Five steps that helps you to set up and solve FEA problems faster

One of the things I learned from my previous company is a focus tool to improve the productivity. It is nothing but writing down your results expected in doing a particular work and the different steps that may lead to the results. Although tough in the beginning I found this to be irritating, but soon I found that this greatly improved my productivity. This bought in   focus to the work undertaken. On thinking into the matter I found that this tool can be applied to everything we do. In my this article I will be giving the steps that need to be planned before an FEA analysis so that we have more clarity on the steps we do and help us sail through the process. Below listed are the steps to be planned before starting your FEA:

1. Master the phenomenon

This is a onetime investment. Every time a new phenomenon is presented to be simulated we have to make sure we that we know every aspect of it. We have to know the different forces that act on the model in the real world. Many of the . This provides insight choosing the right type of analysis, all the useful data that can be obtained from a particular phenomenon. Many of the physical phenomenon does not behave in the ideal way in real worlds. Digging deeper can help us to reveal this. We will also be aware on how to best simplify our FEA model so as to save time without compromising the accuracy of the outcome. This also helps us to determine if we are going to need a static or dynamic analysis to capture the phenomenon.

1. Start with the end in mind

This is the step were we will decide what are the results we need by going through the whole process. Simulation of a particular phenomenon can give us a whole array of different results. But to save resource, time and get clarity on the process we need to identify what are the particular results we are interested in.

 This step is particularly important since because the analysis procedure depends on the results we need to derive. Stresses for example require finer meshes and much better model than if you are interested to get displacements.  We have to get clear of the agenda of our analysis like if we are carrying out the analysis to optimise an existing model, to measure the life of the product, try out different materials to select a particular material, etc.

1. Garbage in garbage out

Garbage in garbage is one of the foremost principles of FEA analysis. FEA software gives output even if the input values are wrong. These are designed to process the input values. The beauty and bane of recent FEA analysis software is that it can give you beautiful results even if you have very less technical knowledge.

This step involves identifying all the input values required to simulate the phenomenon studied in point 1. After identifying the quantity of the inputs required we have to find out the quality of the inputs, i.e the values of the input provided. Even a single wrong assumption in the input value can have large variations in the result.

If you are doing CFD analysis you may need to know the temperature dependent viscosity, the density and the specific heat of different interacting fluids to get accurate results. Sometimes we may not get some input values and it is tempting to just fill in the blanks with any parameter that may seem appropriate, but this may lead to bad results if it isn’t carefully done with a deep reflection of the quality of data being used. The loads in real condition may also vary from the laboratory test values.

1. Both the end and the means are important

Everybody must have heard of the quote the end justifies the means. It is derived on the philosophy that even if the means of achievement is wrong if the end result is good then the wrong way can be justified.

But in FEA this rule is not applicable because here the method of solving a particular problem is as important as getting the result. In FEA the type of method determines everything like the time of solving, the accuracy of the results, how the software assumes the phenomenon etc .We may be deceived by the apparent external simplicity of the model because even simple models can be ruled by complex behavioural laws that require advance computer simulation to be solved.  

One of the things that come under the type of methods is solving the problem  in the simplest way possible. This includes capturing the phenomenon such that it takes the least possible time without sacrificing the accuracy.  After all why would anyone spend a week analysing a model if the results can be had in a day. For example analysing a model as a truss instead of a solid model can be a huge time saver. Using a shell element instead of a solid model in instances where the accuracy is not affected can be another example.

1. Make your result bullet proof

One of my managers used to say that the duty of the engineer does not end with finding the results for a particular problem, the engineer has to go a step ahead to create a procedure to validate his own results. Due to the large number of input values that can go wrong in a FEA problem the principle should be applied on a stringent basis.   

For many the criteria for getting a wrong answer are not able to completely solve the model. Once we get the result we feel that we have done our purpose. But people with experience know that getting through till the result is only the first step. Every time we analyse something we have to thing about the different assumptions we have made and how these assumptions may not be completely true in certain cases.

One of the methods of verification can be changing our boundary conditions, forces applied, geometry dimensions etc to see if the result obtained by a particular method is consistent with the input. Hand calculations should be used to compare the obtained results and made sure they are in sync.