CFD simulation of a safety relief valve for improvement of a one-dimensional valve model in RELAP5

Detta är en Master-uppsats från Chalmers tekniska högskola/Institutionen för teknisk fysik

Sammanfattning: In the Swedish nuclear power plants a structural verification of the pipe systems is a necessity to ensure that the pipes are strong enough to withstand the forces which can result from a sudden event. One example of a component which generates forces in the systems while operating is the safety relief valve. Safety relief valves are used in order to prevent overpressure in a process system by releasing a volume of fluid from the process when a predetermined maximum pressure is reached.In order to analyze the forces from water and steam in the pipe systems the software RELAP5, which performs calculations in one dimension, is commonly used within nuclear engineering. The valve model which is currently used when simulating a safety relief valve in RELAP5 is the motor valve model. However, the usage of this model with present settings results in forces higher than in reality in the pipe systems.The purpose of this project was to investigate how a safety relief valve can be modeled with CFD and to find interesting parameter relations to be implemented in RELAP5 in order to obtain more realistic results of generated forces in the pipe systems. The aim was to modify the currently used motor valve model and to develop a servo valve model which is a more flexible model to use in RELAP5. The purpose of this project was also to investigate if a CFD simulation in 2D of the valve gives similar results as a 3D simulation.The investigated valve in this project was a proportional valve. It starts to open at a set pressure of 31 bar(g) and is completely opened at 10 % overpressure, i.e. 34.1 bar(g), where the maximum lift of 8.5 mm is reached. The movement of the spindle is determined by the different forces acting on it. In this project the hydraulic forces, the spring force and the gravity force were considered.The CFD simulations were performed in ANSYS FLUENT v.13. Dynamic layering was used in order to change the mesh during the opening process of the valve. The 2D and 3D geometries were created and meshed in ANSA v.13.2.1. Axisymmetry was used as a boundary condition in the 2D model, and in the 3D model mirror symmetry was used. The used turbulence model was SST k-omega. A sensitivity analysis was performed in order to investigate if and to which extent different mesh densities, turbulence models and time step sizes influence the results of the CFD simulations.A verification of the 3D geometry and force calculations was performed, with the conclusion that they seem to be consistent with reality. The transient 2D and 3D simulations were conducted with both an instant and a gradual increase of inlet pressure. Differences could be observed between the 2D and 3D simulations but similarities were also evident. The simulations performed with a gradual increase of inlet pressure were verified with experimental data. Interesting relations were found such as that the total hydraulic force acting on the spindle is a function of different pressures in the valve and the mass flow through the valve.In the currently used motor valve model in RELAP5 an opening time of 1 ms, an instant increase of inlet pressure and the abrupt area change model are used. This model was modified by using an opening time of 41 ms which was a result from the 3D CFD simulation. This modification resulted in lower forces generated in the pipe right after the valve. The generated forces also reached more realistic magnitudes than the forces generated from the currently used model.A servo valve model was developed in RELAP5 by specifying all necessary relations, needed for the valve to function, in control variables. One relation from the CFD simulations, describing the total hydraulic force acting on the spindle, was implemented successfully. The usage of the abrupt area change model in combination with short pipes resulted in a stable system and realistic forces. The trends in the opening process were fairly consistent with reality when the inlet pressure was gradually increased.Both the motor and the servo valve model were also modified by using the smooth area change model including the implementation of a Cv table. This modification did not decrease the magnitude of the forces and instabilities were observed in the system.The opening process of the valve, simulated both with CFD and in RELAP5, is faster than the opening process observed in experimental data. This concludes that the models are conservative, which is a requirement within the nuclear industry.