Hello. I am carrying out some CFD analysis on SolidWorks Flow Simulation and when I click on Run for an external flow it can take up to 40 minutes. I only have the mesh resolution set at 3 which is quite low. I have checked and 100 % of my CPU is being used during this time. I have an Intel i7 CPU and 8 GB of RAM installed on my PC. Any recommendations?
Thanks Jared,
40 hours?? I don't feel so bad now.
You say to make sure my solution is in ram. How can I check that?
look at the task manager and how much memory is being used, compare against your installed ram
Hi Chris,
Thank you for your answers. I was looking at my Ram and was noticing the solver is only on my CPU. Can you help me about how i can adjust it to use my RAM as well?
How many iterations does it take to converge?
Have you used goals to converge or is it using a default of 4 travels?
I recently went overboard and added dummy volumes to apply unique initial conditions in numerous locations of a large model and my convergence dropped from 35ish hours to under 7 hours.
The closer you get your initial conditions to final the faster it will solve. This isn't a great deal of time savings on a once-run model, but if you're doing development and making changes and re-simulating the time savings to converge can easily justify adding more initial conditions.
good point on initial conditions
Thanks Chris,
I have since closed the Solver window and I cannot find how many iterations it converged on. How can I find that retrospectively?
I will be carrying out many simulations on different models so any hints for reduction of time would be great. What did you do with the dummy volumes?
When you load the results there will be a topic underneath (cut plots, surface plots, etc.) that is Goals Plot. If you select this it will make a plot of your goals vs (iterations or travels or time). You should also be able to RMB on the Load Results and select summary, I believe this tells you. Or when you load a set of results the right side of the pane will show the iterations.
I have a large heat transfer model that has certain areas at room temperature and others at 1000C+. If you set a single fluid temperature to start it has to iterate long enough to raise or lower the local temperature. But if you make a volume that represents a certain region you can apply an Initial Condition for the gas in that area.
Example:
I have a 10ft long tube that has 0C gas entering one end.
Heat is applied to the outside of the tube.
The gas exits around 1000C (based on a previous long simulation solve)
If I break this up into 1ft sections where each one is approximately 100C hotter than the last (or whatever the previous results indicate).
Make a cylinder the ID of the tube and put it inside as a solid plug. Repeat this along the length (likely with a linear pattern in this case).
Choose the first plug and set an Initial Condition. Click the box to "disable solid body". Flow will now only use this body to define the conditions of any point inside of it, but it will be hollow (fluid). Set this initial condition to 100C.
Repeat this step for each volume but setting the temperatures of each one hotter than the last.
Now when you run Flow will start with a temperature progression instead of assuming a single uniform temperature. The more complex your system the more this helps. Especially if you have forced flow in areas resulting in rapid heat transfer but natural convection in others which can take forever to converge.
Thanks for your help Chris
how fast do you want your simulations to be exactly?
Thanks Jared. It's not that I want my simulations to be done in seconds. I am just trying to find out is there ways in which I can increase the performance. Once I know that my machine is being used to the best of its ability I can work away.
are you still stuck on smoethign specific?
40minutes, not bad. i have some that take 40hrs.
best way to speed things up > optimize the analysis setup
after that your i7 should be fine
get an ssd
make sure your solution is in ram (which it probably is if it only takes 40 mins)
how fast do you want it to be? sometimes things just take time