Mechanics of Time-Dependent Materials 1: 241–268, 1998. 241
c 1998 Kluwer Academic Publishers. Printed in the Netherlands.
Viscoelastic Creep Crack Growth: A Review of Fracture Mechanical Analyses
W. BRADLEY1, W.J. CANTWELL2 and H.H. KAUSCH2 1Department of Mechanical Engineering, Texas A&M University, College Station, TX, U.S.A.; 2Materials Science Department, EPFL, DMX-LP, Swiss Federal Institute of Technology, 1015 Lausanne, Switzerland
(Received 8 February 1997; accepted in revised form 2 October 1997)
Abstract. The study of time dependent crack growth in polymers using a fracture mechanics approach has been reviewed. The time dependence of crack growth in polymers is seen to be the result of the viscoelastic deformation in the process zone, which causes the supply of energy to drive the crack to occur over time rather than instantaneously, as it does in metals. Additional time dependence in the crack growth process can be due to process zone behavior, where both the flow stress and the critical crack tip opening displacement may be dependent on the crack growth rate. Instability leading to slip-stick crack growth has been seen to be the consequence of a decrease in the critical energy release rate with increasing crack growth rate due to adiabatic heating causing a reduction in the process zone flow stress, a decrease in the crack tip opening displacement due to a ductile to brittle transition at higher crack growth rates, or an increase in the rate of fracture work due to more rapid viscoelastic deformation. Finally, various techniques to experimentally characterize the crack growth rate as a function of stress intensity have been critiqued. Key words: fracture, fracture mechanics, polymers, viscoelastic crack growth
1. Introduction The subject of slow, stable crack growth in polymeric materials is growing in importance. The use of plastic pipe for natural gas and water distribution and for transport of raw sewage has become quite common during the past 15 years. Such applications are made with the expectation of service lifetimes of at least 30 years. However, such lifetimes under service conditions of constant stress, due to ground loading and internal pressure, will require polymeric materials with a high resistance to slow, stable crack growth, sometimes called static fatigue. In metals at above half of their melting temperature (expressed in degrees K), time dependent plastic deformation at the tip of cracks can result in stable, time dependent crack growth, usually referred to as creep crack growth. For most met- als, neither creep deformation nor creep crack growth is significant at ambient temperature. On the other hand, many polymers experience considerable creep at room temperature, especially for long term service. This is a consequence of the fact that ambient temperature is a significant fraction of the glass transition tem- perature (again, expressed in degrees of absolute temperature) for most polymeric materials. This creep, which results from the viscoelastic character of polymeric 242 W. BRADLEY ET AL. materials, can also give creep crack growth, or more appropriately, viscoelastic creep crack growth. The usual mechanical properties that are measured in materi- als characterization such as loss modulus, storage modulus, tensile yield strength and ultimate elongation may not be very useful in predicting a polymer’s resistance to viscoelastic crack growth. Thus, new approaches are being developed to better predict a polymer’s resistance to viscoelastic creep crack growth for engineering applications and materials development The susceptibility of polymers to viscoelastic creep crack growth has resulted in some rather expensive lessons in service. For example, it has been common in the United States for polyethylene pipe used for city distribution of natural gas in a metropolitan area to be pinched clamped to stop gas flow and allow repairs to be made. This practice makes it possible to install natural gas distribution systems in metropolitan areas with a minimum of valving compared to distribution systems made with cast iron pipe. It has been assumed that the polyethylene can be pinched clamped without introducing any damage to the pipe. However, it has been found that pinch clamping may introduce damage in the form of surface cracks at the inside diameter of polyethylene pipe that will subsequently have viscoelastic creep crack growth over a period of 5–10 years before causing leaks (Jones and Bradley, 1987). The use of plastic pipe in connection with metal fittings with sharp edges to connect city water service to residential customers resulted in the initiation and propagation of viscoelastic creep crack growth, causing leaks to develop. However, the viscoelastic creep crack growth that developed where the plastic pipe was bent around the sharp corners of the fittings took several years to produce leaks. By the time the problem was identified, the improperly designed fitting had been widely used and several thousand such applications subsequently leaked, resulting in enormous replacement costs. Metal crimps used in conjunction with plastic fittings for plastic pipe for water service in residential homes have been found to produce viscoelastic creep crack growth leading to leaks, but again only after several years of service. Thus, the problem did not “surface” until the use of the crimped fittings had become widespread. Buried PVC pipe has been found to be susceptible to viscoelastic creep crack growth when the pipe is of inferior quality and/or the installation leads to excessive ground loading (U.S. Department of Transportation, Research and Special Programs Administration, 1994). However, failures in service may not occur for several years and conventional mechanical properties tests such as tensile or internal pressure tests do not always identify pipe that is susceptible to viscoelastic creep crack growth (Jones and Bradley, 1993; Richard et al., 1959; Kausch von Schmelig and Niklas, 1963). It is clear that applications of polymers involving long term loading resulting in either constant or intermittent stresses can cause viscoelastic creep crack growth. Thus, a better understanding of how to evaluate a materials resistance to viscoelastic creep crack growth and how to produce polymeric materials with a high degree of resistance to such cracking is essential to the successful retention of some existing markets and expansion into some new markets for polymeric materials. VISCOELASTIC CREEP CRACK GROWTH 243
The application of fracture mechanics to viscoelastic media goes back to the middle 1960s (Williams, 1965; Vincent and Gothan, 1966; Retting and Kolloid, 1966). Extending Griffith’s work to linearly viscoelastic materials Williams (1965) found that the crack initiation criterion depends on the loading history. Vincent and
Gotham (1966) and Retting and Kolloid (1966) were among the first to note that _ the work of fracture in polymers, 2 , was a function of the crack growth rate, a. Kostrov and Nikitin (1970), following the lead of Dugdale (1960) and Barenblatt (1962) for time independent materials, were the first to note that a failure zone needs to be introduced ahead of the crack if the time dependence of the fracture process is to be properly modeled. Various approaches to model the viscoelastic fracture in the process zone have been taken, depending on the assumed geometry of the process zone ahead of the crack tip and the constitutive properties for the material in the process zone. If a process zone, or failure zone, is assumed to be finite, then the constitutive properties of the material in the process zone need to be specified (Wnuk and Knauss, 1970; Knauss 1993). Alternatively, the failing material in the process zone can be rep- resented by a stress-displacement relationship for a zone of zero thickness as was done by Knauss (1970). Knauss (1970) solved the Griffith problem for an assumed finite thickness process zone assuming a linear, viscoelastic constitutive behavior for the material in the process zone and used his model to analyze center-cracked panels of Solithane 50/50. Subsequently, Mueller and Knauss (1971) studied crack propagation in a linearly viscoelastic strip. Knauss (1974) also modeled the steady- state crack propagation in a viscoelastic sheet. Thus, by the mid-seventies, Knauss and co-workers had established a framework for the linear viscoelastic fracture of polymers through the application of cohesive crack models. At the same time Knauss and co-workers were developing their theories of viscoelastic fracture mechanics, Williams and Marshall (1972–1975) put forth the idea that fracture mechanics for viscoelastic materials could be treated with an approach that is similar to the traditional fracture mechanics developed for metals by simply replacing the time independent values for modulus and flow stress with equivalent viscoelastic relaxation moduli and flow stresses. The time scale for fracture was estimated by using a Dugdale (1960) calculation of the process zone size, which was divided by the crack growth rate. Concurrent with the work of Knauss and Williams and Marshall but indepen- dently, Schapery (1975a, 1975b) demonstrated that indeed viscoelastic creep crack growth could be described using the approach developed for metals by Barenblatt (1962) with the replacement of the elastic modulus with a viscoelastic modulus. He further indicated how to determine the appropriate time at which this viscoelastic modulus should be evaluated using linear viscoelasticity. Finally, by assuming a power law dependence for the viscoelastic constitutive relationship, he was able to obtain a simple analytical relationship between the steady-state crack growth rate and the applied stress intensity. 244 W. BRADLEY ET AL.
These theoretical developments, along with experimental support, have been addressed in recent books by Williams (1987), Kausch (1987), Kinloch and Young (1983), Atkins and Mai (1985), and Anderson (1991). However, there is still a general lack of appreciation for the significance of viscoelastic creep crack growth in polymeric materials development and characterization. The variety of relationships found in the literature to relate viscoelastic creep crack growth rate to fundamental material properties can further confuse the person unfamiliar with the field. The role of viscoelastic creep crack growth in the newly emerging standards to measure fracture toughness in polymeric materials needs to be addressed. Finally, the role of viscoelastic creep crack growth in slip-stick crack growth and the effect of
adiabatic heating at the tip of a growing crack need to be further clarified, especially G for systems where the critical energy release rate, 1c , increases with increasing temperature. More generally, the effect of crack growth rate and temperature on the fracture process of polymeric materials cannot be understood outside the context of an understanding of viscoelastic creep crack growth. It is to these issues that we will address our attention in this paper. In the next section, we will present an abbreviated and simplified treatment of Schapery’s theory of viscoelastic fracture mechanics (Schapery, 1975a, 1975b). In Section 3, we will show how the variety of relationships found in the literature based primarily on the work of Williams and co-workers (Marshall and Williams, 1973; Williams, 1972; Williams and Marshall, 1975) can be related to Schapery’s
results, noting especially the difference in assumptions made and how they affect
p
a_ ' K the exponent, p, in the general expression that .InSection4,wewillexplore the relationship between viscoelastic creep crack growth and unstable crack growth, including slip-stick crack growth, and evaluate when adiabatic conditions at the
crack tp are responsible for unstable crack growth. In Section 5 we will discuss the role of viscoelastic fracture mechanics in fracture toughness characterization of polymeric materials, including a brief discussion of the new American Society for Testing and Materials (ASTM) and the European Society for Integrity of Structures (ESIS) standards on fracture toughness measurements in polymeric materials. We will also consider experimental techniques for conveniently determining crack growth rates as a function of stress intensity in Section 6. In Section 7, we will illustrate the use of viscoleastic fracture mechanics in the prediction of service performance in polymeric materials and indicate the limitations in such applications at present.
2. Schapery’s Theory of Viscoelastic Fracture Mechanics Schapery’s model divides a material with a crack and subjected to loading into two regions (as seen in Figure 1), a process zone at the crack tip which has experienced highly nonlinear viscoelastic (and possibly plastic) deformation and a so-called far field region where the deformation may be appropriately characterized as linear viscoelastic, as seen in Figure 1. He then proceeds to show that the stress VISCOELASTIC CREEP CRACK GROWTH 245
Figure 1. Stress around tip of a growing crack in Schapery model (14). distribution in the far field, as calculated by Barenblatt (1962) for elastic materials, also describes the stress distribution for viscoelastic materials. He further argues following Barenblatt that the stresses at the tip of the process zone will be finite
(i.e., all singularities must cancel) if and only if
a
Z 2 1=2
1=2
K = r =r r;
1 f d (1)
0
r
where f is the stress in the process zone as a function of distance, , from the tip of
the crack (see Figure 1) and is the length of the process zone. As in the Dugdale analysis, the size of the process zone in the Barenblatt analysis is determined by
the stresses in the process zone. For the case of uniform stresses in the process zone
= y (which is a key assumption in the Dugdale analysis), where f the Barenblatt
analysis gives the Dugdale result as a special case; namely, !
2 K
1
= ;
(2)
8 y
K where 1 is the applied stress intensity and y is the yield stress in a non-strain hardening material. Schapery next proves that the displacements at the boundary of the process zone in viscoelastic materials may be determined to a good approximation by using the elastic solution, but replacing the elastic compliance with a viscoelastic creep compliance. He is then able to integrate the work required to fracture a small strip of material as it passes through the process zone (or more appropriately as the crack tip and process zone move past this strip of material) and is gradually stretched to failure using the relationship
246 W. BRADLEY ET AL.
Z
vt; r
r = r;
2 f d (3) r
0
t; r where v is the displacement at the boundary of the process zone and is the work per unit area of fracture surface created. It should be noted that Equation (3) is
essentially (stress) (displacement) calculated by integrating over the process zone boundary to give work per unit area. The result of the integration of Equation (3), which requires some very clever manipulation to get it into a form that allows use of Equation (1), is
2
~