An Automated Strategy for Gas Turbines Off-Design Predictions with a CFD-based Throughflow Method

TitleAn Automated Strategy for Gas Turbines Off-Design Predictions with a CFD-based Throughflow Method
Publication TypeJournal Article
Year of PublicationSubmitted
AuthorsRicci M, Pacciani R, Marconcini M, Macelloni P, Cecchi S, Bettini C
JournalApplied Thermal Engineering
ISSN Number1359-4311

With the increasing importance of renewable energy sources in the power generation scenario, traditional fossil-fuel power generation systems are subjected to relevant and rapid load variations. Consequently, designers have become more and more interested at predicting the transient behavior of fossil fuel power plants. Tools allowing off-design performance predictions of turbomachines have consequently become desirable even in the first design phases. The paper presents the development of a strategy for gas turbines off-design analyses that exploit a novel CFD-based throughflow method, and its application to a heavy-duty, medium size, F-Class, 4-stage gas turbine designed and manufactured by Ansaldo Energia. The throughflow code is based on the axisymmetric Euler equations with tangential blockage and body forces, and inherits its numerical scheme from a state-of-the-art CFD solver (TRAF code), including real-gas capabilities. The effects of three-dimensional flow features, like secondary and leakage flows, are accounted for in the meridional analysis both in terms of flow distorsions and losses. In order to allow a realistic reproduction of spanwise distributions of flow quantities a radial mixing model is included. Coolant and purge flow injections in the meridional flow-path are modelled by means of distributed source terms and mixing loss contributions. The strategy starts from the calibration of the throughflow method in order to match the results of 3D CFD analyses at design point, whereupon the computational framework is frozen and used for off-design simulations. Such a matching is achieved by prescribing the S2 flow surface, the aerodynamic blockage of the blades and a suitable loss distribution. This is done by employing an automated procedure that extracts such information directly from the circumferentially averaged CFD solution at design point without the need of user inputs for the calibration. The proposed methodology is fairly general and will be discussed in details in the paper. The analysed operating conditions of the turbine encompass a wide range of expansion ratios and corrected rotational speeds. The feasibility of the procedure is assessed by a detailed comparison with 3D CFD results in terms of span-wise distributions and performance figures. It will be shown how the generality and reliability of the proposed method demonstrates its feasibility for an intensive use in the design of gas turbines. In particular, throughflow predictions can compete with the ones provided by state-of-the-art 3D CFD approaches and can be obtained with a small fraction of the computational time.

Refereed DesignationRefereed