Error Analysis of Multi Level Graded Precision Model for Stability Monitoring on the Large Power System

To keep computational burden in stability analysis of the large power system at manageable levels the order of the problem must be reduced. An effective way to accomplish this starts with identifying dynamic clusters [1] within the system than using these to identify weak boundaries where system separation is probable [17] in a particular emergency. Reduction of the order is then accomplished by a graded precision model [18], [19] which retains the backbone (EHV system, tie lines) at full nonlinear representation, linearizes and aggregates the cluster interiors and treats the interface at an intermediate level. This representation is utilized in this paper to further reduce computational load by approximating the effect of the fast, intracluster, oscillations by applying a new averaging method, as opposed to customary singular perturbation. A detailed error analysis of this technique is presented. Excellent approximation is demonstrated in defining the stability boundary through the error analysis and also by simulation.