Keywords: modular modelling, constitutive modelling, software engineering, soil mechanics, Fortran, verification.

There is an increasing need for good practice in the computational modelling of mechanical systems. Significant advances are being made in virtually every aspect of the field, and researchers face ever greater challenges in keeping up to date. A new research problem typically requires the solution of a combination of sub-problems some of which may have been tackled already; yet quite often it is not practical to take direct advantage of past efforts even in closely related areas. This applies both to theoretical advances, where the essential principles may be difficult to abstract and to apply in different contexts, and to software developments, where the effort involved in (checking, adapting and) reusing existing code may outweigh any advantage to be gained.

The use of modular modelling techniques is beneficial to the reliability and intelligibility of computational mechanical systems. Using the simple and fundamental concepts of abstraction, modularity and structure, together with the principle of information hiding [1], it is possible to design, develop and construct models and systems with integrity. Systems built in this way also have a high potential for reuse and adaptation. These qualities make the modular modelling approach ideally suited to the development and testing of new models in a research environment: they allow attention to be focused on the topic of interest with a minimum of distraction and constraint.

The type of problem considered here is characterized by a significant degree of complexity in terms of both model formulation and programming. The natural way to deal with this complexity is to break the problem down in a process of successive refinement. This is a fundamental principle in software engineering, where the necessity for modularity in the design of operating systems, compilers, hardware systems and so on is beyond question. The question of how to apply this idea to the solution of modelling problems in computational mechanics is the subject of this paper. The term 'modular modelling' is meant to suggest that the same principles that apply to the programming of a model apply also to its mathematical formulation.

The approach is illustrated by showing how these concepts apply to the development of a constitutive modelling system for natural clays. The concept of modularity is useful at several levels. It affords almost all of the advantages usually attributed to object-oriented methods, including reusability, ease of verification and testing, and the potential for further development. Crucially, in a research context, modularity allows researchers to focus on the real topic of interest, to incorporate up-to-date methods from other fields, and to avoid wasting time repeating past efforts.

1

D.L. Parnas, "On the Criteria To Be Used in Decomposing Systems into Modules", Communications of the ACM, 15(12), 1053-1058, 1972. doi:10.1145/361598.361623

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £75 +P&P)