Development of a decision support system for furrow and border irrigation

McClymont, David (2007) Development of a decision support system for furrow and border irrigation. [Thesis (PhD/Research)] (Unpublished)

[img]
Preview
Text (Introductory Pages)
McClymont_2007_front.pdf

Download (301Kb)
[img]
Preview
Text (Whole Thesis)
McClymont_2007_whole.pdf

Download (11Mb)

Abstract

[Abstract]: Furrow and border irrigation practices in Australia and around the world are typically inefficient. Recent advances in computer-based surface irrigation decision support technology have the potential to improve performance, but have had little uptake. Despite considerable academic achievements with individual components of the technology, the implementation of this knowledge into usable tools has been immature, hindering adoption. In particular, there has been little progress in encapsulating the different decision support components into a standalone system for surface irrigation. Therefore, the research problem addressed in this dissertation aims to develop a new decision support system for furrow and border irrigation aimed at increasing the usability of the technology, and improving decision making capabilities. Specifically the research hypothesis is: “That calibration, optimisation, and parameter analysis capabilities can be developed and integrated with an accurate and robust simulation model into a decision support system to improve furrow and border irrigation performance.” Six research objectives have been identified to support the hypothesis including: (RO1) investigate existing surface irrigation modelling technology to determine a model and solution technique structure suitable for incorporating into a decision support system; (RO2) develop a robust reliable simulation engine for furrow and border irrigation for automation within a decision support system under optimisation and systematic response evaluation; (RO3) investigate and develop parameter estimation (calibration) capabilities for the decision support system; (RO4) investigate and develop optimisation capabilities for the decision support system; (RO5) investigate and develop parameter response (design charts) capabilities for the decision support system; and (RO6) develop an objectoriented framework to combine the components developed in Research Objectives 2 to 5 with data management facilities and a graphical user interface. Successful completion of these objectives has resulted in the development of a decision support system for furrow and border irrigation featuring an automationcapable hydrodynamic simulation engine, automated full-hydrodynamic inverse solution, automated optimisation of design and management variables, and automated user-definable real-time generation of system response. This was combined with a highly flexible object-oriented program structure and webbrowser-like graphical user interface. Each of these components represents a unique implementation of the required functionalities, differing from the established software packages (such as SIRMOD and WinSRFR) that use alternate technologies with no automation or optimisation capabilities. Development of the hydrodynamic simulation engine has involved the refinement of the commonly used implicit double-sweep methodology with the objectives of achieving robustness and reliability under automation. It was subsequently found that only subtle changes and manipulations were required in much of the numerical methodology, including derivation of simplified solution equations. The main focus of this research has targeted the computational algorithms that drive the numerical solution process. Key factors effecting robustness and reliability were identified in a study of simulation operation, and treated through these algorithms. Validation was undertaken against output from the SIRMOD simulation engine, with robustness and reliability tested through tens of thousands of simulations under optimisation and automated system response evaluation. The calibration facilities demonstrated that the inverse-solution using the fullhydrodynamic model is a viable and robust methodology for the unique identification of up to three infiltration/roughness parameters. Two optimisationmethods were investigated during this research with objective-functions based upon either a volume-balance time-of-advance equation, or complete simulations of the hydrodynamic model. A simple but robust optimisation algorithm was designed for this purpose. While the volume-balance method proved fast and reliable, its accuracy is reduced due to the underlying assumptions and simplistic model structure. The hydrodynamic method was shown to be accurate, although it suffered slow execution times. It was therefore decided to use the two methods in tandem during the solution process where the faster volume-balance method is used to provide starting estimates for the more accurate hydrodynamic method. Response-surface investigation for the advance-based objective function identified a unique solution when solving for three parameters. It was found that the automated unconstrained optimisation of design and management practices is limited to the selection of one solution variable (time to cut-off) due to non-unique multi-variable solutions. Nevertheless, the developed facilities provide a unique benchmarking of irrigation performance potential. This research has used the earlier-developed optimisation algorithm to automate simulations using a prototype objective-function based upon user-defined weightings of key performance measures. A study of the response-surfaces of different configurations of the objective-function identified parabolic ridges of alternate solutions, so, in practice, the optimisation process simplifies down to optimising only one parameter: time-to-cutoff. It was also recognized that the performance-based objective functions are highly sensitive to numerical discretisation inconsistencies that occur between simulations, which impede solution convergence. The highly customisable, automated, system response evaluation facilities developed in this research offer potential as both a research and practitioner tool, capable of multidimensional analysis of irrigation systems subject to temporal and spatial infiltration variations. A preliminary study demonstrated the importance of infiltration variation on irrigation decision-making, and provided initial guideline layout designs that combined the effects of variable infiltration and three decision variables using a fixed management strategy of minimising runoff. A limited range of response outputs for a fixed management objective negated the potential benefit of visualising a large number of dimensions. Nevertheless, this study provided direction for the subsequent software development with recommendations including: representing system outputs as contours and iso-curves, rather than by the chart axes; representing different infiltration conditions in separate design charts; allowing the user to assign variables to each chart axis; and representing only two decision variables in each chart. Finally, the simulation, calibration, optimisation and parameter analysis components were combined with a database and graphical user interface to develop the FIDO (Furrow Irrigation Decision Optimiser) decision support system. There were three focus areas during this marriage of components; firstly, an object-oriented structure was developed to accommodate program elements concentrating on separating the graphical user interface components from other task related objects for flexible future development; secondly, a database was developed using XML-based technologies to store property, paddock, event and model information; and thirdly, a user-friendly graphical user interface was created with web-browser-like functionality. The software design evolved through many different prototypes with its current design being heavily influenced from the successes and mistakes of the previous attempts. This work represents the first coordinated attempt to develop a decision support system for furrow irrigation linking a database, simulation engine, calibration facilities, optimisation facilities, and parameter analysis capabilities. A major feature of this work is that all components of the system have been developed from first principles using an object-oriented structure, with the primary goal of implementation into a decision support system. This research has contributed to the development of a professional-quality software package to improve the decision-making capabilities of researchers, irrigation consultants, and irrigators.


Statistics for USQ ePrint 4083
Statistics for this ePrint Item
Item Type: Thesis (PhD/Research)
Item Status: Live Archive
Additional Information: Doctor of Philosophy (PhD) thesis.
Depositing User: epEditor USQ
Faculty / Department / School: Historic - Faculty of Engineering and Surveying - Department of Agricultural, Civil and Environmental Engineering
Date Deposited: 21 Apr 2008 03:58
Last Modified: 16 Oct 2013 05:35
Uncontrolled Keywords: furrow irrigation; border irrigation; decision support system
Fields of Research (FOR2008): 07 Agricultural and Veterinary Sciences > 0799 Other Agricultural and Veterinary Sciences > 079901 Agricultural Hydrology (Drainage, Flooding, Irrigation, Quality, etc.)
URI: http://eprints.usq.edu.au/id/eprint/4083

Actions (login required)

View Item Archive Repository Staff Only