Validation of recently proposed methodology of electromigration (EM) assessment in power networks was achieved by fitting to measurements performed on dedicated test-structures when non-uniform temperature generated by specially designed on-chip heaters and interconnect Joule heating was accounted. Analysis of the predicted voiding locations confirms the essential impact of temperature non-uniformity on EM-induced IR drop degradation. Tuning the model parameters allowed to achieve a good fit between model predictions and measurements for different test-grid structures.