The project focused on maintenance of the main channel system of an irrigation system, and had 3 stages:

1) A suite of methods that simulate the link between the observed condition of an irrigation system to its performance
2) A method that uses (1) to predict the impact of various maintenance options (i.e. changes in condition), and so sets priorities. Other issues affecting scheduling maintenance are to be included.
3) Testing of the procedure developed at two schemes (see under Results below).

Fuzzy logic is a mathematical tool that appears well suited to achieve Stage (1) because the observations of system condition are usually uncertain or in linguistic form (e.g. 'poor condition', 'inadequate deliveries', etc).

Work under Stage 1 progressed on four fronts:

1) Development of trial applications of fuzzy logic procedures to irrigation schemes. The inputs have been asset condition and outputs have been related to the hydraulic performance of irrigation systems. The use of farmer questionnaires in diagnosis was also investigated, and questionnaire results were analysed partly using fuzzy logic procedures.

2) Development a simplified hydraulic modelling system to undertake the same task as (1) just above.

3) Development of risk assessment techniques to link the structural condition of assets to a measure of performance in terms of reliability.

4) Development of procedures to convert measures of hydraulic and reliability performance of an irrigation system to an overall performance of more direct use to maintenance planners who need to include benefit-cost considerations. A method based on Fault Trees was developed.

The principal approach used under Stage 2 was the computation of benefit-cost ratios for each potential maintenance task. Engineers responsible for maintenance scheduling can use benefit-cost ratios to improve their subjective judgements of priority. The ratios provide a 'score' for each potential maintenance task. The engineers can then include other considerations, such as farmer preferences, to select a final set of tasks. A key advance on the current subjective process by which potential maintenance tasks are assessed has been made here: benefit-cost ratios enable both restorative and preventative maintenance activities to be directly compared. Maintenance of drains can also be assessed using benefit-cost ratios in the same way and can also be directly compared against the other categories of maintenance.

More sophisticated methods for deriving priorities for maintenance were developed to enable comparisons against the simple benefit-cost ratios. The more sophisticated methods were a system-wide benefit-cost analysis and multicriteria analysis (using two different methods).

The fieldwork enabled a full hydraulic model of the channel system to be developed. The link between maintenance activities and deliveries to outlets were assessed using the full model and these results were used as a 'benchmark' against which to test the new hydraulic analysis methods. The fuzzy logic method was found to require some calibration and its predictions had reasonable accuracy. The minimalist model proved more accurate.

A larger system, the 2400 ha Tirunelveli Channel System, was used to test the practical applicability of the methods. The Minimalist Model is found to be an appropriate technique for assessing the impact of restorative maintenance in the main channels of the system. It could be applied without imposing a requirement for excessive data collection and the results were found to be not unduly sensitive to uncertainties in the input information.

The Fuzzy Logic Method was found to provide a good prediction for the irrigation system. However the application at Tirunelveli was found to highlight practical difficulties in using the technique.

The Fuzzy Logic Method and the Minimalist Model are alternative techniques to fulfil the same function. Each has its advantages:

The benefits resulting from preventative maintenance on structures and canal banks was assessed using Risk Analysis. The method based on Fault Trees was tested at both field sites. The failure probabilities were derived by interviewing farmers and the Irrigation Assistants, all of whom had long experience of the system. The applications at Radha and Tirunelveli were found to provide reasonable assessments of preventative maintenance with an acceptable level of data collection from the farmers and engineers.

The simple benefit-cost ratios were found to provide very informative results at Tirunelveli, the test system. It was found that benefit-cost ratios can be used as a suitable basis for setting priorities for maintenance when applied in combination with engineering judgement. The more sophisticated methods for deriving priorities for maintenance were found to provide no significant additional information or accuracy.

Application of the techniques at Tirunelveli has resulted in a curve of returns to be produced for the maintenance. The curve indicates that when a normal level of scheduled maintenance is applied, about 250 Rupees per hectare or about US$2.5 per hectare, then the predicted returns are relatively very high: about 40 Rupees of benefit (improved or saved farmer income) per 1 Rupee spent on maintenance. When rehabilitation, i.e. deferred maintenance, is applied then predicted returns are an order of magnitude lower at Tirunelveli. The same pattern can be expected at other systems, but with different values.

A further outcome has been assessments of the economic returns provided by scheduled maintenance activities for an irrigation system in Tamil Nadu, India. The returns were found to be an order of magnitude higher than those for rehabilitation at the system. Application of the procedure at other systems will enable equivalent assessments to be made.

HR Wallingford, Indian Institute of Technology Madras and HydroControl, 2000. Prioritising maintenance and rehabilitation of irrigation/drainage systems on the basis of multi-parametric descriptors of scheme performance. Final Report EX, HR Wallingford Ltd, Oxfordshire, UK.

Tel: +44 1491 835381
Fax: +44 1491 826352
Email: dfid-kar-water@hrwallingford.co.uk