AbstractIn power system load flow studies, it is common to ignore wind park (WP) losses completely, since WPs are obliged to provide reactive power as a free‐of‐charge ancillary service. Strongly simplified WP loss estimations are also widespread. Prevailing loss modelling approaches may result in substantial and so far usually ignored reactive power allocation inefficiencies. The innovation of this paper lies in demonstrating this and in presenting an alternative modelling approach that combines speed, ease‐of‐use and accuracy.Calculations are based on a generic WP connected to 110 kV. It is simulated once using gearless permanent magnet synchronous generators and once using doubly fed induction generators. In this way, for the first time, the consequences of neglecting differences between wind generator types when calculating reactive power‐related losses are systematically quantified. Loss‐minimal dispatch is assessed using heuristic optimal power flow. Relevance of findings is demonstrated for the application of reducing system losses based on reactive power control of WPs in closed‐loop optimal power flow schemes. It is shown that non‐consideration of WP losses may lead to dispatch decisions where the increase in WP losses outweighs the achieved loss reduction in the 110 kV system by 400%. The study facilitates comparing reactive power ancillary services from WPs to that of other technologies from an economic perspective. It concludes by a benchmarking overview of pros and cons of modelling alternatives. Copyright © 2017 John Wiley & Sons, Ltd.