The rate-equation for viscous flow in BB model

Hello everyone,

  I recently used the particle swarm method to calibrate the BB model in the incompressible case, and then I will consider the compressible case. If the material is compressible, the stretch part of the rate-equation sometimes seems to be negative (I haven't rigorously proven it). If there is indeed this problem, how can I solve it?

Thanks, by advance,

Carl