Ricardo Wave Experiment Problems

[Marios]

Hey guys,

i am using Ricardo WAVE 8.1 for the simulation of our engine and I am facing a problem when I try to run the experiment function. The problem is that the software looks like it is simulating but at the end instead of getting the results i get an error that says: 'Summary file does not exist for: (a bunch of files)' and 'There is insufficent data for curve fitting'. Did anyone else faced this problem and know a way to solve it??

[Marios]

Hey guys,

i am using Ricardo WAVE 8.1 for the simulation of our engine and I am facing a problem when I try to run the experiment function. The problem is that the software looks like it is simulating but at the end instead of getting the results i get an error that says: 'Summary file does not exist for: (a bunch of files)' and 'There is insufficent data for curve fitting'. Did anyone else faced this problem and know a way to solve it??

[samphlett]

Contact WAVE support via your supervisor. The FSAE grant provides for this.

[ibanezplayer]

I've run into this many times. Typically whenever this has happened, a regular simulation( Run > Wave > Screen mode) wouldn't produce results either. Troubleshooting in this mode I find to be easier because it allows you to see what part of your model is causing problems.

On the occasion that your "screen mode" simulation runs fine, but your wave experiment fails I have found that I had multiple cases set up, with something "iffy" about the first case compared to the others.

But as Samphlett said, you should get 10 hours of support through Ricardo and they are very good.

Hope this helps...

[Hub]

Like ibanezplayer said, running less case is more stable. Also "central composite" seems to be more stable for me.

[Mikey Antonakakis]

I'd say also make sure your range of values is reasonable. I'm assuming you're running an experiment for runner lengths. Sounds dumb, but the only time I ever really got those errors was when I was using a huge range of values, unreasonable lengths, and/or had the wrong units (i.e. mm instead of inches) on one or more of my runners. Go back and double check your model. As far as I can remember, it was usually my fault for messing up an experiment.

[samphlett]

Central composite just means more test points points (and a higher order fit), so more data to fit to. I don't understand the comment about stability. If a point doesn't work, it just doesn't work. If too many points don't work, the curve fit won't work through lack of data (too few equations).

[Hub]

Yes samphlet, but in my case I can run larger value with composite than full/half factorial. Don't know why.

[samphlett]

Well Hubert, there's clearly some information missing here. Each point is a standalone steady-state case. Probing the .out files for the failures is where I'd start. Send it in.