Cross-WLF model coefficients

[Anonymous]

Dear All,

I wonder whether there are any specific numerical boundaries for the Cross-WLF model coefficients that can be imported into MoldFlow database.

I have measured and fitted the viscosity data of SABIC 56M10 (PP) and BASF Terluran GP22 (ABS) grades. For data fitting I used Excel`s solver to minimise the RMS between the fitted and measured data (220-240-260 C).

I utilised a rotational rheometer (shear 0.01 - 0.1 - 1 1/s) and capillary rheometer (50-10000 1/s) to obtain the viscosities and applied Bagley and Rabinowitsch corrections.

Eventually I got completely different coefficients when comparing with the ones that are in the database by default. I also downloaded the fitted viscosity of the BASF GP22 from MoldFlow and compared with my fitted data, which viscosity values looked very similar, but as mentioned, the coefficients were different.

For instance I noticed that A2 (is always?) 51.6 K in the database but if I fit my own data and constrain A2 in the solver then I got a little bit worse fitness, i.e. higher RMS.

Is that a good approach to minimise RMS and NOT constraining any of the coefficients?

For example, testing an unknown grade only for a filling analysis with the Cross-WLF model, are there specific rules that one should apply when fitting the measured data to the Cross-WLF model?

The second issue:

Can the crystallisation temperature Tc be used as T* for crystalline grades, similarly as Tg is used for amorphous ones?

Is this T* used to determine the freeze point or should that be inputted separately in the "T trans" cell "Rheological Properties" tab?

It`s a bit confusing as for BASF GP22 Ttrans is 103 C but, D2 in CrossWLF is 343.15 K = 70 C.

Thank you very much for any clarification, comments!

[maxmarautodesk]

Hi Martin,

I have no clue about your second issue, but perhaps I can help you on the first one:

Instead of fitting the data yourself, you can get a data fitting software tool from Autodesk free of cost. You can use your measured viscousity data as an input. The software is able to calculate all necessary values for the Cross-WLF viscosity model, if you provide a sufficient data set. If I remember correctly, it should be viscosity vs shear rate for at least three different temperatures.

The software from Autodesk ist actually made for creating new data base entries for Moldflow based on material data that you collected yourself. If you provide also information about thermal and mechanical properties, you can get a data base file (.udm). It will contain the fitted data sets and can be directly used in Moldflow, when you select the material for your simulation.

I cannot directly answer your question, if your mathematical approach is correct or sufficient. But I believe using the software of Autodesk will save you some work. I cannot look into the software, but I think the calculations to find the model coefficients will probably follow the same and standardized way of Autodesk. Hence, the outcome will likely be more correct and reliable than juggling with numbers in Excel.

I find the software tool quite convenient (given having enough/all data available to feed it). You get out the ready-to-use data base file which can furthermore be shared with Autodesk (if you like), so that in the future the material will appear in the Moldflow data base and be available for the whole community. The software is called "MPL DataFitting". Please contact the Autodesk support for getting it, since I am not sure, if I am allowed to forward it to you.

I hope this helps, regards,
Max

[Anonymous]

Hi Max,

thank you very much for your detailed answer and sorry for getting back to you late.

I have contacted the datafitting guys at Autodesk, I hope they can help me to work out the best solution.

thanks again,

Kind regards,

Martin

[Anonymous]

Martin, did you try the data fitting software from Autodesk and how did it work?  I recently bought a trial edition of IRIS rheology software and I'd like to compare it with this 'free' version discussed in this chain.  Thanks!

[Anonymous]

Hi Patrick,
thanks for your query. I have the tool but haven`t used it.
As far as I remember entered data is actually uploaded to/and fitted by an Autodesk server. After fitting, it  may be saved into MoldFlow database. Please confirm the conditions with Autodesk.

Since my data was kind of proprietary, I skipped this option and have used Excel`s Solver.

For me the key thing is that Excel gives more flexibility and direct insight when looking at the effect of change of any model coefficient.
My approach was to fix D2 (Tg of material) and A2=51.6K and get the rest optimised based on minimising the squared difference between the measured and fitted data.
So far I have had reasonably good fitting, confirmed by validation and direct comparison of MoldFlow viscosity coefficients with own data.

I haven`t heard about IRIS that must be a special and powerful one.

Good luck with your studies,

kind regards
Martin

[maxmarautodesk]

Dear Patrick, dear Martin,

I have an update for you. My last post was mainly a rephrasing of the information I got from Autodesk together with the software. The software tool is indeed useful for creating new data base files for Moldflow. However, I found that the fitting performance of the software regarding the pVT data is really poor. So far, my trials gave wrong results. Why wrong and why am I so sure? Because the fit for the pVT data looks like a straight line, it does not match the input data points at all, does not catch the shape of the actual curve described by the data points, and the coefficients don't make sense at all. The transition temperature of the materials is usually known, but the respective coefficient given by the software after fitting does not match it at all. Moreover, some of the coefficients the fitting software yields are negative (I have seen very small numbers, but no negative values).

I would be happy to get some feedback on how this IRIS software performs. Perhaps it turns out to be a good alternative. I also tried Origin for fitting the data, but basically the same like in the Autodesk tool happened. It yielded wrong coefficients and fitting plots that did not match the data points whatsoever. I could accomplish very good fits by means of Sigmasoft (competitor to Moldflow) which does not have any problems with fitting the data. The simulation software comes already with the ability to fit measured material data. It seems to work very well, but I don't know why other software messes up the fitting so tremendously.

Best regards,
Max