Experimental evaluation of ability of Relap5, Drako, Flowmaster2™ and program using unsteady wall friction model to calculate water hammer loadings on pipelines

Mechanical loadings on pipe systems caused by water hammer (hydraulic transients) belong to the most important and most difficult to calculate design loadings in nuclear power plants. The most common procedure in Sweden is to calculate the water hammer loadings on pipe segments, according to the classical one-dimensional (1D) theory of liquid transient flow in a pipeline, and then transfer the results to strength analyses of pipeline structure. This procedure assumes that there is quasi-steady respond of the pipeline structure to pressure surges—no dynamic interaction between the fluid and the pipeline construction. The hydraulic loadings are calculated with 1D so-called “network” programs. Commonly used in Sweden are Relap5, Drako and Flowmaster2—all using quasi-steady wall friction model. As a third party accredited inspection body Inspecta Nuclear AB reviews calculations of water hammer loadings. The presented work shall be seen as an attempt to illustrate ability of Relap5, Flowmaster2 and Drako programs to calculate the water hammer loadings. A special attention was paid to using of Relap5 for calculation of water hammer pressure surges and forces (including some aspects of influence of Courant number on the calculation results) and also the importance of considering the dynamic (or unsteady) friction models. The calculations are compared with experimental results. The experiments have been conducted at a test rig designed and constructed at the Szewalski Institute of Fluid Flow Machinery of the Polish Academy of Sciences (IMP PAN) in Gdansk, Poland. The analyses show quite small differences between pressures and forces calculated with Relap5, Flowmaster2 and Drako (the differences regard mainly damping of pressure waves). The comparison of calculated and measured pressures and also a force acting on a pre-defined pipe segment shows significant differences. It is shown that the differences can be reduced by using unsteady friction models in calculations. Recently, such models have been subjects of works of several researches in the world.