Feasibility of transient model tests in a closed-loop test rig with the example of a reversible pump turbine

Abstract

The growing share of renewable energies in power generation increases the risk of unintended exceedings of frequency limits in an interconnected grid. To counteract this, a so-called redispatch, a correction of the power balance in the grid, takes place. Due to their flexible operating capability, pumped storage power plants often perform this correction. However, the frequent change of the operating point of these power plants leads to an increase in transient events and thus raises the mechanical stress on the plant components. In addition to the known methods for analyzing fluid machinery, transient model tests are an important element for investigating the increasing demands on mechanical stress capacity and operating behavior. In this work, a method to implement transient model test in a closed-loop test rig is presented. For this propose, various transient load cases of a real pumped storage power plant with a reversible pump turbine are simulated. The time-dependent quantities of the machine are transferred to the scaled model using suitable laws of similarity. A special feature is the use of the Strouhal number to convert the time scale. These transferred data serve as setpoints for the model machine when simulating the test rig. An iterative optimization procedure is used to adapt the behavior of the test rig actuators until the setpoints are reached with the specified accuracy. With the knowledge about the entire control sequence, the transient experiment is carried out on the laboratory test rig. The use of a bypass increases the spectrum of transient load cases under investigation. This enables, for example, a fast change between pump and turbine operation of the model machine. In order to examine the developed method for its capability, various influencing factors were considered. By means of a parameter study, the speed of sound of the numerical test rig model was varied. Especially for load cases with modified guide vane opening, different values of the speed of sound have an influence on the simulation results. Therefore, the speed of sound must be modeled as close as possible to the same values as in the real test rig. Furthermore, it is shown that the readout procedure of the characteristic diagram of the hydraulic machines can influence the simulation results. Responsible is the linear interpolation between the grid points used by the simulation software during the transient process, which leads to discontinuities. The general applicability of the method was investigated with the help of various transient load cases, such as operating point changes and fast transitions. Due to the lack of automated guide vane adjustment on the test rig used, only the fast transitions were carried out experimentally. For this purpose, the model rotational speed was adjusted between pump and turbine operating point at constant guide vane opening. The findings were then transferred to the simulated load cases with active guide vane adjustment. The applied methodology is confirmed by the good agreement between setpoints and measurement results. Limitations of the transient model test are given by the electrical machines. To reduce the differences in dynamic similarity between model and prototype, high rotational speeds of the model are required. For load cases with ambitious speed changes, such as fast transitions or total load rejections, the drives reach their acceleration restrictions. Another challenge is the measurement of the transient discharge, which influences the results especially at high discharge gradients. It is revealed that the used indirect measurement of the model discharge in the pump branch in conjunction with the bypass branch using two sensors with too high delay times is insufficient. These challenges need to be addressed in future research.