Water Journal : Water Journal November 2013
WATER NOVEMBER 2013 50 Feature Article without cavitation issues (minimum net positive suction head or NPSH). This minimum required suction pressure also places a constraint on the transfer capacity of Ravenswood Pump Station, because the pump station draws down the suction pressure in order to increase the ow rate through the system. The ability to meet the design capacity of the pump station therefore relied on being able to draw down the suction pressure without causing problems for the pumps. The testing involved incorporating the control strategies into the WANDA model and simulating how the control system responded to a number of normal operating scenarios and failure scenarios for the future Southern System. Normal operating scenarios included the starting and stopping of either the Tamworth Bank or the Dandalup Bank and ow setpoint changes with both banks running or either bank running by itself. Simulated failure scenarios included pump trip (sudden stop) of the pumps in one bank while the other bank remains running, and the sudden closure of a source regulating valve (sudden cease of supply). The simulations made it possible to compare the performance of the different control strategies in terms of response, stability and the ability to limit the suction pressure from dropping too low. This resulted in the selection of one control strategy to develop further into the control system design for Ravenswood Pump Station. DETAILED DESIGN In the detailed design stage the focus shifted to working out the details of the control system design and documenting the control logic so that it could be programmed. During this phase of the design process the WANDA model was used to simulate design options for problems that required solving. This involved developing potential solutions, incorporating these solutions in the model and running simulations to test their performance. This often involved an iterative process until the desired result was achieved. After the control logic was fully developed and documented, the model was used to determine preliminary control settings for proportional-integral controllers as a starting point for the commissioning of the control system. An aspect that required attention during the detailed design stage was the control logic for automatically starting and stopping pumps within a bank. The pumps in the Tamworth Bank and Dandalup Bank are able to operate safely between 50 per cent and 110 per cent of their best ef ciency point (BEP) ow rate. Pump operation outside of this range for an extended period of time may cause damage to the pumps. In terms of energy ef ciency, it is most ef cient to operate the banks with the fewest number of pumps possible. The control logic for automatically starting and stopping pumps within a bank was designed to operate the bank with the fewest number of pumps possible for a given ow rate, while also protecting the pumps from operating outside of their safe operating ranges. Another aspect related to pump operation that required attention was the sequences for stopping and starting of the banks. The simulations in the early design stages highlighted that each bank had its own unique problem when it came to starting the bank. The Tamworth Bank always starts with already established gravity ow in the Stirling Trunk Main. The bank therefore starts with pump ows that are often too high for the pumps to operate safely, until the pumps reach a speed where the pump ow falls below 110 per cent of best ef ciency pump ow. To ensure the pumps operate within the safe operating range during start-up, the Tamworth Bank always starts with three pumps simultaneously and against a partially closed valve. The valve is slowly opened after the pumps reach a high enough speed for the pump to operate without the aid of the valve. Getting the start-up sequence right required a number of simulations with the WANDA model. The Dandalup Bank experiences problems with low pump ow during start-up of the bank. To overcome this issue, the speed of the starting pump is ramped up quickly until the rst ow develops. After this point is reached, the control system is switched on and the ow controller adjusts the pump speed more gradually until the desired ow setpoint is reached. COMMISSIONING Although the system modelling made it possible to test the detailed design of the control system for Ravenswood Pump Station before implementation, the control system still needed to be commissioned to prove that everything would work in the actual pump station. Because control system modelling is still a relatively new tool, the decision was made to validate the control system model for Ravenswood Pump Station during the commissioning stage. The validated model could then be used to simulate commissioning tests that were particularly hard to set up in the real system, saving time and effort. The hydraulic model of the Southern System was rst calibrated with a steady state eld test that measured pressures along the Stirling Trunk Main for a number of different ow rates. This calibrated hydraulic model was then used to replicate a number of test scenarios for Ravenswood Pump Station's control system. The control system model was rst calibrated by replicating the eld data of one set of commissioning tests. This involved adjusting the control settings in the model until the simulated pump speeds, ow rates and pressures showed close agreement with the pump speeds, ow rates and pressures recorded by the SCADA system. The eld data of a second series of commissioning tests was then used to validate the model by simulating the test scenarios, without further adjustment of the control settings in the model. RESULTS SYSTEM CONTROL STABILITY One of the results of the control system modelling in the early design stages was that the model showed that the two banks could be operated in a stable manner, with both pump banks controlled by ow control loops. The ow control loops of the two banks do interact with each other, but this interaction was not nearly as signi cant as was initially anticipated (GHD, 2009). As it turned out, the Tamworth System and the Dandalup System behave quite differently from a dynamic point of view. The Tamworth System, by design, has high ow rates and low static head, while the Dandalup System has relatively low ow rates and high static head. Figure 4. Control diagram -- pump bank control system.
Water Journal September 2013
Water Journal December 2013