Enhancing Prediction Accuracy in the Evaluation of Power Plant Uprates Utilizing a Novel ‘Big Data’ Approach

Abstract

The evaluation of power plant uprates has traditionally been based on the definition of several ‘typical’ operating modes based on historical data and a — more or less detailed — model of the plant that is compared in current configuration against the same base model including the modifications under consideration. For the economic assessment of the uprate, annual operating hours are allocated to the operating points, and fuel savings and/or additional output predicted by the model due to the modifications are evaluated against the expected investment cost. In this study, the authors demonstrate that this classic approach contains risks in several aspects, in particular: • the representativeness of the ‘typical’ operating modes, • the accuracy of the model, and • the correctness of the assumptions in the allocation of operating hours. Utilizing the example of an actual uprate of a heat recovery steam generator (HRSG) in a large utility plant of an Austrian steel company, a new approach for an evaluation based on ‘big data’ is presented that uses a full year of operational data in hourly granularity for both, the verification of the accuracy of the plant model, and the evaluation of the effect of the uprate. The authors also provide details of the underlying technologies that allow for both, excellent match of operational data with a fully-fledged heat balance software and fast evaluation of tens of thousands of calculation cases.