It is often difficult to simulate the dynamic response of structure consisting of number of screws or bolts, since it is expensive to represent every single hardware items in the simulation. It not only consumes time in setting up the simulation, but also eats up the computational time significantly. How then a design engineer can get the best answer from finite element analysis with less number of iterations?
The most important part of performing dynamic simulation is to establish clear goals. If the need is to calculate stresses and strains in the fasteners or within the parts itself that are fastened, your simulation process will be similar. However, if your concern is only about the dynamic responses of the product, i.e. natural mode shapes, resonance frequencies or peak deflections, the simulation process will vary.
Being the local properties of mesh elements, stress and strain values in each element is important and elements are required to be small enough to capture the physics as well as are good in relation with their neighboring elements. Thus, meshing is critical part of FEA simulations along with proper application of screwed or bolted conditions. However, with fine mesh, the stress and strain results will be slow.
The dynamic response is a property obtained by considering combined aggregate response of all individual elements. Thus determining stress and strain is more about eliminating problematic elements, which are big and tend to provide stiffness to the overall geometry. On the contrary, calculating displacement means allowing every element to add their effects.
Such displacement based FEA problems however can be solved faster using few elements and applying less effort in establishing boundary conditions for fasteners. Solving the displacement problem can also be expedited by telling the solver to compute only the displacements rather than also calculating stress and strain.
While FEA users have the choice of selecting linear and non-linear solvers for dynamic problems, the fastest is the linear dynamics study. Linear here refers that the stiffness matrix of the structure is linear and do not consist of non-linear materials like rubber or non-linear boundary conditions like sliding contact. However, if in your application the effect of bolt pre-load is of an interest, non-linear solver should be selected. To perform a non-linear study faster, it is always better to perform simpler, linear test cases, so that you have reasonable answers that can speed up the time required in performing actual non-linear case.
For any complex structural vibration problem, it is often good to gain as much knowledge as you can through tests that are simple and cheap. Adding the details to this knowledge can help you in tuning the accuracy further, precisely on the regions of the model that require attention. Eventually, with all the knowledge, you can easily identify levels of mesh refinement required, types of boundary conditions that suit your goals and all other micro details that can be integrated to perform the final study.
Thus, if the purpose of your simulation is to analyze the structure for stiffness and not for stress, and when there are no non-linear materials or contact elements, linear dynamic solver can give you the desired results faster. However, if you also need to determine stress in the fasteners, you will require the CAD model to be developed with fasteners. Either way, you can make use of a bolt connector to include contact effects, pre-load stress and a detailed stress details on fasteners. This however requires using a nonlinear solver.