On the determination of the optimal modelling conditions for higher performance finite element analyses

In the present days, performing a finite element analysis (FEA) is mostly a day-to-day obligation in every modern and technologically updated production and design industry. Never has the need for efficient and accurate FEA been more relevant. Ensuring a correct relation between the design, calculation and analysis stages of an FEA has always been a major problem for analysts. It is frequent to adopt unsuited solid models in a misleading attempt to avoid analysis errors. Nevertheless, this often leads to a huge loss of performance and, consequently, an increase in the cost of the process. In the present paper, the authors propose a semi-quantitative procedure to determine the ideal modelling stage leading to a good compromise between the total design and CPU times and the overall quality of the results. The procedure is based on two coefficients: (i) a relation between the average error levels and the total CPU time of the analysis process and (ii) a semi-quantitative relation between the volumes of successive solid modelling phases. The proposed procedure is described and tested with a practical example. Reductions in the total CPU time are over 70% with associated error levels under 2%.