of the grading curve is not so strong as in

the case of the absorbed fluid volume. However, a closer look to the small varia-

tions reveals that the trends of grain size and cement content are the same as in

the case of absorbed volume.

Fig. 6 shows the smallest absorption and infiltration depth as has been ex-

pected since these concretes are high performance ones.

Table 4 contains the values of the sorptivity and the penetration coefficient.

The sorptivity is the quotient of absorbed volume per area divided by the square

root of time. The penetration coefficient gives the penetration depth divided by

the square root of time. Both quantities characterise physical properties of a ma-

terial. Both material constants have been derived from sorption and infiltration

tests, So and Sp and Bo and Bp respectively.

The sorptivity So is in the range of 1.0 to 1.49 l m-2 h-1/2 for concrete with a

water-cement ratio of 0.55. The corresponding value Sp lies in the range of 1.11

to 1.97 l m-2 h-1/2. The difference between sorption test and infiltration test is

consistent. High performance concretes M6 to M10 show considerably lower

values So between 0.38 and 0.66 l m-2 h-1/2 and Sp between 0.47 and 0.76 l m-2

h-1/2 . The mixes with blast furnace slag cement fit into the ranges of mixes with

Portland cement except M12 in the infiltration test with a high value of 2.24 l

m-2 h-1/2.

The penetration coefficient Bo ranges between 12.2 and 14.1 mm h-1/2 for

mixes with a water-cement ratio of 0.55. Bp lies between 13.7 and 15.2 mm h-1/2,

i. e. a slight increase due to the pressure of 20 kPa. With a lower water-cement

ratio of 0.32 the Bo drops to 7.2 and 10.3 mm h-1/2 and Bp drops to 8.0 and 11.0

mm h-1/2. All results are consistent as the influence of grain size, water-cement

ratio and pressure are concerned. The B-values for blast furnace slag cement

concrete are similar to those of Portland cement concrete.

74


The effective pore radius r can be calculated from Bo and Bp as shown in

Eq. (4) or equivalently also from the sorptivities since penetration coefficient

and sorptivity are linked together via the porosity ε (see Eq. (5)). Since the po-

rosity levels out a similar equation occurs for the sorptivity as for the penetration

coefficient.

Table 4 contains the results. It can be seen that the pore sizes range be-

tween about 0.2 to more than 1.0 µm when the content angle is taken to zero.

The values decrease when the cosine of the contact angle is taken as 2/π [3]. The

absorbed values increase with the water-cement ratio. A deviation is obvious for

M12 with blast furnace slag cement and a high cement content.

The values of r calculated from the sorptivity are always larger than calcu-

lated from the penetration coefficient. This feature is certainly due to the fact

that the single size tube model is only a rough approximation of reality. It reality

the smallest pores have the greatest capillary suction form while the complete

filling of the pores are lacking behind. This means that the penetration coeffi-

cient should take into account smaller pores than the sorptivity does.

As the absolute values of r are concerned these are rather large compared to

pore sizes which are calculated from many intrusion experiments [8]. Obviously,

the pores which are reached by the organic fluid are the larger ones and the very

small pores are either filled by water of are inaccessible due to other reasons, for

instance due to the viscosity of the fluid or of the size of the molecule. This

could also mean that the model of the sharp wetting front is questionable. On the

other hand, it means that the selection of various fluids could give an impression

of the pore sizes which can be detected.

CONCLUSIONS

∗ The experiments with n-decane have proven the capillary suction law which

states that the absorbed volume and the penetration depth are a function of

the square root of time.

∗ The water-cement ratio is the main parameter governing the absorption

properties.

∗ The infiltration test with 20 kPa leads only to a minor increase of the pene-

tration and absorption.



∗ The pore sizes determined from the absorption test are larger than from the

penetration test indicating a different access to pores by different mecha-

nisms.

∗ Maximum aggregate size and various cement contents lead to different

physical properties.

REFERENCES

[1] Reinhardt, H. W. (ed.): Penetration and permeability of concrete: barriers

to organic and contaminating liquids. London: E&FN Spon, 1997

[2] Aufrecht, M.: Beton als sekundäre Dichtbarriere gegenüber umweltgefähr-

denden Flüssigkeiten - Technologie und Konzept für den Schadensfall,

Dissertation Universität Stuttgart, 1994

[3] Sosoro, M.: Modell zur Vorhersage des Eindringverhaltens von organi-

schen Flüssigkeiten in Beton, DAfStb, H. 446, Berlin 1995

[4] Brauer, N.: Analyse der Transportmechanismen für wassergefährdende

Flüssigkeiten in Beton zur Berechnung des Medientransports in ungerisse-

ne und gerissene Betondruckzonen, DAfStb, H. 524, Berlin 2002

[5] Paschmann, H., Grube, H., Thielen, G.: Untersuchungen zum Eindringen

von Flüssigkeiten in Beton sowie zur Verbesserung der Dichtheit des Be-

tons. DAfStb, H. 450, Berlin 1995

[6] DAfStb Guideline "Betonbau beim Umgang mit wassergefährdenden Stof-

fen", Part 4, Berlin 1996

[7] Pfingstner, A.: Determination of concrete pore structure parameters from

penetration tests with n-decane, Otto Graf Journal 10 (1999), pp. 113-127

[8] Reinhardt, H.-W., Gaber, K.: From pore size distribution to an equivalent

pore size of cement mortar. In: Materials & Structures 23 (1990), pp. 3-15

76

Analysis of crystalline materials contained in a palestine kohl vessel from the 4th century A.D.

ANALYSIS OF CRYSTALLINE MATERIALS PRESERVED IN A PAL-ESTINE KOHL VESSEL FROM THE 4TH CENTURY A.D.

UNTERSUCHUNGEN AM KRISTALLINEN INHALT EINES KA-JALGLASES AUS PALÄSTINA, 4. JH. A.D.

ANALYSE DU CONTENU CRISTALLIN D'UN RECIPIENT A KHOL DE PALESTINE DATANT DU 4ÈME SIÈCLE A.D.

Friedrich Grüner

SUMMARY

The crystalline content of a Late Roman glass vessel used to hold cosmetic

eye shadow (kohl) was analysed. The analytical techniques used were X-ray

powder diffraction and scanning electron microscopy. The materials detected are

described, indicating that they may have been used as kohl.

ZUSAMMENFASSUNG

Es wurde der kristalline Inhalt eines spätrömischen Doppelglasgefäßes aus

Palästina mit Röntgenpulverdiffraktometrie und am Rastelektronenmikroskop

untersucht. Der Gefäßinhalt wurde wahrscheinlich als Augenschminke (Kajal)

benutzt.

RESUME

Le contenu cristallin d'un récipient romain provenant de Palestine a été ana-

lysé au diffractomètre poudre aux rayons X et au microscope électronique à ba-

layage. Le contenu du récipient était probablement utilisé comme maquillage

pour les yeux (khôl).

KEYWORDS: kohl, glass vessel, galena, anglesite, cerussite, x – ray diffraction

1. INTRODUCTION

The following is a report on a study of the materials contained in a double –

tube flask from the collection of the “Württembergisches Landesmuseum” in

Stuttgart. A typical glass vessel for holding cosmetic eye – paints might have

one, two or four individual tubes.


F. GRÜNER

Studies of kohl previously reported in the literature have dealt with Egyp-

tian material /1/ and Late Roman to Byzantine material /2/. Galena (lead sulfide)

and the basic copper carbonate, malachite were widely used in Egypt for this

purpose. Both types were used in the Predynastic period, but the use of mala-

chite had stopped by the end of the New Kingdom. The use of galena continued

into the Coptic period. In Palestine glass vessels from the mid 4th to early 7th

century only galena was found in previous studies /2/.

The double – tube flask of the collection of the Württembergische Landes-

museum was made out of one long glass bleb, which had been divided into two

sections. Than both sections had been blown separately. The glass is light green

in colour and shows many bubbles. Both tubes contained a chunk of altered,

dark grey kohl (Fig. 1). One tube with a broken fragment shows part of a bronze

or copper rod, sticking in the altered kohl. The total height of the vessel is 9.9

cm, the diameter of each tube is approximately 1.7 cm.

Fig. 1: Double tube flask made of light green glass with 4 bails. One tube is broken and shows the preserved residue of the kohl and the corroded bronze rod.

78


Fig 2: Detailed photograph of the rod sticking in the kohl.

In Fig 2 some details of the chunk and the sticking rod are shown. The rod

is partly covered with green, blue – green and red coloured corrosion products.

The surface of the kohl is dark grey in colour and shows sometimes metallic

brightness.

2. EXPERIMENTAL PROCEDURES

For the detailed analyses at least one sample of the altered kohl was re-

moved from each tube. The samples were prepared for X – ray diffraction

(XRD), using a Siemens D 500 diffractometer. Scanning electron microscopy

with energy dispersive spectrometry (SEM/EDS) were used to identify the

chemical elements present. A Camscan scanning electron microscope including

a Noran Voyager energy dispersive x-ray analyzer was used for microscopic in-

vestigation.

3. ANALYTICAL RESULTS

The results of the analyses of the kohl vessels are presented below. Two

samples were removed from the surface of the solid chunk in both tubes. In both

samples the most common alteration products of galena, anglesite (PbSO4) and

cerussite (PbCO3) were present in major amounts. But galena (PbS) was also

observed in minor amounts (see Fig. 2). Both samples are nearly identical in

composition and could not be distinguished with x – ray diffraction.


F. GRÜNER

47-1734 (*) - Cerussite, syn - PbCO3 - Y: 10.30 % - d x by: 1. - WL: 1.5406 - Orthorho

05-0592 (I) - Galena, syn - PbS - Y: 14.79 % - d x by: 1. - WL: 1.5406 - Cubic - a 5.936

36-1461 (*) - Anglesite, syn - PbSO4 - Y: 75.98 % - d x by: 1. - WL: 1.5406 - Orthorho

83-1720 (C) - Anglesite - Pb(SO4) - Y: 69.31 % - d x by: 1. - WL: 1.5406 - Orthorhombi

SchminkeP2 - File: SchminkeP2.RAW - Type: 2Th/Th locked - Start: 5.000 ° - End: 70.

Sqr (Counts)

0

1

10

100

200

300

400

500

600

700

800

2-Theta - Scale

5 10 20 30 40 50 60 70

Fig. 2: XRD plot of the powdered kohl sample. Anglesite and cerussite occurred as common

alteration products, but galena is also present.

It is reasonable that finely ground galena for use as kohl would have

enough time to alterate into anglesite and cerrusite during ca. 1500 years of stor-

age under unknown archaeological conditions. The analyses of a small piece of

the rod, sticking inside the kohl showed cuprite and brochantite (see Fig. 3).

47-1734 (*) - Cerussite, syn - PbCO3 - Y: 0.61 % - d x by: 1. - WL: 1.5406 - Orthorhom

36-1461 (*) - Anglesite, syn - PbSO4 - Y: 8.38 % - d x by: 1. - WL: 1.5406 - Orthorhom

83-1720 (C) - Anglesite - Pb(SO4) - Y: 3.25 % - d x by: 1. - WL: 1.5406 - Orthorhombic

43-1458 (I) - Brochantite-M - Cu4SO4(OH)6 - Y: 2.72 % - d x by: 1. - WL: 1.5406 - Mon

75-1531 (C) - Cuprite - Cu2O - Y: 90.16 % - d x by: 1. - WL: 1.5406 - Cubic - a 4.26000

Schminke P1 - File: SchminkeP1.RAW - Type: 2Th/Th locked - Start: 5.000 ° - End: 70

Sqr (Counts)

0

10

100

1000

200

300

400

500

600

2000

3000

2-Theta - Scale

5 10 20 30 40 50 60 70

Fig. 3: XRD plot of a mixed sample with kohl (anglesite, cerussite) and alteration products of

the bronze rod (cuprite, brochantite).

80


The complete vessel was placed under the scanning electron microscope

for further investigations. The original surface of the kohl was studied in the

broken tube. Elemental analysis showed high concentrations of lead and sulphur

and some copper in the surrounding material of the rod, indicating a bronze al-

loy or copper metal. Other elements like Sb were absent, eliminating the use of

stibnite as possible component in the kohl material.

Fig. 4: Elemental analysis of a galena cube at the surface showing mainly Pb,

S is buried by the Pb peak.

Photomicrographs of the surface are shown in Fig. 5 and 6. The kohl con-

sists of a very fine grained groundmass with hypidiomorphic intergrown cubes

of galena.

Fig. 5: Photomicrograph of the fine grained groundmass with intergrown galena cubes.


F. GRÜNER

Fig. 6: Detailed photomicrograph of a galena cube.

4. CONCLUSIONS

In this study evidence was found only for galena as material used for the

production of kohl. Both flasks contained identical materials. For its use as an-

cient make up (eye shadow) it should be ground very fine. It is reasonable to

assume that most of the galena is altered to anglesite and cerussite during the

long period of storage under unknown archaeological conditions.

5. ACKNOWLEDGEMENT

The author wish to thank Mrs. Dr. Honroth at the Württembergisches Lan-

desmuseum for providing the sample material.

REFERENCES

[1] LUCAS, A., 1962: Ancient Egyptian Materials and Industries, 4th ed., revised

by J.R. Harris (Edward Arnold, London, 1962), pp 80-84

[2] BLANCHARD, W.D., STERN, E.M., STODULSKI, L.P., 1992: Analysis of Mate-

rials contained in Mid-4th to Early 7th Century A.D. Palestinian Kohl Tubes,

Mat. Res. Soc. Symp. Proc. Vol. 267, pp 239-254

82

Acoustic emission analysis of SFRC beams under cyclic bending loads

ACOUSTIC EMISSION ANALYSIS OF SFRC BEAMS UNDER CYCLIC BENDING LOADS

SCHALLEMISSIONSANALYSE AN STAHLFASERBETON UNTER ZYKLISCHEN BIEGEVERSUCHEN

ANALYSE DES ÉMISSIONS ACOUSTIQUES DE BETONS RENFORCES PAR FIBRES D'ACIER SOUS FLEXION CYCLIQUE

Florian Finck

SUMMARY

To further understand the failure processes within steel fibre reinforced

concrete members under cyclic load, a series of 3-point bending tests was

performed on notched beams using quantitative acoustic emission (AE)

measurements. AE measurements supplement the mechanical test data by

providing a large quantity of information about the progress of damage in terms

of time, location and cause. Quantitative analysis of acoustic signals consists of

an accurate localization of the fracturing and under certain assumptions an

inversion for the moment tensor can be performed to gain information about the

total energy released and the orientation of the rupture plane. After

decomposition of the moment tensor, the type of rupture process can be

quantified and visualized using ostensive crack models like those for shear and

opening. In this article some first results of the fatigue test series and the

analysis of the AE-data are presented.

ZUSAMMENFASSUNG

Zur Untersuchung von Schädigungsprozessen innerhalb

stahlfaserbewehrter Betonbauteile unter zyklischer Last wurde eine Reihe von

3-Punkt-Biegeversuchen durchgeführt und die auftretenden Schallereignisse

aufgezeichnet. Neben den mechanischen Prüfdaten können so Informationen

über den Zeitpunkt und den genauen Ort der fortlaufenden Schädigung

gewonnen werden. Darüber hinaus kann unter bestimmten Voraussetzungen

eine Inversion auf den Momententensor durchgeführt werden, welcher

bruchmechanische Parameter, wie z. B. die Bruchenergie und die Orientierung

der Bruchflächen enthält. Nach einer geeigneten Zerlegung des Tensors können

die enthaltenen Bruchmoden quantifiziert und durch anschaulich Bruchmodelle,


F. FINCK

wie die des Öffnungs- oder des Scherbruches beschrieben werden. In diesem

Artikel werden erste Ergebnisse der Ermüdungsversuche und der

Schallemissionsanalyse vorgestellt.

RESUME

Afin d'analyser les processus de détérioration à l'intérieur d'éléments en

béton armé de fibres d'acier sous chargement cyclique, nous avons réalisé une

série d'essais de flexion 3 points et enregistré les émissions acoustiques. Ainsi

nous avons pu gagner, outre les données mécaniques, des informations sur la

nature, le moment et le lieu exacts des émissions acoustiques, et, par là, sur la

progression de la rupture. Dans certaines conditions, le tenseur des moments

peut être calculé par inversion. Celui-ci contient des informations sur l'énergie

libérée et l'orientation de la surface de rupture. Après la décomposition du

tenseur de moment, les modes de rupture peuvent être quantifiés et visualisés à

l'aide de modèles simples, comme ceux pour le cisaillement et l'ouverture. Dans

cet article les premiers résultats des essais de fatigue et de l'analyse des

émissions acoustiques sont présentés.

KEYWORDS: fatigue test, steel fibre reinforced concrete, acoustic emission,

moment tensor

INTRODUCTION

Steel fibre reinforced concrete (SFRC) has been in use since the late 60s,

mainly as shotcrete for underground constructions and flooring. Some

advantages of SFRC are a minimization of crack widths and permeability or an

increased toughness. Although various basic works on the behaviour of SFRC

members have been published [e.g. WEILER 2000], there remain open questions

about the mechanical laws and processes that exist during failure. The

interaction between steel fibre reinforcement and a cementitious matrix, as well

as the characterization of failure of SFRC members, are mayor topics of the

subproject A6 in the collaborative research centre SFB 381.

In a fatigue test series with steel fibre reinforced concrete (SFRC) beams

under cyclic 3-point bending load we studied the behaviour of ongoing failure.

Thereby, the investigation of acoustic emissions under changing conditions

(e. g. load, amplitude and frequency) was the main focus, not an accurate

statistical investigation of the members. The external, i. e. visible, fatigue is

84


given by mechanical test data containing the deflection in dependence on load

and the number of load cycles. Additionally, acoustic emission analysis yields

information about the internal processes of failure, which correspond to the

emission of seismic energy due to cracking. Each single crack (event) is

localized and for a selection of events moment tensors are evaluated. A suitable

decomposition of this tensor [JOST & HERMANN 1989] yields parameters such as

the energy released during rupture, the orientation and the size of the rupture

plane and a combination of ostensive fracture modes. From these parameters the

stress regime in the member and the mechanics of failure can be derived.

SETUP OF THE FATIGUE TEST SERIES

For the test series beams with dimensions 15 cm X 15 cm X 70 cm with a

1.5 Vol.% reinforcement of Dramix® RC 80/60 BN steel fibres (length: 60 mm,

diameter: 0.75 mm) were used. On the bottom surface in the middle of the beam

a notch with a depth of approximately 3.3 cm caused a well-defined start of a

crack. The transmission of force by the servo hydraulic 100 kN test frame was

realized using three steel cylinders. The two fixed lower supports had a distance

of 60 cm and the upper support at the centre was free to rotate around the

longitudinal axis of the beam to avoid torsional stress. Figure 1 shows details of

the test setup, with the AE sensors attached to the specimen.

Figure 1: Sketch of a notched SFRC beam under a cyclic 3-point bending load. AE sensors are mounted around the area of damage.


F. FINCK

Piston displacement, load and crack opening were recorded over time. The

acoustic emissions were recorded by an 8-channel transient recorder with a

sample rate of 2.5 MHz per channel and an amplitude resolution of 12 Bit. Eight

piezo-electric accelerometers were evenly distributed on the surfaces of the

beam.

First, the range of the failure load was evaluated in two static bending tests.

Although, a large variation of this value has to be expected due to a change in

the distribution of fibres and the composition of concrete in the beam, this value

provides an estimate of the load profile for the dynamic tests. During the

dynamic tests, load, amplitude and frequency were adapted to the progress of

damage and the AE activity.

PRESENTATION OF THE RESULTS

In the following section the results of one test run are presented. A load of

7.5 kN was applied statically before the load cycles began. In the second plot in

figure 2 the different phases with changing load, amplitude and frequency

during the fatigue test are labelled and coded by blue (dark grey) and green

(light grey) respectively to be identified in the load over crack opening

(displacement) plot and the crack opening over time or cycles plot. The

frequency of the sinusoidal load cycles was 1 Hz from phase G. From that point

the number of load cycles equals time in seconds plus 5000. On the bottom a

histogram of the acoustic emission activity can be correlated with the ongoing

failure in the beam.

With each increase of the maximum load the crack opening, as well as the

AE activity, increases rapidly but a relaxation is visible with the continuation of

the test. The extending areas of the hysteretic ellipses in the load over crack

opening plots are another indicator for the damage progress.

86


Figure 2: Mechanical test data of one fatigue test. From top: load deflection curve,

load over time profile, crack opening over time and the AE activity.


F. FINCK

During the test a total of 385 acoustic emissions were recorded from which

377 could be localized. The data quality was very good regarding noise due to

the cyclic bending and the accuracy of the localization lies in a range of about

1 cm.

Figure 3 shows the located events from three prospectives: from above, a

front view and a side view. The markers representing the sources of the acoustic

emissions are given in the legend, corresponding to the test periods starting with

the according labels (see also figure 2, 2nd plot). This illustrates the temporal

growth of the damaged zone.

Figure 3: Projection of the localization of acoustic emissions. The different markers correspond to various test periods, as indicated in the legend.

88


Nearly all acoustic emissions lie in the central region of the specimen in the

vicinity of the main crack. Due to the steel fibres, some smearing of the damage

zone takes place. The early events come from the lower half of the specimen

since the crack starts in the edge of the notch due to tension. Then steel fibres

are activated and accommodate load as they are pulled out. The crack grows

towards the top of the specimen under a relative constant spatial AE activity

from the complete region under fatigue. The width of the damage zone in y-

direction is more or less in the range of the fibre length (i. e. 60 mm). This

suggests that always the short end of the fibre is being pulled out, as expected.

THE INVERSION OF MOMENT TENSORS

To gather more information about the mechanical reasons of failure, we

calculate moment tensors with a relative moment tensor inversion (RMTI)

technique developed by DAHM 1993. The application and some theory of this

method on acoustic emission data has been described previously [e. g. FINCK

2002, FINCK 2001, GROSSE 1999]. An advantage of the RMTI is the elimination

of the Green’s functions [AKI & RICHARDS 1980] of the medium by an inversion

for a cluster of events. Two circles in figure 3 indicate the orientation of two

clusters of 16 events each which were inverted for their moment tensors. C1 is a

cluster from very early events in the tension zone, events in C2 originate from

an advanced stage of the test.

Figure 4: Radiation patterns of seismic energy and results from the moment tensor inversion for selected events from cluster C1 an C2 (see figure 3). Mr is the relative seismic moment,

ISO is the isotropic component of the event and DC is the double-couple portion of the deviatoric component.


F. FINCK

A combination of two different crack modes is expected for the performed

test. First, the opening of the main crack should radiate energy similar to event

EV 29 in cluster C1. An opening mainly perpendicular to the vertical crack-

surface with particle motion outwards parallel to the y-axis and a significant

remaining isotropic component. This conforms to mode 1. Second, a great

number of events should correspond to the pull-out of steel fibres. Mainly a

double couple mechanism for shear failure is expected in this case, with a small

isotropic component only (conforming mode 2 or 3).

A selection of the results is shown in figure 4. The first row contains results

for cluster C1, the second for C2. Under the top view projection of the radiation

patterns of seismic energy the relative seismic moment Mr, the isotropic

component ISO and the double-couple portion DC of the deviatoric component

are given with errors. The best results from a boot strap analysis [EFRON &

TIBSHIRANI, 1986] can be found in the brackets. The majority of the moment

tensors consist of a very small positive isotropic component. For event EV 29

the errors are very high, so the results must be doubted, though the radiation

pattern fits to first expectations. For the other events in C1 the DC component is

rather small. The deviatoric components of these events can not be explained by

one pure shear crack. Other deviatoric phenomena seem to take place. But the

early events in C1 vary from the results for C2. The events occurred at an

advanced stage of the test, where the fibre-pull out seems to be the major reason

for acoustic activity. Here, the DC component is large.

The results have a great stability for a changing composition of the

investigated clusters. In earlier investigations the results for single events were

dependent on the composure of the cluster, meaning that the existence or non-

existence of other events had an influence on the results. Also the errors are

small.

CONCLUSIONS

We successfully obtained high quality acoustic emission data from cyclic

bending tests of steel fibre reinforced concrete beams. The majority of these

events could be localized and an inversion for the moment tensor of a selection

of events was performed. Stable results from the moment tensor inversion can

partially be correlated with the expected mechanisms of failure – an opening of

the crack (mode 1) and mainly the pull-out of fibres (mode 2 or 3). Acoustic

90


emission analysis helps understanding complex mechanisms of failure even over

a large period of time.

The decomposition of the moment tensor into crack modes known from

geological investigations seem not to be suitable for experimental data from the

laboratory. A decomposition taking crack modes from engineering models in to

account, is needed. This subject will be of intensive interest in future

ACKNOWLEDGEMENTS

These investigations are part of our work in the collaborative research

centre SFB 381 at the University of Stuttgart which is financially supported by

the Deutsche Forschungsgemeinschaft (DFG). We gratefully acknowledge this

support. The author would also like to thank Lindsay Linzer, Rock Engineering

Dept., CSIR Miningtek for providing the radiation pattern generator.

REFERENCES

AKI, K., RICHARDS, P.G.: Quantitative Seismology; Volume 1. Freeman and

Company, New York, 1980.

DAHM, T.: Relativmethoden zur Bestimmung der Abstrahlcharakteristik von

seismischen Quellen. Dissertation, Universität Karlsruhe, 1993.

EFRON, B. TIBSHIRANI, R.: Bootstrap methods for standard errors, confidence

intervals and other measures of statistical accuracy. Statistical Science 1,

pp.54-77, 1986.

FINCK, F.: Application of the moment tensor inversion in material testing. Otto-

Graf-Journal, Vol. 12, pp. 145-156, 2001.

FINCK, F., MOTZ, M., GROSSE, C.U., REINHARDT, H.-W., KRÖPLIN, B.:

Integrated Interpretation and Visualization of a Pull-Out Test using Finite

Element Modelling and Quantitative Acoustic Emission Analysis. Online

publication: http://www.ndt.net/article/v07n09/09/09.htm, 2002.

GROSSE, C.U.: Grundlagen der Inversion des Momententensors zur Analyse von

Schallemissionsquellen. Werkstoffe und Werkstoffprüfung im Bauwesen.

Festschrift zum 60. Geburtstag von Prof. Dr.-Ing. H.-W. Reinhardt, Libri

BOD, Hamburg, pp. 82-105, 1999.


http://www.ndt.net/article/v07n09/09/09.htm

F. FINCK

JOST, M.L., HERMANN, R.B.: A students guide to and review of moment tensors.

Seism. Res. Letters, Vol. 60, pp. 37-57, 1989.

WEILER, B.: Zerstörungsfreie Untersuchung von Stahlfaserbeton. Dissertation an

der Universität Stuttgart, Shaker Verlag, 2000.

92

About the Improvement of US measurement techniques

ABOUT THE IMPROVEMENT OF US MEASUREMENT TECHNIQUES FOR THE QUALITY CONTROL OF FRESH CONCRETE

GERÄTETECHNISCHE FORTSCHRITTE BEI DER QUALITÄTS-SICHERUNG VON FRISCHBETON MIT ULTRASCHALL

AMÉLIORATION DES TECHNIQUES DE MESURE ULTRASONIQUES POUR LE CONTRÔLE DE QUALITÉ DU BÉTON FRAIS.

Christian U. Grosse

ABSTRACT

Over the last decade a testing method based on ultrasound was developed

at the Institute of Construction Materials of the University of Stuttgart to control

the hardening process of cementitious materials by means of non-destructive

testing. This paper describes the systematic improvement and re-design of the

testing system and the investigation methods.

ÜBERSICHT

Am Institut für Werkstoffe im Bauwesen der Universität Stuttgart wurde in

den letzten zehn Jahren ein Ultraschallverfahren zur die Analyse des Erstarrens

und Erhärtens von zementgebundenen Materialien entwickelt. Der Artikel

beschreibt die fortdauernde Verbesserung der Messtechnik im Hinblick auf die

Qualitätskontrolle von Frischbeton und –mörtel.

RESUME

A l'université de Stuttgart, un procédé ultrasonique de contrôle de la prise

et du durcissement des matériaux cimentaires a été développé au courant des dix

dernières années. L'article présent décrit l'amélioration continue des dispositifs

et de la procédure de mesurage en ce qui concerne le contrôle de la qualité des

béton et mortiers frais.

KEYWORDS: Fresh concrete, non-destructive testing, ultrasound


C. U. GROSSE

INTRODUCTION

Nowadays the characterization of cement-based materials during the

stiffening process by ultrasound measurement techniques is well established.

This paper deals with the ultrasound technique used in through transmission. In

numerous publications [e. g. GROSSE & REINHARDT 1994, GROSSE ET AL. 1999,

REINHARDT ET AL. 1999a] the patented test method [REINHARDT ET AL. 1999b]

developed at the University of Stuttgart was described earlier. Methods based on

ultrasound are better suited for the characterization of the setting and hardening

of cement based materials than traditional test methods like the Vicat-needle-

test, the penetrometer test or the flow test, because the travel time, the

attenuation and the frequency content of ultrasound waves sent through the

material are closely correlated with the elastic properties of concrete or mortar.

These parameters can be continuously monitored during the stiffening giving a

comprehensive picture instead of snapshots of workability for example.

A sophisticated device was developed and numerous experiments have

been conducted in the past, investigating the influence of water-to-cement ratio,

the type of cement, the use of additives and admixtures, the air bubble content

and so far, for the setting and hardening of concrete or mortar. Newer features

are the extraction of the initial and final setting time out of the signals [GROSSE

& REINHARDT 2000] and the parallel registration of the state of hydration.

However, the earlier described device lacks of handiness and several features,

which could improve the art of such measurements further.

EVOLUTION AND SURVEY OF DEVICES EXISTING AT THE UNIVERSITY OF STUTTGART

The first measurements to control the setting and hardening of concrete at

the University of Stuttgart using ultrasound are dated back to the early 1990’s.

These experiments have been conducted in the frame of a research project

sponsored by the German Reinforced Concrete Committee (DAfStb, V 345) and

are published in the 1994th volume of the Otto-Graf-Journal [GROSSE &

REINHARDT 1994]. The tests were carried out with a rough set-up using a

container made of 40 mm thick styrene foam (Styropor) plates and the dimensions

300 mm × 300 mm × 80 mm (Fig. 1). The emitter was a simple steel ball impactor

dropping a ball of 4 mm diameter on to a small aluminium plate, which was placed

in contact with the fresh concrete.

94


Fig. 1: Set-up with steel ball impactor and receiver. Dimensions of the container: 300 mm x 300 mm x 80 mm

Later on the ideas were proofed by numerous students during their Diploma

thesis, technician and student research assistants. Jochen FISCHER [1994], Bernd

Weiler and the author [Grosse 1996] developed a device using three long styrene

foam walls and two smaller rigid side walls out of aluminium plates and the same

simple impactor as used earlier (Fig. 2).

Fig. 2: Set-up of the smaller styrene foam container with two aluminium side walls. Dimensions of the container: 200 mm x 80 mm x 60 mm.


C. U. GROSSE

While the first set-up caused problems to determine the correct travel

distance of the pulse to the receiver, what is essential for velocity measurements,

the second set-up was unsatisfactory as well, because of interfering waves

resulting from the walls.

A re-design of the device described in REINHARDT ET AL. [1996], WINDISCH

[1996], HERB [1996] and REINHARDT ET AL. [1998] for concrete measurements

was patented later [Reinhardt et al. 1999] and consisted of a mould completely

out of PMMA of the dimensions 160 mm × 200 mm × 70 mm (Fig. 3), but the

handling of this device was poor and the leakiness of the container caused a

penetration of fluids especially during the compaction process. However, the

device was modified by BEUTEL [1999] and tested to be suitable for field

measurements.

Fig. 3: Set-up of the first container out of PMMA only. Dimensions of the container: 160 mm x 200 mm x 70 mm.

In the meantime the development of a test set-up adjusted to mortar

materials run parallel. Due to smaller grain sizes (usually less 2 mm) the

dimensions of a mortar device can significantly be reduced. Not all steps of the

development can be described in detail. Figure 4 gives an impression of the

iterative process of finding a suitable shape for the mould. The final container

[GROSSE ET AL. 1999] had two walls of PMMA and a U-shaped rubber foam

with an inner volume of 40 cm³ for the mortar (Fig. 5).

96


Fig. 4: Evolution of containers, tested for mortar applications.

There are two main advantages in respect to the concrete set-up. The

amount of material necessary to be tested is significantly reduced and so is the

amount of waste. Secondly, the pulse is not excited by an impactor, what is an

advantage in terms of reliability and handiness.

Fig. 5: Final set-up of the mortar device showing the mould (rubber foam and PMMA-walls) and the transducers.


C. U. GROSSE

Consequently, a new device, illustrated in Fig. 6, was developed by

STEGMAIER [2000] and Herb, whereby the dimensions were changed to 400 mm

× 59 mm ×130 mm in accordance to the smaller mortar device. Similar to

former concrete devices the wave is generated using a steel ball exciter, referred

to as Ultrasound Impactor (USIP), hitting a small plate fixed on the PMMA

casing. The resulting excitation can be seen as broad banded, having a relatively

wide frequency bandwidth of up to 100 Hz.

Fig.6: FreshCon device for concrete measurements developed on the basis of the older mortar device (see Fig.5).

Though many difficulties were eliminated the system still shows up

unresolved problems. Specifically, a wave travelling through the container wall

which onset is detected before the irradiating primary wave can be observed.

Further on, the energy evolution during the hardening of concrete is still difficult

to analyze since the steel ball transmitter USIP, as a mechanical system,

provides unreliable energy data and the plate where the steel ball is shot on

easily disbond so that the coupling of the excited energy into the PMMA

container changes during tests. These factors influence the obtained results and

the reproducibility of tests.

98


To summarize the pros and cons of the concrete device the following

statements can be given:

• Less reproducibility of impact energy results in energy determination uncertainties.

• Contact problems of steel plate at PMMA container (delaminations).

• Unreliable generation of impacts due to steel balls sticking in the impactor rod.

• Possible side wall waves disturbing the measurement at early ages during investigations of very “slow” materials.

• Pressure air equipment necessary for the impactor.

It should be stated, that at the end of 2000 no possibility to record the

hydration temperature in the same sample during the ultrasound measurements

as a secondary control technique was available using the existing FreshCon

software.

ANALYSIS METHODS

Using ultrasound methods the degree of hardening is characterized by the

change of significant parameters. Not only the travel time of the ultrasonic pulse

through the testing device, consequently the velocity of compressional waves

but also the frequency content and the relative energy are recorded.

On the basis of suitable parameters, e.g. the frequency content of the signal

over the time, additionally a wavelet transformation (WT) is carried out in order

to gather as much information as possible from the raw signal to evaluate

concrete and mortar, respectively. The program AutoCWT, able to apply the

WT was implemented by MANOCCHIO [2001], where the calculation kernel is

taken from the program IWB-CWT, coded by BAHR [2001a]. More information

about the application of wavelets in the characterization of the setting and

hardening of cementitious materials can be obtained from Grosse [2001],

GROSSE & REINHARDT [2001] or MANOCCHIO [2001].

Further on as a new feature of the FreshCon system the ability to record the

temperature evolution over the time is introduced as well as the determination of

the associated hydration heat, following DIN-EN 196 part 9.


C. U. GROSSE

Fig. 7: Set-up (top) for measurements of elastic parameters (velocity, energy, frequency) as well as the temperatures. Bottom: Screenshot of the new program version 2.04 of FreshCon.

100


MEASUREMENTS OF HYDRATION TEMPERATURES

The program FreshCon was extended to enable temperature measurements

using the multi-channel National instrument computer board NI 4351. A

screenshot of this program version is represented in Fig. 7, where also a picture

of the test setup is given. The temperature distribution occurring during the

hydration process is a characteristic for the state of hardening of cement-based

materials. Therefore, statements can be deduced according the relations between

two different materials. It should be mentioned that the hydration process in the

semi adiabatic container in comparison to the testing device for ultrasound

measurements is faster due to the accumulation of heat in the temperature

container. Consequently, the sound velocity and temperature distribution cannot

be correlated directly. A picture of the testing device for the determination of the

heat of hydration, taken from KÖBLE [1999] is given in Fig. 8.

Stützvorrichtung aus Polystyrol

Gehäuse aus Holz

Polystyrolscheibe

Mörteldose, h = 12cm

Dichtung

Polystyrol, d = 5cm

Zellgummidichtung, d = 2cm

Holzdeckel

Dewar-Gefäß

Plexiglasscheibe zum fixieren des Temperaturfühlers

16cm

Thermokabel geführt in einem Plexiglasröhrchen

Dichtung

181

292

232

328

Luft

cable to digital thermometer

wooden lid

wooden box

rubber foam, d = 20 mm

seal

seal

polystyrene, d = 50 mm

polystyrene

plate of lucite to fix thermocouple

can filled with mortar

Dewar container Ø 160 mm

polystyrene to hold Dewar container in upright position

air

Fig. 8: Set-up of the calorimeter device according KÖBLE [1999], dimensions in mm.

Regarding the determination of the heat of hydration DIN EN 196 - 9 is

followed, accordingly. The aim of the semi-adiabatic method, namely the

Langavant - method, applicable to mortar, is the determination of the released

amount of heat during the hydration process. For this purpose the online version

of the program FreshCon, implemented by BAHR [2001b], was modified. The

system is now able to record the temperature in the calorimeter (Fig. 8), the

temperature in the tested material (Fig. 7, top) and the air temperature. All these

data are obtained automatically and stored together with the data of the

ultrasound measurements. A typical result is represented in Fig. 9, showing all


C. U. GROSSE

three temperatures as a function of the concrete age. The temperature effect is

dominant at the curve obtained using the calorimeter (straight line) due to the

semi-adiabatic conditions in the Dewar container. Testing concrete materials a

hydration effect is clearly seen at the temperature data obtained in the ultrasound

container (dotted line) compared to the air temperature (dashed line).

0 200 400 600 800 1000 1200 1400

0

5

10

15

20

25

30

35

40

45

Rilem Round Robin TestsiBMB: 18/19 Apr 2002

Mixture: RB03 (concrete)

Remarks: very dry, water added

Hyd

ratio

n H

ea

t [°

C]

Concrete

Air

Air

an

d C

on

cre

te te

mp

era

ture

[°C

]

Time [min]

0

5

10

15

20

25

30

35

40

45

Hydration

Fig. 9: Results of temperature measurements using the new FreshCon version.

DEVELOPMENT OF A NEW CONCRETE FRESHCON DEVICE

Comparing the two devices for mortar and concrete measurements, the

advantages of the mortar set-up should be summarized:

• Good reproducibility of signal generation (energy).

• Easy onset time determination due to signals with good reproducibility.

• No pressure air equipment necessary.

• Full automatic measurement and storing of waveforms.

• Automatic determination of velocity and energy as well as additional parameters.

• Full control of measurement parameters.

To enhance the handling of concrete experiments accordingly, a new

design of the device shown in Fig. 6 was suggested. For the new device the

102


impactor was replaced by an US transmitter in combination with a wideband

power amplifier and a function generator. No pressure air is need for this device;

a control sensor next to the impactor recording the emitting pulse is no longer

required.

0 120 240 360 480 600 7200

500

1000

1500

2000

velocity impactor

velocity piezo

rel. energy impactor

rel. energy piezo

age [min]velo

city [m

/s]

1E-6

1E-5

1E-4

1E-3

0.01

0.1

1

Fig. 10: Comparison experiments between impactor and US emitter.

rel. e

nerg

y [-]

In several preliminary experiments the new set-up was tested by

MANOCCHIO [2001] to compare the results of measurements by the impact

generated signals and by piezo-electric emitters in parallel (Fig. 10). The two

curves at the bottom of the right side in Fig. 10, recorded at the same time using

the same material, represent the velocity evaluation of gypsum. Gypsum was

used as a test material due to its fast hydration evolution. The two curves at the

top position in Fig. 10 represent the relative energy. A decrease of the velocity

and energies values is caused by shrinkage effects. Both curve pairs look very

similar in respect to differently used pulse generation methods.

This successful first test triggered the re-design of the concrete device (as

well as of the mortar device). A flow chart of the new experimental set-up is

given in Fig. 11. The electronic pulse is generated by a frequency generator and

amplified by a power amplifier. Broadband piezo-electric transducers generate

the ultrasound signal to be transmitted through the material. A transducer of the

same type is used as a receiver and the signal is passed through a pre-amplifier

to the PC-board A/D-converter, denoted as “computer-based signal processing”

in Fig. 11. Special attention is given to the correct trigger time of the signal,

what is essential for velocity measurements. A power amplifier of the companies

KROHN-HITE CO. or DEVELOGIC GMBH is used along with sensors of the

company VALLEN INSTRUMENTS.


C. U. GROSSE

Fig. 11: Flow chart of newly developed FreshCon experiments.

Testing device (concrete or mortar)

The new container/sensor design for concrete as well as for mortar

experiments is shown in Fig. 12, demonstrating the similarity of these two. The

U-shaped rubber in the middle of the container is essential. Regarding the

concrete device, a special “long wall” container was produced for very “slow”

materials to avoid waves propagating along the walls to be faster than the direct

waves. The distance of the screw joints can be adjusted to the material

properties.

Fig. 12: Re-designed FreshCon container/sensor for mortar (left) and concrete (right) measurements.

First experiments in the frame of a master thesis [KALCKBRENNER 2002]

and during round robin test of a RILEM technical committee showed very

promising results.

104


ROUND ROBIN TESTS – STATUS

The International Union of Testing and Research Laboratories for Materials

and Structures, RILEM, as a non profit-making, non-governmental technical

association is structured in groups of international experts, the so called

Technical Committees (TC). In the framework of advanced testing of cement-

based materials during setting and hardening the TC 185 - ATC organized a

round robin test series. The purpose of these tests is to assess the capability of

existing test methods based on non-destructive techniques in terms of suitability,

sensitivity and accuracy. Results will be summarized in a state of the art report

and a test recommendation is planned to be released. In the context of providing

a direct comparability, experiments are carried out by different members at the

same place using the same charge of materials/mixtures.

The technical realization of the experiments is in the responsibility of the

TC secretary (C. Grosse) and the local organizers. The ongoing test series

started in 2001 with experiments in Vaulx-en-Velin (France) and was continued

in Evanston/Chicago (USA) in spring 2002 and Brunswick (Germany) in

summer 2002. The next round robin test is scheduled for spring 2003 in Delft

(The Netherlands). In detail the following groups have been involved so far:

• Ecole Nationale des Travaux Publics de l’Etat (ENTPE), Vaulx-en-Velin,

France; Dr. L. Arnaud and Prof. C. Boutin.

• Center for Advanced Cement-Based Materials (ACBM) at Northwestern

University, Illinois, USA; Prof. S. Shah and Dipl.-Ing. T. Voigt.

• Institute of Structural Materials, Solid Structures and Fire Protection

(iBMB) of the Technical University of Brunswick, Germany; Prof. H.

Budelmann, Dipl.-Math. M. Krauß.

• Fraunhofer Institute for Non-Destructive Testing (IZFP) in Saarbrücken,

Germany; Dr. G. Dobmann and Dr. B. Wolter.

• Institute of Construction Materials (IWB) at the University of Stuttgart,

Germany; Prof. H.-W. Reinhardt, Dr. C. Grosse and Dipl.-Ing. A. Kalck-

brenner (M.Sc.).

An experimental test program was compiled to be the basis for all

experiments [GROSSE & REINHARDT 2002]. Six different mixtures are

recommended to be tested – five other mixtures are tested additionally. Some of

the results obtained by the Institute of Construction Materials (IWB) at the


C. U. GROSSE

University of Stuttgart are published by KALCKBRENNER [2002] and correlated

to the results of other groups. A comprehensive report will follow.

To give an example of the data obtained during one test series Fig. 13

demonstrate the variation of the velocities over the age of the material.

Concerning these velocities an S-shaped curve is typical for cementitious

materials. After a certain time at the beginning, while the velocity variation is

small, the gradient is increasing significantly. Regarding the data RE5 from a

mix with added retarder this increase occurs relatively late. To make the basic

statements more evident the curves are smoothed and bad data points are

removed. It is obvious that concrete mixes are “faster” than mortar mixes in

respect to hardening, while the RE5 mix with retarder is the “slowest“ material.

Fig. 13: Comparison of the velocity measurements testing mixtures RE 1-6.

0 200 400 600 800 1000 1200 1400

0

1000

2000

3000

4000

5000

Ve

locity [m

/s]

Age [min]

RE1: concrete, w.c. 0.45

RE2: concrete, w.c. 0.60

RE3: mortar, w.c. 0.60

RE4: mortar, with plasticizer

RE5: mortar, with retarder

RE6: mortar, with air entrainer

smoothed!

It should be stressed that only material properties related to the elastic behavior

can be analyzed with ultrasound techniques. As far as the chemical properties

are not related to the elastic properties, other measurement techniques have to be

used in combination with ultrasound to get more data. The results of the round

robin tests should indicate the value of the described ultrasound through-

transmission technique in comparison to other techniques like ultrasound

reflection, nuclear magnetic resonance, electric and maturity methods.

106


SUMMARY AND OUTLOOK

The measuring device developed at the University of Stuttgart is able to

analyze the setting and hardening of cementitious materials in a comprehensive

way. The method is based on ultrasound and can be used for numerous

applications, where reliable and reproducible data are required, what addresses

material parameters like the water-to-cement-ratio, the type of cement or the

effect of additives as retarders or accelerators. At the concreting site, where

efficiency and a low budget are boundary conditions, the application of this new

technique can help to enhance the stability during construction or the progress of

the construction work saving both: time and money. Some examples are the

development of admixtures, the in-situ quality control, the slip form concreting

or the precasting. Certainly, the applications are not restricted to cementitious

materials.

Further improvements are concerning the velocity evaluation. Since the

device consist of an analogue-to-digital converter of 5 MHz only, the resolution

of the velocity calculations varies over age. Actually, the resolution decreases

with increasing velocities. This is the reason of the so-called bit-pattern

occurring usually at ages of 400 minutes and later. To ease the interpretation the

velocity curves are smoothed using adjacent averaging (10 points), but it is

suggested to plot the original data points into the smoothed curves as well.

Using the offline version of the FreshCon picking algorithm the data can be re-

evaluated after the test concerning the onset times of the signals only.

Surprisingly, curves re-picked by the operator are usually very similar to the

automatically processed data so that a time consuming manually picking is not

improving the results anymore.

Formerly, the comparison of energy evaluation results was sophisticated

due to the application of two different devices. Energy values as measured by

the FreshCon software are basing on the squared amplitudes of the signal

beginning at the signals onset of compressional waves. These values strongly

depend on the energy released by the impact to the container. The

reproducibility of the transmitter energy is low of impactor devices compared to

devices using an ultrasound emitter. Changing the set-up as described made the

interpretations regarding energies more reliable. There is still the disadvantage

of energies emitted by piezo-driven devices to be of several magnitudes lower

than impactor pulses. A new impactor device without pressure-air giving broad-

band pulses of reproducible magnitude is under development.


C. U. GROSSE

Talking about the scientific aspects of the ultrasound technique, the method

developed at the University of Stuttgart is under further progress. This is

especially true concerning wavelet algorithms. The degree of automatization is

enhanced and additional analysis techniques will be implemented in future.

With regard to the international activities of the RILEM technical

committee more information can be obtained from the author or at the TC’s

homepage: http://www.rilem.org/atc.html. Colleagues working in this scientific

field are offered to collaborate in this initiative.

ACKNOWLEDGEMENTS

The described design and re-design of ultrasound devices are the result of

many years of scientific work. It is difficult to address the thanks to everybody

who was involved. However, some colleagues should be mentioned in no

particular order: Dr. B. Weiler, Dipl.-Ing. J. Fischer, Dipl.-Ing. I. Kolb, Dipl.-

Ing. N. Windisch, Dipl.-Ing. A. Herb, Dipl.-Ing. S. Köble, Dipl.-Ing. R. Beutel,

Dipl.-Ing. C. Manocchio, Dipl.-Ing. A. Kalckbrenner (M.Sc.), Mr. G. Bahr and

Mr. G. Schmidt. A special acknowledgement is going to Prof. H.-W. Reinhardt

who initiated this research project and contributed during the years in numerous

ways.

The results shown regarding measurements in the frame of the RILEM TC

185-ATC were obtained during a collaboration with the research group of Dr.

Laurent Arnaud, Laboratoire Géomatériaux, Département Génie Civil et

Bâtiment, of the Ecole Nationale des Travaux Publics de l’Etat (ENTPE) in

Vaulx-en-Velin near Lyon, France.

108

http://www.rilem.org/atc.html


REFERENCES

Bahr, G.: Entwicklung von Algorithmen für die kontinuierliche Wavelet Trans-

formation mit LabView. University of Stuttgart, internal report (2001a).

Bahr, G.: Bedienungsanleitung FreshCon 2.04. University of Stuttgart, Institute

of Construction Materials, manual (2001b).

Beutel, R.: Praktische Anwendbarkeit der Ultraschallwellenmessung als Instru-

ment zur Bestimmung des Erhärtungsgrades von Beton. Diploma thesis,

University of Stuttgart, 2000.

Fischer, J.: US-Messungen an Frischbeton. Diploma thesis, University of

Stuttgart, 1994.

Grosse, C. U., H.-W. Reinhardt: Continuous ultrasound measurements during

setting and hardening of concrete. Otto-Graf-Journal 5 (1994), pp 76-98.

Grosse, C. U.: Quantitative zerstörungsfreie Prüfung von Baustoffen mittels

Schallemissionsanalyse und Ultraschall. PhD Thesis, University of Stuttgart,

1996, 168 pages.

Grosse, C. U., B. Weiler, A. Herb, G. Schmidt, K. Höfler: Advances in ultra-

sonic testing of cementitious materials. Festschrift zum 60. Geb. von Prof.

Reinhardt (C. U. Grosse, Ed.), Libri publishing company, Hamburg (1999),

pp. 106-116.

Grosse, C. U., H.-W. Reinhardt: Ultrasound technique for quality control of

cementitious materials. Proc. of 15. World Conf. on NDT, Rom 2000, (on

CD-ROM and in the internet at www.ndt.net).

Grosse, C. U.: Verbesserung der Qualitätssicherung von Frischbeton mit

Ultraschall. Concrete Plant and Precast Technology, Vol. 67, No. 1 (2001),

pp. 102-104.

Grosse, C. U., H.-W. Reinhardt: Fresh concrete monitored by ultrasound

methods. Otto-Graf-Journal Vol. 12 (2001), pp. 157-168.

Herb, A.: Frischbeton: Korrelation zwischen Ergebnissen klassischer Konsis-

tenzmessungen und Ultraschall-Verfahren. Diploma thesis, University of

Stuttgart, 1996.

Kalckbrenner, A.: On the modification of non-destructive ultrasound

measurement techniques for quality control of cement based materials.

Master Thesis, University of Stuttgart, 2002.


http://www.ndt.net/

C. U. GROSSE

Köble, S.: Physikalisch-chemischer Hintergrund des Hydratationsvorgangs von

Frischmörtel im Hinblick auf Ultraschalluntersuchungen. Diploma thesis,


Manocchio, C.: Verwendung der Wavelet-Transformation zur Charakterisierung

von Frischbeton mittels Ultraschall. Diploma thesis, University of Stuttgart,

2001.

Reinhardt, H.-W., C. U. Grosse: Setting and hardening of concrete continuously

monitored by elastic waves. Proc. of the Int. RILEM Conf. "Prod. methods

and workability of concrete", Paisley/Schottland (1996), pp. 415-425.

Reinhardt, H.-W., C. U. Grosse, A. Herb: Kontinuierliche Ultraschallmessung

während des Erstarrens und Erhärtens von Beton als Werkzeug des

Qualitätsmanagements. Deutscher Ausschuss für Stahlbeton, No. 490

(1999a), pp. 21-64.

Reinhardt, H.-W., C. U. Grosse, A. Herb, B. Weiler, G. Schmidt: Verfahren zur

Untersuchung eines erstarrenden und/oder erhärtenden Werkstoffs mittels

Ultraschall. Patent pending under No. 198 56 259.4 at the German Patent

Institution, Munich (1999b).

Stegmaier, M.: Zerstörungsfreie Prüfung des Erstarrens und Erhärtens von

Beton – Weiterentwicklung des Ultraschallprüfverfahrens. Diploma thesis,


Windisch, N.: Untersuchung der Erhärtung von Beton – hochfester Beton bzw.

Fließbeton – mit Ultraschallwellen, Diploma thesis, University of Stuttgart,

1996.

110

Α δισχρετε βονδ µοδελ φορ 3∆ αναλψσισ οφ τεξτιλε ρεινφορχεδ ανδ πρεστρεσσεδ χονχρετε ελεµεντσ

A DISCRETE BOND MODEL FOR 3D ANALYSIS OF TEXTILE REIN-FORCED AND PRESTRESSED CONCRETE ELEMENTS

DISKRETES VERBUNDMODELL FÜR 3D-FE-BERECHNUNGEN VON TEXTILBEWEHRTEN UND VORGESPANNTEN BETONKONSTRUK-TIONEN

UN MODELE DISCRET DE L'ADHERENCE POUR L'ANALYSE 3D DE STRUCTURES EN BETON RENFORCEES ET PRECONTRAINTES AVEC DES ARMATURES TEXTILES

Μαρκυσ Κργερ, ϑοκο Οβολτ, Ηανσ−Ω. Ρεινηαρδτ

SUMMARY

Τεξτιλε ρεινφορχεδ χονχρετε στρυχτυρεσ σηοω σεϖεραλ σιγνιφιχαντ αδϖανταγεσ

χοµπαρεδ το στεελ ρεινφορχεδ χονχρετε στρυχτυρεσ ωηιχη αρε ωελλ κνοων υπ το

νοω. Ηοωεϖερ σοµε δισαδϖανταγεσ λικε τηε λοω υτιλιζατιον φαχτορ οφ τηε τεξτιλε

ρεινφορχεδ ελεµεντσ βεχοµε οβϖιουσ. Ασ ιν ανψ ρεινφορχεδ στρυχτυρε, α τρανσφερ

οφ φορχεσ φροµ ρεινφορχεµεντ το χονχρετε ισ αχχοµπλισηεδ τηρουγη βονδ. Τηερε−

φορε υνδερστανδινγ ανδ φυρτηερ ιµπροϖεµεντ οφ βονδ προπερτιεσ βετωεεν τεξτιλε

ανδ χονχρετε ισ ιµπορταντ. Ιν τηε παπερ βονδ προπερτιεσ βετωεεν διφφερεντ τεξτιλεσ

ανδ ηιγη περφορµανχε φινε γραιν χονχρετε αρε δισχυσσεδ.

Νυµεριχαλ σιµυλατιονσ ωιτη α ϖαριατιον οφ ινπυτ δατα ωερε περφορµεδ υσινγ

α νονλινεαρ φινιτε ελεµεντ χοδε βασεδ ον τηε µιχροπλανε µοδελ φορ χονχρετε ανδ

τηε δισχρετε βονδ µοδελ. Τηε βονδ µοδελ ισ βασεδ ον α δισχρετε Φινιτε ελεµεντσ

φορµυλατιον ωηιχη χαν βε υσεδ φορ στεελ ρεινφορχεδ χονχρετε ασ ωελλ. Τηε νυ−

µεριχαλ σιµυλατιονσ ανδ ϖαριατιον οφ παραµετερσ σηοω τηε ινφλυενχε οφ διφφερεντ

βονδ χηαραχτεριστιχσ οφ τεξτιλε ρεινφορχεµεντσ ανδ τηερεφορε γιϖε σοµε ηιντσ ον

ποσσιβλε οπτιµισατιον οφ τεξτιλε στρυχτυρεσ.

ZUSAMMENFASSUNG

Ωιε βερειτσ ιν δερ νευερεν Λιτερατυρ ερωηντ, ζειγεν τεξτιλε Βεωεηρυνγσ−

µατεριαλιεν ιν ϖερσχηιεδενεν Ανωενδυνγσγεβιετεν δευτλιχηε ςορτειλε γεγενβερ

κονϖεντιονελλερ Σταηλβεωεηρυνγ. Αβερ αυχη εινιγε Ναχητειλε ωιε διε γερινγε

νυτζβαρε Φεστιγκειτ τεξτιλερ Βεωεηρυνγ ιν Βετονβαυτειλεν υνδ διε δαµιτ ϖερβυν−

δενεν ηοηεν Κοστεν σπρεχηεν γεγεν εινεν Εινσατζ σολχηερ Βεωεηρυνγσµατερια−

Οττο−Γραφ−ϑουρναλ ςολ. 13, 2002 111

Μ. ΚΡ⇐ΓΕΡ, ϑ. ΟΒΟΛΤ, Η.−Ω. ΡΕΙΝΗΑΡ∆Τ

λιεν. Ιµ Αλλγεµεινεν κοµµτ δεµ ςερβυνδ ζωισχηεν Βεωεηρυνγ υνδ Βετον βει

δεραρτιγεν ςερβυνδωερκστοφφεν εινε ηοηε Βεδευτυνγ ζυ, σινδ διεσε δοχη υντερ

ανδερεµ µα⇓γεβενδ φρ δασ Τραγϖερηαλτεν. Ιµ ϖορλιεγενδεν Βειτραγ ωερδεν

δαηερ ωεσεντλιχηε ςερβυνδειγενσχηαφτεν ϖερσχηιεδενερ τεξτιλερ Βεωεηρυνγεν ιν

Βετον δισκυτιερτ υνδ ερλυτερτ.

Ανηανδ ϖον νιχητλινεαρεν Φινιτε−Ελεµεντ−Βερεχηνυνγεν µιτ Παραµετερϖα−

ριατιονεν ωιρδ ειν νευεσ ςερβυνδµοδελλ ζυρ Χηαρακτερισιερυνγ τεξτιλερ Βεωεη−

ρυνγεν ιν Βετον ϖοργεστελλτ. ∆ασ ιν δεν ΦΕ−Χοδε ΜΑΣΑ εινγεβυνδενε ςερ−

βυνδµοδελλ βασιερτ ιµ Ωεσεντλιχηεν αυφ δεν γλειχηεν Ανναηµεν ωιε σιε φρ δεν

Σταηλ−/Βετονϖερβυνδ γελτεν υνδ ωυρδε ιν εινιγεν ωενιγεν Πυνκτεν φρ τεξτιλε

Βεωεηρυνγεν ανγεπασστ. Νυµερισχηε Σιµυλατιονεν ζειγεν, ωιε Εινφλσσε τεξτι−

λερ Βεωεηρυνγεν αυφγρυνδ υντερσχηιεδλιχηερ Στρυκτυρ υνδ Αρτ βερχκσιχητιγτ υνδ

ωιε ζυδεµ τεξτιλε Βεωεηρυνγεν ηινσιχητλιχη δεσ Τραγϖερηαλτενσ τεξτιλβεωεηρτερ

Βαυτειλε οπτιµιερτ ωερδεν κννεν.

RESUME

Λεσ αρµατυρεσ τεξτιλεσ οντ πλυσιευρσ αϖανταγεσ σιγνιφιχατιφσ παρ ραππορτ αυξ

αρµατυρεσ χονϖεντιοννελλεσ εν αχιερ. Χεπενδαντ χερταινσ ινχονϖνιεντσ χοµµε

λε βασ ταυξ δ∋εξπλοιτατιον δε λα ρσιστανχε δε λ∋αρµατυρε τεξτιλε ετ λεσ χοτσ λεϖσ

θυι εν ρσυλτεντ σ∋οπποσεντ ◊ λευρ αππλιχατιον ◊ γρανδε χηελλε. Λ∋αδηρενχε εντρε

λ∋αρµατυρε ετ λε βτον ϕουε υν ρλε ιµπορταντ δανσ λεσ µατριαυξ χοµποσιτεσ, ελλε

εστ σουϖεντ δχισιϖε πουρ λε χοµπορτεµεντ σουσ χηαργε δ∋υνε στρυχτυρε. ∆ανσ

λ∋αρτιχλε πρσεντ, λεσ χαραχτριστιθυεσ δε λ∋αδηρενχε δε διφφρεντεσ αρµατυρεσ τεξ−

τιλεσ σοντ δχριτεσ ετ δισχυτεσ.

∆εσ σιµυλατιονσ νυµριθυεσ αϖεχ υνε ϖαριατιον δεσ παραµτρεσ οντ τ εφ−

φεχτυσ εν υτιλισαντ δεσ λµεντσ φινισ νον−λιναιρεσ βασσ συρ λε µοδλε ∀µιχρο−

πλανε∀ πουρ λε βτον ετ λε νουϖεαυ µοδλε δισχρετ δε λ∋αδηρενχε. Λε µοδλε δε

λ∋αδηρενχε εστ βασ συρ υν µοδλε δισχρετ δε λ∋αδηρενχε θυι πευτ τρε γαλεµεντ

εµπλοψ πουρ λε βτον αϖεχ υνε αρµατυρε εν αχιερ. Λεσ σιµυλατιονσ νυµριθυεσ

ετ λα ϖαριατιον δεσ παραµτρεσ µοντρεντ λ∋ινφλυενχε δε διφφρεντεσ χαραχτριστιθυεσ

δε λ∋αρµατυρε τεξτιλε. Ον πευτ εν δδυιρε δεσ µεσυρεσ πουρ οπτιµισερ λε χοµπορ−

τεµεντ δεσ στρυχτυρεσ αϖεχ δεσ αρµατυρεσ τεξτιλεσ.

ΚΕΨΩΟΡ∆Σ: υνχοατεδ τεξτιλεσ, ιµπρεγνατεδ τεξτιλεσ, χονχρετε, Χαρβον, ΑΡ

γλασσ, βονδ, βονδ µοδελ, 3∆ ΦΕ αναλψσισ, πρεστρεσσ

112


INTRODUCTION

Βονδ βεηαϖιουρ οφ τεξτιλε ρεινφορχεµεντ ιν χονχρετε ισ εξπεχτεδ το ϖαρψ

φροµ τηατ οφ ΦΡΠ βαρσ ορ χονϖεντιοναλ στεελσ βαρσ. Μοστ τεξτιλε ροϖινγσ υσεδ ασ

χονχρετε ρεινφορχεµεντ χονσιστ οφ τηουσανδσ οφ σινγλε φιλαµεντσ ανδ τηερεφορε

χαν νοτ βε δεφινεδ ασ α σινγλε ροδ. Ιφ συχη α ροϖινγ ισ εµβεδδεδ ιν χονχρετε τηε

σηαπε οφ τηε χροσσ σεχτιον δετερµινεσ τηε βονδεδ αρεα ανδ ιτ µυστ βε χλαριφιεδ

ηοω µανψ φιλαµεντσ ωερε ιν διρεχτ χονταχτ ωιτη χονχρετε. Α γρεατ δεαλ οφ ρε−

σεαρχη ηασ βεεν δονε ρεχεντλψ το χηαραχτεριζε βονδ βεηαϖιουρ οφ συχη µυλτιφιλα−

µεντ ελεµεντσ ιν χονχρετε βυτ θυιτε νεω ιννοϖατιονσ νεχεσσιτατε φυρτηερ ρεσεαρχη

/ΒΡΑΜΕΣΗΥΒΕΡ, 2000/, /ΝΑΜΜΥΡ, 1989/, /ΟΗΝΟ, 1994/. Μορεοϖερ θυιτε α νυµ−

βερ οφ εξπεριµενταλ ινϖεστιγατιονσ ηαϖε βεεν χαρριεδ ουτ το υνδερστανδ βονδ βε−

ηαϖιουρ οφ πρεστρεσσεδ ανδ/ορ ιµπρεγνατεδ τεξτιλεσ ορ ροϖινγσ.

Ονε παραµετερ τηατ µαψ στρονγλψ ινφλυενχε τηε βονδ περφορµανχε ισ τηε διφ−

φερενχε ιν τηε χοεφφιχιεντ οφ τηερµαλ εξπανσιον φροµ τηατ οφ στεελ ορ χονχρετε. Ιτ

ισ αλσο κνοων τηατ τρανσϖερσε πρεσσυρε ιµπροϖεσ βονδ ωηιχη ισ νεγλεχτεδ ιν

µανψ βονδ µοδελσ. Ηοωεϖερ, τηισ εφφεχτ σεεµσ το βε νοτ ιµπορταντ φορ εµβεδ−

δεδ µυλτι−φιλαµεντ ροϖινγσ ωηιχη ηαϖε νοτ βεεν φυλλψ ινφιλτρατεδ ωιτη χεµεντ

δυε το ϖοιδσ βετωεεν τηε ιννερ φιλαµεντσ. ∆εσπιτε τηισ τηε Ποισσονσ εφφεχτ βε−

χοµεσ σιγνιφιχαντ ανδ ινφλυενχεσ τηε τρανσϖερσε στρεσσ φιελδ ιφ τηε ροϖινγ ισ ιµ−

πρεγνατεδ ανδ/ορ πρεστρεσσεδ. Σοµε τεστ ρεσυλτσ οφ χαρβον ρεινφορχεδ ανδ

πρεστρεσσεδ σπεχιµεν αρε ιλλυστρατεδ ιν Φιγυρε 1 /ΚΡ⇐ΓΕΡ, 2001Β/.

0 1 2 3

0

5

10

15

20

25

30

35

40

45

50

55

60

4

slip ∆s*, mm

P/(

c-∆

s*)

, N

/mm Carbon

no prestressing prestress 150 N/roving

prestress 250 N/roving

Carbon, epoxy impreg.

no prestressing



Φιγυρε 1: Βονδ στρεσσ περ υνιτ λενγτη ϖερσυσ σλιπ βασεδ ον 20µµ δουβλε σιδεδ πυλλ−ουτ (στορεδ ατ 20°Χ, 65% ΡΗ φορ 40 δαψσ) /ΚΡ⇐ΓΕΡ, 2001Β/



Ιτ χαν βε σεεν τηατ αν ιµπρεγνατιον οφ α χαρβον ροϖινγ ωιτη αν εποξψ ρεσιν

γενεραλλψ ρεσυλτσ ιν α βεττερ βονδ ωηερεασ α ροϖινγ τηατ ωασ νοτ ιµπρεγνατεδ

σηοωσ α λοω µαξιµυµ βονδ στρεσσ ανδ αφτερ βονδ φαιλυρε α ϖερψ λοω φριχτιοναλ

ρεσιστανχε. Ιτ ισ ασσυµεδ τηατ τηε µαιν ρεασον φορ τηισ ισ τηε ριββεδ συρφαχε

φορµεδ βψ τηε βινδερ τηρεαδσ ανδ τηε χηανγε οφ τηε ροϖινγ διαµετερ οϖερ ιτσ

λενγτη, εσπεχιαλλψ ατ τηε χροσσινγ ποιντσ ωηερε τηε περπενδιχυλαρ ωοοφ ροϖινγ ισ

φιξεδ. Τηε βινδερ τηρεαδσ αρε χαυσεδ βψ τηε ωαρπ κνιττινγ προχεσσ (Φιγυρε 2) ανδ

ωερε φιξεδ βψ τηε εποξψ ρεσιν. Ιτ χαν βε σεεν φροµ φιγυρε 1 τηατ πρεστρεσσινγ

λεαδσ το α ηιγηερ βονδ στρενγτη. Ασ δισχυσσεδ αβοϖε, βονδ περφορµανχε οφ τεξτιλε

ρεινφορχεµεντ ιν χονχρετε δεπενδσ ον µανψ διφφερεντ παραµετερσ. Τηισ λεαδσ υσ

το χονσιδερ α φορµυλατιον οφ α συχη βονδ µοδελ ιν ωηιχη τηεσε ασπεχτσ ωουλδ βε

αχχουντεδ φορ.

10 mm

Φιγυρε 2: ∆εταιλ οφ αν εποξψ ιµπρεγνατεδ χαρβον φαβριχ

DISCRETE BOND MODEL FOR FINITE ELEMENT ANALYSIS

Φορ νυµεριχαλ στυδιεσ τηε βονδ προπερτιεσ βετωεεν τεξτιλεσ ανδ χονχρετε, δισχρετε

ελεµεντσ ωερε υσεδ. Τηε βονδ µοδελ προποσεδ βψ /ΟΒΟΛΤ, 2002/ ηασ τηερεφορε

βεεν µοδιφιεδ φορ τεξτιλε ρεινφορχεµεντ ανδ υσεδ τογετηερ ωιτη σολιδ φινιτε ελε−

µεντσ ιν α 3∆ ΦΕ στυδιεσ.

Ιν τηε νυµεριχαλ στυδιεσ βονδ βετωεεν τηε τεξτιλεσ ανδ χονχρετε ωασ σιµυλατεδ

βψ δισχρετε βονδ ελεµεντ τηατ ηαϖε ρεχεντλψ βεεν ιµπλεµεντεδ ιντο 3∆ ΦΕ χοδε

ΜΑΣΑ /ΟΒΟΛΤ, 2002/. Χονχρετε, ωηιχη ισ δισχρετιζεδ βψ τηε τηρεε διµενσιοναλ

φινιτε ελεµεντσ, ισ µοδελλεδ βψ τηε µιχροπλανε µοδελ /ΟΒΟΛΤ, 2001/. Τηε βονδ

ελεµεντσ χοννεχτ τηε χονχρετε φινιτε ελεµεντσ ωιτη τηε ρεινφορχεµεντ τηατ ισ ρεπ−

ρεσεντεδ βψ τηε τρυσσ φινιτε ελεµεντσ (σεε Φιγυρε 3). Ονλψ δεγρεεσ οφ φρεεδοµ ιν

τηε βαρ διρεχτιον αρε χονσιδερεδ. Ηοωεϖερ, βεσιδε τηε τανγεντιαλ στρεσσεσ παραλλελ

το τηε βαρ διρεχτιον, τηε ραδιαλ στρεσσεσ περπενδιχυλαρ το τηε βαρ διρεχτιον αρε

γενερατεδ ασ ωελλ. Ιτ ισ ασσυµεδ τηατ ατ α γιϖεν σλιπ τηε ραδιαλ στρεσσ δεπενδσ ον 114


ερατεδ ασ ωελλ. Ιτ ισ ασσυµεδ τηατ ατ α γιϖεν σλιπ τηε ραδιαλ στρεσσ δεπενδσ ον τηε

γεοµετρψ οφ τηε βαρ ανδ τηε βαρ στραιν ασ ωελλ ασ ον τηε γεοµετρψ ανδ τηε βουνδ−

αρψ χονδιτιονσ οφ τηε χονχρετε σπεχιµεν. Τηε ιντεραχτιον βετωεεν τανγεντιαλ ανδ

ραδιαλ στρεσσεσ ισ αχχουντεδ φορ ιν τηρεε διφφερεντ ωαψσ: (ι) διρεχτλψ, τηε σηεαρ

στρεσσ δεπενδσ ον τηε νονλοχαλ (ρεπρεσεντατιϖε) ραδιαλ στρεσσ οβταινεδ φροµ τηε

χονχρετε ελεµεντσ χλοσε το τηε ρεινφορχινγ βαρ, (ιι) τηε λοχαλ στραιν οφ τηε βαρ ελε−

µεντ ανδ ιτσ λατεραλ εξπανσιον ορ εξτενσιον ανδ (ιιι) ινδιρεχτλψ, ιν α ωαψ τηατ τηε

λαργερ σηεαρ στρεσσ (ηιγηερ βονδ στρενγτη δυε το λαργερ ριβσ ορ ρουγηνεσσ οφ τηε

βαρ ελεµεντ) χαυσε ηιγηερ αχτιϖατιον οφ στρεσσεσ ιν τηε ραδιαλ διρεχτιον.

Ιν τηε πρεσεντ µοδελ, σπλιττινγ οφ χονχρετε ισ ινδιρεχτλψ αχχουντεδ φορ.

Ναµελψ τηε ιντεραχτιον βετωεεν σηεαρ ανδ ραδιαλ στρεσσεσ ρεσυλτσ ιν χορρεσπονδ−

ινγ τανγεντιαλ τενσιλε στρεσσεσ τηατ χαυσεσ χραχκινγ οφ τηε συρρουνδινγ νον−λινεαρ

χονχρετε ελεµεντσ ανδ, τηερεφορε, φαιλυρε οφ βονδ ρεσιστανχε.

Concrete element

fibre element

Bond element(zero width)

Repeated nodes

Φιγυρε 3: Βονδ ελεµεντσ ωιτη ζερο ωιδτη.

Bond stress-slip relation in a 2D consideration

Τηε εξπεριµενταλ εϖιδενχε /ΧΕΒ ΒΥΛΛΕΤΙΝ 230, 1996/ ινδιχατεσ τηατ τηε

λοαδ τρανσφερ βετωεεν ρεινφορχεµεντ ανδ χονχρετε ισ αχχοµπλισηεδ τηρουγη βεαρ−

ινγ οφ τηε ρεινφορχεδ στεελ λυγσ ον συρρουνδινγ χονχρετε ανδ τηρουγη φριχτιον. Ασ

δισχυσσεδ βψ /ΨΑΝΚΕΛΕςΣΚΨ, 1987/, τηε τοταλ βονδ ρεσιστανχε χαν βε δεχοµ−

ποσεδ ιντο τωο χοµπονεντσ: (ι) µεχηανιχαλ ιντεραχτιον χοµπονεντ !µ, ανδ (ιι)

φριχτιον χοµπονεντ !φ. Τηε φριχτιον χοµπονεντ χαν βε σεπαρατεδ ιντο α ρεσιδυαλ

φριχτιον !ρ ανδ α ϖιργιν φριχτιον !ϖ χοµπονεντ. Τηε ρεσιδυαλ φριχτιον ρεπρεσεντσ

φριχτιοναλ ρεσιστανχε υπον σλιπ ρεϖερσαλ ωηερεασ τηε ϖιργιν φριχτιον χοµπονεντ ισ

δυε το τηε αδδιτιοναλ φριχτιοναλ ρεσιστανχε δεϖελοπεδ υπον λοαδινγ το πρεϖιουσλψ

υνδεϖελοπεδ σλιπ λεϖελσ. Ιτ ισ ασσυµεδ τηατ τεξτιλε ρεινφορχεµεντ βεηαϖεσ σιµι−

λαρλψ ασ στεελ ρεινφορχεµεντ δοεσ, ωιτη τηε διφφερενχε τηατ µαινλψ τηε αδηεσιον οφ

τηε τεξτιλε ανδ τηε ρουγηνεσσ οφ τηε συρφαχε ιµπροϖε τηε µεχηανιχαλ βονδ ινστεαδ

οφ τηε στεελ λυγσ.



Βασεδ ον τηε εξπεριµενταλ ρεσυλτσ /ΕΛΙΓΕΗΑΥΣΕΝ, 1983/, /ΜΑΛςΑΡ, 1992/

ανδ ασ ωελλ δοχυµεντεδ βψ /ΛΟΩΕΣ, 2002/, τηε βονδ σλιπ ρελατιονσηιπ οφ στεελ

ρεινφορχεµεντ ιν χονχρετε χαν βε δεσχριβεδ βψ τηε παραµετερσ τηατ αρε συµµα−

ριζεδ ιν Ταβλε 1. Τηε σαµε παραµετερσ αρε υσεδ φορ τεξτιλε ρεινφορχεµεντ, βυτ ιν α

σλιγητλψ διφφερεντ µαννερ. Τηε χυρϖε οφ τηε βονδ στρεσσ ϖερσυσ σλιπ ρελατιονσηιπ

υσεδ φορ τηε νυµεριχαλ στυδιεσ ισ ιλλυστρατεδ ιν Φιγυρε 4.

Ταβλε1: Συµµαρψ οφ τηε µοδελ παραµετερσ

∆εσχριπτιον οφ τηε µοδελ παραµετερ Μοδελ παραµετερ

πεακ µεχηανιχαλ βονδ στρενγτη !µ = !µ,0 ∀ ∗ [ΜΠα]

πεακ φριχτιοναλ βονδ στρενγτη !φ = !φ,0 ∀ ∗ [ΜΠα]

πεακ ϖιργιν φριχτιον βονδ στρενγτη !φ,ϖ = (1−0.4) !φ [ΜΠα]

πεακ ρεσιδυαλ φριχτιον βονδ στρενγτη !φ,ρ = 0.4 !φ [ΜΠα]

σεχαντ το βονδ ρεσπονσε χυρϖε φορ ινιτιαλ λοαδινγ κσεχ [ΜΠα/µµ]

σλιπ ατ ωηιχη πεακ βονδ στρενγτη ισ αχηιεϖεδ σ1 = (!µ+!φ)/κσεχ [µµ]

σλιπ ατ ωηιχη βονδ στρενγτη βεγινσ το δεχρεασε σ2 = σ1+σ2∗ [µµ]

σλιπ ατ ωηιχη µεχηανιχαλ βονδ ρεσιστανχε ισ λοστ σ3 [µµ]

τανγεντ το τηε λοαδ−δισπλαχεµεντ χυρϖε υπον υνλοαδινγ κυνλοαδ [ΜΠα/µµ]

ινιτιαλ τανγεντ το τηε βονδ−σλιπ ρεσπονσε κ1 [ΜΠα/µµ]

τανγεντ το τηε βονδ−σλιπ χυρϖε ατ πεακ ρεσιστανχε κ2 = α⋅κσεχ [ΜΠα/µµ]

∗ ∀ σεε νεξτ χηαπτερ

Τηε παραµετερ !µ,0 ανδ !φ,0 ρεπρεσεντ τηε στρενγτη οφ τηε µεχηανιχαλ ανδ φριχτιοναλ

χοµπονεντ (συβσχριπτ µ ανδ φ), ρεσπεχτιϖελψ, φορ τηε χασε οφ νο χονφινινγ πρεσ−

συρε, νο δαµαγε ανδ ελαστιχαλλψ βεηαϖεδ ρεινφορχινγ βαρ ελεµεντ.

Υπ το τηε σλιπ σ1 ατ ωηιχη πεακ βονδ στρενγτη ισ ρεαχηεδ (σεε Φιγυρε 4), αλλ

ρεσπονσε χυρϖεσ αρε δεφινεδ βψ Μενεγοττο−Πιντο (ΜΠ) εθυατιον /ΜΕΝΕΓΟΤΤΟ−

ΠΙΝΤΟ, 1973/. Τηε χυρϖε δεφινεσ α χυρϖε χοννεχτινγ τωο λινε σεγµεντσ ανδ ιτ

ρεαδσ:

( )

1

Ρ

0 Ρ

1σ σ β (1 β)

1 σ 0

! ∀# ∃% &τ = τ ⋅ τ = ⋅ + − ⋅ ⋅ τ∋ (% &+) ∗

+ ,

!!!

(1)

ωηερε β ισ τηε ρατιο βετωεεν τηε ταργετ ανδ ινιτιαλ τανγεντσ, ανδ σ αρε νορµαλ−

ιζεδ στρεσσ ανδ δισπλαχεµεντ, ρεσπεχτιϖελψ, ανδ Ρ δεφινεσ τηε ραδιυσ οφ τηε χυρϖα−

τυρε. ανδ σ αρε τηε παραµετερσ το χαλχυλατε τηε αβσολυτε στρεσσ ανδ δισπλαχε−

µεντ φροµ τηε νορµαλιζεδ παραµετερσ.

τ! !

0τ 0

116


2

1

κβ

κ= (2)

2 σεχαντκ κ , ωιτη 0=α⋅ ≤α≤1 (3)

0

σσ

σ=! (4)

σεχαντ 20 1 1 σεχαντ

1 2 1 σεχαντ

(κ κ ) (1 )σ σ σ κ

κ κ κ κ

−=⋅ =⋅⋅− −

−αα⋅

1

(5)

0 0σ κτ=⋅ (6)

k1

k2=α·ksecant

ksecant

= m+ f

kunload

Cyclic loading

Monotonic loading

s1 s3 Slip s s2

f,r

m

f,v

f= f,r+ f,v

Bond stress

s0

0

f,r

Φιγυρε 4: Βονδ στρεσσ−σλιπ ρελατιον οφ τηε βονδ ελεµεντ µοδελ

Variation of bond strength in a 3D stress field

Α φαχτορ ∀ (σεε Ταβλε 1) αχχουντσ φορ τηε δεπενδενχψ οφ τηε βονδ στρεσσ ον

τηε στρεσσ−στραιν στατε οφ χονχρετε ανδ στεελ ιν τηε ϖιχινιτψ οφ τηε βονδ ζονε. Ασ α

ρεσυλτ, τηε τωο διµενσιοναλ βονδ στρεσσ ϖερσυσ σλιπ ρελατιονσηιπ σηοων βεφορε ισ

ινφλυενχεδ βψ λατεραλ εξπανσιον ορ εξτενσιον ιν διφφερεντ ωαψσ ανδ βεχοµεσ α

τηρεε διµενσιοναλ µοδελ.



Τηε παραµετερ ∀ ισ χαλχυλατεδ ασ σηοων ιν εθυατιον (7). Τηρεε παραµετερσ

αρε χονσιδερεδ: ∀Σ χοντρολσ τηε ινφλυενχε οφ τηε ψιελδινγ οφ στεελ ρεινφορχεµεντ

ον τηε βονδ ρεσπονσε ανδ ισ σετ το ∀Σ=1 φορ τεξτιλε ρεινφορχεµεντ; ∀Χ αχχουντσ

φορ τηε ινφλυενχε οφ τηε λατεραλ στρεσσεσ βετωεεν ρεινφορχεµεντ ανδ χονχρετε

χαυσεδ βψ τηε στρεσσ ιν χονχρετε ανδ τηε λοχαλ στραιν οφ τηε βαρ ελεµεντ ανδ ιτσ

λατεραλ εξπανσιον ορ εξτενσιον; ∀χψχ χοντρολσ τηε ινφλυενχε οφ τηε λοαδινγ−

υνλοαδινγ−ρελοαδινγ ον τηε βονδ ρεσπονσε.

Σ Χ χψΩ = Ω⋅Ω⋅Ω χ (7)

Ασ σηοων ιν εθυατιον (8) ανδ Φιγυρε 5, τηε παραµετερ ∀Χ, ωηιχη χαν τηεο−

ρετιχαλλψ ϖαρψ βετωεεν 0 ανδ 2, αχχουντσ φορ τωο διφφερεντ εφφεχτσ. Τηε φιρστ ισ τηε

ινφλυενχε οφ τηε λατεραλ στραιν οφ τηε στρεσσεδ βαρ ελεµεντ. Τηε παραµετερ ηΡ ισ α

χονσταντ τηατ ρεπρεσεντσ τηε συρφαχε ρουγηνεσσ οφ τηε ρεινφορχεµεντ βαρ. Χοµ−

παρεδ το τηε ριββεδ ρεινφορχεµεντ, ηΡ ισ χλοσε ρελατεδ το τηε ηειγητ οφ τηε στεελ

λυγσ. ισ τηε ρεινφορχεµεντ στραιν, δσε

σ = 2ρσ τηε βαρ διαµετερ ανδ σµ ισ τηε Ποισ−

σονσ ρατιο οφ τηε υσεδ ρεινφορχεµεντ ελεµεντ. Τηε φαχτορ αρ χοντρολσ τηε ινφλυ−

ενχε οφ τηε ραδιαλ χονχρετε στρεσσ ανδ φορ τηε χαλχυλατιονσ ισ σετ το 1. Τηε παραµε−

τερ αφ χοντρολσ τηε ινφλυενχε οφ τηε ρουγηνεσσ οφ τηε ρεινφορχεµεντ ηΡ ον τηε

βονδ ρεσπονσε. Ιν τηε πρεσεντ στυδψ ιτ ωασ σετ το 2.

Τηε παραµετερ επ,0 ισ τηε στραιν δυε το πρεστρεσσινγ οφ ρεινφορχεµεντ. Χονσε−

θυεντλψ ιν τηε χασε οφ πρεστρεσσινγ ανδ νον εξτερναλ λοαδινγ τηε βονδ ισ ιν−

χρεασεδ ονλψ βψ τηε ραδιαλ στρεσσ ιν χονχρετε νεαρβψ τηε ρεινφορχινγ βαρ.

( )

Ρχ ρ φ σ σ π,0 2

σχ2

σ Ρ

11,0 τανη ( )

ρ0,1 φ1

ρ η

# ∃∋ (σ∋Ω = + α⋅ −α⋅µ⋅ε−ε⋅∋ ⋅

−∋ (∋ (+) ∗

((

(8)

-3,0 -2,0 -1,0 0,0 1,0 2,0 3,0

0,0

1,0

2,0

( )

Ρφ σ σ π,0 2

σχ2

σ Ρ

1, ( )

ρ0,1 φ1

ρ η

σ−α⋅µ⋅ε−ε⋅⋅−

+

χΩ

Φιγυρε 5: ∆εφινιτιον οφ !Χ ασ α φυνχτιον οφ λατεραλ στρεσσ ανδ στραιν

118


Τηε ινφλυενχε οφ τηε ραδιαλ στρεσσ ιν χονχρετε ιν τηε ϖιχινιτψ οφ τηε ρεινφορχ−

ινγ βαρ ισ αχχουντεδ φορ βψ αν αϖεραγε ραδιαλ στρεσσ ρσ περπενδιχυλαρ το τηε βαρ

διρεχτιον. Τηε παραµετερ φχ ισ τηε υνιαξιαλ χοµπρεσσιϖε στρενγτη οφ χονχρετε. Ιν

τηε φινιτε ελεµεντ αναλψσισ τηε αϖεραγε ραδιαλ στρεσσ ασσοχιατεδ το τηε ν−τη βαρ

ελεµεντ ισ χαλχυλατεδ ασ:

Ν Νι

ρ ρ ι Ρι 1 ι 1Ρ

ωιτη1

ς ςς = =

σ = σ ∆ = ∆− ις− (9)

ωηερε ∆ ςι δενοτεσ τηε ϖολυµε ωηιχη χορρεσπονδσ το τηε ι−τη ιντεγρατιον

ποιντ οφ τηε φινιτε ελεµεντ ανδ τηε στρεσσ περπενδιχυλαρ το τηε ρεινφορχεµεντ.

Ν ισ α τοταλ νυµβερ οφ ιντεγρατιον ποιντσ τηατ φαλλ ιντο α χψλινδερ οφ α διαµετερ ∆

(σεε Φιγυρε 6). Ιν τηε πρεσεντεδ µοδελ ∆ ισ ασσυµεδ το βε τηρεε τιµεσ α βαρ δι−

αµετερ (∆ ≈ 3 δ

ιρσ

σ). Ιν (9) ςΡ ισ τηε ρεπρεσεντατιϖε ϖολυµε, ι.ε. τηε ϖολυµε οφ τηε

χονχρετε χψλινδερ οφ διαµετερ ∆ τηατ ισ ασσοχιατεδ το τηε τρυσσ φινιτε ελεµεντ

ωηιχη ρεπρεσεντσ α ρεινφορχινγ βαρ.

Φιγυρε 6: Ρεπρεσεντατιϖε ϖολυµε

Εξπεριµεντσ σηοω τηατ φορ χψχλινγ λοαδινγ−υνλοαδινγ−ρελοαδινγ τηε βονδ

στρενγτη σιγνιφιχαντλψ δεχρεασεσ ωιτη ινχρεασε οφ νυµβερ οφ λοαδινγ χψχλεσ

/ΕΛΙΓΕΗΑΥΣΕΝ, 1983/, /ΒΑΛΑΖΣ, 1991/. Ιν τηε πρεσεντ µοδελ τηισ εφφεχτ ισ αχ−

χουντεδ φορ βψ τηε φαχτορ ∀χψχ τηατ ρεαδσ:

1.1

χψχ0

εξπ 1.2# # ∃Λ∋Ω = − ⋅ ∋ (∋ Λ) ∗) ∗

∃((

(10)

ωηερε # ισ τηε αχχυµυλατεδ σηεαρ ενεργψ δισσιπατιον ανδ #0 ισ α χονσταντ

ρεπρεσεντινγ τηε αρεα υνδερ τηε µονοτονιχ βονδ−σλιπ χυρϖε οφ ρεσπεχτιϖε σηεαρ

χοµπονεντ. Τηε αβοϖε εθυατιον ηασ βεεν προποσεδ βψ /ΕΛΙΓΕΗΑΥΣΕΝ, 1983/ ανδ

ιτ ισ βασεδ ον α λαργε νυµβερ οφ χψχλιχ τεστ δατα οφ στεελ ρεινφορχεδ χονχρετε.



Σιµιλαρ βεηαϖιορ σεεµσ αλσο το βε αππροξιµατελψ ϖαλιδ φορ τεξτιλε ρεινφορχεδ χον−

χρετε, ηοωεϖερ, ιτ ισ νοτ χλαριφιεδ υπ το νοω.

NUMERICAL STUDIES

Ασ σηοων ιν τηε λαστ χηαπτερ τηε βονδ περφορµανχε οφ τηε µοδελ ισ ινφλυ−

ενχεδ βψ ∀Χ ωηιχη αχχουντσ φορ: (ι) τηε ινφλυενχε οφ ραδιαλ στρεσσ οβταινεδ φροµ

τηε συρρουνδινγ χονχρετε ελεµεντσ ανδ (ιι) τηε ινφλυενχε οφ ρεινφορχεµεντ στραιν.

Το δεµονστρατε τηε εφφεχτσ οφ τηεσε τωο διφφερεντ ινφλυενχεσ νυµεριχαλ στυδιεσ

ηαϖε βεεν χαρριεδ ουτ.

Τωο ΦΕ µοδελσ (ΜΙ ανδ ΜΙΙ) ωερε εµπλοψεδ το σηοω τηε ινφλυενχε οφ τηε

τρανσϖερσε στρεσσ φιελδ ανδ τηε ρεινφορχεµεντ στραιν ον τηε βονδ προπερτιεσ. Τηε

ΦΕ µεση οφ τηεσε µοδελσ ισ σηοων ιν Φιγυρε 7. Ιτ ρεπρεσεντσ χονχρετε σπεχιµεν

χονφινεδ ιν διρεχτιον ξ ανδ ψ. Τηε βουνδαρψ χονδιτιονσ ωερε σλιγητλψ διφφερεντ.

Μοδελ Ι (ΜΙ) ηασ σοµε ρεστραινεδ νοδεσ ιν τηε ζ−διρεχτιον ονλψ ατ τηε βοττοµ συρ−

φαχε, ωηερε τηε λοαδ ισ αππλιεδ το τηε βαρ ελεµεντ. Ιν Μοδελ ΙΙ (ΜΙΙ) αλλ τηε νοδεσ

οϖερ τηε σπεχιµεν ηειγητ ωερε φιξεδ ιν τηε ζ−διρεχτιον. Χονσεθυεντλψ, τηε τρανσ−

ϖερσε στρεσσ φιελδ αρουνδ τηε βαρ ελεµεντ ισ διφφερεντ. Τηε βαρ ελεµεντ ιτσελφ ισ

πλαχεδ ιν τηε µιδδλε οφ τηε σπεχιµεν ανδ ισ πυλλεδ ιν ζ διρεχτιον ατ τηε ποιντ

ζ = 20 µµ (βοττοµ συρφαχε).

Φιγυρε 7: Φινιτε ελεµεντ µοδελσ ωιτη διφφερεντ βουνδαρψ χονδιτιονσ

Ασ σηοων ιν ταβλε 2, τηε µαιν βονδ παραµετερσ ωερε σετ το χονσταντ φορ αλλ

χαλχυλατιονσ. Νοτε τηατ τηεσε παραµετερσ ωερε χηοσεν ονλψ το θυαλιτατιϖελψ σηοω

τηε περφορµανχε οφ τηε µοδελ ανδ ωερε νοτ χαλιβρατεδ φορ τηε υσε ιν τηε πραχτιχαλ

120


αππλιχατιονσ. Αδδιτιοναλλψ, ιτ ηασ το βε νοτεδ τηατ τηε ρεινφορχεµεντ ισ ασσυµεδ

το βε λινεαρ ελαστιχ φορ αλλ τηε χαλχυλατιονσ, ι.ε. ∀Σ = 1.

Ταβλε2: Συµµαρψ οφ τηε υσεδ µοδελ παραµετερσ φορ τηε ρεινφορχεµεντ

∆εσχριπτιον οφ τηε µοδελ παραµετερ Μοδελ παραµετερ

πεακ µεχηανιχαλ βονδ στρενγτη !µ = 9.0 [ΜΠα]

πεακ φριχτιοναλ βονδ στρενγτη !φ = 4.0 [ΜΠα]

σεχαντ το βονδ ρεσπονσε χυρϖε φορ ινιτιαλ λοαδινγ κσεχ = 50.0 [ΜΠα/µµ]

σλιπ ατ ωηιχη βονδ στρενγτη βεγινσ το δεχρεασε σ2∗ = 0.03 [µµ]

σλιπ ατ ωηιχη µεχηανιχαλ βονδ ρεσιστανχε ισ λοστ σ3 = 0.50 [µµ]

τανγεντ το τηε λοαδ−δισπλαχεµεντ χυρϖε υπον υνλοαδινγ κυνλοαδ = 220.0 [ΜΠα/µµ]

ινιτιαλ τανγεντ το τηε βονδ−σλιπ ρεσπονσε κ1 = 220.0 [ΜΠα/µµ]

τανγεντ το τηε βονδ−σλιπ χυρϖε ατ πεακ ρεσιστανχε κ2 = 22.0 [ΜΠα/µµ]

ραδιυσ οφ τηε χυρϖατυρε Ρ = 8.0 [−]

Ποισσονσ ρατιο µσ = 0.5 [−]

Ψουνγ µοδυλυσ Εσ = 74000.0 [ΜΠα]

ρεινφορχεµεντ αρεα Ασ = 0.93 [µµ″]

βαρ διαµετερ 2 ⋅ ρσ = 1.0 [µµ]

συρφαχε ρουγηνεσσ ηΡ = 0.01 [µµ]

Μορεοϖερ, το δεµονστρατε τηε εφφεχτ οφ πρεστρεσσινγ, τηρεε διφφερεντ χασεσ

(α,β,χ) ωερε χονσιδερεδ ωιτη:

(α) (11) Χ,α 1.0Ω =

(β) ΡΧ,β

χ

1.0 τανη0,1 φ

#σΩ = + ∋ ⋅

) ∗

∃( (12)

(χ)

( )

ΡΧ,χ φ σ σ π,0 2

σχ2

σ Ρ

11.0 τανη ( )

ρ0,1 φ1

ρ η

# ∃∋ (σ∋Ω = + −α⋅µ⋅ε−ε⋅

∋ ⋅−∋ (∋ (+) ∗

(( (13)

Influence of the 3D stress field on the bond

Ιν Φιγυρε 8 τηε ρεσυλτσ οφ τηε πυλλ ουτ στρεσσ ϖερσυσ σλιπ αρε σηοων. Ιν τηε

χασε (α) τηε στρεσσ ϖερσυσ σλιπ χυρϖε οφ βοτη µοδελσ λοοκσ αλµοστ τηε σαµε φορ

πρεστρεσσεδ ανδ νον πρεστρεσσεδ στατε, ι.ε. τηε χονχρετε στραιν ιν ζ διρεχτιον ισ

νεγλιγιβλε. Τηερεφορε ιτ ισ σηοων ονλψ ονε χυρϖε. Ηοωεϖερ, ιτ χαν βε σεεν τηατ ιφ



τηε ρεινφορχεµεντ ισ πρεστρεσσεδ µαξιµυµ πυλλουτ στρεσσ ινχρεασεσ σλιγητλψ δυε

το τηε µορε ηοµογενεουσ βονδ στρεσσ διστριβυτιον οϖερ τηε εµβεδµεντ λενγτη.

Φορ διφφερεντ χασεσ ατ µαξιµυµ λοαδ αλσο σεε Φιγυρε 9. Τηισ εφφεχτ χαν αλσο βε

σεεν ιν αλλ τηε χαλχυλατιον δισχυσσεδ λατερ.

0,0 0,1 0,2 0,3 0,4 0,5 0,6

0

200

400

600

800

1000

1200 MI & MII, case (a) MI & MII, case (a), prestressed MI, case (b)

MI, case (b), prestressed MII, case (b) MII, case (b), prestressed

loa

d, N

slip, mm Φιγυρε 8: Χαλχυλατεδ πυλλ ουτ λοαδ ϖερσυσ σλιπ

Ιφ χασε (β) ισ χονσιδερεδ ανδ τηε ινφλυενχε οφ τηε ραδιαλ στρεσσ οφ τηε χον−

χρετε ισ τακεν ιντο αχχουντ, τηε χηανγε οφ βονδ ρεσπονσε βεχοµεσ οβϖιουσ. Τηε

ινφλυενχε χαλχυλατεδ ιν µοδελ ΜΙΙ ισ αλµοστ ινσιγνιφιχαντ ωηερεασ ιν µοδελ ΜΙ

µαξιµυµ πυλλ ουτ στρεσσ ινχρεασεσ οϖερ 30 περχεντ χαυσεδ βψ τηε δεφορµατιον οφ

τηε χονχρετε ελεµεντσ. Αδδιτιοναλλψ τηε βονδ στρεσσ ισ χαλχυλατεδ ασ 1.5 τιµεσ ασ

ηιγη ασ φορ χασε (α) ατ ζ = 15µµ ωηιχη χαν βε εξπλαινεδ βψ τηε βουνδαρψ χονδι−

τιονσ ανδ τηε ρεσυλτινγ στρεσσ φιελδ οφ χονχρετε.

0 2 4 6 8 10 12 14 16 18 20

8

10

12

14

16

18

20

load

MI & MII, case (a) MI & MII, case (a), prestressed MI, case (b) MI, case (b), prestressed MII, case (b) MII, case (b), prestressed

Bo

nd

str

ess, N

/mm

2

embedment length z, mm Φιγυρε 9: Χοµπαρισον οφ χαλχυλατεδ βονδ στρεσσ ατ µαξιµυµ πυλλ ουτ λοαδ

122


Influence of the reinforcement strain on the bond performance

Τηε ρεσυλτσ οφ τηε χαλχυλατιονσ οφ τηε πυλλ ουτ στρεσσ ϖερσυσ σλιπ φορ χασε (χ)

αρε σηοων ιν Φιγυρε 10. Φορ τηε µοδελ ΜΙ τηε µαξιµυµ λοαδ ισ ϕυστ αβουτ 15

περχεντ ηιγηερ τηαν φορ χασε (α) ανδ λοωερ τηαν φορ χασε (β). Αλσο τηε βονδ στρεσσ

οϖερ τηε εµβεδµεντ δεπτη ισ λοωερ φορ χασε (χ) χοµπαρεδ το χασε (α), ασ σηοων

ιν Φιγυρε 12.

0,0 0,1 0,2 0,3 0,4 0,5 0,6

0

200

400

600

800

1000

1200

MI & MII, case (a) MI, case (c) MII, case (c)

loa

d, N

slip, mm

Φιγυρε 10: Χαλχυλατεδ πυλλ ουτ λοαδ ϖερσυσ σλιπ

0,0 0,1 0,2 0,3 0,4 0,5 0,6

0

200

400

600

800

1000

1200

MI & MII, case (a)

MI, case (c), prestressed MII, case (c), prestressed

loa

d, N

slip, mm

Φιγυρε 11: Χαλχυλατεδ πυλλ ουτ λοαδ ϖερσυσ σλιπ οφ πρεστρεσσεδ σπεχιµεν



Ιν τηε χαλχυλατεδ βονδ στρεσσ σλιπ χυρϖεσ φορ πρεστρεσσεδ ρεινφορχεµεντσ,

σηοων ιν Φιγυρε 11, τηε ινφλυενχε οφ τηε στεελ στραιν ον τηε βονδ περφορµανχε ισ

αλσο οβϖιουσ. ∆υε το τηε πρεστρεσσινγ οφ αδηεσιϖε τψπε, ατ ινιτιαλ στατε τηε ρειν−

φορχεµεντ λατεραλ στραινσ αρε νεγατιϖε (χοµπρεσσιον). Χονσεθυεντλψ, τηε χοντραχ−

τιον οφ ρεινφορχεµεντ ισ λεσσ τηαν ιν τηε χασε οφ υνπρεστρεσσεδ ρεινφορχεµεντ ανδ

τηερεφορε ιτ δοεσ νοτ ρεσυλτ ιν συχη α λαργε ρεδυχτιον οφ βονδ στρεσσεσ. Φιγυρε 12

σηοωσ τηε ινφλυενχε οφ ∀Χ ον τηε βονδ στρεσσ φορ υνπρεστρεσσεδ ανδ πρεστρεσσεδ

ρεινφορχεµεντ.

0 2 4 6 8 10 12 14 16 18 20

8

10

12

14

16

18

20

load

MI & MII, case (a) MI & MII, case (a), prestressed MI, case (c) MI, case (c), prestressed MII, case (c) MII, case (c), prestressed

Bo

nd

str

ess, N

/mm

2

embedment length z, mm

Φιγυρε 12: Χοµπαρισον οφ χαλχυλατεδ βονδ στρεσσ ατ µαξιµυµ πυλλ ουτ λοαδ

Comparison of calculations and tests of textile reinforced elements in a bending test

Ιν Φιγυρε 13 τηε ρεσυλτ οφ α φουρ−ποιντ βενδινγ τεστ ισ χοµπαρεδ ωιτη τηε ρε−

συλτσ οφ τηε ΦΕ χαλχυλατιονσ. Τηε σπαν οφ τηε πλατε ωασ 250 µµ ανδ τηε λοαδ ωασ

αππλιεδ ατ τηε τηιρδ ποιντσ. Τηε τεστ σπεχιµεν ωασ α εποξψ ιµπρεγνατεδ χαρβον

τεξτιλε ρεινφορχεδ χονχρετε πλατε (300µµ ξ 60µµ ξ 10µµ) ωιτη α φινε γραιν

χονχρετε (φχ ≈ 80ΜΠα). Τηε ινπυτ παραµετερσ φορ τηε χαλχυλατιον οφ τηε βενδινγ

τεστ ωερε χαλιβρατεδ ατ δουβλε σιδεδ πυλλουτ τεστσ. Φορ δεταιλσ σεε /ΚΡ⇐ΓΕΡ,

2002/. Φορ τηε χαλχυλατιον χονχρετε ισ µοδελλεδ βψ τηε µιχροπλανε µοδελ. Τηε

αγρεεµεντ βετωεεν σιµυλατιον ανδ εξπεριµενταλ ρεσυλτσ ισ γοοδ. Ηοωεϖερ, χοµ−

παρεδ το τηε τεστ ρεσυλτσ, τηε νυµεριχαλ ρεσυλτσ σηοω ιν τηε ποστ πεακ ρεγιον µορε

δυχτιλε βεηαϖιουρ.

124


0 5 10 15 20 25

0

200

400

600

800

1000

1200 Calculated, Case (a) Calculated, Case (c) test data

Displacement, mm

Loa

d,

N

Φιγυρε 13: Λοαδ−δεφλεχτιον χυρϖε φορ α χαρβον ρεινφορχεδ ελεµεντ υνδερ µονοτονιχ λοαδινγ

Ασ χαν βε σεεν φροµ Φιγυρε 14, τηε στραιν ιν ξ διρεχτιον (δαρκ ζονεσ ινδιχατε

χραχκσ) σηοω α γοοδ αγρεεµεντ ωιτη τηε χραχκ διστριβυτιον οβσερϖεδ ιν τηε εξ−

περιµεντ. Ιτ ηασ το βε νοτεδ τηατ τηε χραχκ διστριβυτιον ιν τηε τεστεδ σπεχιµεν ισ

γρεατλψ ινφλυενχεδ βψ τηε τρανσϖερσε τεξτιλε ρεινφορχεµεντ, ωηιχη λεαδσ το λεσσ

χραχκσ βυτ α ηιγηερ χραχκ ωιδτη.

X

Y

Z

0.03

0.0287

0.0275

0.0262

0.025

0.0238

0.0225

0.0212

0.02

0.0187

0.0175

0.0163

0.015

0.0138

0.0125

0.0113

0.01

0.00875

0.0075

0.00625

0.005

1

utput Set: MASA3 pbzfCEP6072eformed(20.26): Total nodal disp.

χραχκσ τεξτιλε ρεινφορχεµεντ

Φιγυρε 14: Χοµπαρισον οφ πρινχιπλε στραινσ οφ χαλχυλατεδ µοδελ ιν ξ διρεχτιον ατ µαξιµυµ λοαδ ανδ χραχκ διστριβυτιον οφ τεστεδ σπεχιµεν.



Ιν Φιγυρε 15 τηε στραινσ οφ τηε χονχρετε ελεµεντσ ιν ψ διρεχτιον (ηοριζονταλ

χραχκσ) αρε σηοων ανδ χοµπαρεδ ωιτη σπεχιµεν. Τηε δαρκ ζονεσ ρεπρεσεντσ

χραχκεδ χονχρετε. Ιν τηε εξπεριµεντ αλµοστ τηε σαµε χραχκ διστριβυτιον ανδ τηε

σαµε φαιλυρε µοδε ωασ οβσερϖεδ.

X

Y

Z

0.005

0.00475

0.0045

0.00425

0.004

0.00375

0.0035

0.00325

0.003

0.00275

0.0025

0.00225

0.002

0.00175

0.0015

0.00125

0.001

0.00075

0.0005

0.00025

0.

L1

C1

Output Set: MASA3 pbzfCEP6080

Deformed(22.23): Total nodal disp.

C

onto r A rg E stra

Φιγυρε 15: Χοµπαρισον οφ τεστεδ σπεχιµεν αφτερ τεστ ανδ πρινχιπλε στραινσ οφ χονχρετε ελε−µεντσ ιν ψ διρεχτιον ατ µαξιµυµ λοαδ.

CONCLUSIONS

Α νεω δισχρετε βονδ µοδελ τηατ ισ βασεδ ον α βονδ στρεσσ−σλιπ ρελατιονσηιπ

ηασ ρεχεντλψ βεεν ιµπλεµεντεδ ιντο α 3∆ φινιτε ελεµεντ χοδε. Τηε βονδ µοδελ

αχχουντσ φορ τηε ινφλυενχε οφ ελαστιχ ανδ πλαστιχ ρεινφορχεµεντ στραινσ, τηε ινφλυ−

ενχε οφ τηε ραδιαλ στρεσσ οφ τηε συρρουνδινγ χονχρετε ασ ωελλ ασ φορ τηε ινφλυενχε

οφ τηε χψχλιχ λοαδ ηιστορψ ον τηε βονδ ρεσπονσε.

Ασ σηοων ιν τηε νυµεριχαλ εξαµπλεσ, τηε τρανσϖερσε στρεσσ φιελδ ανδ τηε ρε−

ινφορχεµεντ στραιν µαψ ηαϖε σιγνιφιχαντ ινφλυενχε ον τηε λοχαλ βονδ στρεσσ. Ιφ α

ρεινφορχεµεντ ωιτη α ρουγη συρφαχε ισ υσεδ τηε λοχαλ βονδ στρεσσ ισ µαινλψ ινφλυ−

ενχεδ βψ τηε ραδιαλ στρεσσ οφ τηε συρρουνδινγ χονχρετε. Ηοωεϖερ, τηε ινφλυενχε οφ

τηε ρεινφορχεµεντ στραιν ινχρεασεσ ασ σµοοτηερ τηε ρεινφορχεµεντ συρφαχε ισ.

Τηε παραµετερ ∀Χ ινφλυενχεσ τηε λοχαλ φαιλυρε οφ χονχρετε χλοσε το τηε βαρ

νεαρβψ α χραχκ ωηερε ρελατιϖελψ ηιγη στρεσσεσ ιν ρεινφορχεµεντ αρε πρεσεντ. Τηε

βονδ στρενγτη ισ ρεδυχεδ ανδ τηερεφορε χραχκ ωιδτη ανδ τηε διστριβυτιον οφ χραχκσ

ισ αφφεχτεδ. Ιτ ισ ωελλ κνοων τηατ αλσο ψιελδινγ οφ στεελ ρεινφορχεµεντ ενλαργε τηισ

εφφεχτ ωηιχη ισ αχχουντεδ φορ βψ ∀Σ ιν τηε βονδ µοδελ.

126


Τηε βενεφιτ οφ τηε πρεσεντεδ βονδ µοδελ βεχοµεσ οβϖιουσ ιφ ονε χονσιδερ

διφφερεντ τψπεσ οφ ρεινφορχεµεντσ, ε.γ. διφφερεντ διαµετερ, συρφαχε στρυχτυρεσ ορ

στεελ λυγσ ανδ στρεσσ στραιν προπερτιεσ. Ιτ ισ ασσυµεδ τηατ τηε γενεραλ παραµετερσ

οφ τηε βονδ στρεσσ σλιπ−ρελατιον σηοων ιν Φιγυρε 1 µαινλψ δεπενδ ον τηε χονχρετε

παραµετερσ ανδ χαν βε σετ το χονσταντ φορ α γρουπ οφ ρεινφορχεµεντ ελεµεντσ οφ

τηε σαµε τψπε. Τηισ χαν βε φορ εξαµπλε α σετ οφ στεελ βαρσ οφ διφφερεντ διαµετερ

ορ τεξτιλε ρεινφορχεµεντ τψπε ωιτη διφφερεντ Ψουνγσ µοδυλυσ βυτ αλµοστ τηε

σαµε συρφαχε ρουγηνεσσ.

Νεϖερτηελεσσ τηε δισχυσσεδ µοδελ ηαϖε το βε χαλιβρατεδ βασεδ ον α σεριεσ οφ

διφφερεντ εξπεριµενταλ τεστσ ιν ορδερ το φινδ ουτ τηε ρεαλ ινφλυενχε οφ τρανσϖερσε

στρεσσεσ ανδ στραινσ ον τηε βονδ ρεσπονσε ανδ τηυσ ον τηε στρυχτυραλ ρεσπονσε ασ

ωελλ.

REFERENCES

Βαλαζσ, Γ.Λ. (1991): Φατιγυε οφ βονδ. ΑΧΙ Ματεριαλσ ϑουρναλ, 88 (6), 620−629,

1991.

Βραµεσηυβερ, Ω.; Βανηολζερ, Β.; Βρµµερ, Γ. (2000): Ανσατζ φρ εινε ϖερειν−

φαχητε Αυσωερτυνγ ϖον Φασερ−Αυσζιεηϖερσυχηεν. Βετον− υνδ Σταηλβετονβαυ, Νο.

95, Ηεφτ 12, ππ. 702−706, 2000.

ΧΕΒ Βυλλετιν 230 (1996): ΡΧ ελεµεντσ υνδερ χψχλιχ λοαδινγ. Στατε οφ τηε αρτ ρε−

πορτ, Εδ. Βψ Τ. Τελφορδ, Τηοµασ Τελφορδ Σερϖιχε Λτδ, Λονδον, 1996.

Χυρβαχη, Μ.; Ζαστραυ, Β. (1999): Τεξτιλβεωεηρτερ Βετον Ασπεκτε αυσ Τηεοριε

υνδ Πραξισ. Βαυστατικ−Βαυπραξισ 7, Μεσκουρισ (Εδ.), Βαλκεµα, Ροττερδαµ, 1999.

Ελιγεηαυσεν, Ρ.; Ποποϖ, Ε.Π.; ανδ Βερτερο, ς.ς. (1983): Λοχαλ Βονδ Στρεσσ−Σλιπ

Ρελατιονσηιπσ οφ ∆εφορµεδ Βαρσ υνδερ Γενεραλιζεδ Εξχιτατιονσ. Ρεπορτ

ΥΧΒ/ΕΕΡΧ−83/23. Βερκελεψ: ΕΕΡΧ, Υνιϖερσιτψ οφ Χαλιφορνια, 1983.

Κργερ, Μ. (2001α): Πρεστρεσσεδ Τεξτιλε Ρεινφορχεδ Χεµεντ Χοµποσιτεσ. ΙΩΒ−

Μιττειλυνγεν, ϑαηρεσβεριχητ 2000/2001, Υνιϖερσιτψ οφ Στυττγαρτ, Ινστιτυτε οφ Χον−

στρυχτιον Ματεριαλσ (ΙΩΒ), 2001.

Κργερ, Μ.; Ρεινηαρδτ, Η.−Ω.; Φιχητλσχηερερ, Μ. (2001β): Βονδ βεηαϖιουρ οφ

τεξτιλε ρεινφορχεµεντ ιν ρεινφορχεδ ανδ πρεστρεσσεδ χονχρετε. Ιν: Οττο−Γραφ−

ϑουρναλ. ςολ. 12 2001, ππ. 33−50.

Κργερ, Μ.; Ρεινηαρδτ, Η.−Ω. (2001χ): Πρεστρεσσεδ τεξτιλε ρεινφορχεδ χεµεντ

χοµποσιτεσ. Προχ. 11. Ιντερνατιοναλε Τεχητεξτιλ−Σψµποσιυµ φρ τεχηνισχηε Τεξ−

τιλιεν, ςλιεσστοφφε υνδ τεξτιλαρµιερτε Ωερκστοφφε, Νο. 338, Φρανκφυρτ, Απρ. 2001.



Κργερ, Μ.; Ξυ, Σ.; Ρεινηαρδτ, Η.−Ω.; Οβολτ, ϑ. (2002): Εξπεριµενταλ ανδ νυ−

µεριχαλ στυδιεσ ον βονδ προπερτιεσ βετωεεν ηιγη περφορµανχε φινε γραιν χον−

χρετε ανδ χαρβον τεξτιλε υσινγ πυλλ ουτ τεστσ. Ιν: Βειτργε αυσ δερ Βεφεστιγυνγσ−

τεχηνικ υνδ δεµ Σταηλβετονβαυ (Φεστσχηριφτ ζυµ 60. Γεβυρτσταγ ϖον Προφ. ∆ρ.−

Ινγ. Ρ. Ελιγεηαυσεν), ππ. 151−164, Στυττγαρτ, 2002.

Λοωεσ, Λ. Ν.; Μοεηλε, ϑ.Π.; Γοϖινδϕεε, Σ. (2002): Α χονχρετε−στεελ βονδ µοδελ

φορ υσε ιν φινιτε ελεµεντ µοδελλινγ οφ ρεινφορχεδ χονχρετε στρυχτυρεσ. Ιν πριντ,

2002.

Μαλϖαρ, Λ.ϑ. (1992): Βονδ ρεινφορχεµεντ υνδερ χοντρολλεδ χονφινεµεντ. ΑΧΙ

Ματεριαλσ ϑουρναλ 89 (6), 711−721, 1992.

Μενεγοττο, Μ.; Πιντο, Π. (1973): Μετηοδ οφ αναλψσισ οφ χψχλιχαλλψ λοαδεδ ρειν−

φορχεδ χονχρετε πλανε φραµεσ ινχλυδινγ χηανγεσ ιν γεοµετρψ ανδ νονελαστιχ βε−

ηαϖιουρ οφ ελεµεντσ υνδερ χοµβινεδ νορµαλ γεοµετρψ ανδ νονελαστιχ βεηαϖιουρ

οφ ελεµεντσ υνδερ χοµβινεδ νορµαλ φορχε ανδ βενδινγ. Προχεεδινγσ οφ τηε

ΙΑΒΣΕ Σψµποσιυµ ον τηε ρεσιστανχε ανδ υλτιµατε δεφορµαβιλιτψ οφ στρυχτυρεσ

αχτεδ ον βψ ωελλ−δεφινεδ ρεπεατεδ λοαδσ, Λισβον, 1973.

Ναµµυρ, Γ.; Νααµαν, Α. (1989): Βονδ Στρεσσ Μοδελ φορ Φιβερ Ρεινφορχεδ Χον−

χρετε Βασεδ ον Βονδ Στρεσσ−Σλιπ Ρελατιονσηιπ. ΑΧΙ Ματεριαλσ ϑουρναλ, Νο. 86,

ππ. 45−55, ϑαν./Φεβρ. 1989.

Οηνο, Σ.; Ηανναντ, ∆.ϑ. (1994): Μοδελλινγ τηε στρεσσ−στραιν Ρεσπονσε οφ Χον−

τινυουσ Φιβρε Ρεινφορχεδ Χεµεντ Χοµποσιτεσ. ΑΧΙ Ματεριαλσ ϑουρναλ, Νο. 91,

ππ. 306−312, Μαρ. 1994.

Οβολτ, ϑ.; Λι Ψ.; Κοαρ, Ι. (2001): Μιχροπλανε µοδελ φορ χονχρετε ωιτη ρελαξεδ

κινεµατιχ χονστραιντ. Ιντ. ϑ. οφ Σολιδσ ανδ Στρυχτυρεσ, 38, 2683−2711, 2001.

Οβολτ, ϑ.; Λεττοω, Σ.; Κοαρ, Ι. (2002): ∆ισχρετε βονδ ελεµεντ φορ 3∆ ΦΕ αναλψ−

σισ οφ ρεινφορχεδ χονχρετε στρυχτυρεσ. Ιν: Βειτργε αυσ δερ Βεφεστιγυνγστεχηνικ

υνδ δεµ Σταηλβετονβαυ (Φεστσχηριφτ ζυµ 60. Γεβυρτσταγ ϖον Προφ. ∆ρ.−Ινγ. Ρ.

Ελιγεηαυσεν), ππ. 239−258, Στυττγαρτ, 2002.

Ρεινηαρδτ, Η.−Ω.; Κργερ, Μ. (2001): ςοργεσπανντε δννε Πλαττεν αυσ Τεξτιλβε−

τον. Προχ. Τεξτιλβετον 1. Φαχηκολλοθυιυµ δερ Σονδερφορσχηυνγσβερειχηε 528

υνδ 532, εδιτεδ βψ ϑ. Ηεγγερ, ππ. 165−174, Ααχηεν, 2001.

Ψανκελεϖσκψ, ∆.Ζ.; Ρεινηαρδτ, Η.−Ω. (1987): Ρεσπονσε οφ πλαιν χονχρετε το χψ−

χλιχ τενσιον. ΑΧΙ Ματεριαλσ ϑουρναλ 84 (5), 365−373 1987.

128

Experimental realisation of a pretentious testing task on the field of pioneer bridge structures

EXPERIMENTAL REALISATION OF A PRETENTIOUS TESTING TASK ON THE FIELD OF PIONEER BRIDGE STRUCTURES

VERSUCHSTECHNISCHE REALISIERUNG EINER NICHT ALLTÄGLICHEN PRÜFAUFGABE AUS DEM BEREICH DER PIONIERBRÜCKENKONSTRUKTIONEN

REALISATION D'UN ESSAI DE CHARGEMENT COMPLEXE D'UN PONTON DU GENIE MILITAIRE

Wolfgang Harre

SUMMARY

An extraordinary test-setup and a pretentious test procedure is described for

investigation of a ponton, submitted to the very manifold and complex loading

conditions of pioneer bridge structures.

ZUSAMMENFASSUNG

Es wird der aufwendige Versuchsaufbau und die anspruchsvolle

Versuchstechnik erläutert, um im Prüflabor die komplizierten

Beanspruchungsverhältnisse mit allen Randbedingungen eines in eine belastete

Schwimmbrücke eingebundenen Pontons nachzufahren, mit dem Ziel, die

Reaktionen (Tragverhalten, Schwingfestigkeit) derartiger geschweißter

Aluminium-Leichtbau-Konstruktionen zu untersuchen.

RESUME

Le dispositif et la procédure d'essai complexes servant à simuler en

laboratoire les conditions de chargement très compliquées d'un ponton faisant

partie d'un pont flottant sont décrites.

KEYWORDS: Testing of Pioneer Bridge Structures, Aluminium-Bridge-

Structures


W. HARRE

1. INTRODUCTION

The development of dismountable bridges (bridge systems, military

bridges, pioneer bridges) requires, besides the extensive design work and

detailed theoretical analysis, also experimental investigations on materials,

structural details, complete substructures (modules) and even on complete

bridge structures. Since many decades, the Otto-Graf-Institute is the leading

testing institution on this field of dismountable bridges.

A very important branch of the dismountable bridges are wet gap military

bridging systems, the so-called floating bridges representing very efficient and

universal useful structures to overcome big rivers, water surfaces and obstacles

(caused for instance by catastrophes, floods etc.).

In the following, it will be reported of a recently finished test project

concerning floating bridge systems, which was outstanding pretentious if

compared with the usual test projects, carried out commonly in the department

2.

2. TEST PROJECT

Basic elements of floating bridges are generally the welded hollow box

girders in aluminium, the so-called pontons or bays, which will – dependent on

the respective demand – be coupled together in appropriate number to form in

composite action a complete floating and load carrying road way, see the

following pictures (Fig. 1 and 2).

Fig. 1: Floating bridge in service

130


Fig 2.: Cross-section of floating bridge (1 = Roadway ponton, 2 = Bow ponton)

An excellent example of floating bridges, the so-called RIBBON-Bridge,

was developed about 25 years ago in Germany by EWK (Eisenwerke

Kaiserslautern). Meanwhile the RIBBON-Bridge is in service in 11 armies

worldwide. Based on the positive experiences and the perfect performance all

over the world in the past, even the US-Army was interested finally in this

floating bridge.

However in order to provide extensively the US-Army with the Ribbon

Bridge, EWK had to satisfy some American proposals concerning a better

handling and carrying capacity of the bridge system. At last, the Americans

wished, that the successful demonstration of the demanded improvements

should be realized by an appropriate full scale test in the Otto-Graf-Institute.

So, in cooperation with EWK, a testing program was developed to prove

the accomplishment of the requested improvements, quasi as a certification of

the IMPROVED RIBBON BRIDGE (IRB).

Essentially, the testing program included the static and dynamic loading of

a complete ponton in a suitable special test set-up. Testing should be carried out

in a procedure, which simulates all the load cases and load characteristics

happening in practical use of the IRB.


W. HARRE

To create that kind of most unfavourable and harmful loading situations in

the ponton means concretely to apply bending moments and longitudinal as well

as transverse loads in the same way and magnitude as it occurs, when a battle

tank MLC 70 either stands on the floating bridge or passes the bridge, like

demonstrated in the following diagram (Fig 3.).

F1

F1

F2

F F3

F F6

FF5

FF4

F

F7

F8

F9

F10

Ponton

Fig. 3: Tank loading in reality and through laboratory simulation

132


3. EXPERIMENTAL REALIZATION

3.1 Test set-up

There were mainly two problems to be solved:

a) Installation and program-operation of a relatively high number

(13) of hydraulic jacks with different capacities (50 kN to 2 MN)

b) Application of high bending moments (partly > 2 MNm by

means of very high horizontal forces)

Even the comparatively abundant and well assorted equipment of the

Department 2 of the Otto-Graf-Institute was not able and sufficient to solve

satisfactorily the two problems in a direct way. Especially the wanted number

and capacity of the hydraulic jacks (at least 4 jacks with more than 2 MN)

represented a considerable challenge. So, some reflection was necessary to find

a way for the experimental realization, using the available equipment.

The central idea of the solution was the consistent application of the lever-

action (Hebelgesetze): The mainly in pairs acting high horizontal forces with

opposite signs could be replaced by a lever-structure as shown basically in the

following sketch.

F12

F1-

6

F1

3

F11

Ponton

Fig. 4: Forces on the ponton


W. HARRE

F7

F8

jack > 1MN

jack > 1MN

task solutionb c

Ponton Ponton

jack < 1MN

a

Fig. 5: Solution of the testing task

Proceeding that way, several profitable effects could be achieved: the

number of the required jacks was reduced from 13 to 5, the capacity of the used

jacks could be adapted perfectly to the jacks available in the department 2 by

corresponding choice of the lever-ratios b:c and last not least – this is very

important with regard to the test set-up – the introduction of the lever structures

opens the possibility to anchor the initially horizontal forces now vertically

either in the strong floor directly or by means of test frames at hand indirectly.

The anchorage of high horizontal forces is – as experience shows – on principle

very difficult and expensive in a laboratory, because these forces have to be

turned round sooner or later to pass and anchor them finally into the floor.

After the concept of the experimental realization was found, the proceeding

was evident: after checking the available and suitable jacks in the department,

the different lever-ratios were calculated for the different loading points. After

that, all the other details were designed. Then all parts were manufactured by

EWK together with the ponton to be tested.

The following pictures will try to give an impression of the complicate and

complex test set-up:

134


Fig. 6. Testing arrangement on the strong floor


W. HARRE

Fig. 7.a: Mid section of testing structure

Fig. 7.b: End section of testing structure

136


Fig. 7.c: Oil supply system

Fig. 7.d: Complex multi-dimensional loading arrangement


W. HARRE

3.2 Test run

The loading program required the independent, however exactly balanced,

synchronous controlling of altogether 5 jacks for static as well as for dynamic

running. The following systematic presentation of the loading functions

illustrates the dynamic test run (Fig. 8).

Test run

jack 2

jack 3

Load-time-diagram

jack 4

time

time

time

time

Compression

Tension

Compression

Compression

Tension

Tension jack 1und 5

Fig. 8: Loading sequence

138


The implementation of this working load test procedure supposed the

electronic coupling of the controlling units (S-59 Regler) of all the jacks.

The exact run down of the whole test program was realized by means of a

„managing“ computer, which directed the different controlling units according

to the test program. In certain intervals, that is at times after reaching

reconceived numbers of cycles, the test program also provided breaks in the

dynamic loading. During these breaks, different static extreme load

configurations were tested. All the measurements (loads, displacements, strains)

happened automatically by a multipoint measuring system. The data were stored

on CD for further evaluation. The described test set-up and equipment allowed a

dynamic loading frequency of 0.4 Hz. The estimated lifetime of the bridge was

50 000 cycles, so that the bare net time for carrying out the dynamic tests

amounted to ca. 56 hours.

The main result finally was, that the test specimen, will say the ponton

(bay), as well as the test set-up itself passed the test procedure successfully.

Apart from some insignificant cracks in the welds on uncritical structural points

of the ponton, no serious damage could be observed on the test specimen. The

test set-up also showed a perfect performance with regard to function and

reliability.

As a final statement, it can be concluded, that the experimental realization

of this pretentious testing task on the field of pioneer bridge structures was an

complete success for EWK and OGI.


W. HARRE

140

Restoration of the sarcophagus of Duke Melchior von Hatzfeld

RESTORATION OF THE SARCOPHAGUS OF DUKE MELCHIOR VON HATZFELD – THE ACCOMPANYING SCIENTIFIC AND TECHNICAL INVESTIGATIONS

SCHADENSURSACHEN DES ZERFALLS DES HATZFELD-SARKOPHAGS UND ENTWICKLUNG EINER RESTAURIERUNGSMETHODE

RESTAURATION DU SARCOPHAGE DU DUC MELCHIOR VON HATZFELD - INVESTIGATIONS SCIENTIFIQUES ET TECHNIQUES

Gabriele Grassegger

SUMMARY

The article shows the investigations that led to the cause of decay of the

alabaster sarcophagus and new methods for the restoration of the resin

impregnated piece of art. The main reason was thermal decomposition of the

gypsum and rapid rehydration accompanied by mismatching properties of the

resin. The restoration used different formulas of cold-hardening PMMA resins

combined with fillers and special coatings for 5 different steps of structural

strengthening, adhesion, gluing of cracks, reshaping and retouching. The

restoration has been successfully completed by a team of restorers.

ZUSAMMENFASSUNG

Der Alabaster-Sarkophag des Grafen von Hatzfeld (in Laudenbach) zeigte

durch problematische Restaurierungen schwere Schäden, deren Ursachen

festgestellt wurden. Die Hauptprobleme waren Wasserabspaltungen und Zerfall

des Gipses, Spannungen durch Rückhydratisierung und andere physikalische

Eigenschaften des Harzes, das zur Tränkung verwendet worden war. Es wurden

dazu passende Restaurierungsmethoden entwickelt, die auf einer 5-stufigen

Behandlung mit kalterhärtenden PMMA-Harzen beruhen. Die Behandlungen

haben die Funktionen: strukturelle Festigung, Rißverklebung, Rißverfüllung und

Antragungen, um die alte Form der Ornamente wieder herstellen zu können. Das

Verfahren ist durch ein Restauratorenteam erfolgreich umgesetzt worden und

der Sarkophag konnte wieder aufgestellt werden.


G. GRASSEGGER

RESUME

Suite à une restauration problématique, le sarcophage en albâtre du duc

Melchior von Hatzfeld avait de sévères dégradations, dont les causes furent

déterminées. Les causes principales étaient la décomposition thermique du

plâtre, les tensions mécaniques dues à une réhydratation rapide, ainsi que les

propriétés physiques mal adaptées de la résine utilisée. Nous avons développé

une nouvelle méthode de restauration basée sur un traitement en cinq phases

avec des résines PMMA durcissant à froid. Le traitement avait pour buts: le

renforcement de la structure, le colmatage des fissures et le remodelage des

ornements. La restauration a été accomplie avec succès par une équipe de

restaurateurs, le sarcophage est à nouveau exposé.

KEYWORDS: Restoration, alabaster, sarcophagus, conservation

1. INTRODUCTION

The sarcophagus of the late Duke Melchior von Hatzfeld was created in

1659 by the famous stonemason Archilles Kern from Forchtenberg

(Unterfranken). In the year 1657 Melchior von Hatzfeld had been the liberator

of Krakau against the Swedish army sent by the German Emperor. The

sarcophagus was finely carved out of a famous alabaster coming close by

Forchtenberg. It shows the Duke in a suit of armour on the cover plate and

scenes of his battles on the sides. Because of his legacy Archilles Kern created

two tombs with very similar sarcophagus one in Prausnitz (Silesia) and one in

Laudenbach (Hohenlohe) in a little chapel in the mountains, called Bergkirche

(fig. 1).

142


Figure 1: The Hatzfeld Sarcophagus after the successful restoration and reconstruction in the Bergkirche chapel in Laudenbach (picture by Georg Schmid, restorer).


G. GRASSEGGER

Figure 2: Severe decay forms like warping, cracks and lamellar disintegration after the false restoration on one of the plates showing scenes from a battle

(picture by Georg Schmid, restorer).

144


2. THE CAUSES OF DESTRUCTION

In 1982–1984 the object, which had been restored several times, underwent

further restoration, this time by full impregnation with acrylic resin after

preliminary tests on a sample slab had proved successful (Fig. 1 and 2).

After thorough preliminary treatment (sealing of cracks, coatings, etc.), the

process comprised the following stages: drying at up to 100°C for several days,

vacuum treatment at up to 0.2 Torr/0.9 bar, flooding with PMMA monomer

solution at up to 20 bar to saturate the object, followed by hardening at a raised

temperature, i. e. at up to 80°C.

Immediately after treatment the sarcophagus showed good superficial

strengthening but damage ranging from warping to cracking was found as early

as September 1984 after the object had been mounted on an aerated-concrete

core and replaced in the Bergkirche church. By May 1985, the damage was

more apparent. Numerous cracks had appeared and a large proportion of the

joints had opened.

In October 1986 the State Office for the Preservation of Historic

Monuments in Stuttgart (LDA B.-W.) called us in to ascertain the causes of the

damage and to try to repair it. Numerous scientific and technical tests were

performed on the object to discover the cause of damage. Our main findings

were as follows (Grassegger, 1987):

"As a result of drying and the process of impregnation with acrylic resin, the

alabaster itself had partially dehydrated into semi-hydrate and anhydrite. This

was proven in numerous surface and deep-section samples by means of phase

analysis by x-ray diffraction. The degradation behaviour of gypsum had been

underestimated because the literature often states that “plaster burning” starts

at temperatures from 120°C. (In fact, water desorps in 2 steps and this

process starts from as low as 40°C.)

"Due to its heterogeneous structure, the object was very unevenly

impregnated, which gave rise to stresses. However, 1H NMR (1H [hydrogen]

nuclear magnetic resonance spectroscopy) measurements showed that the

sarcophagus itself had been impregnated through to the centre. The PMMA

(polymethyl methacrylate synthetic resin) content in the drilling core taken

from the dog was approx. 13% PMMA by weight on the surface, falling to


G. GRASSEGGER

approx. 5% PMMA in the centre (results gained by Günther Krause, Ref. 35

FMPA).

"Examination of the alabaster under a scanning electron microscope revealed

clear coating of synthetic resin on the gypsum crystals and only very small

vacuoles and air bubbles within the resin (Figs. 3 and 4).

"Selective moistening of samples proved the existence of residual swelling

stresses in the impregnated material, leading to further warping and crack

formation. In this material, water absorption was still approx. 0.5% by

weight, whereas in a pure PMMA sample it should be 0%.

"The thermal expansion αT of the resin treated material fluctuated markedly

and unsystematically between 1.7 and 3.1 10 -5m/mK and was non-linear. Its

break point likewise fluctuated widely between + 20°C and + 80°C. The

PMMA itself was 7.3 10-5 m/mK up to a break point of + 25°C. Above that, it

was 10.5 10-5 m/mK up to 60°C. This indicates that expansion is

heterogeneous and that the expansion properties of the alabaster and the

PMMA overlap in different ways. This leads to stresses when the material is

subjected thermal strain.

"According to the αT tests and determination of the glass transition point by

differential thermoanalysis (DTA), the PMMA’s glass transition point was

approx. 60°C.

"A deep-section drilling core with an apparently even density showed very

large variations in permeability to steam. On the surface was a dense zone

with a coefficient of resistance to steam diffusion of µ = 1,200, while deeper

down the values fluctuated between µ = 380 –2,100.

After damage had occurred, due to the contact with atmospheric humidity

and in particular to the high moisture level when the sarcophagus was mounted

on an aerated concrete pedestal in the very damp church, rehydration to gypsum

was very rapid and was accompanied by stresses in line with the heterogeneity

of material as described above. This led to severe deformation and in some areas

to cracks and to disintegration of the structure in the form of expansions or

swelling of the stone texture (see fig. 2).

Hence from the scientific and technical point of view there was a most

unfortunate combination of harmful factors that could neither have been

expected nor foreseen.

146


Fig. 3: Stalk-like gypsum structure of very fine-grained alabaster formation impregnated and coated with PMMA. The large bubbles (vacuoles, below) are occasional places where air was

trapped. Sample taken from the dog sculpture at a depth of approx. 10 cm (SEM picture).

Fig. 4: Coarse gypsum crystal (left) with synthetic resin coatings. The ring-shaped objects are cavities in the crystal which are lined with a film of resin. The flaky body on the right is

probably a newly developed anhydrite with a porous structure. Sample taken from a depth of approx. 10 cm (SEM picture).


G. GRASSEGGER

3. DEVELOPMENT OF RESTORATION METHODS

Several avenues had to be explored before a successful, practicable method

was found. The conservation testing process lasted from 1986 until into 1998.

Experiments aimed at removing the resin by means various solvents were

unsuccessful. Either hardly anything dissolved, or the process of dissolving

caused severe swelling of the entire structure, so this approach was rejected. It

also proved impossible to destroy the resin physically.

Subsequently, in the years up to 1996 various cold-hardening PMMA-

based artificial resins were tested. PMMA products made by a company in

Frankfurt proved to be the best base and had been tried out several times

previously. The products in question were 3 types of synthetic resin, as follows:

1) Finish X30 (or X40) PMMA resin, which hardens physically and is

dissolved in Xylene. This was used in different concentrations as:

"A stabiliser and strengthener in the form of impregnation

"As a preliminary treatment on powdery decay areas.

2) A PMMA-based synthetic resin adhesive that constitutes a reactive MMA

resin that interlaces into a PMMA resin when mixed with hardeners and

catalysts. It was used as:

"A binder for a mortar with adhesive properties and for new shaping

"A binder for materials for injecting into cracks (strongly adherent)

"An adhesive

3) Motema-WPC, a water-soluble acrylic suspension that can also be diluted

with water. This was used as:

"A base for retouching paint, for which it was blended with pigments.

"A binder for shaping and repair mortars

A team of restorers headed by Georg Schmid (of Messrs. Aedis, Stuttgart-

Möglingen) devised the recipes for the various materials based on the above

resins, optimised their properties for the purposes of application, and

coordinated colours. Our job was by conducting stress tests to identify the

versions that were the best technically, and the most stable (cf Table 1).

148


Table 1: Composition of the restoration materials employed (Restoration expert Georg Schmid and team, Möglingen).

Application Description/Recipe

Adhesive mortar, fine formula for repair and shaping

FM1= Acrylic resin suspension (Motema WPC) binder blended with Lenzin (natural gypsum) filler to a stiff, doughy consistency.

Injection mortar for adhesive and bridging of the cracks

I2= 1 part*) resin (Motema injection PMMA 220) + 1% by weight (in relation to the resin) peroxide catalyst plus 2 parts glass pellets <50µ.

Structural strengthening of the damaged alabaster regions and pre-treatment of sides of cracks

Impregnation with 5% PMMA solution Finish X40 (further diluted with Xylol to 5% resin content, results cf. Table 2).

Intermediate treatment (intermediate varnish) Impregnation with Finish X40, diluted to 5% resin content with ethyl acetate (to prevent previously applied coatings from dissolution.)

Final treatment, retouching paint. R1 = Retouching of repair mortars (pigments and Motema-WPC binder, i. e. water polymer coating, Pigments: Mixol: ground, natural standard pigment mixtures).

*) = parts by weight

Preliminary testing of the resins to establish their suitability

The synthetic resins were investigated in several tests to establish their

suitability and durability. This included artificial ageing simulations, tests to

establish their penetration into the alabaster – which in cracks included

measuring adhesion and determining their strengthening effect. All this was

done using various recipes. Here we show, by way of example, only the increase

in resistance to pressure, and adhesion (Tables 2 and 3).

Strengthening the alabaster

Strengthening was required because some of the alabaster had become

powdery in defective parts and on the surface. Powdery layers along the sides of

some cracks also needed to be stabilised before sealing. The gypsum was fully

impregnated, to saturation point in the case of samples 1 to 6, with solvents and

with Finish X 40 methacrylate solution to strengthen the structure. For

comparison, untreated, freshly quarried alabaster material (samples A–C) was

measured at the same time (Table 1).


G. GRASSEGGER

Table 2: Pressure resistance in various strengthened gypsum samples (block, dimensions approx. 5x5x2.5 cm, test following standard DIN EN 1926, test vertical to height).

Sample Treatment of sample Bulk density

[kg/dm³]

Breaking load

[N]

Compressive strength

[N/mm²]

1 strengthened 2.26 23480.00 17.78

2 strengthened 2.22 40750.00 30.57

3 strengthened 2.20 34910.00 25.25

4 strengthened 2.22 59850.00 46.61

5 strengthened 2.21 25630.00 19.18

6 strengthened 2.20 60050.00 44.96

Average 2.22 40778.33 30.72

A gypsum, freshly quarried 2.16 29870.00 19.41

B gypsum, freshly quarried 2.24 28560.00 20.41

C gypsum, freshly quarried 2.21 21120.00 17.88

Average 2.21 26516.67 19.23

Based on these findings, the result of strengthening could be rated very

good. There was a substantial increase in resistance to pressure, from 19 to 30

N/mm2 on average, equivalent to a rise of c. 50%. An even greater improvement

in strength was to be expected in the case of disintegrating gypsums like those in

the sarcophagus, since a larger quantity of saturating material could be absorbed

and residual strength had dropped to almost zero because of the destruction

process.

150


Figure 5: Before the restoration, small putto statue with most severe damage as warping,

cracks and almost complete disintegration (picture by Georg Schmid).

Figure 6: The same statue after restoration and treatment with 4 steps according to the

methods proposed (picture by Georg Schmid).


G. GRASSEGGER

Sealing the cracks in the alabaster

Numerous cracks in the alabaster and the open joints between the sections

had to be tied positively without visible changes. For this, original alabaster

material was glued together with various mixtures based on Motema 220 (Table

3).

Table 3: Tensile strength of gluing of two pieces of gypsum with various adhesives based on Motema 220 (measured in accordance with the DIN EN 12 372 standard).

Sample Breaking load,

total (N)

Tensile strength (N/mm2)

K1-1 gluing with filled resin*) 1531 0.61

K1-2 gluing with filled resin 1542 0.61

K1-3 gluing with filled resin 2044 0.82

K1-4 gluing with filled resin (premature failure due to crack in gypsum)

144 0.06

K2 PMMA resin + 1% hardener, unfilled 2049 0.82

K3 PMMA resin + 1% hardener, unfilled 822 0.33

Average 1355 0.54 *) comparable to Recipe I2 with the addition of 1% Aerosil (precipitated silicic acid) as a filler.

The findings showed that all tensile tests on glued samples (both filled and

unfilled adhesives) have a high level of tensile strength, higher than that of the

stone itself. This is shown by the path of the fracture in the stone itself, i.e. a so-

called cohesion fracture occurs in the stone.

UV resistance and ageing tests on the finished mixtures

For the sake of certainty, to check the durability of the finished mixtures

they were tested for UV resistance. The plan was to expose all recipes that might

be considered for use (20 in all) so as to rule out future changes.

A climate simulator of the global UV testing type, model UV 200 RB/20

DU, system Weiss, construction type BAM, was used. In this case, only UV

radiation in long-term climatic conditions corresponding to the room climate

was used. UV exposure took place in 2 cycles for a total of 300 hours. The

samples were inserted vertically and half of each was covered with opaque foil

(cf Figs. 4 and 5).

UV radiation was by means of fluorescent lamps that approximate the

short-wave part of sunlight. In particular, radiation simulates the high-energy

152


UV-A and UV-B rays (λ = 300–420 nm) that could trigger photo-oxidation. The

combination of fluorescent lamps employed corresponded to the spectral

distribution as per Method B of DIN 53 384 E.

Results of UV ageing

No kind of UV ageing or other damage was found to result from storage in

the room climate conditions and UV radiation. In this respect, the restoration

materials must be described as durable and stable.

CONCLUDING REMARKS

These extensive tests created excellent conditions for the tomb’s lasting

restoration. Skilful implementation by the team of restorers led by t Mr. Georg

Schmid/Stuttgart Möglingen reinstated the tomb to its former beauty (see Fig. 1

and 5).

By way of additional protection, the grave chapel containing the

sarcophagus is to be air-conditioned with the aim of avoiding alternating strains

in future. For the reasons stated at the beginning, a constant climate of approx.

10°C and a maximum of 50% relative humidity is to be aimed for.

ACKNOWLEDGEMENT

Thanks to the whole group of people who participated for the long period

of investigations and trials until the sarcophagus could be restored, especially to

Mr. Otto Wölbert and Mr. Meckes from the LDA who were the driving force of

the project and never gave up.

The whole history of the piece of art and it’s restoration will be published

in the upcoming issue of the “Nachrichtenblatt der Denkmalpflege in Baden-

Württemberg”, 4/2002 by a team of authors Otto Wölbert (restoration history),

Georg Schmid (restoration), Judith Breuer (art history), Robert Vix

(Architecture) and Gabriele Grassegger (technical investigations).

REFERENCES/INTERNAL REPORTS (SELECTION)

Grassegger, G. (1987): Hatzfeld-Grabmal, Bergkirche Laudenbach -

Untersuchung zur Schadensursache an einem Alabaster-Sarkophag nach einer

Kunstharz-Volltränkung, Nr. D3 140 008/GR (LDA internal report dated

18.9.1987)


G. GRASSEGGER

Grassegger, G. (2001): Restaurierung des Hatzfeld-Grabmals, Mechanische

Untersuchung von Festigungen und Probeklebungen auf Alabastergips“, Nr.

32-804073 (internal report of the Otto Graf Institute, Research and Testing

Establishment for Building and Construction [FMPA] dated 2.7.2001)

Grassegger, G. (2002): Restaurierung des Hatzfeld-Grabmals – Test der UV-

Alterungsbeständigkeit bei den fertigen Restaurierungsmateralien (Kittmörtel,

Klebungen, Injektagen und Retouchen), Berichtsnummer: 32 804 073 000-2

(FMPA internal report dated 3.5.2002).

154

Geotechnical Aspects and Observations of a Quarry Reclamation

GEOTECHNICAL ASPECTS AND OBSERVATIONS OF A QUARRY RECLAMATION

GEOTECHNISCHE ASPEKTE BEI DER WIEDERVERFÜLLUNG EINES STEINBRUCHS

ASPECTS GEOTECHNIQUES DU REMPLISSAGE D'UNE CARRIERE

Hermann Schad, Geoffrey Gay

SUMMARY

A disused quarry was refilled with mainly cohesive soil from excavations

from the local area. During the refilling slip movements took place. The

stabilisation methods used and the measurement and analysis of the movements

that took place during the filling are described.

ZUSAMMENFASSUNG

Ein ausgebeuteter Steinbruch sollte mit bindigem Material aus der

Umgebung – überwiegend Löss- und Verwitterungslehmen – verfüllt werden.

Bei der Verfüllung traten trotz der Stabilisierungsmaßnahmen

(Sandwichbauweise und Geokunststoffbewehrung) Rutschungen und größere

Bewegungen auf. Eine Ergänzung dieser Maßnahmen durch Betonscheiben und

einen Schotterfuß reduzierte die Verschiebungsgeschwindigkeit auf das bei

Erddeponien übliche Maß. Durch die Langzeitbeobachtungen wurde es möglich,

ein Kriechgesetz für die Bewegungen anzugeben.

RESUME

Une carrière désaffectée a été remplie avec des excavations de la région,

principalement des sols cohérents. Pendant le remplissage, des glissements ont

eu lieu. Les méthodes de stabilisation employées sont décrites, ainsi que les

mesures et l'analyse des déplacements qui ont eu lieu pendant le remplissage.

KEYWORDS: Limestone quarry, slip movement, stabilisation, refilling,

geotextiles


H. SCHAD, G. GAY

1. INTRODUCTION

In 1985 it was decided to refill and recultivate part of a limestone quarry in

the south west of Germany near Neuffen in the state of Baden-Württemberg.

The quarry had been used for the production of crushed limestone mainly for the

use in road construction. An aerial photograph of this stage of the refilling is

shown in fig. 1.

Fig. 1: Aerial photograph of first stage of refilling 24.4.1988

In 1989 it was decided to refill a further part of the quarry. It was soon

realised that the planned slope of 1:2 was not possible using conventional earth

works construction methods so an arched retaining wall was planned at the foot

of the slope and the fill was to be reinforced using geotextiles. The refilling was

to be carried out using mainly cohesive soil from excavations in the vicinity. A

cross-section of the planned refilling is shown in fig. 2. Later the slope was

flattened to 1:2.5.

156


Fig. 2: Cross-section of second stage of the refilling 1990

2 SLIP MOVEMENTS AND STABILISATION

In November 1997 it was noticed that relatively large slip movements must

have taken place because a natural stone wall surrounding a manhole had been

deformed considerably. (see fig. 3). The refilling was not complete at this time.

Fig. 3: Deformed natural stone wall


H. SCHAD, G. GAY

It was therefore decided to set up a grid of measuring points to observe the

movements of the slope. The grid points are shown in fig. 4.

Fig. 4: Plan of grid points used between 24.11.97 and 21.11.00

The first measurement took place in November 1997. After the first two

measurements with a weeks difference between them it appeared that the

deformation rate was slowing down. It had reduced from 9mm/d to 5mm/d. In

the third week however the speed increased to 20mm/d so it was decided to

increase the factor of safety by stabilising the foot of the slope using concrete

buttresses with crushed rock between them as shown in figs. 5 and 6.

158


Fig. 5: Section through concrete buttress

Fig. 6: Plan of concrete buttresses with crushed rock filling


H. SCHAD, G. GAY

The concrete buttresses were used as a supporting element and the crushed

rock as a drainage. Grid point measurements on the 23.12.97 showed a further

increase in the deformation rate (37mm/d). It was therefore decided to fill the

volume between the concrete buttresses with a well graded crushed rock instead

of the coarse crushed rock as planned. The deformation rate slowed down

considerably (see Section 3). At first it was not clear whether this was due to a

frost period between 21.1.98 and 4.2.98.

A slope stability calculation showed that for a factor of safety of 1.0 the

shear parameters in the horizontal direction have to be ϕ = 10.21° and c = 0

kN/m² and in the vertical direction ϕ = 20 ° and c = 0 kN/m² (see fig. 7).

Fig. 7: Elements for slope stability calculations

Calculations with the Kinematical Elements Methods (Gussmann et al.

2002) for the stabilised state showed that the increase in stability factor due to

the concrete buttresses and the lowering of the water table by 2m was relatively

small (0.04). In the long term however an increase in the shear strength due to

consolidation and “age hardening” is to be reckoned with. Under the phenomena

“age hardening” is understood that when a cohesive soil is placed, especially in

wet weather, there is a relative large amount of water between the soil

aggregates and the soil is very soft or even “liquid”. In the course of time this

160


free water is absorbed by soil aggregates. This reduction in free water leads to

an increase in strength.

3 RESULTS OF DEFORMATION MEASUREMENTS

The deformation measurements are divided into two phases:

"Phase 1 between 24.11.97 and 22.11.00

In this phase the 19 grid points were placed. Some of these were damaged

during the earthworks and had to be replaced.

"Phase 2 between 22.11.00 and 4.4.00

After the earth works were completed new grid points were put in place

(see fig. 8).

Fig. 8: New grid points after 21.11.00


H. SCHAD, G. GAY

The decisive deformation kinematics as derived from the deformation

measurements between 24.11.97 and 8.2.98 are shown in fig. 4. It can be seen

that the part which is reinforced with geotextiles moved horizontally as a block.

The displacement rates of the points 16 and 17 (fig. 4) are characteristic for

the lower part of the slope. The average deformations and the resulting

deformation rates are shown in the following diagram.

Fig. 9: Displacements of the lower part during the first 72 days

After construction was completed a horizontal deformation of 16mm in the

course of 1.5 years was measured at grid point 5 (see fig. 8). The vertical

deformations of this point during the same time interval were 36mm. The

maximum settlement measured on the “plateau” was 48mm. At the foot of the

slope the maximum deformations at grid point 10 were horizontal 14mm and

vertical 3mm. Of special note is the similarity of the settlements of the grid

points 1 to 5 on the “plateau”. Inside 1.5 years (21.11.00 to 4.4.02) they were

40mm, 48mm, 39mm, 36mm and 35mm as shown in fig. 10.

162


Fig. 10: Displacements of the plateau in the second phase

4 INTERPRETATION OF THE MEASUREMENTS

In the following diagram (fig. 11) the average time deformation curves

with the time on a logarithmic scale are plotted. It can be seen that up to 68 days

after the start of the measurements slipping took place. After that the creep

phase started.

Fig. 11: Displacements as function of time


H. SCHAD, G. GAY

In the semi-logarithmic plot the creep phases approximate to straight lines

which can be represented to a good approximation by the black lines with circles

as points. Using logarithms to the base 10 the following relationships are

obtained. This logarithmic creep is often observed in soil but a lot of other

rheological models could be used (e. g. Schad/Breinlinger 1991).

In the next 10 years horizontal deformations of 5 to 10 cm. are to be

expected and that similar deformations are to be expected in the next 90 years.

5 LITERATURE

Gussmann, P.; Schad, H.; Smith, I. (2002): Numerical Methods in Geotechnical

Engineering Handbook, Vol. 1, Ernst & Sohn Berlin, 437 - 479

Schad, H.; Breinlinger, F. (1991): Numerical analysis of visco-elastoplastic soil

behaviour considering large deformations. Proc. 10th European Conference

on Soil Mechanics and Foundation Engineering, Florence/Italy, 255 - 260.

164

Non-destructive detection of longitudinal cracks in glulam beams

NON-DESTRUCTIVE DETECTION OF LONGITUDINAL CRACKS IN GLULAM BEAMS

ZERSTÖRUNGSFREIE MESSUNG VON LÄNGSRISSEN IN BRETTSCHICHTHOLZ-TRÄGERN

DÉTECTION NON DESTRUCTIVE DES FISSURES LONGITUDINALES DANS LES POUTRES DE BOIS LAMELLÉ COLLÉ

Simon Aicher, Gerhard Dill-Langer, Thomas Ringger

SUMMARY

The paper reports on the detection and length characterisation of

longitudinal cracks in glued laminated timber (glulam) beams by means of

ultrasound (US) pulse transmission method. In the preliminary study one large

glulam beam with a crack starting at one end-grain face and ending at about one

third of total beam length has been evaluated. For the transmission

measurements US pulses have been applied to the narrow faces of the beam,

thus propagating parallel to cross-sectional depth perpendicular to fibre. The

beam has been scanned by a transmitter / receiver pair of US transducers shifted

along the longitudinal beam axis. The recorded full US wave signals were

evaluated for three different scalar parameters being “time of flight”, “peak-to-

peak amplitude” and “first amplitude”. The comparison of the visual inspection

with the US parameters, showing significantly different scatter ranges, yielded a

satisfactory agreement with respect to the determination of crack length. The

NDT crack detection based on the parameter “time of flight” was also

satisfactory when the crack extended only over a part of the beam width, i. e. not

being visually detectable from one of both side faces. The latter can be very

important for in-situ inspection of beams in buildings with assumed or partial

cracks.


S. AICHER, G. DILL-LANGER, T. RINGGER

ZUSAMMENFASSUNG

Der Aufsatz berichtet über den Nachweis und die Längenmessung von

longitudinalen Rissen in Brettschichtholz (BSH) -Trägern mittels der Methode

der Durchschallung mit Ultraschallpulsen. In der orientierenden Studie wurde

ein großer Brettschichtholzträger mit einem Riss untersucht, wobei der Riss von

einer Hirnholzfläche ausging und innerhalb des ersten Drittels der Trägerlänge

endete. Für die Transmissionsmessungen wurden Ultraschallpulse auf die

Schmalseiten der Träger aufgebracht, so dass diese sich parallel zur

Querschnittshöhe und damit rechtwinklig zur Faserrichtung ausbreiteten. Der

Träger wurde mit einem Ultraschall Geber / Empfänger-Paar in Richtung der

Trägerachse abgerastert. Die aufgezeichneten vollständigen Ultraschallsignale

wurden hinsichtlich dreier verschiedener skalarer Parameter ausgewertet:

Signal-Laufzeit, Signalstärke und erste Amplitude. Der Vergleich der visuellen

Charakterisierung mit den Ultraschallparametern, die jeweils deutlich

unterschiedliche Streubreiten aufwiesen, ergab eine zufriedenstellende

Übereinstimmung hinsichtlich der Bestimmung der Risslänge. Die

zerstörungsfreie Risserkennung auf Grundlage des Parameters "Signal-Laufzeit"

war auch dort noch zufriedenstellend, wo sich der Riss nur über einen Teil der

Querschnittsbreite erstreckte, also auf einer der beiden Seitenflächen schon nicht

mehr sichtbar war. Letzteres kann für die in-situ Beurteilung von Trägern in

realen Bauwerken mit vermuteten oder teilweise vorhandenen Rissen sehr

wichtig sein.

RÉSUMÉ

L’article traite de la détection et de la caractérisation des fissures

longitudinales dans les poutres en bois lamellé collé, au moyen d’une méthode

de transmission des impulsions ultrasons. Lors de travaux préliminaires, une

poutre présentant une fissure commençant à l’une des extrémités et s’étendant

jusqu’au tiers de la longueur totale a été étudiée. Les impulsions d’ultrasons sont

appliquées sur les deux faces les plus étroites de la poutre et se propagent

parallèlement à sa section, c’est à dire perpendiculairement aux fibres du bois.

Les capteurs ultrasons (émetteur et transmetteur) balaient alors la poutre suivant

son axe longitudinal. Le signal enregistré fournit trois paramètres différents: le

temps de propagation du signal, son amplitude pic à pic et sa première

amplitude. La comparaison entre les indications obtenues par la méthode des

166


ultrasons et l’évaluation visuelle sont en accord quant à la détermination de la

longueur de la fissure. La détection basée sur le seul paramètre « durée de

propagation » est également satisfaisante lorsque la fissure ne s’étend que sur

une partie de l’épaisseur, c’est à dire lorsqu’elle ne traverse pas la poutre de part

en part. Ce dernier cas est très intéressant pour l’inspection in-situ des

constructions présentant des fissures ou des risques de fissure.

KEYWORDS: non-destructive testing, ultrasound, pulse transmission, crack in

glulam beams, scalar ultrasound parameters

1. INTRODUCTION

Non-destructive evaluation of the state of integrity resp. of defects or

partial damages in structural building elements generally represents an important

issue. The capability of NDT based reliable assessment of components enhances

the acceptance of building systems or materials, may affect safety factors and

enables assessment of upgrading or rehabilitation works. Timber and glulam

beams despite all positive aspects are prone to longitudinal cracks generally

resulting from interaction of poor constructive detailing and unaccounted

climatic stresses. Longitudinal cracks primarily occur at i) support areas due to

interaction of shear stresses and climate and ii) at notches, holes and in apex

areas of curved / tapered beams due to tension stresses perpendicular to grain

bound to undue load actions and / or often climate stresses. Finally,

iii) longitudinal cracks can occur from poor glue lines generally bound to

trespass of open / closed curing times of the adhesives.

The reliable assessment of the extent of damage and of the result of the

repair works represent the two equally important aspects of the NDT assessment

of damaged or upgraded construction elements. In many cases a visual

inspection of the beams is very costly or almost impossible. Today reliable NDT

tools for employment in the sketched area are missing for lumber / glulam

beams being contrary to constructions with some other important building

materials. The reason for this NDT lag consists i.a. in the anisotropy,

inhomogeneity and the high damping characteristics of the natural material

wood.



Timber Department of Otto-Graf-Institute being deeply involved in the

assessment / expertises of damages, repair proposals and technical approval of

rehabilitation works has started to focus on active NDT methods about a year

ago. This paper gives some preliminary results of one of the on-going projects.

It is reported on the detection and length characterisation of longitudinal cracks

in glued laminated timber (glulam) by means of ultrasound (US) pulse

transmission method. The US method has been chosen due to its sensitivity to

impedance changes at discrete boundaries. Former literature known attempts in

this field [KLINGSCH, 1991; KIMURA ET AL 1991] dealt with artificial defects,

being rather thick saw cuts of defined length parallel to beam axis and over full

cross-sectional width. The results and the practical relevance of the exclusive

focus on such slots / cracks has been discussed controversially in the involved

engineering community. In the investigation reported here a fully practice

relevant crack resp. cracked beam was investigated.

2. EXPERIMENTAL SET-UP

The investigated specimen represented a part cut from a large beam with a

round hole loaded until failure in bending, compare Fig. 1a. The beam had failed

with two large cracks initiated at the hole periphery by high local stresses

perpendicular to grain. The crack propagation was then driven by both, shear

and tension stresses perpendicular to grain.

The investigated NDT specimen (Fig. 1b) showed an open crack over full

width b at end grain face Y closer to the former hole location and no visible

crack at the opposite end grain face Z. Thus, despite disputable accuracy of

visual inspection the crack must end within the specimen as the latter consisted

of one massive piece.

168


a)

l = 2100

p/2 p/2

Lc

h = 440

T

B

Y

Z

cracks atultimate load

900

120

investigatedNDT specimen

originallyted beamtes

b)

l = 2100

b = 120

h =

44

0

Top narrow face (T)

Bottom narrow face (B)

visible part of the crack

end grainface Y

end grainface Z

"right" wide face II

10

0

33

33

xy

Fig. 1a,b: Original location and dimensions of the employed NDT specimen. a) larger cracked beam from which the NDT specimen was cut out after failure of the beam b) view and dimensions of partially cracked NDT specimen



The dimensions of the block were (width b × depth h × length l): 120 mm ×

440 mm × 2100 mm. Starting at end-grain face Y and proceeding in the

longitudinal (x-) beam direction, the crack is visible at both side-faces I and II

for the major part of the crack length (compare chap. 4).

The ultrasonic NDT evaluation / assessment of the crack (length) was

throughout performed by means of a pair of piezoelectric (US) transducers.

Transmitter and receiver were positioned oppositely at mid-width of the narrow

faces T and B of the beam specimen and aligned parallel to depth. Starting at the

end-grain face Y with the opened crack the transducer pair was moved along

beam length l with increments of

x = 50 mm towards the end grain face Z. At

each position an ultrasonic pulse synthesized by a generator unit was put to the

specimen by the piezoelectric transmitter. Figure 2 shows a schematic

representation of the experimental set-up. The fixation of the transmitter and of

the receiver differed. The transmitter was throughout fixed to the surface by a

hot melt adhesive also serving as coupling agent. Contrary, the receiver was not

glued to but applied to the surface by hand pressure without using any kind of

coupling agent.

At each location x a number of 25 repetitive measurements were performed

in order to enable noise reduction. As the crack was not centred in the middle of

the cross-sectional depth but much closer (~70 mm) to narrow face B it was

questioned whether there might be an influence if the transmitter is at a closer or

more remote distance to the crack. Therefore two test series S1 and S2 were

performed with the transmitter first being at narrow side T and then at narrow

side B.

170


a)

crack at endgrain face A

ultrasoundtransmitter

ultrasoundreceiver

US-pulsegenerator unit

amplifier

narrowface T

narrowface B

-3

0

3

0.0 0.5time [ms]

signal [V]

b)

narrow face T

end grainface Z

narrow face B

end grainface Y

crack

x

l = 2100

Fig. 2a,b: Schematic representation of the experimental set-up



3. NDT EQUIPMENT

The generator unit (USG 20, Geotron Electronics), originally optimised for

NDT of concrete, produced high voltage pulses with main frequencies between

20 kHz and 350 kHz. The duration of a single pulse was less than 1 ms.

The ultrasonic transducers (UPG-D 3037, UPE-D 3038, Geotron

Electronics) used in the experiments were piezoelectric converters with a

coupling-surface of 3 mm in diameter. The transducers showed a multi-resonant

frequency characteristic with main peak values between 20 and 100 kHz.

The received ultrasonic pulses have been amplified by a broadband

amplifier (AM 502, Tektronix) with an amplification factor of 100 dB. The

complete signals were recorded by a PC based transient recorder with 12 bit

amplitude resolution and 20 MHz time resolution.

4. VISUAL CHARACTERIZATION OF THE CRACK DIMENSIONS

For correlation of the ultrasound NDT parameters with the length of the

crack in longitudinal beam direction and with the crack opening, the dimensions

of the crack were determined by visual inspection at both wide side faces I and

II of the specimen. The crack openings were measured with a feeler gauge.

Figures 3 a and 3 c give a schematic illustration of shape, position and

dimensions of the crack according to the visual characterization and feeler gauge

measurements at the two wide faces I and II, while Fig. 3 b shows a top view

indicating the projected crack area. According to the visual findings the crack

can be divided into three different sections.

In section A (0 ≤ x ≤ 53 cm), the crack is characterized by measurable

openings of 1.2 mm (x = 0) to 0.4 mm (x = 53 cm) at side face I; at side face II

the respective dimensions are: 0.4 mm (x = 0) to 0.05 mm (x = 53 cm).

In section B (53 cm < x ≤ 67 cm) a crack-opening was only measurable at side

face I with openings in the range of 0.35 mm to 0.25 mm. At side face II, the

closed crack was visible as a small displacement edge within the surface.

Finally in section C (67 cm < x < 88.4 cm), the crack was still measurable

at side face I with openings from 0.25 mm to 0.05 mm. The end of the crack at

x = LC =88.4 cm almost coincides for measurable (0.05 mm) crack opening and

visual inspection. At side face II, the crack is not visible at all.

172


The true extension and shape of the crack front might well be somewhat

ahead of x = LC what will be determined at the end of the ongoing experiments.

5. CHARACTERIZATION OF SIGNAL-PARAMETERS

Once an ultrasonic pulse is generated and applied to the narrow face of the

glulam beam, it is proceeding through the specimen perpendicular to the

direction of the glued lamellas, i.e. perpendicular to fibre direction and is

detected by the receiver at the opposite surface.

The recorded full wave signals purged from noise by multiple pulse

measuring method were so far evaluated for three different scalar parameters,

being:

• “Peak-to-peak amplitude” (pp amplitude) of the signal, which represents

the difference between the recorded absolute maximum and minimum of

the complete signal. The parameter is correlated to the transmitted energy

of the pulse. Figure 4 shows one exemplary wave signal, including the

determination of the pp-amplitude.

• “Time of flight” (TOF) of the signal is defined as the time lag between the

external trigger edge given by the pulse generator and the on-set, i.e. the

begin of the recorded signal. The signal-parameter TOF and also the

below specified parameter |1st a| are exemplarily depicted in Fig. 5 for the

signal given in Fig. 4 now presented with a close-up at the begin of the

signal.

• “First amplitude” (1st a) of the signal is defined as the maximum (or

minimum) amplitude of the first observable half cycle. In detail, for signal

characterization, the absolute value of the first amplitude has been used.



a)

0

20

40

0 70 140 210x [cm]

A B Ch [cm]

left face I

LC

beam

depth

h =

440 m

m

b)

0

20

40

0 70 140 210

end of measurable crack(openings > 0.05 mm)x

53.0

cm

67.0

cm

88.4

cm

x [cm]

A B C “left“ face I

“right“ face IIface

Y

face Z

exact shape of crack front unknown

end of visible crackbeam

wid

th

b =

120 m

m

c)

0

20

40

0 70 140 210x [cm]

A B Ch [cm]

right face II

LC

beam

depth

h =

440 m

m

Fig. 3a-c: Schematic illustration of the appearance of the crack at different faces of the specimen. The graphs 3a) and 3c) give measured crack lengths and crack openings (100-times enlarged) at the left and right wide side faces (I and II). Fig 3b shows a projection of the crack area revealing the three crack sections A-C.

174


-3.0

-1.5

0.0

1.5

3.0

0.00 0.20 0.40 0.60 0.80 1.00

time [ms]

signal [V]

pp a

mplit

ude

close upin Fig. 5

Fig. 4: Recorded signal with evaluation / definition of the “peak-to-peak amplitude” (pp amplitude)

-0.4

-0.2

0.0

0.2

0.4

0.00 0.05 0.10 0.15 0.20 0.25

time [ms]

signal [V]

1st a

TOF

Fig. 5 Evaluation / definition of “time of flight” (TOF) and of “first amplitude” (1st a); the graph represents a close-up of the recorded ultrasound pulse given in Fig. 4



6. RESULTS OF THE ULTRASOUND MEASUREMENTS

The reproducibility of the signal parameters for repeated independent

measurements at a specific location x (uncoupling and new coupling of the

transducers for each measurement) differed considerably between parameter

TOF on the one side and parameters pp and |1st a| on the other side. In case of

TOF in average an extremely small coefficient of variation (C.O.V.) of 0.3 %

was obtained whereas for the parameters pp and |1st a| the considerably higher

C.O.V.’s were 21% and 24 %, respectively. For the reproducibility test a number

of 10 repeated measurements have been evaluated.

The major results of the performed preliminary investigations are compiled

in Figs. 6 to 8, showing the signal parameters TOF, pp and |1st a| along specimen

axis x. In all graphs the results of the two test runs S1 and S2 with the alternative

transmitter positions at narrow specimen sides T or B are given, and the mean

value of both test runs is shown additionally. Further, the quality of the signal

parameter reproducibility is indicated in all Figures by an error bar with a height

of 2 times of the respective standard deviation (the error bars are not visible in

Fig. 6 due to the very small C.O.V.’s). The visually determined crack length

segments A, B and C are indicated in the graphs, too. Following the results are

discussed in more detail.

Figure 6 specifying “time of flight (TOF)” vs. beam axis x reveals almost

throughout a steep decrease of parameter TOF along crack length segments A, B

and C. It should be emphasized, that the TOF decrease in the investigated case is

apparently not affected by the fact that the crack is not visually detectable at

surface II in crack zone C. For positions x > LC a rather constant TOF value of

253.2 ms is obtained. This gives a mean phase velocity in transverse direction to

fibre of v90 = 0.44 / (253.2*10-6) = 1738 m/s which is in good agreement with

literature data [Buchur 1989] on phase velocities perpendicular to fibre of wood

/ glulam made of European spruce. A comparison of test series S1 and S2

indicates apart from one exception in the crack range A, that the TOF results are

obviously not influenced by transmitter location closer resp. more far from the

crack plane.

176


240

270

300

330

360

0 30 60 90 120 150 180 210

S1: transmitter at face T

S2: transmitter at face B

253.2

TOF [µs]

specimen axis x [cm]

visually determined crack length, wide side face I

visual crack length, wide side face II

TOF mean value ofuncracked specimen

A B C

LC

mean valueof S1 and S2

Fig. 6 Results of time of flight (TOF) measurements along the beam axis x. The x-axis gives the distance x [cm] of the transducers position to the end grain face Y. The crack was supposed, according to visual inspection, to end at x = 88.4 cm.

Figure 7 shows the “peak-to-peak amplitude” (pp amplitude) of the

transmitted signals. Qualitatively a slow increase of pp-amplitude values from

mid- length of segment A through to C and the increase continues to about 20

cm beyond C; into the visually uncracked part of the beam. Quantitatively high

scatter of the measured data, especially in the uncracked section is observed.

Exemplarily at a distance x = 160 cm from end grain face Y, certainly well

ahead of the crack front, the pp-amplitudes exhibit a local minimum with values

comparable to those measured within the crack at x = 70 cm (in section C). The

transition from the cracked to the undamaged section is rather smooth without a

clearly marked step in the pp amplitude course.



0.0

3.0

6.0

9.0

12.0

15.0

0 30 60 90 120 150 180 210



pp amplitude [V]

vis. crack lengthwide side face I

vis. crack lengthwide side face II

A B C


LC


Fig. 7 Results of peak-to-peak amplitude (pp amplitude) measurements along beam axis x. The x-axis gives the distance x [cm] of the transducers position to the end grain face Y. The crack was supposed according to visual inspection to end at x = 88.4 cm.

Thus, the signal-parameter pp-amplitude does not allow a clear

identification of the crack length. The relatively high uncertainty due to coupling

conditions makes it even more difficult to quantitatively estimate the location of

the crack tip. However, in spite of the scatter, the clearly visible trend of

decreased attenuation for decreasing crack openings is not affected qualitatively.

The presented results for the behaviour of the peak-to-peak amplitude in

case of cracks can be compared to the observations of [KLINGSCH, 1991], where

no damping of pp amplitudes has been measured in the case of saw-cuts.

In Fig. 8 the results of |1st a| along beam axis x are shown for the two

performed test series S1 and S2 together with the boundaries of the different

crack sections.

178


0.00

0.20

0.40

0.60

0.80

0 30 60 90 120 150 180 210



| 1st a | [V]

vis. crack lengthwide side face I

vis. crack lengthwide side face II

A B C


LC


Fig. 8 Results of measurements of absolute values of the first amplitude (|1st a|) along the beam axis x. The x-axis gives the distance x [cm] of the transducers position to the end grain face Y. The crack was supposed according. to visual inspection to end at x = 88.4 cm.

The measured course of the |1st a| values can roughly be described as a step

function with quite constant low values within all three sections A to C of the

crack and a sharp increase at the assumed crack tip. It is interesting to note that

in section C still strong damping of |1st a| is observed, while the crack is solely

visible at one side face of the specimen.

Although the scatter among |1st a| values within the undamaged part of the

beam is significant, the results between cracked and uncracked parts of the beam

are clearly separated, which is especially true for the mean values of the two test

series with interchanged transmitter / receiver conditions.



7. CONCLUSIONS

The performed preliminary study on the feasibility of crack detection in

glulam beams by means of ultrasonic pulse transmission method revealed

promising results.

All three evaluated scalar signal parameters, being time-of-flight (TOF),

peak-to-peak-amplitude (pp-amplitude) and first amplitude (1st a) showed

significant correlations with the occurrence of the completely or partly visually

detectable crack.

Both, pp-amplitude and 1st a exhibited relatively high scatter among the

values within the undamaged part of the beam accompanied by a quite poor

reproducibility bound to the performed non-ideal coupling conditions. The

pp-amplitude values showed a smooth transition zone from cracked to the

uncracked sections of the beam. Contrary hereto the 1st a values exhibited a

pronounced step indicating the end of the crack clearly.

The TOF-results showed best reproducibility and a clear, although smooth

change at the end of the crack. The different crack sections with one-sided

throughout measurable crack openings respectively one sided first measurable

and then visible crack opening were best represented by the course of TOF

results.

For all three characteristic signal parameters, no significant differences due

to interchanged positions of transmitter and receiver were observed. Thus, no

detection of the location of the crack with respect to depth direction could be

performed, being in good accordance with the results of [KLINGSCH, 1989,

1991].

Although feasibility of the applied NDT methods and evaluation for crack

detection could be shown by the presented study, the results from only one

exemplary specimen may not be generalized. Additional tests also with

investigate different beam / crack configurations have to be performed to obtain

a statistically more reliable data basis.

In order to improve the presented ultrasonic method for applications in real

structures, the coupling problem has to be solved and the feasibility for beams

with realistic heights of about 1 to 1,5 m has to be shown. Advanced signal

processing techniques for the evaluation in the frequency domain (i.e. Fourier-

180


and Wavelet transforms) should be used for noise reduction, enhancement of

resolution and defect sensitivity.

ACKNOWLEDGEMENTS

The authors are very much indebted to Dr. Catherine Lidin (Collano AG,

Switzerland) for the utmost valuable favour translating the abstract and title of

this paper into technically and linguistically correct French.

The financial support of German Science Community (DFG) via grant to

Sonderforschungsbereich 381 "Characterisation of damage evolution in

composite materials using non-destructive test methods" and hereby to sub-

project A8 "Damage and NDT of the natural fibre composite material wood" is

gratefully acknowledged.

REFERENCES

KLINGSCH, W.: Zerstörungsfreie Lokalisierung äußerlich nicht sichtbarer

Holzschädigung. Bauen mit Holz 6, 1989, pp. 421-423

KLINGSCH, W.: Erarbeitung anwendungstechnischer Grundlagen zur

zerstörungsfreien Qualitätsüberwachung von Holzleimbauteilen mittels

Ultraschall. Forschungsbericht, 1991

BUCUR, V.: Acoustics of wood. Boca Raton, New York, London, Tokyo, 1995,

p. 121

KIMURA, M., KUSUNOKI, T., OHTA, M., HATANAKA, K., KOZUKA, H., ITO, H.:

Ultrasonic pulse test on glulam glued connection. Proc. Int. Timber Eng.

Conf., part 2, London, 1991, pp. 2.250 – 2.257



182

Determination of local and global modulus of elasticity in wooden boards

DETERMINATION OF LOCAL AND GLOBAL MODULUS OF ELAS-TICITY IN WOODEN BOARDS

BESTIMMUNG DES LOKALEN UND GLOBALEN ELASTIZITÄTS-MODUL IN HOLZBRETTERN

DETERMINATION DU MODULE D’ELASTICITE LOCAL ET GLO-BAL SUR DES PANNEAUX A BASE DE BOIS

Simon Aicher, Lilian Höfflin, Wolfgang Behrens

SUMMARY

The paper reports on an efficient method for determination of the local

modulus of elasticity by means of elongation/strain measurements. Further, the

effect of local weak sections on the global modulus of elasticity determined by

deflection measurement is revealed. The global modulus of elasticity computed

on the basis of the partly extremely varying locally measured moduli of elastic-

ity complies well with the globally measured MOE.

The experimental investigations were performed with edgewise bent beech

boards. First, the elongation /strain measurement method was verified exem-

plary with a board which was inflicted successively with artificial defects

(holes). For each defect state the local and global moduli of elasticity were

measured and the differences are discussed. Second, the variation of local

modulus of elasticity and its high spatial correlation with the location of bending

failure is shown exemplarily by means of four beech boards of a larger test se-

ries.

ZUSAMMENFASSUNG

Es wird über eine effiziente Methode zur Bestimmung des lokalen Elastizi-

tätsmoduls mittels Längsverschiebungs-/Dehnungsmessungen berichtet. Des-

weiteren wird die Auswirkung lokaler Schwachstellen auf den mittels Durchbie-

gungsmessung bestimmten globalen Elastizitätsmodul gezeigt. Der aus den teil-

weise extrem variierenden lokal gemessenen Elastizitätsmoduln berechnete glo-

bale Elastizitätsmodul stimmt sehr gut mit dem gemessenen globalen Elastizi-

tätsmodul überein.


S. AICHER, L. HÖFFLIN, W. BEHRENS

Die experimentellen Untersuchungen wurden mit hochkant biegebean-

spruchten Buchebrettern durchgeführt. Zuerst wurde die Längsverschiebungs-

/Dehnungsmeßmethode exemplarisch an einem Brett verifiziert, welches stu-

fenweise mit künstlichen Defekten (Löchern) versehen wurde. Für jeden De-

fektzustand wurden die lokalen und globalen Elastizitätsmoduln gemessen. In

einem zweiten Schritt wird die Änderung des lokalen Elastizitätsmoduls und

dessen hohe Korrelation mit dem Ort des Biegeversagens exemplarisch an vier

Brettern einer größeren Versuchsreihe aufgezeigt.

RESUME

Cet article présente une méthode efficace permettant de déterminer le mo-

dule d’élasticité local par une mesure couplée déplacement/déformation. D’autre

part, on fait apparaître l’effet des sections localement faibles sur le module

d’élasticité global déterminé par la flèche. Le module d’élasticité global obtenu

par intégration des modules locaux mesurés, extrêmement variables, est en bon

accord avec le module global mesuré.

L’étude expérimentale a porté sur des panneaux fléchis à chant. La mé-

thode couplée déplacement/déformation a été préalablement vérifiée sur un pan-

neau présentant des défauts artificiels (trous). Pour chaque défaut, on détermine

les modules local et global, et les différences sont discutées. Par la suite, la va-

riation du module local et sa forte corrélation spatiale avec la résistance en

flexion est mise en évidence sur 4 panneaux de hêtre extraits d’une campagne

expérimentale plus importante

KEYWORDS: local modulus of elasticity, global modulus of elasticity, stiff-

ness variation, artificial defects, weak sections

1 INTRODUCTION

It is reported on a method for determination of the local modulus of elastic-

ity (MOE) in bending tests with timber beams and respective results. In hetero-

geneous materials such as wood the modulus of elasticity can vary strongly

along the length of the boards. Based on a positive stiffness - bending strength

correlation, the footprints of locally low MOE values determine the strength

class (or grade) of boards in grading machines based on the bending principle.

Local MOE obviously depends strongly on the length of the board segment used

184


for the MOE determination which in most cases is larger than a local weak area,

mostly created by a knot or by sloping grain.

A local MOE determined over a board segment length of 5 times the cross-

sectional depth as specified in EN 408 still represents an integral (constant)

value over a considerable length and there may be strong local MOE deviations

within that length. The stated averaging effect of concentrated zones of low

MOE areas occurs in all bending type grading machines which bend at consecu-

tive locations, as there are practical limits for the length of the span. This is the

major reason for the moderate coefficient of correlation between bending

strength and MOE. Interesting approaches on how to reconstruct the variation of

the true MOE function from MOE data collected from a consecutively bent

board in order to derive the true local MOEs based on Fourier transforms were

proposed by Bechtel (1985) and Foschi (1987).

Apart from strength grading the knowledge of the actual local MOE and of

the associated local strength variation along the length of boards is very impor-

tant for (stochastic) modelling of boards and glulam subdivided in unit cells of

small length, i.a. Foschi and Barrett (1980), Ehlbeck et al. (1985), Isaksson

(1999) and Serrano (2001). Hereby the length of the unit cell has a considerable

modelling influence on load sharing in adjacent glulam lamellas.

For modelling and calibrating the MOE variation along board length sev-

eral approaches are known (i.a. Foschi and Barrett (1981), Ehlbeck et al. (1985),

Kline at al. (1986) and Taylor (1991)). All models are based on a calibration vs.

global (and partly local) modulus of elasticity necessitating extensive empiric

data and leaving model dependent considerable uncertainties.

The experimental determination of local MOEs which at first view seems

to be a very simple task is demanding in case a bending method is applied and

has limits concerning the smallness of the segment length. Reliable results be-

low span to depth ratios of about 3 are questionable; limits were revealed by

Kaas (1975) employing the so-termed “middle ordinate method”. The method is

based on the assumption that short segments of a bent board approximate arcs of

circles with varying radii.

The work reported here was conducted in the frame of establishing a realis-

tic empirical data basis for the variation of bending MOE and bending strength

values along the length of beech wood boards bent about the major axis. It was



intended to measure the local bending MOE over distances or unit cell lengths in

the range of 1 to 2 times of the depth of the boards.

The paper shows a method for determination of the local modulus of elas-

ticity by means of displacement/strain measurements and reveals the effect of

local weak sections in comparison to the global modulus of elasticity determined

by deflection measurement. The experimental verification of the method which

in principle is independent of specific materials is performed with edgewise

loaded beech boards. First, the displacement/strain measurement method is veri-

fied exemplary with a board which was inflicted successively with artificial de-

fects. For each defect state the local and global modulus of elasticity were meas-

ured and the differences are discussed. Finally, the variation of local modulus of

elasticity and its local correlation with the position of failure is shown exempla-

rily on four beech boards of a larger test series.

2 GLOBAL AND LOCAL METHOD FOR DETERMINATION OF MODULUS OF ELASTICITY IN BENDING

It is important to note that the terms “local MOE” and “global MOE” here

are defined different as in European standard EN 408. In the mentioned standard

the so-termed local MOE is determined in a 4point bending test with loads in the

3rd points via deflection measurement within the constant moment length of 6

times the depth h of the beam. The deflection w1 is actually determined over a

length of `m = 5 h (Fig. 1a). Contrary thereto, so-termed “global MOE” is de-

termined acc. to EN 408 from deflection measurement w2 over full span of 18 h

including effects of shear and of indentations at the support locations (Fig. 1a).

In this paper global MOE is determined similar as local MOE acc. to EN

408 via deflection measurement w1 within an enlarged constant moment area of

9 h. Local MOE is determined by elongation/strain measurement over the length

of a small segment of the beam (see below).

2.1 Global modulus of elasticity

Global MOE based on deflection w1 in the constant moment area is (see

Fig. 1a)

1

2ma

globwI8

FE

``= (1)

with I = moment of inertia vs. major axis.

186


This measurement delivers an integral value over the length of `m.

2.2 Local modulus of elasticity

For determination of the local modulus of elasticity the bending compli-

ance of the beam was no more determined by deflection measurement but in-

stead by local elongation/strain measurements. The employed measuring princi-

ple is illustrated in Fig. 1b. The method consists of elongation/strain measure-

ments over “small” lengths at the bending tension and compression edge of the

beam. Based on the strains of the segment, εc and εt, the curvature κ of the seg-

ment of length L is given by (h = beam depth)

( )h

tc ε+ε=κ where

L

u )t(c

)t(c

∆=ε . (2a, b)

The bending MOE of the segment is then obtained from the curvature-moment

relationship

I

MEseg

κ= with M = F `a (3)

The introduced elongation/strain measurement method is limited to small

deflections; the arc length of the bending line of the beam segment has to be ap-

proximately equal to its chord length. This is the case for small ratios of L/h.

Here length L was chosen with respect to modelling aspects (smallest employed

“cell” size), not discussed in this paper, in conjunction with existing equipment

as L = 200 mm. This is roughly 1.5 times the depth of the investigated beams.

The determination of the elongations can be performed very accurately, say re-

producible, by means of so-called on-set strain extensometers. Despite the mis-

leading term “strain extensometer” the measurement actually represents an

elongation measurement of an exactly determined base length L. In order to es-

tablish the base length, small fixing plates (∅ 8 mm) with a conical hole for the

pin pointed extensometer legs are glued (wax) to the measured object, here to

the narrow faces of the beam.



w1

w2

M M L

∆ uc

∆ ut

Detail

Detail:

`a `m

F F

a)

b)

h

Fig. 1: Schematic illustration of global and local MOE determination

a) global MOE based on deflection measurement of w1 within constant mo-ment area (Note: this is different from EN 408 where global MOE is related to deflection w2)

b) local MOE based on elongation/strain measurement at top and bottom edge of the beam over a small length

3 TEST CONFIGURATION AND PROGRAM

The reported investigations comprise two test sets A and B of experiments,

all with beam specimens of equal size and loading. In test set A the differences

of the specifically employed local and global MOE determination were regarded

in detail, in test set B the (spatial) correlation of local MOE and the location of

bending failure was investigated.

All investigations were performed as 4point bending tests with span to

depth ratio of about 19 h. The constant moment length was chosen fairly large as

about 9 times the depth of the board (see Fig. 2). The constant moment length

was then divided into six equal sized segments of L = 200 mm, which is ap-

proximately 1.5 times the beam depth. For each segment the local modulus of

elasticity was determined. Additionally the global modulus of elasticity was

188


global deflection measurement w1 over ≈ 9 h

Detail, seeFigs. 2b - d

` = 2460

`a = 555 `m = 6 x L = 1200 `a = 555

b = 40

h =

129

L = 200

75 75

1350

a)

seg-ment 1 2 543 6

seg-ment 1 2

543 6

seg-ment 1 2 54

36

Detail (test A1 and test set B)

3 holes ∅

25 mm

3 holes ∅

25 mm

Extensometer

Detail (test A2)

Detail (test A3)4 x 50

4 x 50

h/4h/4

h/2

h/4h/4

h/2

b)

d)

c)

Fig. 2a-d: investigated specimen and loading configurations in test sets A and B

a) general test set-up

b) test A1(with board No. 1) and test set B; only natural defects (if at all) in segments 1 – 6

c) test A2; 3 artificial holes in segment 5 of board No. 1

d) test A3; 3 additional artificial holes in segment 3 of board No. 1



determined via deflection measurement over the length `m = 6 L. A distance of

75 mm between the load application points and the outer segments was chosen

in order to avoid strain disturbances due to the load concentration. The distance

between support and load application was 4 times the board depth in order to

avoid shear failure in test set B.

The elongation has been measured with a “strain” extensometer with a

resolution of 0.01 mm. Each segment length L has been measured two times,

first in the unloaded state and second in the loaded state at 3 kNm (σm = 27

N/mm2), being roughly 1/3 of the mean failure moment.

Test set A: In order to prove for the employed local MOE method in an ex-

emplary manner the ability to reveal the pronounced effect of local MOE varia-

tion, first three tests A1, A2 and A3 were performed with one board (No. 1)

which was inflicted in tests A2 and A3 with artificial defects. All tests com-

prised the six local and one global MOE measurements. In detail, in test A1 the

native board, being free of knots, was investigated (see Fig. 2b). In test A2 three

holes all with a diameter of 25 mm were drilled into beam segment 5 close to the

bending compression edge (see Fig. 2c). In test A3 three additional holes (∅ 25

mm) were drilled into beam segment 3, now close to the bending tension edge

(see Fig. 2d).

Test set B: In on-going tests the described local and global MOE meas-

urements were applied so far to 30 boards loaded to failure after compliance de-

termination at intermediate load stop at 3 kNm. One result evaluation reported

here was related to the spatial correlation of the failure location with the local

MOE distribution.

4 RESULTS OF TEST SET A

Tables 1 and 2 contain the results of the tests A1 to A3, i.e. the local strains

and the MOEs of the six segments and the global MOE. In addition, Tab. 2 con-

tains finite element calculated global MOEs (Eglob,calc) based on the experimen-

tally determined local MOEs (Eseg). The theoretical global MOE determined as

in the experiment from the global deflection in the constant moment area, serves

as a plausibility control of both, the locally and globally determined MOE. Fig-

ures 3 to 5 give a graphical representation of the strain and MOE results.

Hereby, the local MOEs are given as constant values within the specific segment

190


whereas the strains of the segments are shown for the center of the segment al-

lowing a better visual differentiation of strains and MOEs.

In all three tests A1 to A3 throughout a very good agreement between the

experimentally and computationally obtained global MOEs was observed. The

deviation was maximally 2%. Following the results are discussed in detail:

Test A1: Figure 3 reveals a very moderate local MOE variation (13350 to

14420 N/mm2) around the constant global value (13920 N/mm2). The extreme

deviations of the local MOEs (Eseg) vs. Eglob are + 3.6% and – 4.2%.

Test A2: Figure 4 shows a pronounced drop of the local MOE in segment 5

where the holes were placed. The difference between the extreme local MOEs

along span is 19%; the extreme deviations of the local MOEs vs. Eglob now are

+ 8.6% and – 9.1%. The effect of the local stiffness decrease on the global

MOE, however, is still moderate; Eglob now is 4.2% less compared to test A1

whereas Eseg5 decreased by 13%. It should be noted that the measured strains

show clearly the position of the defect application (see also Tab. 1). Whereas

strain εt at the bending tension edge of segment 5 is almost unchanged as com-

pared to the measurement in test A1, strain εc at the bending compression edge

of segment 5 shows a pronounced increase of 30%. In all other segments the lo-

cal strains remain very similar to those measured in test A1. In this context it

should be stated that the employed local strain based MOE determination gives

an error of maximally ± 300 N/mm2 at repeated measurements in conjunction

with the used extensometer.

Test A3: It can be seen from Fig. 5 that the additional artificial defects

(equal sizes and numbers as in test A2) applied close to the bending tension edge

in segment 3 lead to a pronounced decrease of Eseg3 of 14% vs. the former value

in test A2. The obtained reduction of local MOE resembles very closely the de-

crease of Eseg5 in test A2. The difference between the extreme local MOEs along

`m is 14%; the extreme deviations of the local MOEs vs. Eglob now are + 12.4%

and –2.9%. This second set of defects now reduces the global MOE by 3.5%

compared to the A2 result. Again the strains clearly show the depth location of

the new defect; now the strain at the bending tension edge increases by 24%,

whereas the strain at the bending compression edge of segment 3 remains rather

unchanged. The largest difference between local and global MOE now increased

to 12.4%.



Table 1: Compilation of local compression and tension strains of test set A

1 2 3 4 5 6

εc 10-5 -62.4 -60.4 -58.7 -64.9 -61.3 -60.2

εt 10-5 71.2 66.8 66.7 70.2 67.5 64.9

εc 10-5 -63.3 -59.5 -57.8 -66.7 -79.4 -62.9

εt 10-5 70.3 66.8 66.6 70.2 68.4 64.9

εc 10-5 -61.5 -57.7 -61.4 -66.7 -74.0 -61.1

εt 10-5 70.3 66.8 82.8 72.0 68.4 64.9

A2

A3

compression

and tension

strain per 1kNm

local strains in segments 1 to 6test

A1

Table 2: Compilation of results for local and global MOEs of test set A

global

measured

MOE

calculated MOE based

on the measured local

MOEs

Eseg1 Eseg2 Eseg3 Eseg4 Eseg5 Eseg6 Eglob Eglob, calc

- N/mm2

N/mm2

N/mm2

N/mm2

N/mm2

N/mm2

N/mm2

N/mm2

A1 13495 14172 14381 13347 13989 14415 13919 13932

A2 13494 14272 14493 13171 12197 14110 13340 13576

A3 13686 14479 12502 13000 12660 14311 12879 13141

test measured local MOEs in segments 1 to 6

12000

12500

13000

13500

14000

14500

15000

segment

modulu

s o

f ela

sticity [

N/m

m2]

55

60

65

70

75

80

85str

ain

[10

-5/k

Nm

]

1 2 3 4 5 6

Eseg Eglob |εc| εt

Fig. 3: Local strains, local and global MOEs of test A1 with board No. 1; no artificial defects

192


12000

12500

13000

13500

14000

14500

15000

segment

mo

du

lus o

f e

lasticity [

N/m

m2]

55

60

65

70

75

80

85

str

ain

[10

-5/k

Nm

]

1 2 3 4 5 6


Fig. 4: Local strains, local and global MOEs of test A2 with board No. 1; artificial defects in the bending compression part of segment 5

12000

12500

13000

13500

14000

14500

15000

segment

mo

du

lus o

f e

lasticity [

N/m

m2]

55

60

65

70

75

80

85

str

ain

[10

-5/k

Nm

]

1 2 3 4 5 6


Fig. 5: Local strains, local and global MOEs of test A3 with board No. 1; additional arti-ficial defects in the bending tension part of segment 3



5 RESULTS OF TEST SET B

Table 3 specifies the measured local and global MOEs, the computed

global MOEs and the location of the failure of four beech boards. The chosen

examples are exemplary for 30 tests performed so far. Figures 6 to 9 give

graphical representations of the results including the local strain variations.

Figure 6 reveals the case of a nearly homogeneous board with no knots and

no apparent grain deviation. Local and global MOEs show very little differ-

ences. Despite the small stiffness variations, the bending tension failure occurred

in segment 3 with the lowest local MOE and with highest tension strain. The

minimum local MOE differed only by 3% from the global MOE and only by

1.3% from the next weakest segment 1.

Figure 7 also depicts the strains and local MOE variations of a board with-

out knots, but nevertheless with pronounced differences (maximally 18%) of

local MOEs ranging from 12600 to 15400 N/mm2. The extreme deviations of

the local MOEs vs. the global MOE are + 7.7% and – 11.8%. The strains of seg-

ments 2 and 3 with lowest local MOEs show an interesting feature being that

maximum tension and compression strain occur successively in segments 2 and

3, indicating sloping grain. The specimen failed in bending tension at the transi-

tion of segments 2 and 3.

Figure 8 relates to a board with a knot of 22 mm diameter and associated

strong fibre deviations around the knot located in the upper bending compres-

sion part of segment 3. The very pronounced difference between the extreme

local MOEs of 10360 and 14200 N/mm2 was 27%; the extreme deviations of the

local MOEs vs. Eglob were –17% and 13.6%.

Table 3: Compilation of local and global MOEs of test set B

global

measured

MOE

calculated MOE

based on the

measured local

MOEs

location of

failure

Eseg1 Eseg2 Eseg3 Eseg4 Eseg5 Eseg6 Eglob Eglob,calc segment 1)

B1 13555 14140 13375 13746 13746 14138 13779 13713 3

B2 15395 13687 12597 14032 14586 14212 14288 13703 2 - 3

B3 13808 11696 10356 13435 14203 12911 12508 12222 2 - 3

B4 17844 17843 16270 15825 15842 15366 16830 16333 41) x - y means the intersection betw een tw o segments

test measured local MOEs in segments 1 to 6

194


12000

13000

14000

15000

16000

17000

segment

modulu

s o

f ela

sticity [

N/m

m2]

40

45

50

55

60

65

str

ain

[10

-5/k

Nm

]

1 2 3 4 5 6


area of failure initiation

Fig. 6: Local strains, local and global MOEs of test B1 (board No. 416)

11000

12000

13000

14000

15000

16000

segment

mo

du

lus o

f e

lasticity [N

/mm

2]

45

50

55

60

65

70

str

ain

[10

-5/k

Nm

]

1 2 3 4 5 6




The specimen failed as sole specimen in the tests so far in bending compression

at the transition of segments 2 and 3 with highest compression strains and lowest

local MOE, respectively.



10000

11000

12000

13000

14000

15000

segment

mo

du

lus o

f e

lasticity [

N/m

m2]

50

60

70

80

90

100

str

ain

[10

-5/k

Nm

]

1 2 3 4 5 6




14000

15000

16000

17000

18000

19000

segment

mo

du

lus o

f e

lasticity [

N/m

m2]

40

45

50

55

60

65

str

ain

[10

-5/k

Nm

]

1 2 3 4 5 6




Figure 9 shows strains and MOEs of a board without knots and absolutely

very “high” MOEs with a global MOE of 16830 N/mm2. Bending compression

and tension strains are very similar. The specimen failed in bending tension in

segment 4 with the second lowest MOE. However, local MOEs in segments 4, 5

and 6 are very similar and deviate maximally by 2% from their respective mean.

196


Again, as in test set A, a very good agreement between the experimentally

and computationally obtained global MOEs was observed. The deviation was in

average 2.4% and maximally 3.7%.

6 DISCUSSION

The results show that the employed method is able to deliver local MOEs

and therefore to reveal the MOE variation within a board. However, the meas-

ured local MOEs still do not represent the true MOEs of the zones with or with-

out defects within the board. The measured MOE depends to a great extent on

the gauge (segment) length, L, the length of the weak area and also on the rela-

tive differences of the stiffness within the gauge length. The smaller the chosen

segment length, the smaller the difference between the measured and the “true”

MOE. The employed gauge length of about 1.5 times the board depth seems to

be in the size range of typical defect zones of the regarded wood species as the

results show a good correlation between the minimum localized MOE value and

location of bending failure. However, some improvement should still be ob-

tained by a further reduction of gauge length L.

7 CONCLUSIONS

The presented results show that determination of the local modulus of elas-

ticity in (edgewise) bending can be well performed by elongation/strain meas-

urement at the bending tension and compression edges.

The measured local MOEs and the experimental global MOE obtained from de-

flection measurement, are consistent. This results from the fact that the global

MOE can be predicted by beam theory or FE analysis with an average error of

about 2% on the basis of the local MOE of the segments, here chosen with a

length of 200 mm.

It was revealed that the locations of failure comply well with the locations of

minimum MOE along beam length (the study so far comprised 30 beech

boards). The presented method seems to enable the prediction of the type of

bending failure either at the tension or compression edge.

The data of the on-going study serve as a calibration basis for modelling of the

variation of modulus of elasticity and bending strength along beech wood boards

as input data for glued compound elements with edgewise bent lamellas.



ACKNOWLEDGEMENTS

The authors want to express sincere thanks to Patrick Castera, head of La-

boratoire du Bois de Bordeaux (LRBB), for his repeated favour in performing

the translation of the French abstract.

REFERENCES

BECHTEL, F.K. (1985): Beam stiffness as a function of point wise E, with applica-

tion to machine stress rating. Proceedings International Symposium on

Forest Products Research, CSIR, Pretoria, South Africa

CORDER, S.E. (1965): Localized deflection related to bending strength of lumber.

Second Symposium on the Non-destructive Testing of Wood, Washington

State University, Pullman, WA, pp. 461 – 473

EHLBECK, J., COLLING, F., GÖRLACHER, R. (1985): Einfluß keilgezinkter Lamellen

auf die Biegefestigkeit von Brettschichtholzträgern. Entwicklung eines Re-

chenmodells. Holz Roh- Werkstoff, 43, pp. 333 – 337

FOSCHI, R.O., BARRETT, D. (1980): Glued-laminated beam strength: A model. J.

of the Structural Div., Vol. 106, No. ST8, pp. 1735 – 1754

FOSCHI, R.O. (1987): A procedure for the determination of localized modulus of

elasticity. Holz Roh- Werkstoff, 45, pp. 257 – 260

ISAKSSON, T. (1999): Modeling the variability of bending strength in structural

timber; length and load configuration effects. Report MBK-1015, Div. of

Structural Eng., Institute of Technology, Lund, Sweden

KASS, A.J. (1975): Middle ordinate method measures stiffness variation within

pieces of lumber. Forest Products J., 25 (3), pp. 33 – 41

KLINE, D.E., WOESTE, F.E., BENDTSEN, B.A. (1986): Stochastic model for modulus

of elasticity of lumber. Wood and Fibre Science, 18 (2), pp. 228 - 238

SERRANO, E. (2001): Mechanical performance and modeling of glulam. Manu-

script for „Timber Engineering“, Edts. S. Thelandersson and H.J. Larsen,

Wiley & Sons, in press

TAYLOR, S.E., BENDER, D.A. (1991): Stochastic model for localized tensile

strength and modulus of elasticity in lumber. Wood and Fibre Science,

23(4), pp. 501 - 519

198

Transient temperature evolution in glulam with hidden and non-hidden glued-in steel rods

TRANSIENT TEMPERATURE EVOLUTION IN GLULAM WITH HIDDEN AND NON-HIDDEN GLUED-IN STEEL RODS

TRANSIENTE TEMPERATURENTWICKLUNG IN BRETTSCHICHT-HOLZ MIT VERDECKT UND NICHT VERDECKT EINGEKLEBTEN STAHLSTANGEN

EVOLUTION TRANSITOIRE DE LA TEMPERATURE DANS DU LAMELLE COLLE COMPORTANT DES GOUJONS COLLES EN ACIER, APPARENTS OU NON

Simon Aicher, Dirk Kalka, Ralf Scherer

SUMMARY

A recently terminated European research project on glued-in steel rods in

timber structures (GIROD) – with participation of Timber Department of Otto-

Graf-Institute – revealed a strong strength reducing influence of elevated tem-

peratures, not expected to that extent. This affects especially the duration of load

behavior of the connections. The maximum temperature level acting in service

on the glued-in rod connections thus sets performance requirements on the shear

modulus-temperature relationship and on the glass transition temperature of ap-

propriate adhesives.

Today’s prevailing intuitive conviction of practitioners is, that rods bonded

hidden in the interior of glulam cross-sections experience, due to the low ther-

mal conductivity and specific heat of wood, considerable lower temperatures

compared to ambient climate. The paper gives some experimental and computa-

tional results proving, depending on cross-sectional width, only a moderate re-

duction of peak temperatures combined with a pronounced phase shift vs. ambi-

ent temperature varying roughly sinusoidally during a day.

ZUSAMMENFASSUNG

Ein kürzlich abgeschlossenes Europäisches Forschungsvorhaben betreffend

in Holz eingeklebter Stahlstangen (GIROD) – mit Beteiligung des Fachbereichs

Holz des Otto-Graf-Instituts – zeigte einen in dieser Ausprägung nicht erwarte-

ten großen festigkeitsmindernden Einfluss erhöhter Temperaturen. Dies beein-

flusst insbesondere auch das Zeitstandverhalten der Verbindungen. Das maxi-


S. AICHER, D. KALKA, R. SCHERER

male Temperaturniveau, das im Gebrauchszustand auf Verbindungen mit einge-

klebten Stangen einwirkt, definiert somit Leistungsanforderungen an die

Schubmodul-Temperaturbeziehung und an die Glasübergangstemperatur geeig-

neter Klebstoffe.

Die heute in der Praxis vorherrschende Meinung ist, dass verdeckt in das

Innere eines Brettschichtholzquerschnitts eingeklebte Stangen infolge der nied-

rigen Wärmeleitfähigkeit und Wärmekapazität von Holz beträchtlich niedrigeren

Temperaturen im Vergleich zum einwirkenden Umgebungsklima ausgesetzt

sind. Der Aufsatz berichtet über einige experimentelle und rechnerische Ergeb-

nisse, die belegen, dass abhängig von der Querschnittsdicke lediglich eine

schwache Reduzierung der Spitzentemperaturen verbunden mit einer ausgepräg-

ten Phasenverschiebung gegenüber den Außentemperaturen, die näherungsweise

sinusförmig über den Tag variieren, vorliegt.

RESUME

Un projet de recherche Européen portant sur les goujons collés en acier

dans les structures en bois (GIROD) récemment achevé – auquel participait le

département bois de l’Otto-Graf Institute – a mis en évidence un effet négatif

marqué de températures élevées sur la résistance, qui affecte principalement la

durée de vie des joints. La température maximale agissant sur les joints en ser-

vice impose donc des exigences de performance sur la relation température –

module de cisaillement et la température de transition vitreuse des adhésifs ap-

propriés.

La conviction intuitive des praticiens aujourd’hui est de penser que les gou-

jons collés cachés à l’intérieur des sections de lamellé collé sont soumis, du fait

de la faible conductivité thermique et chaleur spécifique du bois, à des tempéra-

tures considérablement plus faibles que celles du climat ambiant. Cet article pré-

sente des résultats expérimentaux et numériques montrant, selon la largeur de la

section, une faible réduction seulement des températures de pic combinée à une

transition de phase prononcée, par rapport à la température ambiante qui varie

grossièrement de manière sinusoïdale au cours d’une journée.

KEYWORDS: glued-in rods, glulam, elevated temperatures, transient tempera-

ture evolution

200


1. INTRODUCTION

Today’s prevailing conviction of practitioners is, that steel rods bonded

hidden in the interior of glulam cross-sections experience, due to the low ther-

mal conductivity and specific heat of wood, considerable lower temperatures

compared to ambient climate. A European research project on glued-in rods in

timber structures (GIROD) revealed an unexpected strong strength reducing in-

fluence of elevated temperatures in duration of load tests with connections

bonded by epoxy and polyurethane adhesives [BENGTSSON, JOHANSSON, 2002;

AICHER, 2002].

In the performed tests the temperature increase was applied after mechani-

cal loading thus suppressing positive post-curing effects. The experiments re-

vealed clearly the crucial importance of a sufficiently high glass transition tem-

perature. Performance requirements on temperature stability – especially shear

modulus-temperature relationships and/or glass transition temperature – have to

be set in view of realistic temperature loading scenarios. Eventually the tempera-

ture loading should also be considered in probabilistic manner, in case a specific

adhesive shows high post-curing potential.

The reported experimental investigations were performed in first instance

to substantiate the GIROD results. In addition thereto a major point of interest

was the transient temperature evolution in the timber-bond line interface.

2. TEST PROGRAM

In order to verify the different temperature-strength behavior of glued-in

steel rods either protruding or fully hidden in the wood, two types of specimens

shown in Fig. 1 were investigated. The performed experiments concerned the

strength verification at variable temperature and static long term loads and fur-

thermore the temperature evolution in the bond lines. This paper reports on the

temperature evolution.

The temperatures in the bond line were measured with thermo-elements

consisting of copper/constantan (Cu/Cu-Ni) wires. In order to measure the tem-

peratures in the bond line with little interfering influences of leakages to ambient

climate, the application of the thermo-element wires to the bond line was per-

formed as following: first an oversized specimen was sawn up lengthwise with a

saw blade thickness of 2 mm. Then the two parts were clamped together and a

hole with a diameter of 13 mm for the glued-in rod was drilled. The thermo-



wires were glued into small notches as shown in Fig. 2a. The end or actual

measuring point of the thermo-wire was flush with the surface of the drilled

hole. Then the two parts of the specimen were glued together again. In a second

step the steel rod (Ø 12 mm) was glued into the wooden piece (see Fig. 2b). All

gluing were performed with a special epoxy adhesive.

a) b)

specimen part A

specimen part B

thermo-elements

thermo-elements

support rod M24

glued-in test steel rod M12

115

115 T1

T5

T2 T3 T4

40

40

40

40 5

15

240

180

180

600 insulation tape

support rodM24

600

support rodM24

glued-in test steel rod M12

115

240

180

180

600

115 T1

T5

T2 T3T4

40

40

40

405

15

Fig. 1a,b: Geometry and schematic test set-up of a) specimen No. I with protruding steel rod and b) specimen No. II with hidden steel rod

In order to obtain a specimen with a fully hidden rod which could be sub-

jected to temperature and mechanical loads, specimen No. II, shown in Fig. 1b,

202


was used. First, specimen part A was manufactured as specified above and after

curing of the adhesive, part A incorporating the protruding test rod was glued

into the rod hole of specimen part B. In order to enforce exclusively load trans-

fer between the specimen parts A and B via the glued-in rods, a Teflon sheet

with a thickness of 0.5 mm was inserted between the two parts of the specimen

(see Fig. 2c) . The surrounding edge of 10 mm width and 2 mm depth of the two

specimen parts was sealed with an elastic insulation tape compressed to 0.5 mm

(see Fig. 1b and 2d).

a) b)

c) d)

Fig. 2a-d: Views of the specimens No. I (a,b) and No. II (c,d)



3. TEMPERATURE LOADING

As in the GIROD project a cyclic sinusoidal variation of warm and dry

climate was applied [AICHER ET AL., 2002]. Contrary to GIROD, where a full

temperature cycle consisted of 8 hours, now a practically relevant cycle length

of 24 hours was chosen. A sinusoidal variation of temperature within a time

span of 24 hours represents a very good approximation of daily temperature

courses. This is shown exemplarily in Fig. 3 with recorded temperature data

(sheltered outdoor, well ventilated shed in Stuttgart) for a period of three succes-

sive days.

14

16

18

20

22

24

26

28

30

28.7.02 29.7.02 30.7.02 31.7.02

time

tem

pera

ture

T [

°C]

sinusoidal approx.

measured

Fig. 3: Course of temperature (sheltered outdoor conditions) of typical summer days and

sinusoidal approximation of the temperature

As in the GIROD project the minimum and maximum set temperatures

were chosen as 25°C and 55°C, resulting in a peak-to-peak temperature ampli-

tude of 30 K. These temperature boundaries might be regarded as an upper, yet

realistic temperature range, which can occur under a dark, little ventilated roof

in very warm summers in the Southern part of Europe. The course of the applied

temperature and of the relative humidity is given in Fig. 4. The controlling of the

relative humidity was limited, due to technical restrictions of the climate cham-

ber, to 45% RH during a time of 6.5 h of a full cycle of 24 h, as shown in Fig. 4.

204


The actual temperatures obtained in the climate chamber showed a mini-

mum and maximum of 24.9°C and 54.7°C, respectively, with a peak-to-peak

temperature amplitude of 29.8 K. The relative humidity roughly ranged from 5

to 50 %, exceptionally temporary up to 67 % for about 0.5 hours.

0

10

20

30

40

50

60

70

0 12 24 36 48 60 72 84 96 108 120

time t [h]

tem

pera

ture

T [

°C]

/

rela

tive h

um

idit

y R

H [

%] T

RH

Fig. 4: Course of the applied ambient temperature and relative humidity variation in the climate chamber

4. NUMERICAL INVESTIGATIONS

In an early paper the evolution of temperature in a specimen with a glued-

in rod protruding into ambient air was investigated numerically and experimen-

tally [AICHER ET AL, 1998], taking into account the timber, the adhesive layer

and the steel. An additional fourth layer, representing a steel/adhesive interface,

was introduced in order to account for the problem that the used FE-code does

not enable the specification of contact conductance of inner surfaces. By means

of the interface layer a good agreement of measured and calculated transient

temperatures was obtained.



In this paper the preliminary numerical study is exclusively related to

specimen No. II with the hidden rod. In a first crude approximation the inner

steel rod was omitted, so only the heat transfer through a quadratic block of tim-

ber was regarded in a 2 dimensional analysis. The cross-sectional dimensions of

the timber, a = 115 mm, were reduced by the hole diameter of 13 mm (rod

diameter + 1 mm) to 102 mm.

In the calculations a constant thermal diffusivity perpendicular to grain of

h

²mm700

C

kD

P ρ

was employed. Hereby k , CP and ρ were assumed as

Km

W13.0k

thermal conductivity perpendicular to fiber

acc. to DIN 4108, part 4

Kkg

kJ6.1CP specific heat [BATZER, 1985]

³m

kg420ρ

mass density of glulam at dry status of about

u = 7 %

The convection heat transfer coefficient h was chosen as a fitting parameter

in the range of 10 to 20 W/(m²·K). Literature data for forced convection of gas

media vary roughly between 10 and 100 W/(m²·K); a value of 25 W/(m²·K) is

assumed for convection at exterior walls in DIN 4108, part 4.

5. TEMPERATURE EVOLUTION IN CYCLIC CLIMATE

Figs. 5a,b show the temperature evolution of both specimen types I and II

at different thermo-element positions. The evolution of the ambient temperature

in the climate chamber is given, too. For a better visualization of phase shift and

differences in amplitudes Figs. 6a and b show the temperature evolution at a cy-

cle length of 24 hours; additionally finite element computed temperature evolu-

tions based on the revealed approach are specified in case of specimen type II

with a hidden rod.

206


a)

20

25

30

35

40

45

50

55

60

0 6 12 18 24 30 36 42 48 54 60 66 72

time t [h]

tem

pera

ture

T [

°C]

T1

T2

T3 - T5

T0T1T2

T4T3

ambient climate (climate chamber)

T1

T2

T3

T4

T5

20

25

30

35

40

45

50

55

60

0 6 12 18 24 30 36 42 48 54 60 66 72

time t [h]

T1 - T5

T

TTTT

ambient climate (climate chamber) finite element calc.

measured temperatures

calc_1calc_2calc_3

T5

T4

T3

T2

T1

part B

part A

[°C

]

tem

pera

ture

T

b)

Fig. 5a,b: Temperature evolution over 3 days at the positions of the thermo-elements a) specimen No. I with protruding steel rod b) specimen No. II with hidden steel rod



a)

20

25

30

35

40

45

50

55

60

24 27 30 33 36 39 42 45 48

time t [h]

tem

pera

ture

T [

°C]

T1

T2

T3 - T5

T0T1T2

T4T3

ambient climate (climate chamber)

T1

T2

T3

T4

T5

20

25

30

35

40

45

50

55

60

24 27 30 33 36 39 42 45 48

time t [h]

T1 - T5

ambient climate(climate chamber)

finite element calc.

measured temperatures

calc_1calc_2calc_3

T5

T4

T3

T2

T1

part B

part A

[°C

]

ure

T

tem

pera

t

b)

Fig. 6a,b: Temperature evolution over 1 day at a cycle length of 24 hours a) specimen No. I with protruding steel rod b) specimen No. II with hidden steel rod

208


The differences between the two specimen types are rather small. Purely

qualitatively the temperature in the wood-bond line interface of specimen No. II

shows slightly decreased amplitudes and a slightly more pronounced phase shift

vs. ambient climate when compared to specimen No. I with the protruding rod.

Quantitatively the results are specified in Tab. 1.

Tab. 1: Temperature evolution in the wood-adhesive interface of specimens No. I and No. II at thermo-element positions T1 and T5

thermo-element T1

(“protruding” end of steel rod) thermo-element T5

(embedded end of steel rod)

maximum tempera-

ture Tmax

minimum tempera-

ture Tmin

peak-to-peak

amplitude

∆T

phase shift

∆t

maximum tempera-

ture Tmax

minimum tempera-

ture Tmin

peak-to-peak

amplitude

∆T

phase shift

∆t

[°C] [°C] [K] [h] [°C] [°C] [K] [h]

ambient climate

54.7 24.9 29.8 - 54.7 24.9 29.8 -

experimental results:

specimen

No. I (protrud-ing rod)

53.4 28.1 25.3 1.7 52.6 28.2 24.4 2.4

specimen

No. II (hidden rod)

51.2 28.7 22.5 3.3 51.6 28.8 22.8 3.2

It can be seen that the maximum temperatures at the embedded ends of the

rods (thermo-element T5 for specimens No. I and No. II) differ only very little

by about 1°C. The reduction of maximum temperature vs. ambient climate in

case of specimen No. II (hidden rod) was only 3 K. The phase shift between the

maximum temperature of the ambient climate and the maximum of recorded

temperatures was 2.4 and 3.2 hours in case of specimens No. I and No. II, re-

spectively.

In case of specimen No. II no difference of temperature amplitudes, peak

temperatures and phase shifts between thermo-element T1 close to the sealed

joint of both specimen parts A and B as compared to the embedded end (thermo-

element T5) was observed. In case of leakages at the sealing of the joint of the



parts A and B a different result should be obtained. This is substantiated by the

calculations.

Table 2 represents the results of the rough finite element calculation com-

pared to the experimental results of specimen type II and the applied ambient

climate. It can be seen that either the maximum and minimum temperatures Tmax

and Tmin (result calc_1) or the phase shift ∆t (result calc_3) of the measured ex-

perimental results can be fitted by tuning of the convection heat transfer coeffi-

cient h. A roughly acceptable approximation of both , the temperatures and the

phase shift, is obtained with a convection heat transfer coefficient of h = 15

W/(m²·K). This number is within the plausible range.

Tab. 2: Maximum temperatures and phase shift for specimen No. II according to experi-mental test results and finite element calculation

calculation result

convection heat transfer

coefficient h

[W/(m²·K)]

maximum temperature

Tmax

[°C]

minimum temperature

Tmin

[°C]

peak-to-peak amplitude

∆T = Tmin-Tmax

[K]

phase shift

∆t [h]

ambient climate

- - 54.7 24.9 29.8 -

experimental results

- - 51.2 28.7 22.5 3.3

calc_1 10 51.4 28.6 22.8 4.2

calc_2 15 52.6 27.4 25.2 3.6

2D finite element calculation

calc_3 20 53.2 26.8 26.4 3.3

6. INFLUENCE OF TIMBER THICKNESS

The presented test results and hereby the small damping of the temperature

maxima is obviously related to the cross-sectional dimensions of the specimens

(quadratic cross-section of 115 mm · 115 mm). In order to verify the influence

of an increased timber thickness some more calculations similar to those out-

lined in chapter 4 were performed. The convection heat transfer coefficient was

chosen as h = 15 W/(m²·K) being the value which forwarded a reasonable good

agreement of the simplified analysis with the hidden rod specimen No. II. The

imposed temperature varies again sinusoidally between 25 and 55°C with a

phase length of 24 hours.

210


Fig. 7 gives the temperature courses for square cross-sectional dimensions

with thicknesses of a = 50, 102, 150 and 200 mm. The value a = 102 mm relates

to the discussed results of specimen No. II, whereby the reduction of the real

thickness of 115 mm to 102 mm is bound to the simplified approach of omitted

rod cross-section. The graph shows the qualitatively somewhat trivial result of a

decreasing temperature amplitude and an increasing phase shift with growing

cross-sectional dimensions.

20

25

30

35

40

45

50

55

60

24 27 30 33 36 39 42 45 48

time t [h]

tem

pera

ture

T [

°C]

T1 - T5

ambient climate(climate chamber)

finite element calc.a = 50, 102, 150, 200 mm

measuredtemperatures

a

T5

T4

T3

T2

T1

Fig. 7: Temperature evolution at a cycle length of 24 hours for finite element calculations with varying timber thicknesses compared to experimental results (specimen No. II with hidden rod)

A quantitative summary of the results is given in Tab. 3. In detail the

changes of minimum and maximum temperature, the peak-to-peak amplitude

∆T = Tmax-Tmin and the phase shift ∆t are specified. It can be seen that the reduc-

tion of the maximum temperature in the regarded range of cross-sectional di-

mensions is rather moderate. For a medium sized glulam thickness of 150 mm

the maximum value is still rather close to 50°C, which marks the limit of type I

adhesives acc. to EN 301.

The quantitative results of the very rough approximation of the problem

shall be regarded with a more refined modeling considering the true build-ups.



Tab. 3: Extreme temperatures, temperature differences and shifts of phase depending on timber thickness according to a simplified calculation

calculation result

cross-sectional thickness

a [mm]

maximum temperature

Tmax

[°C]

minimum temperature

Tmin

[°C]

peak-to-peak amplitude

∆T = Tmin-Tmax [K]

phase shift

∆t [h]

ambient climate

- - 54.7 24.9 29.8 -

experimental results

- - 51.2 28.7 22.5 3.3

calc_1 50 54.0 26.0 28.0 2.5

calc_2 102 52.6 27.4 25.2 3.6

calc_3 150 48.7 31.3 17.4 5.7

2D finite element calculation

calc_4 200 46.3 33.6 12.7 7.2

7. CONCLUSIONS

The performed experiments on transient temperatures in glue-line/wood

interfaces of steel rods bonded into glulam and subjected to cyclically varying

ambient climate revealed

• relatively low damping of maximum temperatures for a cross-sectional

thickness of 115 mm,

• pronounced phase shifts and

• only minor differences between the cases of protruding or hidden rods.

The results were extrapolated to different cross-sectional thicknesses by

means of numerical calculations with a simplified model. The calculation results

yielded rather moderate damping within the typical range of glulam thicknesses

up to 200 mm. Roughly it can be concluded that the maximum ambient tempera-

ture level acting in service on the glued-in rod connections sets the performance

requirements on the shear modulus/temperature relationship resp. on the glass

transition temperature of appropriate adhesives.

ACKNOWLEDGEMENTS

The authors are cordially indebted to Dr. Patrick Castera, Head of Labora-

toire du Rheologie du Bois Bordeaux (LRBB), for performing the french transla-

tion.

212


REFERENCES

AICHER, S. (2002): Duration of load tests on full-sized glued-in rod specimens.

GIROD_WP5: Technical Report for work package 5, Research Report,

Otto-Graf-Institute, University of Stuttgart.

AICHER, S.; KALKA, D.; HÖFFLIN, L. (2001): Duration of load tests on full-

sized glued-in rod specimens. GIROD_WP5: Technical Report for work

package 5, workpart by FMPA. Research Report, Otto-Graf-Institute,

University of Stuttgart.

AICHER, S.; WOLF, M.; DILL-LANGER, G. (1998): Heat flow in a glulam joist

with a glued-in steel rod subjected to variable ambient temperature.

Otto-Graf-Journal Vol. 9, pp. 185-204 , Otto-Graf-Institute, University of

Stuttgart.

BATZER, H. (1985): Polymere Werkstoffe. Volume I, Georg Thieme Verlag

Stuttgart. New York.

BENGTSSON, C.; JOHANSSON, C.-J. (2002): GIROD – Glued in rods for timber

structures. SP Report 2002:26. SP Swedish National Testing and Research

Institute.



214

Modelling of concrete hydration

MODELLING OF CONCRETE HYDRATION

MODELLIERUNG DER BETON HYDRATATION

MODELISATION DE L'HYDRATATION DU BETON

Sven Mönnig

SUMMARY

DuCOM is a finite element program which can show the hydration of

concrete with any concrete mixtures, to any given time step and different

environmental conditions. Comparing calculated temperature distribution,

hydration and heat growth rates with measurements a high accuracy was proven.

ZUSAMMENFASSUNG

DuCOM ist ein FEM-Programm, welches die Hydratation von Beton mit

beliebigen Betonmischungen, zu beliebigen Zeitschritten und verschiedenen

Umgebungsbedingungen darstellen kann. Bei dem Vergleich von errechneten

Temperaturverläufen, Hydratationskurven und Wärmezuwachsraten mit

gemessenen Laborwerten, zeigt sich eine hohe Genauigkeit von DuCOM.

RÉSUMÉ

DuCOM est un programme d'éléments finis capable de décrire l'hydratation

de bétons de compositions arbitraires, à des intervalles arbitraires et pour

différentes conditions d'environnement. La comparaison des gradients de

température, des courbes d'hydratation et des taux de chaleur calculés avec les

valeurs mesurées en laboratoire indiquent une précision élevée de DuCOM.

KEYWORDS: DuCOM, heat growth rate, concrete hydration

1. INTRODUCTION

DuCOM [1] is a program working with concrete finite elements. It is able

to deliver a linear description of the hydration of concrete. It provides solutions

for pore pressure and temperature at each node of each element for given time

steps and environmental condition, i.e. relative humidity and temperature.


S. MÖNNIG

Results of porosity, the degree of hydration for every clinker, shrinkage and

strength are obtained, too. By implementing DuCOM, a program developed by

the University of Tokyo, into MASA [2] the simulation of the influence of

hydration on the bearing capacity is possible. To estimate the computational

accuracy of DuCOM extended calculations were compared with publications of

test results.

2. THEORETICAL PRINCIPALS OF DUCOM

Physical processes like humidity and vapour transport, the hydration of

concrete and the development of pore structure are integrated over the volume of

a standard reference element. The transport behaviour is simulated on a macro

scale. Hydration is simulated by a multi component system which includes the

heat development and the amount of available water. The heat development is

dependent on the amount of free water. Size and structure of the pores are

dependent on the degree of hydration. The pore structure influences the transport

behaviour inside the concrete. All the single processes are dynamically linked

and dependent on each other as figure 1 should point up.

FEMAP

Visual Basic program

provides input data: nodes and elements; geometry of model

addional input data: mixture; environmental conditions

DuCOM

model for hydration

pore structure

pore pressure

convergence

T > T soll

output of data

VB program Program rewrites DuCOM output format into neutral ASCII files

FEMAP display of results

DuCOM developed by University of Tokyo

output data: temperature; hydration degree

dispersal of porosity

relative humidity; dispersal of moisture; pore pressure

Figure 1: application flow

216


For further information it is recommended to read “Modelling of concrete

performance” [1].

3. LIMITATIONS OF DUCOM

There are some constrictions of DuCOM that should be mentioned. The

pore structure is simulated by a consistent distribution of average sized grains.

The size is dependent on the amount of cement, fly ash and blast furnace slag.

The distance between the grains is based on Blaine values and the size of the

grains. Pores are considered to be cylindrically shaped. For the calculation of the

hydration the gel and capillary pores are treated as one type.

The assumptions for the moisture transport are non deformable and

isothermal material behaviour. Furthermore it is assumed that the total mass of

vapour can be neglected compared to the total amount of water. Gas pressure

within the material is constant and equals the air pressure. Liquid transport is

performed with constant velocity. Thermal effects are negligibly small. These

assumptions are based on a representative volumetrically element. All

calculations refer to this element.

4. IMPLEMENTATION OF DUCOM

With FEMAP as input and output program it is possible to use a common

used program for the visualization of the models. The output file from FEMAP

is written in an ASCII format which is translated into the input file for DuCOM.

For this transformation a Visual Basic (VB) program was developed by the

IWB. While translating the file, the program asks for additional input data.

Necessary input information are the time period of the analysis, mixture, Blaine

values, temperature of the concrete mixture, temperature and relative humidity

of the environment. After the end of the calculation another VB program will

translate the result files of DuCOM into an ASCII format which can be read by

FEMAP. The results can be graphically presented and furthermore they can be

read from MASA and used for continuous analysis of the structure.

5. RESULTS OF SIMULATIONS AND COMPARISON WITH TEST RESULTS

All calculations were based on a 20×20×20 cm3 cube. The resulting values

of the curves were taken at nodes arranged along a line by the centre of a cube.


S. MÖNNIG

OPC was simulated with these fractions of clinker:

C3S 47,2 %

C2S 27,0 %

C3A 10,4 %

C4(A,F) 9,4 %

The model of the concrete cube has been scaled by the input program to the

size desired by the user. Figure 2 shows the cube before scaling.

X

0.

1.

2.

3.

4.

5.

6.

7.

8.

9.

10.

Y

0.1.

2.3.

4.5.

6.7.

8.9.

10.

Knoten 121

Knoten 96

Knoten 71

Knoten 46

Knoten 21

Element 27

node 121

node 96

node 71

node 46

node 21

element 27

Figure 2: Model of a 20×20×20 cm3 cube

Temperature distribution in a cube

Of interest was the influence of the environment on the development of

heat inside the cube. Simulations with an adiabatic system have been performed

as well as calculations with one, two, three, four and five sides opened to the

environment. The mix temperature was 20°C. The environmental conditions

have been assumed constant with 15°C and 100% relative humidity. The

concrete mix contained 375 kg/m3 cement and 1885 kg/m3 aggregates.

218


0.00 0.01 0.10 1.00 10.00 100.00

Figure 3: Temperature distribution inside a cube

Degree of Hydration in dependence on the water/cement ratio

The proportion has been changed to a mixture with very high cement

content. It contained 836.3 kg/m3 cement and 1032 kg/m3 aggregates. These

fractions have been chosen to minimize the influence of the aggregates on the

water diffusion and the hydration. The reference mixture had a water/cement

ratio of 0.4 but the same cement and aggregate content as the others. This

mixture has been calculated with five sides open to the environment which had a

constant temperature of 15°C and a relative humidity of 100 %.

Progress of hydration

0,0

0,1

0,2

0,3

0,4

0,5

0,6

0,7

0,8

0,9

1,0

0 4 8 12 16 20 24

Hours [h]

Hyd

rati

on

Deg

ree [

%]

W/C 0.4 Five Open

Sides

W/C 0.20 Adiabatic

W/C 0.40 Adiabatic

W/C 0.60 Adiabatic

1.0

0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0.0

Figure 4: Degree of hydration in dependence on the water/cement ratio


S. MÖNNIG

Heat growth rate of clinker

DuCOM provides the overall generated heat for each clinker at each time

step. By subtracting the accumulated heat at one time step from the previous one

it was possible to calculate the heat growth rate. The curves presented in figure 5

have been interpolated with Excel to abrade them. DuCOM calculated 2500 time

steps to reach 24 hours.

0,00

0,05

0,10

0,15

0,20

0,25

0,30

0 4 8 12 16 20 24

Time [h]

He

at

Gro

wth

Ra

te [

kc

al/k

g]

C3A

C3S

C4AF

C2S

Total Heat

0.30

0.25

0.20

0.15

0.10

0.00

0.05

Figure 5: Heat growth rate of clinker

6. DISCUSSION

The maximum temperature of the hydration as shown in figure 3 does

reach the extent as expected. Different methods of gaining an approximated

value do provide similar results, e.g. the approximation formula for adiabatic

heat growth as given in [3] provides likewise results. The results presented in

figure 5 follow the expected curves. The influence of the low temperature of the

environment on the rate of hydration is reasonable, too. The curves in figure 5

show the same behaviour of the clinker as it was expected due to the specific

enthalpy of each clinker.

7. SUMMARY OF RESULTS

DuCOM proved to be very reliable being used for the simulation of

hydration of ordinary Portland cements and their mixtures. Temperature

development and hydration degree are corresponding with measured values

given in the literature. More calculations and experiments should be performed

to estimate the accuracy of calculated strength and shrinkage.

220


REFERENCES

[1] Maekawa, K.; Chaube R., Kishi, T.: Modelling of concrete performance,

E&FN Spon, London, 1999

[2] Ožbolt, J.: MASA – Macroscopic Space Analysis. Internal Report, Institut

für Werkstoffe im Bauwesen, Universität Stuttgart, 1998

[3] Zement Taschenbuch 2000, Verlag Bau+Technik, Düsseldorf, 2000


CORROSION INDUCED FAILURES OF PRESTRESSING STEEL ... · CORROSION INDUCED FAILURES OF PRESTRESSING...

Documents

Transcript of CORROSION INDUCED FAILURES OF PRESTRESSING STEEL ... · CORROSION INDUCED FAILURES OF PRESTRESSING...