[Ocp-europe] RuggedPOD v2 sealing feedbacks and improvements

Carter, Bill bill.carter at intel.com
Tue Jul 7 20:46:46 UTC 2015


I'm assuming that the aluminum is a cast plate.   Does the heat transfer property change (gets worse) with cast Aluminum?
As I recall from experience in my own garage building race vehicles, a machined surface transfers heat better.

The O-ring design is exactly what I envisioned - nice work!    Again, from my racing experience, an O-ring design can benefit from alignment pins to keep the plates from shifting during assembly and avoid potential tears or damage to the o-ring.    If you are concerned with loads, maybe look at a double O-ring option that can be tested during the next prototype build.

Bill


From: ocp-europe-bounces at lists.opencompute.org [mailto:ocp-europe-bounces at lists.opencompute.org] On Behalf Of Jean-Marie Verdun
Sent: Tuesday, July 07, 2015 12:45 PM
To: ocp-europe at lists.opencompute.org; team at ruggedpod.qyshare.com
Subject: Re: [Ocp-europe] RuggedPOD v2 sealing feedbacks and improvements

Hi Bill,

This is what the team worked on today ;).

Attach are some interesting data. We ran static mechanical simulation as to evaluate the stress that the top cover is experiencing. As you can see it is quite high, but displacement with a thickness of 12 mm aliminum or 5mm stainless steel shall be ok. The 5 mm stainless steel is the current design (but it is really heavy), while the 12 mm aluminum is the one we are thinking about.

So we plane to move to aluminum top cover.

Regarding o-ring this is the new design (Calque 1.png). We moved to a 4 mm O-Ring instead of 3.5. The locking mechanism doesn't use anymore threaded rod, but "screw" which are bolted. We are also welding a new part at the top of the tank as to improve flatness and rectify die casting poor quality process.

Feedbacks welcome !

Jm



Le 06/07/15 23:38, Carter, Bill a écrit :
Can you use a large rubber o-ring around the entire top cover.
The o-ring will allow for movement and expansion/contraction without affecting the seal.

Bill


From: ocp-europe-bounces at lists.opencompute.org<mailto:ocp-europe-bounces at lists.opencompute.org> [mailto:ocp-europe-bounces at lists.opencompute.org] On Behalf Of Thierry Dupeux
Sent: Monday, July 06, 2015 12:16 AM
To: jean-marie.verdun at splitted-desktop.com<mailto:jean-marie.verdun at splitted-desktop.com>; team at ruggedpod.qyshare.com<mailto:team at ruggedpod.qyshare.com>; ocp-europe at lists.opencompute.org<mailto:ocp-europe at lists.opencompute.org>
Subject: Re: [Ocp-europe] RuggedPOD v2 sealing feedbacks and improvements

Hi Jean-Marie,

I think a screw with large head going from the top of the top cover into threaded hole, avoid need for glue. I think it would be better solution and nicer look.
However, I cannot comment on dimension and number of screws.

Best Regards, Thierry.


------------------------------------
Thierry Dupeux
mobile: +33 (0) 6 75 86 48 89

From: ocp-europe-bounces at lists.opencompute.org<mailto:ocp-europe-bounces at lists.opencompute.org> [mailto:ocp-europe-bounces at lists.opencompute.org] On Behalf Of Jean-Marie Verdun
Sent: Sunday, July 05, 2015 10:33 PM
To: team at ruggedpod.qyshare.com<mailto:team at ruggedpod.qyshare.com>; ocp-europe at lists.opencompute.org<mailto:ocp-europe at lists.opencompute.org>
Subject: [Ocp-europe] RuggedPOD v2 sealing feedbacks and improvements


Hi Folks,

I spent some time drawing the concept of the top sealing that we used for RuggedPOD v2.

The sealing concept is quite easy to understand. We roughly used a 3D torus joint which is standing at the top of the edge of the tank. This joint is placed under pressure from the top cover which is bolted to the tank using threaded rod which have been pre-assembled into the tank.

This principle is currently not working efficiently and we must find the root cause of the issue. The first thing we observed is that most of the theaded rod are getting bent or unlocked from the tank. They are currently based on stainless steel a strong material with a 3 mm diameter.

We used also specific glue to lock them down into the hole which was threaded.
The top surface of the tank edge has been post processed to be flatened as much as we could.

Roughly the assembly process is the following one:-

- Solder the tank side to each other
- Solder the bottom
- Post process the edge of the top surface of the tanke to flaten it
- use CNC to create 3D torus joint placement area
- Put the thread rod in place
- Run the whole assembly

Looking at some math, when we put the tank under vacuum, I believe that the pressure on the top cover is slightly moving to 1kg/square cm as the pressure inside the tank is going to 0.007 bar, which means about 1.4 tons for the whole cover. This pressure is transferred in some way to the tank edges, and stainless steel elasticity is exercising a strong force on the threaded rod which are breaking there glue link and create some small leakage.

I might be totally wrong, and this is where I need your help !

We urgently need to have a look to this issue, as I would like to upgrade asap this design flow. The following options might be explored and will be discussed with some of our manufacturing partners this week like:

- Embedding the threaded rod during the panel manufacturing process rather than gluing them
- Use higher diameter and longer threaded rod
- Put one at each corner (currently there is none, and most of the leakage are observed in that place)
- Put them in a Z shape around the 3D torus joint instead of a standard aligned position
- Increase the number of Threaded rod
- Use bolt with wide pressure area (we are using standard 3mm bolt, perhaps using 3mm hole bolt but 6mm contact diameter surface might be better)
- We could also increase the thickness of the top cover, but what will be the right one ?

I like to avoid increasing the thickness of the tank panel (which will increase the contact surface with the top cover), as it will lower heat exchange capacity.

Feel free to jump into the discussion ! JJ will try to run some static mechanical simulation tomorrow with solidworks as to understand how much pressure each threaded rod have to address. But we are not expert in mechanical simulation and if you have some options to help us feel free !

Jm
--

  Jean-Marie Verdun
  Splitted-Desktop Systems
  President
  Batiment Aristote
  Parc des Algorithmes
  Route de l'orme des merisiers
  91190 Saint Aubin FRANCE

  jean-marie.verdun at splitted-desktop.com<mailto:jean-marie.verdun at splitted-desktop.com>

Â
[cid:image001.jpg at 01D0B8B3.FFB05F70]

Splitted-Desktop Systems is a proud member of the Open Compute Foundation
and a Gold member till July '15. We can provide you consulting and engineering
services related to any OCP hardware. We also drive RuggedPOD and Daap Open
Hardware projects








_______________________________________________

Ocp-europe mailing list

Unsubscribe: http://lists.opencompute.org/mailman/options/ocp-europe



Ocp-europe at lists.opencompute.org<mailto:Ocp-europe at lists.opencompute.org>

http://lists.opencompute.org/mailman/listinfo/ocp-europe

--

  Jean-Marie Verdun
  Splitted-Desktop Systems
  President
  Batiment Aristote
  Parc des Algorithmes
  Route de l'orme des merisiers
  91190 Saint Aubin FRANCE

  jean-marie.verdun at splitted-desktop.com<mailto:jean-marie.verdun at splitted-desktop.com>


[cid:image001.jpg at 01D0B8B3.FFB05F70]

Splitted-Desktop Systems is a proud member of the Open Compute Foundation
and a Gold member till July '15. We can provide you consulting and engineering
services related to any OCP hardware. We also drive RuggedPOD and Daap Open
Hardware projects


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.opencompute.org/pipermail/ocp-europe/attachments/20150707/e81115f1/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image001.jpg
Type: image/jpeg
Size: 7706 bytes
Desc: image001.jpg
URL: <http://lists.opencompute.org/pipermail/ocp-europe/attachments/20150707/e81115f1/attachment.jpg>


More information about the Ocp-europe mailing list