The present report includes five parts:
· The paper first explains how irrigation scheduling, delivery scheduling and operation of the delivery system are inter-related.
· The second section is devoted to the qualitative features of deliveries, i.e., reliability, equity and flexibility.
· The third section sets out the means and tools that can be applied to increase operational flexibility and reliability at either farm level (on-farm storage, conjunctive use of groundwater) or delivery system level (proper design, automation). It also includes developments about the computer programs now available for improving operational procedures or the overall water management in a project.
· The fourth section deals with the problem of discharge control and measurements, a key point for the application of crop-based, water-saving irrigation scheduling methods.
· The last section sums up the conclusions and recommendations that can be drawn from the above developments.
Irrigation scheduling at farm level results in a delivery schedule, i.e. a plan or, in the best cases, a real-time decision on when, how long and with which flow rate, expressed in absolute value or as a sharing proportion, water should be delivered at the farm gate. Then the problem arises of supplying water accordingly.
It would be nice if every farm could be supplied with enough irrigation water from an individual well or spring. The farmers would be then in a position to fully master their irrigation schedules and to best manage irrigation farming for their crops and profit, according to the on-farm constraints and only the external constraints from economics and the environment. In reality, such cases are exceptions and most irrigated areas throughout the world are partly or fully supplied from collective delivery systems.
The dependence on a collective system implies social, cultural and policy constraints which are dealt with under Theme 6 of the Workshop. It also means that effective application of any irrigation scheduling method and effective implementation of the corresponding delivery schedule are subject to the physical capability of the collective system for delivering water according to this schedule, and to the capacity of the management for operating the system properly.
As an example of the necessary consistency between irrigation scheduling and the operational features of the delivery system, it is clear that crop-based scheduling methods can hardly be applied with systems in which flow control consists in splitting a more or less known incoming flow in fixed proportions related to the field areas. At the opposite end of the spectrum, a pressure pipe or canal system under downstream control delivering water on demand allows farmers to apply the most advanced irrigation scheduling methods.
It should not be inferred from the above examples that the only way to achieve sustainable irrigated agriculture is by converting all existing systems into on-demand systems. We only mean that on farm irrigation scheduling, delivery scheduling and the configuration and control mode of the delivery system are closely inter-related and must be considered as a whole when planning and designing a new irrigation scheme or the rehabilitation and/or modernization of an existing one.
In addition to the technical aspects, special attention should be given to the non-physical factors that may affect operation, such as the social, educational and policy constraints, so as to avoid the often reported discrepancies between the planned delivery schedule and operational method on the one hand, and those actually applied on the other.
RELIABILITY, EQUITY AND FLEXIBILITY OF DELIVERIES
As recalled by Burt (1996), reliability of deliveries is a prerequisite for the application of any irrigation scheduling programme. Whatever the delivery schedule, either dictated or resulting from an agreement, either rigid or flexible, it is necessary that water should be supplied to the users in conformity with the expected level of service.
Besides the technical advantage of irrigating in the best possible conditions with respect to possible limitations that are known and accepted by the farmers, reliability of deliveries is an essential condition for establishing, sustaining or restoring a climate of confidence between the water supplier and the water users. Such a climate is indeed a prerequisite for due payment of the water charges, and for the success of any move towards technical or institutional measures aiming at better water management, such as improved delivery scheduling and farmers' participation in operation, maintenance and management through water users' associations.
The requirements for reliability are:
· A design of the conveyance and delivery system that must be not only technically sound from the hydraulic, constructional and operational points of view, but also consistent with both the service required and the local conditions.
· A good-quality construction for both the works and the equipment, all in accordance with the specifications and drawing resulting from the above design.
· Proper operation and maintenance (O&M) of the system, which implies good organization and dedication from the managing level to the field O&M level, as well as enough budgetary resources to cover maintenance costs.
Reliability over time will be all the better since provisions have been made at the design and implementation stages to accommodate any changes in the physical condition of the system or the service required.
Equity versus flexibility
Equity, i.e., deliveries in proportion to the cropped areas served, has long been, and still is, the leading principle of delivery scheduling in most projects, especially run-of-the-river schemes and where water is scarce. This principle is, moreover, in accordance with the traditions or laws governing water rights in many countries. Such systems can be implemented and operated with very simple non-adjustable structures and equipment and no communication facilities. Operational and managing requirements are thus reduced to a minimum.
In fact, theoretical equity is seldom a reality. Two categories of reasons can be considered:
· It is virtually impossible to design non-adjustable outlets that work properly over a wide range of flow rates and levels. In addition, the functioning of the dividing structures is inevitably affected by the drift of canal hydraulic parameters over time (Plusquellec et al., 1994).
· Especially in the case of loose management, the actual sharing out of water may deviate widely from the official scheduling, as a result of the dominant position of some farmers, non-registered transactions on water rights and land ownership, or uncontrolled extension of irrigated area (Bandaragoda, 1996).
On the other hand, except for mono-cropping and cases where water requirements of different crops are similar, equity-based delivery scheduling is obviously inconsistent with crop-based and water-saving irrigation practices.
However, case studies presented by Bandaragoda (1996) show that attempts to move from equitable, or so presumed, delivery schedules to more flexible crop-based schedules may fail because of resistance to change from not only the managing agencies but also the users.
Whether the decision on when and with how much water to irrigate is up to the supplier or to the users, flexibility of deliveries is still acknowledged as a precondition for applying efficient methods of irrigation scheduling. Adjusting water supply to irrigation water needs varying over time and space obviously implies that deliveries to each farmer or group of farmers could be varied in frequency, flow rate and/or duration and, consequently that, the whole delivery system could be operated in a flexible and efficient way. This can be achieved by applying relevant design concepts, equipment, and operational tools, referred to in the following section.
However, as developed by Burt (1996), delivering water with a high degree of flexibility and reliability depends not only on technical means, but also requires:
· a decentralization of decisions and responsibilities based on the breakdown of the delivery system into 'levels of service' (e.g. main canal level, secondary canals level, etc.), each one being responsible for providing a well-defined service to the next lower one;
· the institution of seasonal or yearly water allocations, so as to avoid or, at least, minimize the water wastage and inequities that may result from flexible schedules, and especially from on-demand schedules (cf. Horst, 1996).
Finally, applying flexible deliveries requires a number of technical, social, educational, organizational and financial conditions that are seldom met in developing countries, as set out by Horst (1996) and Mangano (1996). Both have reservations about the suitability and applicability of flexible delivery schedules in such countries. They consequently recommend sticking to the simplest scheduling and operating modes, and ensuring that they are actually applied and trying to obtain the most from their low flexibility. We can only agree upon with realistic approach, based on the fact that 'successful irrigation scheduling must be simple to implement and easy to understand from both the farmer and project management personnel standpoints' (Burt, 1996) and that the excess deliveries resulting from simplified schedules are likely to be compensated for by the decrease in operational losses resulting from simpler operation (Horst, 1996).
However, considering the absolute necessity for water savings and sustainable food production, we are convinced that improving the flexibility of delivery schedules and of delivery system operation to some degree should be a permanent concern, at least for the longer term.
HOW TO IMPROVE OPERATIONAL FLEXIBILITY AND OVERALL WATER MANAGEMENT
Flexibility through on-farm measures in spite of rigid deliveries
In spite of rigid delivery scheduling, flexibility of on-farm irrigation scheduling can be obtained by providing a storage capacity below the delivery point, so as to compensate for the expected mismatch, especially in time, between deliveries and consumption. Construction costs are not so high, but, in the general case where the reservoir area has to be taken from cultivable land, it should be ascertained that the benefits resulting from more efficient irrigation practices do offset the loss of cropped area. Favourable topography will allow gravity irrigation, otherwise water must be pumped from the reservoir, which entails additional investment and operating costs but may allow the application of water saving irrigation techniques such as drip irrigation.
A second way, which may be combined with the one above, is the conjunctive use of groundwater, where available and of a quality good enough for its application as it is or mixed with surface water.
Flexibility of main system operation through a proper design process
It is beyond the scope of this paper to detail the control concepts and techniques applicable to canal and pipe systems. We will only say that they cover all possible degrees of operational flexibility and responsiveness to demand. People wishing to go deeper into the matter may refer to papers and publications specially devoted to the subject (Goussard, 1993; Plusquellec et al., 1994; Zimbelman, 1987).
As the above references are mainly related to canal systems, we must point out that pressure pipe systems have the advantages of eliminating seepage losses and being responsive to demand by nature, of course within the limits of their conveyance capacities. However, their application is restrained by a comparatively high cost in the cases of long distances and large flow rates. Nevertheless, a low-pressure piped distribution system, which can be implemented at a reasonable cost, is often an advantageous alternative to tertiary canals.
However, there is no recipe for an ideal control system that would be applicable to all projects. Proper design of irrigation delivery systems is indeed a matter for engineering art and science, subject, however, to a preliminary comprehensive definition of the frame and objectives of the project. As pointed out by Plusquellec et al. (1994), satisfactory performance implies that the design should start with the definition of a proper operation master plan that combines and reconciles the various perspectives and expectations of the parties involved, i.e., the farmers, the field operators, the project management, the government and the evaluators. According to the authors, the essential components of such a plan are: level of service anticipated, levels of control in major, minor and watercourse canals and pipes, water ordering procedure, communications, decision support system, data collection and processing, who will do what, plus, at a later stage, specific operating instructions for all control structures.
In a similar perspective, Ankum (1996) presents a classification and a selection process of the control methods applicable to canal delivery systems together with a logic diagram showing the inter-relations between those methods, the modes of water allocation and water distribution, and the decision-making procedures.
As it eliminates uncertainties and inadequacies of human interventions, automation of some degree is acknowledged as an efficient means to simplify operation, increase the reliability and flexibility of deliveries and reduce operational water losses (Goussard, 1993; Plusquellec et al., 1994; Zimbelman, 1987). In addition, control methods responsive to demand cannot be implemented without automation.
However, besides unquestionably successful applications, not only in developed countries but also in developing countries (e.g., in north Africa), there are a number of cases where the introduction of automatic control has failed. The reasons for such failures are a design that is too sophisticated with regard to the local conditions and, above all, a lack of sufficient preparation, information and training of the farmers, operators and managers. Automatic control thus passes for costly, cumbersome to understand, master and maintain, and consequently requiring a highly skilled staff with sound knowledge in electro-mechanics, electronics and computers (Horst, 1996).
We will here take issue with this widespread but hasty judgement which prevents developing countries from taking advantage of modernization through well-planned and well-engineered automation. It should yet be clear that, in many cases, efficient automatic control, especially of water levels, can be implemented with very simple structures, such as fixed weirs, or equipment, such as float-operated gates and valves. As for the extra cost, it ranges from around 6 to 8% of the total cost of a new manually-operated system, an amount which can generally be recovered in a few years by water savings, increased yields and improved social order.
Computer-aided manual operation and overall water management
The use of an operation simulation model to elaborate setting points for the control equipment of the delivery system does not generally increase the potential reliability and flexibility of the system but may contribute to making them a reality. Numerous user-friendly and versatile simulation programs, that can be run on personal computers by non-specialists in computer science after only a short training, are now readily available (CEMAGREF, 1992; Montañés, 1995). However, due to the hazards of calibration and to the inescapable drift of hydraulic parameters over time, the outputs of the model should be considered with caution and frequently re-validated through feedbacks from the field.
On the other hand, if a model can be really helpful for appreciating routing times and adjusting the operating schedules of discharge regulators accordingly, it should not be forgotten that steady manual upstream control can only be based upon level control. In fact, except for the discharge regulators located at the heads of the main and branch canals, the instructions to operators should simply be to maintain a constant water level upstream of each check structure, something which does not require any simulation. Finally, such models appear to be more suitable for tuning up a system design and testing various operational strategies and procedures than for day-to-day operation.
The so-called 'integral' models have a much more comprehensive purpose than just the improvement of delivery system operation. They aim to improve the overall water management in a project, in integrating the various agronomic, economic and operational factors that affect the efficiency of water use. They typically combine simulation of crop water requirements, on-farm scheduling, delivery scheduling and/or main system operation, and can process management indicators.
Three models of this type have been presented:
· The SIMIS program (Scheme Irrigation Management Information System, Morábito et al., 1995) is essentially an information and evaluation system. The reported application to a scheme in Argentina has shown that it can be helpful for the managing staff in detecting discrepancies between plans and reality and in localizing the weak points in the physical infrastructure, operation and management of the delivery system.
· The INCA (Irrigation Network Control and Analysis) software (Makin and Cornish, 1996) is both an information and analysis tool for general management and a decision support tool for day-to-day operation of medium to large-scale irrigation schemes. In particular, it can be used for determining not only optimum delivery schedules according to computed crop water requirements, but also the corresponding operating schedule of the control structures of the main system. Its application to several schemes in Asia, combined with systematic monitoring, has reportedly led to appreciable water savings while at least maintaining and often improving the reliability and equity of deliveries.
· The model package described by D'Urso et al. (1996), currently under trial application in Italy, is designed as a decision-support tool combining a mapping of crop water requirements by means of remote sensing techniques with models simulating the crop-soil system, farmers' irrigation preferences and delivery system operation. Its original feature is that it has been specifically developed for on-demand irrigation systems and therefore includes a module simulating farmers' behaviour.
Such models are undoubtedly powerful and useful tools for project planning, design and evaluation, especially in centrally managed projects and in water scarcity conditions. However, considering their heavy requirements in terms of management organization, extensive data collection and monitoring, calibration, training and installation time, their use for current system operation appears much more questionable. We are afraid that the applications of such sophisticated models will always remain rather restricted and one may ask whether improving the operational reliability and flexibility of the system through some modernization would not be simpler and more beneficial.
It should be also noted that, in developed countries, where the responsibility for irrigation scheduling is left to farmers, the use of irrigation scheduling models has moved from day-to- day operational applications by the water supplying agencies to on-demand assistance available to farmers from public or private irrigation advisory services (Burt, 1996).
FLOW-RATE CONTROL AND MEASUREMENT AT DELIVERY POINTS
The institution of volumetric allocations is acknowledged as a real incentive, and even an essential condition (Burt, 1996), for farmers to practise efficient, water-saving irrigation scheduling. This requires the application of proper procedures and equipment for effective control and measurement of deliveries. No paper has been devoted to this problem, so some elaboration on the matter follows.
Whatever the distribution system and the delivery schedule, flow-rate or volume control and measurement at every delivery point is an essential factor for good water management at both farm level and overall system level:
· for the farmers to know the quantity of water they are receiving, as necessary to arranging their irrigation within the farm accordingly and to checking whether deliveries are in accordance with the agreed or dictated schedule;
· for the turnout operator (farmer or supplier's agent) to adjust the quantity delivered according to the pre-defined schedule or, in the case of on-demand deliveries, to the actual needs;
· for the supplier, as knowledge of deliveries is one of the essential data for water management, as it is also the basis for water charges assessment and, as the case may be, for volumetric invoicing and possible measures against overconsumption.
In reality, from a global point of view, proper measurement of deliveries, either directly in volumes or through integration of flow rates over time, appears to be rather an exception than the rule, at least for canal systems.
In the case of pressure irrigation, where water is delivered through a pressure pipe system, the simple, inexpensive solution consisting of a totalizing propeller meter, possibly combined with a discharge limitor, has been adopted worldwide.
In low-pressure piped distribution systems for surface irrigation, the discharge can be measured from time to time with a weir installed on a masonry box just at the outlet, but, even with a fixed outlet valve setting, volume values derived from such measurements at intervals are all the less reliable as the pressure, and consequently the discharge, varies greatly. Where pressure variations are too high for reasonably considering the discharge as constant, there is no other satisfactory solution than fitting the outlets with suitable totalizing propeller meters, or else with recently developed 'constant discharge outlet valves', which, in the fully open position, deliver a specified flow and maintain it nearly constant with well-known margins over a relatively wide pre-defined range of pressures.
Numerous methods have been applied to measure or at least assess the discharge of canal turnouts. They are based either on level measurements just upstream of a weir or flume located downstream of the turnout or integrated with it, or on the discharge characteristics of a calibrated orifice which may be the turnout gate itself. More seldom, velocity measurement through a propeller meter within a short pipe section can. be used. However, practice has revealed that they are not so suitable for the purpose, and one cannot keep count of the projects, especially in developing countries, where measuring facilities have been provided but are either misused or, more often, no longer used at all.
All the above methods have two major disadvantages:
· The flow over or through a turnout gate is highly sensitive to upstream level variations. As a result, unless the upstream water level is permanently kept within a narrow range or, in the case of control through an orifice, upstream level variations are insignificant in comparison with the water head on the orifice, the flow delivered is likely to vary greatly once the turnout gate has been set. Proper integration of flow rates over time, as necessary to obtaining reliable volume figures, would then require continuous recording of flow rates. This may be considered for some key points in a system, but not for every turnout.
· The procedures generally required for a correct application, such as accurate readings of scales combined with simultaneous adjustment of the turnout gates, or the use of charts, are often beyond the capabilities and dedication level of the average turnout operator.
However, it is possible to move beyond this deadlock:
Firstly, as unsteady turnout discharge results from unsteady upstream level, any improvement of the level control in the supply canal, if only through weirs, will reduce turnout discharge variations and thus greatly help assess the volumes delivered. For example, if the level variations in front of a turnout sluice gate are less than 20% of the average head on the gate, the discharge will vary by less than ±5% maximum around its average value, a margin small enough in irrigation practice for the discharge to be considered as virtually constant. The discharge can be then estimated, if not really measured, from the number of opening turns of the operating device, which of course implies a preliminary calibration of the gate. In the same conditions, the discharge can also be derived from velocity measurements through a propeller meter, possibly a portable one, placed in a short pipe section following the turnout gate.
Where level variations are too high to consider the flow through a simple gate as constant, a satisfactory solution consists in installing so-called 'constant-flow distributors' or 'modules'. Such pre-calibrated units, generally made of steel, combine three functions: on-off, step-by-step adjustment of the discharge to the required value from zero to maximum by on-off operation of shutters with conveniently stepped widths, and maintaining it nearly constant, i.e., within a ± 5% margin in current practice, with no re-adjustment in spite of upstream level variations within a relatively wide pre-defined range. As the nominal discharge of each calibrated sluice is marked on its shutter, measurement comes down to reading and totalizing the discharges of the open shutters.
It should be noted that the application of effective water measurements and the institution of measures such as volumetric invoicing and volumetric allocations often come up against strong opposition from farmers, who then destroy or bypass the measuring devices. Such changes in the rules governing irrigation water deliveries cannot be instituted overnight. Serious preparation is required to win the users' agreement and, as the case may be, to revise existing water laws and water rights according to the new practices.
We hope to have made it clear that, in the continuous chain from the water source to the crop-soil system, one of the major obstacles to effective implementation of crop-based and water-saving irrigation scheduling is the inability of most conveyance and delivery systems to deliver water at the farm gates with the reliability and flexibility required. Another problem is the lack of convenient flow-rate measurement devices at most delivery points, which prevents the institution of volumetric water allocations, a measure recommended for efficient, water-saving on-farm irrigation scheduling.
It appears from the papers presented, as well as from our own experience, that a number of concepts, methods and equipment types for improving the reliability and flexibility of system operation and for proper delivery measurements do exist, but have not yet been extensively applied. For example, inherent reliability and flexibility of a delivery system can be improved through a proper design or re-design including some modernization and possibly some kind of automation. From the more general point of view of water management over a whole project, integral models are undoubtedly useful for project planning, design and evaluation, but they seem to be too cumbersome tools for day-to-day system operation.
On the other hand, effective implementation of the planned operational procedures, including the delivery schedule is subject to the application of non-physical measures, that have been explained and discussed under Theme VI of the Workshop.
In fact, besides administrative and financial constraints, the main reason for the slow progress in improvements in this field is a lack of knowledge and understanding of available techniques, combined with an ill-founded, though natural, resistance to change. Faced with this situation, which prevails in most developing countries, especially those with extensive irrigation, some authors recommend sticking to the simplest operational and scheduling procedures. This realistic approach can be accepted for the short term. However, along with other authors, we are convinced that the competition for water and the necessities of increased food production and crop diversification will make it essential, sooner or later, to improve the flexibility of irrigation schedules, delivery schedules and hence the operation of the delivery systems, whatever the country.
However, the farmers in countries with extensive irrigation will never be in a position to support the cost of the physical and non-physical measures required to effectively turn an existing rigid, equity-based system into a more flexible, crop-based one. In the end, this cost can only be covered by subsidies.
It is then essential that all people who are conscious of the need for efficient water management and possess some pieces of relevant know-how in this field make every effort to disseminate knowledge, improve education and training at all levels, transfer technology, incite decision-makers to changes, involve the farmers in the decision process and urge the funding agencies and governments to set up the financial means required.
Ankum, P. 1996. Selection of operation methods in canal delivery systems. In: Irrigation Scheduling: From Theory to Practice, Proceedings ICID/FAO Workshop, Sept. 1995, Rome. Water Report No. 8, FAO, Rome.
Bandaragoda, D.J. 1996. Institutional conditions for effective water delivery and irrigation scheduling in large gravity systems: evidence from Pakistan. In: Irrigation Scheduling: From Theory to Practice, Proceedings ICID/FAO Workshop, Sept. 1995, Rome. Water Report No. 8, FAO, Rome.
Burt, C.M. 1996. Essential water delivery policies for modem on-farm irrigation management. In: Irrigation Scheduling: From Theory to Practice, Proceedings ICID/FAO Workshop, Sept. 1995, Rome. Water Report No. 8, FAO, Rome.
CEMAGREF (ed.) 1992. Application of Mathematical Modelling for the Improvement of Canal Operation. Proceedings of a 1992 IIMI-CEMAGREF International Workshop, CEMAGREF, Montpellier, France.
D'Urso, G., Menenti, M., and Santini, A. 1996. Remote sensing and simulation modelling for on-demand irrigation systems management. In: Irrigation Scheduling: From Theory to Practice, Proceedings ICID/FAO Workshop, Sept. 1995, Rome. Water Report No. 8, FAO, Rome.
Goussard, J. 1993. Automation of Canal Irrigation Systems and L'Automatisation des Réseaux d'Irrigation en Canaux. ICID, New Delhi.
Horst, L. 1996. The discrepancy between irrigation scheduling and actual water distribution. In: Irrigation Scheduling: From Theory to Practice, Proceedings ICID/FAO Workshop, Sept. 1995, Rome. Water Report No. 8, FAO, Rome.
Makin, I.W. and Cornish, G.A. 1996. Irrigation scheduling at system level: an analysis of practical applications of the INCA software. In: Irrigation Scheduling: From Theory to Practice, Proceedings ICID/FAO Workshop, Sept. 1995, Rome. Water Report No. 8, FAO, Rome.
Mangano, G.V. 1996. Applicability of irrigation scheduling in developing countries. In: Irrigation Scheduling: From Theory to Practice, Proceedings ICID/FAO Workshop, Sept. 1995, Rome. Water Report No. 8, FAO, Rome.
Montañés, J.L. 1995. Mathematical models as a way to modernize the scheduling and delivery of water from irrigation canal networks. Paper submitted to ICID/FAO Workshop, Sept. 1995, Rome. Irrigation Scheduling: From Theory to Practice. FAO, Rome.
Morábito, J.A., Fornero, L., Emili, L. and Thomé, R. 1995. SIMIS & an integrated administration of an irrigation project - Case study: Matriz Gil secondary canal, Mendoza, Argentina. Paper submitted to ICID/FAO Workshop, Sept. 1995, Rome. Irrigation Scheduling: From Theory to Practice. FAO, Rome.
Plusquellec, H., Burt, C. and Wolter, H.W. 1994. Modem Water Control in Irrigation, Concepts, Issues and Applications. World Bank Technical Paper No. 246, Irrigation and Drainage Series, World Bank, Washington DC.
Zimbelman, D.D. (ed.) 1987. Planning, Operation, Rehabilitation and Automation of Irrigation Water Delivery Systems. Proceedings of a 1987 Symposium sponsored by the Irrigation and Drainage Division of the American Society of Civil Engineers and held in Portland, OR, ASCE, New York, USA.