Water Journal : Water Journal November 2013
NOVEMBER 2013 WATER 49 Feature Article When the total source production is low, water is transferred to Tamworth Reservoir, North Mandurah Tank and Caddadup Tank by gravity. When the total source production is high, Ravenswood Pump Station boosts water transfers with two banks of pumps, the Tamworth Bank and the Dandalup Bank. The preferred transfer route for the source water is north via Tamworth Reservoir using the Tamworth Bank. When the total source production exceeds the transfer capacity of the Tamworth System, the Dandalup Bank of Ravenswood Pump Station is started. This bank then boosts the excess source production to either Serpentine Pipehead for transfer to Perth or to North Dandalup Dam for seasonal storage. RAVENSWOOD PUMP STATION Ravenswood Pump Station consists of two separate pump banks, the Tamworth Bank and the Dandalup Bank, that operate as two separate pump stations (Figure 2). The Dandalup and Tamworth Banks have each been con gured with three variable speed pump sets. The Tamworth Bank operates with relatively high ow rates between 125 ML/d and 265 ML/d and low heads up to 120m. The Dandalup Bank operates with relatively low ow rates between 30 ML/d and 130 ML/d and high heads up to 225m. The maximum ow rate achievable for the Ravenswood Pump Station with both banks in operation is 285 ML/d, which is governed by the residual head at the suction side of Ravenswood Pump Station and hence the Net Positive Suction Head (NPSH) available to the pumps. METHODOLOGY The Southern System is relatively complex, with a number of connected sources and supply points, as well as supply directly to reticulation systems in Mandurah and Caddadup. In designing the control system for Ravenswood Pump Station, both the current and the future operation of the system had to be taken into account. Because the future operation of the Southern System depended on the design of Ravenswood Pump Station's control system, the design process was required to be iterative. It was also important to involve key stakeholders to ensure the nal control system design would meet the Water Corporation's operational requirements. At the start of the design process, a System Modelling Guiding Group was formed to guide the system modelling for Ravenswood Pump Station. The group consisted of a well-rounded range of disciplines and included an asset capability manager, system planners and a pump station design manager from the Water Corporation, as well as a water systems consultant (from GHD at the time). In order to simulate the dynamic behaviour of the Southern System with the new Ravenswood Pump Station, a purpose- built model was created in hydrodynamic simulation package WANDA (Deltares, 2008). WANDA is one of a small number of software packages that have the ability to simulate the dynamic behaviour of control systems in pipe networks. The WANDA model created was based on Water Corporation WATSYS models of the Southern System and the preliminary mechanical design of Ravenswood Pump Station (Figure 2). In WANDA, control behaviour is modelled by measuring hydraulic parameters in the hydraulic model and using these as inputs to control logic components such as proportional-integral controllers (Figure 3). The control logic components calculate control actions, such as a change in pump speed, based on the control parameters (e.g. proportional gain, integral time constant) and the difference (error) between the measured hydraulic parameter and the setpoint for the parameter. The output of the control components is then fed back into the hydraulic model to change pump speeds or valve positions. EARLY DESIGN In the early design stage, the focus of the system modelling was on developing an understanding of the Southern System's dynamic behaviour and testing a number of control strategies for Ravenswood Pump Station. A major concern was whether the ow controllers of the Tamworth Bank and the Dandalup Bank would be able to operate together in a stable manner. The two pump banks draw water from a common trunk main without an intermediate tank and, therefore, interact hydraulically. The concern was that the ow control loops would interact in a way that would result in the banks "hunting" for their ow setpoints, without ever reaching a stable condition where both banks would be running steadily, side by side. Another concern was the ability of the control system to prevent the pressure at the suction side of the pump station from dropping too low. The pumps require a certain minimum suction pressure to operate Figure 2. Hydraulic model -- Ravenswood Pump Station. Figure 3. Pump bank control system model.
Water Journal September 2013
Water Journal December 2013