Secure Stochastic Unit Commitment in Low Rotational Inertia Power Systems With The Inclusion of Frequency Stability Constraints

This thesis work studies the Unit Commitment (UC) problem from a low-rotational inertia power system perspective. The novel framework used to approximate the non-linear term of the Frequency Nadir into a linear form is the separable programming technique. It has been manifested that using auxiliary variables that guarantee an special order set 2 (SOS2) condition, we can approximate into linear such non-linear term into the UC problem. Furthermore, we have included a fast-screening contingency ranking algorithm that takes into account the reliability data of the power system. This modelling is known as the Reliability Performance Index (RPI). This index can serve in addition to the implementation of Optimal Power Flow (OPF) studies of simple load flow studies using the scheduling results of the UC. The strength of this index is that it encapsulates the probabilistic behaviour of a contingency, and the severity of it at the same time, yielding a different contingency ranking when it is compared with traditional contingency ranking approaches. The uncertainty that renewable energy sources (RES) bring into the power system operations planning analysis, specifically the UC problem, is addresed via a technique known as stratified sampling. With this approach, we generate the scenarios to include into a stochastic framework of UC. Moreover, we modelled the traditional deterministic UC into a stochastic framework, specifically following a two-stage stochastic unit commitment approach (TSSUC). Under this practice, we ensure that the frequency stability constraints were enforced into a stochastic framework as well. The results indicate that it is possible to use a stochastic modelling of the frequency stability constraints using separable programming for the first time, without the need to perform pre-processing activities to include them into the UC modelling. The concept of value of information is included in this stochastic UC analysis. The results indicate the cost of the available information when making a commitment decision regarding the generation fleet in the power system. It was seen as well that it is possible to include a virtual inertia response in the scheduling, and this helps to alleviate a potential loss of largest in-feed in the system.