FG Baustatik, Stahlbau, FEM
Design codes require to demonstrate that certain limits of inelastic deformation are not exceeded during the service life of a structure. If the loading is cyclic, inelastic strains may accumulate cycle by cycle and may exceed specified allowables after a number of cycles before a stationary state (elastic or plastic shakedown) is achieved, or the inelastic strains may grow unboundedly due to a ratchetting mechanism.
In principle, inelastic deformations can be calculated by performing evolutive (step-by-step) inelastic analyses. These require specific information, which is, however, not always available, such as detailed constitutive modelling and loading history. Furthermore, evolutive inelastic analyses are very costly. Therefore, simplified inelastic analyses are desirable to provide at least partial information about structural behaviour: more specifically, upper bounds on, or estimates of, elastic-plastic-creep deformations.
Some simplified methods are envisaged by design codes (as pointed out in Chapter 2). However, they are based on specific configurations of geometry and loading or they adopt heuristic assumptions, the reasonability of which is not always evident for general applicability within the scope of these codes. Accordingly, design codes seem to require improvements.
Several simplified methods are reviewed in the present Report and might serve as alternatives to those suggested by design codes. Simplified methods can be grouped in two classes: (a ) procedures intended to determine a safety factor against a critical event of the global structure (such as collapse); (b ) techniques apt to provide information on local quantities (such as inelastic strain) associated to inelastic structural responses. Procedures of class (a ) and some of their recent extensions are briefly surveyed in Chapter 3. Subclasses of category (b ) are discussed in the subsequent Chapters, with emphasis on their operative peculiarities and on their practical usefulness or potentialities.
Upper bounds in plasticity (Chapter 4 and Appendix A) can be computed by various approaches, basically by satisfying a set of equations and inequalities and by carrying out some optimisation procedure. To within the consequences of modelling errors, residual post shakedown quantities are guaranteed to be bounded from above: this circumstance is referred to by the adjective "rigorous". A number of bounding inequalities can be proved. Usually, the better (lower) the bound, the more expensive is its computation. However, computational advantages over other simplified approaches can hardly be ascertained in general.
Upper bounds in creep (Chapter 5) are based on general rigorous mechanical foundations, but the applications available mostly concern particular cases employing "ad hoc" imaginative, sometimes heuristic assumptions which are not easy to transfer to other cases. Mostly, elastic-perfectly plastic material behaviour and the Bailey-Orowan creep model are assumed.
Some simplified methods, such as the British shakedown method (Chapter 6), intend to estimate residual stress fields after elastic shakedown (the British method also, in certain circumstances, after plastic shakedown). They adopt empirically corroborated conjectures, rather than rigorous arguments. Material hardening is neglected.
Zarka's method (Chapter 7 and 8 and Appendix B) provides estimates of the mean strain in case of elastic shakedown and, in addition, a lower and an upper estimate of strain range in case of plastic shakedown, by adopting some heuristic assumptions. The validity of these assumptions is difficult to assess in practical applications. Material hardening is required.
The method developed by Ladevèze and coworkers (Chapter 9 and Appendix C) is not, strictly speaking, a simplified method, in the sense that it provides the same kind of information as rigorous evolutive analyses. The simplification lies in the solution process, which can be stopped after a few iterations since each one of these concern the whole
time interval of interest. Thus estimates are achieved of the structural inelastic response over a large time interval (much larger than the time step in an evolutive analysis). General material models are admitted.
None of the simplified methods reviewed in the present Report can directly be recommended for general practical use in nuclear design situations. However, some of them seem to have at least no less potentialities than the methods mentioned so far by design codes. Further work is necessary to clarify the conditions under which they are advantageous.
The European Fast Reactor (EFR) collaboration with the EFR Associates Design and Construction Rules Committee, and the R&D Agreement (AGT9B) produced significant developments in design-by-analysis procedures for high temperature plant. Many of these developments are judged to be relevant to the non-creep conditions of LWR plant, and for this reason, the CEC DGXI Working Group Codes and Standards supported this study to review and make recommendations on their potential application for improving LWR design code procedures. The topics considered are judged to be those where the most significant and relevant developments have been made and the list, although not exhaustive, is as follows:
- Negligible creep criteria
- Design-by-analysis procedures for weldments
- Shakedown design rules
- Design-by-analysis methods for tubeplates
- Buckling rules
- Interaction diagrams for assessing ratcheting
- Rules for the prevention of elastic follow-up in piping
- Strain range enhancement
- Constitutive equations for inelastic analysis
- Margins on Level D criteria
- Zarka's method
Life assessment of a structure subject to cyclic loading rests on quantifying strain accumulated prior to shakedown and the strain range experienced after plastic shakedown has been achieved. Few methods exist to predict these quantities. Zarka's method is one of these methods. It is evaluated by analyzing several examples of structures and comparing the quality of the results obtained and the numerical effort required with evolutive analyses by using a commercial Finite Element program.
As elastic–plastic fatigue analyses are still time consuming the simplified elastic–plastic analysis (e.g. ASME Section III, NB 3228.5, the French RCC-M code, paragraphs B 3234.3, B 3234.5 and B3234.6 and the German KTA rule 3201.2, paragraph 7.8.4) is often applied. Besides linearly elastic analyses and factorial plasticity correction (Ke factors) direct methods are an option. In fact, calculation effort and accuracy of results are growing in the following graded scheme: a) linearly elastic analysis along with Ke correction, b) direct methods for the determination of stabilized elastic–plastic strain ranges and c) incremental elastic–plastic methods for the determination of stabilized elastic–plastic strain ranges.
The paper concentrates on option b) by substantiating the practical applicability of the simplified theory of plastic zones STPZ (based on Zarka's method) and – for comparison – the established Twice-Yield method. The Twice-Yield method is explicitly addressed in ASME Code, Section VIII, Div. 2. Application relevant aspects are particularly addressed. Furthermore, the applicability of the STPZ for arbitrary load time histories in connection with an appropriate cycle counting method is discussed.
Note, that the STPZ is applicable both for the determination of (fatigue relevant) elastic–plastic strain ranges and (ratcheting relevant) locally accumulated strains. This paper concentrates on the performance of the method in terms of the determination of elastic–plastic strain ranges and fatigue usage factors. The additional performance in terms of locally accumulated strains and ratcheting will be discussed in a future publication.
In case of cyclic loading, strain may accumulate due to a ratcheting mechanism until the state of shakedown is possibly achieved. Design Codes frequently require strain limits to be satisfied at the end of the specified lifetime of the structure. In addition, the strain range is required for performing fatigue analyses in case of plastic shakedown. However, little guidance is usually provided by Design Codes on how the accumulated strains and strain ranges are to be calculated, and some of the guidelines implemented in Design Codes are not well founded and may therefore be misleading. This is, for example, true for the ASME B&PV Code, Section III. Of course, strains and strain ranges can be determined by means of incremental elastic-plastic analyses, which require to go step-by-step through many cycles of a given load histogram until the state of shakedown is reached. This is rather costly in terms of engineering time and numerical effort. As an alternative, simplified methods can be adopted, e.g. the Simplified Theory of Plastic Zones (STPZ) as used in the present paper. Being a direct method, effects from load history are disregarded. The theory is described shortly and illustrated by some examples. It is shown that the Simplified Theory of Plastic Zones is well suited to provide reasonable estimates of strains accumulated in the state of elastic and plastic shakedown at the cost of few linear elastic analyses.
If a mechanical structure is to be designed for operation under cyclic loading, primarily two kinds of failure must be guarded against: (1) low cycle fatigue which may occur due to strains cycling between two states (controlled by the strain range exceeding twice the yield limit);
(2) ductility exhaustion which may occur due to accumulating strain from one load cycle to another.
These two kinds of failure are local failure modes so that strains need to be calculated and then assessed by comparison with code allowables such as the 1%, 2% and 5% strain limits set by the ASME nuclear codes. Elastic-plastic strains can be calculated by incremental (or step-by-step or evolutive) analyses. Unfortunately, this can be extremely costly if thousands of cycles are required to achieve shakedown. Therefore, simplified elastic-plastic analysis methods are desired allowing to obtain specific information at reduced effort, nevertheless accounting for the main features controlling strain such as kinematic hardening. Zarka’s method, early versions of which are available since twenty years, appears promising to provide both strain ranges and accumulated strains in the saturated cycle, i.e. after shakedown has been achieved. However, several attempts to use this method in the nuclear industry failed to qualify the method as a reliable analysis tool. This was due to several reasons:
(1) the publications describing the method were written in a highly scientific language the design engineers in industry were not familiar with;
(2) in some cases Zarka’s method provided excellent results (compared with incremental analyses), but bad ones in others.
Nevertheless, there remained some interest to uncover the potential of this method. For that purpose some calculations are performed for simple configurations of structure and loading (so that the structural response can be interpreted relatively easily). More insight into the performance of the method may thus be gained in terms of computational steps to be followed, the numerical effort required, the quality of the results obtained, and the sensibility with respect to material data and load level.
The basic idea of Zarka's method is to redefine the elastic-plastic problem by an equivalent elastic problem with suitably defined modified elastic material parameters and initial strains. This requires estimating (and iteratively improving) the geometry of the plastic zone and of transformed internal variables. A particular class of material models is admitted, the simplest of which is the linear kinematic hardening model.