Role of sediment in the design and management of irrigation canals : Sunsari Morang Irrigation Scheme, Nepal

Sediment transport in irrigation canals The sediment transport aspect is a major factor in irrigation development as it determines to a large extent the sustainability of an irrigation scheme, particularly in case of unlined canals in alluvial soils. Investigations in this respect started since Kennedy published his channel-forming discharge theory in 1895. Subsequently different theories have been developed and are used around the world. All of them assume uniform and steady flow conditions and try to find the canal dimensions that are stable for a given discharge and sediment load. In the past irrigation schemes were designed for protective purposes with very little flow control, hence steady and uniform flow conditions could be realised to some extent. Modern irrigation schemes are increasingly demand based, which means that the water flow in a canal is determined by the crop water requirements. Accordingly the flow in the canal network is not constant as the crop water requirement changes with the climate and the growing stages of the crops. Also the inflow of the sediment is not constant throughout the irrigation season in most schemes. The situation is even worse for run-of-the-river schemes where fluctuations in the river discharge have a direct effect on the inflow of water and sediment. The conventional design methods are not able to predict accurately the sediment transport behaviour in a canal, firstly due to the unsteady and non-uniform water flow conditions and secondly due to the changing nature of the sediment inflow. Hence, the actual behaviour of a canal widely diverges from the design assumptions and in many cases immense maintenance costs have to be met with to tackle the sediment problems. An irrigation scheme should not only be able to deliver water in the required amount, time and level to the crops on the field, but also should recover at least its operation and maintenance cost. Cost recovery is, to some extent, related to the level of service provided by the irrigation organization and the expenditure for operation and maintenance of the scheme. Past experiences in Nepal have shown that modernization of existing irrigation schemes to improve the level of service has also increased the operation and maintenance costs. These costs are, in some cases, high compared to the generally low level of ability of the water users and farmers to pay these costs. The search of making schemes more equitable, reliable and flexible has resulted in the introduction of new flow control systems and water delivery schedules that may, if not carefully designed, adversely affect the sediment transport behaviour of a canal. In quite some schemes unpredicted deposition and/or erosion in canals have not only increased the operation and maintenance costs but also reduced the reliability of the services delivered. Irrigation development in Nepal and the study area Nepal is a landlocked country in South Asia lying between China and India. It is situated between 26º22' N to 30º27' N latitude and 80º4' E to 88º12' E longitude of the prime meridian. Roughly rectangular in shape, the country has an area of 147,181 km2. It is 885 km in length but its width is uneven and increases towards the West. The mean North-South width is 193 km. Nepal is a predominantly mountainous country, with elevations ranging from 64 m+MSL (Mean Sea Level) at Kechana, Jhapa to 8,848 m+MSL at the peak of the world highest mountain, Everest, within a span of 200 km. Nepal has a cultivated area of 2.64 million ha, of which two third (1.77 million ha) is potentially irrigable. At present 42% of the cultivated area has some sort of irrigation, out of which only 41% is receiving year round irrigation water. The existing irrigation schemes contribute approximately 65% of the country’s current agriculture production. Nepal has a long history of irrigated agriculture. Most of the existing large-scale irrigation schemes are located in the southern alluvial plain (Terai). The canals are unlined and the sediment load forms an integral part of the supplied irrigation water. The schemes are predominantly supply based and have a very low duty for intensive cropping. In view of the increased competition among the different water using sectors and low performance of these schemes, many of them are undergoing modernization. For example, the Sunsari Morang Irrigation Scheme (SMIS) is one of the schemes under modernization, and it has been taken as a study case for this research. A better understanding of the sediment transport process under changing flow and sediment load conditions, a shifting management environment and different maintenance scenarios will be very useful in pulling out the schemes from the present vicious cycle of construction-deterioration-rehabilitation. The Sunsari Morang Irrigation Scheme (SMIS) is located in the eastern Terai. The Koshi River is the source of water. A side intake for the water diversion, an around 50 km long main canal of capacity 45.3 m3/s for water conveyance and 10 secondary canals and other minor canals of various capacities for water distribution were constructed to irrigate a command area of 68,000 ha. The system was put into the operation in 1975, but faced a serious problem of water diversion and sediment deposition in the canal network. Hence from 1978, after 3 years of operation, rehabilitation and modernization work of the scheme has been started. During modernization the intake has been relocated to increase the water diversion and reduce the sediment entry. Besides, a settling basin with dredgers for continuous removal of sediment has been provided near the head of the main canal. Apart from that the command area development and modernization of existing canal network is in progress and till third phase (1997-2002), around 41,000 ha area has been developed. Sediment transport research The aim of this research is to understand the relevant aspects of sediment transport in irrigation canals and to formulate a design and management approach for irrigation schemes in Nepal in view of sediment transport. In the process, the design methods used in the design of irrigation schemes in Nepal and their effectiveness on sediment transport have been studied. The impact of operation and maintenance on sediment movement has been analysed taking the case study of SMIS. An improved design approach for sediment transport in irrigation canals has been proposed. A mathematical model SETRIC has been used to study the interrelationship of sediment movement with the design and management and to evaluate the proposed design approach for irrigation canal based on the data of the SMIS. The mathematical formulation of sediment transport process in an irrigation canal is based on the previous works in this field, most notably the work of Mendez on the formulation of the mathematical model SETRIC. Subsequent analysis, improvement and verification works by Paudel, Ghimire, Orellana V., Via Giglio and Sherpa have been used. The model SETRIC has been verified and improved where found necessary and has been used to analyse the irrigation scheme and to propose an improvement in the design and management from sediment transport point of view. Assessment of design parameters The methods of selecting the design discharge and sizing of canals for modern irrigation schemes based upon the present concept of crop based irrigation demand, water delivery schedules and water allocation to the tertiary units have been analysed. The selection of a crop depends upon the soil type, water availability, socio-economic setting and climatic conditions. The type of crop together with the soil type determines the irrigation method and irrigation schedules, while the type of crop and climatic condition determines the irrigation water requirement. The required flow in a canal is then derived based on the water delivery schedule from that canal to the lower order canals or to the field to meet the water requirement. The factors that influence the roughness of an irrigation canal have been analysed and a proposal for a more rational roughness determination process has been formulated based on the available knowledge. The roughness in the sides depends upon the shape and size of material, vegetation and surface irregularities, while the roughness in the bed is a function of shape and size of material and the surface irregularities (bed form in case of alluvial canals). For the prediction of roughness in the bed mostly two approaches are in use – methods based on hydraulic parameters (water depth, flow velocity and bed material size) and the methods based on bed forms and the grain related parameters. In this research, the method based on the bed form and grain related parameters, as suggested by van Rijn, has been used. Similarly, for the determination of roughness in the sides, the influence of surface irregularities have been included by dividing the maintenance condition as ideal, good, fair and poor and accordingly applying the correction to the standard roughness value for the type of material. The influence of vegetation has been accounted based on the concept of V.T. Chow. The various methods of computing the equivalent roughness have been compared and the method proposed by Mendez has been found to be better when tested with the Kruger data. Most of the sediment transport predictors consider the canal with an infinite width without taking into account the effects of the side walls on the water flow and the sediment transport. The effect of the side wall on the velocity distribution in lateral direction is neglected and therefore the velocity distribution and the sediment transport are considered to be constant in any point of the cross section. Under that assumption a uniformly distributed shear stress on the bottom and an identical velocity distribution and sediment transport is considered. Majority of the irrigation canals are non-wide and trapezoidal in shape with the exception of small and lined canals that may be rectangular. In a trapezoidal section the water depth changes from point to point in the section and hence the shear stress. The effect would be more pronounced if the bed width to water depth ratio (B-h ratio) is small. The change in velocity distribution in a canal in view of the change in boundary shear and water depth along the cross section has been analysed and evaluated with the field measurements. The change in velocity and shear stress in a canal section has been used to evaluate the influence of B-h ratio and side slope in the prediction of sediment transport capacity by selected predictors (Brownlie, Engelund-Hansen and Ackers-White). The evaluation with the available data set showed that the proposed correction improved the predictability for non-wide irrigation canals. Canal design approaches for sediment transport in Nepal For the design of canals having erodible boundary and carrying sediment loads two approaches are in practice, namely the regime method and the rational method. The regime design methods are sets of empirical equations based on observations of canals and rivers that have achieved dynamic stability. The rational methods are more analytical in which three equations, an alluvial resistance relation, a sediment transport equation and a width-depth relationship, are used to determine the slope, depth and width of an alluvial canal when the water and sediment discharges as well as the bed material size are specified. In Nepal, the design manuals of the Department of Irrigation recommend Lacey’s regime equations and White-Bettess-Paris tables with the tractive force equations for the design of earthen canals carrying sediment. But in practice, there is no consistency in the design approaches that has been found to vary from canal to canal even within the same irrigation scheme. The use of Lacey’s equation for computing the B-h ratio has generally resulted in wider canals. This is so, because flatter side slopes than predicted by the Lacey’s equations are used from soil stability considerations. The White-Bettess-Paris tables are derived from alluvial friction equations of White, Bettess and Paris (1980) and sediment transport equations of Ackers and White (1973). No records regarding the use of this method for the design of canals was found and hence its performance in terms of sediment transport could not be verified. However, the Ackers and White sediment transport equations over-predicted the sediment transport capacity of a canal when tested with the SMIS data. The sediment load entering into the canals of SMIS is mostly fine (d50 < 0.2 mm) and most of the large scale irrigation schemes in Nepal have similar geo-morphological settings. That means that the White-Bettess-Paris tables will result in a canal with a flatter slope than actually required to carry the type of sediment prevailing in SMIS and other similar irrigation schemes of Nepal. Analysis showed that the Brownlie and Engelund and Hansen equations are more suitable for the type of sediment that has been found in SMIS. During the modernization, the secondary canals (S9 and S14) of SMIS have been designed by two different approaches. Secondary Canal S9 has been designed using Lacey’s regime concept while Secondary Canal S14 has been designed using an energy approach. In the energy approach the erosion is controlled by limiting the tractive force and the deposition is controlled by ensuring equal or non-decreasing energy of the flow in the downstream direction. Both the canals have been evaluated for their sediment transport capacity for the prevailing sediment characteristics. The carrying capacities of both canals (~ 230 ppm) have been found to be less than the expected sediment load (~ 300 – 500 ppm) in the canal. The energy concept assumes that the sediment transport is proportional to the product of velocity and bed slope. The carrying capacity of the canal designed by this principle has been found to be variable along its length. It means that the sediment transport capacity is not only a function of bed slope and water depth as assumed in the energy concept. An improved approach for the design and management of irrigation canals In general the reliability of sediment transport predictors is low and at best they can provide only estimates. As per Vito A Vanoni (1975) a probable error in the range of 50-100% can be expected even under the most favourable circumstances. There is no universally accepted formula for the prediction of sediment transport. Most of them are based upon laboratory data of limited sediment and water flow ranges. Hence they should be adjusted to make them compatible to specific purposes, otherwise the predicted results will be unrealistic. An improved rational approach has been proposed for the design of alluvial canals carrying sediment loads. To find the bed width, bed slope and water depth of a canal for a given discharge and sediment characteristics three equations, namely a sediment transport predictor (total load), resistance equation (Chézy) and a B-h ratio predictor are used. A canal design program DOCSET (Design Of Canal for SEdiment Transport) has been prepared for the improved approach including the above mentioned improvements. The program can also be used to evaluate the existing design for a given water flow and sediment characteristics. Basic features of the new approach are:  concept of dominant concentration. Instead of using the maximum concentration, the approach suggests to look for a concentration that results in net minimum erosion/deposition in one crop calendar year;  determination of roughness. The proposed method makes use of the elaborated and more realistically determined roughness value in the design process. The roughness of the cross section is adjusted as per the hydraulic condition and sediment characteristics. Moreover the influences of the side slopes and the B-h ratio are included while computing the equivalent roughness of the section. This should result in a more accurate prediction of hydraulic and sediment transport characteristics of the canal and hence, a better design;  explicit use of sediment parameters. The sediment concentration and representative size (dm) is explicitly used in the design. That will make the design process more flexible as different canals might have to divert and convey sediment loads of different sizes (dm) and amounts;  Use of an adjustment parameter. An adjustment parameter has been used that includes the influence of non-wide canals, sloping side walls and exponent of velocity in the sediment transport predictor. This adjustment should increase the accuracy of the predictors when they are used in irrigation canals, an environment for which they were not derived;  holistic design concept. This approach uses one canal system as a single unit. The canal system may have different canals of different levels, but the water and sediment management plans are prepared for the whole system. Then the hydraulic design of the individual canal can be made to meet the design management plan for that canal;  Selection of B-h ratio. A B-h ratio selection criterion has been proposed considering the side slope selection practices in Nepal as well as the sediment transport aspects. Since, the sediment transport process is influenced by the management of the irrigation scheme, the design should focus to have a canal that is flexible enough to meet the demand and still have a minimum deposition/erosion. The provision of sufficient carrying capacity up to the desired location (conveyance), providing controlled deposition options if the water delivery plans limit the transport capacity (provisions of settling pockets) and preparation of maintenance plans (desilting works) are some of the aspects that would have to be analysed and included in the design to reduce the sediment transport problems. The canal design methods can give the best possible canal geometry for a given water flow and sediment concentration only. For water flows and sediment concentrations other than the design values, there may be either erosion or deposition. The aim of the design would have to be to balance the total erosion and deposition in one crop calendar year. So, a design may not be based on the maximum sediment concentration expected during the irrigation season, but on a value that results in the minimum net erosion/deposition. The best way to evaluate a canal under such scenario is to use a suitable sediment transport model. Besides, the roughness of the canal depends upon the hydraulic conditions, sediment characteristics and the maintenance plans that are constantly changing throughout the irrigation season. The canals are designed assuming a uniform flow and sediment transport under equilibrium condition. However, such conditions are seldom found in irrigation canals due to the control in flow to meet the variation in water demand. Hence, the design of a canal would have to be evaluated using a sediment transport model for the selection of proper design parameters and to evaluate the design for the proposed water operation plans. The mathematical model SETRIC The mathematical model SETRIC is a one-dimensional model, where the water flow in the canal has been schematised as a quasi-steady and gradually varied flow. This one dimensional flow equation is solved by the predictor-corrector method. Gallappatti’s depth integrated model for sediment transport has been used to predict the actual sediment concentration at any point under non-equilibrium conditions. Galappatti’s model is based on the 2-D convection-diffusion equation. The mass balance equation for the total sediment transport is solved using the modified Lax’s method, assuming a steady condition of the sediment concentration. For the prediction of the equilibrium concentration one of the three total load predictors: Brownlie, Engelund and Hansen or Ackers and White methods can be used. The model SETRIC was evaluated using other hydrodynamic and sediment transport models (DUFLOW and SOBEK-RIVER) and was validated by the field data of SMIS. Predictability of different predictors has been compared. The Brownlie and Engulund and Hansen methods predicted reasonably for the sediment size of 0.1 mm (d50), while predictability of Ackers and White for the sediment size was found to be poor. The sensitivity of Brownlie’s method was more uniform than the other two methods for a sediment size range of 0.05 to 0.5 mm. Field data collection For the field measurements of the sediment transport process, one of the secondary canals of SMIS (S9) was selected. Since, the objective of field data was to test the design approach for sediment transport; preference was given for a canal that was recently designed and constructed. The field measurement of water and sediment flow was carried out in 2004 and 2005. During field measurements the water inflow rate into Secondary Canal S9 system was measured. A broad crested weir immediately downstream of the intake for Secondary Canal S9 was calibrated and used for discharge measurement. For sediment concentration measurements, dip samples just downstream of the hydraulic jump were taken on a daily basis. The samples were then analysed in the laboratory and the sediment concentration was determined. Point sampling across the section using pump samplers were also taken and the calculation results showed that the dip samples underestimated pump samples by around 8% in case of the total load and by around 35% for the sediment of size > 63 μm. At the end of the irrigation season, the deposited sediment samples along the canal were taken to determine the representative sediment size and ot