3D runners or beam runners?

[Anonymous]

Hello everybody,

I was wandering why i see some important difference in these 2 simulations (please see the images attached):

1: I imported the model of a part, meshed in 3d and created the runner an the gate inside moldflow, then i rotated it for 120° two times and drew the central sprue and hot gate. In this way I obtained the model in the first image attached.

2: For the second analysis I imported the model with the cold runners and sprue already drawn in 3D. I then created in Moldflow the same hot gate.

The mesh for this part and the previous one are very similar: i used the edge lenght as criterion to buld it and all the parameters for the mesh creation were set the same.

 

Please note that the geometry of the runners was drawn to be the most similar possible for the 2 studies.

I chose the same material for both the analysis (gold indicator in fill pack and warp) and the exact same process settings.

What I obtain are very similar results in filling and pressure for the cavity, but different results for pressure in the runners, pressure xy at injection and other results as flow front temperature.

So, did i do something wrong or not considered something? And wich result should I consider the best?

I have my opinion but I would really like to know yours.

Thanks

[Sanjay_paul]

Hello,

Did you change the meshed runner property to cold gate surface for the second study?

You need to select the meshed runner and then right click on mouse will show you change property type . In change property type you will find the cold gate surface.

Regards

Sanjay Paul

[madhan_rm]

Hi,

Yes, you got to change the mesh property to cold runner & cold gate respectively before running analysis.

And just in curiosity am asking, why would such sprue configuration is required?  Are you simulating something like just 1 quarter of mold..

Best Regards,

Madhan.

[Anonymous]

Thank you both for your contribution, unluckily the result that i obtained is almost the same.

I duplicated the study and i changed the mesh properties as you suggested.

You can see in the picture on the left the new simulation with the sprue and gate changed to cold runner properties, on the right the old simulation.

As you can see, in the injection pressure nothing has significantly changed...

I also wonder how changing the property prom part to cold runner would have done some difference in the results that i am looking for... For the pressure for exemple, the pressure fall shouldn't be related just to the geometry? (of course considering all the other parameters the same, as well as the mold temperature etc.) 

About your question Madhan yes, is just a part of the mold, but i'm taking it just as exercise to better understand all those differences we are seeing, i'm not really interested in the value of the results right now, just in why 2 simulations that should give almost the same answer are actually so different...

Does anyone has other ideas or explanations?

Thanks to all

[ikv]

Hi,

interesting question.

I have some other potential explanations for the difference, although I am not so sure about the second one:

1.) As far as I know, the assigned cross sectional geometry of the beam elements is always transformed into an equivalent circular cross section with the same flow resistance ("hydraulic diameter") for the computations within Moldflow. As your channels seem to be rectangular in shape, the heat transfer from a circular reference channel might be different from the real rectangular geometry, which is exactly represented in your 3D-model - maybe this causes different cooling situations in the runner and thus different temperatures and pressure drops.

2.) I think I read about it somewhere (cannot remember where) that the (radial) temperature distribution, which is calculated in a beam element, gets lost in the transition from the last beam element of the runner system to the first tetra element of your 3D mesh - so I understood it as a transfer problem from the 1-dimensional beam element to the 3D tetra elements (maybe the average temperature is transfered, but the distribution gets lost. which might cause the different temperature values in the part as well) - not really sure about that one, though.

3.) Could it have something to do with the discretization of the 3D runner system? From the pressure curves it seems like the biggest difference comes from the final tapered part of the runner system (gate region) - maybe a finer discretization (more elements over the thickness) might result in a different result as you can better capture the temperature gradient from the surface to the center of the flow channel... Maybe you could compare the frozen layer fraction of the beam and the 3D channels in the gate region?

Just some ideas...

Thomas

[thomas_b_peeler]

I have noticed differences as well.  From experience with pressures in the past, I seem to trust the beam elements more for the majority of the feed system, but I do like the results that can be obtained using 3D gates, although sometimes I need to use beams for the entire runner sytem..

[Anonymous]

Thanks to both, starting from your observations i tried to do some more tests.

Firstly i refined the mesh at the gate for both the beams and the 3d version, then i created a "midversion" with just the submarine part of the gate modeled in 3d and the rest in beams. As you can see in the images below, the new mid-version seems to be more similar in the results with the all 3d one. 

As you expected the most important role is probably played by the gate part, infact in the new version also the injection time-pressure XY plot is much more similar to the 3d version.

In the frozen layer resuslts you can see a significant difference between the 3d and the mid-3d  just in the runners part, so i wonder if maybe is really something that has to do with the way the software calculates the heat transfer?

Then i created the same injection system but using runners and gates with circular shape, in this way if the problem was the fact that the beams are always resized as a circular section with an equivalent area, i should have obtained much more similar results between the simulation with beams or 3d.

As you can see below i plotted the pressure at injection location for all these simulations. It appears to me that the difference stays the same even changing the shape of the section, so would consider that factor negligible.

Now if I want to choose the most correct one, wich one is it? If i well understand Thomas states that he would go with the midway version.

First time i'll have the opportunity to check the results with a real situation i'll try to update the conversation.

If anyone would like to give more support or ideas is really welcome.

[thomas_b_peeler]

Gregorio,  You used a hot runner into your 3D cold runner, so not sure if the following is related to your results but I thought I should add this to this discussion anyway because it could be related. 

I performed a test a few months back where I used a 3D mesh for everything including cold sprue, runner, gate and part.  The material used was Celanese Hostaform S 9364 POM (gold standard).  I noticed higher than expected (I believed) volumetric shrink and Z warp in part.  I then took exactly the same study and added a short hot beam to the start of the cold sprue (2 mm diam.) and injected into that instead of directly into the cold sprue (2 mm diam. at small end).  The resulting volumetric shrink decreased by a substantial amount as well as the z warp.  I was surprised that such a small change could have such an impact on the results.  Especially when the nozzle acts like a hot runner anyway, so this change is not invalid.  After inspection of the injection area near the cone on the original model (just 3D cold sprue), it was evident that excessive cooling was occurring in that area.

After further testing I noticed that the size of my gate contact area had a major impact on the results in question.  Even when using a contact area the same as the short hot beam diameter (2 mm), the vol. shrink and warp results were as high as in the first analysis.  When I increased the contact diameter beyond the actual size (4 or 6 mm), the vol shrink and warp decreased to a similar magnitude to the short 2 mm hot beam results.

I now know what to look for and a work-around; however, I think that maybe this should be looked at by the Autodesk Moldflow team.

Thank you.

[Anonymous]

Hello Thomas,

this is another important information for sure, even if i'm not sure it's related to my situation since both the simulations have hot runner and gate in beams as first elements. Anyway it underlines the fact that something is missing in this whole runners simulation part, probably there is some differences in the conditions when you choose one way instead of the other that we don't know.

It would be helpful if someone from Autodesk could give us some explanation about these aspects.

I'll maybe try to open a ticket and see if i get an answer.

In the meantime merry Chrismas to everybody.

[Anonymous · Accepted solution]

Hi everybody.

I got help from Paul of Autodesk Support to better understand what's happening. (thanks Paul)

The problem is that beams are linked to the 3d model with just one node and it doesn't matter the angle of the submarine gate and what i see graphically, the section considered by the solver for the gate is always the one given from the gate dimensions that i chose in the gate properties.

So if i have a gate like the one here below and i want the same results using beams i will need to have the final section of my beam to be equal to just the part of the section of my 3d gate wich is actually in contact with the model. So in this case the section of the beam in the final part of the gate will need to be equal to the upper part of the gate in my 3d model, the one wich is actually "touching" the part.

[Anonymous]

Hello Gregorio,

you posted this topic some time ago, but maybe it is still relevant.

Did you look at the shear rate within the gates for both modelling versions?

From my experience the shear rate in gates modelled with beams is often higher than in tetra modelled gates (also very depending on the element size in the gating area).

So if it is the case in your model, this might be an explanation to the significant difference in the pressure results. As with higher shear rates the viscosity and also the pressure decreases. In addition to that the temperature increases (due to shear heating what can be seen in the temperature at flow front result, see attached picture) which leads also to a drop in viscosity and pressure.

Best Regards,

Michael

[Anonymous]

Hello Michael,

thanks for your answer.

I don't have that study anymore so i can't check the shear now,  but i think i remember that at the beginning they were quite different.

Anyway as you can read in the last post before yours, after changing the dimension of the gate made in beams in order to have it matching the actual section of the 3D one (just the part of the section that is letting the flow pass through) i anyway got similar results between beams and 3D.

Best regards