Development of Recommended Practice and Toolkit for accurate load prediction on high head Francis turbines

The HiFrancis project Background The average lifetime of a Norwegian hydro power plant is 45 years, and many show signs of fatigue with the need of maintenance. Some power plants in Norway have experienced failures after installation of new Francis runners. The main problem is formation of cracks in the turbine runner due to cyclic loads. Being able to correctly predict these loads and the dynamic response of the turbine is acknowledged by the industry as a prerequisite to achieve reliable operations in the future. The energy market is changing, with larger contributions from sun and wind power, and to solve the problems of today is essential for maximizing of the possibilities of tomorrow. In this research project, the goal has been to secure safe and reliable operation, as well as reducing the risk for failures due to resonance problems. This has been done by developing methods for fast and accurate numerical simulations of rotating machinery. High head Francis turbines has been the main focus, specifically the Francis-99 model at NTNU, but the methods have been developed for generic use in all industries. Method Simulations of the water (CFD) and the structure in the turbine runner (FEA) have been done. At the same time, methods for analysis of the complex fluid-structure interaction (FSI) have been developed. The results from the simulations have been validated with experimental results from the Waterpower Laboratory at NTNU. In addition to accurate results, there have been a focus on reducing the simulation time, without compromising on the accuracy. It has therefore been developed several software components in out Digital Lab delivery to facilitate faster simulations than at the start of the project. Results In addition to the software deliveries, the project have produced several written guides, as references for the industry in future development and procurement of high head Francis turbines; - Buyers Guide: A guide to be used in the procurement of turbines. The goal is that all parties, the buyer, consultants and seller, all have a better basis of knowledge regarding potential problems. This will lead to better discussions and lower risks for all involved. - Recommended Practice CFD/ FEA: A technical guide that describes numerical simulation of high head Francis turbines. In addition, the main technical results of the project are as follows; - CFD analyzes of the turbine match very well with experimental data. Especially for off-resonance operation, the CFD analyzes are accurate. Studies also show that under resonant conditions, CFD can only find part of the fluid pressure. In order to find the last (acoustic) part of the pressure one has to do an FSI analysis which involves the structure of the impeller. - A procedure has been proposed to evaluate the probability of resonance, as well as the structural response at resonance. - A Model Order Reduction method based on Krylov vectors has been developed. This method can give time savings of 20 times or more, without effect on accuracy. The method is mainly used in structural simulations. - A Model Order Reduction method based on modal decomposition has been developed. This method offers great time savings over a regular two-way simulation. In addition, the method manages to find phenomena that cannot be simulated with a one-way link. To evaluate the value of this research, the consulting firm Impello has done an independent value creation analysis. They estimated that value creation, in the form of reduced risk of accidents, is in the order of NOK 500 million in Norway, as well as several billion worldwide. The entire hydropower industry, from manufacturers and consultants to suppliers, has been included, and this has led to a significant increase in knowledge. In addition, the knowledge gained can be in against other industries, leading to several potential synergies.

På vegne av NFR undersøkte Impello effektene av forskningen på miljøvennlig energi. Resultatene ble presentert i rapporten «Effekter av energiforskningen». Rapporten konkluderte med redusert sannsynligheten for resonans og påfølgende havari i nye HF-turbiner fra 10 % til 5 %. Verdiskapingspotensialet ble estimert til 500 mill. NOK i Norge og 1.5 mrd NOK globalt for utskiftning av eksisterende HF-turbiner. For nye HF-turbiner kan reduserte havari-kostnader utgjøre mer enn 5 mrd. kr globalt. Alle de store turbinleverandørene er med på prosjektet og kan få nytte av kompetanseutviklingen, i tillegg lages det en rekke dokumenter som skal støtte opp under både produsenter, konsulenter og leverandører under innkjøp og design av nye aggregat. Statkraft har estimert en besparelse på 80 millioner kroner, som følge av redusert risiko for havari i to nyinnkjøpte turbiner. For EDRMedeso har vi sett en markant økning i aktivitet mot vannkraftsbransjen, samt forsking og utviklingsprosjekter

The underlying idea for value creation is to develop a solution for accurate and fast numerical simulations of high head Francis turbines. The project will deliver two ready-to-use innovations: 1. A Recommended Practice for numerical simulations of high head Francis turbines. The document will describe modeling techniques in detail as well as valid assumptions, verified to give reliable results. The ambition is that this document will be the industry standard in future simulations on high head Francis turbines. 2. A software Toolkit supporting the developed workflow. Existing mathematical models that reduce computational time for the numerical simulations will be implemented. To secure reduced simulation time with respect to human interaction, the Toolkit will be packaged in a user-friendly environment. The average age of a Norwegian hydro power plant is 45 years, and many will be maintained or refurbished the coming years. Some newly refurbished high head power plants have experienced failures after installing new and modern Francis turbines, where the turbine develops cracks in the blades. Using numerical simulations, the cyclic fluid loads and the behavior of the turbine in operation can be predicted. When performed early in the design process, improved turbine designs can be achieved. This is highlighted by the industry as a prerequisite to secure reliable operation in the future. The Recommended Practice along with the Toolkit will be a vital and valuable innovation to meet the future requirements safely. The central R&D challenges addressed in this project is: 1. Accurate prediction of fluid flow through the turbine 2. Accurate prediction of natural frequencies of the turbine 3. A simulation workflow that handles the interaction between the fluid flow and the turbine 4. Proving the validity for the simulation workflow 5. Reduction of the computational time for the simulation workflow 6. Improvement of the human interaction for the simulation workflow