# QUESTION:I have 2 models of a MEMS scale membrane. One has EMT126 elements with VOLT=0 on all nodes (and so should not exert any force on the membrane), the other has no EMT126 elements at all. Nonlinear geometry effects are on, and I apply pressures in a prestress load step then remove the pressures and recalculate the displacements. Final results between the 2 models differ, but they shouldn't because the EMT126 voltages are all zero. What's happeneing?ANSWER:The presence of EMT126 in the database forces ANSYS to turn off solution control. In the model without EMT126, solution control is, by default, on. If you turn off solution control in the model without EMT126 elements, the results will match those with the model that has EMT126 elements in it.The stiffness in the thickness direction of a 1 um thick membrane spanning 560 um is very small. Normal deflections of ~1e-4 um in this direction are therefore probably numerical noise. This probably explains why the use of different nonlinear algorithms (SOLCON,OFF/ON) predict different results.

 QUESTION:I have 2 models of a MEMS scale membrane. One has EMT126 elements with VOLT=0 on all nodes (and so should not exert any force on the membrane), the other has no EMT126 elements at all. Nonlinear geometry effects are on, and I apply pressures in a prestress load step then remove the pressures and recalculate the displacements. Final results between the 2 models differ, but they shouldn't because the EMT126 voltages are all zero. What's happeneing?ANSWER:The presence of EMT126 in the database forces ANSYS to turn off solution control. In the model without EMT126, solution control is, by default, on. If you turn off solution control in the model without EMT126 elements, the results will match those with the model that has EMT126 elements in it.The stiffness in the thickness direction of a 1 um thick membrane spanning 560 um is very small. Normal deflections of ~1e-4 um in this direction are therefore probably numerical noise. This probably explains why the use of different nonlinear algorithms (SOLCON,OFF/ON) predict different results.