SW working method: complex mechanically welded part

Out of category
1

Hello

 

my design office has a recurring problem with SolidWorks: parts, and more particularly the associated drawings, are extremely heavy to handle. By this I mean that a simple change can take 4 hours, where normally 5 minutes is enough.

The problem comes from the working method. We model mechanically welded assemblies into a single ". PRT" containing more than 100 functions (see attachment). We end up with giant shots (7 well-filled A0 pages), which bring all the computers to their knees.

I would like to know how you would model and draw an equivalent mechanically welded structure (make an assembly? a part? how to reduce the heaviness of drawings?)

 

Yours sincerely.


exemple_de_piece_chaudronne.jpg
2

Hello

I think it also depends on your network. We work with sub-assemblies in the main assembly as well as parts and for MEPs it's the same. A MEP not under assembly then the overall plan with the accessories...

3

Hello S.guillaume

I would also lean towards the network: because between us your piece may be large in size but is not very big or very complex for SW.

Can you tell us the file size of your example.

On the other hand, making all the drawings of MEP in a single A0 or several AOs must represent more constraints than advantages for the fab.

I can see them cutting with scissors the part of the plan that interests them, not to mention that to feed the laser and plasma cuts; oxycoupe it must be sporty.

Certainly, you'd be better off working with a multitude of parts, then subsets, then a meta-set.

If you take your own production as an example, you know that boilermakers (affectionately called chaudracs) always use the smallest piece cut before any assembly.

It would be wise to start from the sub-assemblies that are made naturally by the boilermakers because they correspond for one reason or another to a manufacturing logic linked to the constraints of welding and handling and assembly for screwing and riveting.

In general, sub-assemblies correspond to either a manufacturing or subcontracting logic when certain parts are purchased or manufactured externally.

For example, you can't weld an inspection hatch until the main body is itself made, etc...

I think that if @OBI WAN goes through this it will surely give you an interesting opinion.

On the other hand, if you proceed, as most companies probably do, from individual plans, it means that you have to have a rigorous management of these plans (versions, etc.) and surely in the long run you will have to use PLM or other software.

But the question you ask about the slowness of the computer can lead to profound cultural change and it can be managed because change management cannot be improvised, it is very cumbersome even if it saves time later.
We can help you with that, surely.

Kind regards

EDIt: I had scratched the nickname of OBI WAN

2 Likes
4

A bit like the others for us, a sheet metal part = a file and a plan, then mechanically welded assembly, then another assembly with everything that is assembled by screwing, casing or other without welding.

2 Likes
5

PRT size:  7MB (not too many worries on that side)

DRW size: 30MB 

I just did a test: I copied the PRT/DRW pair to my hard drive. No noticeable improvement in performance. Any regeneration in the DRW takes 2 to 5 minutes. I think this comes from the fact that solidworks regenerates all the views of the 7 A0 pages.

I also think that the part should be broken down into a multitude of parts and sub-assemblies. (Boilermakers complain that the plan is too complex and unpleasant to use.)

My predecessors saw fit to consider that any set of welded parts must be modelled as a single part, with a single plan. Apparently, it's easier to manage with EPDM....  

 

1 Like
6

The example play does not seem to me to be extremely complex.
I would lean either towards file access problems (network), or, or even especially, for external references that need to be updated (classic case: hole spacing). Because in this case, it is easy to go up in memory without being aware of the entire machine...

@Zozo_mp: a multi-folio for this kind of part is not necessarily an outlier if it is structured correctly: one board for welding, one for machining, one for drilling (for example). I had a client who made individual views of the bodies to be welded (so let's say oxy-cutting).

In any case, it doesn't take 4 hours to modify.

3 Likes
7

@ stefbeno

Very interesting your remark on the multi-foliot and especially the good track you give for external references and  hole spacing.

8

@ S.Guillaume

Very good reflex to have taken everything out of the network to do the test on an isolated machine.

I'm surprised that we can say that it's easier for EPDM because when I see very, very large assemblies in aeronautics, it doesn't put EPDM on its knees.

EPDM is an indexed database (DBMS) which gives very, very fast accesses even for 3 or three million items.

On the other hand, SW basic manages flat files more or less indexed according to the settings and is not structured internally to do multi-piece, so necessarily relative slowness compared to a DBMS.

Kind regards

9

I hadn't seen S.Guillaume's message about the "off-grid" test.

The time of 2 to 5 minutes (we are far from the initial 4 hours) for 7 boards does not seem abnormal to me: for a customer who makes offshore equipment, we model much more complex parts (in terms of number of bodies) in mechanically welded mode but with fewer boards, we are on the same order of magnitude of time. For this customer we use EPDM but if it improves access times, I don't see how EPDM could improve a regeneration time.

On this kind of part, the way of modeling also has its influence: if you use functions "until the next surface" it takes longer to calculate than "one-eyed".

2 Likes
10

The 4 hours are a bit of an exaggeration, but not that far from the truth. Below, a typical process when I work on this kind of part (or assembly)

  1. Some 3D edits ==> 15 minutes
  2. 2D Opening ==> 10 minutes
  3. Addition view, and cross-sectional view ==> 10 minutes
  4. Modification of an existing section view => 20 minutes because SW crashed
  5. Rating ==> 10 minutes
  6. Regeneration/backup ==> 10 minutes
  7. Bug, views have disappeared, so regeneration/ save==> 10 minutes
  8. Thread representation bug, so regeneration/backup ==> 10 minutes
  9. it didn't work, I restart => 2 min
  10. 2D Opening => 10 minutes
  11. Regeneration, backup => 10 minutes
  12. Archiving ==>5 minutes
  13. Later in the day, damn I forgot 3 holes, so 2D opening ==>10 minutes
  14. Ect...

Sometimes it goes well. But as soon as bugs show up (on a drawing containing +- 80 views, it's very common), it takes a long time. Because each operation is slow. And if the chef asks how long it takes to make the change, the answer is 4 hours.  ^^

 

I note for the blind hole function.

 

 

 

 

11

Hello Guillaume,

I also often find myself with large mechanically welded parts, which can be spread over dozens of A0s.

After a long time of trial and error, the strategy I now systematically adopt is to have only one folio per drw file. Typically, one or more DRWs for flow rates (plates, unfolded sheets, etc.)  and then a file for each soldered set .

1 Like