Thanks for your response.
I would really like to se how would this design work atleast in wirtual world. Back in time it was big and expensive waterblock. But maybe it would fit todays threadripper cpus.
I have design it with my friend while I was on high school. It was created in Pro Engeneer Wildfire 3 i think. Maybe I still have copy of educational version somewhere so I could try to conwert it to step files if needed.
Files in latest folder are for ProE, Invertor and igs. I hope that files will work for you.
-
I maybe know why there is so much turbulence. I though that was problem with Catia but it seems, that ProE exported threads as a cylinder. Correct surface is green and fitting should not exceed that surface. That fitting were used from older desing just for screeshots. It is fitting from hardware store. Fittings like a EK have shorter thread.
-
So I ran the simulation today. I made some pictures and put that in a zip folder. If you want me to show some specific part of the simulation, just tell me.
I ran the cooler for a 250W CPU (2990WX - has similar surface area to the one of your cooler)
For the problem with the turbolence: the simulation was just not accurate enough for the numerical method to be stable in certain areas, so it converges to a solution, that doen't make a lot of sense. Aside from this digression into numerical mathematics , this can be solved by making the small cells (like in the finite element method) smaller.
After I ran this simulation, the results looked quite good, except that the inflow was not symmetrical (on one picture with the streamlines you can see it). I discovered that this is because of the boundary condition I had to set: in addition to a inlet and outlet it needed also some information about pressure. Therefore I set the surface on the inside of the fitting as source of a slight overpressure, but as the cylindrical surfaces are all separated into two parts on this model, it probably generates this pressure just on one side of the fitting making the flow assymetrical (more towards the front)
Said that, here's the ZIP
-
Thanks a lot.
It shows what we have been expecting (Good cooling under jets and ribs are mostly just for directing flow) and what we have to improve.
Area under middle fitting seems to profit from change shape to cone and shorter fitting. I see lot of restriction there.
It look like working concept for me.
I would like to see it compared in lab in real conditons with other blocks. too bad, that czech watercooling lab is closed for a years now (CTTL.cz). Very precise tests with lot of measurement. It has been bought by czech retailer and closed. Just few videos on youtube remains. For example:
CPU is replaced with heating element and switches are for sensors
-
You're welcome.
Yes, it definitely seems, that the area in the middle is the best cooled. However i moved the tubes up by half a millimeter (because at first i thought the simulation was not working because the gap was to small). So maybe the flow restriction of the cooler in that area could be a problem: Either your waterflow gets reduced compared to a standard cooler, or the pump has to generate a higher pressure. However otherwise to me this design approach seems to be really strong when it comes to cooling that specific area.
I didn't even know, there was something like a watercooler-testing-lab. It's a pity that they closed it. This would have been an interesting way of testing the cooler and comparing it to the simulation.