If you use first-order elements, then yes, usually they will shear lock with full integration. Switching to reduced integration (i.e. one Gauss point for a linear element) will lead to a softer behaviour of your element and has the danger of hourglassing, so you may need to employ hourglass control.

If you cannot switch to second-order elements, incompatible mode elements may also help, but their behaviour strongly relies on the element not being distorted, so these elements will yield good results only if they are as regular as possible.

If you use first-order elements, then yes, usually they will shear lock with full integration. Switching to reduced integration (i.e. one Gauss point for a linear element) will lead to a softer behaviour of your element and has the danger of hourglassing, so you may need to employ hourglass control.

If you cannot switch to second-order elements, incompatible mode elements may also help, but their behaviour strongly relies on the element not being distorted, so these elements will yield good results only if they are as regular as possible.

Many thanks dear Dr. Baeker for your valuable answer. Based on your answer, to avoid shear locking or hourglass effects, the best way is to use 2nd order element, is that true?

Moreover, can we use the 20-node brick element instead of 27-node hexahedral element and avoid the mentioned problems?

Quadratic hexahedral elements should be o.k. for avoiding shear locking; the reduced int. elements can show very weak hozrglassing, IIRC, but this is usually not a problem.

20-node bricks are quadratic, so they should be fine in most cases.

Note. though, that there are situations where quadratic elements may perform worse than linear elements (may happen in contact, sometines with large localised deformation).

I am modelling the large deformation analysis of thick plates based on the total Lagrangian method, Do you think that employing 20-node elements would be a wise choice?

It should be at least a good starting point.

You could always try different element types for a test case to see whether effects are large (this will also give you some insights into whether your mesh is fine enough, although a mesh convergence study is always a good idea...)

It also depends on your material law - if plastic deformation is large, it may be good to use 20-node bricks with hybrid formulation.

 Thank you very much Dr. Baker. Your comments have truly helped me.

Hi,

Shear locking happens only in those problems in which bending of the part is involved. Hence we use reduced integration. Using reduced integration (evaluating the primary field variables at a single gaussian point i.e at the centre of the element ) would decrease the stiffness of the overall element. This is the cause of hourglassing (unusual deformations) of the element. One could enhance the hourglass control mode or use quadratic elements. However, using quadratic element formulation would increase the computation time. If one uses the enhanced hourglass control mode, it is always recommended to verify quantitatively (checking the artificial strain energy). As a thumb rule, the ratio ALLAE/ALLSE should not be more than 5%.

If someone finds any flaws or could improve the answer, please feel free to do so.

Thank you very much,

BMNK