Identification of Pool Characteristics of Irrigation Canals

In the Western part of the United States of America, more and more irrigation canal networks are being modernized, which often includes some form of real-time automatic water level control. This real-time control consists of water level measurements that are communicated to a controller. The controller calculates the required control actions and communicates these back to the adjustable gates. The objective of the controller is to maintain the water levels in the canal pools at a given target level. The control loop is repeated at a fixed time interval. To attenuate noise and undesired signal contents, filtering of the water level measurements is required. The reasons that irrigation districts decide to automate their canals include: improved water delivery service, improved efficiency of water delivery, improved utilization of labor, and water resource stewardship. The design of a water level controller is not obvious. On one hand, the reaction to flow disturbances or changes in target level should be fast. On the other hand, tuning the controller to react too strongly to these disturbances can result in overreaction and water levels that severely oscillate. Research on control of open channels has resulted in design (tuning) rules for water level control and filtering that are based on estimates of certain pool characteristics. These characteristics are the storage area of the back water part of the pool, the delay time from upstream side to downstream end and for short, flat and deep pools the frequency and peak magnitude of the basic reflecting wave (resonance) as well. This article presents a procedure to determine the pool characteristics required for controller and filter design. The procedure applies system identification tools and was tested on several canal pools at the Central Arizona Irrigation and Drainage District, Eloy, AZ. The identification procedure results in realistic characteristics for the pools.