Hi.

Order of interpolation within the element will depend on order of an element. If you have a low order element usally calculation nodes are located only at the edges of the element and interpolation will be linear. If you have nodes in the midpoint too, quadratic interpolation can be used etc. Node number will increase calculation time but high order elements provide smoother fields of deformation/strain etc. 

Zsolt

https://en.wikipedia.org/wiki/Finite_element_method

http://math.nist.gov/~WMitchell/papers/MitchellICCS15.pdf

First order elements suffer from shear locking (if fully integrated) or hourglassing (with reduced integration) and may suffer from volumetric locking.

If your solution is expected to be smooth, second-order is usually a better choice. Second-order elements may have problems in contact and when deformation is very severe.

Details also depend on your FE software because each program has a different selection of elements.

Any reasonable introduction to FE should discuss the implications of this choice in detail.

I'm sorry to have to contradict Claudio, but the most common error in FEA is to just use the default element type and assume that this will work.

Unless the default is quadratic and your problem is sufficiently simple (like linear elasticity), there is a very high probability your result will be wrong. For example, in ABAQUS the default element is linear with reduced integration; this is prone to hourglassing.

If you plan doing serious  FE simulations, please, please take a few days to understand the basics of the element choice - it will save you a lot of time in the long run.

Martin Baeker, is there any  document you recomend to anderstand the basics of the element choice?