25 Test Method Cyclic crack growth tests with CRB specimens for the evaluation of the long-term performance of PE pipe grades.pdf

ARTICLE IN PRESS

POLYMER

TESTING

Polymer Testing 26 (2007) 180–188

www.elsevier.com/locate/polytest

Test Method

Cyclic crack growth tests with CRB specimens for the evaluation

of the long-term performance of PE pipe grades

Gerald Pinter a, , Markus Haager b , Werner Balika b , Reinhold W. Lang a,b

a Institute of Materials Science and Testing of Plastics, University of Leoben, Franz-Josef-Strasse 18, A-8700 Leoben, Austria

b Polymer Competence Center Leoben GmbH, Parkstrasse 11, A-8700 Leoben, Austria

Received 1 August 2006; accepted 14 September 2006

Abstract

It is well known that resistance against slow crack growth is important for the lifetime of pressurized polyethylene (PE)

pipes. Thus, several methods have been proposed in recent years to evaluate the long-term performance of PE using

fracture mechanics. It is generally believed that this leads to results more quickly compared to internal pressure tests. In the

presented research work, a method was implemented using fatigue loading of cracked round bar (CRB) specimens to

characterize crack growth resistance. The method was applied to ﬁve commercially available PE pipe materials and the

results were compared with the full notch creep test (FNCT). The same ranking was found with both methods, but it was

obvious that fatigue crack growth (FCG) experiments were faster by up to two orders of magnitude, especially when

characterizing modern (bimodal) PE types.

Keywords: Polyethylene; Slow crack growth; Fatigue; Ranking

1. Introduction

Internal pressure tests on pipe specimens are the

traditional way to determine the long-term proper-

ties of PE pipes. Unfortunately, this method is

expensive and very time-consuming, especially when

information on the CCG behavior is needed [3] .

Due to this, strong efforts have been put into the

development of fracture mechanics tests, which are

able to simulate CCG behavior of pipes in a

laboratory test and to obtain information on the

long-term behavior of PE pipes within a reasonable

time frame. CCG tests evaluated according to linear

elastic fracture mechanics (LEFM), the full notch

creep test (FNCT), the Pennsylvania notch test

(PENT), the notched pipe test (NPT) and the cone

test, among others, were introduced and are widely

used throughout

Pressurized polyethylene (PE) pipes have been

used successfully for more than 40 years, primarily

in fuel gas and water supply systems. Consequently,

substantial experience concerning the failure beha-

vior and the ﬁtness for purpose of PE piping

systems is available. Among other things, it is well

known that crack initiation followed by creep crack

growth (CCG) is the most important long-term

failure mechanism, which is reﬂected by numerous

publications dealing with this topic [1–7] .

Corresponding author. Tel.: +43 3842 402 2104;

fax: +43 3842 402 2102.

E-mail address: pinter@unileoben.ac.at (G. Pinter).

the industry, as well as

the

scientiﬁc community [2,5,7–14] .

doi: 10.1016/j.polymertesting.2006.09.010

ARTICLE IN PRESS

G. Pinter et al. / Polymer Testing 26 (2007) 180–188

181

Parallel to the development of these methods, the

inﬂuence of the molecular structure on crack growth

was investigated and the distribution of the mole-

cular mass and, in particular, the distribution of the

short-chain branches were identiﬁed as the most

important parameters [15–18] . Speciﬁc improve-

ments in the polymerization process of PE followed

and the development of third-generation resins in

particular has led to outstanding CCG resistance.

This was achieved by a bimodal distribution of the

molecular mass and the speciﬁc placement of the

short-chain branches on the high molecular mass

fraction. Even with the above-mentioned fracture

mechanics methods, testing of these materials

exceeds practicable time frames. Consequently,

quicker test methods are needed to evaluate the

long-term behavior of modern PE pipe materials.

One such method to propagate cracks, even in

high-performance PEs, by purely mechanical means

is to apply cyclic loads and to characterize the

fatigue crack growth (FCG) behavior. Despite the

differences in loading conditions, investigations on

PE showed that the micro-mechanisms of crack

growth have at least qualitative similarity, and

rankings based on FCG generally correlate well

with those obtained for CCG [1,7,14,19–24] .

Furthermore, it is possible to predict CCG from

FCG experiments by extrapolation [21,22,25] , and

currently major efforts are directed towards calcu-

lating the lifetime of PE piping systems based on

FCG test results [26,27] .

As much as 15 years ago, the ﬁrst reports about

static and cyclic crack growth tests on round

circumferentially notched bars were published in

Japan [24,28] ; later, some work was done also in

Europe under static [29,30] as well as cyclic loads

[7,14,19,31] . In these studies, the failure behavior

and failure mechanisms were studied under different

testing parameters. A big advantage of this speci-

men type is that it is geometrically very simple so

that it can be manufactured easily from compres-

sion-molded plaques as well as from pipes. More-

over, a well-deﬁned plane strain condition prevails

along the whole notch, and the stress in the

remaining ligament is in the same range as in pipes

under internal pressure.

The main objective of this work was to establish a

quick screening and ranking tool for PE pipe

materials that correlates with existing tests and

offers the potential to be used in an industrial

environment. Therefore, preliminary tests on two

PE pipe materials were performed where also crack

growth initiation and fracture surfaces were closely

investigated. Afterwards, a test procedure was

established and three additional PE pipe materials

were characterized. The ranking was compared with

results from the FNCT, which is widely used,

especially in Europe.

2. Experimental

All investigations in this study were performed on

commercially available PE-HD pipe materials with

a minimum required strength of 8MPa (MRS 8

or PE 80) and 10MPa (MRS 10 or PE 100),

respectively. This material classiﬁcation is estab-

lished from internal pressure tests on pipes, and

means that pipes made from these materials, loaded

with a hoop stress of 8 and 10MPa, respectively,

have a durability of 50 years at 23 1C. Some

characteristic material properties are summarized

in Table 1 . Compression-molded plates with a

thickness of 10 and 15mm were manufactured from

these materials. Subsequently, FNCT specimens

(1010100mm) ( Fig. 1 ) as well as cracked round

bar (CRB) specimens (D ¼ 14mm, L ¼ 100mm)

( Fig. 2 ) were machined from the plates.

The FNCT can be described as a constant load

tensile test, measuring the failure time of notched

specimens at elevated temperatures in a surface

Table 1

Characteristic material properties of the investigated materials (from data sheets)

r (g/cm 3 ) M n (kg/mol) M w (kg/mol)

E (N/mm 2 )

s y (N/mm 2 )

Material-code

Color

SCB (1/1000 C)

Comonomer

PE80-A

Black

0.955

290

Hexene

1000

PE80-B

Black

0.948

190

5.5

Hexene

700

PE80-C

Yellow

0.94

190

5.5

Hexene

700

PE100-D

Black

0.96

365

3.8

Butene

1100

PE100-F

Black

0.959

7.5

230

—

Butene

1400

r: density; M n and M w : number and weight average molecular mass; SCB: number of short side-chain branches; E: Young’s modulus;

s y : yield stress.

ARTICLE IN PRESS

182

G. Pinter et al. / Polymer Testing 26 (2007) 180–188

mode or pure tensile loading conditions) were

measured and plotted in a double-logarithmic

diagram. In this diagram, the stress in the ligament

of the specimens at the start of the test, s 0 ,is

additionally given. Crack lengths were monitored

during the tests with the aid of a traveling

microscope equipped with a linear variable trans-

ducer for displacement measurements. Further-

more, the crack opening displacement (COD) was

recorded using a strain gage. DK I was calculated for

the CRB specimens according to the equation from

Benthem and Koiter [32,33] :

Fig. 1. FNCT specimen and schematic drawing of test setup.

pab

DK I ¼ DF

pb 2

(1)

¼ 1

þ 3

1 þ 1

4 #

þ 0:731

0:363

ð2Þ

Fig. 2. CRB specimen and fatigue loading condition.

where DF is the difference of F max and F min , R the

radius of the specimen, b the radius of the remaining

ligament and a the crack length (see Fig. 3 ).

The fracture surfaces of selected specimens were

examined in detail with a scanning electron micro-

scope (SEM; Zeiss-DSM 962, Germany) in order to

further investigate the failure mechanisms in fatigue

loaded CRB specimens.

active environment ( Fig. 1 ). In this investigation,

the experiments were carried out according to ISO

16770 [12] at 80 1C with 2 wt% Arkopal N110

in deionised water in a test apparatus designed by

the Polymer Competence Center Leoben GmbH,

Austria. Prior to testing, the specimens were

notched by pressing a razor blade to a depth of

1.6mm into all four sides of the specimens. After the

test, the exact ligament area of every specimen was

determined using a microscope (Wild-M3Z, Swit-

zerland) and, taking into account the applied load,

the stress in the ligament area was calculated.

For each material the failure time was recorded

at different stress levels. For data presentation

failure times were plotted versus stress in a double

logarithmic diagram.

FCG experiments were performed on CRB

specimens using a servo-hydraulic closed-loop test-

ing machine (MTS 858 Table Top, MTS Systems

GmbH, Germany). A sinusoidal load at a frequency

of 5Hz and R ¼ 0.1 (R ¼ minimum load F min /

maximum load F max )at231C as well as 80 1C was

used to run the tests ( Fig. 2 ). Before testing, the

specimens were pre-cracked using razor blades and

then they were mounted into the testing machine by

screwing into especially designed clamps.

Failure times at different initial stress intensity

factor ranges, DK I,0 , (index I stands for opening

Fig. 3. CRB specimen with variables used in Eqs. (1) and (2).

ARTICLE IN PRESS

G. Pinter et al. / Polymer Testing 26 (2007) 180–188

183

3. Results and discussion

failure regimes, the PE 100 type showed higher

failure times. Typically, the changeover from ductile

to brittle failure mode occurred at DK I,0 ¼ 0.8M-

Pam 0.5 for the PE 80 and around 0.95MPam 0.5 for

the PE 100 material, respectively. Later, it turned

out that these values are typical for PE 80 and

PE 100.

Typical fracture surfaces for both failure modes

are presented in Fig. 5 for PE100-D. Starting from

the initial razor blade notch, specimens, which failed

in a fully ductile mode ( Fig. 5a ) showed necking

followed by ﬁnal ductile failure. In contrast, the

specimens, which failed in a brittle mode, are

characterized by typically smooth fracture surfaces

after crack initiation ( Fig. 5b ). However, with

increasing crack length, the load in the remaining

ligament increases rapidly and, consequently, after

reaching a critical crack length, ﬁnal ductile failure

occurred.

From Fig. 5b , it can be seen that the cracks were

not growing symmetrically for all specimens. It is

assumed that even slight asymmetries in the speci-

mens, the clamping and the whole testing machine

are responsible for this phenomenon. Moreover,

inhomogeneities in the specimen itself (internal

stresses, morphology, voids, etc.) may also have

an effect. The equation for the calculation of DK I in

CRB specimens is based on symmetrically growing

cracks and does not take into account that

asymmetrically growing cracks will lead to addi-

tional bending loads at the crack tip. Apart from the

fact that CRB specimens are generally not well

suited for the generation of crack growth kinetics,

because of their very high gradient in DK I , this was

3.1. FCG: test method development

At the beginning of the investigations, FCG

experiments were performed on PE80-A and

PE100-D at 5Hz, R ¼ 0.1 and 23 1C in order to

develop and to optimize the test methodology. In

[19] , it was already shown that under these

conditions hysteretic heating is negligible and does

not inﬂuence FCG. At high loads, ductile failure

was observed ( Fig. 4 ). Below a certain level, the

failure mechanism changed and, in the second part

of the diagram, crack initiation followed by crack

growth led to the failure of the specimens (see

Fig. 4 ). Similar to other creep rupture tests, in each

regime the failure times were located on straight

lines in a double-logarithmic diagram. In both

1.4

ductile failure

brittle failure

1.2

PE100-D

0.8

PE80-A

0.6

CRB-specimen

5 Hz

R = 0.1

0.4

0.1

100

failure time [h]

Fig. 4. FCG behavior of PE80-A and PE100-D from tests with

CRB specimens.

Fig. 5. Fracture surfaces of CRB specimens of PE100-D after fatigue loading at 23 1C; (a) ductile failure (DK I,0 ¼ 1.09MPam 0.5 ),

(b) brittle failure (DK I,0 ¼ 0.61MPam 0.5 ).

ARTICLE IN PRESS

184

G. Pinter et al. / Polymer Testing 26 (2007) 180–188

Fig. 6. SEM pictures of a brittle CRB fracture surface of PE80-A after fatigue loading (DK I,0 ¼ 0.6MPam 0.5 ,231C): (a) overview,

(b) 0.1mm after crack initiation, (c) 0.5mm after crack initiation and (d) 2mm after crack initiation.

the main reason that at this stage of the work no

efforts were taken to determine crack growth

kinetics of the investigated materials. Important,

however, is to emphasize that asymmetric crack

growth has no effect on the calculation of DK I,0 ,

because at the start of the test the notch is always

symmetrical. Moreover, the tests showed good

reproducibility of failure times (

length and increasing DK I , respectively, the ﬁbrillar

structure became rougher. By about 3.5mm after

crack initiation, a highly plastically deformed

fracture surface was present, which is a result of

the ﬁnal ductile failure of the CRB specimen.

If the fracture surfaces of PE80-A and PE100-D

tested at identical DK I,0 are compared, a similar

ﬁbrillar structure is obvious, although the ﬁbrilla-

tion seems to be ﬁner in PE100-D ( Fig. 7 ). On the

one hand, this can be explained by the smaller

plastic zones in PE 100 (higher yield stress of PE 100

compared to PE 80) and on the other hand by the

lower crack growth rates in PE 100 (longer failure

times). In the region of ductile failure, no differences

could be identiﬁed.

In addition to the failure times, crack initiation

times, t i , were determined by evaluating the COD

data. By way of example, in Fig. 8 ,COD max as well

as DCOD ( ¼ COD max COD min ) is plotted versus

10%), so that

an effect of asymmetrically growing cracks on

failure times can be excluded.

A more detailed analysis of the fracture surfaces

was done by SEM. On the brittle fracture surfaces, a

micro-structure typical of CCG and FCG in PE,

was identiﬁed that is characterized by the remnants

of ﬁbrils and micro-voids, the typical attributes of

craze formation and breakdown. In Fig. 6 , the

fracture surface of PE80-A tested at an initial stress

intensity factor range of 0.6MPam 0.5 is documen-

ted at different crack lengths. With increasing crack

Plik z chomika:

Inne pliki z tego folderu:

Inne foldery tego chomika: