Strain rate effects on the response of stainless steel corrugated firewalls subjected to hydrocarbon explosions

Abstract

The yield stress of a steel coupon can be enhanced beyond its quasi-static test values when subjected to dynamic load. A number of techniques, such as the overstress constitutive model, have evolved in the past to take account of the enhancement effects. This paper investigates the influence of strain rate modelling on the response of stainless steel corrugated firewalls subjected to blast loading generated from typical hydrocarbon explosions, paying particular attention in the event when the connecting weld between the firewalls and supporting angles has failed. Two failure models, namely the spot-weld model and the shear failure model, have been incorporated into the constitutive material model to account for the weld failure. The investigation involves the use of the commonly adopted overstress strain rate model and dynamic factors proposed in a recent report by the Health and Safety Executive to study the effects of high strain rates on offshore steels. Both approaches are compared and the limitations of the overstress model are highlighted.

Introduction

The effects of strain rate on structural steels have gained much attention in recent years due to the increased need to design structures for accidental dynamic loads. In addition, the rapid increase in the usage of high-strength structural steel in various industries has warranted the need for further research on the effects of strain rate. In particular, published data involving the rate effects on stainless steel are rather scarce. Structural stainless steel has shown very strong dependency on strain rate, especially in the region of 0.2% proof stress. Studies carried out by the Advantica [1] (formerly British Gas Plc) and a more recent report by the Health and Safety Executive [2] (HSE) have demonstrated some differences in the behaviour of ‘normal’ and ‘high’ strength steel. Furthermore, there are indications that the effects of strain rate are governed by parameters such as thicknesses of the structural components, stress state of test piece, stress concentration and work hardening.

In general, for most structural steel such as mild steels, both the upper and lower yield strain and stress increase with strain rate until the ultimate load is reached without yielding. However, the effect decreases as the yield strength increases. Both the ultimate stress and strain are relatively insensitive to strain rate but the rate effects can be significant on strain at initiation of strain hardening. Rates of hardening tend to decrease with increase in strain rates, approaching a rigid perfectly plastic behaviour. These observed behaviours are well documented by Jones [3], Soroushian and Choi [4] and Davies and Magee [5]. Beg et al. [6] have also noted that the effects of strain rate can propagate cracks in a brittle manner. In their work, they have presented test results on the influence of strain rates in relation to temperature, specimen sizes, weld types/sizes and welding procedures on fully welded moment connections. However, the influence of strain rate on the fracture (rupture) strain cannot be conclusive, possibly due to the difficulty of carrying out the test on material up to the point of rupture. For example, while Soroushian and Choi [4] have noted a slight increase in rupture strain for strain rates up to 6 s⁻¹, Davies and Magee [5] have observed no such effects for a range of aluminium alloys and structural steels including 302 and 310 stainless steel, for strain rates from 1.6×10⁻⁴ to 8.33×10² s⁻¹. On the other hand, in a series of tests carried out by the HSE [2], the rupture strain reduces with increase in strain rates for low carbon mild steel, and Grade 1.4404 (316L), Grade 1.4362 (SAF 2304) and Grade 1.4462 (2205) stainless steel. To add to the list of uncertainties, Dieter [7] has noted that for many materials, the elongation to fracture increases with strain rate beyond the usual metal working range until a critical strain rate is reached where ductility falls off rapidly. It is also noted that the strain at initiation of strain hardening will either increase or remain constant with increasing strain rate. Apart from the experimental efforts made by the investigators to study the strain rate behaviour of structural materials, much work has also been carried out in the development of constitutive equations to describe the material behaviour when subjected to strain rate effects. Harding [8] and Malvern [9] have made a comprehensive review on the theoretical basis and applicability of various equations. The importance of strain rate histories and microstructural considerations on the prediction of the strain rate effects has also been discussed. In a study by Stout and Follansbee [10] on 304L stainless steel, material work hardening has been found to be strongly influenced by temperature and rate of martensitic transformation. Adiabatic heating due to high strain rates and plastic straining may reduce the tendency towards martensitic transformation, possibly resulting in lower hardening rates. At very high strain rates of the order of 10³ s⁻¹, they state that a single overstress model or power law to describe the hardening behaviour may not be adequate as the strain rate sensitivity can increase dramatically.

Although there is abundance of literature on the strain rate phenomenon, many studies are applicable only to their investigated domains, and sometimes even contradictory with other studies. In this paper, a finite element study is carried out to investigate the strain rate effects on two failure models, namely the spot-weld model and the shear failure model. Both failure models have been demonstrated by past studies [11], [12] of their capabilities to describe the responses of a structural stainless steel corrugated panel when subjected to hydrocarbon explosion. Both the overstress constitutive strain rate model and the dynamic factors, called enhancement model here, proposed by the HSE [2] will be incorporated into the finite element model to take account of the strain rate effects on the responses of the firewall, in particular when weld failure occurs. Large scale experimental results from three test panels subjected to different blast loadings will be presented and compared with the numerical studies. These will be discussed in greater details in the following sections. In addition, the effects of strain rate on the post-failure behaviour of the idealised spot-welds will also be presented.