There are a number of ways of predicting low-cycle fatigue life. I infer from prex's suggestion that the plastic strains are accumulated in the model. For very low cycle fatigue or rather ratchetching (ie in tens to hundreds of cycles) this is possible. However, a failure criteria is still needed. However, low-cycle fatigue may go up-to 10^5 cycles and modelling each and every one of these cycles is not practical. In such cases the model should be taken to a state where a stable hysteristis loop is formed, this will give the strain range. You still need a failure criteria though. There are three generally accepted ways of predicting life from this strain range.
1) Strain Range
Relate the strain range to a strain-life curve. You may have to generate your own curves experimentally.
2) Equivalent Elastic Stress Approach
Assume the strains are elastic and find the equivalent stress range from the product of the strain and the modulus. Use standard SN curves and extrapolate back to the low-cycle fatigue regime. I once carried out a research project on the low-cycle fatigue of welded high strength steels using this method and it worked well. However still use with care!
3) Local Stress-Strain Approach (Neuber)
This is a little more involved, see any text on LCF for further details.
TERRY