from tensile creep experiments I have test data for different nominal stresses. I want to use this data to calibrate a generalized Maxwell model of the form
E(t) = E_inf + sum_i (E_i * exp(-t / tau_i)
I came across a procedure with the following steps:
1. Calculate creep modulus E(t) = sig_nom / eps (t), with sig_nom as the nominal stress and eps (t) as the time-dependent creep strain
2. Find E_inf, hence E(t=t_inf) with t_inf as the last point in time
3. Calculate the first Prony parameter N=1 (E_1, tau_1) --> E(t) = R_1(t) + E_1*exp(-t/tau_1), using logarithmic transformation LN ( E(t) ) ~ LN (E_1) - (1/tau_1)*t with linear regression
4. Follow same steps as in 3. using R_1(t) = E(t) - E1*exp(-t/tau_1) --> R_1(t) ~ E_2*exp(-t/tau_2)
Now my question is how should I preferably determine E_inf (long-time creep modulus)? If it is exactly E(t=t_inf) the slope of LN ( E(t) ) does not remain constant for t--> inf. However this is needed to perform the linear regression.
Does anyone have some hints or different approaches to determine E(t)?
Thanks in advance!