Embodiments of the present invention provide a method for performing lumped-process model calibration. The method includes creating a plurality of sub-process models for a set of sub-processes; creating a lumped-process-model incorporating said set of sub-processes; calculating a first set of output patterns from a set of test patterns by using said plurality of sub-process models; calculating a second set of output patterns from said set of test patterns by using said lumped-process-model; and adjusting process parameters used in said lumped-process-model to calculate said second set of output patterns to match said first set of output patterns. A computer system for performing the lumped-process model calibration is also provided.