PROPERTY RELATIONSHIPS USED IN EN 338 STRENGTH GRADES OF NORWAY SPRUCE STRUCTURAL TIMBER Ren Steiger1, Martin Arnold1, Robert Jockwer1,2 ABSTRACT: For structural use solid timber has to be strength graded. In order to stay economic, the grading process usually focuses on the most important physical and mechanical properties: density, bending strength (MOR) and modulus of elasticity (MOE). Grading standards also state how further mechanical properties can be derived from these basic property values by empirical relationships. The aim of this study was to review some of these property relationships as a contribution to a further development of strength class systems and grading standards. Based on mechanical tests of Norway spruce structural timber with different cross-sections, the following topics were evaluated with respect to the European strength class system (EN 338): a) simultaneous matching of basic values: characteristic values of density, MOR and mean MOE parallel to the grain, b) relationship of characteristic values of tension and compression strength parallel to the grain with respect to the corresponding characteristic value of MOR, c) ratio of 5th-percentiles and mean value of density and MOE, and d) ratio of MOE in bending, compression and tension. With regard to machine strength grading, the correlation of (flexural) MOE and MOR as well as the potential of dynamic MOE derived from bulk density and ultrasonic wave speed as an indicating property for strength were evaluated. Over all the test results confirmed the property relationships and strength class profiles regarding MOR, tension strength, compression strength, density and MOE given in EN 338. KEYWORDS: strength grading, characteristic values, property relationships, dynamic MOE, ultrasound, EN 338 1 INTRODUCTION 12 1.1 THE EN 338 STRENGTH CLASS SYSTEM For structural uses solid timber has to be strength graded according to grading standards which provide a strength class system to assign timber samples to specific strength classes. In order to stay simple and economic the grading process usually focuses on the most important physical and mechanical properties: density, bending strength (MOR) and modulus of elasticity (MOE) parallel to the grain. To reflect the variation of mechanical properties, the classification is based on so-called characteristic values, which are 5th-percentile or mean values, respectively. In Europe the classification of structural timber is carried out according to a set of three linked standards: Test methods to determine mechanical properties as

well as dimensions, moisture content (MC) and wood density of test pieces are specified in EN 408 [1].

1 Ren Steiger, Martin Arnold, Robert Jockwer, Empa, Swiss Federal Laboratories for Materials Testing and Research, Wood Laboratory, Ueberlandstrasse 129, CH-8600 Dbendorf, Switzerland Email: rene.steiger@empa.ch , martin.arnold@empa.ch , robert.jockwer@empa.ch 2 Robert Jockwer ETH Zrich, Institute for Structural Engineering, Steel, Timber and Composite Structures, Wolfgang-Pauli-Strasse 15, CH-8093 Zrich, Switzerland

Characteristic values of mechanical properties and density are derived from the test data according to EN 384 [2].

The timber is finally classified in strength grades according to EN 338 [3]. This standard defines twelve strength classes for softwoods. A specific sample can be assigned to a certain strength class, if the characteristic values of density and MOR (both of them are 5th-percentiles) as well as MOE (mean value, usually derived from bending tests) match or exceed the values of the desired class. Additional mechanical properties needed for the design of timber structures are derived from these basic values by empirical relationships.

1.2 BASIC CHARACTERISTIC VALUES AND

PROPERTY RELATIONSHIPS The decisive parameters in strength grading (density, MOR, MOE parallel to grain) are called "basic" values. Density correlates well with many mechanical properties and is e.g. used as basis for calculating characteristic values of compression strength perpendicular to the grain and of embedding strength. Characteristic values of MOR fm,k are needed along ultimate states design. They have to be derived from bending tests. Bending MOE Em is primarily used in verification of serviceability limit states. It can be determined either by static bending tests or by dynamic methods (vibration, ultrasound) [4, 5].

1.3 CHARACTERISTIC VALUES OF TENSILE AND COMPRESSION STRENGTH PARALLEL TO THE GRAIN

Respective values of tension and compression strength parallel to grain ft,0,k and fc,0,k can be determined by tests as well or may be derived from characteristic MOR fm,k using empirical relationships. According to EN 384 the characteristic values of tensile strength parallel to the grain ft,0,k for softwood species can be calculated with Equation (1):

kmkt ff ,,0, 6.0 (1)The characteristic values of compression strength parallel to the grain fc,0,k for softwood species are derived with Equation (2) from the corresponding MOR fm,k:

45.0,,0, 5 kmkc ff (2) 1.4 RATIO OF FIFTH PERCENTILES AND

MEAN VALUES OF DENSITY AND MOE In EN 338 the in-grade characteristic value of density is derived from the grade's mean value assuming a specific coefficient of variation (COV). The ratio of the characteristic to the mean value of density given in EN 338 varies between 0.82 and 0.85 with an average of 0.84. This variation results from rounding. Former editions of EN 384 explicitly assumed a constant ratio of 0.84 for softwood species, which in case of a normally distributed sample is identical to a COV of 10%. 5th-percentiles of MOE E0,05 are used in calculations where member stiffness takes direct influence on member strength, e. g. overall structural stability, buckling and lateral torsional stability. EN 384 and EN 338 set a constant COV of 20% for variability in MOE. Hence, the ratio of 5th-percentile (E0,05) to mean value (E0,mean) of MOE assuming normal distribution is:

E0,05/E0,mean = 0.67 for softwood species. (3)

1.5 RATIO OF MOE IN BENDING, TENSION

AND COMPRESSION PARALLEL TO GRAIN EN 338 does not differentiate between MOE depending on the type of loading (bending (Em), tension (Et), compression (Ec)) and gives one single value E0. 1.6 AIM OF THE STUDY The aim of this study was to review some property relationships based on recently compiled large data sets as a contribution to a future revision of the grading standards. Based on several series of tests with Norway spruce structural timber, the following property relationships were analysed: relationship of bending MOE and MOR as derived by

dynamic and static testing simultaneous matching of basic values: characteristic

values of density and MOR and mean MOE parallel to the grain

relationship of characteristic values of tension and compression strength parallel to the grain with respect to the corresponding characteristic value of MOR

ratio of 5th-percentiles and mean value of density and MOE

ratio of MOE in bending, compression and tension. 2 MATERIAL AND METHODS 2.1 DATA SETS Three data sets consisting of a total of 1644 specimens from different extensive research projects referenced (1) to (3) served as data base: 1. In the early 1990s a test program was launched at

the ETH Zurich, which aimed to evaluate the mechanical properties of Swiss grown Norway spruce structural timber [6-8].

2. The second source of data is a research project on storm damages observing wind-induced compression failures [9]. Property relationship analysis of this data source was carried out using data resulting from reference tests on specimens without any storm damages.

3. A third source of data is a PhD thesis [10] dealing with the influence of long-term log storage in the forest on the mechanical properties of timber.

Detailed information on sample sizes, dimensions of test pieces and data pooling are given in [11]. 2.2 TEST PROCEDURES All 3 studies (1)-(3) used similar test procedures. Testing was carried out in accordance with EN 408 and the characteristic values were determined according to EN 384. The critical section was always positioned in such way that it was subjected to maximum load. For the bending tests the tension edge was selected at random. Tension tests were performed on full cross-section timber with special test equipment [12]. Compression tests were carried out with lateral restraints at half-length to prevent the specimens from buckling. Wood moisture content (MC) was measured by the electric resistance method. Density was determined from the mass and the volume of the whole specimen. The bulk density values were adjusted to the density of small defect-free prisms by dividing by 1.05 [2, 13]. 2.3 ASSESSMENT OF TIMBER QUALITY The MOE, regarded as the best direct parameter to estimate strength, was used as an indicator for timber quality. Instead of the static MOE, dynamic MOE Edyn was used. Edyn was calculated from the velocity v of an ultrasonic wave passing the specimen longitudinally and from the specimens bulk density r. Such measurements, originally described by Goens [14], have been successfully used for assessing the stiffness properties of structural timber [4, 5]. Edyn was used to group test values according to timber quality. Grouping always aimed to get at least 40 specimens in each group. Usually groups contained more than 60 80 specimens. 2.4 ADJUSTMENT OF DATA TO REFERENCE

CONDITIONS According to EN 384, the reference MC shall be consistent with 20C and 65% relative humidity, which

for most softwoods corresponds to a MC of about 12%. For specimens with a MC in the range of 10 to 18%, an adjustment of values to 12% MC was made (EN 384). Bending and tensile strength values assessed not according to the geometrical reference conditions (height, width and length of member) requested by EN 384 were adjusted using factors given in the respective standard. 3 RESULTS AND DISCUSSION 3.1 ASSESSMENT OF TIMBER QUALITY The correlation of MOR fm and bending MOE Em from static bending tests is shown in Figure 1. The coefficient of correlation amounts to R = 0.66. Although this is lower compared to other studies, the bending MOE can be taken as a reliable indicator for MOR.

y = 0.003x + 10.6R = 0.66 (R 2 = 0.43)n = 525

0

10

20

30

40

50

60

70

80

90

0 2000 4000 6000 8000 10000 12000 14000 16000 18000 20000Bending MOE E m [N/mm2]

Ben

ding

stre

ngth

fm

[N/m

m2 ]

Regression linelower 5% confidence limit

+ Test values EN 338 strength class values f m,k and E 0,mean

Figure 1: Correlation of static bending MOE Em with MOR fm. The dots indicate pairs of characteristic values (fm,k; E0,mean) as given by EN 338

Dynamic MOE Edyn is known to be a good indicator of actual stiffness (MOE) and strength of timber in bending, tension and compression. As two examples, the correlation of Edyn with static bending MOE Em and MOR fm is shown in Figures 2 and 3.

y = 0.8586x - 1357R = 0.77 (R 2 = 0.60)

n = 668

0

5000

10000

15000

20000

25000

0 5000 10000 15000 20000 25000Dynamic MOE E dyn [N/mm

2]

Stat

ic M

OE

in b

endi

ng

Em

[N/m

m2 ]

Figure 2: Correlation of dynamic MOE Edyn with static MOE in bending Em

y = 0.0034x - 5.25R = 0.62 (R 2 = 0.39)

n = 522

0

10

20

30

40

50

60

70

80

90

100

0 5000 10000 15000 20000 25000Dynamic MOE E dyn [N/mm

2]

Ben

ding

stre

ngth

f m

[N/m

m2 ]

Figure 3: Correlation of dynamic MOE Edyn with MOR fm

Further correlations are depicted in [11]. Hence, it can be concluded, that MOE can be estimated by a much simpler dynamic measurement (together with density) instead of carrying out static bending tests. Static and dynamic MOE show a strong correlation with R = 0.77 to 0.9. Between dynamic MOE and (bending, compression, tension) strength the correlation is weaker (R = 0.59 0.72) but still reliable enough to use dynamic MOE as one of the indicators for timber strength. 3.2 SIMULTANEOUS MATCHING OF BASIC

CHARACTERISTIC VALUES OF DENSITY AND MOR AND MEAN MOE

According to measured dynamic MOE Edyn, the data set derived from bending tests (sample size n = 522) was divided into 6 equally sized groups. The characteristic values of MOR and density were calculated for each group and compared to the code values given in EN 338. Figure 4 shows that the test data match the strength class profile given by EN 338 well. For strength classes with MOE > 13000 N/mm2 the tests exhibit somewhat lower characteristic values of bending strength fm,k than assigned by EN 338. Still, based on our tests results, there is no reason for changing in-grade discrimination of basic characteristic values fm,k, E0,mean and rk.

0

10

20

30

40

50

60

70

80

90

100

0 2000 4000 6000 8000 10000 12000 14000 16000 18000 20000

Bending MOE E m [N/mm2]

Cha

r. v

alue

of b

endi

ng st

reng

th f m

,k [N

/mm

2 ]

0

50

100

150

200

250

300

350

400

450

500

Cha

ract

erist

ic v

alue

of d

ensit

y r

k [k

g/m

3 ]

fm,k test values with linear trend linefm,k EN 338 valuesrk test values with linear trend linerk EN 338 values

Figure 4: Comparison of basic strength class values, rk, E0,mean and fm,k and test values (with linear trend lines)

3.3 RELATIONSHIP BETWEEN CHARACTE-RISTIC VALUES OF TENSION AND COMP-RESSION STRENGTH PARALLEL TO GRAIN AND CORRESPONDING CHARAC-TERISTIC VALUE OF MOR

The data for the comparison of bending and tensile or compression strength was grouped according to the dynamic MOE Edyn with equal class boundaries for both bending and tensile strength data sets. The ratio of characteristic values of tensile to MOR is shown in Figure 5. The ratio obviously is not constant but rather depends on timber quality. Ratios of mean values and characteristic values exhibit the same trend. Assuming a constant ratio of ft/fm = 0.6 according to Equation (1) gives safe estimates for characteristic MOR values fm,k 22 N/mm2 (EN 338 strength classes C22 and higher). For fm,k < 22 N/mm2 however, as it results from extrapolation of the trend line, the ratio derived from the test results in the present study is smaller than 0.6.

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

10 20 30 40 50 60

Bending strength f m,k [N/mm2]

Rat

io f

t / f m

Test values with linear trend line: char. level

Test values with linear trend line: mean level

EN 338: ft,k/fm,k = 0.6

Figure 5: Ratio of MOR fm and parallel to the grain tensile strength ft,0

Figure 6 depicts the relationship between parallel to the grain compression strength fc,0 and MOR fm. Compared to our test results the EN approach (Equation (2)) is more conservative for characteristic MOR values fm,k 24 N/mm2 (EN 338 strength classes C24). The trend line fitting the test results progresses equally on the mean and on the characteristic level. An extrapolation of the trend to characteristic MOR fm,k < 24 N/mm2 however, results in lower fc,0,k values compared to the characteristic values assigned by Equation (2) and thus Equation (2) overestimates fc,0,k in strength classes lower than C24. EN 338 values as well as test values (4 groups graded according to Edyn) exhibit a linear relationship between the compression strength parallel to the grain fc,0 and the wood density r (Figure 7). The slope of the regression line found in our tests however is almost twice the slope of the trend line fitted to the code values. This confirms again the overestimation of fc,0,k by Equation (2) for fc,0,k < 20 N/mm2. As an alternative to the code approach (Equation (2)), the compression strength parallel to the grain could be derived from the wood density using a linear model.

(2) y = 2.56x -0.34; R 2 = 0.88

(1) y = 2.65x -0.35; R 2 = 0.93

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

20 30 40 50 60Bending strength f m,k [N/mm

2]

Rat

io f

c / f

m

EN 338: f c,0,k /f m,k = 5(f m,k )-0.55

Test values with power trend line(1): char. level

Test values with power trend line (2): mean level

Figure 6: Ratio of MOR fm and parallel to the grain compression strength fc,0

(3) y = 0.077x - 6.46; R 2 = 0.99

(2) y = 0.13x - 22.8; R 2 = 0.96

(1) y = 0.14x - 30.6; R 2 = 0.98

0

5

10

15

20

25

30

35

40

200 300 400 500 600

Wood density r [kg/m3]

Com

pres

sion

stre

ngth

f c,0

[N/m

m2 ]

Test values with linear trend line (1): char. level

Test values with linear trend line (2): mean level

EN 338 values with linear trend line (3)

Figure 7: Relationship between compression strength parallel to the grain fc,0 and wood density r: Test results and EN 338 values

3.4 RATIO OF 5TH-PERCENTILE AND MEAN VALUE OF DENSITY AND MOE

The ratio between characteristic and mean value of density (MC = 12%) of all 1640 specimens results in rk/rmean = 0.84. The mean value is 450 kg/m3 and the COV amounts to approximately 9.7%. The normal probability plot (Figure 8) confirms the correctness of assuming a normal distribution for density. The test results are in line with the code procedures of EN 338 and EN 384 (COV = 10% and ratio rk/rmean = 0.84). The ratio of characteristic to mean value of MOE calculated based on all results of bending tests (n = 668) yields in E05/Emean = 0.70. The COV is 18.4% and a normal distribution fits the data reasonably well (Figure 9). These results confirm the ratio and COV given in EN 338 and EN 384 which are 0.67 and 20% respectively.

250 300 350 400 450 500 550 600 650.01

.1

1

5102030

50

70809095

99

99.9

99.99

Density r [kg/m3]

[%]

n = 1640

Min = 321 kg/m3

Max = 623 kg/m3

Mean = 450 kg/m3

Stdev = 43.5 kg/m3

COV = 9.68 %

rk = 378 kg/m3

rk/r

mean = 0.84

Figure 8: Normal probability plot of density r (MC = 12%)

5000 10000 15000 20000.01

.1

1

510203050

70809095

99

99.9

99.99

MOE in bending Em

[N/mm2]

[%]

n = 668

Min = 4823 N/mm2

Max = 19226 N/mm2

Mean = 13111 N/mm2

Stdev = 2409 N/mm2

COV = 18.4 %

E05

= 9150 N/mm2

E05

/Emean

= 0.70

Figure 9: Normal probability plot of bending MOE Em (MC = 12%)

3.5 RELATIONSHIP BETWEEN MOE IN BENDING AND IN COMPRESSION OR TENSION PARALLEL TO THE GRAIN

Figures 10 and 11 show parallel to the grain tensile MOE Et or compression MOE Ec respectively versus bending MOE Em.

y = 0.808 x + 2523R = 0.91; R 2 = 0.83

n = 109

5000

10000

15000

20000

25000

5000 10000 15000 20000 25000

MOE in bending E m [N/mm2]

MO

E in

tens

ion

Et [

N/m

m2 ]

Figure 10: Bending MOE Em versus parallel to the grain MOE in tension Et for Norway spruce structural timber with cross-section 80/160 mm

y = 0.745 x + 3357R = 0.85; R 2 = 0.72

n = 81

5000

10000

15000

20000

25000

5000 10000 15000 20000 25000

MOE in bending E m [N/mm2]

MO

E in

com

pres

sion

Ec

[N/m

m2 ]

Figure 11: Bending MOE Em versus parallel to the grain MOE in compression Ec for Norway spruce structural timber with cross-section 80/160 mm

Based on the respective regression equations shown in Figures 10 and 11 MOE values Et and Ec and ratios Et/Em and Ec/Em were calculated and compared to MOE values given in EN 338 for the whole strength class spectrum C14 to C50 (Figure 12). On average, bending MOE derived according to EN 408 4-point bending test differed only by 1% from MOE in tension or in compression. However, the ratios are not constant; they depend on timber quality. Timber members assigned to classes C35 and C40 are free of defects. That is why such members behave similar to small clear specimens resulting in identical MOE Et, Ec, Em.

0.80

0.90

1.00

1.10

1.20

1.30

6000 7000 8000 9000 10000 11000 12000 13000 14000 15000 16000

MOE in bending E m [N/mm2]

MO

E r

atio

Et,0

/ E

m o

r Ec,

0 / E

m

Ratio Ec,0/EmRatio Et,0/Em

C16

C24

C30

C40

Figure 12: MOE ratios: tension/bending and compression/bending

Figure 13: Sawing pattern of interaction test specimens

The relationship of MOE in bending, tension or compression is known to also depend on the sawing pattern. The specimens in our tests were free of pith-

associated wood (Figure 13) and tested in edgewise position. Heartwood may lead to other ratios of MOE. To account for the influence of sawing pattern, timber has to be modelled as a composite material with zones of different MOE within cross-section [15]. 4 CONCLUSIONS The presented study led to the following conclusions: Overall, the test results confirmed the property

relationships and strength class profiles regarding MOR, tension strength, compression strength, density and MOE given in EN 338.

For strength classes with MOE > 13000 N/mm2 the tests exhibited somewhat lower characteristic values of bending strength fm,k than assigned by EN 338. Still, based on our tests results, there is no reason for changing in-grade discrimination of basic characteristic values fm,k, E0,mean and rk.

Static bending MOE was proven to be a good (single) indicator for MOR, although the respective coefficient of variation (R = 0.66) was lower compared to other studies.

The good correlation between dynamic MOE and static MOE (bending, tension, compression) as well as strength (MOR, tension / compression parallel to the grain) recommends using ultrasonic wave speed together with density to assess timber quality.

Tensile to MOR ratios were found not to be constant, but to depend on timber quality. Ratios for characteristic MOR fm,k 22 N/mm2 (EN 338 classes above C22) amounted to 0.6 0.75 but were < 0.6 for lower quality timber. However, for design practice the use of a conservative value of 0.6 is reasonable.

The relationship between the characteristic values of compression and MOR parallel to the grain given in EN 384 is more conservative for higher quality timber than our test results indicate. As an alternative to the existing approach, compression strength parallel to the grain could be derived from wood density using a linear model. Doing so, the importance of density as a classification criterion would be increased.

The calculated fractile ratios for density rk/rmean and for MOE parallel to the grain E0,05/E0,mean were 0.84 and 0.70. These values correspond well to the ratios given in EN 384 and EN 338 (0.84 and 0.67) assuming a COV of 10% and 20% respectively.

The ratios of MOE of Norway spruce structural timber in tension, compression and bending differed depending on timber quality. Differences up to 9% between tensile and bending MOE were found. Regarding compression MOE, the maximum difference to the bending MOE was 12%. For normal quality timber of classes C24 to C30 (mostly used in practice), the differences between Et, Ec and Em are not higher than 6%. With regard to a simple design process, the current practice of using one single MOE value should therefore not be changed. But when assigning timber populations to strength classes based on tension MOE instead of bending MOE, the differences have to be taken into account.

REFERENCES [1] Comit Europen de Normalisation CEN: EN 408:

Timber structures - Structural Timber and glued laminated timber - Determination of some physical and mechanical properties. 2003.

[2] Comit Europen de Normalisation CEN: EN 384: Structural timber - Determination of characteristic values of mechanical properties and density. 2004.

[3] Comit Europen de Normalisation CEN: EN 338: Structural timber - Strength classes. 2003.

[4] Hearmon R.F.S.: Vibration testing of wood. Forest Products Journal, 16(8):29-40, 1966.

[5] Kollmann F. and Krech H.: Dynamische Messung der elastischen Holzeigenschaften und der Dmpfung. Holz als Roh- und Werkstoff, 18(2):41-54, 1960.

[6] Steiger R.: Biege-, Zug- und Druckversuche an Schweizer Fichtenholz, Institut fr Baustatik und Konstruktion IBK, ETH Zrich, Vol. 207, Birkhuser Verlag Basel, 1995.

[7] Steiger R.: Versuche an Fichten-Kanthlzern: Biegemoment - Normalkraft - Interaktion, Institut fr Baustatik und Konstruktion IBK, ETH Zrich, Vol. 209, Birkhuser Verlag, Basel, 1995.

[8] Steiger R.: Mechanische Eigenschaften von Schweizer Fichten-Bauholz bei Biege-, Zug-, Druck- und kombinierter M/N-Beanspruchung, Institut fr Baustatik und Konstruktion IBK, ETH Zrich, Vol. 221, Birkhuser Verlag, Basel, 1996.

[9] Arnold M. and Steiger R.: The influence of wind-induced compression failures on the mechanical properties of spruce structural timber. Materials and Structures, 40(1):57-68, 2007.

[10] Mischler-Schrepfer V.: Der Einfluss der Waldlagerung von Fichten-Rundholz auf die Lngs-Zugeigenschaften des Schnittholzes. Ph.D. thesis, ETH Zrich, Switzerland, 2000.

[11] Steiger R. and Arnold M.: Strength grading of Norway spruce structural timber: Revisiting property relationships used in EN 338 classification system. Wood Science and Technology, 43(3-4):259-278, 2009.

[12] Steiger R., Arm H., and Gehri E.: Einspannvorrichtung fr Zugversuche an Holzproben grsseren Querschnitts, Institut fr Baustatik und Konstruktion Eidgenssische Technische Hochschule etc. Zrich, 1994.

[13] International Organization for Standardization ISO: ISO 3131 (1975-11): Wood: Determination of density for physical and mechanical tests. 1975.

[14] Goens E.: The determination of the elasticity module of rods with the help of bending fluctuations. Annalen der Physik, 11(6):649-678, 1931.

[15] Gehri E.: CIB-W18/30-6-3: Timber as natural composite: explanation of some peculiarities in the mechanical behaviour - Case: Assessment of the modulus of elasticity of timber parallel to grain. In: Proceedings of CIB-W18 Meeting Thirty, 1997.