You are on page 1of 19

Managing sedimentation in spate irrigation schemes Philip Lawrence1

Independent consultant, (p.lawrence@sediments.plus.com)

Contents 1.0 2.0 Introduction Overview of sedimentation problems encountered in spate systems 2.1 2.2 2.! 2." !.0 Sediment transport in wadis Sediment mana ement in traditional systems Sediment mana ement in modernised systems # sediment mana ement strate y for spate irri ation systems

#pplication of models to aid t$e desi n of sediment mana ement structures in spate sc$emes !.1 Sediment e%cludin inta&es !.1.1 !.1.2 !.1.! !.1." !.2 '%amples of sediment e%cludin inta&es ($ysical models )$ree dimensional numerical modellin *asic inta&e model

Secondary sediment control + e%tractors and settlin basins !.2.1 !.2.2 ,esi n of ravel traps -ybrid e%tractor.flus$ed basin for lar e sc$emes

".0 /.0 0.0

Spate canal desi n met$ods (redictin future command levels Sediment mana ement options 0.1 0.2 1atc$ment conservation Sediment mana ement options for a ran e of sc$eme types

2.0 3.0

Sediment mana ement desi n software 4eferences

1.0

Introduction

'ffective mana ement of sedimentation is a &ey factor in improved spate irri ation systems. )$is paper describes t$e sediment problems encountered in spate sc$emes and describes met$ods t$at now can be used to estimate5ate t$e performance of sediment control structures. 6ost of t$e sediment control strate ies described in various papers in t$e first 7#O sponsored spate irri ation e%pert consultation, (7#O 1832), remain valid 9 t$e laws of p$ysics $ave not c$an ed over t$e last two decades. :$at $as c$an ed is t$e development of numerical models t$at now, wit$in limits, allow t$e performance of sediment mana ement structures to be investi ated wit$out t$e use of p$ysical models, w$ic$ enerally cannot provide ;uantitative predictions in t$is application. <umerical models provide ;uantitative predictions of t$e si=es and ;uantities of sediment diverted at spate irri ation off ta&es, t$e performance of sediment e%tractors and settlin basins, and t$e sediment transportin capacity and patterns of deposition wit$in canal networ&s. #doptin a ;uantitative approac$ to sediment mana ement based on numerical simulations $as many advanta es, t$e most important bein t$e capability to rapidly compare a ran e of sediment mana ement options, optimisin t$e performance of sediment control structures, and enablin costs of constructin sediment control structures to be compared wit$ a realistic estimate of t$e benefits. )$e ;ualification above arises from t$e ;uality of predictions provided by even t$e best of currently available sediment transport functions, w$ic$ underpin all sediment transport simulations, and t$e eneral lac& of reliable field data, particularly from spate systems, w$ic$ is needed to set up and verify numerical sediment transport models. 6odels can only be improved and developed, and sedimentation issues properly ;uantified, t$rou $ comparisons wit$ reliable field information. -owever w$en applied by en ineers wit$ an understandin of t$e models and t$e processes bein simulated, numerical sediment transport models can provide an invaluable aid to t$e desi n of spate irri ation diversions and water distribution systems. )$e first section of t$e paper summarises sedimentation problems encountered in spate sc$emes, and sets out a sediment mana ement strate y. 6odels t$at can be applied to aid t$e desi n of sediment control structures and canals are t$en described, wit$ some e%amples. )$e last section of t$e paper lists sediment mana ement options for sc$emes wit$ a ran e of c$aracteristics. 2.0 2.1 Overview of sedimentation problems encountered in spate systems Sediment transport in wadis

6orp$olo y and sediment transport processes in spate wadis are described in uidelines prepared in 200/2. In summary t$e main features of t$e sediment loads arrivin at typical spate diversion sites are>

:adi bed materials can ran e from boulders and cobbles to silts, wit$ t$e lower reac$es of wadis usually $avin sand beds. Sediments of all t$e si=es represented in a wadi bed are transported in t$e lar est floods. )otal load sediment concentrations risin to and e%ceedin 100,000 ppm, or 10 percent by wei $t, occur in floods in some wadis. Sediment concentrations up to / percent by wei $t in floods are common.

Improvin community spate irri ation, ( ?awrence and ,r. 7 van Steenber en. 4eport O, 1/" -4 :allin ford, @A. (1an be downloaded free of c$ar e from> $ttp>..eprints.$rwallin ford.co.u&.1!".

Sediment transport is dominated by t$e finer sediment fractions transported in suspension. )ypically t$e proportion of fine silts and clay in annual sediment load ran es between /0B and 80B. -i $ concentrations of fine sand are also transported in suspension. )$e suspended sand load is ;uite fine, enerally between 0.1 and 1 mm, w$en compared wit$ t$e parent bed material, as s$own in 7i ure 1 !. )$e proportion of t$e sediment load represented by lar est sediments, transported mostly by rollin and slidin alon a :adi bed, (bed load), is typically only / percent or less of t$e annual sediment load. Sediments in t$is si=e ran e, (coarse sand, ravel, cobbles, and in some cases small boulders), will, if diverted, settle and bloc& inta&es and canals.

Wadi Zabid Wadi Zabid 100 90 80 70 60 % Finer 50 40

Wadi Laba Wadi Laba

Suspended Sediment lar er t!an 0.06 mm

Bed Material 30 20 10 0 0.01

0.1

1 Size mm

10

100

1000

Figure 1 Bed material and suspended bed material (sand) load size grading for two Wadis #s only t$e very lar est floods are usually allowed to pass beyond spate irri ated areas most of t$e sediments transported to t$e first diversion point in a spate sc$eme are deposited wit$in t$e irri ated area. 1oarser sediments settle in t$e :adi c$annels and canals, and finer sediments are deposited on t$e fields, w$ere farmers welcome sedimentation as a source of fertility. Spate systems build up t$eir own soils, and older systems are c$aracterised by fine sediment deposits t$at are many metres deep. Sediment deposition rates on spate irri ated fields reported in t$e uidelines ran e between 1 to more t$an /0 mm .year. 2.2 Sediment management in traditional spare irrigation systems

)raditional spate systems $ave several features t$at t$at $elp control sedimentation>

Inta&es are was$ed away by lar e floods, reatly reducin t$e volume of water diverted to a canal durin periods of very $i $ flow in t$e wadi, w$en $i $ concentrations of coarse sediments are bein transported. )raditional canals are steep, and flow at muc$ $i $er velocities t$an would be allowed in conventionally desi ned irri ation c$annels. @sually all t$e flow in a canal is diverted fields at a sin le point, maintainin t$e full canal disc$ar e and sediment transportin capacity from t$e wadi inta&e to t$e field.
!

)$e si=e ran e of suspended bed material can be appro%imately estimated from t$e bed material si=e distribution. )$is facility is available in -4 :allin fordCs S-#41 software.

:$en command starts to be lost inta&es can easily be moved furt$er upstream to re ain command. Sediment management in modernised systems

2.3

:$en systems are modernised wit$ new permanent diversion structures muc$ lar er disc$ar es can be diverted from flood flows. 1onse;uently water carryin $i $ concentrations of coarse sediments is diverted to canals. In some systems modernised in t$e 1830Cs t$e canals were constructed wit$ ;uite low slopes, and a limited sediment transportin capacity. )$e combination of a lar e input of sand and lar er sediments and a limited canal sediment transportin capacity inevitably resulted in severe canal sedimentation problems. :$ile later systems were provided wit$ canals wit$ steeper slopes, canal de+siltin remains as a si nificant maintenance burden in many modernised sc$emes. Dated scour sluices are usually provided at inta&es. )$ey are intended to be operated to e%clude t$e coarse sediments t$at are transported alon and close to t$e wadi bed floods. 6anual operation of sluice ates in rapidly varyin spate flows, to follow idealised ate operation rules $as enerally proved difficult or impossible. #part from t$ese practical difficulties, t$e first priority of farmers is to divert as muc$ water as possible. )$ey may be e%tremely reluctant to open, (or allow a ency employed ate operators to open) sluice ates, e%cept durin t$e very lar est floods. In some lar er sc$emes flus$ed sedimentation basins are provided, w$ere t$e incomin water is slowed down, and lar er sediments are deposited in a lined basin. )$ese are t$en flus$ed bac& to t$e wadi w$en t$e basin is full. If water for flus$in is not available, or flus$in is re arded as bein too difficult to or anise, basins are e%cavated mec$anically. :$en properly desi ned t$ese structures function well at inta&es in perennial rivers, but t$ey $ave $ad a mi%ed trac& record w$en used in spate sc$emes. If twin sedimentation basins or a bypass c$annel are provided it is possible, at least in t$eory, to maintain canal supplies and flus$ basins durin t$e periods w$en t$e wadi flows e%ceed t$e disc$ar e needed for t$e canal. Eery slic& operational procedures, almost certainly reliant on electrically powered ates, are re;uired for t$is to be feasible. :$ile a flow duration curve may s$ow t$at substantial volumes of water are available w$en t$e wadi disc$ar e e%ceeds t$e canal demand, t$ese Fe%cessF flows occur durin numerous flood pea&s, at unpredictable times over a flood season. In many floods e%cess flow is only available for very s$ort periods, sometimes only a few minutes. )$us unless a very lar e flood $appens to occur w$en a basin needs flus$in it is necessary to utilise canal flows for sediment flus$in . )$is is always t$e case w$en a sin le sedimentation basin is provided. 7armers $ave stron obGections to w$at t$ey perceive as wasta e of water t$at could be diverted for irri ation, and t$ere are e%amples of inta&es provided wit$ sop$isticated sediment mana ement facilities t$at are rarely operated due to pressure from farmers, or ot$er powerful local interests, on ate operations staff. )$e result of t$ese operational difficulties is t$at substantial ;uantities of sand and lar er sediment are diverted to canals w$ere t$ey settle, reducin diversion and conveyance capacities. )$e need for fre;uent, often annual, removal of sediment deposits from t$e upstream reac$es of canals places a $eavy burden on farmers or sc$eme operators. 4isin command levels due to sediment deposition on t$e fields eventually results in some parts of upstream irri ated areas oin out of command. In sc$emes wit$ permanent inta&es weir crests and t$e sill levels of water control structures eventually $ave to be raised in e%pensive re$abilitation proGects.

2.4

sediment management strategy for spate irrigation systems

# sediment mana ement strate y for modernised spate systems can be summarised as>

Limit the diversion of coarser sediments


)$is re;uires e%cludin as muc$ as possible of t$e lar er sediments t$at will settle in and bloc& canals at t$e inta&e. Some form of secondary sediment control at t$e $ead of t$e main canal may be feasible in lar er systems to furt$er limit t$e concentrations of lar e sediments t$at are admitted to t$e canal system. #t simple inta&es wit$out a sediment e%cluder t$e only option is to limit t$e flows allowed to enter a canal system.

Transport fine sediments through canals to the fields )$is re;uires canals and water control structures to be desi ned to transport lar e concentrations of fine sediments. Making provision for the inevitable rise in command levels
)$is re;uires t$at inta&es canals and water control structures are desi ned to continue functionin wit$ t$e substantial rise in command levels t$at will occur over t$e desi n life of structures. Accept the need for canal de-silting and plan for it In spite of t$e measures listed above routine canal de+siltin will be necessary in some sc$emes. )$is s$ould be planned for by providin access to plant, and puttin in place sustainable fundin mec$anisms to support it use. Ensure that proposed sediment control measures and understood and accepted by farmers 'ffective well en ineered sediment control measures will only wor& if t$ey are perceived to be beneficial by farmers and are operated. 3.0 pplication of models to aid t!e design of sediment management structures in s pate sc!emes

Huite a wide ran e of numerical models t$at can be applied to aid t$e desi n of sediment control structures are now available, a encies and consultants in different countries will eac$ $ave t$ere own preferences. )$e software pac&a es described in t$is section were developed by -4 :allin ford wit$ fundin from t$e *ritis$ Dovernments ,epartment for International ,evelopment specifically for application to sediment mana ement in irri ation systems. 3.1 3.1.1 Sediment e"cluding inta#es $"amples of sediment e"cluding inta#es

)$e fi ure below s$ows a conventionally desi ned river inta&e used to divert water from a relatively steep seasonal ravel bed river in t$e ($ilippines, (left), and an inta&e desi ned for spate irri ation diversions, (ri $t).

%igure 2 &iversion structures wit! sediment e"cluders for perennial and spate irrigation inta#es )$e structure on t$e left diverts only a small proportion of t$e river flow durin t$e wet season. ,urin t$ese periods t$e sluice ates are left partially open, providin a stron sluicin flow sweepin t$e bed load past t$e canal ates. In t$e dry season, w$en most if not all of t$e river flows are diverted, t$e sluice ates are opened intermittently (still pond operation) to flus$ t$e sediment deposits t$at build up in t$e sluicin poc&et formed by t$e divide wall. )$e canal ates are closed durin sluicin . )$is type of perennial river inta&e is not suitable for spate irri ation, w$ere all t$e wadi flows are diverted to canals for most of t$e time t$at t$e wadi is flowin , t$e sediment loads are muc$ lar er, and still pond operation is not feasible. In t$e spate diversion structure s$own on t$e ri $t t$e inta&e and sluice ates are placed downstream from t$e weir. )$ere is no divide wall, w$ic$ would obstruct flows approac$in t$e inta&e from across t$e wadi. ()$e an le at w$ic$ flows a spate inta&e can vary durin , and between floods, and in flood recessions flows may approac$ t$e inta&e parallel to t$e weir.) Sluice ates are provided and supposed to be opened durin t$e s$ort periods of $i $ flow, w$en e%cess water is available for sluicin , and t$e $i $est concentrations of coarse sediment are transported. )$e curved c$annel provides some additional e%clusion of coarse sediments due to t$e bed load sweep effect t$at moves lar e bed load sediments towards t$e inside of t$e bed, and t$rou $ t$e sluiceway.

%igure 3 'ed load sweep at a c!annel bend ()s * flow at surface )b * flow at bed + )$e canal inta&e is ali ned at a s$allow an le to t$e flow direction in $i $ flows, so as to minimise t$e diversion of bed load." 3.1.2 ,!ysical models

($ysical models, usually wit$ a mobile bed, $ave often been used to aid t$e desi n of sediment e%cludin inta&es. )$e desi n of many of t$e lar er spate inta&es constructed since t$e 1830Cs were developed wit$ t$e aid of p$ysical model tests. 7or e%ample model tests were carried out on several curved c$annel e%cluders similar to t$at s$own in fi ure 2, (Smit$ 1832, and )osswell, 1838), and s$owed t$at in a p$ysical model curved c$annel e%cluders can e%clude 2/ B or more of (bed) sediments from a canal, w$en operated wit$ a flus$in disc$ar e of around !0 B of t$e canal disc$ar e. -owever usin a p$ysical model to ma&e ;uantitative predictions of sediment e%clusion is frau $t wit$ difficulty. )$e problem is principally one of scalin . If sediment si=e is scaled in proportion to t$e main model scale, t$en t$e material re;uired becomes so fine t$at it e%$ibits very different properties from t$e prototype. #rtifices suc$ as t$e use of oversi=e li $twei $t sediments or tiltin models $ave been used wit$ some success but Ialin (1821) s$owed t$at it is impossible to satisfy t$e p$ysical laws for scalin sediments w$en water is used as t$e model fluid. In models of a reasonable scale only t$e lar er sediments, movin as bed load can be represented. )$ese difficulties are demonstrated in fi ure ", w$ic$ is based on studies carried out by -ydraulic 4esearc$, :allin ford, @A, and includes comparisons of sediment e%cludin performance predicted in p$ysical model studies wit$ t$at later measured in t$e field. In t$e fi ure sediment e%cludin performance is e%pressed as a performance ratio ((4) defined as> (4 J 1+ sediment concentration enterin canal 1oncentration bein transported by t$e river

# performance ratio of 1.0 indicates complete sediment e%clusion, =ero indicate neutral performance, wit$ t$e inta&e neit$er reducin nor en$ancin sediment concentrations, w$ile a
"

)$is contradicts t$e advice iven in 7#OCs 2002 irri ation desi n manual, (fi ure !8 volume 2 module 2), w$ic$ recommends ri $t an led inta&es for silt laden rivers. ($ysical and numerical models and field e%perience all demonstrate t$at frontal inta&es divert t$e minimum of bed load to canals. )$e proportion of bed load in t$e diverted flow increases as t$e diversion an le is increased. )$e reason is t$at t$e lower momentum of flows near t$e bed of a c$annel ma&es t$em more easily divertible if t$e flow is turned. In spate inta&es t$e an le of diversion is of course only relevant durin $i $ flows, w$en water is passin t$rou $ t$e sluiceway or over t$e weir

ne ative ratio would indicate t$at t$e inta&e is wit$drawin a $i $er sediment concentration t$an t$e mean concentration in t$e river. 7i ure " clearly demonstrates t$e overestimation of sediment e%cludin performance in p$ysical models, w$ic$ is not always made clear in reports from modellin or anisations. It is stressed t$at t$is conclusion relates only to sediment e%clusion at inta&es, p$ysical models are a valuable tool t$at are still widely used for a lar e ran e of applications.

%igure 4 Observed at predicted performance of some sediment e"cluding inta#es . 3.1.3 -!ree dimensional numerical modelling

<umerical modellin of t$e flow and sediment movement in t$e vicinity of an inta&e $as t$e potential to ma&e ;uantitative predictions of its sediment e%cludin performance, wit$out problems in representin a wide ran e of rain si=es. # sediment e%cluder $as a lar e effect on t$e sediment rain si=e distribution carried in flows diverted to a canal, and t$is can be more important t$an its effect in reducin t$e overall sediment concentrations. #n inta&e wit$ a moderate overall e%clusion performance may e%clude all t$e coarser material, so t$at only fine material enters t$e canal. If t$e canal is able to transport t$is fine material, t$e inta&e will prevent w$at would ot$erwise be a severe sedimentation problem. Suc$ conditions can only be assessed wit$ a si=e+by+si=e prediction of sediment e%cludin performance. # second advanta e of numerical modellin is t$at it can be very muc$ c$eaper t$an commissionin a p$ysical model. -4 :allin ford developed a t$ree dimensional computational model specifically for studies of t$e performance of sediment e%cludin inta&es in 188/. (-4 :allin ford, 188/). )$e model represents t$ree dimensional effects, for e%ample t$e $elicoidal flow w$ic$ develops at a river bend or curved c$annel sediment e%cluder, and also ta&es account of> momentum, turbulence, bed eometry, bed friction, sediment settlin velocities, and of course t$e eometry of an inta&e and associated any sediment e%clusion features. 4esults from t$is model are compared wit$ field performance data in fi ure ". # reement between t$e model predictions and t$e sediment e%cludin performance measured in t$e field

is, (for sediment transport predictions), e%cellent. )$e model enables predictions of t$e variation of sediment e%cludin performance wit$ rain si=e, w$ic$ is essential if t$e impact of an e%cluder in on limitin canal sedimentation is to be assessed, see fi ure /.

%igure . Sediment e"clusion as a function of sediment si/e )$e settin up, calibration and operation of t$is type of model re;uires specialist numerical modellin and sediment transport e%pertise, and application of t$e model would only be relevant in lar e spate modernisation proGects, w$ere lar e sums are to be spent in improved infrastructure. 7or t$ese applications it provides desi ners wit$ a capability to rapidly assess and optimise t$e sediment e%cludin performance of a comple% inta&e. 3.1.4 'asic inta#e model 6ore basic inta&es are provided on smaller wadis. #n e%ample of a newly constructed inta&e in 'ritrea is s$own below as fi ure 0.

%igure 0 'asic inta#e in a sand bed river 'stimates of sediment e%cludin performance of t$is type of inta&e, wit$ a simple sluicin arran ement, can be made usin a simpler model, as described in -4 :allin ford, 2001. )$is model $as a flow component, and a sediment component. )$e flow component determines t$e ori in of t$e flow diverted to t$e canal from a cross section located upstream. #n envelope containin t$e diverted flow wit$in t$is cross section is predicted, based on an assumption t$at t$e flow re;uirin t$e least momentum c$an e to be diverted will enter t$e canal inta&e. )$e sediment component of t$e model computes t$e sediment si=es and concentrations transported in t$e diverted flow envelope, t$ereby predictin t$e sediment si=es and concentrations enterin t$e inta&e. #n e%ample of t$e output provided by t$is model for a simple inta&e wit$ a sluice is s$own in fi ure below>
Impact of Sluicing
1 0.9 0.8 0.7 0.6 PR 0.5 0.4 0.3 0.2 0.1 0 0 2 4 6 8 10 12 14 16 18 Eexcluder discharge m3/s

%igure 1 $ffect of sluicing disc!arge on predicted sediment e"cluding performance . )$is model can also be used to estimate e%clusion for t$e ran e of sediment si=es present in t$e c$annel bed upstream from t$e inta&e, t$us provin uidance on t$e sluicin disc$ar es needed to ma%imise e%clusion of selected (lar e) sediment fractions.
//

In t$e fi ure (4 is a performance ratio defined earlier in t$is section.

3.2

Secondary sediment control 2 e"tractors and settling basins

'ffective sediment e%clusion at an inta&e may be not be possible, or may not be snsufficient to e%clude all t$e lar er sediments t$at will settle in canals. 'ven at well desi ned inta&es t$ere is always t$e possibility t$at sluicin will not be carried out. Some form of secondary sediment control is often considered, particularly, in lar er sc$emes. )$e c$eapest and most tec$nically attractive option is a vorte% tube e%tractor. )$is e%tracts flow and sediments from near t$e canal bed in t$e canal $ead reac$, and pass it bac& to a river. )$e advanta e of t$is approac$ is t$at t$e si=e ran e of t$e sediments t$at are e%tracted can be controlled to some e%tent by adGustin t$e proportion of t$e canal flow, often around 10B, w$ic$ is e%tracted. -owever as t$ese devices re;uire water to be continuously KwastedF t$ey are not usually appropriate for spate sc$emes. (6odels used to desi n e%tractors are described in (-4 :allin ford, 2001). Settlin basins are a second met$od of secondary sediment control, t$at in perennial sc$emes enable sediment to be removed from t$e flows enterin a canal at t$e e%panse of only t$e few B of t$e incomin flow volume, w$ic$ is used for intermittent flus$in . 3.2.1 &esign of gravel traps

In spate sc$emes t$e obGective is to trap lar e sediments at t$e $ead of a main canal. )$e relatively small basins used to ac$ieve t$is are often called ravel traps. Several e%amples of sedimentation basins used in spate sc$emes are s$own in -4 :allin ford 200/. ()$e problems of flus$in ravel traps in spate sc$emes were discussed earlier.) )$e advanta e of a ravel trap over t$e alterative of removin sediment deposits from a canal $ead reac$ is t$at some sedimentation can occur wit$out limitin t$e flows t$at can be diverted to or conveyed by a canal, w$ic$ may be important w$ere t$e available $ead is limited. If sediments are removed and e%cavator rat$er t$an by flus$in t$en de+siltin is mostly focussed at t$e basin, rat$er t$an alon t$e canal system. Settlin basin desi n can be carried out usin -4 :allin fordCs ,ossbas software pac&a e. )$is includes two numerical models t$at simulate t$e performance of basins operatin in t$e deposition and sluicin modes. )$ese models provide several advanta es over earlier basin desi n met$ods, w$ic$ can only be used to predict t$e trappin efficiency of an idealised basin, as t$ey assume =ero sediment transportin capacity wit$in t$e settlin =one, and t$us do not predict t$e variations in trappin efficiency as t$e basin fills wit$ sediment. @sin t$e models a desi ner can ma&e several &ey predictions t$at assist in refinin and optimisin a settlin basin desi n. )$ese are>

)$e variation in sediment concentrations and rain si=es passin t$rou $ a basin and enterin t$e downstream canal networ& as t$e basin fills wit$ sediment. 'stimates of t$e fre;uency of sediment sluicin or de+siltin operations. :$en a basin is flus$ed, t$e time period re;uired to flus$ t$e basin and t$e volume of water needed for flus$in . )$e slope and cross section re;uired for t$e escape c$annel to convey sediment flus$ed from a basin to t$e river or disposal point.

)$e sediment trap efficiency for different sediment si=es varies widely as a basin fills wit$ sediment. :$en empty, and supplied wit$ a very low disc$ar e, a conventionally desi ned basin will trap considerable volumes of fine silts. )$ese ta&e up stora e in t$e basin, and are anyway usually valued by farmers. -owever basins for spate sc$emes can be desi ned to minimise trappin fine sediments, t$e results of numerical simulations s$owin t$at>

*asins s$ould be relatively narrow, so as to &eep t$e operatin velocities fairly $i $, wit$ sediment stora e obtained by increasin t$e len t$, rat$er t$an t$e widt$ or dept$ of t$e basin. (See for e%ample t$e basin desi ned for :adi 4ima in Iemen s$own in (7#O, 1832) ,eposition of some fine sediment is inevitable if a series of low flows carryin $i $ concentrations of fine sediments enter t$e basin, particularly w$en t$e basin is initially empty. )$is may occur ;uite often in spate sc$emes, w$en a series of small floods arrive at an inta&e. If it is considered necessary substantial reductions in t$e trap efficiency for fine sediments can be made if t$e tail water level in t$e basin is lowered for very low basin disc$ar es. )$is can obviously be ac$ieved by providin a ate to control downstream water levels, but ate operation can be problematic in spate flows. #not$er possibility is to provide a notc$ed weir at t$e basin e%it, so t$at tail water levels are substantially lowered w$en t$e basin disc$ar e is very low. 3ybrid e"tractor4settling basin for large sc!emes

3.2.2

It is possible to combine t$e benefits of vorte% tube sediment e%tractor, wit$ t$e muc$ lower wasta e of water needed for flus$in provided by a settlin basin, w$ic$ can be =ero if a basin is mec$anically e%cavated. # $ybrid system of t$is type was proposed by consultants for t$e wadi 6awr system in Iemen, but was reGected by t$e client and :orld *an& due to concerns of t$e sustainability of a Get pump t$at was to be used to evacuate t$e basin. -owever a $ybrid vorte% tube .settlin basin system was later constructed in a sc$eme supplied from a seasonal ravel bed river in t$e ($ilippines in 188/. )$e river transports very $i $ sediment concentrations, some derived from mine tailin s. In t$is system, desi ned usin t$e models described in (-4 :allin ford, 2001), a vorte% tube sediment e%tractor was located upstream from t$e first drop structure alon t$e main canal. It disc$ar es to a small flus$ed settlin basin, wit$ t$e settlin basin disc$ar e bein passed bac& to t$e canal system downstream from t$e drop. In t$is case a flus$ed basin was used, but in spate sc$emes a mec$anically e%cavated basin could be considered, allowin e%traction of coarse sediments from t$e canal wit$out any loss of water. )$e system proved to be e%tremely successful, t$e substantial reduction in de+siltin re;uirements provided by t$e e%tractor installation $altin , and t$en reversin , a lon term decline in t$e irri ation service area, and providin very lar e economic returns, (-4 :allin ford, 1880, 6erritt, 2002). Systems li&e t$is could be considered for lar e spate sc$emes, w$ere investment in infrastructure can be Gustified by a reatly reduced O and 6 costs.

%igure 5 )orte" tube used in combination wit! a settling basin 4.0 Spate canal design met!ods

1anals for perennial sc$emes are often desi ned usin ma%imum and minimum water velocities set by Kno scourin 9 no siltin F criteria, as described in, (7#O 2002). #s concentrations of sand diverted to canals is far lar er in spate sc$emes t$an is t$e case in perennial sc$emes, and canals are operated at a fraction of t$eir desi n capacity for most of t$e time, spate canals desi ned in t$is way rapidly silt up. K4e imeF desi n met$ods $ave been applied in spate sc$emes. )$ese are sets of empirical e;uations derived from observations of canal systems t$at are relatively stable, or Lin re imeL. )$e met$ods are simple to apply as usually only t$e disc$ar e $as to be specified. Some met$ods, suc$ as t$e (?acey, 18!0), e;uations are still widely used in t$e Indian subcontinent 0 but as most are based of observations of canals t$at carried low sand loads, t$ey are not appropriate for use in spate sc$emes. #n e%ception is t$e Simons and #lbertson re ime e;uations, based on a lar e data set collected from canal systems in India and <ort$ #merica t$at includes e;uations for canals wit$ sand beds and co$esive ban&s carryin K$eavyF sediment loads (2000 9 3000 ppm). 1amac$o in (7#O, 1832) reports t$at t$ese e;uations were used to desi n canals in some of t$e modernised Iemen systems. Sediment transport t$eory s$ows t$at t$ree e;uations are re;uired to determine t$e slope, dept$, and widt$ of a strai $t alluvial c$annel w$en t$e incomin water and sediment disc$ar es, and bed material si=e is specified. )$e first two are provided by alluvial friction and sediment transport e;uations. )$e t$ird, a widt$ relations$ip, can be obtained from minimum stream power or ma%imum sediment transport efficiency concepts. # roup of canal desi n met$ods based on t$ese relations$ips (rational met$ods) is available, and provide t$e most lo ical met$od of desi nin canals to ac$ieve a specified sediment transportin capacity. Of t$ese t$e met$od due to 1$an (183/) provides predictions of slopes and bed widt$s t$at are similar to wide s$allow canals observed in many spate systems.

#lbeit wit$ a number of adGustments to t$e coefficients

1onventionally t$e pea& or desi n disc$ar e is used to determine canal bed slopes and cross sections. If t$is is followed for spate canals t$ere could be serious siltation problems, as spate canals flow at t$eir full desi n disc$ar e for very s$ort time periods. Indicative fi ures can be derived from spate flow duration curves, s$ow t$at a typical canal supplied from a wadi mi $t flow at disc$ar es lar er t$an /0 B of its desi n disc$ar e for only a few B of t$e time. 1amac$o, in (7#O, 1832), su ests t$at t$e slopes of spate canals s$ould be computed for a disc$ar e of 2/B of t$e ma%imum, but notes t$at t$is is in fact ;uite conservative. )$e desi n disc$ar e capacity is provided by specifyin ade;uately si=ed cross sections. )$ese computations can be conveniently carried out usin -4 :allin fordCs K,orcF software pac&a e, w$ic$ includes a wide ran e of canal desi n met$ods. Slopes obtained wit$ t$e 1$an desi n met$od are compared wit$ t$e avera e slope of traditional canals in t$e wadi Mabid system in Iemen in t$e table below. )$e computations are for a median bed sediment diameter of 1mm, and an input sediment concentration of 2000 ppm, a typical sand load sediment concentration carried by a small flood. &isc!arge (m34s+ /0 (from survey) /0 !2./ (2/B) 2/ (/0B) Sediment concentration (sand load6 ppm+ na 2000 2000 2000 Slope (m4#m+ ".0 (from survey) !.2 !.3 !.8

:it$ t$e selected bed sediment si=e and sediment concentration t$e 1$an met$od predicts a slope similar to t$at observed if it is based on /0 B of t$e full supply disc$ar e. -owever t$is would c$an e if a different sediment concentration or median sediment diameter $ad been selected. It is recommended t$at canal desi ns in modernised sc$emes are based on t$e slopes and cross sections of stable e%istin traditional canals. )$e desi n of enlar ed, e%tended or new canals can t$en be derived usin t$e 1$an e;uation, wit$ a Gudicious c$oice of input parameters to provide a ood matc$ wit$ t$e slopes and cross sections observed in e%istin canals. In new sc$emes parameters suc$ as canal bed material si=es and t$e incomin sediment concentration $as to be estimated if rational desi n met$ods are to be used. )$is is beyond t$e scope of t$is paperN suitable procedures are described in (-4 :allin ford, 2001). ..0 ,redicting future command levels

In e%istin sc$emes t$e future rise in command levels is estimated from $istorical rates of rise of field and command levels, derived from surveys, corin or trial pits, and t$e e%tent of upstream movement of traditional diversion structures. In new sc$emes, w$ere t$ere are no spate systems in t$e vicinity, t$e limited sediment yield information t$at is available for catc$ments in re ions w$ere spate irri ation is practiced can be used. )$e 7#O data base of sediment yields of t$e worlds rivers (part of #;uastat) is a ood startin point, alt$ou $ a searc$ s$ows t$at t$ere is now far more information available in t$e public domain t$an w$en t$is information was compiled in 188/.

'%amples of sediment yield data from catc$ments in 't$iopia and 'ritrea are iven below. )$e information is derived from t$e 7#O data base, supplemented wit$ data collated durin proGect studies.
10000

1000

"ield # 3209 $rea%0.21 &2 # 0.36

't!i(pia 'ritrea

"ield t/ m!/# 100 10 1 10 100 1000 Catchment Area m


!

10000

100000

1000000

%igure 5 sediment yields from catc!ments in $ritrea and $t!iopia )$is sort of information can be used derive preliminary, albeit appro%imate, estimates of future increases in command level in spate irri ated areas. :$ere sediment yield data is available for nearby catc$ments of similar si=e, rainfall, slopes and ve etation etc., it can be used directly. Ot$erwise it mi $t be assumed t$at spate irri ated areas will be located in re ions wit$ $i $ sediment yields, and estimates based on yield fi ures from t$e upper ;uartile of t$e available data, for catc$ments of similar si=e to t$e catc$ment of t$e wadi supplyin t$e sc$eme t$at bein considered. 7ield rise rates can be derived usin assumptions as to t$e proportion of t$e annual sediment load t$at will be diverted to a sc$eme, t$e sc$eme command area, a bul& density for settled silts, and an indication of t$e li&ely variation in sedimentation rates between upstream and downstream fields. 2 0.0 0.1 Sediment management options Catc!ment conservation

It is sometimes assumed t$at t$e introduction of soil conservation pro rammes in river catc$ments will rapidly reduce river sediment loads, and t$us sedimentation problems in downstream irri ation canals. )$e information t$at is available su ests t$at t$is is not t$e case over t$e lifetime of typical irri ation structures, e%cept for sc$emes supplied from very small catc$ments. )$e principle reason is t$at in semi arid re ions t$e vast store of sediments available for remobili=ation in catc$ments t$at $ave suffered from $i $ rates of soil erosion in t$e past, continue to contribute to downstream sediment yields for lon periods, decades or even centuries, even if catmint wide conservation was possible. . Soil and water conservation pro rammes will not usually provide si nificant s$ort term reductions in canal sedimentation in spate irri ation systems. -owever construction of series of low c$ec& dams can provide some benefit. #lt$ou $ t$e stora e provided by c$ec& dams soon fills wit$ coarse sediments, t$e reduction in and control of river bed radients provided by c$ec& dams can reduce bed and ban& scour durin floods. #s t$ese measures reduce t$e (unwanted) sand si=es transported by wadis t$ey mi $t be considered as part of wider soil and moisture conservation pro rammes w$en spate irri ation is bein developed in new areas.
2

In wadi ?aba in 'ritrea mean sedimentation rates in upstream fields were about twice t$e mean rate.

0.2

Sediment management options for a range of sc!eme types

)$ese are summarised in t$e table below> ntake!scheme type *asic inta&e wit$out a weir O O O O O *asic (probably small) inta&e wit$ a low weir O O O O O O O O -i $er cost inta&es O O O O O O "ediment management options ?ocate inta&e at t$e outside of a c$annel bend ?imit flows enterin canal wit$ flow t$rottlin structure and reGection spillway. If provided close ates durin periods of very $i $ wadi flows. (rovide steep canals, minimise pondin and flow division. 1onsider arran ements for and sustainability of canal de+siltin ?ocate inta&e at t$e outside of a c$annel bend (rovide a simple sediment sluice. #li n canal inta&e to minimise diversion an le. ?imit flows enterin canal, close ates durin periods of very $i $ wadi flows Pflow t$rottlin structure.reGection spillway. 1onsider if mec$anically e%cavated ravel trap is appropriate. (rovide steep canals, minimise pondin and flow division. 6a&e provision for risin command levels. 1onsider arran ements for and sustainability of canal de+siltin ?ocate inta&e at t$e outside of a c$annel bend Incorporate sediment sluice, consider curved c$annel sediment e%cluder if bed sediments are coarse. #li n canal inta&e to minimise diversion an le. ?imit flows enterin canal, close ates durin periods of very $i $ wadi flows Pflow t$rottlin structure.reGection spillway. 1onsider if mec$anically e%cavated ravel trap is appropriate, or w$et$er flus$ed settlin basin mi $t be feasible. :$ere $i $ investments costs mi $t be Gustified by reduced de+siltin costs, consider $ybrid e%tractor .settlin basin system located in t$e canal $ead reac$. (rovide steep canals, minimise pondin and flow division. 6a&e provision for risin command levels. 1onsider arran ements for and sustainability of canal de+siltin .

O O O

n all cases it is essential to ensure that any planned sediment management measures are both understood and accepted by farmers# and that they are also sustainable over the long term$

Sediment management design software

)$e software pac&a e (S-#41)3, enables a ;uantitative approac$ to be adopted for selection and desi n of sediment mana ement options. It contains a simple inta&e model for assessin sediment e%clusion at basic inta&es, sediment e%tractor and a sluiced settlin basin desi n models, a canal sediment routin model, a canal desi n module, and as well as ot$er modules to assist in analysis of irri ation sc$eme sedimentation problems. )$e pac&a e and supportin tec$nical and software manuals is available free of c$ar e, and can be downloaded from> $ttp>..www.dfid+&ar+water.net.w/outputs.software.$tml

%igure 7 8S!arc9 opening screens

)$e software runs on windows Q( and some earlier operatin systems. It $as also run in :indows Eista usin t$e Q( mode, but t$e rap$ics settin s need to be adGusted.

:eferences

1$an - -, 183/. Design of stable alluvial canals in a system. #S1' Rourn -ydr 'n , Eol 11.1 <o -II. 7#O , 2002, Planning, Development Monitoring and Evaluation of Irrigated Agriculture with Farmer Participation, Eolume I, 6odule 2 7#O.@<,(. 1832. Spate Irrigation Proceedings of the Subregional E!pert "onsultation on #adi Development for Agriculture in the $atural %emen . 0+10 ,ecember, 1832. #den, (,4 Iemen. @nited <ations ,evelopment (ro ramme. 7ood and # riculture Or anisation of t$e @nited <ations. 130pp. -4 :allin ford , 188/ A $umerical Model for Predicting Sediment E!clusion at Inta&es , 'dmund #t&inson, 4eport O, 1!0, 7ebruary 188/. -4 :allin ford, 1880, A method for evaluating the economic benefit of of sediment control in irrigation systems, 1$ancellor 7, ?awrence ( #t&inson '. -4 :allin ford report O, )< 31 -4 :allin ford ,2001, Procedure for the selection and outline design of canal sediment control, )ec$nical and software manuals ( ?awrence ' #t&inson ( Spar& 1 1ounsell. 8 -4 :allin ford, 200/, Improving community spate irrigation, ( ?awrence and ,r 7 van Steenber en, -4 :allin ford report O, 1/"10 ?acey D, 18!0. Stable channels in alluvium 6inutes of t$e (roc., Institute of 1ivil 'n ineers, ?ondon, @A. <o 228, pp2/8+282. 6erritt S 2002, #ater for Agricultures, Irrigation economics in international perspective , Spon press, ?ondon and <ew Ior&, IS*< 0+"1/+2/2!3+/ Smit$ AE-, -ewlett 46D S ?awson R,. # -eadwor&s for Spate Irri ation Systems. 1ivil 'n ineerin 6a a=ine, Sept 1832, )osswell (, 1838. 'apunga (ice Pro)ect, report on hydraulic model investigation for irrigation headwor&s. 1ivil 'n ineerin ,epartment, )$e @niversity, Sout$ampton, @A. Ialin 6S 1821 *heory of hydraulic models, 6acmilan ?ondon

8
10

$ttp>..www.dfid+&ar+water.net.w/outputs.software.$tml $ttp>..boo&s.$rwallin ford.co.u&.acatalo .International,evelopmentpa e1.$tml

You might also like