Available online at www.sciencedirect.

com

ScienceDirect
Procedia Engineering 101 (2015) 2 – 9

3rd International Conference on Material and Component Performance

under Variable Amplitude Loading, VAL2015

Fatigue life prediction and damage modelling of High-density

polyethylene under constant and two-block loading
Djebli Abdelkadera*, Bendouba Mostefaa, Aid Abdelkrima, Talha Abderrahimb,c,
Benseddiq Noureddinec, Benguediab Mohamedd
a
LPQ3M, Mascara University, 29000, Algeria
b
Material Laboratory, (HEI) Lille, France
c
LML, IUT, University of Lille1, France.
d
Department of Mechanical Engineering, University of Sidi Belabbess, 22000, Algeria

Abstract

An experimental analysis for determining the fatigue strength of HDPE-100, under constant and variable amplitude
loading is presented. Further, the cumulative fatigue damage behavior for HDPE-100 was experimentally investigated.
First, The S–N curve was obtained to establish the fatigue life of The HDPE-100 subjected to constant stress
amplitude. Secondly, the Cumulative fatigue damage was estimated by different cumulative model such as Miner rule,
damage stress model and Energy model (Damage energy model). Comparison between predictions and experimental
results showed different trends depending on the prediction model used.
© 2015
© 2015The TheAuthors.
Authors. Published
Published by Elsevier
by Elsevier Ltd.is an open access article under the CC BY-NC-ND license
Ltd. This
Peer-review under responsibility of the Czech Society for Mechanics.
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the Czech Society for Mechanics
Keywords: polyethylene; Cumulative damage; Fatigue damage; viscoelastic behavior; Two-block loading

1. Introduction

The fatigue behavior of polymers is gaining more and more importance in the study of materials, such interest
dictated by the increasing use of polymers in complex applications, and this as a direct consequence of their advantages
like technical performance and of course their relatively low cost [1]. High density polyethylene (HDPE) is one of the
most used in industry because of the diversity of its applications and the multiple benefits it provides over conventional
materials.

* Corresponding author. Tel.: +213-5-52-46-06-23; fax: (+213) 045 813 998.

E-mail address:djebliabdelkader@yahoo.fr

1877-7058 © 2015 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the Czech Society for Mechanics
doi:10.1016/j.proeng.2015.02.002
 Djebli Abdelkader et al. / Procedia Engineering 101 (2015) 2 – 9 3

This material is extensively used in water and gas distribution systems. This latter has provided cheaper solutions
in the industry of pipes constituents and network, from manufacturing to exploitation, through installation and
maintenance.

Nomenclature

DEM Damage energy model

DSM Damage stress model
Nri: number of cycles to failure for level i under V i
Ui the work transferred to the specimen for the solicitation of amplitude Vi applied for ni cycles.
Wi: total work transmitted to the specimen at the time of fatigue failure.
Wedi energy due to damaged stress
Wi energy due to the applied stress
Wu energy due to the ultimate stress of the material
W energy density
V edi Damage stress
Vi Applied stress
Vu Ultimate stress
V Stress
H Strain

The tube system into service is subject to internal pressure or to external forces that vary in magnitude and
frequency. The repetition of these fluctuating loads, generates responses that differ depending on the type of loading
[2]. Such cyclic loads cause cumulative damage and cracks that can contribute to rupture of the pipe system [3]. It is
important to make clear that these polymeric pipes have a problem that needs special consideration. Experience has
shown that low stress may cause rupture of the pipe in relatively short service time, especially if the loading is cyclic
[4]. It is probable that the problems of material fatigue under cyclic loading are associated with the presence of intrinsic
defects during the manufacturing process. Such defects can lead to pipe rupture, causing economic and human losses.
This will have direct consequences on the durability of the whole system [5].

Until now, Industrial laboratory carries out hydrostatic tests to pipes. Such tests provide information on material
resistance to monotonic loads, but give no information on the behavior of the material under cyclic loading. Even, if
stresses are lower than those considered in the design step [6, 7].

The aim of this paper is to determine the fatigue strength of HDPE-100 material, and analyze its fatigue behavior
by characterizing and determining its life curve. Based on the experimental results, we propose to examine and evaluate
some cumulative damage models proposed for metallic materials, such as a Miner's rule, damage stress and the energy
(damage energy) model. Predictions of different models were carried out. Then, they are compared to experimental
results.

2. Material and experimental procedure

2.1. Tensile tests

Specimen tests were cut directly from HDPE pipe in the longitudinal direction [8, 9]. Cylinder sections were cut
axially and transversely by a guillotine cutter. A total of 60 fatigue specimens and 10 tensile specimens were machined
by the CNC machine Mill 55 (EMCO- Concept Mill). The geometry and dimensions of the specimens are shown in
Figure 1.
4 Djebli Abdelkader et al. / Procedia Engineering 101 (2015) 2 – 9

The monotonic tensile tests are made on LLOYD electromechanical testing machine referenced LR150K, carried
out in the material’s laboratory at Higher School of Engineer (HEI, Lille). The tensile machine has a load cell capacity
of 5 KN and a crosshead displacement that reaches up to 1150 mm. For the measurement of the local strains in the
central zone of the specimen, we used an extensometer (Epsilon Clip-on STAGE) having a base length L0=50 mm
and allowing displacement measurement between -10 and 100% of L0. The monitoring of the tests and the data
acquisition are performed by software of materials analysis named "NEXYGEN-Plus" developed by LLOYD
Instruments.

Tests were carried out at a crosshead speed of 5, 10, and 50 mm/min corresponding to the strain rates of 1.5×10-3,
3×10-3 and 1.5×10-2 s-1 respectively, this choice was taken to examine the monotonic tensile response of the HDPE
and to assess the contribution of viscosity in its mechanical behavior through the influence of the strain rate.
Typical stress-strain curves at different strain rates are given in Figure 2. It is noted that the mechanical behavior
of the HDPE-100 is nonlinear and depends on the strain rates. Therefore, it would be of great interest, for a fatigue
study, to analyze such an observation.

Fig.1 Tensile Specimen geometry according to ASTM D-638- Fig. 2 Stress-Strain curves at different strain rates for HDPE-100
Type I, dimensions are in mm

One can see that, the behavior at different strain rates have the same trend. In the domain of small deformations
the behavior is elastic. Then, it is linear viscoelastic. This step is followed by a plastic domain, reached at the maximum
point in the stress-strain curve. After that, stress decreases with increasing strain. It is noted that the strain at the
threshold of plasticity decreases with increasing strain rate. The elastic modulus E and the yield stress are given in
table 1.

Table 1: Quasi static Mechanical properties of HDPE-100

Strain rates Yield stress Yield strain Young’s modulus
(s-1) (MPa) (%) (MPa)
10-3 22.5 11 750
2×10-3 24.5 10 825
10-2 28.75 7.5 1400

2.2. Fatigue tests

All fatigue tests were done at room temperature in a single sequence at a frequency of 2 Hz and a load ratio R =
0.0 until the specimen is broken or maximum elongation is reached. The platform of multi-axial fatigue test was
provided by the Laboratory of Mechanics (LML). The main experimental means and information about the platform
and its accessories are given in [9]. Table 2 resumes the fatigue tests (Test No. 1 to Test No. 9). Where, the boxes
 Djebli Abdelkader et al. / Procedia Engineering 101 (2015) 2 – 9 5

signed with a cross indicate that the specimen was eliminated for mishandling and NR means specimen not broken
after one day test.

Table 2 Protocol and results of fatigue tests at constant load.

Test Number Specimen Lifetime cycles Amplitude Stress (MPa)
E1 274400
N1 E2 NR 15
E3 ×
E1 12934
N2 E2 × 16
E3 ×
E1 4222
N3 E2 15581 17
E3 ×
E1 6000
N4 E2 1819 18
E3 3002
E1 1774
N5 E2 × 19
E3 ×
E1 3223
N6 E2 959 20
E3 383
E1 431
N7 E2 438 22
E3 218
E1 218
N8 E2 180 24
E3 127
E1 31
N9 E2 × 26
E3 ×

2.2.1. Cumulative damage tests with two blocks loading

In the two-block cumulative fatigue tests, two stress amplitudes were selected as the controlled conditions for each
block loading. If the first stress level applied three cycles is higher than the second stress level, the loading sequence
is said high to low (H-L). Reversely, it is said low to high (L-H). The residual fatigue life n12 is the number of cycles
reached until failure of the specimen in the second loading block. N f1 and Nf2 are lifetimes of the material under V 1
and V 2 , respectively.
Details of these tests are given in Table 3 with the results of the residual life times obtained.

Table 3 Sequences of cumulative damage tests and results

V1 V 2 Nf1 Nf2 n1 n12 Exp

L-H 1 16 18 27034 4783 2000 998

L-H 2 18 22 4783 250 900 181
H-L 1 18 16 4783 27034 900 19924
H-L 2 22 18 250 4783 181 851
6 Djebli Abdelkader et al. / Procedia Engineering 101 (2015) 2 – 9

3. Fatigue damage modelling:

In the next sections, we propose to model the lifetimes of our HDPE material. The cumulative damage models
chosen are: Miner's linear model, the model of the damage stress and the model of damage.

3.1. The linear Miner's rule

For its simplicity and no parameter requirement, we used the Miner's rule to verify how it performs. Miner assumed
that the damage represented by the variable Di, and generated by ni cycles of level i is such that [11]:
Ui ni
Di (1)
Wi N fi
By hypothesis, for a sequences loading with p stress levels (p> 0), Miner supposed that the rupture occurs when :
i p

¦D
i 1
i 1.

3.2. Damage stress model DSM

The second model used is known as a damage stress model [12, 13, and 14]. As the Miner's rule, it is simple; it
only requires the knowledge of the Wöhler curve. An advantage over the Miner's rule, this nonlinear cumulative model
takes into account the load history. Expression of the DSM indicator is given as follows:
V edi  V i
D (2)
V u V i
Damage is calculated from level i to level i+1 by the relation:
V edi  V i V equiv  V i 1
D (3)
V u V i V u  V i 1

3.3. A nonlinear energy model (damage energy model)

Similar to the damage stress model, damage energy model has no parameters to be determined. It only requires
Wöhler’s curve converted into energy versus cycle numbers [15]. Thus, the strain energy density is taken as a variable
in the fatigue damage calculation:
Wedi  Wi
D (4)
Wu  Wi
Where energy is calculated by the relation:
1
W V .H (5)
2

4. Results discussion

4.1. Wöhler Curve and damage accumulation

As illustrated by figure 3, we obtained a linear trend between the logarithm of life times and applied stress.
Therefore, it is judicious to model the Wöhler curve by Basquin's equation [16]. This latter is commonly used to model
the S-N curve of HDPE material [9, 10, 17]. Thus, parameters of Basquin's equation are given in equation 6:
 Djebli Abdelkader et al. / Procedia Engineering 101 (2015) 2 – 9 7

V 0.0728 log N r  33.5 (6)

It is noted that, the dispersion results from the same level of loading is already observed. Such a phenomenon is
due to several factors, such as construction defects during the process of pipe extrusion and the machining process of
specimens.

Fig. 3 Wöhler’s curve for the studied HDPE-100 material.

4.2. Fatigue life prediction with the nonlinear energy model (DEM)

The DEM model is connected to Wöhler’s curve (converted to energy curve). Therefore, damage is calculated for
each cycle and accumulated until reaching the unit value (D=1). Examples of results obtained from this model are
presented in Figures 4 and 5, for low-high and high-low sequences, respectively.

Fig. 4 Accumulated damage by the DEM model: Low to high two- Fig. 5 Accumulated damage by the DEM model: High to Low two-
blocks loading (Energies 0.163-0.244 correspond to 18-22 MPa). blocks loading.

Figures 4 and 5 show that the level loading significantly affects the evolution of the damage. Furthermore, the order
of application of stress is considered. Effectively, increasing loading leads to an acceleration of the damage upon the
second block. Inversely, when applying a decreasing sequence, the damage diminishes upon the second block loading.
This latter observation is interpreted by the slope of the curves, when passing from first to second block.
8 Djebli Abdelkader et al. / Procedia Engineering 101 (2015) 2 – 9

4.3. Fatigue life prediction with the damaged stress model

As the damage energy model, the damage stress model (DSM) is connected to Wöler curve, cycle by cycle. The
damage is calculated using a Matlab program. Using the same calculation principle, except that, the damage parameter
is a function of the equivalent stress at a given cycle.

Predictions obtained from this model, showed the same trends observed previously.

Fig. 6 Accumulated damage by the DSM model: Low to high two- Fig. 7 Accumulated damage by the DSM model: High to Low two-
blocks loading. blocks loading.

The predictions obtained from this model showed the same trends observed above. Examples of results are given
in Figures 6 and 7. Where, compared to the DEM Model, one can note that the same sequences of loading gave
different predictions.
Figure 8 summarizes different residual lives from different models. As is illustrated in figure 8, The Miner’s rule
gave predictions closer to experimental results.

Fig.8 Different model predictions compared to the experimental results.

 Djebli Abdelkader et al. / Procedia Engineering 101 (2015) 2 – 9 9

The results obtained from the models are optimistic. Indeed, and given the dispersed nature of the fatigue, one can
judge the relevance of the predictions obtained from different models.

5. Conclusion

In this paper, we have determined the fatigue strength of HDPE-100 material. Then, we have analyzed its fatigue
behavior by characterizing and determining its life curve. The experimental results served as the basis for modelling
accumulation of damage. For this task, we used Miner's rule, DEM model and DSM model, that were originally
proposed for metallic materials. Predictions from different models were carried out. Then, they are compared to
experimental results.
In general, the identified findings are as follows:

x The Miner's rule seems to be a good predictive tool for Low-High and high-Low sequences.
x The DSM nonlinear model gave results closer to that obtained with the Miner's rule. It seems an adequate
model to predict lifetimes of HDPE. However, results for the two sequences, low to high and high to low, are
non-conservative (above experimental results).
x The DEM nonlinear model overestimates the lifetimes for both levels of loading, especially, for lowest loading
levels. In fact, in this paper, we neglected the effect of the viscous dissipated energy, which is added to the
elastic energy. This problem will be treated with special attention in future research.

References

[1] Brewin J. and Chapman P. Recent Developments in the Design and Manufacture of Plastic Pipes, Vinidex Tubemakers PTV limited, 2000.
[2] Joseph SH. Fatigue Failure and Service Lifetimes in PVC Pressure Pipes, Plastics and Rubber Processing and Applications 1984; 4(4): 325-30.
[3] Dowling NE. Mechanical Behavior of Material: Engineering Methods for Deformation, Fracture and Fatigue. 2nd ed. Prentice Hall, Upper
Saddle River, NJ, USA; 1999.
[4] Gardner AK, King KB and Cooper J.R. In-Service Care Considerations for High Pressure Vessels in a Commercial Polyethylene Process,
Denver, CO, USA (1981); 48: 151-64.
[5] Oral E, Malhi Arnaz AS and Muratoglu OK. Mechanisms of Decrease in Fatigue Crack Propagation Resistance in Irradiated and Melted
UHMWPE, Biomat. (2005.); l 27: 917-25,
[6] Kannappan S. Introduction to pipe stress analysis, John Wiley and Sons, USA, 1986.
[7] Goncalves R. Introducción al Análisis de Esfuerzos, Segunda Edición, Editado por R. Goncalves, Caracas, Venezuela, 2002.
[8] STPM. CHIALI, Tubes, Polyéthylène (PE) et Accessoires , Catalogue technique.
[9] Djebli A, Bendouba M, Aid A, Talha A, Benseddiq N and Benguediab M. Uniaxial Fatigue of HDPE-100 Pipe: Experimental analysis, ETASR,
2014; 4(2):600-4.
[10] Gonzalez M. Machado R, Gonzalez J. Fatigue analysis of PE-100 pipe under axial loading, Proceedings of the ASME 2011 Pressure Vessels
& Piping Division Conference; July 17-21, 2011, Baltimore, Maryland, USA.
[11] Miner MA. Cumulative damage in fatigue. J. of App. Mec., (1945): 159-164.
[12] Mesmacque G, Garcia S, Amrouche A and Rubio-Gonzalez C. Sequential law in multiaxial fatigue, a new damage indicator. Int. J. of Fat.
2005; 27(4): 461-7.
[13] Aid A, Bendouba M, Aminallah L, Amrouche A, Benseddiq N. & Benguediab M. An equivalent stress process for fatigue life estimation under
multiaxial loadings based on a new non linear damage model. Mat. Sci. and Eng; 2012, 538: 20-27.
[14] Aid A Amrouche A, Bachir BA, Benguediab M. Fatigue life prediction under variable loading based on a new damage model. Mat. & Des.
2011, 32(1):183–91.
[15] Djebli A, Aid A, Bendouba M, Amrouche A, Benguediab M. & Benseddiq N. A non-linear energy model of fatigue damage accumulation and
its verification for Al-2024 aluminum alloy. Int. J. of Non-Linear Mech., 2013 51: 145-151.
[16] Basquin, O. H. The exponential law of endurance tests. In Proc. ASTM (1910). 10, II: 625-630.
[17] Berrehili A, Castagnet S & Nadot Y. Multiaxial fatigue criterion for a highǦ density polyethylene thermoplastic. Fat. & Frac. of Eng.Mat. &
Structures, 2010; 33(6): 345-357.