Although it is well known that it is the deviatoric component of any stress state that drives creep, there were previously very little data to show whether the creep coefficient is identical in both dilational and compressive stress states. Recent papers by Hall,185,186 Neustroev,187 and Garzarolli188 show that creep coefficients are unchanged by the sign of the hydrostatic stress.

As shown in the next section, additional confirma­tion of the independence of creep compliance on the sign of the hydrostatic stress component can be found in some stress relaxation experiments.

4.02.9.6

Stress Relaxation by Irradiation Creep

There are situations where the applied load is initi­ally fixed and then declines during irradiation. There is usually a transient followed by an instantaneous creep rate defined by B0, but the load is constantly falling, leading to an exponentially declining load. Two examples of in-reactor creep relaxation experi­ments are shown in Figure 82, both conducted on a high-nickel alloy Inconel X-750.

Foster and coworkers have very convincingly demonstrated that creep coefficients derived from creep experiments could be used to successfully pre­dict stress relaxation for the same steel in similar neutron spectra.163

Note that the creep coefficient derived for X-750 from the EBR-II experiment is 1.6 x 10-6 (MPa dpa)-1, just slightly larger than B0 and probably enhanced by low levels of voids or bubbles in this high-nickel alloy. In NRU, however, the creep relax­ation proceeded much faster, partially due to a larger transient, but also because the steady-state creep rate is larger. In this experiment the thermal-to-fast ratio was ^10, so there was significant 59Ni enhancement of dpa rate and probably also bubble formation to enhance the creep rate. The greater scatter at very low residual stresses in the EBR-II experiment is mostly due to frictional variations on the compressed
springs and grain-to-grain interactions that come into play at low stress levels.

Stress relaxation experiments can be conducted using a wide variety of specimen types and usually yield similar results, although the transient regimes often vary with specimen geometry, preparation, and texture versus stress field relationship, as shown in Figure 83.

Creep relaxation by irradiation is important in that it can reduce the opportunity for irradiation — assisted stress corrosion cracking. It does so by decreasing internal or surface stresses produced by deliberate or inadvertent damage, as well as by reducing internal stresses arising from welding, abrupt cooling, etc. Figure 84 demonstrates the radiation-induced relaxation that occurs in a weld that proceeds with a creep compliance of B0 that is independent of the sign of the hydrostatic stress.189 Therefore, it appears that the creep compliance B0 can be confidently applied to any stress state.

As a rule of thumb one can anticipate that by 10dpa, >90% of any preload will be relaxed even in the absence of a transient. The fractional unload­ing is not dependent on the magnitude of the preload as long as the bolt or component was not loaded beyond the yield point.

Stress relaxation in structural components of operating reactors is not always operating in isolation. Frequently, a component experiences time-dependent stresses that develop with time as a result of the growth or movement of adjacent components. In pressurized water reactors there are bolts that join baffle plates to former plates. These bolts are usually cold-worked 316 but the plates they join are annealed 304 stainless, a higher swelling steel. Initially, the bolt will start to relax its preload but if the plates are swelling faster than the bolts, then differential swelling will begin to reload the bolt. Additionally, if a bolt is replaced with a fresh bolt, the reloading can be even stronger due to larger amount of

differential swelling. Figure 85 shows several calcu­lated histories of bolt loading for PWR-relevant tem­peratures and dpa rates.1

While bolts are generally preloaded to a specified level, there is always some range of attained loads. It is difficult to measure the stress level in a bolt while it is still in place, but a rough measure of the remaining load can be made from the torque required to remove the bolt. While this is not an exact measurement with friction, corrosion, irradiation-induced self-welding, and other complications possibly participating to define the torque, Figure 86 shows that the measured torques are in reasonable agreement with predictions of creep equations based on experiments conducted in BOR-60 fast reactor. The fact that most of the data lie above the predictions may indicate that many of the bolts are indeed being reloaded by differential swelling to some degree.