Mohammad Ali
Converter transformers are the largest single source of controllable loss in a line-commutated ultra-high-voltage direct-current (UHVDC) converter station, yet they are usually specified with loss-capitalization rules developed for sinusoidally loaded power transformers. This paper develops and applies a three-layer framework that minimizes the transmission losses and the overall operating cost of 1100 kV converter transformers. Layer 1 is an analytical electromagnetic–thermal–economic design model of a single-phase, two-wound-limb unit whose total owning cost (TOC) is minimized by differential evolution and cross-checked by particle swarm optimization over seven variables, including the strand thickness of continuously transposed conductors, the core-steel grade and magnetic tank shunts, subject to impedance, over-excitation, hot-spot, transport-height and 600 t transport-mass limits. Layer 2 is a harmonic-aware loss model that computes the valve-winding spectrum from the commutation overlap at every load and corrects the IEC h^2 eddy-loss law for skin effect. Layer 3 coordinates the firing-angle and on-load tap-changer set-point and stages the coolers for minimum loss over an 8760-h load and ambient profile. Applied to a ±1100 kV, 12 GW bipole with 611.8 MVA units, the TOC-optimal design cuts rated losses from 1288 kW (conventional design) and 1725 kW (minimum-first-cost design) to 1077 kW, raises full-load efficiency to 99.79 % and lowers the TOC by 2.75 % and 9.54 %, respectively. At station level, the three layers together reduce annual converter-transformer energy losses from 152.0 GWh to 122.9 GWh (−19.2 %), saving about USD 1.75 million and 16.6 kt of COâ?? per year. The study also shows that rated-point loss factors underestimate the annual load-loss energy by 4.4–7.1 % and that tap tracking lowers the no-load energy by 11.7 %.