Modelling of metal forming process

I am working on the Finite Element Analysis of a metal forming process. I have created the rollers and the mandrel and the roller as analytically rigid surface and the blank as 3D deformable solid. 

My issues are the follows :

1.I have specified the contact between the roller and the blank as surface to surface contact and that between the blank and the mandrel as surface to surface contact. The result turns out that the internal energy generated (ALLIE) is greater than the work done(ALLWK). Also, the total energy (ETOTAL) is not a constant- it is monotonously increasing.

2. I changed my contact definitions from surface to surface contact to general contact (All with Self) The total energy is found to decrease and the internal energy becomes less than the work done.

I have specified 10 as the frequency of ALE and 15 as the number of re-meshing sweeps per increment. I understand that the internal energy is 95% dependent on the material property and the rest 5% on the Artificial energies. I changed my material model from Isotropic hardening to Johnson Cook. The results show the same trend. I consulted a friend of mine who suggested that the increase in energy could be due to excessive mesh distortion and so I should try increasing the frequency of ALE.

I would like to have your comments on the following :

1. The change in trend of the total energy with change in contact definition.
2. Possible reasons for variation of Total energy
3. Generation of Viscous dissipation energy, though i have not specified any values for viscous dissipation.

I would be grateful if you could throw some light on these.