Control of pumped storage power plants

Project focus

  • Development of efficient models to describe the flow in long pipelines
  • Development of scalable, physics-based digital twins of pumped storage power plants
  • Design of optimal (model-predictive) control strategies
  • Analysis of the possibilities to estimate non-measureable quantities

Description

Due to the increased implementation of renewable energy production (wind energy, solar plant), the amount of power plants with strongly fluctuating power production has significantly increased in the European electric grid. Thus, suitable storage systems are necessary to guarantee the stability of the electric grid. The most common way to store energy is given by pumped storage power plants, which basically comprise two water tanks which are connect by a pipeline and a turbine, cf. Fig. 1. In cases of excessive electric power in the electric distribution grid, water is pumped from the lower tank to the upper tank. Contrary, power can be supplied to the grid by transporting water from the upper tank via the turbine to the lower tank.

Typical setup of a pumped storage power plant

In this research project in cooperation with Andritz Hydro, control strategies for the optimal dynamical operation of pumped storage power plants are developed. Fixed speed (synchronous machine) and variable speed (doubly-fed induction machine or synchronous machine with a full-rated converter) plants are considered. A typical setup of a pumped storage power plant which utilizes doubly-fed induction machines (DFIM) is depicted in Fig. 2.

Components of a variable speed pumped storage power plant

A main difficulty for the dynamic control of pumped storage power plants are the long pipelines (in the range of up to several kilometers), which connect the turbines with the tanks. For fast changes of the turbine operation, this results in pressure waves (water hammer) which distribute along the pipelines and can result in a severe load (up to a damage) of the pipelines. Fig. 3 depicts a typical distribution of the pressure and volume flow at a pipeline for fast changes in the turbine operation point.

Spatial and time distribution of the pressure and volume flow in a pipeline

To take into account these effects, approximations of the infinite-dimensional equations describing this effect are developed. A focus is laid on methods, which allow for a very high numeric efficiency. This is important in particular for the utilization of the models in model based control strategies. Based on these models, optimal control strategies (model-predictive control) are examined. Based on these models and models of the electrical system, a scalable, physics based digital twin (including sensors and actuators) is developed. The physics based digital twin allows for easy adaptation to different topologies of pumped storage power plants (number of pipelines, types of turbines and generators, different elements in the electrical system, …). The digital twin is used for examination of optimal control strategies (model-predictive control). The main goal is to increase the dynamics of the closed-loop system (and thus the potential for the stabilization of grid variations) while keeping or increasing the energy efficiency of the power plant. Optimal open-loop control of high dynamic transients (e.g. load rejection) is also investigated.

Another focus of the project lies in the optimal power plant dispatching. One aspect is the optimal economic power plant utilization with day-ahead optimizations and control of short-time overloads in order to generate additional income during intraday trading in high volatile power grids. Another aspect is the power plant operation under minimal wear in order to extend the power plant lifespan.

Selected publications

  • J. -F. Mennemann, L. Marko, J. Schmidt, W. Kemmetmüller, and A. Kugi, The spectral element method as an efficient tool for transient simulations of hydraulic systems, Applied Mathematical Modelling, vol. 54, p. 627–647, 2018.
    [BibTex] [Download]
    @Article{Mennemann2018,
    author = {Mennemann, J.-F. and Marko, L. and Schmidt, J. and Kemmetm{\"u}ller, W. and Kugi, A.},
    title = {{The spectral element method as an efficient tool for transient simulations of hydraulic systems}},
    doi = {10.1016/j.apm.2017.10.010},
    issn = {0307-904X},
    pages = {627--647},
    volume = {54},
    journal = {Applied Mathematical Modelling},
    year = {2018},
    }
  • J. Schmidt, W. Kemmetmüller, and A. Kugi, Modeling and static optimization of a variable speed pumped storage power plant, Renewable Energy, vol. 111, p. 38–51, 2017.
    [BibTex] [Download]
    @Article{Schmidt17,
    Title = {Modeling and static optimization of a variable speed pumped storage power plant},
    Author = {Schmidt, J. and Kemmetm\"uller, W. and Kugi, A.},
    Journal = {Renewable Energy},
    Pages = {38--51},
    Volume = {111},
    Year = {2017},
    Doi = {10.1016/j.renene.2017.03.055},
    ISSN = {0960-1481},
    }
  • J. -F. Mennemann, J. Schmidt, W. Kemmetmüller, and A. Kugi, Simulation von Welleneffekten in Pumpspeicherkraftwerken mit Hilfe der Spektral-Element-Methode, at – Automatisierungstechnik, vol. 64, iss. 8, p. 681–695, 2016.
    [BibTex]
    @Article{Mennemann16,
    Title = {{Simulation von Welleneffekten in Pumpspeicherkraftwerken mit Hilfe der Spektral-Element-Methode}},
    Author = {Mennemann, J.-F. and Schmidt, J. and Kemmetm{\"u}ller, W. and Kugi, A.},
    Journal = {at -- Automatisierungstechnik},
    Pages = {681--695},
    Volume = {64},
    Year = {2016},
    Number = {8},
    Doi = {10.1515/auto-2016-0061},
    ISSN = {0178-2312},
    }
  • J. Mennemann, J. Schmidt, W. Kemmetmüller, and A. Kugi, Anwendung der Spectral-Penalty-Methode zur effizienten Diskretisierung der Massen- und Impulsbilanzgleichungen in Rohrleitungssystemen von Pumpspeicherkraftwerken, in Tagungsband GMA-Fachausschuss 1.30 “Modellbildung, Identifikation und Simulation in der Automatisierungstechnik”, Anif/Salzburg, Austria, 2015, p. 56–95.
    [BibTex]
    @InProceedings{Mennemann15,
    author = {Mennemann, Jan-Frederik and Schmidt, J. and Kemmetm\"uller, W. and Kugi, A.},
    title = {{Anwendung der Spectral-Penalty-Methode zur effizienten Diskretisierung der Massen- und Impulsbilanzgleichungen in Rohrleitungssystemen von Pumpspeicherkraftwerken}},
    booktitle = {Tagungsband GMA-Fachausschuss 1.30 "Modellbildung, Identifikation und Simulation in der Automatisierungstechnik"},
    year = {2015},
    month = {9},
    isbn = {978-3-9815012-9-2},
    pages = {56--95},
    address = {Anif/Salzburg, Austria},
    }
  • J. Schmidt, W. Kemmetmüller, and A. Kugi, Modellierung und Analyse eines Pumpspeicherkraftwerks mit drehzahlvariablem Generator, in Tagungsband GMA-Fachausschuss 1.30 “Modellbildung, Identifikation und Simulation in der Automatisierungstechnik”, Anif/Salzburg, Austria, 2014, p. 38–68.
    [BibTex]
    @InProceedings{Schmidt14,
    author = {Schmidt, J. and Kemmetm\"uller, W. and Kugi, A.},
    title = {Modellierung und {A}nalyse eines {P}umpspeicherkraftwerks mit drehzahlvariablem {G}enerator},
    booktitle = {Tagungsband GMA-Fachausschuss 1.30 "Modellbildung, Identifikation und Simulation in der Automatisierungstechnik"},
    year = {2014},
    month = {9},
    pages = {38--68},
    address = {Anif/Salzburg, Austria},
    }
  • L. Marko, J. -F. Mennemann, L. Jadachowski, W. Kemmetmüller, and A. Kugi, Early- and late-lumping observer designs for long hydraulic pipelines: Application to pumped-storage power plants, Int. Journal of Robust and Nonlinear Control, vol. 28, iss. 7, p. 2759–2779, 2018.
    [BibTex]
    @Article{Marko2018,
    author = {Marko, L. and Mennemann, J.-F. and Jadachowski, L. and Kemmetm{\"u}ller, W. and Kugi, A.},
    title = {{Early- and late-lumping observer designs for long hydraulic pipelines: Application to pumped-storage power plants}},
    doi = {10.1002/rnc.4049},
    issn = {1099-1239},
    number = {7},
    pages = {2759--2779},
    volume = {28},
    journal = {Int. Journal of Robust and Nonlinear Control},
    year = {2018},
    }

Project partners and funding