How to Solve Error 5903 "Fail to allocate base search memory for CONSLI contact"?

Anuj Nandal_22610

I wanted to run a Clip (Non Linear LGDisp) analysis in my workstation through Optistruct Solver. While running on the solver deck, I am getting this error 5903 "Fail to allocate base search memory for CONSLI contact, required 8898208 bytes". My system has 64Gb Ram and 48GB Graphics, my hard drive also has space. Just to be sure I ran this in another system, even in servers which have more than 300Gb Ram, still getting same issue. I am pretty sure memory is not an issue, also 8898208 bytes is equivalent to 8Mb, which I'm sure can easily be allotted. Not sure what is causing it, Can anyone please help? (PS Just 2 days ago I had run a similar analysis and that worked just fine). Please find the image of the error attached for reference

Prasad Reddi_20512

Anuj Nandal_20286 said:

Thanks Prasad for a quick response,

1. The size of the model is around 60Mb, I have around 2Tb+ space in my drive which should be sufficient.

2. I did use S2S contacts, which might have caused this issue.

    I can't really use tie in this case since I want a sliding contact for my analysis

3. In my model I have material, geometric and contact non linearity, strains are also large around 15-20%. Switching off LGDisp may solve but the result may not be accurate.

Thank you for your suggestions, I'll try to implement and get back to you.

Hi Anuj,

Few points to clarify:

1. Size of the model means no of nodes and elements in the model. However, looking at the file size and the disk space availability, seems not an issue with the ROM

2. Suggestions about TIE Contacts and switching off LGDISP is some ways for debugging and to sort out root cause for the issue. Once the root cause is identified and fixed, you can restore your requirements. 

3. S2S -> N2S could work.

Please post your inputs once something is sorted out.

Cheers,

Prasad

Prasad Reddi_20512

Hi Anuj,

Normally memory issues comes due to some parameters like size of the model, complexity of the model that includes contacts and other non-linearities. So, we need to address these points one by one or by going with some combinations:

1. Size of the model: check by reducing the size of the model (no of nodes and elements)

2. Contacts: 

1. Convert any defined S2S contacts to N2S for Sliding contact
2. If point no 1 does not work, then convert all Sliding contacts to TIE contacts

3. Run the model just by ignoring LSDISP (Geometric non-linearity) defined in the model.

Also, note that the memory requirement is the ROM requirement. Its checks between the disk space needed for calculations vs available disk space. Though you may have huge disk space availability, however, while calculations and using parallel processing's, solver need very huge requirements.

I hope above points will be useful to you. Please let me know if you need any other information.

Cheers,

Prasad

Anuj Nandal_22610

Prasad Reddi_20512 said:

Hi Anuj,

Normally memory issues comes due to some parameters like size of the model, complexity of the model that includes contacts and other non-linearities. So, we need to address these points one by one or by going with some combinations:

1. Size of the model: check by reducing the size of the model (no of nodes and elements)

2. Contacts: 

1. Convert any defined S2S contacts to N2S for Sliding contact
2. If point no 1 does not work, then convert all Sliding contacts to TIE contacts

3. Run the model just by ignoring LSDISP (Geometric non-linearity) defined in the model.

Also, note that the memory requirement is the ROM requirement. Its checks between the disk space needed for calculations vs available disk space. Though you may have huge disk space availability, however, while calculations and using parallel processing's, solver need very huge requirements.

I hope above points will be useful to you. Please let me know if you need any other information.

Cheers,

Prasad

Thanks Prasad for a quick response,

1. The size of the model is around 60Mb, I have around 2Tb+ space in my drive which should be sufficient.

2. I did use S2S contacts, which might have caused this issue.

    I can't really use tie in this case since I want a sliding contact for my analysis

3. In my model I have material, geometric and contact non linearity, strains are also large around 15-20%. Switching off LGDisp may solve but the result may not be accurate.

Thank you for your suggestions, I'll try to implement and get back to you.

Prasad Reddi_20512

Anuj Nandal_20286 said:

Thanks Prasad for a quick response,

1. The size of the model is around 60Mb, I have around 2Tb+ space in my drive which should be sufficient.

2. I did use S2S contacts, which might have caused this issue.

    I can't really use tie in this case since I want a sliding contact for my analysis

3. In my model I have material, geometric and contact non linearity, strains are also large around 15-20%. Switching off LGDisp may solve but the result may not be accurate.

Thank you for your suggestions, I'll try to implement and get back to you.

Hi Anuj,

Few points to clarify:

1. Size of the model means no of nodes and elements in the model. However, looking at the file size and the disk space availability, seems not an issue with the ROM

2. Suggestions about TIE Contacts and switching off LGDISP is some ways for debugging and to sort out root cause for the issue. Once the root cause is identified and fixed, you can restore your requirements. 

3. S2S -> N2S could work.

Please post your inputs once something is sorted out.

Cheers,

Prasad

Anuj Nandal_22610

Hi prasad, the N2S conversion worked. The model is working now. Thanks for the help

Prasad Reddi_20512

Anuj Nandal_20286 said:

Hi prasad, the N2S conversion worked. The model is working now. Thanks for the help

Hi Anuj,

Thats nice to hear!

Keep on posting your queries and solving others as well.

Cheers,

Prasad