Submitted to: North American Water and Environment Congress Proceedings
Publication Type: Proceedings
Publication Acceptance Date: 12/19/1996
Publication Date: N/A
Interpretive Summary: For many large irrigation projects, the operation of the canals that convey and distribute water to the farms affects the farmers' ability to efficiently use the resource. Infrastructural improvements (e.g., conversion to pressurized pipelines) can improve the reliability of water deliveries but are typically very expensive relative to changes in operations. Operations can be improved by developing analytical tools for determining canal gate motions. One such tool is a mathematical model that computes the water delivery schedule needed at the upstream end of a canal to provide a specified downstream water demand. Calculation of the needed inflow is complex because the system serves various users simultaneously, and changes at one location in the canal affect deliveries at other locations. Several schemes have been developed to solve this problem. In this paper, we compare three schemes by use of examples. Furthermore, we propose an improvement to one scheme that provides reasonable results with fewer computational difficulties than a theoretically more precise scheme. Beneficiaries of this research are other scientists working in various aspects of canal control as well as designers and operators of open-channel water delivery systems.
Technical Abstract: Open-channel water distribution systems are difficult to operate because changes in withdrawal rates at one point in the canal affect water levels and delivery rates at other points in the system. Inadequate canal operation in response to changing demands causes unreliable and inequitable deliveries to users and may endanger the physical integrity of the delivery structures. Improving canal operations is, thus, an important strategy for improving agricultural water management and efficiency of use. A method for computing canal operations to satisfy predetermined demand schedules by solving the unsteady flow equations backward in space, has been available for several years. Practical application of the method has been limited, partly because of computational limitations of the algorithm, which is based on a method- of-characteristics solution. Two finite-difference schemes have been proposed recently for solving the inverse unsteady flow problem. A study was conducted to compare these methods with the more theoretically correct method-of-characteristics. One of the proposed finite- difference methods was found to use an incorrect boundary condition specification, resulting in unstable computations for many flow conditions. The second scheme, which locally linearizes the governing equations, produced inaccurate results under certain design conditions. When we made this scheme fully nonlinear, results were as accurate as those obtained by the method-of-characteristics, and the procedure was found to be more robust.