I was performing sequential coupled thermo-mechanical analysis using model change for element deactivation & activation in a multi-step analysis. What I found that the thermal history information in current activated elements don't effect other newly activated elements. It becomes even more clear when the .odb data from thermal analysis are carried forward for the mechanical analysis, in that the elements which are only part of initial model and are active in initial step show some stresses & deformation effects. However the elements that are getting activated in later step through the "model change" don't show any stress & deformation corresponding to their relevant thermal history in the later thermal analysis.