1. Introduction

The verification of scientific software consists in ensuring the absence of errors in the programming of the implemented modeling. A classic and very reliable approach is to compare the result of the software to that of an analytical solution. Unfortunately, for complex and/or non-linear problems, analytical solutions become very difficult to obtain. It is in this context that the Manufactured Solutions Method is particularly interesting.

The Manufactured Solutions Method is a method for verifying scientific software. It is a systematic method of obtaining analytical solutions for problems that can be very complex and non-linear. Its principle is simple: we explicitly give ourselves the solution to look for in the form of an analytical expression and, from this solution, we build the data (blockages, loads) necessary to obtain this solution.

It is interesting to note that this method is very commonly used in fluid mechanics. Thus the « American Institute of Aeronautics and Astronautics » (AIAA) included it in its software quality assurance standards in 1998 [bib1] This practice is much rarer in solid mechanics; thus the « American Society of Mechanical Engineers » (ASME) only integrated it into its standard in 2006 [bib2], i.e. much more recently. This is not linked to any particular difficulty, as extensive studies in solid mechanics have been carried out using this method [bib3].

In the rest of this document, we will explain the principle of this method on a simple example and then the applicable validation procedure. We will then detail the IT implementation in Code_Aster.