Should the spinning reserve procurement in systems with wind power generation be deterministic or probabilistic?

It is a common practice by the system operators to set the spinning reserve (SR) provision based on deterministic approaches. But is this the best operating policy? This paper presents a comparison between the traditional N-1 criterion and two probabilistic approaches to estimate the optimal amount of SR requirements. This comparison shows that methods based on a cost-benefit analysis achieve a better economical balance. An approach based on an exogenous cost-benefit analysis capable of handling systems of realistic size is then presented. This approach is then extended to include demand and wind power generation forecast errors.