26 An alternative approach to the evaluation of the slow crack growth resistance of polyethylene resins used for water pipe extrusion.pdf

J Polym Res (2007) 14:181-189

DOI 10.1007/s10965-006-9095-1

An alternative approach to the evaluation of the slow crack

growth resistance of polyethylene resins used for water

pipe extrusion

Fabiano Moreno Peres

Cláudio Geraldo Schön

Received: 8 August 2006 / Accepted: 25 November 2006 / Published online: 23 January 2007

Abstract High-density polyethylene (HDPE) pipes

have been largely employed in water and gas distrib-

ution systems. In spite of offering signiﬁcative advan-

tages over other materials, HDPE pipes suffer from

premature failures due to creep fracture. The current

industrial criterium for design and sizing of HDPE

pipes is discussed. The concept of ‘regression curve,’

i.e. of a time-to-failure criterium based in long-term-

hydrostatic strength (LTHS) tests, is criticised and

concluded to be unsatisfactory for this purpose. An

alternative approach is suggested, which is based on

shorter-term tests. This is illustrated by testing ﬁve

HDPE resins designed for pipe extrusion and com-

paring with their standard ‘regression curves’. The

obtained results are consistent with the ‘regression

curve’-based analysis, justifying the use of the alterna-

tive approach in the industry.

Introduction

High-density polyethylene (HDPE) is currently used

for the production of plastic pipes for water and gas

distribution systems. This material offers several advan-

tages over its competitors, like ﬂexibility, low cost, en-

vironmental attack resistance and ease of installation.

HDPE pipes for water distribution systems are, in prin-

ciple, designed for long lifetimes in service: current life-

times speciﬁed by industrial standards are expected to

be of the order of 50 years [ 1 , 2 ]. In spite of this, HDPE

pipes are known to be subject to premature in-service

failures due to time-dependent (creep) fracture, which

leads to leakage and, consequently, to environmental

and economic costs (water losses and increased main-

tenance costs). The search for improved materials leads

the petrochemical industry to continuous development

of new HDPE resins. Strategies such as increase in

the molecular mass, copolymerization (chain branch-

ing) and molecular mass distribution engineering have

been suggested to improve resistance to failure [ 2 , 3 ],

which is known to occur via a slow crack growth (SCG)

mechanism [ 4 ].

The resistance to in-service failure is usually eval-

uated in long-term hydrostatic strength (LTHS) tests,

in which the pipes are submitted to different inner

hydrostatic pressures, P , at different temperatures, T ,

for long periods of time. When subject to a inner

hydrostatic pressure, the tube wall’s stress state may

be approximated by a plane-stress state with principal

tensile stresses along the tube axis (the so called axial

stress) and along the tube circumference (the so called

circumferential, or hoop, stress,

Key words HDPE pipes

slow crack growth

water distribution systems

regression curve

ramp test

Abbreviations

LTHS

Long-term hydrostatic strength

MRS

Minimum required strength

LCL

Lowest conﬁdence level

SCG

Slow crack growth

F. M. Peres

)

Department of Metallurgical and Materials Engineering,

Escola Politécnica da Universidade de São Paulo,

Av. Prof. Mello Moraes, 2463, CEP 05508-900

São Paulo, SP, Brazil

e-mail: schoen@usp.br

C. G. Schön (

), the magnitude of

the later being approximately twice as larger as the for-

mer [ 5 ]. As a response to this permanent stress state the

182

F.M. Peres, C.G. Schön

(σ)

from the tests to the praxis, since the pipe is supposedly

tested in the same conditions as in-service. This trans-

ferability, however, must be criticised, since:

ductile

σ c

–

The LTHS tests are developed in laboratory-

controlled situation (in particular, constant P and

T is observed during the entire test schedule), while

they experiment ﬂuctuations of these conditions in

service;

brittle-like

–

The failure of a HDPE pipe will be controlled also

by extrinsic factors, like damage introduced during

installation or use, and the lifetime of the installed

pipe will be, eventually, shortened in comparison

with the one obtained in the LTHS tests;

ln(t r )

Fig. 1 A typical ‘regression curve’ as obtained in LTHS tests of

HDPE pipes

–

Part of the room temperature ‘regression curve’ is

extrapolated from high temperature tests using the

thermo-temporal superposition principle and the

use of this principle in extruded semi-crystalline

polymers should be questioned, since reorienta-

tion of the macromolecules from the initially tex-

tured material, characteristic of the extruded pipe,

is likely to happen during these high temperature

tests [ 8 ].

material deforms by creep and two major failure modes

are identiﬁed: a ‘ductile’ mode, characteristic of high

stresses (high P

⇒ σ ≥ σ c ) – followed by bulging or

ballooning of the tube wall – and a ‘brittle-like’ mode,

characteristic of lower stresses (low or moderate P

⇒

σ<σ c ) – followed by negligible macroscopic plastic

deformation, with the development of a small through

longitudinal crack (slit fracture) [ 1 , 2 ]. Service failures,

except in the case of accidents, are brittle-like since the

pipe is designed to withstand the inner pressures which

would lead to ductile fracture.

The results of the LTHS tests are usually evaluated

in terms of a graphic, known in the praxis as ‘regres-

sion curve’: a bi-logarithmic plot of the hoop stress,

The LTHS tests and the ‘regression curve,’ therefore,

cannot be directly related with the life expectation of

the pipe in the water distribution system. However,

indirectly they allow us to compare resins under similar

conditions: a PE-100 resin is expected to have superior

in service performance in comparison with a PE-80 one.

In addition, the determination of the ‘regression curve’

of a polymer requires several long-term creep tests (up

to 10 4 h) and hence the LTHS test are considerably

expensive, becoming in most cases impracticable as

a quality control tool. Finally, the ‘regression curve’

philosophy implies a very long lifetime for the pipes

(50 years), while experience shows that in-service fail-

ures are observed in a much shorter timescale.

The aim of the present work is to suggest an al-

ternative procedure for comparison of different resins

based on the ‘ramp test,’ originally proposed by Zhou

et al. [ 9 ]. This will be tested by comparing the re-

sults obtained in ﬁve HDPE resins designed for water

and gas pipe extrusion, obtained from four traditional

suppliers, with the available data on their respective

regression curves.

versus time-to-failure, t r . The results for each of

the above mentioned failure modes in such plots are

linear with negative slopes (See Fig. 1 ). 1

The ductile-

to-brittle transition stress,

σ c is characteristic of a given

formulation (i.e. base resin plus additives) and is not

easily observed in room temperature tests, requiring

extrapolation from results at higher temperatures for

its determination. The LTHS test is also used for the

establishment of a ‘minimum required strength’ (MRS)

at 50 years of lifetime, according to the ISO 9080

Standard [ 6 ], which, in its turn, is used to classify the

material (in form of extruded pipe). For example, a

HDPE formulation which is predicted to fail in the

LTHS tests at 50 years for 8

0 MPa has a

MRS of 8.0 MPa and is designated as PE-80, according

to the ISO 12162 Standard [ 7 ].

The LTHS test, the ‘regression curve’ and MRS

concepts are widely accepted by the water distribution

industry for the speciﬁcation of HDPE pipes. This is

probably due to the apparent transferability of results

≤ σ<

Methodology

The ‘ramp test’ consists in the assessment of the engi-

neering yield stress (

1 In other words, the hoop stress and time-to-failure are power-

law-related, with negative exponents.

σ y ) and engineering draw stress

Slow crack growth resistance in Polyethylene

183

σ dr ) of standard tensile specimens 2

(

as a function of

index of 0.85 g (190 ºC/5 Kg/10 min) according

to ISO 1133. Extracted data from the ‘regres-

sion curve’ (Fig. 3 ), according to ISO 9080:

LTHS (50 years/20 ºC) = 9.38 MPa, LTHS/LCL

(50 years/20 ºC) = 8.64 MPa (MRS = 8.0 MPa).

MDPE 8818 – Non-pigmented polyethylene com-

pound, supplied by PBBPolisur S. A. (Dow

Latin America), classiﬁed as PE-80, according

to ISO 12162, designed (after pigmentation),

for water and natural gas pipe extrusion. The

material has density 0.940 g/cm 3 and melt ﬂow

index 0.77 g (190ºC/5 Kg/10 min) according

to ASTM D 1238. Extracted data from the

‘regression curve’ (Fig. 4 ) of the compound

produced with this resin and a yellow pigment

(MDPE 8818 YW), according to ISO 9080:

LTHS/LCL (50 years/23 ºC) = 8.03 MPa (MRS

= 8.0 MPa).

HP 0155 – Experimental non-pigmented compound,

supplied by Braskem S. A., classiﬁed as PE-

100 according to ISO 12162, designed, after

pigmentation for water and natural gas pipe

extrusion. The compound with black pigment

(carbon black) has density of 0.955 g/cm 3 and

melt ﬂow index of 0.3 g (190ºC/5 Kg/10 min).

Extracted data from the ‘regression curve’

(Fig. 5 ) of the black compound, according to

ISO 9080: LTHS (50 years/20ºC) = 10.7 MPa,

LTHS/LCL (50 years/20ºC) - 10.1 MPa (MRS

10.0 MPa).

MP 0240 – Experimental non-pigmented polyeth-

ylene compound, supplied by Braskem S. A.,

designed after pigmentation for water and

natural gas pipe extrusion. The compound

produced with this resin in the yellow colour

has density 0.939 g/cm 3 and melt ﬂow index

0.8 g (190ºC/5 Kg/10 min). Extracted data

from the ‘regression curve’ (Fig. 6 )ofthe

yellow compound, according to ISO 9080:

LTHS (50 years/20ºC) = 8.81 MPa, LTHS/LCL

(50 years/20ºC) = 8.28 MPa (MRS = 8.0 MPa).

the strain rate (

)[ 9 ]. Both functions are linear with

different positive slopes when plotted against the log-

arithm of the strain rate. The intersection of the two

lines, according to the original proposal, corresponds to

the critical stress for the ductile-to-brittle transition (

σ c )

[ 9 ]. In spite of its minimalistic structure, this method

is based on solid hypotheses about the damage mi-

cromechanics of polyethylene (nucleation and growth

of crazes and subsequent rupture of the ﬁbrils) [ 10 ].

The procedure here suggested consists in combining

a limited number of shorter-term hydrostatic strength

tests at ambient temperature, to determine the ductile

wing of the ‘regression curve,’ with the ‘ramp test,’ to

establish the lowest limit of hoop stress (

σ = σ c )for

which the the extrapolation of this ductile wing would

be valid. The procedure for the determination of the

MRS, described in the ISO 12162:1995 Standard [ 7 ]

would be then adopted for classiﬁcation of the polymer

provided the ductile wing of the curve can be extrapo-

lated to t f

50 years. This suggested methodology will

be, from now on, brieﬂy referred to as “STHS”.

≥

Experimental

Materials

The following HDPE resins have been investigated in

the present work:

GM 5010 T2 – Black polyethylene compound 3 con-

taining 2.2% carbon black, supplied by Ipi-

ranga Petroquímica S/A, classiﬁed as PE-80

according to ISO 12162 and designed for wa-

ter pipe extrusion. This is a HDPE resin with

bimodal molecular mass distribution with den-

sity 0.954 g/cm 3 and melt ﬂow index of 0.53 g

(190ºC/5 Kg/10 min) according to ISO 1133. Ex-

tracted data from the regression curve (Fig. 2 ),

according to ISO 9080: LTHS (50 years/20ºC)

= 10.232 MPa, LTHS/LCL (50 years/20ºC) =

9.901 MPa (MRS = 8.0 MPa).

Rigidex PC 002-50R968 – Blue polyethylene com-

pound, supplied by Solvay Indupa do Brasil

S.A., classiﬁed as PE-80 according to ISO 12162,

designed for water pipe extrusion. The mate-

rial has density of 0.944 g/cm 3

Test procedure

ASTM D638 Type IV standard tensile specimens have

been extracted from 200

and melt ﬂow

3 mm plates obtained

by compression molding at 190ºC and using 19.6 MPa

closure pressure for 5 min, followed by air cooling.

Tensile tests were performed in a universal mechanical

testing machine at 25

200

2 In the original proposal, Zhou et al. [ 9 ] extract the tensile

specimens from the pipe, oriented along the extrusion direction.

3 Polyethylene compound, in the present context, corresponds to

the base polyethylene resin plus antioxidants and other additives,

among which, depending on the case, pigments.

2ºC in displacement control

(ramp test) with strain rates ranging from 0.0001 to

0.1 s − 1

(calculated as an engineering strain rate us-

184

F.M. Peres, C.G. Schön

Fig. 2 ‘Regression curve’ of

the GM 5010 T2 resin, as

furnished by the supplier

(LPL is an alternative

designation for the lowest

conﬁdence limit, LCL,

obtained in the linear

regression based on the

experimental data)

ing the grip displacement rate). Three samples have

been tested for each resin and value of

Results and discussion

. Average

values of

σ dr have been calculated (as well

as the respective standard deviations). Standard linear

regression has been employed to calculate

σ y and

Figure 7 shows a typical stress-strain curve for the

case of the Rigidex PC002-50R968 resin. As ex-

pected for semi-crystalline polymers like HDPE, yield

point and cold-drawing plateau are easily identiﬁed in

Fig. 7 . Nevertheless, some ambiguity in the deﬁnition

of drawing stress is possible. In the present work this

quantity is deﬁned as the stress corresponding to 100%

σ y ×

and

lines, from which the stress corresponding to

the intersection (

σ c ) has been calculated and compared

with the respective ‘regression curve,’ furnished by the

resin supplier.

Fig. 3 ‘Regression curve’ of

the Rigidex PC 002-50R968

resin, as furnished by the

supplier

Slow crack growth resistance in Polyethylene

185

100

sion curves”. The ISO9080 Standard requires that the

sample fails in the ductile mode for a valid estimation

of LTHS (50 years/20ºC), hence

T = 23 o C

regression line

LCL (97.5%)

σ c should be lower in

modulus for the resin i.e. the brittle-like fractures occur

at lower stresses and, therefore, after t f

50 years).

In the cases of resins GM5010T2 and PC002-50R968

the values obtained in the “ramp test” are compati-

ble with the ones estimated based on the “regression

curve” (see Fig. 2 ). The suggested STHS methodology

would eventually give the same results as the full LTHS

test, since, according to

8.03 MPa

σ c obtained in the “ramp test,”

the ductile wing of the 20ºC curve may be extrapolated

safely up to 50 years. For the MDPE 8818 and MP0240

resins, on the other hand, the “ramp test” results clearly

overestimate

50 years

100

1000

10000

100000

time [hours]

Fig. 4 ‘Regression curve’ of the MDPE8818YW resin, drawn

based on the data furnished by the supplier

σ c leading to the erroneous conclusion

that the pipe would fail in the “brittle-like” mode with

t f

50 years, in contradiction with the measured “re-

gression curves”.

The results for the “ramp test” in resin HP0155

suggest that

extension of the sample i.e.

1.0). This criterium is

arbitrary and was empirically determined, based on the

observation that at this deformation level all samples

presented a well-developed neck, which extended over

a signiﬁcative portion of its length.

Figure 8 presents the results for the evaluation of

ε =

σ c is much lower than the values obtained

in the regression curve at ambient temperature, and, in

principle, the STHS method would give a valid estimate

of MRS for this compound, as in the cases of resins

GM5010T2 and PC002-50R968. A closer analysis of the

regression curve of the HP0155 resin (Fig. 5 ), however,

shows that the estimated

σ c

of the resins using the “ramp test” (points are averages

of three tests for each strain rate and error bars repre-

sent the corresponding standard deviations).

Table 1 shows the values of

σ c at 20ºC is too small. In fact,

the data shown for the LTHS tests at 60ºC

present

σ c

8 MPa. The transition at ambient temperature,

therefore, must occur at a higher stress. The measured

“regression curve” indeed “suggests” that

σ c obtained in the “ramp

test” of the investigated resins and the value of LTHS

(50 years/20ºC) obtained from the respective “regres-

σ c would be

Fig. 5 ‘Regression curve’ of

the HP 0155 Black resin,

as furnished by the supplier

Plik z chomika:

Inne pliki z tego folderu:

Inne foldery tego chomika: