Cover Image
close this bookDesign and Operation of Smallholder Irrigation in South Asia (WB, 1995, 134 p.)
close this folderChapter 8 - Hydraulics of canal regulation and types of control structures
View the documentBackground
View the documentDownstream control with limited demand
View the documentUpstream control with rotational delivery
Open this folder and view contentsHydraulic controls on secondary and tertiary canals
View the documentHydraulic controls on primary canals
View the documentProduction of small hydraulic structures


The subject of canal regulation, particularly as applied to third world situations, has generated a considerable volume of literature over the last decade, and considerable contention. Publications include recent monographs by the American Societies of Civil Engineers and of Agricultural Engineers, proceedings of the International Commission on Irrigation and Drainage, internal papers of international development agencies including the World Bank publications of academic and research institutions associated with agricultural development, and papers by professionals in developing countries (many of which are available through the Overseas Development Institute and the International Irrigation Management Institute).

Views on the subject expressed by competent professionals and agencies differ substantially. Differences stem principally from the perennial question of how the individual smallholder will react to situations of varying supply and demand. On the one hand is the view that only a simple preordained system of rotational supply will survive, as it certainly has in some situations. The view is tempered, however, by the experience that in other situations such a system is so obviously contrary to cultivator needs that the cultivators themselves reject it. On the other hand, there is the view that with modem technology water deliveries can be, and should be, much more closely matched to cultivator needs, a view which some experienced practitioners treat with considerable reservation, if not cynicism.

The subject of supply of irrigation from a partially regulated or unregulated source, in an area with capricious monsoonal climate, with highly independent small cultivators, is in fact a difficult one. Much as international lending institutions prefer standard solutions to commonly encountered problems, irrigation in developing countries requires a situation-specific approach.

A feature of most South Asian irrigation schemes is that the supply of water in some seasons, particularly in some years, is less than what the cultivators demand. Distribution in such circumstances has to be on the basis of available supply, rather than on demand, i.e. it is to some extent a system of rationing.

The various supply/demand situations have been discussed in the last chapter, with comments on options available for canal management. Systems without storage (run-of" river) or with fixed high-flow season and low-flow season diversions call for no further comment. They are essentially supply-driven, with upstream control. On the other hand, systems with upstream storage (even where providing partial regulation only) offer the possibility of tailoring irrigation releases much more closely to supply and demand. The extent to which this possible flexibility is to be exercised determines the degree of sophistication of the hydraulic controls required and the intensity of management.

For purposes of illustration, a case is considered in which the storage reservoir will have net capacity of about one-third of the mean annual inflow. Rainfall in the irrigation area averages some 1300 mm annually, mainly recurring in the single monsoon season. Soils include both light well-drained uplands and heavy poorly drained low-lands, in close association. Most of the area is in holdings of less than one hectare, primarily owner cultivated. The area is underlain by hard-rock with limited groundwater potential (dug wells only). Basic irrigated crops expected to be grown in the area when the project comes into service are paddy and maize in the monsoon season, paddy being either monocropped or double-cropped, wheat and oil-seeds in the dry season, and sugarcane as an annual crop. However, there is also keen interest in specialty crops including potatoes, tomatoes, and other vegetables, as well as bananas and citrus. Market projections indicate increasing desirability of diversification into these areas.

Design studies are to include a range of options for regulation of water distribution including possible provision of a high degree of flexibility extending either down to the individual farm, or alternatively to the 30 to 40 ha tertiary command.

Downstream control with limited demand

From the cultivator's viewpoint, the most desirable situation would be to have water available "on demand", preferably demand by the individual. During periods in which supply of water to the system is not a constraint, such demands could be met, as far as the hydraulics of the operation are concerned, by a "level-top" arrangement of the tertiary. The tertiary in this case is virtually an elongated pool, or stepped series of pools if gradients so require, the levels in which are maintained constant, regardless of amount of withdrawal, by float operated gates at the head of each reach. Each cultivator has access to the tertiary, and within limits of the capacity of his outlet he can take water whenever he pleases and in whatever amount. This is "downstream control" both in the hydraulic sense, the gate at the head of each reach being controlled by the level downstream of the gate, and in the popular sense, supply being determined from the downstream end of the system (the irrigator) rather than being dictated by upstream agency.

However, when supply to the system is limited, there can be problems with this utopian system. The total demands on the tertiary are reflected, reach by reach, up to head of the tertiary, at which point they become demands on the secondary or sub-secondary. If they are to be met, the gate on the outlet from the secondary must respond to the level in the first reach of the tertiary, maintaining it at its design elevation by discharging into the tertiary whatever flow is required to achieve that objective. The total demand of all the tertiaries served by a secondary becomes the demand at the head of the secondary, and this is passed on to the primary and eventually reaches the head of the primary, which is the outlet from the reservoir. Here a basic problem may arise. The reservoir may have insufficient water in storage to continue supplying at that rate or storage may be being held back for critically important irrigation at the beginning of the next season. Under these conditions, the controlling factor becomes supply rather than demand. Restrictions in water use must be passed on down the line, finally reaching the cultivator and unrestricted demand has to be substituted by limited demand or managed demand.

There are in fact two types of constraint on the downstream demand type of operation. One is in the hydraulics of the system, particularly in the control of the primary and secondary canals. While theoretically an infinite degree of hydraulic flexibility can be provided with capability of instant response to changes in demand regardless of length of canal the required control structures and their management can become very sophisticated. Work continues on more advanced computer-assisted dynamic operation of canal systems and float operated automatic controls. It can be assumed that most present problems will eventually be solved. More intractable is the second type of problem, the management of deficiencies in supply to cultivators while still preserving some degree of freedom in demand.

If the interests of the individual cultivator could be submerged into a collective common interest, a tertiary command of many small holdings with varying cropping patterns and varying water requirements could be served very well by a level-top demand-type of tertiary system. It would be communally managed with due regard to upstream supply constraints and benefitting from the flexibility in water distribution within the tertiary command which such a system provides. Decisions on the timing and amount of irrigation on each unit would presumably be on rational grounds.

However, for various reasons such a communal situation is not generally in prospect in South Asia, although it is approximated in some sugar-cane cooperatives, and the problem remains of how to regulate the use of a restricted supply of water by 8 group of cultivators acting as individuals. One possible solution is the Water Users Association discussed in a later chapter. Levying water charges in proportion to use, with rates possibly scaled upwards if consumption exceeds a certain rate or during periods of critical supply, could provide some degree of control. However, water meters have generally been short-lived in smallholder installations and collection of water charges remains a problem.

To summarize, the downstream demand type of system, while providing complete hydraulic flexibility in withdrawals from the tertiary, nevertheless, requires operational restraint during periods of limited supply. The same considerations apply to demand operation of the secondaries. When there is limited supply to the primary canal it would be unreasonable to respond to unrestricted demand in any secondary, although the hydraulic control at the head of each secondary is designed with that capability. While limited upstream supply may prevent full utilization of the hydraulic flexibility provided by downstream control, such constraint can be minimized for short-term operation (e.g. 24-hourly or weekly) by provision of pondage at some point on the primary canal, if a suitable site can be found.

To reiterate, the key question determining whether a demand-responsive type of system would be workable is whether cultivators would exercise restraint in withdrawals during periods of restricted supply. If not, and in the absence of effective metering at the farm turnout, a demand system in the form of a level-top tertiary can become a "license to steal". The alternative system employing rotational supply within the tertiary command has a tertiary channel of much smaller capacity than the level-top. Even if the rotation breaks down and many cultivators divert from the tertiary simultaneously rather than singly, the total flow which can be diverted is strictly limited to the capacity of the tertiary intake.

However, a limited demand system does provide the ultimate degree of flexibility for irrigation of diversified cropping, in circumstances where the social structure and the character of the cultivator permit its use. The latter factors should be the starting point in consideration of possible application of a demand system in a smallholder situation.

Upstream control with rotational delivery

The term "upstream control. is used in the popular sense, meaning that water releases to the primary canal are decided by the operating- agency, with consideration being given to the supply/demand situation, rather than being directly (hydraulically) responsive to downstream demand. The term will later also be used in the technical sense, implying operation of a gate (usually automatically) in response to the water level immediately upstream of it.

The alternative approaches to the upstream control type of system have been noted in the previous chapter. They are as follows:


(a) Continuous flow throughout, including tertiaries, with division of flow in proportion to area served, through use of flow dividers This system is widely used in small village schemes. In larger public systems the flow divider may be used at branches in primary and possibly secondary canals, but not at offtakes to tertiaries, as the resulting flow during periods of limited supply could be too small for efficient conveyance in unlined tertiaries and field channels, and for field application for non-paddy crops. Continuous flow throughout is not further considered herein.


(b) Continuous regulated flow in primary canals, with full flow supply in fixed rotation to secondaries and their tertiaries. This system requires no operation of controls on the secondaries or the tertiary offtakes. However the proportionate reduction in supply/demand which can be accommodated depends on the divisibility of the secondaries into groups (discussed later). The system is particularly applicable to situations in which the supply in high and low flow seasons is assured (a special situation), and the ratio between the two can be accommodated numerically in designing the grouping of secondaries.


(c) Continuous regulated flow in both primary and secondary canals, with full flow supply in fixed rotation to tertiaries. This removes the above constraint of divisibility of the number of secondaries and also permits tailoring of supply to the needs of individual tertiary commands. Any proportionate reduction in supply can be accommodated, including adjustment to conserve storage during rainfall. It is a very flexible system, but requires operation of hydraulic structures down the length of the secondaries.


(d) Continuous regulated flow in primary canals, with full-flow supply in rotation to secondaries and their tertiaries, subject to a system of rotating priorities contingent upon the amount of water available. This system, widely used in N.W. India, can accommodate any degree of reduction in supply. Its principal disadvantage is the relatively long time between irrigations which may occur in low priority rotations. However, in the area in which the system is currently in use, canal irrigation is widely supplemented by farmer-owned tubewells. The system would not be appropriate in an area with limited availability of groundwater and planned diversified cropping, such as the case under consideration. It is not further considered herein.


(e) Supply to portions of the service area is deleted for the entire season in dry years. This type of operation is appropriate only where a large proportion of the seasonal supply comes from storage and the deficiency is predictable. It is not further considered.

Of the systems listed above the two which could be appropriate to the project situation under discussion are (b) and (c). The difference between the two is that in alternative (b) supply to the secondaries is full-flow and rotational in periods of reduced supply or demand. While in alternative (c) it is variable (regulated) and continuous. As a corollary, in (b) rotational supply begins at the secondary; in (c) it begins at the tertiary. (In another usage, system (b) is "structured" down to the head of the secondary, system (c) down to the head of the tertiary). There are advantages and disadvantages with both alternatives and choice should be project specific.

Such group rotations, operated in conjunction with the project reservoir, may be sufficient to meet all seasonal variations in supply and demand. However, the period between irrigations may be a problem. The minimum practical duration of running time of a secondary operating on/off depends upon its length. If filling and emptying are not to take up too great a proportion of the "on" cycle the minimum practical running time for a secondary may be ten days. With a one on/two off rotation of secondaries (supply at one-third of maximum) the corresponding period between irrigations of a particular holding supplied once from its tertiary during each rotation would be thirty days. For a basic crop such as wheat this irrigation interval would not be a problem, provided that the dates of supply were know in advance. For specialty crops it would be excessive unless cultivators had farm storage, such as a large-diameter dug well. (It is presumed that only limited groundwater is available in the case under discussion).

Apart from the practical length of the "on" period of the secondary, as previously noted there are soil conditions in which rotational operation of unlined secondaries would be unfeasible due to excessive sloughing of banks. There are also ground conditions (high watertable) in which rotational operation would be destructive to a lined secondary, due to back-pressure on the lining.

Summarizing, full-flow rotation of secondaries permits operation of the system at two or three levels of delivery, in addition to delivery at full capacity. As the rotational cycle is likely to be as much as three or four weeks this is not a rapid response type of operation, capable of short-term adjustment. It is more appropriate to making seasonal changes in rate of delivery which can be planned in advance, a situation which generally requires substantial reservoir capacity. The irrigation interval is relatively long during periods of reduced delivery, being determined by the length and filling time of the secondaries. The use of such a system is contingent upon ground conditions which permit on/off cycling of secondaries without deterioration of the canal section. The system is most appropriate where sufficient groundwater is available to provide supplemental irrigation of stress sensitive crops. Its principal advantage is simplicity of operation, gate operation being required only down to the head of the secondary.

In alternative (c), the secondaries and the primary canal operate continuously at variable flow. The tertiaries served by a particular secondary operate rotationally at full-flow, the tertiaries being grouped for this purpose in the same manner as the secondaries in alternative (b). However, there are many more tertiaries than secondaries, with more possible groupings, and the rotational period can be much shorter due to the shorter filling time of tertiaries. Furthermore stability of the channel section under on/off conditions is not generally a problem with the much smaller tertiaries.

For example, with alternative (c), operation the situation is taken in which the supply to the area is reduced to one-quarter of system capacity, either due to reduced availability or low demand. The flow in the primary canal is reduced to one-quarter of capacity by operation of control structures, and also the flow in the secondaries. (The division of flow between primary and secondaries may be by flow dividers on the primary canal, discussed later). To reduce flow in the tertiaries to one-quarter of capacity would be ineffective due to high proportionate seepage losses in these small channels with such low flow and low field efficiency. Consequently, the tertiaries on each secondary are divided into four groups, one group at a time taking the whole flow in the parent secondary and operating at full design capacity. Each group operates for three days on and nine days off, for a rotational cycle of twelve days.

While such an arrangement may appear straight-forward, and it is indeed operable, it poses a number of hydraulic and other operational problems. First, to permit full-flow diversion to tertiaries the water-level in the parent secondary must be maintained at or near full supply level, even while flow in secondary is reduced to one-quarter of capacity. This requires a considerable number of hydraulic structures on the secondary and their operation. Second, and probably more importantly, the tertiary intakes have to be gated and the gates must remain closed other than during the appropriate rotational turn. The question is what agency operates the tertiary intake gates, and ensures that they are not opened out of turn, particularly in periods of severely reduced supply when crop survival is at stake.

Downstream control

As noted earlier, limited-demand downstream-control type of operation could be provided through a level-top canal or stepped series of such canals, constituting the tertiary, to which all farms would have access. Controls on the tertiary would consist of a series of weirs, each with gates automatically activated by the level in the reach immediately downstream of the weir, serving to maintain the downstream reach full at all times. Gates to each farm turn-out would need to be adjustable by the cultivator, providing flexibility as to the rate of flow diverted to the farm at any time. A flow recording device either at the head of the tertiary or at each turn-out would also be required.

In view of the problems of maintaining level-top float controls in open tertiary canals a more feasible arrangement could be to substitute a buried pipe for the tertiary canal if ground-slopes provided sufficient gradient for economic use of pipe. A valved outlet would serve each farm. If the ground-slope were excessive for the use of low-head pipe in a demand type of operation, the head could be reduced where required by the use of float-operated stand-pipes or other pressure-reducing devices. It is noted that a closed pipe system with valved outlets, providing supply on demand, subjects the pipe to significant hydrostatic pressure. The alternate use of pipe as part of an upstream control system, sometimes referred to as a "buried channel" system, does not.

Structures on the secondary canal supplying a demand-type of tertiary operation would automatically maintain the required level in the secondary at each outlet, also the setting of the tertiary intake gate required to maintain the level in the first reach of the level-top tertiary. Structures on the primary canal would serve the same type of function for supply to each secondary, but the operation would be more complex in the case of the primary canal due to its greater length and dynamic effects (surges) in its operation.

It will be evident from the above discussion that the structures required for the operation of a system providing limited-demand type of service down to the individual smallholder, while technically feasible, would not be particularly simple in nature. In the small irrigator environment, where usually only the most robust structures survive, such a system could be entertained only under the most favorable conditions of cultivator support.

Upstream control

As previously discussed, there are two alternatives types of operation based on upstream control. In the first (alternative (b) above) both secondaries and their associated tertiaries and are run rotationally on full flow. Although there are limitations in the applicability of this system, noted earlier, it is undoubtedly the simplest with respect to provision and operation of hydraulic structures. At the tertiary level, particularly if the system is designed to supply one farm sum-out at a time, the structures required are simply turn-outs, checks, branch structures, and drops. All of them are gated, the gates being either fully open or fully closed. The only problems are leakage of the closed gates, and theft of gates. Whether they are made of wood, sheet steel, or concrete tile, gates found useful for household purposes are widely subject to theft. In place of the missing gate, closure is usually made by brushwood and mud, excavated from the channel bottom and wedged across the gate opening. In one design, the conventional gate is backed up by a grillage of embedded steel rods extending across the opening, behind the gate slot. The rods facilitate such extemporaneous closure.

The structures down the length of the tertiary (farm turnouts), in turn, divert the full flow in the channel. On the other hand, the structure at the head of the tertiary (the tertiary intake) diverts only part of the flow in the parent secondary. As the secondary always runs at full capacity with this system, control of flow to each tertiary intake is relatively straight forward. It could be achieved by use of a flow divider at each intake, as the aim is to divert a proportion of the flow in the secondary to each tertiary. However, flow dividers are not regularly employed at tertiary intakes in view of their cost (the divider is essentially a weir, and extends across the full width of the secondary), the number required, and the fact that a hydraulic drop in the secondary is required at each divider and sufficient head may not be available for a series of such drops. The type of structure normally used for the tertiary intake is hydraulically a submerged orifice, in some cases simply a pipe or culvert extending through the canal bank. More sophisticated structures have shaped entry and may incorporate energy recovery at the exit from the culvert. The virtue of the orifice type of intake is that its discharge is affected by head (i.e. by level in the secondary) to only a relatively small degree only. In fact, it is proportional to the square root of head, and over a range of level in the secondary the proportionate change of flow to the tertiary intake, due to a change in level in the secondary, can be approximately the same as the proportion change in flow in the secondary itself. This feature, sometimes referred to as "modularity" is desirable, as the flow or the level in the secondary may not always be exactly as planned. It is preferable that the tertiary intakes divert a proportionate share in the deficiency or excess of the flow in the parent canal. The A.P.M (Adjustable Proportionate Module) of the N. W. Indian irrigation systems is of this type, diverting not only a "modular" share of the flow in the secondary but also a corresponding share in the silt being carried by it. An ingenious type of intake structure of French design has capacity near-constant over a range of head. Such a structure is appropriate to a system in which the flow in the secondary is closely regulated, and the flow to be diverted at each tertiary is consequently also closely defined. In a less well regulated system, however, where the flow in the secondary is likely to vary, proportionality of diversion at the intake rather than a fixed amount of diversion is the desirable feature.

With the type of operation in question, the secondary runs at full capacity only (other than during filling and emptying), which greatly simplifies requirements for control structures on the secondary. The water-level profile, and the level opposite each tertiary intake, is determined by canal geometry and within limits is predictable. It is required to be known, for purposes of tertiary intake design, at one flow only, i.e. full flow. Particularly if the capacity of the intakes is adjustable (one-time adjustable) and can be matched to actual water levels in the secondary determined during initial operation, formal hydraulic structures for control of level in the secondary at tertiary intakes need not normally be required. If for any reason such structures are, in fact, needed for stabilization of water-levels (such as at drops), their setting can remain fixed, and related to full flow only.

In the second alternative ((c) above) the flow in the secondaries is not rotational. It is continuous, the rate varying in accordance with supply or demand. The tertiaries operate at full capacity, but in groups, rotationally. For instance, if the secondary is running at one-third design capacity, one-third of the tertiaries operate at a time. The tertiaries may be grouped as the upper one-third, followed by the middle third, and finally the downstream third, the cycle then being repeated. Alternatively every third tertiary down the whole length of the secondary, beginning with the first, may operate as a rotational group, followed by a similar group beginning with the second tertiary, and finally the last group beginning with the third tertiary. The cycle is then repeated. The structures required within the tertiary command (the farm turn-outs, etc), and their operation, are the same as in the alternative (b) system discussed above. However, the situation in the secondary, including the outlets to the tertiaries, is substantially different. Because the flow in the secondary may vary from full capacity to as little as one-quarter of full-flow, hydraulic controls are required down the length of the secondary to maintain water levels at or near full supply level adjacent to tertiary outlets, thus ensuring that the outlets operate at design capacity.

Several types of structure could be used for this purpose, with either fixed or variable geometry. Conceptually, the simplest is a fixed weir with a crest sufficiently long that the range in depth over the crest between low-flow and full-flow (i.e the range in level in the secondary at that point) would have an acceptably small effect on the discharge in the adjacent tertiary outlet. As an example, a secondary of 1.0 m³/sec (35 ft³/sec) full flow capacity would require a weir of crest length of about 8.5 m (28 ft) if the range of level between full flow and one-quarter flow over the weir were to be kept at no more than 10 cm. The latter range would cause a change in discharge capacity of the adjacent tertiary intake (design head 30 cm) of about 20%, which would probably be acceptable. The point of significance is that the required weir crest length is 8.5 m whereas the width of a canal of this capacity is about 2.5 m. The weir would consequently need to be U-shaped in plan, extending about 4.25 m downstream to provide the desired crest length. This is the classical "duck-billed" weir. It is not a particularly low-cost structure, and the practicability of using it in a specific case would depend upon the number of such structures required, i.e. the spacing along the canal, which is determined by the gradient of the canal alignment.

The alternative to a fixed weir is a weir with moveable gates, adjusted to maintain levels in the secondary whenever the flow is changed. For small canals the gates could take the form of manually placed vertical "needle-beams" (which control the width flow over the weir crest), or radial or leaf gates. More sophisticated automatic float-controlled gate are also available.

Although the type (c) system with continuous regulated flow in the secondaries provides highly desirable flexibility in supply, the fact that hydraulic structures on the secondary have to be adjusted in the course of such operation is a disadvantage. Tertiary intake gates necessarily have to be opened and closed rotationally. Gates on the level control weirs in the secondary also have to be operated, unless duck-billed weirs or automatic float-controlled gates are used. However, with appropriate design it is possible to confine such gate operation to accommodating changes in rate of delivery in the secondary, and to avoid the need for adjustment at each rotation of the tertiary groups.

Maintaining the latter rotations and avoiding unauthorized opening of tertiary intakes out-of-turn is, in fact, the principal concern with this type of operation. It is necessary to have sufficient staff by the operating agency to maintain supervision of secondary canal structures including tertiary intakes or farmer organizations capable of operating the secondary canals as well as the tertiaries.

Hydraulic controls on primary canals

As the small cultivator is not generally involved in operations at the primary canal level, technical details of primary canal control structures are not relevant to this discussion. However, the extent to which irrigation systems are designed to be demand-responsive at the tertiary level has a major influence on the operational requirements of primary canal controls.

With currently available technology it would be technically possible to make primary canals instantly responsive to changes in demand. It would, however, be very costly to do so and unjustified when problems in management of limited-demand at the small cultivator level, discussed earlier, rule out such operation at least for the immediate future.

Of more immediate concern is the need to design into the primary system controls which can accommodate possible future developments, such as a change in the direction of greater crop specialization in various areas of the project command, with consequent changes in allocation of water to particular secondaries and in capacity required at structures. The possibility of a future change from on/off full flow rotation of secondaries in favor of continuous regulated flow, which could affect design requirements of outlet structures on the primary canal, should also be taken into account (Plusquellec 1985 and 1988, Le Moigne 1988).

Production of small hydraulic structures

Supply to smallholders requires a large number of hydraulic structures at the tertiary level. For instance, a 20,000 ha service area with holdings averaging one hectare in size would need 500 to 600 tertiary intakes and some 20,000 farm turnouts, together with a large number of "junction structures", checks, and drops.

As previously noted, installation of tertiary structures has been responsible for much of the delay (often several years) between completing the main canal system and bringing the whole command into effective operation. Traditional brick construction of small hydraulic structures is satisfactory technically, but installation is very slow and limited to the dry season. The alternative material is pre-cast concrete. Manufacture of the units can continue throughout the year, and installation, although also limited to the dry season, is relatively rapid and requires largely unskilled labor.

While pre-cast concrete is undoubtedly the indicated solution to the problem, experience with small pre-cast structures underlines several design considerations unique to the smallholder situation. The tertiary outlet, in particular, is generally regarded by cultivators as the main constraint on the rate of diversion to their holdings, which is certainly true. The structure is consequently subject to attack and much ingenuity is devoted to increasing its discharge. Small structures in general are also particularly susceptible to destructive soil pressures and movement in expansive clays, a factor which operates against the use of light-weight sectionalized construction in pre-cast units. Robust, relatively heavy construction is preferable, economy being served by low labor requirements in installation rather than in cost of materials.

The problem of theft of gates on tertiary outlets and farm outlets, also on junction structures, has been discussed earlier. No material has proved immune and chains and padlocks are equally susceptible. The only material not subject to theft is soil, and this is eventually the fall-back material for closures. Repeated taking of mud from the channel-bed adjacent to structures, for closure purposes, can result in depressions which remain water-filled after each rotation, eventually seeping away and representing water-loss. The problem cannot be entirely avoided, but it can be reduced by designing the structures so as to minimize the amount of soil, or mud, required for closure. Steel bars embedded for this purpose in the throat of the opening have been incorporated in some designs.

Means of providing a tamper-proof adjustable gate for tertiary outlets has received much attention in the past, but without notable success. None of the operational systems discussed above in fact require adjustment of the intake in normal service; they are all full-flow rotational systems. Capability of one-time adjustment of capacity at the time of initial installation is, however, a desirable feature, for two reasons. First, because the size of tertiary command unavoidably varies, the appropriate capacity of outlet also varies from one to another. It is convenient to deliver standardized pre-cast units to the field and to make the appropriate final adjustment to capacity at each location. Second, the head in the intake and its capacity are functions of water-level in the parent secondary and in some cases water-level in the tertiary (if the intake is operating "submerged"). Both levels can be estimated in advance approximately only. It is convenient to make final adjustment of the intake in the field, in accordance with actual measured levels. One method of doing so is to supply the intake in the form of a standard body with oversized opening in which an insert sleeve is permanently grouted at the site. The sleeves are supplied in a range of sizes of opening, from which selection is made appropriate to the size of service area and the actual head on the particular intake.