Failure of polyethylene gas pipe materials

N. Brown and X. Lu Department of Materials Science, Engineering, University of Pennsylvania, Philadelphia, PA, USA

Abstract Two of the pipe resins that are commonly used in the USA are compared with respect to their relatiue life times. The improvement in the performances of the currently used materials compared to the older material from the sameproducer have been measured The laboratory conditions for producing brittle failure in notched specimens in the minimum time are presented

Introduction This paper has several purposes: o) to compare the lifetime of currently used polyethylene gas pipe materials; (2) To show the conditions for prod- ucing brittle failure in the min imumm testing times; (3) to show how the performance of resins from the same resin producer have changed with time; and (4) to show how to predict the effect of temperature on the reIative lifetimes.

Experimental Materials The resins from two of the most prominent gas pipe resin producers in the USA were investigated. Through the years the performances of their resins have improved. From each producer, an old resin and recently used resin were compared. The materials are described in Table !. Resins from Producer I were not available so only material from pipe could be tested.

Extensive tests were done on slow cooled compression molded resin from Producer ]] because this state gives very reproducible results and gives a standard baseline from which to compare all of the other materials. Tests on specimens from pipe from Producer ii were also made. Test Method After extensive tests with different specimen geometries, the single edge notched specimens in Figure 1

Table ! Materials Investigated

Designation** History Mw Mn Density Co-Polymer

IO Pipe Mfr. 0.940 Ethylene-Butene 1965 taken ~4.4 ethyl branches out of 1000C ground 1987

IN New pipe, 0.936 Ethylene-Octene Mfr. 1987

IIO Resin Mfr. before 1980 0.935 Ethylene-Hexene Compression~170,000~15,000 4.5 branches/10O0C motded, slow cooled

IIN Resin Mfr. 1985 0.935 Ethylene-Hexene Compression 170,000 15,000 4.5 butyl branches molded, slow 10OOC cooled

**Producer I; Producer II

5 0 m m

2 5 m m

1 j s,.eoo,c. l O m m . _ L _ _ J 1 mm

I l a o = 3 . 5 m m ~ao

a b

Fig I a) Geometry of compression molded specimen b) Geometry of specimen from pipe.

44 CONSTRUCTION & BUILDING MATERIALS Vol. 3 No. 1 MARCH 1989

were used because ~1~ the fracture is plane strain; ¢2~ the ratio of notch depth to specimen thickness of 1/3 mini- mizes the time to failure without introducing excessive creep of the ligament; ~3~ the side grooves enhance plain strain fracture as occurs in service and also reduces the time to failure. The notch was made by very slowly pressing a fresh razor blade into the specimen.

The kinetics of the failure process were obtained by measuring with a microscope the notch opening dis- placement as a function of time as shown in Fig. 2. The specimens were exposed to a constant load and the duration of the tests ranged from 10 minutes to over I year. The test temperatures ranged from 80 to 23°C.

Results Typical Data Typical data for brittle fracture is represented in Fig. 3 where the COD is plotted against time at a given temperature and stress. The initiation of crack growth occurs at the point where the curve begins to accelerate. In general, the time for crack initiation is on the order of 60% of the lifetimes as previously reported by Dickinson and Ewing ¢1) for pipe material used in the UK. Note that at 80°C the lifetime of the specimen from new pipe of Producer II is about 100 greater than from new pipe of Producer I. Lifetime of llN-Ethylene.Hexene Copolymer The curves of stress versus lifetime for material IIN, compression molded and slow cooled, at various temp- eratures are shown in Fig. 4. The behaviour is typical for all polyethy- lenes in that the ductile range of failure shows a shallow slope and the low stress brittle region has a much steeper slope. The nose in the curve is the transition region where anoma- lous behaviour occurs in that a decrease in stress produces a decrease in lifetime. Detailed microsopic observ- ations by Lu and BrownC2~ of the damaged zone at the root of the notch shows that extensive crack blunting occurs in the ductile region. In the transition region from blunting is reduced by the decrease in stress so that the local stress concentration is increased; thus, the lifetime decreases with the decrease in stress. Once in the brittle region, the sharp- ness of the notch does not change

0

1 I J 1 1

Fig 2 Experimental method for measuring the notch opening.

E

{..9 Z

z w

0

- r

0 z

TIME (rain)

0 2 0 0 4 0 0 6 0 0 8 0 0 1000 , J i i ~ 1 7 3 9 ) i i i

Z N PIPE 1 5 0 0 - 1 f : 1 1 8 2 0

1 0 0 0 -

5 0 0

X

J o

!" /

I 0 L t v ! t I t I I

0 2 0 4 0 6 0 80 T I M E (103 min)

100

Fig 3 Notch opening displacement versus time at 80°C and 2.4 MPa for new pipe from Producer I and 11. t r is lifetime.

greatly so that the reduction in the applied stress monotonically increases the lifetime.

Note that the slope of the curves in the brittle region vary with temp- erature as pointed out previously by Ayres ¢3~. This point is significant because current suggestions¢4.5~ for

modifying the specifications for qual- ifying resins, ASTM D2837, are based on the assumption that the slope in the brittle region is independent of temperature.

It takes a long time to produce brittle failure in Resin IIN even with our notched specimen. When this

C O N S T R U C T I O N & B U I L D I N G M A T E R I A L S VoI. 3 No. 1 M A R C H 1 9 8 9 4 5

resin is tested by the conventional ASTM D2837 test in the form of unnotched pipe under an internal pressure, it has not been possible to obtain brittle failure at 80 ° in tests that have lasted for two years (Ayres (3)). It is interesting to show the test cond- itions for producing brittle failure in our specimens in the shortest time. Figure 5 shows the stress as a function of temperature that will produce brittle failure in the shortest time. When this curve is extrapolated to 23°C, the maximum stress that can be used to produce purely brittle failure is 6 Mpa. A higher stress would produce ductile failure. Figure 6 shows the dependence of minimum lifetime for brittle failure on temp- erature. When this data is extra- polated to 23°C, the lifetime would be about 30 years at 6 MPa. This result only relates to our specimen geomeby with a 3 .5mm notch as produced bya razor blade. In order to substantially reduce the time to produce brittle failure in this material it is necessary to use an environmental agent such as Igepal or to do a fatigue test. Lifetime of lN.Ethylene-Butene Copolymer Fig. 7 shows stress versus lifetime for IN specimens from 4" diameter pipe. The brittle failure region is empha- sized. With this material it is possible to produce brittle failure in 10 hrs. at 80°C and in 15 days at 23°C. Comparison of Lifetimes I0, IN, !I0, and IlN at 80 and 23°C Fig. 4 shows stress versus lifetime for the above materials in the brittle region. In order to determine the relative lifetimes it is necessary to choose a specific stress level since the curves have somewhat different slopes. In comparing IO and IN, it is seen that at a stress of 1 MPa the lifetime of IN is 19 times greater than IO at 80°C and at a stress of 4 MPa the lifetime was improved by a factor of 12 at 23°C. In comparing IIO and ilN, at 1.5 Mpa at 80°C, the lifetime of lln is 90 times that of IIO; at 23°C it is not possible to determine the change between IlN and IIO because the IIN test has yet to fail in over 1 year. However, the extrapolated data from Figures 4 and 5 show IlN would fail at 23°C under 6 MPa in 30 years, whereas ilO should fail at 23°C under 6 MPa in about 3 days. The producers have made a significant improve- ment in their resins over the period of about 15-20 years.

I i

i i i i l l I

.1.

I 1 I I I I I

I0 4

TIME

i i i i I i J | I i I i i i l i TR -418

(3 o= 3.5 rnm

RT ,,k

~.o<~ . t~.,~,. '~"-,-.5 o" c

• ~ 8 0 ° C

I I I I I I I I I I I I I I I I

10 5 I 0 6

TO FAILURE (rain)

Fig 4 Stress versus time to failure for ethylene-hexene co-polymer-II compression molded and slow cooled, at various temperatures.

,oo f j 8o

re- 6 0 . J F- < r r \ \ LI.I \ 13- 4,0 • UPPER LIMIT OF BRITTLE \ \ - 2; FAILURE \ ', uJ \ \ I'-- o LOWER LIMIT OF DUCTILE \ x

- FAILURE \ x

20

\ \ x \ \ \

1 I I I I I ~

/

2 3 4 5 6 7 8

STRESS (MPa)

Fig 5 Temperature versus maximum stress for brittle failure(..) and versus minimum stress for ductile failure (o o) (from Fig. 4).

~ 1 0 0 o

~ 5 0

F- < n~ b.I

I 0 4

o LOWER LIMIT OF DUCTILE FAILURE • UPPER LIMIT OF BRITTLE FAILURE

I I ! I I T I I I I I I I I f ' ! I ! ! I I ! f !

10 5 10 6 10 7

T I M E TO F A I L U R E {rn in)

Fig 6 Temperature us. minimum time for brittle failure ( . . ) and us. maximum time for ductile failure (oo) (from data in Fig. 4)

46 CONSTRUCTION & BUILDING MATERIALS Vol. 3 No. 1 MARCH 1989

I.O

"" 5 :E

(/3 (/3

I.O I-

0.5

i i i I I I | I i I I I l l l l l I I I i l l ; , i I I I I J i l l l j | I

o ° + o

~ O o C

0.2 , r t ;,ti+Irl ! ! r ~IIIIl I t + i t , , t l + + , '~, , , ,! 10+ 102 10 3 10 4 iO s

TIME TO FAILURE (rnin)

Fig 7 Stress versus time to failure for ethylene-butene co-polymer I from new 4 inch pipe at various temperatures.

I O ~ ' - ~ - - - ° ~ o - - - - . ~ . ---nIN --;fIN v--*

L , A . , - s o ' c I o,o -Wc Oi I i i i ~ t t l I I I t l l l t l t i i t i i l l l i t l i t

"10 z" 10 3 104 105 'I' 6

T IME TO FAILURE (min)

Fig 8 Stress versus time to failure for the four materials listed in Table I at 80°C and 23°C

~ - ~ CM-SC /4" PIPE

20001 I , / E r I II z3o6= iN) /

I ~t [ [ 2.4 mPo ~

/

o ooot ] o I [ / / / FA,LORE AT Z F ? IX y 1 9 6 0 0 0 '

I / J K _ . x ~ ' .. x x v 1 ~

O | I I I I I 1 0 40 80 120

TIME ( |03ra in)

Fig 9 Notch opening versus time for (ooo) compression molded slow cooled resin II and (xxx) pipe IIN material.

In comparing the lifetimes of IN and IIN at 23°C and 6 MPa, the data indicate that IN would last 17 days versus 30 years for IIN. Based on the data in Figs. 4-7, the ratio of lifetimes of IN and fIN are not expected to vary appreciably with the stress level. Effect of Temperature on Lifetime The effect of temperature on lifetime was measured at a constant stress in the brittle region. All data obeyed the following equation:

t~ = A e (;VRT where A depends on the material and stress and Q hardly depends on stress, R, the gas constant ---= 8.31 J-moll K -' and T is the absolute temperature. The value of Q are as follows:

Material Q(KJ/mol) IO 105 IN 83 IIN 119

Here is an example of the utilization of the above data. Consider a south- ern gas company where the average ground temperature is 22°C who uses a pipe with a Q = 105 KJ/mol and assume that this company exper- iences significant leakage in their piping system 20 years after instal- lation. After what period of time will a northern company where the average ground temperature is 11°C exper- ience significant leakage if they use the same pipe and experience the same type of ground stresses? Signifi- cant leakage in the northern comany should occur in: tf = 20 d l°5~°/Sa'xlr~" ,r~5) = 105 years Comparison of Pipe and Compression Molded Material The processing of the resin into pipe has an important influence on the lifetime as emphasized by Marshall oL Fig. 9 shows that the fIN pipe has about 4 times the lifetime of IlN- compression molded and slow cooled. Fig. 10 shows that IN pipe has about 4 times the lifetime of IN resin that was compression molded and slow cooled. Both residual stresses and differences in morphology are respon- sible.

Comparison of IN and IIN in the section on "Comparison of Lifetimes..." should now be modified so that IN pipe and IIN pipe are compared. The ratio of their lifetimes at 23°C are estimated to be about 600 for a stress of 6 MPa.

CONSTRUCTION & BUILDING MATERIALS VoL 3 No. 1 MARCH 1989 47

E

<.9 z 7 I,I I:L 0

F- O 7

2 5 0 0

2 0 0 0 -

1500 -

1000 -

5 0 0 ~

01 I 0 100

I 1 1

tf = 1 7 6 /

I N { C M - S C )

N P I P E -

r

2 3 0 6 I (N) 8 0 ° C 2 .4 M Pa

t I 1 I I 1 2 0 0 3 0 0 4 0 0 5 0 0 6 0 0 7 0 0

T I M E (rain)

Fig I 0 Same as 9 for IN material (o o o) compression molded s low cooled (° ° °) new pipe.

Discuss ion Our results are based on the single edge notched tension specimen geometries shown in Figures 1 a and

b. The absolute value of the lifetimes will vary with specimen geometry and notch depth. However, it is contended that if the deformation zone at the

notch tip is small compared to the dimensions of the specimen and if plane strain fracture is produced, then the ratio of the lifetimes of the various materials will be relatively independent of the test geometry and the materials will be ranked in the same order.

As was so clearly presented by MarshalP~, it is important to test specimens with a well defined notch geometry in order to properly deter- mine the relative lifetime. If a test lasts too long before brittle failure is observed, because the specimen is not notched, then the test has limited value.

References 1. Dickinson A and Ewing L, Int. Gas

Research Conf. Toronto, pub. GRI 1986, VoL 1, Pfi2.

2. I_u X and Brown N, Polymer Preprints pub. Am. Chem. Soc . - Conf. on LA, September 1988.

3. Ayres R L, Tenth Plastic Fuel Gas Pipe Symposium in New Orleans, pub. AGA 1987, P206.

4. t4ruk S M, Ninth Plastic Fuel Gas Pipe Symposium, pub. AGA 1985 p202.

.5. PaIermo E F and DeBlieu I K, ibid, p215. 6. Marshall G P - Private communication. 7. Marshall G P and Birch M W - Sixth Int.

Conf. Plastic Pipe 1985, Paper 18.

A c k n o w l e d g e m e n t s This research was supported by the Gas Research institute and the U.S. Department of Energy. N. Brown is solely responsible for the contents of this report.

This paper was presented at the Plastics VII conference, September 1988, and is available in the Proce- edings Plastics VII published by the Plastics & Rubber Institute.

48 CONSTRUCTION & BUILDING MATERIALS Vol. 3 No. 1 MARCH 1989