SM

IWMI is a Future Harvest Centersupported by the CGIAR

Postal Address:P O Box 2075ColomboSri Lanka

Location:127, Sunil MawathaPelawattaBattaramullaSri Lanka

Tel:+94-1-787404

Fax:+94-1-786854

E-mail:iwmi@cgiar.org

Website:http://www.iwmi.org

ISSN 1026-0862ISBN 92-9090-471-2

Water ManagementI n t e r n a t i o n a l

I n s t i t u t e

SM

IWMI is a Future Harvest Centersupported by the CGIAR

Fundamentals of

Smallholder Irrigation:

The Structured System Concept

RESEARCH

58

Water ManagementI n t e r n a t i o n a l

I n s t i t u t e

B. Albinson and C. J. Perry

R E P O R T

Research Reports

IWMI’s mission is to improve water and land resources management for food,livelihoods and nature. In serving this mission, IWMI concentrates on the integrationof policies, technologies and management systems to achieve workable solutions toreal problems—practical, relevant results in the field of irrigation and water and landresources.

The publications in this series cover a wide range of subjects—from computermodeling to experience with water user associations—and vary in content fromdirectly applicable research to more basic studies, on which applied work ultimatelydepends. Some research reports are narrowly focused, analytical and detailedempirical studies; others are wide-ranging and synthetic overviews of genericproblems.

Although most of the reports are published by IWMI staff and their collaborators,we welcome contributions from others. Each report is reviewed internally by IWMI’sown staff and Fellows, and by external reviewers. The reports are published anddistributed both in hard copy and electronically (www.iwmi.org) and where possible alldata and analyses will be available as separate downloadable files. Reports may becopied freely and cited with due acknowledgment.

Research Report 58

Fundamentals of Smallholder Irrigation:The Structured System Concept

International Water Management InstituteP O Box 2075, Colombo, Sri Lanka

B. AlbinsonandC. J. Perry

IWMI receives its principal funding from 58 governments, private foundations, andinternational and regional organizations known as the Consultative Group onInternational Agricultural Research (CGIAR). Support is also given by the Governmentsof Ghana, Pakistan, South Africa, Sri Lanka and Thailand.

Albinson B.; and C. J. Perry. 2002. Fundamentals of Smallholder Irrigation: The StructuredSystem Concept. Research Report 58. Colombo, Sri Lanka: International Water ManagementInstitute.

/ Irrigation systems / small-scale systems / irrigation design / farm size / water management/ water delivery / water distribution / irrigation scheduling / water scarcity / productivity / watersupply /

ISBN 92-9090-471-2

ISSN 1026-0862

Copyright © 2002, by IWMI. All rights reserved.

Please direct inquiries and comments to: iwmi-research-news@cgiar.org

The authors: B. Albinson is an independent irrigation consultant and C. J. Perry is aResearch Fellow of the International Water Management Institute (IWMI), Colombo, Sri Lanka.

This paper is dedicated to the memory of Leslie Shanan,whose earlier work inspired many of the ideas it contains.

iiiiii

Contents

Summary v

Introduction 1

Background 2

What is a Structured System ? 3

The Advantages of Structured Systems 5

Productivity of Land and Water 5

Applicability of Structured Systems 7

The Design of Structured Systems 7

Determination of System Duty 8

Determination of the Service Area and Structured Level 8

Determination of Appropriate Modular Watercourse Size Range 10

The Special Case of Paddy 10

Conclusions 11

Technical Annex 13

Literature Cited 21

iv

v

Summary

Smallholder irrigation systems—where farm sizesgenerally range from a fraction of a hectare to 10hectares—pose special management problems,especially where the water available for irrigation isfrequently less than the demand. The intensity ofsystem adjustments required to meet individualfarmer demands, and the administrative complexityof measuring and accounting water deliveries havegenerally proven excessive when attempting tomeet “on demand” schedules, resulting in chaos(often characterized by illegal tampering withinfrastructure, and vast differences of water useintensity at different locations in the system).

The alternative—provision of a simple service,based on proportional sharing of available supplieson the basis of landholdings—has been resilient formany years over vast areas.

The approach is based on a clear delineationbetween the part of the irrigation system that isactively managed (at various flow rates and waterlevels) and the part of the system that operateseither at full supply level (with proportionaldivision of water down to the level at whichfarmers rotate among their individual farms), or iscompletely shut.

This operational design is known as a“structured” system, and has well-definedhydraulic characteristics, simplifying operationand management, in turn allowing a clearerdefinition of water entitlements and theresponsibilities of agency staff and farmers.

The approach is particularly suited to areaswhere water is scarce and discipline is neededto ration water among users. An additionalbenefit, which has been demonstrated inmodeling studies using a well–proven modelrelating to water and yield, is that theproductivity of water (which is more importantthan the more traditional productivity of landwhen water is scarce) is substantially increasedwhen deficit irrigation is practiced—a widelyobserved and predictable response to rationedwater supplies.

Structured systems are most suited wherewater is scarce, clear definition of waterentitlements is needed, management capacity islimited, and investment resources are limited.

The approach to determining critical aspectsof a structured system design is described in thisreport.

1

Introduction

This report describes what is meant by, and theprinciples involved in designing and operating, acorrectly “structured” system. We believe it isimportant to provide some background as to whythe proposed approach has particular merit incertain circumstances, and what thesecircumstances are. On this basis, we hope thatdesigners and managers can decide whether theirparticular needs will be well served by a structuredsystem.

In principle, it seems a simple matter to takea large channel of flowing water and divide itsuccessively into smaller channels, eventually tosupply individual farmers with the amount of waterneeded, as and when required—irrigation “ondemand.” However, supplying irrigation on demandrequires a system that is capable of upwardcommunication of information and downwarddelivery of water—an interactive managementprocess. There is strong evidence that thetechnical, political, and social environments inmost developing countries create a need for anon-demand service beyond the capability of manyof the responsible institutions (World Bank 1991).

The increasingly prevalent problem of waterscarcity and competition greatly exacerbates theinherent complexities of providing tens ofthousands of farmers with customized individualschedules—and this is the situation where the“structured” approach has particular strengths.

The approach proposed here is not entirelynovel (Shanan 1992) and is essentially in place in

Fundamentals of Smallholder Irrigation:The Structured System Concept

B. Albinson and C. J. Perry

the warabandi (Malhotra 1982) areas ofnorthwest India and Pakistan. Nevertheless, asmore modern infrastructure, communications,and management approaches have beendeveloped since the “structured” approach wasfirst implemented more than 100 years ago,many believe that more modern approaches canbetter serve today’s needs. To that extent,however, there is no full agreement that thestructured approach should be widely followedfor the design of new projects and therehabilitation of old projects. We hope thematerial in this report will help to resolve the (sofar) endless arguments that have arisen overthe subject—though even advocates of “ondemand” irrigation concede that the past severaldecades of effort have failed to produce asustainable success in any smallholder system(Davids 2001).

Additional background reading on the case forstructured systems includes Malhotra (1982),Shanan (1992), Jurriens et al. (1996), and Horst(1998). The case for more flexible deliverysystems is well summarized by Burt andPlusquellec (1990) and Plusquellec et al. (1994).Merriam (1992) went even further, recommendingthat both the timing and quantity of waterdeliveries should be unrestricted, in anticipation ofa decline in demand as confidence in systemreliability improved. Success, however, wastransitory, and the system proved incompatiblewith scarcity of water.

2

Background

The modern era of smallholder irrigation thatbegan in the nineteenth century continues to thisday. Until the mid-twentieth century the majority ofthe development was under the aegis of variouscolonial powers. Water rights of individual farmerswere not an issue—in Asia, the colonial powers’interests were in achieving maximum output(which generally coincided with maximum droughtinsurance) in order to enable the autocraticadministration to enforce equitable distribution ofwater—although it is interesting to note that theNorth West India Canal and Drainage Act (1873)did provide for some fundamental water rights.Typically, the defined responsibilities of agenciesand farmers focused on who constructed, paid for,and maintained which parts of the system.

The past 50 years have seen two fundamentalchanges in the smallholder milieu: thestrengthening of at least partly democratic forces,and the ever increasing excess of demand overavailable supplies of water. The first makesessential the entrenchment of individual rights—and a transparent mechanism for meeting thoserights. The second makes essential the rationalplanning of the resource allocation.

Those arguing for more modern managementsystems have pointed to the need to increaseproduction to meet the needs of a growingpopulation, and to improve farm incomes througha service that will meet the needs of more watersensitive, higher value crops such as fruits andvegetables.

Water has become a scarce commodity inmany areas for one or both of the followingreasons: the progressive expansion of irrigation

and other water demands within basins so thatsupply and demand curves have intersected; andthe intensification of demand within projects,which has put farmers into a situation ofcompetition rather than cooperation in local watermanagement. The familiar “head versus tail”problems have developed, along with far moreinterference in system management, ranging fromattempts to influence agency officials tointerventions to steal water and even destructionof control structures.

Very often the background institutional trendhas been a progressive deterioration in theauthority of operating agencies. The democraticforces in newly independent countries havereduced the power of the bureaucracies of colonialrule, strengthening the power of the local people—and in parallel have degraded both the officiallysanctioned rewards and the social authority of thebureaucracies. In turn, the quality of the agencyemployees attracted to work in this lesscompetitive (Kikuchi et al. 2000) and lessrespected public sector system has declined.

A common symptom of failure from the abovecauses is the destruction of system infrastructureby the farmers. This is not just vandalism, butcan also be an expression of frustration by thefarmers that the systems have been unable tofulfill the promised levels of service (Horst 1998).Typical attempts at demand operation disintegratein the complexities of technology management;the farmers have no means of verifying fairness ofdelivery and there is no clarity or transparency inthe operation—an invitation to corrupt practices(Wade 1982).

3

What is a Structured System?

A structured system includes a regulated uppercanal network feeding groups of Service Areas.Within the Service Areas (SAs) the flow isdistributed proportionally to individualwatercourses. Flow in the upper regulated portionis variable and canals usually run at partialcapacity. Within the SA the canals always runapproximately at full design discharge or arecompletely closed.

A key operational characteristic of a structuredsystem is the “structured level”—the interfacebetween the regulated variable flow of the upperpart and the proportional full flow of SAs. Theimportance of this division is founded in thefundamentals of open channel hydraulics.Regulation of variable open channel flow whilemaintaining full supply command requires constantmanagement intervention, whereas full flowingproportional systems can be designed to distributethe incoming flow without managementintervention.

In present common practice, the regulatedupper part and the proportional full flow SAs areresponsibilities of the system managers, and theflow within the tertiary areas is farmer managed.However, the inherent simplicity of the proportionalflow management within the SAs makes it anideal subject for the next segment of farmermanagement.

Location Management Hydraulics Responsibility

Main system Active Regulated Irrigation Agency

Interface Active Gate

Service Areas Passive Proportional Irrigation Agency

Possible take over by farmers

Interface Passive Division structure

Watercourse Area Active Distributed Farmers

TABLE 1.Hydraulic characteristics of the various levels of the structured system facilities.

Table 1 summarizes the hydrauliccharacteristics of the various levels of thestructured system facilities.

The complexity of management is directlyrelated to the structured level. Systems can bestructured at any level from the dam or barragedown to the minor level. An irrigation system can,therefore, be broadly specified in terms of spatialand temporal resolution. The spatial resolution, orservice area, is the smallest unit that can beindividually served, and the temporal resolutioncan be considered as the shortest period ofirrigation delivery.

The more detailed the spatial and temporalresolutions, the greater the system’s flexibility butthe more difficult to manage. The resolutionsshould, therefore, be set at levels that meet thedeclared service to the farmer, and that are withinthe capability of the owning and operatingagencies to plan, design, construct and manage.

The structured concept provides a means ofclassifying aspects of physical design and operationof irrigation facilities and allows for an unambiguousdefinition of proposed and existing standards. Thereis a common language and framework for describingsystems. Having a common language is important;the debate about the appropriateness, relevance andfeasibility of various operational plans andassociated designs suffers greatly when the

4

proposal is not set out so that the responsibilities,infrastructure and rules can be seen in their entirety,and interrelationships are defined.

The essential merit of the structuredframework is that it forces an orderly definition ofthe various levels in the system, and allowsexamination of whether the whole package issound, internally consistent and implementable.This, in turn, defines the legal framework requiredto support disciplined operation by all parties.

The approach provides for the sharing of wateramong the Service Areas in accordance withavailability, and predefined rules—most commonlyuniform amounts of water are assigned perhectare of command.

The schedule of deliveries is determined bytwo factors: first, the total quantity of wateravailable (which may be precisely known, if theproject is served from storage, or estimated if in-season river flows are significant); and second,the expected general schedule of demand—basedon some typical, overall cropping plan. The firstfactor determines the number of total days in theseason that the system can be run at full supplylevel; the second determines how those “fullsupply days” should be scheduled over the cropseason.

The operators do not undertake obligationsbeyond following the proposed schedule andensuring that the proper discharges are supplied tothe head of the service area and, then,proportionally to the watercourses (within whichthe distribution of water is the responsibility of thefarmers) during the designated irrigation periods. Inpractice, the schedule will be adjusted in the lightof actual water availability and in response torainfall.

When supplies fall short of expectations, theshortage is shared by rotating the supply channels

in accordance with predefined priorities that ensureany shortfall in the course of the irrigation seasonis equally shared in all Service Areas. Mostimportantly, canals are run full if they are run atall, because running canals at partial dischargewill always result in the same distribution ofscarcity—usually the head-end service areas willtake their share, while the tail enders lose out.1

The simpler set of obligations placed on theoperators—to meet a defined uniform scheduleover large areas—combined with a physicalinfrastructure that is designed to distribute thescheduled supplies in proportion to the land area,result in virtually automatic operation of thesystem. This automation further allows less scopefor undetected farmer intervention and forfavoritism on the part of the system operators.

Individual farmers are free to grow whatevercrops they wish and are aware of the schedulewhen irrigation will be provided. Because of thesimilarity of crop water requirements for a widerange of crops (most crops except rice, bananasand sugarcane have rather similar waterrequirements), the farmer’s choice is notsignificantly restricted. From the farmer’sperspective, the main difference between demandirrigation and supply irrigation is that supply-basedsystems require farmers to select and matchcrops to a predefined water supply—in returnoffering a firm guarantee of the overall pattern ofdeliveries.

For the managers, in supply-based systems,the calculated crop water requirements of thedesign cropping plan are a planning tool to assistoverall scheduling, not a management tool to meetlocal demand. Once the schedule is fixed, noattempt is made to force the farmers to grow anyparticular crop. The entire effort of management isdirected towards fulfilling the promised schedule.

1Interestingly, when the system was first introduced in northwest India, farmers preferred to be at the tail ends. Discipline was severeand no one would interfere with the flow in the larger channels, so that all low flows and water left after a turn was complete woulddrain to the tails, who thus got a better service!

5

The structured approach to system designaddresses several of the legal, technical, andorganizational problems discussed above:

• Water entitlement is transparently and simplydefined.

• Operationally, the system clearly demarcatestwo zones of management—one theresponsibility of the agency, and the other theresponsibility of the farmers.

• The operational rules are clear and controlsystems are simple, so that the demands onoperating staff are reasonable and transparent.

The approach is well adapted toconditions of water scarcity because water isallocated based on a proportional sharing ofthe available supply, rather than being anattempt to match demand with the availablewater. As will be clarified later, this does notmean that the schedule of supply isformulated without reference to demand, butrather that the prescribed scheduleapproximates the demand of an aggregatedcropping pattern, and the feasible croppingintensity is a matter for the farmers todecide, based on experience and generaladvice the agency can provide in advanceabout overall seasonal supply.

The Advantages of Structured Systems

Productivity of Land and Water

As a final element in justifying the structuredapproach, we address the issue of productivity—which would seem a likely casualty in a systemwhere scheduling and operations are simpler,management more rigid, and infrastructure lesssophisticated.

We note that as water becomes scarce,system designers and managers must focus noton crop yield per hectare, but rather on theproductivity of water, or yield per cubic meter. Formany systems this is a completely newobjective—the original designs were based onassumptions about the cropping intensities thatfarmers could achieve under the prevalentconditions of labour availability per unit of land,degree of farm mechanization, and profitability ofcrops. Today, farms have fragmented, laboravailable per hectare has often increased—andoften been supplemented by mechanicalequipment, and crop yields per hectare haveincreased sharply. All these factors point to more

intensive agriculture and higher demand forinputs—including water—per unit of land.

Recent research indicates that:

• returns per cubic meter of water for commonfield crops are little affected by the level ofsophistication of the irrigation service, and

• rationing water to levels below that requiredfor maximum production per unit of landsharply increases water productivity.

Sarwar et al. (2001) have applied the SWAPmodel calibrated for Pakistan to assess theimpact on local drainage requirements of the levelof water availability and different managementstrategies. As part of this study, yield indicatorswere generated in order to assess the productionimpacts of meeting drainage objectives. Theseresults can be reinterpreted to provide indicators

6

of the productivity of land and the productivityof water for different management scenarios,where water is abundant and where water isscarce.

Sarwar (2000) considered threemanagement scenarios: the first providedirrigation “on demand” in terms of timing andquantity; the second provided irrigation with alimited amount of flexibility in timing but fixedquantities; the third provided irrigation accordingto a fixed time and fixed quantity schedule. Theresponse of common field crops—wheat andcotton—was simulated over a 15-year period ofactual rainfall conditions. Irrigation scheduleswere optimised for conditions of excess water,where the objective was to obtain maximumyield per hectare, and for conditions of waterscarcity, where the objective was to maximisereturns to water.

The results show that:

• where water is plentiful, the maximumreturns to land were achieved with a fullyresponsive, on-demand system of operation,

• where water is scarce, returns to water were30-50 percent higher when water wasrationed rather than supplied to meet the fullcrop requirement (whether with fixed orflexible schedules), and

• returns to water for a fixed schedule werewithin 3 percent of returns for demand-oriented schedules.

The importance of the last conclusion is allthe more significant because the initial cost ofinfrastructure and ongoing management costsfor a structured system (which is designed to

provide the simple, uniform, rationed service) arefar less than the cost of a system providing on-demand, differentiated services to very smallfarms.

In summary:

• rationed water substantially increases theproductivity of water by encouraging farmersto follow a strategy of limited under-irrigation,and

• the productivity of water, where rationing is inplace, is almost identical for both simple fixedschedules and for complex “on demand”management.

These results at first glance are surprising:Why do returns to water increase as crops areunder-irrigated? And why is the productive benefitof responding to crop demand so small?

On the first point, a number of authors andanalyses (Vaux and Pruitt, Hargreaves, Jensen,FAO) have identified diminishing returns to wateras full evapotranspirative demand is met. As withmost production relationships, the initial returnsare large, but the returns to increased resourceuse become progressively smaller.

The ability of the crop to perform as well witha rather unresponsive irrigation schedule is relatedto the role of the soil matrix as a store of water,carrying over unutilized excess supplies from oneweek to the next. This stored water correspondsto a financial “bank balance,” where income doesnot have to equal expenditure every day, providedthe overall balance over a week or month ispositive.

We now elaborate the details of how fixedschedules are established and then set out thetechnical approach to system design.

7

The background has set out the scenarios wherethe introduction of a structured system is likely tobe most relevant and most suitable. Theparameters to consider are:

• Availability of water—structured systems areparticularly appropriate to conditions ofcompetition and scarcity.

• Need to define water entitlements—wherewater entitlements are undefined or disputed,a structured system provides a transparentand systematic approach to allocation.

• Management capacity—structured systemsprovide clear and limited demands on systemoperators, and well-defined responsibilities for

the farmers. This makes structured systemsapplicable where management capacity islimited.

• Investment resources—structured systems willgenerally be cheaper to construct and operatethan the alternatives.

The corollary of this list should also be stated.Structured systems may not be the best choicewhere:

• Water is plentiful.

• Water rights are well defined.

• Management capacity is strong.

• Investment resources are plentiful.

Applicability of Structured Systems

The Design of Structured Systems

In outlining the general approach to designing andoperating a structured system, we take the verycommon example of an irrigation project servingmany farmers. It has a regulated but not entirelycertain source of water, which is frequentlyinadequate to meet potential demand from thearea that can be commanded from the source—that is, land is plentiful compared to water. Asnoted above, if water is plentiful, then a demand-based system may be the best option. If, as iscommon in monsoon climates, water is plentiful inone season and scarce in another, the system willhave to cope with scarcity (often when water ismost valuable) and as such a structured systemshould be considered.

As in any well-designed system, theoperational plan (or irrigation service) and the

physical infrastructure are intrinsically linked. Wefirst describe the operational plan, though withunavoidable reference to aspects of theinfrastructure.

The key parameters in designing the systemare:

• The system duty, (l/s/ha) or rate of delivery ofwater per unit area.

• The structured level, or “break point” belowwhich the system is unregulated.

• The size of the service area within whichfarmers share water.

• The modular size of watercourses.

8

Determination of System Duty

To assist in describing the analysis, we set outthe data for a hypothetical case below:

• Average available irrigation water: 2,000 m3/haof command.

• Maximum crop demand: 10 mm/day(80 m3/day/ha of crop).

• Total crop demand: 400 mm (4,000 m3/ha ofcrop).

• Duration of crop season: 100 days.

• Soil capacity: 100 mm/m (1,000 m3/ha ofcrop).

From these data we see that in an averageyear, it will only be possible to crop 50 percent ofthe total command—water availability at 2,000m3/ha of command is half of demand per hectareof crop. How should this quantity of water bedelivered? As a rule of thumb, we aim to deliverwater on average every 10–15 days. If theirrigation frequency is much higher, it is difficult tomeet peak requirements and also difficult todeliver the extra water available in a better thanaverage year; if it is much lower, an individualirrigation may exceed available soil capacity, andin low availability years may result in excessivegaps between irrigations.

It is often convenient to deliver irrigation duringa period of 1 week, because this allows each farmerto have a fixed day and time for his turn.

In this case we aim to irrigate eight timesduring the season, and each application will be of250 m3/ha of command, or a depth of 50 mm onthe 50 percent of the command that is cropped.This corresponds to 7.1 mm/day, which we note issomewhat less than the peak demand of 10mm/day and considerably more than the averagedemand for the season of 4 mm/day. Themaximum demand is met by delivering water inadvance of the peak, and then each week duringthe peak, while the lower off-peak demand is metby delivering irrigation every other week, or evenevery third week in periods of very low demand.

We also note that since there are 15 weeks inthe season, and that we are on average irrigatingin only 8 of those weeks, we have the capacity todeliver almost twice the average quantity of waterif supplies are particularly good. If, on the otherhand, supplies are 30 percent below normal, therewill still be about five irrigation deliveries, whichshould be enough to allow scheduling for a smallerirrigated area.

As a hypothetical case, we have designed toprovide 250 m3 of water per hectare of commandin each of eight irrigations in an average year. Ifeach turn lasts 7 days, the system will deliverabout 35 m3/ha/day, or a system duty of0.4 l/s/ha.

Determination of the Service Area and Structured Level

These two parameters are related, the larger theService Area, the higher up the system will bestructured.

As defined above, the structured level is thepoint at which on-off control takes place—abovewhich the canals are regulated and below which

the system is unregulated, comprising a group ofwatercourses that receive the same, undifferentiatedservice.

Determination of the structured level is opento a variety of considerations; the lower thestructured level, the more gates are to be

9

operated. The higher the structured level, the lessthe ability of management to respond to localrainfall or other perturbations in the system—orindeed to consider specific irrigation schedules forspecific areas (cotton areas as distinct frommaize, for example). This decision reopens thewhole issue of responsiveness versus flexibilitythat underlies the “on-demand” versus “supply-based” discussion earlier, and system designersshould think carefully on the balance to be struck.

Some considerations that determine theappropriate structured level are: soils (sets thefrequency of irrigation), response to rainfall(depends on rainfall distribution), establishedrights, time to fill and empty rotational units, thecapability of the institution to operate andmaintain, and the sociological and politicalenvironment.

Soils: They have a major influence on the size ofthe Service Area. Light soils require more frequentirrigation than clay soils, and this affects therotational interval. A general rule is that the fillingtime for the service area should not be more than10 percent of the rotational interval.

Response to rainfall: An assessment should bemade of the spatial distribution of precipitation. Athunderstorm type distribution with highly localizeddistribution would require smaller Service Areas.

Time to fill and empty: The size of the ServiceArea has to be related to the operational plan. Toolarge a Service Area might take too long toimplement the planned rotational schedule.

Institutional capability: The institutional capabilitycan only be inferred by the performance inoperating and maintaining existing projects. As ageneral rule, a good indication of operation is theamount of damage caused by frustrated farmers.

Sociological and political environment: A decisionthat affects the Service Area is whether to allowdiffering water duties in different blocks.

Where water is constrained the availablesupplies should be distributed evenly over theCulturable Command Area (CCA). But in somewater constrained areas where paddy, sugarcane,bananas or other crops with higher waterrequirements are planned, because the soils arenot suitable for other crops, water allocations mayhave to be localized and different water dutiesallowed for different blocks. If the entire block canbe treated uniformly, then it can be structured asa single unit. If wide variations in supply arespecified between one part of a block and another,it must be broken down and a lower level ofstructuring adopted. This means the regulation hasto be taken to a lower level. There may beestablished rights, which have to be also takeninto account.

Farmers’ Rights and Obligation—Equality andEquity: There may be established rights to certaincrops, usually paddy. This may require theprovision of service areas designed to serve theestablished right. Each farmer must have clearlydefined rights and obligations. Equity is defined inthe rights, and equality is established when therules are correctly enforced. Equality in irrigationmeans all farmers are favored equally in periodsof abundance and suffer equally in times ofshortage. A correctly structured system achievesthese objectives because of the inherenttransparency of operation. Any farmer can afford acheap clock, and as long as the water levels areat the standard marks he can be sure that thesupply is fairly distributed.

Complete equity is difficult to achievebecause the sizes of landholdings vary widely.The best we can do in the medium-term is toensure that the supply is distributed proportionallyto the areas commanded. However, if thewatercourses are planned with modular areas andoperated with modular flows, there is no technicalor practical reason why smaller lots should not geta weighted time. The problem is that anyweighting to give advantage to small lots would bearbitrary and open to misuse and has not been

10

used in practice. However, in the warabandisystem of northwest India, weightings are indeedmade for distance from the irrigation outlet.

Common allocation to Service Areas desirable:Differential water allocations to adjacent ServiceAreas are quite possible to account for differentagro-climatic conditions such as cropping patterns,soils or topography. However, great caution isneeded to avoid the appearance of favoritism.

Different allocations presuppose that themanagement will be able to: a) calculate andschedule water requirements for each ServiceArea; and b) operate the system to provide thediffering water requirements in each block. Ingeneral, it is a good practice to have as littlevariation as possible between Service Areas.Whatever the distribution detail, it is essential thatthe farmers know in advance each season whenand how much water will be delivered.

A Service Area comprises one or morewatercourses. Selecting an appropriate range ofsizes for watercourse commands is essential: thesmaller the watercourse, the smaller the irrigationstream to the farmers within the watercourse(given the proportional relationship between flowand area), and vice versa.

While overall system duty determines thecapacity of the various channels, at the farm levelthe size of the irrigation stream is constrained bythe on-farm technology. Typically, a farm streamof 30 l/s is manageable with a range from 20-100l/s depending on soil characteristics, slopes, and

Determination of Appropriate Modular Watercourse Size Range

other factors such as labor availability. Theappropriate range will be a matter of determinationfor the local conditions (paddy will be at the higherend of the range), but continuing with thehypothetical example, we assume that 20-40 l/sshould be the target range for the example. Thisrange, together with the computed system duty of0.4 l/s indicates that the modular watercourse area(farm stream/duty) should range from 50–100hectares. A watercourse of 50 hectares will havean irrigation stream of 20 l/s, and a watercourse of100 hectares will have an irrigation stream of40 l/s.

The Special Case of Paddy

Rice, while not an aquatic plant, is both water-tolerant and water-demanding. Paddy cultivationexploits the water tolerance that gives the plant acompetitive edge over other grasses. Thepresence of standing water for much of thegrowing season also means that: (a) duties arehigh to provide for infiltration, evaporation andevapotranspiration, and (b) the presence of a

surface reservoir provides a buffer against irregularsupplies. This buffer implies that paddy is wellsuited to supply-based irrigation, the maintechnical problem being the design of proportionalsystems, which will handle fairly large differencesin water requirement at different stages of growth.

There are vast areas of smallholder paddy inthe deltas and coastal plains of the world that

11

Conclusions

We have explained how, and described thecircumstances in which, a well-planned supply-based system is capable of providing an irrigationservice that for all practical purposes is equivalentto a demand-oriented system. The record isreplete with irrigation systems that have beenplanned to provide a sophisticated, on-demandservice but have failed to perform as designed.

There is increasing evidence that the reliabilityand simplicity of operation is a compellingargument for farmers and system operators alike.A recent study of the Gediz Basin in Turkey(IWMI and GDRS 2000) has revealed how, inresponse to drought conditions during the late1980s, the State Hydraulics Organization (DSI)introduced a supply-based scheduling system.This replaced a crop-based system, whichpreviously determined canal discharge bytechnical estimates of water deliveries necessaryto avoid crop stress. The drought made itimpossible to continue this system of operationsas water availability was inadequate to meet cropwater requirements. Estimation of waterrequirements for each rotation was replaced with asimple calculation of the number of irrigationdeliveries, at maximum design discharge,available in the reservoir in April. Irrigation

could be well served by correctly structuredsystems. Where irrigation exists it is almostinvariably served by field to field irrigation.Attempts to provide a differentiated demand-basedservice to each farmer have been costly failures.Research (Tabal and Wickham 1981) has shown

that it is much more important to provide equitableflow in the distributary canals rather than focusingon individual on-farm water supply. Correctlystructured systems are a cost-effective and suitablemethod of fulfilling these criteria. This matter isdiscussed further in the Technical Annex.

releases are now made over a shorter period oftime than before. However, canals run full, or areclosed, and farmers are assured of equitableaccess to the available resources.

In another case, in Sri Lanka, the managersof the Kirindi Oya scheme have changedoperations, in consultation with farmers, fromcontinuous variable discharge to a program ofrotations at full supply discharge. This was toreduce water use and to increase equity of thedistribution (Sakthivadivel et al. 2001). During thedry season (yala) 1999, with a severe watershortage, average paddy yield approached 4 t/ha.This contrasts with the yala, 1992 season, whenwater availability was broadly similar, and averageyields were close to zero.

Irrigation systems designed to deliver aservice matched to crop water needs have, ingeneral, failed to perform as intended. Structureddesign with clear operational rules, results inirrigation infrastructure that can deliver reliableservices and also allow the farmers to determinetheir own optimum cropping systems. The inherentsimplicity of operation of hydraulic proportionaldistribution and the transparent monitoringsystems make the infrastructure ideal for transferto participatory management.

12

13

Technical Annex

This part of the paper provides guidance on thespecial requirements of structured systemdesign not found in the standard texts. It is not,however, intended as a comprehensive treatiseon canal hydraulics.

Hydraulic Regulation and Controls

Structured networks comprise a regulatedupstream portion supplying prescheduled flowsto downstream Service Areas, which distributethe flow proportionally to watercourses. Inalmost all cases, it proves convenient toregulate the main and sub-main canals and touse the distributary and minor canals asService Areas. Some notes on the details ofdesign, which are especially important tostructured systems, are given below. This is notan exhaustive treatment of canal design butdraws attention to the special features involvedwith structured systems, and should be read inconjunction with standard texts.

Canal Design: A canal has three designparameters, width, depth and slope. Only one ofthese combinations will produce a stable, non-scouring, non-silting or “in regime” earth canal.For earth canals, it is recommended that asection as close as possible to regime be used.The Lacey Regime f is best determined fromexisting similar systems on the same riverwhich appear to have achieved some level ofstability. For lined systems, it is only necessaryto check for the silting potential (Lacey f abovesilting level) as the scour on lining will notnormally be important.

2This diagram by L. Horst, 1998, is a prime example of a picture being better than a thousand words.

Regulation of the Conveyance SystemUpstream of the Service Area

Main and Branch Canals: This part of the networkmust be capable of a high degree of regulation, asit must be able to supply the distributary channelson demand. The demands will normally bepredictable during the dry season, but will besubject to random variations during the rainyseason as a result of the need for supplementaryirrigation. To provide for varying demands thereare several possible alternatives of differingdegrees of sophistication and cost.

System Response Theory: If Q is the flow in theparent canal and q in the daughter canal, then theresponse at individual nodes is measured by thevalue of F=(dq/q)/(dQ/Q) or the relative change offlow in daughter and parent canals without manualor automatic intervention. Figure 12 shows thenodal response with F=1, >1 and <1.

FIGURE 1.System Response Theory.

14

Regulated Hydraulics

With all nodes at F>1 variations at the parentcanal flow are concentrated in the upper reachesand either passed into the upper branches (if theyare open) or cause overtopping of the parentcanal’s upstream banks. Undershot regulation inthe parent canal combined with overshot regulationinto the daughter canal (the Indonesian Romijngate arrangement) yields very high F values.Automatic downstream control has some of thesecharacteristics—shortages are concentrated in theupstream reaches, but excess flows never occurbecause of the nature of the regulation.

With all nodes at F<1 variations in parent canalflow are concentrated in the lower reaches, alltransients are passed to the downstream ends ofthe parent canal, where they should be dealt withby providing additional freeboard and escapes.Long overshot weirs on the parent canal combinedwith undershot discharge modules to the daughtercanal provide this response. Automatic upstream

control yields the same results. For regulatedchannels, this arrangement is usually preferredbecause the unstable component appears at thedownstream end, where excess flows can be moreeasily passed to the drainage system. Duckbillweirs controlling the parent canal combined withan undershot daughter off-take is a simple cheapand effective way to provide low F values formore constant flow to the daughter canals. Theduckbill should be combined with undershot sluicecontrol, the sluice passing the base flow and silt,while the duckbill handles the variable component.In this way most of the benefits of the long weirare retained at less cost. The stability of flowthrough the off-take can be improved by using oneof several module designs available.

With all nodes at F=1 division at each node isproportional and variations at the head aretransmitted proportionally throughout the system.For proportional distribution within the structuredarea it is important (especially with earth canals)to provide as many critical depth control points as

Canal Control Off-take control F Comments

None, depth D Weir with crest 1 Usually results in impossibly

at 0.9 D narrow off-take weir

None, depth D Orifice and/or 1 Subject to earth canal stability

pipe set at 0.3 D uncertainty. Pipe entrance can be at

0.9 D to allow silt draw equity

Critical depth flume or weir, Orifice and/or 1 Recommended. Pipe entrance can be

depth D; should match pipe set at 0.3 D at 0.9D to allow silt draw equity

canal stage/discharge curve

Critical depth flume or weir, Weir with crest at D 1 Usually results in an impossibly

D; should match canal stage/ narrow off-take weir; makes

discharge curve a silt trap

TABLE 2.Control structure combinations for proportional distribution.

15

3“Flow Measuring Flumes for Open Channels”; M.G. Bos, J.Replogle, A. Clemmens; ASAE, 1991. The United States Bureau of Reclamation(USBR) in conjunction with the United States Department of Agriculture (USDA) and International Institute for Land Reclamation andImprovement (ILRI) has recently developed a full suite of Windows software for the design of Replogle flumes: WINFLUME November1999,4Adjustable Proportional Module, an orifice device which is initially adjustable with a moveable roof shoe but which is concreted in solidafter the system has been tested.5This is contrary to commonly accepted dogma. Although the characteristics of open channel or overshot flow are an H3/2 function andthe characteristics of an orifice are a Z1/2 function, nevertheless it is possible by appropriate design to achieve a useful range ofproportionality.

possible. It is the relative instability of earthcanals that most seriously destroys theproportionality of free off-takes. Critical depthstructures are completely stable. As shown intable 2, a pair of weirs at each off-take provideperfect proportionality, however, there are twoserious problems with this solution:

(a) the ratio of crest lengths in canal and off-takemay be very large, and

(b) weirs in canals form silt traps.

However, every possible opportunity should betaken to provide critical depth conditionsimmediately downstream of all off-takes. Every

foot crossing can be a box culvert drop providingcritical conditions. Replogle3 flumes can be usedwith minimum cost, and in addition to providingstability they give the chance to verify flows.

Off-takes: At the off-take the nodal performanceshould be as near F=1 as possible. A weir withcrest near the canal bed achieves this butunfortunately results in weirs which are too narrowto be practical. The off-take of choice will besome form of APM4 or pipe or a combination ofthe two. The APM is essentially an orifice typecontrol. Using correct design it is possible tomake an orifice or pipe (or a combination of thetwo) off-take proportional over a significant rangeof flows.5 Figure 2 illustrates how this is achieved.

FIGURE 2.Range of proportionality at off-take.

16

A computer program CFO6 is available toassist in the design of appropriate combinations ofcontrol structures to achieve proportional divisionover the required operational range. The exampleillustrated in figure 2 is for a typical distributarychannel with a design discharge Q = 2.25 m3/s,Bed width B = 1.0 m, Depth of flow D = 1.25 m,and canal slope S = 0.0005. The off-take pipe hasa diameter d = 0.25m, a length L = 10m with aninserted sleeve of diameter d = 0.15, and lengthl = 1m. The off-take pipe is set at 0.68 D with adesign flow of 0.075 m3/s or ratio = 30 There issubstantial range of close to proportional flow froma main canal discharge of Q = 0.9 to 2.25 m3/s.

Off-takes should be set to draw an equitableportion of the wash load carried by the parentcanal. This can be achieved by setting the inletsat a lower level than the outlets.

System Tuning: There are, inevitably, small errorsin design and construction that would, if leftuncorrected, result in an inequitable distribution.The practice in the warabandi areas of northwestIndia is to not finalize the setting of the off-takestructures, most notably the adjustment of theAPM off-take iron shoes, until after systemconstruction is completed and tested under fullflow conditions. When all the farmers are satisfiedthat the system is distributing equitably the shoesare permanently grouted in place. If pipe off-takesare used as the structure, a short sleeve can beused for adjustment and it can be permanentlygrouted in once an acceptable distribution isattained. The system is then very difficult tointerfere with.

Field Testing of Outlet Devices Recommended: Ifsome variation of the standard APM is employedusing local materials, for example, a pipe with

sleeves, it is strongly recommended that full scaletests of the proposed outlet characteristics beconducted. These can be quite easily and cheaplydone, using a small pump and some simple Vnotch weirs to measure flow rates. A stagedischarge curve is required to confirm theassumptions made during design.

Arrangements for monitoring operations: Animportant feature of the structured hydraulicregulation system is the simplicity of operations,either full supply or zero discharge. Monitoring thecorrect operation of the distribution system is alsoconsiderably simpler than with fully regulateddistribution to the service area—provided thedischarge entering the distributory canal is withinthe design limit (80-110% of FSQ) and thedistribution to each outlet should be within theacceptable range. This correct functioning of theentire canal can be confirmed by a singlemeasurement of water level at the tail of thecanal, at the final off-take or tail cluster. In NorthIndia, a simple tail flume design is used, in whichdesign water level coincides with a step in theflume walls—hence there is no requirement for awater level gauge.7 The stepped design of theopen flume structure also provides an additionaldegree of protection for safe passage of transientflows that may result from runoff entering thecanal upstream.

The Special Case of Paddy

Watercourse layout for paddy areas in systemsstructured at the distributary level

For cultivation of rice paddy, field to field irrigationrather than rotation through field channels will

6Canals, Flumes and Outlets (CFO), B. Albinson 1996.7The warabandi and its infrastructure. S.P. Malhotra. Central Board of Irrigation and Power. Publication No. 157. New Delhi. 1982.

17

probably remain the preferred method in manyareas. Tabal and Wickham (1981) showed thatproviding equitable and reliable flow at the head ofthe tertiary canal is more important than on-farmdistribution. A reasonable interpretation ofWickham’s data would lead to the conclusion thatthe number of fields crossed is not a significantfactor, although the distance to the nearest farmchannel should not be more than 300 m.

To satisfy these criteria, the watercourse areacan be about 40 hectares, divided into foursubareas. The peak duty is usually about 1.5 l/s/ha for pre-saturation and land preparation and 1.0l/s/ha for maintenance flows during crop growth.The peak flow will be 60 l/s and this can berotated round the four subareas in turn, starting atthe highest level. When preparation is complete,the flow can be reduced to 40 l/s dividedproportionally among the four subareas. The

distributary system will have to supply any flowbetween 1.5 and 1.0 l/s/ha proportionally to allwatercourses in the Service Area, with adequatecommand at all design flows. This is possibleusing correctly sized pipe outlets as illustrated infigure 2. The main canal will have to be regulatedto supply the distributary channels at the flowrates necessary to meet the operational plan.

In the case of systems where paddy is grownduring the wet season and upland crops areirrigated in the dry season, temporary fieldchannels are required for the upland crops. Thesetemporary field channels will have to beconstructed each year, because they will be lostduring the paddy season. With the layoutdescribed the average length of temporary fieldchannel will not be much more than 200 m. Theconceptual layout described is illustrated infigure 3.

FIGURE 3.

Conceptual layout of tertiary unit for paddy cultivation.

18

FIGURE 4.Typical design for a tertiary division box (Not to scale).

Tertiary Division Boxes

Division boxes in tertiary channels often have theirgates stolen. To combat this the design shownbelow makes a formal gate unnecessary. The keyfeature of this design is a relatively wide pad setat about 600 to the horizontal so that noaccurately formed gate is required, the shuttersimply rests on the side and floor pads. Any formof shutter will do including scrap wood, roughlyhewn log, woven palm leaves etc. The seal of theshutter can be improved by rubbing the shutteragainst the inclined pads with some coarse sandas a grinding paste. The box form provides

hydraulic stability. The dimensions and design ofthe box are not critical, except the throat widthshave to be proportional to the area served if thebox is required for proportional distribution. Theboxes can be precast and, for earth tertiaries,provided with grouted pitching transitionsupstream and downstream to suit the soilstability. The boxes are installed at convenientlocations to provide either purely flip-flop controlor both flip-flop and proportional distributiondepending on whether the system design isbased on rotation or continuous flow.

A proposed design, suitable for either linedor unlined watercourses is shown in figure 4.

19

FIGURE 5.Typical layout for low head division.

Low Head Loss Division

In the lower reaches of many irrigation systemsthe terrain is often very flat and there is very

little head available for critical depth divisionboxes. The design in figure 5 is based on thenotched log principle and consumes very littlehead.

20

21

Literature Cited

Albinson, B. 1996. Channels, Flumes and Orifice (CFO) computer program in Visual Basic, Unpublished.

Bos, M.G.; J. Replogle; A. Clemmens. 1991. Flow Measuring Flumes for Open Channels. ASAE.

Burt, C.M.; and H. Plusquellec. 1990. Water Delivery Control. In Managing Farm Irrigation Systems, eds. G. J. Hoffman,T. A. Howell and K. H. Solomon. USA: American Society of Agricultural Engineers. 373-423.

Davids, G. 2001. Personal communication September, 2001.

Horst L. 1998. The dilemmas of water division: Considerations and criteria for irrigation system design. Colombo, SriLanka: International Water Management Institute.

IWMI and GDRS, Turkey. 2000. Irrigation in the basin context: The Gediz Study. Colombo, Sri Lanka: InternationalWater Management Institute.

Jurriens, M.; P.P. Mollinga; and P. Wester. 1996. Scarcity by Design: Protective Irrigation in India and Pakistan.Wageningen, The Netherlands: International Institute of Land Reclamation and Improvement. (Liquid Gold 1996,paper 1).

Kikuchi M.; D. J. Merrey; and L. Dassenaike. 2000. Compensation of irrigation professionals: A comparative study ofthe Philippines, Sri Lanka and Pakistan. IWMI Discussion Paper. Colombo, Sri Lanka: International WaterManagement Institute.

Malhotra, S. P. 1982. The Warabandi System and its Infrastructure. New Delhi, India: Central Board of Irrigation andPower.

Merriam, J. L. 1992. The need on the farm for flexible water supply schedule. In Advances in planning, design andmanagement of irrigation systems as related to sustainable land use: Proceedings of the International Conferenceheld in Belgium, 14-17 September 1992.

Plusquellec, H. L.; C. M. Burt; and H. W. Wolter. 1994. Modern Water Control in Irrigation: Concepts, Issues andApplications. World Bank Technical Paper 246. Irrigation and Drainage Series. USA: World Bank.

Sakthivadivel, R.; R. Loeve; U. Amarasinghe; and M. Hemakumara. 2001. Water Scarcity and Managing SeasonalWater Crisis: Lessons from Kirindi Oya Project in Sri Lanka. Research Report (in review). Colombo: Sri Lanka.International Water Management Institute.

Sarwar, A.; W.G.M. Bastiaanssen; T.M. Boers; and J.C. Van Dam. 2001. Evaluating drainage design parameters forthe Fourth Drainage Project, Pakistan by using SWAP model: Part I—calibration. Irrigation and Drainage Systems,14(4):257-280.

Sarwar, A. 2000. A transient model approach to improve on-farm irrigation and drainage in Semi-arid zones. Ph.D.thesis, Wageningen University and Research Center.

Shanan, L. 1992. Planning and Management of Irrigation Systems in Developing Countries, Agricultural WaterManagement. 22(1 and 2).

Tabal, D. F.; and T.H. Wickham. 1981. Effects of Location and Water Shortages in an Irrigated Area. The Philippines:IRRI.

Wade, R. 1982. Systems of Administrative and Political Corruption: Canal Irrigation in South India, Journal ofDevelopment Studies.

World Bank. 1991. Experience with Rice Irrigation in Monsoon Asia, Washington D.C., USA: Operations EvaluationDepartment.

Research Reports

45. Pedaling out of Poverty: Social Impact of a Manual Irrigation Technology in SouthAsia. Tushaar Shah, M. Alam, M. Dinesh Kumar, R. K. Nagar, and Mahendra Singh,2000.

46. Using Remote Sensing Techniques to Evaluate Lining Efficacy of Watercourses.R. Sakthivadivel, Upali A. Amarasinghe, and S. Thiruvengadachari, 2000.

47. Alternate Wet Dry Irrigation in Rice Cultivation: Saving Water and Controlling Malariaand Japanese Encephalitis? Wim van der Hoek, R. Sakthivadivel, Melanie Renshaw,John B. Silver, Martin H. Birley, and Flemming Konradsen, 2000.

48. Predicting Water Availability in Irrigation Tank Cascade Systems: The CASCADEWater Balance Model. C. J. Jayatilaka, R. Sakthivadivel, Y. Shinogi, I. W. Makin, andP. Witharana, 2000.

49. Basin-Level Use and Productivity of Water: Examples from South Asia. David Molden,R. Sakthivadivel, and Zaigham Habib, 2000.

50. Modeling Scenarios for Water Allocation in the Gediz Basin, Turkey. Geoff Kite, PeterDroogers, Hammond Murray-Rust, and Koos de Voogt, 2001.

51. Valuing Water in Irrigated Agriculture and Reservoir Fisheries: A Multiple Use IrrigationSystem in Sri Lanka. Mary E. Renwick, 2001.

52. Charging for Irrigation Water: The Issues and Options, with a Cast Study from Iran. C.J. Perry, 2001.

53. Estimating Productivity of Water at Different Spatial Scales Using Simulation Modeling.Peter Droogers, and Geoff Kite, 2001.

54. Wells and Welfare in the Ganga Basin: Public Policy and Private Initiative in EasternUttar Pradesh, India. Tushaar Shah, 2001.

55.Water Scarcity and Managing Seasonal Water Crisis: Lessons from the Kirindi OyaProject in Sri Lanka. R. Sakthivadivel, Ronald Loeve, Upali A. Amarasinghe,and Manju Hemakumara, 2001.

56. Hydronomic Zones for Developing Basin Water Conservation Strategies. David J.Molden, Jack Keller, and R. Sakthivadivel, 2001.

57. Small Irrigation Tanks as a Source of Malaria Mosquito Vectors: A Study in North-Central Sri Lanka. Felix P. Amerasinghe, Flemming Konradsen, Wim van der Hoek,Priyanie H. Amerasinghe, J. P. W. Gunawardena, K. T. Fonseka and G. Jayasinghe,2001.

58. Fundamentals of Smallholder Irrigation: The Structured System Concept. B. Albinsonand C. J. Perry, 2002.

SM

IWMI is a Future Harvest Centersupported by the CGIAR

Postal Address:P O Box 2075ColomboSri Lanka

Location:127, Sunil MawathaPelawattaBattaramullaSri Lanka

Tel:+94-1-787404

Fax:+94-1-786854

E-mail:iwmi@cgiar.org

Website:http://www.iwmi.org

ISSN 1026-0862ISBN 92-9090-471-2

Water ManagementI n t e r n a t i o n a l

I n s t i t u t e

SM

IWMI is a Future Harvest Centersupported by the CGIAR

Fundamentals of

Smallholder Irrigation:

The Structured System Concept

RESEARCH

58

Water ManagementI n t e r n a t i o n a l

I n s t i t u t e

B. Albinson and C. J. Perry

R E P O R T