Moldflow Insight Forum
Welcome to Autodesk’s Moldflow Insight Forums. Share your knowledge, ask questions, and explore popular Moldflow Insight topics.
cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 

Cross-WLF viscosity model

18 REPLIES 18
Reply
Message 1 of 19
luca.scrosta
3864 Views, 18 Replies

Cross-WLF viscosity model

Hello,

I'm in charge of moldflow analysis of a “speaker grill” made in ABS with melt flow index 5.

The nominal wall thickness is 3 mm, but the effective pass-section between each hole is 1.2 mm (very thin section).

Comparing the results with real molding, there is a big difference between the peak pressure calculated and the one measured on the injection molding machine (the measured pressure is more or less the double of the one calculated). Probably the pressure-dependency of the viscosity has to be considered, but the "D3 parameter" in the "Cross-WLF viscosity model" of the UDB material characterization is set to ZERO.

Does anyone can suggest me a standard/typical value to be assigned to "D3 parameter" for a generic ABS with low fluidity?

 

Many thanks,

Luka

18 REPLIES 18
Message 2 of 19
abakharev
in reply to: luca.scrosta

Well, the D3 coefficient is usually of the same order of magnitude but lower than b6 in the dual-domain Tait model (PVT). As a wild guess, I would start with D3=b6/10 and if it improves the results I would try b6/4 and b6/2. Strictly speaking when putting non-zero D3 we have to modify the other coefficients (especially, decrease n and modify D1) to fit the experimental data. AFAIK you can order Autodesk Moldflow Lab to measure rheology at elevated pressure and fit the pressure-dependent viscosity data for you.

 

Usually the pressure dpendence of viscosity kicks in after 100MPa injection pressure, if the pressure is below 100MPa then D3 is not important.

 

How do you model the grill? Do you use midplane, dual-domain or three-D?

Message 3 of 19
ealizadeh
in reply to: luca.scrosta

hi,

I simulated a rubber injection molding process with reactive molding module of moldflow, but the injection pressures calculated by the software has almost -35% deviation from reality. I aim to introduce pressure dependance of viscosity into model, but this is not predicted for reactive molding. according to some researches, these effects are not negligible for injection molding simulations.

please advise me what I can do for this problem.

Thank you

Message 4 of 19
abakharev
in reply to: ealizadeh

I don't have much experience with thermosets/rubber but for thermoplastics the effect of pressure dependence of viscosity is usually small if the maximal pressure is well below 100 MPa. I guess the same rule of thumb is true for thermosets. What was the maximal pressure measured or predicted for the process? I would not bother with the pressure dependence if it is below 100MPa.

 

There might be many sources of inaccuracies in pressure predictions: material data, geometrical model, processing conditions. As a wild guess: have you included the dwelling time (the curing time in the barrel of the injection molding machine)? If there is some cure there then the pressure of plastic would increase.

Message 5 of 19
ealizadeh
in reply to: luca.scrosta

Hi,

thank you for the answer.

the filling phase of this process is actually pressure-controlled and the hydraulic pressure settings are equal to an injection pressure of about 110 MPa. but the calculated injection pressure (considering the same injection time for simulation and reality) is 65 MPa.

 

when I run the simulation with full pressure control of filling phase (with hydraulic pressure settings in reality), the filling time calculated by the software is 1.5 sec, while in reality this time is about 9 sec !

 

the barrel temperature setting is 60 celcius degree, so the rubber induction time would not pass too much when it's ready to be injected. by the way, I will try this possibility.

 

another question: what is the best method for measuring visosity? I used RPA 2000 from alpha technology and measured viscosity versus frequency with a fixed strain of 15%. Capillary method is used widely for rhelogical data measurements, but I don't think 50% difference between calculated and real pressure rise from viscosity measurment method.

 

I wait for your response,

Thank you

Message 6 of 19
yuko001
in reply to: abakharev

Hello.

 

I'm trying to optimize D3 parameter of several thermoplastic materials to fit the experimental data. I have a question about your comment: 

Strictly speaking when putting non-zero D3 we have to modify the other coefficients (especially, decrease n and modify D1) to fit the experimental data.

 

Could you tell me how to modify n and D1 parameters? Is there any rule/ equation to calculate them?

 

Thank you in advance,

Message 7 of 19
Alex.Bakharev
in reply to: yuko001

Hi Yuko001,

 

Yes, you are right, if you are using non-zero D3, you have to modify the other Cross-WLF parameters.

 

Autodesk has a free tool named "Autodesk Moldflow Data Fitting" or AMDF for short that simplifies fitting material data including Cross-WLF. It supports non-zero D3. The program is available on Autodesk Subscription Site under the "Product Enhancements". The latest released version is AMDF2018.1 but any version should do the job.

 

Message 8 of 19
yuko001
in reply to: Alex.Bakharev

Thank you Alex,

 

Actually, I have been using AMDF2018 and I know how to input parameters of Cross- WLF model.

The point is the theoretical relation (equations) between the parameters.

 

Let me simplify my question:

When we set 2.0E-7 as D3 for example, how do we calculate the other Cross- WLF parameters?

 

Thank you very much in advance,

 

 

 

Message 9 of 19
Alex.Bakharev
in reply to: yuko001

Hi Yuko001,

 

If you use AMDF then go to Cross-WLF viscosity model tab (if the tab is not there then just fir the data first), enter the desired value of D3, make sure that "Keep constant fixed" check box is checked and press "Fit viscosity button"D3.PNG

 

AMDF should do the work for you.

 

I remember in the earlier versions of AMDF there was a bug with "Keep constant fixed" for D3. Please use AMDF revision 2017.2 or later.

Message 10 of 19
yuko001
in reply to: Alex.Bakharev

Thank you for your quick reply!

 

I'll install revision 2017 and use the fitting tool.

 

 

 

Message 11 of 19
Alex.Bakharev
in reply to: yuko001

Hi yuko001,

 

AMDF2018 should work just fine (have not noticed that you already specified the version of the tool).I was concerned about AMDF 2016 there were problems with handling some GUI check-boxes.

 

Please contact me if any problem with fitting data the way we have discussed,.

 

Message 12 of 19
yuko001
in reply to: Alex.Bakharev

Hi Alex,

 

I tried the way you explained in our discussion, but D3 was reset as 0 when I press "fit viscosity buttom. 

It happened also in AMDF2017 and AMDF 2018 R1.

 

At least I could modify D3 in Data Fitting 2.0. In this case the other parameters should be modified manually.

 

Can you confirm it? If you have any solutions, please let me know.

 

 

Message 13 of 19
Alex.Bakharev
in reply to: yuko001

I have checked with 2017R2 and it works fine if the viscosity is fitted with junction losses D3a.PNG

and resets D3 to zero if no junction losses. You better first fit junction losses without D3 and then enter D3 and refit the viscosity. The behavior might  be because junction losses significantly affect pressures in the capillary and so the pressure effect.

 

Also please check that you have data for L and L/D of capillaries, and there are data for at least two L/Ds.

Message 14 of 19
yuko001
in reply to: Alex.Bakharev

Thank you Alex,

 

Finally I achived to refit viscosity following your advice! Juncture loss is the key...

 

I have 1 concern: when the L/D ratios were identical for all data points in raw viscosity data, The following error message popped out

and I could not finish fitting.

 

"The input data is not valid for fitting juncture loss: L/D ratios cannot be the same for all data points."

 

When the ratios differs like 10, 10.01, 10.02.., it worked.

If we use the same capillary, the L/D ratio must be same for all data points. Why AMDF gives this error message?

Message 15 of 19
yuko001
in reply to: yuko001

My mistake.

L/D should differ for juncture loss calcuation.

 

If I understand correctly, the juncture loss coefficients are not mandatory for 3D simulations.

Then, what should I do if I don't have raw viscosity data with different L/Ds?

 

Thank you very much in advance,

Message 16 of 19
Alex.Bakharev
in reply to: yuko001

Yes, you are right. 

 

Two capillaries are needed to separate Juncture Losses from the Viscosity pressure drop (basically to do Bagley correction).

 

I am afraid you  do need the raw data to fit Cross-WLF with pressure dependency. A pressure dependence makes the pressure drop to grow non-linear (exponential) with the capillary length so the Bagley correction is affected.

 

If you just want a qualitative effect  of pressure dependence on Moldflow predictions then probably the best way is to keep all the other coefficients and just enter D3. Alternatively you can trick the software by entering two similar L/D s (e.g. 9 and 10) and the same apparent viscosity, etc. It should give a very small pressure losses (emerging only from pressure dependence) and it would correct the other coefficients

Message 17 of 19
yuko001
in reply to: Alex.Bakharev

Thank you very much for your advice.

 

I will measure viscosity of the capillary with different length for juncture loss fitting, then fit cross-WLF parameters.

 

I have 1 more question.

How do AMPL tell non-zero D3 materials? Is there any threshold in raw viscosity, or as you mentioned, do they check

the result of maximum pressure in mold flow?

 

Message 18 of 19
Alex.Bakharev
in reply to: yuko001

Hi, yuko001

 

I guess by AMPL you mean the organization: Autodesk Moldflow Plastic Lab. As far as I know they do D3 characterization if they ordered to do so, they have a standard test for this (I think the name is MPL-036). Usually having D3 fitted improves short shot predictions, e.g. in the spiral mold, improves maximal injection pressure predictions, etc. Usually for amorphous materials (e.g polycarbonates) it is more important than for the semi-crystalline materials, etc.

 

Usually D3 is significant at injection pressures above 100MPa but I was told that they can have a noticeable effect below that value.

Message 19 of 19
yuko001
in reply to: Alex.Bakharev

Hello Alex,

 

I have question about cross-WLF viscosity model.

When I add 1 raw viscosity data of different capillary length (upper green point on the attached image) and try fitting

both cross-WLF parameters and juncture loss in AMDF, I got the result as attached.

 

image.png

I expected that fitting curves go higher than raw data points. Why it happens? Do the juncture loss coefficients

and the other cross-WLF parameters affect the result?

 

Can't find what you're looking for? Ask the community or share your knowledge.

Post to forums  

Autodesk Design & Make Report