Design Study - SolidWorks

3D Design Volume Model
,
1

Hello

I am testing the Design Study module. To do this, I made a 10x10 sketch and then a 7 mm extrusion.

This extrusion is the dimension that I want to vary to know for what value I obtain a volume of the prism at 1000 mm3.

It's obviously at 10 mm but I want it to be through an optimization calculation that I get this result.

I deliberately unbalanced the Min from 5mm to the Max at 12 knowing that the right result is 10.

Here's what I get:

That is a better result at 9.81!

Anyone know where the problem comes from?

Thank you

2

If we balance the limits, for example from 5 mini to 15 Maxi, there he finds 10mm for the thickness so 1000 mm3 of volume.

3

My question does not have many amateurs

; °)

4

Good evening Alain

The question may be how many people use this module  :-)

Not to hit my head =======> too late I'm already far away:-)

Kind regards

3 Likes
5

I don't know the module (but I'll be interested in it...).

The 2nd table you put in, indicates 3 iterations. Is there a limit to this?
What result do you get by putting 8 and 15 as the limit?

1 Like
6

Here is the result for 8 minis and 15 Maxi:

10,19 !

It's weird that the choice of interval varies a result that should be unique.

7

I've never used this module, either...

8

I just tested it out of curiosity and same thing with Interval, it seems that SW is taking a step and can't round up.

On the other hand, with interval with step and a 1mm pitch, no worries, it works.


intervalle_et_pas.png
3 Likes
9

I think that according to the interval it makes a division and par on an unrounded step. And stops with a result that is as close as possible but not fair.

EDIT: the 1st calculation is exercised like this, it takes half between the min and max interval, hence the answer 10 for a min of 5mm with a max of 15mm.

So for min 5mm and max 12mm he takes half and looks at the value (here 8.5mm) then redoes a calculation with a ^not calculated I don't know how and comes up with a value close to 9.8125mm.

 

4 Likes
10

It smells like a dichotomy that is not really optimized or that doesn't go far enough.
I just did a test, that's 5 "scenarios" (which I'll call iterations). I didn't see a difference between "fast results" and "high quality" (same result).
I tried to increase the interval (8-25), I have the same result, with 1-25, the value found is 11!

If you have access to support, I'll be curious to have their answer.

3 Likes
11

Yes, I think like you @sbadenis, we have something close in terms of the way we calculate. So the choice of intervals directly influences the accuracy of the results.

3 Likes
12

A trick; by framing with a limit imposed at minimum 999 mm3 to 1001 mm3 Maxi, we have a result closer to 10 mm: 10.01!

13

It's strange I just did a test on SLD 2015 Sp5 and it works well, in fact, my interval corresponds to a number of scenarios with one step, it calculates all the widths between 5 and 120 mm long in steps of 5 in order to always have a volume of 1000mm3. On the other hand I don't have the "iteration" boxes

 

Cdt

edit: As for me, I have the same problem, at the beginning I had put interval with step, but when switching to interval, well I have the same problem. 

2 Likes
14

I did another test with a sphere. I want to know the radius to have a volume of 10000 mm3. By calculation we would have a radius of 13.365...

Here is the result:

Approaching but not the exact result...

15

By moving the cursor, we can see how much the iteration is done; In the example below, it's 0.12 so I can't get the exact value according to the minimum and the maximum chosen

The optimum is not 100% guaranteed, and I don't see how to reduce the iteration to refine the calculation.

16

At best we can use the local trend to refine the mini/Maxi limits and start the calculation again:

17

Indeed, the optimization option has never really convinced me. The notion of steps should not exist since SW is expected to find the solution!! SW should determine on its own by convergence with an increasingly tight interval by minimizing the gap between the result and the target. When I was at school, I made a lot of algorithms for that.

I have turned your example in all directions and have found no better results than you.

Personally I use the "imposed limits" with the "is between between" option, and I refine the step manually. In the cases I have had to deal with so far, it was enough. I recently modeled a cup and I needed to know the level for a capacity of 250 ml at 1%. I did the iterations by hand by tightening the pace, it took me 5 minutes.

 


tasse.jpg
4 Likes
18

It's a shame that the tool doesn't allow more finesse for optimization.

Maybe an evolution to ask SW

3 Likes
19

In the help, we have for the High Quality study:

Finds the optimal solution from a large number of iterations ( Box-Behnken design) and displays the initial scenario, the optimal scenario, and all iterations.

20

And for quick results:

Finds the optimal solution from a few iterations ( Rechtschafner design) and displays the initial scenario and the optimal scenario.