Journal of Irrigation and Drainage Engineering

Historically, the distribution of water in irrigation‐canal networks has been poor in many irrigation projects around the world. This has lead to poor use of the water supply. Unsteady‐flow mathematical models of irrigation‐canal networks are needed to examine the transient nature of canal flows. Uses for such models applied to canal networks are real‐time control, gate scheduling, analysis of operations, and training. The Task Committee on Irrigation Canal System Hydraulic Modeling examined a number of the computer programs available for simulating unsteady open‐channel flow. A number of evaluation and comparison criteria were developed related to technical merit, modeling capabilities, and user considerations. The committee identified current limitations of these models and needs for improvements. Limitations of these canal models are lack of publicity, poor documentation, user confidence and knowledge, and the difficulty of calibration.

Unsteady canal‐simulation model usage requires serious investments of time and personnel. The reasons for deciding to invest in a model, as well as reasons not to invest, are discussed. For most cases, it is better to invest in the improvement of an existing model than the writing of a new model. Many excellent models are available, although very few can be considered user‐friendly for the average design engineer. Unsteady flow‐simulation models are not recommended for real‐time control, but are key tools for the study of various control scenarios. Canal models can simulate an actual canal, but the user must input the necessary canal gate‐control algorithms in order to study the effects of various types of automation and control. The user must also understand the basics of the system to be modeled and have the time and capability to determine if model results are reasonable and sound.

This paper presents a critical review of numerical methods used to model unsteady flow in canals. The significance of the various forms of the governing equations is outlined, and the problems associated with the evaluation of boundary drag and head loss are introduced. The attributes of the numerical solution techniques are described. These attributes include applicability, accuracy, convenience, and robustness. Both legitimate and nonlegitimate methods of achieving robustness are considered. Approximate hydrologic techniques are viewed from the perspective gained by a review of the complete equations. Characteristic and finite‐difference techniques for solving the full equations are compared. Practical difficulties in detecting bore‐wave formation during a simulation are noted. Specific techniques are recommended for difficult problems such as the computation of very shallow flows. Practical considerations concerning testing of the techniques and a cautionary note for users of computer‐simulation models are given.

This is one of several introductory papers resulting from work by the ASCE Task Committee on Irrigation Canal System Hydraulic Modeling. It is intended to serve as background for analyses by the Task Committee of several available programs. Irrigation‐canal modeling is based on the same unsteady‐flow equations used for river modeling. However, the canal and irrigation environment present several unique simulation problems generally not encountered in river modeling. These include zero‐depth conditions, mixed‐regime and reversing flow, gate submergence, flow in inverted siphons and culverts, multiple flow paths, and control‐system interfaces. This paper outlines some of these special problems, discusses their importance, and reviews some solution approaches in general terms.

Numerical models of unsteady flow in canals are used in the design, analysis, and operation of irrigation‐canal systems. Their informed use requires knowledge of the potential pitfalls and sources of error. There are three model parameters that influence errors in the model output: the distance step $\Delta x$, the time step $\Delta t$, and a weighting factor $\theta$. The influence of the Courant number on the accuracy of the results is studied. Calibration of topographical and hydraulic parameters and verification of the model are important steps in the successful implementation of canal models. Two tests are presented that can be used to check the accuracy and stability of the programs. The first is a mass‐conservation test, and the second is a ramp‐discharge test. The results of these tests on several commercially available programs are presented. The intricacies of model calibration, model interpretation, and performance parameters are also discussed.

The Cal Poly Model Canal, located at the Water Delivery Facility, Irrigation Training and Research Center, California Polytechnic State University, San Luis Obispo, Calif., was built to accommodate the study and promotion of automated downstream control. Being a large, fully functional scale model of a canal, it was used to collect data for use by the Irrigation Canal System Hydraulic Modeling Task Committee of the ASCE Irrigation and Drainage Division. The data was used by Task Committee members to help evaluate numerical models for simulation of unsteady free‐surface flow. The Model Canal and the collected data are discussed. This paper is part of a series sponsored by the Irrigation Canal System Hydraulic Modeling Task Committee.

To help evaluate and compare canal unsteady flow models, field data were collected from two operating canal systems: the Central Arizona Project and the Salt River Project. These two large projects in Arizona convey agricultural and municipal water through large canals with numerous pumps and check gate structures. Each project has a central data collection facility that automatically stores field data in a computer data base. Collected data include canal water levels, pump flows, check gate positions, and turnout flows. From each central data base, operations personnel extracted a set of data representing a segment of the canal system for a limited time span. Model evaluators used these data sets to verify numerical accuracy of several existing unsteady flow models. By entering canal dimensions and hydraulic properties, initial depths and flows, gate schedules, and inflow and outflow schedules into a computer model, model output can be compared with field water level measurements.

USM is a hydraulic‐simulation computer program developed by the U.S. Bureau of Reclamation to analyze unsteady‐flow conditions in canals. The program USM uses the method of characteristics to solve the complete Saint Venant equations for unsteady flow in open channels. This solution method yields accurate numerical results but sometimes requires more computer execution time than other methods. USM analyzes canal systems with a variety of structural boundary conditions, including check gates, weirs, siphons, and pumps. Check gates can have either manual or automatic gate control. Branching canals, supercritical flow, and channel dewatering cannot be analyzed. This paper describes the model and compares its strengths and weaknesses with those of other models. Several test data sets compiled by the ASCE Task Committee on Irrigation Canal System Hydraulic Modeling were used in the evaluation. USM simulation results are compared to field data collected from the Central Arizona Project canal system.

This paper is one of a series resulting from the ASCE Task Committee on Irrigation Canal System Hydraulic Modeling, including presentation and evaluation of several programs, one of which is CARIMA. The CARIMA code simulates unsteady free‐surface flow in simple or multiply connected systems of rivers or canals. The simulation uses the Preissmann implicit finite difference method for solution of the complete de St. Venant equations and appropriate equations for hydraulic structures. Special features of the code include automatic topology recognition, a user interface for fully implicit incorporation of specialized supervisory and global control schemes, and the lack of any preset limits on network complexity or size of data tables. CARIMA was used to evaluate a canal automation algorithm at the California Polytechnic Model Canal facility. This investigation provided a useful comparison of CARIMA modeling results with actual results recorded on the Model Canal, and helped to set the specifications for the implementation of many of the computational techniques of CARIMA, in the CanalCAD modeling system.

CANAL is a computer model that was developed at Utah State University to perform unsteady‐flow simulations for branching canal networks. This paper describes the features of the model and discusses its strengths, weaknesses, and potential applications. Some of the features of a more recent version of the model are contrasted with the version used in this study. Compared to other similar models, CANAL was found to be very robust and user‐friendly, and more appropriate for canal‐management applications than for hydraulic‐design and emergency‐operation analyses. Several test data sets compiled by the ASCE Task Committee on Irrigation Canal System Hydraulic Modeling were used to evaluate the model, and simulation results are compared to a set of field data from the Central Arizona Project.

DUFLOW is a microcomputer software package for simulating one‐dimensional unsteady flow in open‐channel systems. The program is designed for simple networks of channels with simple structures. Water levels and flow rates are determined by solving the St. Venant equations of continuity and momentum with the Preissmann scheme. The program has a user interface for conveniently entering a description of the network conditions. DUFLOW is public‐domain software and is distributed at a nominal cost. Although the program cannot handle some of the more‐sophisticated modeling needs (e.g., dry beds, automatic gates), it is useful for first‐time users of canal‐network software. Limited user support is available.

MODIS is an implicit hydrodynamic modeling package that computes the unsteady water flow in open canals. MODIS has been developed to investigate the hydraulic performance of dynamic controlled irrigation systems. The MODIS model can run in various computational modes varying from steady‐state mode to a full dynamic mode in which the complete Saint Venant equations are solved. The applied numerical solution technique is always based on finite differences using the Preissmann implicit numerical scheme. In the MODIS model, priority has been given to robustness rather than accuracy, although warning messages are produced when accuracy considerations are violated. The model's innovative features are its accurate computation of structure flow and the real‐time structure‐operation possibilities. The model can also compute the structure parameters needed to realize a specified water level or discharge. This option is useful to generate operational instructions for manually controlled systems. Furthermore, the model has performance indicators that allow for a fast and diagnostic interpretation of the model results. The models weak points are that the user interface is not yet fully menu driven and that the program is not public domain. The program has been applied in many countries for wide variety of water‐management problems.