SOLIDWORKS Simulation gives users four options for solvers, these are: FFEPlus, Intel Direct Sparse, Direct Sparse, and Large Problem Direct Sparse. When setting up a simulation, many users have questions about which solver is best for their study and what the differences are between the solvers. In order to produce and run simulations in the most efficient way, it can be highly beneficial to understand what the different solvers do and where they can be best used.
For many people who are new to Simulation the choices can be quite intimidating. Thankfully there is the option within simulation to have SOLIDWORKS select the solver for you.
Automatic Solver Selection
The automatic option will choose whether to use FFEPlus by default unless certain conditions are met within the study at which point Direct Sparse is selected. Possible reasons for the switch are:
- Large displacement being turned on
- The use of remote loads or connectors
- A large number of elements in interactions
- Circular symmetry
Should the study exceed around 2 million degrees of freedom then the solver will be changed again to Large Problem Direct Sparse.
A new user should start off using the automatic solver option while they become familiar with the SOLIDWORKS Simulation environment. As their skills develop and studies become more complex it may become necessary to select which solver to use manually. To that end let’s investigate the different solvers.
Types of Solver
FFEPlus Solver
The FFEPlus solver is the solver recommended for most studies due to its efficient us of RAM and generally lower solve times. FFEPlus works by assuming the solution and then evaluating the errors present, SOLIDWORKS uses the errors to estimate a new solution and repeats the processes until the errors are small enough. This method is particularly effective on studies with over many degrees of freedom, becoming more efficient as the number increases.
It is worth noting however that FFEPlus has some limitations. It can fail if there is a large difference in the stiffness between two components or if temperature and pressures have been imported from another study. FFEPlus isn’t always the most efficient solver; if contact interactions, virtual walls, connectors, or soft springs are used then FFE plus may struggle. In those cases, one of the direct sparse solvers may be the best option.
Direct Sparse
There are a few types of direct sparse solver however they all use a similar method to reach a solution. Whereas FFEPlus iterates to a solution as the name implies Direct Sparse reaches a more direct solution to the F=Kx equation. Direct sparse solvers are often most accurate in small to medium sized problems and have stability advantages over FFEPlus when contact interactions, materials of highly different stiffnesses, or connectors are used in a study.
However, these direct sparse solvers are less efficient in RAM than FFEPlus; an issue that becomes particularly evident in larger studies. Typically requiring ten times as much RAM for the same number of degrees of freedom, this means that 16GB of RAM is the suggested minimum with 32GB recommended for anyone who will be using the Direct Sparse solver regularly. A side effect of this inefficiency is that direct solvers are limited in use when it comes to larger problems, between 1.5 and 4 million degrees of freedom the direct solver can start to fail. This necessitates simplifying the study or changing to the Large Problem Direct Sparce or FFEPlus Solver.
That being said there are some varieties of the Direct Sparse solver which can alleviate these issues. The first is Intel Direct Sparse, this allows the solver to use RAM more efficiently, this prevents solver failures but can take longer. Intel Direct Sparse also uses improved multi-core processing where possible to reduce calculation times. There is also the Large Problem Direct Sparse Solver, this works by maximising memory efficiency, this is specifically optimised to work on studies with more than a million degrees of freedom.
We hope that this has helped you to understand more about the types of SOLIDWORKS Simulation solver and when to choose each one. If you are new to simulation and looking to learn more then why not consider attending one of training courses.
If you aren't sure about the best tools or methods to evaluate your designs then please get in touch as we are always happen to recommend a solution appropriate solution.