Guidelines for performance enhancement

Warning

Issue 79

Independent / cache topology

Parallel Computing

Matrix prefactorization

Every iteration of power-flow or state estimation has a step of solving large number of sparse linear equations i.e. AX=b in matrix form. Computation wise this is a very expensive step. One major component of this step is factorization of the A matrix. In certain calculation methods, this A matrix and its factorization remains unchanged over iterations and batches (only specific cases). This makes it possible to reuse the factorization, skip this step and improve performance.

Note: Prefactorization over batches is possible when switching status or specified power values of load/generation or source reference voltage is modified. It is not possible when topology or grid parameters are modified, i.e. in switching of branches, shunt, sources or change in transformer tap positions.