Water Journal : Water Journal August 2011
water AUGUST 2011 75 refereed paper hydraulics Model Simulations and Observations Water Hammer and Surge Protection Design During the concept design (target out- turn cost, or TOC) phase, a decision to use several different transient hydraulic modelling software packages was made to ensure modelling results were not affected by particular numerical features that may be unique to a model, and that may produce different results for some of the critical hydraulic elements of the existing network. This also provided a means of internally reviewing and checking modelling performed using the software package that was predominantly used for design. Multiple modelling approaches were important in providing the required confidence to the WDA. There were tight constraints on the allowable surge performance in the existing system (+/- 5m from normal operating levels) for any transient event created by the new infrastructure, and physical compliance with these limits formed a critical part of the WDA's Key Performance Indicator (KPI) for operating performance. As such, the results of the surge analysis and surge protection design were of particular interest to SWC and the WDA. The key objective was to adopt the model and complexity level that was as functional, efficient and accurate as practical. The approach taken was to produce and run four independent water hammer models of the system, of various network complexities. These were three commercially available models: HAMMER (Bentley), HYTRAN (Hytran Solutions) and WATHAM (Hydraulic Computer Programming), plus a fourth check model developed in-house by KBR. The water hammer models were developed to include both the WDA delivery infrastructure and the existing Potts Hill system. The Potts Hill system model was developed by utilising the trunk hydraulic model developed by SWC (Prospect Trunk Model -- Version No. 0.51) using InfoWorks (Wallingford). SWC's trunk model contained all the required network geometry verified prior to this project by SWC. All models were developed from the origin of this trunk model. Three network extents of differing complexities (see Figure 2) were then developed to allow adequate cross-checking with the available software packages: Full detailed network model: This model included all of the trunk system (as available in SWC's InfoWorks trunk model) and contained 2837 nodes, 3112 pipes, the seven major pump stations, downstream trunk networks and associated reservoirs that deliver water to zones from the Potts Hill gravity system. The full model was analysed in HAMMER only. Cut-down detailed network model: This model contained 608 nodes and 657 pipes, and included the two key gravity tunnels (City and Pressure Tunnels) and the major trunk-connecting network. The major pump stations, other than WP0369, and 'dead end' sections of the gravity trunk system were removed from the full model. As flows to these sections were typically supplied as gravity flow from the Potts Hill Reservoir, and not pumped flow from WP0369, it was proposed that this step could be taken without considerably sacrificing the accuracy of results. To further prevent loss of accuracy, representative node flow demands were required to simulate the flows at the pump stations and deleted gravity sub-zones to maintain flow velocities throughout the system. The cut-down network was analysed in HAMMER and HYTRAN. Basic model: This was a highly simplified model containing only the WDA infrastructure and the City Tunnel to Potts Hill Reservoir, almost reducing the network to a simple transfer pipeline. This model contains 90 nodes and 91 pipes. Representative node flow demands were manually included to simulate the flows at offtakes from the City Tunnel. The basic model was analysed in HAMMER, HYTRAN, WATHAM and the in-house software developed by KBR. The reasoning behind developing models of varying complexity was the large numbers of potential operating scenarios and configurations in the system (approximately 60 base scenarios) that needed to be tested as part of the design on this project. Any reduction in model complexity that did not result in a loss of accuracy would create a distinct advantage in terms of computational time. This was particularly true for the concept design stage and when on-site testing and commissioning activities were taking place. Sensitivity of the modelling results to time-step selection was undertaken, and a uniform time-step of 0.025 seconds for each of these models was found to be suitable. Model Comparisons The biggest advantage HAMMER and HYTRAN had over the other software programs used for modelling in this project was their compatibility with EPANET (public domain software developed by US Environmental Protection Agency's Water Supply and Water Resources Division). InfoWorks also had compatibility with EPANET, allowing the original, verified model geometry and system set-up to be imported into these water hammer programs with minimal manual manipulation. This was important in minimising error and time taken to set up the models. WATHAM was not capable of importing directly from EPANET, requiring manual formatting of data prior to entry. [Note: the latest release of WATHAM now has the ability to import from EPANET.] presented at Figure 2. Model extents.
Water Journal September 2011
Water Journal July 2011