To provide safe and drinkable water for end-user, adequate water treatment control is nec-essary. We intend to control the water system to approach the desired output of the system. For water treatment system, water suppliers want to control the hydraulic stability and also optimize each process of the treatments. For water distribution system, we want to control the system to supply steady quantity of water and qualified water.
2.3.1 Water treatment with process control
With controller to water treatment system, we aim to control the system to maintain water quality and meet time variant consumer demand for water. To control and understand more about the behavior of the water treatment system, simulators are developed to assist water suppliers to realize how systems react under different control strategies and situations [11]. The followings are some currently used simulators :
• OTTER
OTTER is coded in FORTRAN and developed by WRc. OTTER simulate the dynami-cal changes in raw water quality, flow and process operating conditions. OTTER assist the operators to analyze the disinfectants changes under different strategies, to study the effects of pH control on coagulation strategies, to predict the maximum throughput of a treatment works and to evaluating the monitoring strategies on a treatment works, etc..
OTTER is designed to be operated with friendly GUI interface for user [12].
• Stimela
Stimela models the environment of water treatment system developed by DHV Water BV and the Delf University of Technology [13]. Stimela especially focus on the analysis and simulating water quality. Stimela models of drinking water treatment processes calculat-ing the changes of water quality factors, such as pH, oxygen concentration etc.. Stimela also allows different models to be connected, hence the effects between each models could be evaluated.
• WatPro
WatPro is the water treatment simulator for predicting water quality based on specific treatment processes and chemical addition. WatPro provides analysis on steady water condition [11]. WatPro models the behavior of each treatment process and concern the dynamic of the chemicals in water. WatPro is easy for user to generate different water treatment topology [14].
Most of these simulators focus on different procedures of water treatment system. They simulate the behavior of flocculation, disinfection, sedimentation, etc.. Few of these simulators particularly emphasize on the control algorithm of water treatment system.
Most water systems use open-loop control to adjust the systems. For example, chlorine would be joined to the system at the beginning of the treatment, and then if the concentration of chlorine at output doesn't meet the standard, more chlorine would be added at the output position. The concentration of chemicals in water would be affected by the hydraulic dynamic of water, since the volume of water affects the concentration of chemicals strongly. The time-variant volume of water in water system would be a great factor to influence the water quality.
Therefore, excessive chlorine might be injected at output and may cost more with open-loop control. The concept of open-loop control water system could be shown as Figure 2.3.
Figure 2.3: Schema of open-loop water system
Open-loop control can't really stable the system under the time-varying consumer demand and raw water quality disturbance and also meet the system requirements at the same time.
Hence, we want to use the idea of feedback control to balance the injection of chemicals and also the volume of water preserve in water system. The block diagram in Figure 2.4 shows the general idea of feedback control in water system. With feedback control, the control center monitor the output of the system and adjust the control command with the feedback result. With the real-time monitoring and adjusting, the outputs of the system are expected to meet the desired values.
The most difference between open-loop control and feedback control in water system is that feedback control traces the output of the system and then adjusts the control signal to lead the system to approach the desire output. In this paper, we design the feedback controller with model predictive control algorithm.
Figure 2.4: Schema of close-loop water system
2.3.2 Water distribution with process control
The model of water distribution system is developed in past decades. One of the most important and common issues of water distribution system is water quality of the system. Clean water leaving the water treatment system may fall out of the standard in the delivering process.
Since we usually want to keep a detectable concentration of some specific chemicals in water, such as chlorine, the reaction property of these chemicals itself might affect the concentration at consumer position after long delivering time.
Chlorine is widely used in water system to protect the potentially risk of pathogen from public. The concentration of residual chlorine could be consider as an dominated factor for water quality. The concentration of chlorine should be enough to achieve sufficient disinfection, but not too much to produce excessive by-products. Furthermore, chlorine in water will decay with time. To maintain an appropriate concentration of chlorine become one of the major problems in water distribution system.
Several models have been developed to study the changes of chlorine in water distribution system. In [15], a mass-transfer-based model is proposed to predict chlorine decay in drinking water distribution system. The model concerns the reaction between chlorine in water and pipe wall and also the decay rate of chlorine. An input-output model of chlorine transport in drinking water is proposed in [16]. A model tracing chlorine transport using a time-driven approach and evaluating chlorine decay in drinking water distribution system is mentioned in [17].
An general model to describe water quality is proposed in [18]. The model evaluates the changes of chlorine in water according to the dynamic property of water distribution system.
Feedback control algorithm is applied to the model to meet the different position of end-user. In [19], adaptive control algorithm is utilized to control the water distribution system to meet the time-variant consumer demand and water quality.
In this paper, we focus on analyzing water treatment system. We consider only water treat-ment system as we treat-mention water system. We propose s system to model the water treattreat-ment system and apply feedback control to water system. In the proposed model, model predictive control is the control algorithm we apply to control the water system.