[Anna Tian]

Hi,

I like using Workbench ‘updates all design points’ button (as shown in the attachment1). Because it can run or rerun all the design points by just one click.

But there are five things about it I found are inconvenient (even boring) to the users. Hope someone who knows Workbench better could give me some suggestions on operating on CFX.

1. Once I click on the ‘updates design points’ button, it will be troublesome to stop the simulation task during the running of the simulation. For example, I have 10 design points. I click on the button ‘updates design points’ button. Then the simulations are running one by one as planned. But at the third simulations, I find there is a mistake in the stimulation setting. So I decide to stop the simulations queue. But it’s not easy. I have to stop them one by one. It is time consuming to stop one stimulation then wait for the start of the next simulation then stop it. In this case, I have to spend quite some time to stop 8 CFX simulations one by one. So when I get into this situation, I typically have to use task manager to shut down Workbench. Then I lost all the output data of the previous simulations (in this case, first and second simulations). Is there any nicer way to pause the simulations?

2. We can let the Workbench save the design point simulation result files. After that, if I rerun a design point, e.g. design point 2, Workbench doesn’t use the previous running result file of design point 2 as the initial condition for the new simulation. Instead, it still use the ‘Current’ (attachment2) result file as the initial condition. This is not reasonable for some cases. How to change the initial condition setting so that, for example, new simulation of design point 2 could use the previous simulation results file of design point 2 as the initial condition?

3. After I run all the simulations for all the design points, I find I need to extract one more parameter so I add a new output parameter. But after I add this new output parameter, check marks are shown in the list of this parameter (attachment3). I can’t have Workbench to read the value of that output parameter from the design points result files directly even though the simulation results files were all well saved before. In order to read the value of the new added output parameter, I have to either rerun all the simulations or open those design points simulation result files one by one to read it. Both of the two ways are time wasting. How to output new parameters from already saved design points result files more easily?

4. Every time I rerun the simulations of all the design points. The results files are accumulated with the name 001 002 003…..(attachment 4 with 007 008). This accumulation of results files occupies the hardware space very rapidly when the simulations are large. Every time when I got into this problem, I delete the old useless design points result files manually one by one to save hardware space. That’s tedious. How to better deal with this problem?

5. I always got this kind of errors (attachment5) when I update design points and run design points simulations. What does the error message mean?

Hope I have described the CFX Workbench problems clearly. If there’s anything not clear in my questions, let me know and I will explain it more clearly.

[mvoss]

I can definitely confirm all the "bugs/features" that you´ve reported. And there is pretty nothing you can do about it.

One trick i am doing is:

• run the very first dp (dp0, current)
• setup the input and output in post, create report etc.
• save the project
• before filling in the dp-table delete the results in dp0 (via the wb-container, solution)
• fill in the dp-table--> dp1 is a 2nd run of dp0
• calc. all dp

------
since dp0 is empty, no unnecessary results will be copied to the dp-folders.

For the initialization there is an option for it to use the very last result, but i am not sure if this is pointing to the "last result in the dp (e.g. *_dpx_007.res when running for the 8th time)" OR the last result from the "parent setup (which basically is the dp0 case)".

Regarding the last error(5): i think the expression in Pre is causing an error. Did you use any blanks in the names?

[Anna Tian]

Originally Posted by mvoss

I can definitely confirm all the "bugs/features" that you´ve reported. And there is pretty nothing you can do about it.

One trick i am doing is:

• run the very first dp (dp0, current)
• setup the input and output in post, create report etc.
• save the project
• before filling in the dp-table delete the results in dp0 (via the wb-container, solution)
• fill in the dp-table--> dp1 is a 2nd run of dp0
• calc. all dp

------
since dp0 is empty, no unnecessary results will be copied to the dp-folders.

For the initialization there is an option for it to use the very last result, but i am not sure if this is pointing to the "last result in the dp (e.g. *_dpx_007.res when running for the 8th time)" OR the last result from the "parent setup (which basically is the dp0 case)".

Regarding the last error(5): i think the expression in Pre is causing an error. Did you use any blanks in the names?

I use Fx and Fy as my expressions' name. Actually they already represent something. I think that's the reason of error (5). I have already corrected it. Thanks.

Also thanks for that trick. It's helpful. Anyone has any other clue to the other problems?

[Pierre1]

Originally Posted by mvoss

I can definitely confirm all the "bugs/features" that you´ve reported. And there is pretty nothing you can do about it.

One trick i am doing is:

• run the very first dp (dp0, current)
• setup the input and output in post, create report etc.
• save the project
• before filling in the dp-table delete the results in dp0 (via the wb-container, solution)
• fill in the dp-table--> dp1 is a 2nd run of dp0
• calc. all dp

------
since dp0 is empty, no unnecessary results will be copied to the dp-folders.

For the initialization there is an option for it to use the very last result, but i am not sure if this is pointing to the "last result in the dp (e.g. *_dpx_007.res when running for the 8th time)" OR the last result from the "parent setup (which basically is the dp0 case)".

Regarding the last error(5): i think the expression in Pre is causing an error. Did you use any blanks in the names?

You mentioned there is an option to use the very last result. How to select that option?

[Pierre1]

Originally Posted by mvoss

One trick i am doing is:

• run the very first dp (dp0, current)
• setup the input and output in post, create report etc.
• save the project
• before filling in the dp-table delete the results in dp0 (via the wb-container, solution)
• fill in the dp-table--> dp1 is a 2nd run of dp0
• calc. all dp

------
since dp0 is empty, no unnecessary results will be copied to the dp-folders.

I tried this method. Once I deleted the result file in dp0 folder, I won't be able to run simulation of dp1 successfully. 'The solver failed with a non-zero exit code of:1'. Because dp1 won't be able to find any initial condition. You just deleted it.


By the way, what do you mean by 'via the wb-container,solution'?

[mvoss]

Always delete via the wb-project schematic. Right-Click above "solution" (e.g. delete all results...) in the corresponding "FluidFlow" container in the wb-schematic.
I am not sure if the above works (#2) if you want to run subsequent simulations and start one from the results of the previous. I guess "last results" means the last loadable *.res in the dp-folder NOT the last results from another or previous dp.

[Pierre1]

Originally Posted by mvoss

Always delete via the wb-project schematic. Right-Click above "solution" (e.g. delete all results...) in the corresponding "FluidFlow" container in the wb-schematic.
I am not sure if the above works (#2) if you want to run subsequent simulations and start one from the results of the previous. I guess "last results" means the last loadable *.res in the dp-folder NOT the last results from another or previous dp.

By right clicking on the ‘Solution’, thereare two choices related to deleting solution. One is ‘Clear Generated Data’,second is ‘Clear Old Solution Data’.

I tried ‘Clear Old Solution Data’ firstly.It clear all the old solution result files in dp0 folder but not the latest one. So when I run the other design pointssimulations afterwards, workbench will still copy the latest result file to thedp1, dp2… folders which occupy too much hardware space.


Then I tried ‘Clear Old Solution Data’. Inthis way, Workbench does clear all the result files in the dp0 folder andnothing will be copied to the design points folders. This won’t generate anynon-necessary files and save space. But in this way, the design pointsimulations won’t have any initial solution to start from at all. Thesimulations will just start from a CFX default solution. Sometimes we need aclose initial condition to start with to accelerate the calculations.

[mvoss]

You can choose a fixed initial value file when setting up the solver and then save these settings. So the solver settings will be used for all Design Points. I guess a dynamicall ini file from a previous DP is not available out of the box but a fixed one for all dp. And since you can place the ini file anywhere (except for the user file directory of the DP), it won't be copied arround by the Design Point Loop.

[Pierre1]

Originally Posted by mvoss

You can choose a fixed initial value file when setting up the solver and then save these settings. So the solver settings will be used for all Design Points. I guess a dynamicall ini file from a previous DP is not available out of the box but a fixed one for all dp. And since you can place the ini file anywhere (except for the user file directory of the DP), it won't be copied arround by the Design Point Loop.

Do you mean I shall open CFX pre at Workbench and set a simulation initial condition as shown in the attached figure?

I tried and it doesn’t work. Workbench still copies the initial condition to design points folders.

[mvoss]

hmmm.. okay. I wasn't aware of that.
Last chance would be to use the solver entry from the wb-schematic to specify the ini-file instead of the solver spec. in cfx-pre
Carefully check the conflict settings in the wb-schematic if you already setup these parameter in cfx-pre
But i would guess that wb is copying the files anyway, but give it a try.
So, turns out to be a bug looking like a feature.

[Pierre1]

So it seems that nobody could give a solution to this annoying bug?

[chocolater]

I use parameter study in workbench design point in order to run cfx. There is two major way for initilization: update from current or update design points in order.
But i want to initilize ech design point with secefic res that i had. for example initilize dp0 from cfx006.res and initilize dp1 from cfx003.res. Is it possible?

[ghorrocks]

In that case I would not do it using workbench but I would use the command line option to write a batch file which does all your simulations, and you can define any initial conditions for any run you like.

[Veronique Pe]

Good morning,

I met I think all these problems...
In you launch your wb in batch mode, no DesignPointLog.csv will be upgraded if you need to use "slurm" command in you job.
Ansys is aware of this problem and provides no solution....

Véronique