Hierarchical h-adaptive computation in VLSI interconnect capacitance extraction

A hierarchical h-adaptive method based on BEM is proposed for VLSI parasitic capacitance extraction. In h-adaptive methods, high computational accuracy is guaranteed by auto refinement of the boundary elements. A new auto refinement method is introduced by property storage of the boundary elements. It is more effective than the old methods in the stability of adaptive computation. A new error estimator based on superconvergent recovery technology is presented, to determine which of the elements should be refined. Numerical results show that adaptive computation using the new auto element refinement and error estimator is very effective both in accuracy and speed.