A linear‑algebraic procedure that eliminates a subset of internal network nodes from an admittance or impedance matrix by performing a Schur complement on the matrix block associated with eliminated nodes, producing a reduced‑order matrix that preserves the exact terminal (retained‑node) linear steady‑state electrical behavior under the assumptions of linearity and nonsingularity of the eliminated‑node submatrix.

Principle

Given partitioned admittance Y = [[Y_aa, Y_ab],[Y_ba, Y_bb]] with Y_bb invertible, the reduced admittance at retained nodes is Y_red = Y_aa − Y_ab Y_bb^{-1} Y_ba; the reduction preserves nodal voltages/currents at retained nodes for the linear steady‑state model.

Demonstration

Illustrative scenario → A detailed distribution feeder model contains many intermediate buses (b). Recognition → partition the bus admittance matrix into retained transmission buses (a) and distribution interior buses (b). Action → compute the Schur complement to eliminate b and obtain Y_red for system‑level studies. Consequence → system‑level analyses run faster using Y_red while terminal steady‑state responses remain identical under the linear model assumptions.

Misapplication

Applying Kron reduction when the Y_bb block is singular (e.g., islands or ideal open circuits) or eliminating nodes that contain dynamic or nonlinear devices whose internal states affect terminal behaviour; the error is assuming terminal equivalence holds outside the linear steady‑state, frequency‑independent assumptions.

Consequence

Kron reduction yields compact equivalents that reduce computational burden and enable interfacing of subsystem models, but it can hide internal modal dynamics and make the reduced matrix denser (more filled), affecting numerical performance and dynamic fidelity.

Reversal

For frequency‑dependent network elements (skin effect, frequency‑dependent lines) or when dynamic stability and internal modes matter, single‑frequency Kron reduction does not preserve frequency response or dynamics; model‑order reduction methods that retain internal states or frequency dependence are required.

Boundary

Clearly within: linear, time‑invariant phasor‑domain network representations where eliminated‑node submatrix is nonsingular. Boundary case: linearized models of nonlinear devices where small‑signal assumptions hold only locally. Clearly outside: fully nonlinear time‑domain models, frequency‑dependent multi‑port elements without consistent linearization, or cases with singular Y_bb.

Semantic Tension

Reduction (simplicity, computational efficiency) ↔ Fidelity (preservation of internal dynamics and sparsity): Kron reduction simplifies topology but can trade off sparsity and dynamic information.

Synthesis

Kron Reduction is an exact algebraic elimination in the linear steady‑state sense that produces terminally equivalent reduced models only under its invertibility and linearity assumptions; use when terminal equivalence suffices and verify limits when dynamics or frequency dependence are important.