It appears that there are some problems in the current viscoelastic model (ANSYS v.61-v8.1), that could be confusing to ANSYS users.
(1) For TB,Prony input, the data can be entered at different temperatures. I believe that the Prony constants are a function of reference temperature instead of temperature. Because, under each reference, the Prony pairs are obtained based on the master curve which is generated by time-temperature superposition based on modulus-frequency curves at different temperature. Therefore, Prony pairs has already considered different temperature. Why does the design of the Prony Table include different temperature again? I believe that will certainly confuse users.
(2) For the Shift function, there should be only one group C1, C2 and Tref input. However, if you use a different Tref, the C1, C2 should be different and should generate a shift factor A(T) that is quite different. What is the significance of a different Tref?
Your statements are certainly true IF the viscoelastic material is thermorheologically simple (TRS). But TRS is not a requirement for the TB,PRONY version of ANSYS viscoelasticity. It is possible to have a viscoelastic material that has temperature dependent Prony constants that are notthermorheologically simple. For that case, the ANSYS program will perform material property interpolation based upon the available information, just as it would for any other linear or nonlinear property. We believe that this adds flexibility to the material models and removes a limitation on the use of the model that is not warranted. I should point out that the viscoelastic material model input with the TB,VISCO and used with elements VISCO88/89, does not have the temperature dependent coefficient capability, and that model may only be used as a TRS model.
The same sort of reasoning goes with the shift function. According to the TRS assumption, there exists one (and only one) reference temperature, and that is the temperature of the "master curve". One would normally assume that that is the form of thermorhelogocially simple viscoelasticity being used if a shift function is involved. However, extensions of the theory to include shift function coefficients that are temperature dependent is also possible, and that is the basis for the form of the ANSYS input. I personally am not aware of any real material that has temperature dependent shift coefficients. However, that doesn't mean that they don't exist now or will not exist in the future.
If your materials are all thermorheologically simple with a single shift temperature, the TB,PRONY is the material model that you should be using. The greater flexibility of the model is not a limitation to you.