Yes, it really is automatic. In fact, the user spends absolutely no time creating grids for CONVERGE. All of the grid generation is performed internal to the CONVERGE solver (we call this “run-time grid generation”). As a result, highly complicated geometries with moving boundaries are just as easy to set up as simple geometries with stationary boundaries. Great care has been taken to minimize the CPU usage required for run-time grid generation. In fact, this process is so fast that the grid is recreated at the start of each computational time-step when necessary!
There’s no such thing as a free lunch. Am I sacrificing geometric accuracy by using automatic run-time grid generation?
You’re right, there is no such thing as a free lunch. In this case, we’re paying for your lunch! All of the complexity that a user would normally have to deal with when making a mesh has been taken out of your hands and is done at run-time using our innovative techniques. Sacrificing geometric accuracy? You’re actually gaining accuracy over codes that require a grid ahead of time because with CONVERGE, adding resolution at the boundaries (i.e., through fixed embedding or adaptive mesh refinement) results in improved accuracy in near-wall flow predictions. With other codes the accuracy of the geometry is limited to the resolution of the original grid.
I really like the idea of not having to create a grid, but will I still have control over mesh resolution?
Absolutely! In fact, you have ultimate control since changing the global mesh resolution is as easy as changing the values of three numbers (dx, dy, dz) in the input files. Local mesh resolution can be added in specific areas for all or part of a simulation using our fixed embedding options. The locations of increased resolution can also be added automatically using our Adaptive Mesh Refinement (AMR) option.
Yes! CONVERGE is a general CFD code. It is especially powerful for complicated geometries with or without moving boundaries.
I’m writing an article or giving a presentation that includes results generated with CONVERGE. What do I need to know?The proper references for our products are “CONVERGE”, “CONVERGE Studio”, and “CONVERGE Lite” (no TM symbols). To cite the CONVERGE Manual, use the following: Richards, K. J., Senecal, P. K., and Pomraning, E., CONVERGE (v2.3), Convergent Science, Madison, WI (2015). To refer to the company, use “Convergent Science”. Convergent Science maintains a bibliography of papers that cite CONVERGE. Please tell us about your paper after it has been published! Email email@example.com.
You need to run CONVERGE Studio with an OpenGL-compatible video card (OpenGL 3.3+ compatible recommended).
To manipulate geometry in CONVERGE Studio, we recommend the following minimum RAM:
- 10,000 triangles – 200 MB RAM
- 120,000 triangles – 475 MB RAM
- 200,000 triangles – 650 MB RAM
- 400,000 triangles – 1 GB RAM
- 800,000 triangles – 1.8 GB RAM
- 1,000,000 triangles – 2.0 GB RAM
For the CONVERGE solver, the amount of recommended RAM depends on the size of the domain; the level of grid refinement; the number of species, passives and reactions in the mechanism; and other simulation-specific criteria. A small, simple simulation (e.g., an engine sector case) can be efficiently executed in serial on a single machine with less than 4 GB of RAM. Larger simulations (e.g., a port-fuel injected, full-cylinder simulation using the SAGE detailed chemistry solver) may require several multi-core machines to achieve a solution in a reasonable time. Contact the Convergent Science, Inc. Support Team (firstname.lastname@example.org) to discuss processor selection for your simulation.
The screen output also contains error notifications and warnings that are useful for diagnosing problems at run-time. Error notifications are usually followed by a termination of the simulation, while warnings do not necessary indicate a severe problem.
The following errors occur because of missing inputs or a mismatch between the specified input and the input CONVERGE expects based on the type of simulation you specify. Examples of missing input error notifications are:
initialized species not found in mech.datmeans that a species with an initialized value in initialize.in is not part of the reaction mechanism specified in mech.dat.
must provide a file for an arbitrary moving boundarymeans that you have not supplied a motion file (e.g., arbitrary_motion.in) for an arbitrarily moving boundary.
CONVERGE also generates notifications when there is a mismatch between the inputs or when you incorrectly specify the inputs. Examples are:
boundary between disconnect regions can only translate or rotatemeans that you have specified a motion type for a boundary between two disconnected regions other than TR (Translating) or RO (Rotating).
disconnect group between regions 0 1: CLOSED while attached boundary region 1 is movingmeans that a disconnect group between two regions has been defined as CLOSED while a boundary in region 1 is actually moving.
(All OPEN/CLOSE events must occur only when the boundaries adjacent to the disconnect triangles are stationary.)
CONVERGE will also generate error notifications when computational issues arise. For example, when models or sub-models do not converge to specified tolerances. An example is:
equilibrium solver did not converge.
Warning notifications are generated by CONVERGE during run-time. Unlike error notifications, warning notifications do not necessarily imply a severe problem. A warning notification is usually issued if the inputs you supplied to CONVERGE are not typical for that type of simulation. Examples of warnings are:
warning in inputs.in, mult_dt_spray is greater than 1.0. this may lead to large errors in spray calculations
warning in boundary.in, for tke inflow boundary condition for boundary 1 dirichlet or intensity is not specified
CONVERGE, like other software packages that are designed to run on a large number of cores, does not run as efficiently when hyper-threading is enabled. This is true even when the number of processes in a simulation does not exceed the number of physical cores on the machine. For best performance, disable hyper-threading before running a CONVERGE simulation.
On Linux, what does the
File locking failed in ADIOI_Set_lockerror message mean? On Windows, what does
On a Linux machine, this error may be the result of MPI interacting with an NFS-mounted file system. To be able to use mpi_write_flag = 1 on an NFS-mounted file system, use NFS version 3.0+ and ensure that the client machine is mounted to the file system with the noac (no attribute caching) option.
Learn more about how CONVERGE helps you quickly and accurately solve your CFD problems.