TIME TO COMPLETE COOLING ANALYSIS

[rohitr]

Hello all,

I have a problem regarding time to complete cooling analysis.I am analysing simple component with about 65000 elements.

I have analyse the component using fil+flow+warp sequence.To complete the same analysis it hardly takes 4 to 5 hours.

now wish to analyse the component using cool+fill+flow+warp. but cooling anaysis take longer time to complete near about 18 hours or more.

The cooling circuit is simple without any defects.

Kindly let me know suggetions

[teopado89]

ok thank you both for the explanation

[saurabh.essilor]

Hi everyone, 

I am also having a similar problem.  I have been running a transient cool FEM and the analysis is going on 5 days now.  I remember in the past with this same machine doing a similar mold it only took around 1 day max.  I have a very powerful machine:

128Gb RAM

16X Intel Xeon 3.4Ghz processors

1TB internal data. 

I do have RAID backup storage, could that be slowing me down all of a sudden?

[harald_goetz]

I can't add valueable information payinater, instead while reading the specs of your machine, I ask all readers here
if there is a ranklist from the Moldflow Team or AutoDesk regarding the specs for a CPU for Moldflow.

Like: Is it better to have  a 6 core with 3,7 Ghz or an 8 core with 3,2 Ghz (same familiy)?

Sounds like an odd question - but the only option to answer that depends on how good the
parallelisation of Moldflow is today (2016) and what will be improved in the upcoming version.

So a computer bought in the next 6 months will be faced with MF 2017 and MF 2018 at least.

The minimum requirement is still a joke, right? But I have discussions with our internal IT
providers every time when a new computer is purchased because of that.

Regards

Harald

[Idea_Mel]

I think 5 days is too much for that kind of hardware.

Did you unflag the "use mesh aggregation" in the solver option? It could be that a complex 3D model requires lots of equations to be solved, but 5 days is over expectations.

Regarding recommended CPUs for Moldflow, before purchasing my Workstation I read about preferred specs and got to the point that high frequency is preferred to more cores: that means 6-8 cores at about 3.5 or more GHz is best than slower multi-cpu (a multi-cpu with more cores is used when you do more analysis at the same time).

[saurabh.essilor]

Thanks for the help.  

I just turned off the Mesh Aggregation in the process settings.  

I never actually completed the simulations.  I gave up and just looked at the mold temperatures with Simulation CFD.  I'm hoping that I will be able to get the flow simulation someday.  I canceled it at day 8 of running and only being on the 10th cycle.  

I remember a day when I could run 2 at the same time in one day, I just don't know what happened.  I want to say it has something to do with the newest update of the software?  Or is it because of 2016 where I was running primarily 2015 before. 

Dell probably loves me, we are updating our computers on a yearly basis.  We have 5 people who use simulation and we have each computer stagered by a year.   So we always have a top of the line computer in house, right now it is mine and I can't believe how slow it is.  

[Idea_Mel]

What type of analysis is that? How many tetras?
Does the log file shows anything suspicious?
However, if you have a maintainance subscription you can ask for help 🙂

It's for sure Dell loves you, I can't imagine how much your workstation costs! $$

[HugoHerrera7042]

Hello,

in the past I experienced that similar problem, mostly with big parts (Dual Domain ~200,000trias and maybe some 3D's) and fiber filled materials.

What version of the software are you using?

If the model was created on a different version, have you tried to re-select the material so that the ' new software version' can identify/update the material you are using?

Another thing would be to make sure that the water inlet temperature is lower than the target temperature, so that it can converge, but I guess you would see that as a warning message if that would be the case.

Anyway, good luck and interesting post!

Hugo Herrera

[basheerahmed.shihabudeen]

Hello all,

I am facing a similar problem even using Higher RAM Machine ( 128GB ).

I am attempting to Solve Cool+Fill+Pack+warp analysis with 3D elements (Element count - 2765085), It's a quit a big model with 21 coolant circuits. 

                                       " Total Mbytes required for Cool analysis = 18799.55
                                          Total Mbytes of RAM available for analysis = 126151.41
                                          Total Mbytes of disk space available for analysis = 110385.85

                                           Cool will attempt to solve analysis in RAM. "

Using Moldflow 2016, I have launched the calculation 18hrs before still the solution is not done. In the Moldflow job manager it shows 99% of completion but in the log file is not updating after the first iteration. 

I am doubtful whether the calculation is progressing or not ? Shall I wait for some times  ? ( In the task manger of my machine it was showing 62GB Ram utilization and 7% CPU usage ) 

Can anyone please help me to fix the issue...

Thanks in Advance.

[mppkumar]

Hi @basheerahmed.shihabudeen

this is a common issue with BEM cooling; suggest to try the below changes.... 

try to convert your model into cool (FEM), this supports multi threading and analysis runs faster

before starting any analysis, always renumber the nodes (best practice which i follow)

go to - global merge - put any negative number in the tolerance and check Squeeze entity labels, this renumbers the entire model

try to loose the convergence tolerance in cooling analysis; the analysis accuracy will be lost though