Mathematical Reliability
As far as mathematical reliability goes, a general rule of thumb is that well-established methods can be economically used at about 90% accuracy. Greater accuracy can usually be obtained at greater cost. We rarely recommend this, as the extra benefit does not normally justify the extra cost.
Quality of Input Data
Material property reliability can be an issue. The simulation can only be as good as the input data. We sometimes need to have properties measured, as manufacturers' data sheets are too often inadequate. We have good contacts with companies and universities that can perform virtually any measurement we need.
Geometric Accuracy
This sometimes presents the opposite challenge. We can take anything from back-of-the envelope sketches through fabrication drawings to CAD models. CAD models often contain detail that is not really important for performance assessment, and which we have to remove, as it would be very expensive to analyse every last detail.
|
|
What Is Finite Element Analysis: Is Finite Element Analysis Reliable? |
|