Accuracy of the Edgeworth Approximation for Lolp Calculations in Small Power Systems

Because they speed numerical calculations and exhibit functional dependences, Edgeworth-type series are increasingly used to approximate and calculate LOLP's. They are usually sufficiently accurate for large power systems (> 15 000 MW), but can be very inaccurate for small systems or those with low forced outage rates. This is because the approximating Edgeworth-type series are appropriate only for continuous probability densities, while discrete lattice-type density functions describe a typical power systems's probability properties. This paper investigates these inaccuracies for small systems (< 5000 MW), working out specific examples in a numerical approach and examining underlying functional dependences in an analytic approach.