Infeasibility in automated test assembly models: a comparison of different methods