Surge Transients Due to Check Valve Closure in a Municipal Water Pumping Station
AFT Impulse™ Technical Paper
Authors: David Lozano Solé and Roger Bosch Segarra, Aquatec Proyectos para el Sector del Agua SAU (SUEZ Group), Spain; Trey Walters, Applied Flow Technology, USA
Presented at the 13th International Conference on Pressure Surges 2018, Bordeaux, France, November 14-16, 2018 – Copyright © 2018, BHR Group
Abstract
The present study highlights the importance of proper check valve selection to mitigate water hammer and its associated problems. Two different check valves were installed in a pumping station in a municipal water transfer system: a swing check valve and a nozzle check valve. Measurements were taken of pipeline pressures after a pump trip and resulting check valve closure. The field data was compared to predictions from a model using a commercial water hammer tool. Commonly accepted methods for estimating reverse liquid velocity at check valve closure were utilized.
Results were also compared to previous experimental test from other authors. The calibrated model results matched the field data quite well. Comparisons of inferred valve characteristics to previously published results for swing and nozzle valves were not in close agreement for either tested valve.
CONCLUSION
The conclusion and general recommendations drawn from this study are the following:
The use of HFPT on model calibration is strongly recommended when analysing systems with high wavespeeds, short pipes and significant transmission and reflection phenomena.
The slam effect of a swing check valve originates a peak pressure almost 6 times larger (Psteady state+47 m) than the one from a nozzle check (Psteady state +8 m) during the initial instant. The first and second peak pressures caused by the swing check valve slam (217,7 and 202,4 m) exceed instantly the maximum allowable pressure on the system, 20 bar (200 m), determined by the old surge vessel data-plate. On the other, the peak pressure with the nozzle check (177,8 m) is lower than the maximum pressure set before the pumping station revamping. The test data highlights the importance of a proper check valve selection to mitigate initial pressure peaks. In the light of the results, the water operator replaced the swing check valve installed on Pump 1 with a second nozzle check valve, Figure 15.
Dynamic performance data of check valves from Thorley and Ballun (6, 8), predict quite different reverse velocities than estimated from the measured pressures and modelling results for Tibidabo PS. There are numerous uncertainties when trying to compare modeling results to the measurement, among them non-uniform geometry and measured HFPT slightly downstream of the check valve exit where a converging tee exists and the diameter has changed from DN200 to DN300.
Further, the nozzle check valve body is DN150 with a DN150 to 200 expansion fitting at the downstream flange. In the absence of reverse velocity information, the use of charts from both authors provide the best values for modellers. But modelers should use caution if their geometry is variable as is the case at Tibidabo.
The lack of information regarding the dynamic performance of check valves suggests that modellers assess different reverse velocity values to limit the slam impact on the system.
The Commercial 1D modelling tool used on the test (1), based on MOC, provides reliable results with good field data agreement after model calibration to the first pressure spike. The slow energy dissipation of the model explains the main differences observed with field data gathered with pressure transducers. The lack of information from GIS of operators and the need to simplify the models and associated calculation times are partially responsible for the observed differences. On the other hand, the use of steady friction formulas to compute transient head losses would also explain the pressure difference. This is not a major constraint for modellers for designing proper protection devices as energy decay does not affect the initial instant when pressure oscillation is larger.
Below is an excerpt. Use the link above to view the full paper.
Introduction
Water supply systems are strategic infrastructures that guarantee the welfare of the population and ensure the countrys economic development. This means that water utilities must ensure the continuity of supply, the quality and the service pressure at all time. To meet this goal Asset Management Plans have to be implemented to maintain a desired level of service at the lowest life cycle cost. As part of these plans, renewing and replacing system components are essential activities. Pumping stations are strategic assets in water distribution networks and these activities have a direct impact on them.
When design engineers are faced with a pumping station in need of revamping, they are challenged to size and specify new mechanical equipment at the lowest budget while maintaining the system’s operational conditions. The task becomes more complicated when tackling transient analyses with (frequently) limited information available about the existing system. Engineers are thus compelled to hypothesize and make assumptions. The renewed system should perform in such a way that surge pressures reached during operations are at similar or even lower values than the current ones. It has been widely demonstrated by water operators that the origin of burst mains is frequently a result of unexpected and rapid pressure variations, especially in periods of low water consumption in which energy dissipation potential is reduced.
With this in mind, the solution lies in a proper choice of protection devices (e.g., surge vessels, relief valves, surge tanks). Equally important is proper check valve selection, as this is often the origin of the pressure surge. Regarding protection devices, extensive information of their features and performance is supplied by manufactures or available in the catalogue. On the other hand, limited information is traditionally provided by check valve suppliers related to its dynamic performance despite its direct relationship to waterhammer. Manufacturers don’t usually provide this information because the hydraulic conditions of the system affect the check valve performance.
That means that they need to test their valves under a wide range of scenarios to determine the closing time or the expected reverse velocity. Because of this lack of information, the issue has been extensively studied in research projects where tests have been performed under well-controlled conditions (this will be discussed further in Section 2). These documented experiences provide surge modellers with approximate information about the dynamic performance of check valves.
The system studied in this paper had one operating pump and one standby pump. It also had a surge vessel at the pump discharge side. Thus, the check valve slam was driven by gravity and the surge vessel, and not by parallel pump trip and operation.
With the goal of illustrating check valve slam in an installed system and the difficulties modellers encounter when performing transient analyses, the authors present a case study carried out on a recently refurbished pumping station in the Barcelona distribution network. The water operator, aware of the impact of check valves on surge pressure, decided to install two different check valves to compare the pressures reached during a pump trip event. A detailed hydraulic model of the system was developed with the commercial software AFT Impulse (Applied Flow Technology, 2016 (1)) and the results were compared with field data gathered by the SCADA and a high frequency pressure transducer expressly installed for the test.
The goals of the present case study are:
- Compare the slamming effect of two different check valves: a conventional swing check valve vs. a nozzle check valve.
- Reinforce the importance of proper check valve selection.
- Discuss the importance of energy dissipation during a transient event on complex systems.
- Explore agreement between computed and measured pressures during a transient event.
- Improve understanding of the robustness and reliability of computational modelling tools based on the MOC to perform transient analyses.