topic Re: Silicon structrure thermal results going bad in CFD Flex 2016 in CFD Forum

Silicon structrure thermal results going bad in CFD Flex 2016

Anonymous — Wed, 24 May 2017 19:29:54 GMT

HI all,

I have added the "boundary conditions" with heat flex, Just wanted to study the how much heat transferd in different parts of the silicon structure.

3 Volumes I given boundary condition as : 70 W/mm3, and another 3 volumes as :700 w/m3. The results should be controlled .. within 50-70 degree centigrade in different parts of the silicon structure. But it is going beyond that..

I am attaching my report. as well my cfd flex file. Can anybody help me where my Boundary condition or meshing is going bad.

Thanks

Re: Silicon structrure thermal results going bad in CFD Flex 2016

Jon.Wilde — Thu, 25 May 2017 05:49:21 GMT

Your first problem is Boundary Conditions (BCs). You cannot have internal surface conditions like you have. What are the two 10C's meant to be representing?

If you want that entire tube to be 10C, you can assign this to all three surfaces that comprise it but then you need to suppress the part from the mesh (so they are then on the boundary of the domain and not sandwiched between parts).

You have a heat load applied to the air volume, does it somehow generate heat? This seems a little strange right now 🙂

What is actually happening in reality? At the moment, this does not seem like something realistic to me.

Thanks,

Jon

Re: Silicon structrure thermal results going bad in CFD Flex 2016

Anonymous — Thu, 25 May 2017 10:29:51 GMT

Hi Jon,

Many thanks for your quick reply. Could you pls guide me what are the boundary Conditions I should apply, for my below Goal.

Goal is:

I just wanted to see that , the 3 volumes "7um_ILD_metlizazation"( parts) generate the 70 w/mm3 heat, in the entire assembly I created.

I just wanted to see how these sandwiched heat generators spread the thermal temperature across the different pats.

That is it my goal is !! I mean, calculate what are the temps in different parts, after spreading of the heat, from those above 3 sources.

what should be my BCs here ? I applied to air temp as 25W(or 13W), just to say , the experiment is carried out in a room temp. what I should use as my boundary condition for surroundings air ?

If I don't apply 10 C for the both end of the copper wire, I get the "Heat Transfer was disabled because either no thermal boundary conditions were found or the applied conditions are insufficient."

I changed 10C to 1 C also, now I am getting same as old HUGE temperature (10c time) in the other parts.. as you can see in attached Report. Could you pls guide me, to meet my goal.

Many thanks in advance for your great help.

Chandra

Re: Silicon structrure thermal results going bad in CFD Flex 2016

Jon.Wilde — Fri, 26 May 2017 06:32:22 GMT

OK, this makes a little more sense. You need to aim for reality, the more you can tell CFD the better, it won't make guesses for you.

If this is in air, model the air closer to reality 🙂

Use the small guide here, this will allow you to set the ambient temperature to the air and give CFD a nice reference to start with.

Assigning a heat generation to air is a little strange, unless there is something adding heat to it somewhere. If not, simply assign the heat loads to the solid parts, as I suspect is closer to reality 😉

Avoid assigning internal BC's, like you did to the copper. If there is heat extracted here somehow, let me know how it is extracted and we can figure out a way to model it accurately.

Does that make sense and is it useful?

Jon

Re: Silicon structrure thermal results going bad in CFD Flex 2016

Anonymous — Fri, 26 May 2017 06:50:32 GMT

Hi Jon,

Many thanks again for your reply and how to model a air medium.

But my questions still remains same that " How much heat loads to the solid parts" I should assign ?

1. I should start applying " Total heat generation" as "Zero" for rest of the all sandwiched solid parts except those "70 Wats" 3 volumes of the geometry ?

2. Wht the temperature range is "HIGH " as below (red color)?

Part	Minimum Temperature	Maximum Temperature	Volume Averaged Temperature
2017_final_assembly_V2.0 (LevelofDetail1) 10u_dia_45u_copper_tsv:2	6772.88	261601000	6983350
2017_final_assembly_V2.0 (LevelofDetail1) 0.5u_sio2_liner:3	10	2671740000	1128260000
2017_final_assembly_V2.0 (LevelofDetail1) 10u_dia_45u_copper_tsv:3	10	370356000	26736100

etc..

Thanks

Chandra

Re: Silicon structrure thermal results going bad in CFD Flex 2016

Jon.Wilde — Fri, 26 May 2017 07:23:59 GMT

You are welcome 🙂

Don't apply zero, just leave them with nothing assigned.

For the other three parts, you will need to assign the value that corresponds to the level of heat lost from each. I can't really comment as I am not familiar enough with your model.

I have no idea why your temperature range is high yet.

What units are those?

What are you expecting?

Did you make all the changes we discussed?

Thanks,

Jon

Re: Silicon structrure thermal results going bad in CFD Flex 2016

Anonymous — Fri, 26 May 2017 11:55:51 GMT

Jon,

I have applied the pressure a zero on 2 surfaces of the air "ext.volume". and still i get high values of temp. PLs find my BCs as below.

But i get the below warning with the results, I attached.

*** Heat Transfer was disabled because either no thermal boundary conditions were found or the applied conditions are insufficient

boundary conditions

Type

ASSIGNED TO

Total Heat Generation(70 W)

2017_final_assembly_V2.0 (LevelofDetail1) 7u_ILD_metalization_layer:6

2017_final_assembly_V2.0 (LevelofDetail1) 7u_ILD_metalization_layer:4

2017_final_assembly_V2.0 (LevelofDetail1) 7u_ILD_metalization_layer:2

Pressure(0 Pa Gage)

Surface:141

Surface:143

To answer your questions:

What units are those? ==> These are semiconductor (silicon, silicon dioxide, metal inside the silicon dioxide parts).

What are you expecting? ==> I am expecting the heat transfer values should be around 50-70Cengtigrade range, as per my REsitance eqqvivaent models.

Did you make all the changes we discussed? ==> Yes. I did make, pls find my report. Suggest further.

Thanks,

Chandra

Re: Silicon structrure thermal results going bad in CFD Flex 2016

Jon.Wilde — Fri, 26 May 2017 11:59:53 GMT

Hi Chandra,

Please refer to the guide again 🙂

You missed a step that tells you to assign a temperature BC - you'll see it there at the bottom of the page.

Thanks,

Jon

Re: Silicon structrure thermal results going bad in CFD Flex 2016

Anonymous — Sat, 27 May 2017 07:35:01 GMT

Hi Jon,

I applied as 20 C as my room temp now for 1 surface of the "Ext. Valume" created. But, the tool is throwing some garbage values still.

I am afred to believe these values. I am not sute what BCs I should apply to meet my 1 Single goal. Pls find my report, gives more details now.

I am expecting some 50-60 C in all parts of the structure I have, after heat is spread. Later I want to vary the size of the Copper rod etc to see how much temperature varies etc...for my white paper. 1 good result itself is not available since a week, with CFD flex. 😞

Re: Silicon structrure thermal results going bad in CFD Flex 2016

Jon.Wilde — Sun, 28 May 2017 11:25:19 GMT

I see that you have applied 70W/mm3, is that correct? I would have expected a 'Total Heat Generation', otherwise you would be applying a LOT of energy!

You have something like 2000000000000W being applied right now 😉

You need Auto Forced Convection off, as that runs flow first and thermal second - not OK here.

Gravity would also need to be on for air to move right? Apply a value of 1 in the downwards direction (Solve window)

I would also use ADV5 (Solve->Solution Controls->Advection) and the Mixing Length turbulence model.

Does that help?

Jon

Re: Silicon structrure thermal results going bad in CFD Flex 2016

Anonymous — Wed, 31 May 2017 12:15:12 GMT

Dear Jon Sir,

I have applied the BC's as you have guided. Pls find the attached full report.

Finally it took 96 (4days) to conclude the 137 Iterations. In between my laptop was getting auto off.

But still I have below concerns.

1. The parts now showing me 25 Degree Centigrade temperatures, as you can see in attached report. But this was the applied, Air volume temp.

2. The run time is so huge, how to reduce it ? Previously suppose to be 16-20 Mins max before I applied your specified BCs.

Pls note that as per my Resistance eqvi-Valance models. The temperatures of the different parts should be : 50-55 degree centigrade as below.

	r=0.5um
T0	54.09912235
T1	54.20028945
T2	53.36079649
T3	53.37204361
T4	52.92691614
T5	52.91541274
T6	52.91760774
T7	53.30747447

So could you pls advice me , what exact BC's and tuning of the tool I should do, to get more or less above numbers ? Here the T0- T7 are the temperatures of the Silicon bod, metal Plate, copper rod etc. r is the radius of the copper rod inside the structures.

Best Regards,

Chandra

Re: Silicon structrure thermal results going bad in CFD Flex 2016

Jon.Wilde — Wed, 31 May 2017 12:20:52 GMT

This might be related to your gravity, I suggested a value of 1, which will be 'earth'.

Your flow rate is enormous, which I hope is related to this, it would certainly explain why nothing is heating up.

Try that again and/or share the CFZ 🙂

It shouldn't take more than a few hours to run either, that also seems strange, although it does depend on your machine of course.

What is your spec?

Re: Silicon structrure thermal results going bad in CFD Flex 2016

Anonymous — Thu, 01 Jun 2017 11:56:05 GMT

Hi Jon,

I did use the "earth ". The snapshots are attached. I don't know the usages of gravity directions to use (xyz).

Is it depending upon my geometry drawings?

Anyways, Could you pls suggest further, to reach my goal. anything CTM models that I should use here?

I have attached the CFZ file also.

Pls suggest further.

Many Thanks

Chandra

Re: Silicon structrure thermal results going bad in CFD Flex 2016

Jon.Wilde — Thu, 01 Jun 2017 12:48:25 GMT

Hi,

I see, weird then. Could you share the CFZ file please and I will take a look for you 🙂

Thanks!

Jon

Re: Silicon structrure thermal results going bad in CFD Flex 2016

Anonymous — Mon, 05 Jun 2017 09:06:39 GMT

Hi Jon,

sorry for the delayed reply. Pls find the CFZ file.

I kindly request you to pls suggest further.

Thanks.

Chandra

Re: Silicon structrure thermal results going bad in CFD Flex 2016

Jon.Wilde — Mon, 05 Jun 2017 10:12:18 GMT

Hi,

I opened the most recent, what is the difference between the two?

My thoughts:

Refer to the guide I shared before for air domain sizing, this is very small compared to what it needs to be
Leave Intelligent Solution Control on, was there a specific reason for you wanting to turn it off? This usually just leads to instability and should only be done with good reason
You need to switch the air to 'variable' or it's density cannot change and it will not rise

Re: Silicon structrure thermal results going bad in CFD Flex 2016

Anonymous — Tue, 06 Jun 2017 08:59:58 GMT

Hi Jon,

Thanks for ur advice.

I did follow ur info. I am getting the report as attached. Pls have a look at my run snapshot.

Still nothing working for me.. 😞

I wanted to have the at least near approximate results expected. Any other suggestions on this ?

Many Thanks,

Chandra

Re: Silicon structrure thermal results going bad in CFD Flex 2016

Jon.Wilde — Tue, 06 Jun 2017 10:54:30 GMT

Is your air domain now suppressed from the mesh? It is crossed out in the tree, which suggests it has been excluded from the model...

Also, the report says velocities are zero.

Did you also make the domain larger?

Thanks

Jon

Re: Silicon structrure thermal results going bad in CFD Flex 2016

Anonymous — Wed, 07 Jun 2017 10:17:13 GMT

Jon,

Sorry, This time I made it active. somehow it was suppressed. and attached the results again.

Domain remains same. which velocities did you mean? It has inlet/outlet flow rates.

Thanks

Chandra

Re: Silicon structrure thermal results going bad in CFD Flex 2016

Anonymous — Wed, 07 Jun 2017 10:37:56 GMT

Jon., I am thinking is this air

I am thinking is this air around the assembly making this mess?

if so, how to get it rid off ?

Thanks,

Chandra