An Assessment of Margin in Design Steam Hammer Forces for Combined Cycle Power Plants

All combined cycle steam plants have rapid-closing stop valves in steam lines to protect the turbine. The rapid valve closure produces a steam hammer in the piping resulting in large forces for which the piping system and supporting structures need to be designed. These forces are typically calculated using the classical Method Of Characteristics (MOC) solution. An evaluation has been conducted which compares the forces computed using the classical methods with a best-estimate approach. This comparison has been done to define margin, and to benchmark and identify potential refinements in the techniques used for evaluating steam hammer loads. The best-estimate approach involves the use of the RELAP5 computer program. RELAP5 is used extensively in the Nuclear Industry to evaluate fast thermal hydraulic transients. It has the capability to analyze subcooled liquid, two-phase and saturated or superheated steam piping system. The models used in RELAP5 are best estimate results in comparison to the MOC solution which are mathematically derived from theory. The compressible flow program GAFT is used to obtain the MOC solution. The main steam line of a single Heat Recovery Steam Generator combined cycle plant is modeled with both the GAFT program and with a PC version of RELAP5. Identical piping lengths, mass flow rates, pressures are used in each model. Also, a stop valve closure time of 100 milliseconds is modeled. As RELAP5 output results are pressure, flow rate, velocity, and density, the resultant forces are generated using the R5FORCE program, a post-processor to compute associated transient forces on straight piping links. The GAFT program, which is specifically designed to compute steam hammer forces, computes the force history internally on straight piping lengths. A comparison of the peak force from GAFT and from RELAP for every piping link has been generated. Through the comparison, both RELAP5 and GAFT have been verified for the evaluation of rapid valve closure reaction loads. The comparison also shows that the classical method typically over-predicts the best-estimate solution by 15% to 20% for straight piping links. Although not confirmed, a better agreement between the two methods would be expected if a more accurate steam sonic velocity correlation and valve closure model are incorporated into the classical solution. Theis study helps to quantify the degree of conservatism inherent in the classical approach.