Yorkshire Water

ABB

United Utilities

EPSRC

NEPTUNE MONITOR

CONTROL

OPTIMISE

 

Energy and Pressure Management of Oldham Water Supply System United Utilities case study Off-line study

by Hossam AbdelMeguid, Piotr Skworcow, and Bogumil Ulanicki

6/2009 Process control - Water Software Systems, De Montfort University

I

Table of content Table of content ......................................................................................................... I List of figures...........................................................................................................III List of Tables ......................................................................................................... VII Energy and Pressure Management of Oldham Network United Utilities case study ....1 Introduction................................................................................................................1 Operational Control of Water Supply Systems......................................................1 Description of Water Supply and Distribution Systems ........................................1 Components of Water Supply and Distribution Systems ......................................2 Operational control of Water Supply and Distribution Systems ...........................4 Features of Water Supply Systems ........................................................................5 Features of Water Distribution Systems ................................................................5 Control and Decision Structures for Water Supply and Distribution Systems ......6 Pressure control in water distribution systems.......................................................7 Pump scheduling in water distribution systems.....................................................8 Scheduling multi-source, multi reservoir systems ...............................................10 Aim and objectives ..............................................................................................11 Methodology ........................................................................................................12 United Utilities’ expectations ..............................................................................13 ABB’s expectations .............................................................................................15 Requirements ...........................................................................................................15 Case Study .................................................................................................................1 Oldham network topology and structure................................................................3 The schematic diagram of Oldham water supply system ......................................5 Current operation policy ..........................................................................................14 Energy Tariff............................................................................................................16 Elements and components to be scheduled..............................................................17 Pump stations .......................................................................................................17 Water treatment works .........................................................................................18 Valves ..................................................................................................................18 Operational Constraints ...........................................................................................19 Pressure constraints..............................................................................................19 Constraints of the reservoirs ................................................................................19 Constraints of pump stations................................................................................20 Constraints of water treatment works ..................................................................20 EPAnet hydraulic model ..........................................................................................21 Testing and running the EPANET model ............................................................47 Modifications on the EPANET model of Oldham water supply system .............50 FINESSE hydraulic model.......................................................................................59 Descriptive Summary of FINESSE .....................................................................59 Element characterisation in FINESSE .................................................................61 FINESSE simplified model......................................................................................67 Network simplification problem ..........................................................................67 Simplified Oldham water supply model ..............................................................68 Pump Characteristics ...........................................................................................70 Reservoir characteristics ......................................................................................73 Verification of the simplified model....................................................................74 Optimal network scheduling problem......................................................................92 Leakage model .....................................................................................................92 I

Objective function................................................................................................93 Model of water distribution system .....................................................................94 Constraints ...........................................................................................................94 Using a GAMS modelling environment to solve network continuous scheduling problems...............................................................................................................94 Discretisation of continuous schedules..............................................................102 Results....................................................................................................................105 Scenario 1- Full optimisation (Pumps, Valves and WTWs)..............................105 Comparison between optimal schedule and current operation and rules...........115 Scenario 1- Optimisation of pumps schedules only...........................................127 Comparison between optimal schedule and current operation and rules...........137 Scenario 2- Full optimisation (Pumps, Valves and WTWs)..............................149 Comparison between optimal schedule and current operation and rules...........159 Scenario 2- Optimisation of pumps schedules only...........................................171 Comparison between optimal schedule and current operation and rules...........181 Scenario 2 - 7 days optimisation- Pumps and valves only.................................193 Comparison between optimal schedule and current operation and rules...........203 Scenario 2 - 7 day’s optimisation with modified final reservoirs level - Pumps and valves only ..................................................................................................213 Comparison between optimal schedule and current operation ..........................223 Scenario 2 – continuous solution of 7 days schedule with modified initial reservoirs levels - Pumps and valves only .........................................................234 Scenario 3- Full optimisation (Pumps, Valves and WTWs)..............................245 Comparison between optimal schedule and current operation and rules...........255 Scenario 3- Optimisation of pumps schedules only...........................................267 Comparison between optimal schedule and current operation and rules...........277 Scenario 4- Full optimisation (Pumps, Valves and WTWs)..............................289 Comparison between optimal schedule and current operation and rules...........299 Scenario 4- Optimisation of pumps schedules only...........................................311 Comparison between optimal schedule and current operation and rules...........321 Scenario 2- Sources optimisation.......................................................................333 Leakage sensitivity.............................................................................................347 Comparison Between different scenarios ..........................................................358 The effect of changing the lower reservoirs constraints on the continuous solution...............................................................................................................360 Conclusions............................................................................................................374 Overview of optimisation results .......................................................................374 References..............................................................................................................376 Appendix A............................................................................................................380 Appendix B ............................................................................................................381 Appendix C ............................................................................................................382 Appendix D............................................................................................................385

II

List of figures Figure 1 A typical water supply and distribution system ..............................................3 Figure 2. schematics of control and Decision Structures for Water Supply and Distribution Systems......................................................................................................7 Figure 3. schematic representation of leakage reducing options ...................................8 Figure 4 UU’s water supply zones.................................................................................2 Figure 5. Water sources and Aqueduct lines .................................................................4 Figure 6. The connections between Oldham area and Tameside, Rochdale systems and Manchester Ring Main...................................................................................................5 Figure 7. Schematic diagram of Oldham water supply system .....................................7 Figure 8 The annual power consumption and electricity cost of the pump stations for 2006/2007 ....................................................................................................................11 Figure 9 Oldham DMA connectivity schematic ..........................................................13 Figure 10 Daily total demand and exported flow of Oldham water supply system ....14 Figure 11. Weekly total demand of the DMAs of Oldham area in 2007/2008............14 Figure 12 Electricity tariff for 2008/2009....................................................................17 Figure 13. Reservoirs constraints and initial water levels ...........................................20 Figure 14. EPANET hydraulic model of the Oldham water supply system................21 Figure 15. Node elevation and DMAs in Oldham system ...........................................23 Figure 16. Typical demand pattern ..............................................................................24 Figure 17. Nodes of imported flow and the nodes assigned to a leakage ....................24 Figure 18. The leakage patterns ...................................................................................25 Figure 19. Fixed and variable head reservoirs in Oldham area ...................................26 Figure 20 The head patterns of the reservoirs 00014C31 and A00BAF15..................26 Figure 21 The outlet flow from Buckton Castle WTW to feed Oldham system .........27 Figure 22. Tank Volume Curve ...................................................................................27 Figure 23. Pipes of high roughness value ...................................................................32 Figure 24. Example pump curves ................................................................................34 Figure 25. Pump Efficiency Curve ..............................................................................35 Figure 26. The locations of the pump stations in Oldham water supply system .........36 Figure 27. All valve locations in Oldham model .........................................................41 Figure 28. Locations of the partially activated valves .................................................42 Figure 29 The time series plot of the control rules of the valve 24800001 .................44 Figure 30 Schematic of the pump stations controlled by reservoirs water level .........45 Figure 31 The locations of the nodes of negative and low pressure value ..................47 Figure 32. Reservoirs water levels...............................................................................49 Figure 33 Smoothed and original data of the pump stations .......................................52 Figure 34 The differences of the reservoirs level between the original EPANET model and the modified Hazen-William and pump data model .............................................53 Figure 35. The differences of the pump flow between the original EPANET model and the modified Hazen-William and pump data model .............................................53 Figure 36. The differences of the DMAs pressures between the original EPANET model and the modified Hazen-William and pump data model ..................................54 Figure 37 Historical demand data of the DMAs from 01/08/2007 to 07/08/2007.......57 Figure 38. FINESSE software architecture and main functions ..................................59 Figure 39. FINESSE network window ........................................................................60 Figure 40. FINESSE result window.............................................................................60 Figure 41. the simplified model of Oldham water supply system ...............................69 Figure 42. The variation of the head of the Woodgate SR 1 and 2..............................69 III

Figure 43. The leakage nodes in the simplified model ................................................70 Figure 44. The hydraulic and power characteristics for the pump stations in Oldham water supply area .........................................................................................................72 Figure 45 System model of network scheduler............................................................95 Figure 46. System model of GAMS and solver...........................................................96 Figure 47 The structure of GAMS program, i.e. its compilation of control from data to equations and solution. Also the mapping of simulation data to GAMS data and problem formulation to GAMS equations. ..................................................................97 Figure 48 FINESSE: user interface of network scheduler.........................................101 Figure 49. FINESSE: visualisation of optimisation results .......................................101 Figure 50 User interface for schedules discretiser software ......................................104 Figure 51. Continuous and discrete pump schedule ..................................................110 Figure 52. Continuous and discrete reservoir trajectories .........................................113 Figure 53 Valves flow of continuous and discrete schedules ....................................114 Figure 54. Current operation, rules and optimal schedule pump flows .....................117 Figure 55 Current operation, rules and optimal schedule reservoir trajectories ........120 Figure 56. Current operation, rules and optimal schedule valve flow of the scheduled valves .........................................................................................................................121 Figure 57 Current operation, rules and optimal schedule sources flow.....................123 Figure 58 Current operation, rules and optimal schedule leakage flow ....................124 Figure 59. Continuous and discrete pump schedule ..................................................132 Figure 60. Continuous and discrete reservoir trajectories .........................................135 Figure 61 Valves flow of continuous and discrete schedules ....................................136 Figure 62. Current operation, rules and optimal schedule pump flows .....................139 Figure 63 Current operation, rules and optimal schedule reservoir trajectories ........142 Figure 64. Current operation, rules and optimal schedule valve flow of the scheduled valves .........................................................................................................................143 Figure 65 Current operation, rules and optimal schedule sources flow.....................145 Figure 66 Current operation, rules and optimal schedule leakage flow ....................146 Figure 67. Continuous and discrete pump schedule ..................................................154 Figure 68. Continuous and discrete reservoir trajectories .........................................157 Figure 69 Valves flow of continuous and discrete schedules ....................................158 Figure 70. Current operation, rules and optimal schedule pump flows .....................161 Figure 71 Current operation, rules and optimal schedule reservoir trajectories ........164 Figure 72. Current operation, rules and optimal schedule valve flow of the scheduled valves .........................................................................................................................165 Figure 73 Current operation, rules and optimal schedule sources flow.....................167 Figure 74 Current operation, rules and optimal schedule leakage flow ....................168 Figure 75. Continuous and discrete pump schedule ..................................................176 Figure 76. Continuous and discrete reservoir trajectories .........................................179 Figure 77 Valves flow of continuous and discrete schedules ....................................180 Figure 78. Current operation, rules and optimal schedule pump flows .....................183 Figure 79 Current operation, rules and optimal schedule reservoir trajectories ........186 Figure 80. Current operation, rules and optimal schedule valve flow of the scheduled valves .........................................................................................................................187 Figure 81 Current operation, rules and optimal schedule sources flow.....................189 Figure 82 Current operation, rules and optimal schedule leakage flow ....................190 Figure 83. Continuous and discrete pump schedule ..................................................198 Figure 84. Continuous and discrete reservoir trajectories .........................................201 Figure 85 Valves flow of continuous and discrete schedules ....................................202

IV

Figure 86. Current operation, rules and optimal schedule pump flows .....................205 Figure 87 Current operation, rules and optimal schedule reservoir trajectories ........208 Figure 88. Current operation, rules and optimal schedule valve flow of the scheduled valves .........................................................................................................................209 Figure 89. Continuous and discrete pump schedule ..................................................218 Figure 90. Continuous and discrete reservoir trajectories .........................................221 Figure 91 Valves flow of continuous and discrete schedules ....................................222 Figure 92. Current operation and optimal schedule pump flows...............................225 Figure 93 Current operation and optimal schedule reservoir trajectories..................228 Figure 94. Current operation and optimal schedule valve flow of the scheduled valves ....................................................................................................................................229 Figure 95 Current operation and optimal schedule sources flow ..............................230 Figure 96. Continuous pump schedule.......................................................................239 Figure 97. Continuous reservoir trajectories..............................................................242 Figure 98 Valves flow of the continuous schedule....................................................243 Figure 99 sources flow of the continuous schedule ...................................................244 Figure 100. Continuous and discrete pump schedule ................................................250 Figure 101. Continuous and discrete reservoir trajectories .......................................253 Figure 102 Valves flow of continuous and discrete schedules ..................................254 Figure 103. Current operation, rules and optimal schedule pump flows ...................257 Figure 104 Current operation, rules and optimal schedule reservoir trajectories ......260 Figure 105. Current operation, rules and optimal schedule valve flow of the scheduled valves .........................................................................................................................261 Figure 106 Current operation, rules and optimal schedule sources flow...................263 Figure 107 Current operation, rules and optimal schedule leakage flow ..................264 Figure 108. Continuous and discrete pump schedule ................................................272 Figure 109. Continuous and discrete reservoir trajectories .......................................275 Figure 110 Valves flow of continuous and discrete schedules ..................................276 Figure 111. Current operation, rules and optimal schedule pump flows ...................279 Figure 112 Current operation, rules and optimal schedule reservoir trajectories ......282 Figure 113. Current operation, rules and optimal schedule valve flow of the scheduled valves .........................................................................................................................283 Figure 114 Current operation, rules and optimal schedule sources flow...................285 Figure 115 Current operation, rules and optimal schedule leakage flow ..................286 Figure 116. Continuous and discrete pump schedule ................................................294 Figure 117. Continuous and discrete reservoir trajectories .......................................297 Figure 118 Valves flow of continuous and discrete schedules ..................................298 Figure 119. Current operation, rules and optimal schedule pump flows ...................301 Figure 120 Current operation, rules and optimal schedule reservoir trajectories ......304 Figure 121. Current operation, rules and optimal schedule valve flow of the scheduled valves .........................................................................................................................305 Figure 122 Current operation, rules and optimal schedule sources flow...................307 Figure 123 Current operation, rules and optimal schedule leakage flow ..................308 Figure 124. Continuous and discrete pump schedule ................................................316 Figure 125. Continuous and discrete reservoir trajectories .......................................319 Figure 126 Valves flow of continuous and discrete schedules ..................................320 Figure 127. Current operation, rules and optimal schedule pump flows ...................323 Figure 128 Current operation, rules and optimal schedule reservoir trajectories ......326 Figure 129. Current operation, rules and optimal schedule valve flow of the scheduled valves .........................................................................................................................327

V

Figure 130 Current operation, rules and optimal schedule sources flow...................329 Figure 131 Current operation, rules and optimal schedule leakage flow ..................330 Figure 132. Continuous pump schedule.....................................................................341 Figure 133. Continuous reservoir trajectories............................................................344 Figure 134 Continuous valve schedule ......................................................................345 Figure 135 optimal schedule sources flow.................................................................346 Figure 136. Continuous control, speed and flow pump schedule ..............................353 Figure 137. Continuous reservoir trajectories............................................................356 Figure 138 Continuous valve schedule ......................................................................357 Figure 139. Continuous pump schedule at different LRC .........................................368 Figure 140. Continuous reservoir trajectories at different LRC ................................371 Figure 141 Continuous valve schedule at different LRC...........................................372 Figure 142 Sources flows at different LRC ...............................................................373

VI

List of Tables Table 1. United Utilities suggested table to summarise the results .............................14 Table 2. UU's electricity power consumption 2007/2008..............................................1 Table 3. Water sources of Oldham supply system.........................................................6 Table 4 Data of the service reservoirs in Oldham supply system◙ ................................8 Table 5. Shapes, physical dimensions and structure of the reservoirs of Oldham water supply system* ...............................................................................................................9 Table 6 Data of the pump station in Oldham water supply system .............................10 Table 7. Electricity (Seasonal Time of Day and Trading cost) prices 2008/2009.......16 Table 8 Pump stations to be scheduled ........................................................................18 Table 9 The water treatment works costs and limits....................................................18 Table 10 Scheduled valves setting...............................................................................19 Table 11 the scenarios of lower and upper reservoir limits* ........................................19 Table 12. Oldham model summary..............................................................................22 Table 13 Summary of the fixed head reservoirs data ..................................................25 Table 14 Variable head reservoirs data........................................................................28 Table 15. Pipe headloss formulas ................................................................................31 Table 16. The pipe diameter and total length data.......................................................31 Table 17. Pump stations data as represented in the EPANET model ..........................35 Table 18. Pump stations data as represented in the EPANET model ..........................36 Table 19. Data of the valves included in the Oldham supply system model ...............41 Table 20. Minor losses setting of the valves included in the model of Oldham water supply system...............................................................................................................43 Table 21 Variable reservoir fill valve and control levels.............................................46 Table 22 Reservoir initial water level ..........................................................................58 Table 23 The number of components in the full and simplified model.......................68 Table 24 Tank characterstics .......................................................................................73 Table 25 Reservoirs levels and constraints in [m] .....................................................105 Table 26 Reservoir mass balance [Ml] ......................................................................124 Table 27 Pumping Energy and cost ...........................................................................125 Table 28 Sources flow and cost .................................................................................126 Table 29 Leakage flow [Ml] ......................................................................................126 Table 30 Reservoir mass balance [Ml] ......................................................................146 Table 31 Pumping Energy and cost ...........................................................................147 Table 32 Sources flow and cost .................................................................................148 Table 33 Leakage flow [Ml] ......................................................................................148 Table 34 Reservoirs levels and constraints in [m] .....................................................149 Table 35 Reservoir mass balance [Ml] ......................................................................168 Table 36 Pumping Energy and cost ...........................................................................169 Table 37 Sources flow and cost .................................................................................170 Table 38 Leakage flow [Ml] ......................................................................................170 Table 39 Reservoir mass balance [Ml] ......................................................................190 Table 40 Pumping Energy and cost ...........................................................................191 Table 41 Sources flow and cost .................................................................................192 Table 42 Leakage flow [Ml] ......................................................................................192 Table 43 Reservoir mass balance [Ml] ......................................................................210 Table 44 Pumping Energy and cost ...........................................................................211 Table 45 Sources flow and cost .................................................................................212 Table 46 Reservoirs levels and constraints in [m] .....................................................213 VII

Table 47 Reservoir mass balance [Ml] ......................................................................231 Table 48 Total Pumped Flow.....................................................................................231 Table 49 Pumping Energy and cost ...........................................................................232 Table 50 Sources flow and cost .................................................................................233 Table 51 Reservoirs levels and constraints in [m] .....................................................245 Table 52 Reservoir mass balance [Ml] ......................................................................264 Table 53 Pumping Energy and cost ...........................................................................265 Table 54 Sources flow and cost .................................................................................266 Table 55 Leakage flow [Ml] ......................................................................................266 Table 56 Reservoir mass balance [Ml] ......................................................................286 Table 57 Pumping Energy and cost ...........................................................................287 Table 58 Sources flow and cost .................................................................................288 Table 59 Leakage flow [Ml] ......................................................................................288 Table 60 Reservoirs levels and constraints in [m] .....................................................289 Table 61 Reservoir mass balance [Ml] ......................................................................308 Table 62 Pumping Energy and cost ...........................................................................309 Table 63 Sources flow and cost .................................................................................310 Table 64 Leakage flow [Ml] ......................................................................................310 Table 65 Reservoir mass balance [Ml] ......................................................................330 Table 66 Pumping Energy and cost ...........................................................................331 Table 67 Sources flow and cost .................................................................................332 Table 68 Leakage flow [Ml] ......................................................................................332 Table 69. Reservoir mass balance [Ml] .....................................................................358 Table 70 Total pumped flow [Ml] .............................................................................358 Table 71 Pumping cost [£].........................................................................................359

VIII

Introduction

Energy and Pressure Management of Oldham Network United Utilities case study Introduction Water distribution systems, despite operational improvements introduced over the last 10-15 years, still lose a considerable amount of potable water from their networks due to leakage, whilst using a significant amount of energy for water treatment and pumping. The objective of applying energy and pressure managements in water supply systems can be defined as, “supply demands at minimum cost while satisfying operating constraints”. The principal costs involved are usually electricity charges for pumping, treatment costs and charges for imported water. This objective can be achieved by reducing the leakage in the mains (pressure control aspect) and minimising the electric power consumption by the pump stations (optimal pump scheduling aspect). Reduction of leakage, hence savings of clean water, can be achieved by introducing pressure control algorithms (Ulanicki et al. 2008; Ulanicki et al. 2000). Reducing electricity power cost by pump stations can be achieved by increasing electrical efficiency of the drivers and the frequency converters and the mechanical efficiency of the pumps, and scheduling of pumps so that they operate on the off-peak period (cheap electricity tariff) and closer to the best efficiency point. Optimisation of pump schedules and algorithms for control of pressure are traditionally considered separately. However, if the pressure reducing valve (PRV) inlet pressure is higher than required, in many networks it could be reduced by adjusting pumping schedules in the upstream part of the network. Modern pumps are often equipped with variable speed drives, therefore, the pressure could be controlled by manipulating pump speed, thus reduce leakage and energy use. Furthermore, taking into account the presence of pressure-dependent leakage whilst optimising pumps operation is likely to influence the obtained schedules Optimal pressure control, pump scheduling have been investigated by researchers over many years and different approaches have been developed for each aspect separately. The state of art of this work is developing and applying a new method for combined energy and pressure management via coordination of pumps operation with pressure control aspects. The developed methodology is demonstrated on a mediumscale water distribution system.

Operational Control of Water Supply Systems The aim of this section explains how the FINESSE modelling package can be used for operational control of water supply and distribution systems. It describes the major features of water supply and distribution systems. A typical operational control structure for water systems is described. Data exchange between FINESSE and other applications is described. The application functions of FINESSE are discussed in with possible scenarios for using FINESSE in a control room.

Description of Water Supply and Distribution Systems A water supply and distribution system provides clean water to industrial and domestic users. Water is taken from rivers and impounding reservoirs (surface resources), or from boreholes (underground resources). Then the water is purified in treatment works by physical and chemical processes. The clean water is then pumped 1

Introduction into a supply network of pipes. It may be stored in service reservoirs and it is distributed to consumers through distribution pipe networks. Distribution networks may have topologies of grids, branches or a combination of these. The grid topology is advantageous in that any point can be supplied from at least two directions Supply networks typically have a branch topology. Water is transported through the pipenetworks under pressure derived from the force of gravity and pumps. Service reservoirs are an important component of most water distribution systems. The stored water is necessary to meet the fluctuating demands and to equalise operating processes. Some reservoirs are located on natural heights or towers to maintain pressure in their neighbourhood network. The appropriate storage policy is a key issue of operational control and water can be stored in reservoirs in periods of cheap electricity (off peak hours) and used to augment supplies during peak hours.

Components of Water Supply and Distribution Systems The three main parts of the system are: ™ treatment works. ™ supply network of large pipes (“trunk mains”) and main reservoirs. ™ distribution network of small diameter pipes and local reservoirs. Figure 1shows a typical water system including water abstraction, treatment plants and the supply network with main reservoirs. The distribution networks are simplified to arrows from the main reservoirs. The following sections provide a generalised description of a water supply and distribution system. However, there are significant variations in design and operation to suite local circumstantiates. For example, in some systems water is transported under the force of gravity, whereas others require significant energy inputs from pumps. Treatment works processes vary according to the type of source and quality of raw water. Typically borehole treatment processes are much simpler than river-water treatment processes. The interactions between the supply part and distribution part of the system occur where water is pumped, or gravitates from trunk mains into local networks. The main elements of the system are described in more detail in the following sections.

Treatment Works Raw water is screened to remove coarse particles and is also be monitored for quality. It may be stored in raw water reservoirs, which provide a buffer between variations in the source and works input. Storage can also improve raw water quality through settling. Screened raw water is pumped or runs by gravity through successive stages of filtration and chemical treatment to emerge as drinking water, which is stored at the works output in a treated water reservoir. It is then pumped into the supply network by 'high-lift' pumps. The major chemicals used in the treatment process are all delivered in bulk to one area and stored in silos or tanks. All of the plant treatment processes are controlled automatically through closed loops.

2

Introduction

WTW X Tank 1

Reservoir 1

WTW Y

PS 1 Tank 2

1

To the area A

To the area B

PS 2 PS 3 To the area C

Reservoir 2 PS 4

To the area D Figure 1 A typical water supply and distribution system

Pipes and Valves Pipes convey water from sources to users. Pressurised water moves along pipes under gravity or energy supplied by pumps. Pipes can be up to several meters in diameter in the supply part of a network. In the distribution network pipes are smaller and gradually decrease in size down to 0.05m diameter in individual houses. The energy provided by gravity or pumps is dissipated in the pipes. Physically, a pipe section constitutes an analogue to electrical resistance described by head-loss versus flow characteristics. Points of connection between pipes are called nodes of the network. 3

Introduction There are other nodes where reservoirs or demands are located. Flows between different parts of the network are controlled by valves thus providing flexibility in daily system operation. The valves can control both flow and pressure, or even network structure by closing some routes and opening others. Pressure reducing valves adjust pressure to meet consumer needs and to prevent excessive pressures and leakage. Flow reducing valves distribute water between different parts of the network as required by operational conditions. There are also valves to control of flow direction (non-return valves).

Pumps Pumps are active elements to supply water to required elevations and to cover energy losses in pipes and valves. Centrifugal pumps are the most widely used. They have a rotating element which imparts energy to the water. Energy is supplied by electrical motors and is transformed into “water energy” which is often expressed as “head” with units of length. The efficiency transformation is described by a co-efficient called pump efficiency. From the hydraulic point of view, a pump is described by head-increase versus flow characteristic. There are two types of pump from a control point of view including: variable speed pumps (VSP) in which the speed of an electrical motor can be subjected to external control signals, and fixed speed pumps (FSP) with speed fixed at a constant value. The latter are in wider use, especially at heavy duty stations. Pumps are configured in pump stations to provide the desired operation conditions and reliability. If a pump station is composed of FSP pumps, the only control factor available to a system operator is a pump configuration. In the case of VSP pumps there are two control factors, speed, which can change continuously and pump configuration, which is a discrete variable.

Reservoirs Reservoir storage enhances system flexibility by providing a buffer between variations in water supply and demand. It also allows a shift in periods of heavy pumping and high demands in order to reduce pumping cost. Another important function of a reservoir is to sustain pressure in a neighbourhood network. Reservoir capacity can vary enormously from relatively small volumes in water towers to large volumes in ground level reservoirs. The relationship between reservoir depth and volume is often proportional, but some have varying cross-sectional areas and the relationship is more complicated. Reservoirs are usually equipped with special valves to avoid overflowing or emptying.

Operational control of Water Supply and Distribution Systems Operational control of a water system requires co-ordination of the 3 interacting parts, i.e. the retention and treatment, supply and distribution sub-systems. Typically each of these sub-systems is considered as a self-contained system for control purposes with boundary conditions to deal with the interactions. Water systems are spatially distributed and special communication links are required in order to connect remote system areas with the control room. In the past, control consoles contained a substantial number of analogue instruments and indicators difficult to manage by a single person. Now, computer and graphical monitors replaced the analogue instruments, improving significantly compactness of the operator consoles.

4

Introduction For small and medium water networks there is typically one master computer connected to a number of intelligent remote outstations. More control centres are necessary as the scale of the water system increases. Features of Water Treatment Works A treatment works is a process plant with a relatively high level of monitoring and control and dedicated communications network. Many variables are monitored at frequent intervals including flows, tank levels, water quality parameters etc. The key variables are updated continuously on operator consoles. Other variables are monitored but only reported at thresholds (i.e. alarms). The plant contains many local control loops, but from the point of view of the control of an entire water system, the important relationship is between the intake flow from the source (cause) and the outflow (effect). When viewed as an input/output model, the treatment processes impose some operational constraints on the control problem including: ƒ minimum and maximum output flow at any time. ƒ total volume over 24 hours. ƒ rate of change in the set points (i.e. or step-change).

Features of Water Supply Systems Water supply systems share common features that are important from the operational control point of view: 1. Relatively simple network structure with a limited number of connections. 2. Pipes of large diameter to transport bulk quantities of water. 3. Large pump stations composed of numbers of high-lift pumps. 4. Interactions with the distribution part of the system can be modelled as Demands and can be predicted with reasonable accuracy. 5. Systems flows are often insensitive with respect to reservoir level variations. The pipe system has relatively spare measurement and control and, typically, only key flows and pressures are monitored. Pump stations are more densely instrumented with local control loops and monitoring points. Frequently monitored variables include flow, input and output pressures, pump power, pump-operating status (e.g. speed, temperature etc). Stations are usually unmanned and have a dedicated communications links back to a control room. Frequently monitored variables at reservoirs include tanks levels, inlet and outlet flows. There may be digitally controlled valves inlets and outlets, which are also monitored. However, some rely on local control loops (electronic or mechanical). Reservoirs are usually unmanned and may have a dedicated communications link back to a control room.

Features of Water Distribution Systems The following features of water distribution systems are important from an operational control point of view: 1. Complicated network structure with hundreds of connections and many pipeloops. 2. A typical zone contains one service reservoir to sustain supplies and maintain pressures. 3. Reservoir level variations may have significant impact on flows and pressures of the system. 4. Elements such as booster pumps and control valves to control local conditions. The distribution system has relatively spare measurements and control. Typically only a few key flows and pressures are monitored frequently. Local reservoirs are monitored in a similar fashion to those in the supply network. Control elements such 5

Introduction as booster pumps and valves have local control loops and may not be monitored continuously. There are few dedicated communications links. Although measurements may be logged locally at frequent intervals, the time-series may only be downloaded to the control room a few times a day or when alarms occur. Often public communications networks are used. Operators in the field download some loggers. There are often temporary meters and loggers installed to investigate specific features of operation.

Control and Decision Structures for Water Supply and Distribution Systems Complex water systems composed of many sub-systems require an adequate control structure. The water network is divided into treatment, supply and distribution parts. The distribution part can include many sub-systems with well-defined boundary flows where each area has dense network of pipes supplying water to customers. The lower layer of the decision structure directly interacts with the physical system by a distributed telemetry system: applies control decisions to the system and collects measurements. An operator in a local control room aided by appropriate hardware and software assesses the sub-system behaviour and sends essential information to the coordination level. The responsibilities of the operator can vary from following orders from the upper level to solving some parameterised sub-problems. Schedules for major control elements calculated by the co-ordination level are based on abstract mathematical models and the local operator has to convert them into direct control action taking into account detailed physical layouts of the control elements. He also decides the policy for smaller local pumps and valves. Typically a computer model of the water network is the basic tool for the operator so the control decisions before being applied to the physical system are verified by this model. The co-ordination level has a global view of the systems. It works according to the two time horizons. Calculation of the storage policy for one week ahead sets up reservoir level targets for each day. After that the storage policy and schedules for major elements are evaluated for the next day. The co-ordination level uses sophisticated software tools such as demand predictor, state estimator and optimal scheduler described previously. The figure below shows a decision and control structure for a complex water system.

6

Introduction

Figure 2. schematics of control and Decision Structures for Water Supply and Distribution Systems

Pressure control in water distribution systems Water loss form the water distribution systems (WDSs), has long been a feature of the WDS operations management, and occurs in all WDSs. Leaks are expensive for a variety of reasons, including the loss of water and treatment chemicals, the increased risk of water quality deterioration, unnecessary capacity expansion, and the increased energy expenditure required to feed the leaks. The quantity of loss varies and depends on the physical characteristics of the pipe network, operating factors and parameters, and the level of technology and expertise applied to controlling this loss. In Addis Ababa, Ethiopia, nearly 50% of the produced water is lost (Desalegn 2005). The same level is reported for the city of Mutare, Zimbabwe (Marunga et al. 2006). As well, many Asian cities, Non Revenue Water is equivalent to 46% of which over 75% is real losses (Rogers 2005). And in UK, the water utilities recorded the total loss is over 23% (3575 Ml/d) of the total input (OFWAT 2006). Reducing and controlling water loss is becoming very important issue in this age of rapidly growing demand and relative abundance to one of relative scarcity of the 7

Introduction water resources, and climate changes that bring droughts to many locations over the world (Bergkamp et al. 2003; Houérou 1996). And due to today’s high water production, treatment and transmitting costs and rates, many water utilities have been developing new strategies to minimise losses to an economic and acceptable level in order to preserve valuable water resources and to minimise operating expenditures. The options available to reduce the leakage can be represented diagrammatically in Figure 3. This shows that leakage can be reduced by reducing pressure on the system, improving the speed of detection, location and repair of leaks and also by infrastructure improvements. Water companies undertake a mixture of these complimentary actions. general pipe rehabilitation is the most costly and long term action, but is undertaken to improve a number of different factors including leakage and water quality. Operational pressure management is a cost-effective means of reducing leakage over whole sub-networks, and for reducing the risk of further leaks by smoothing pressure variations. Pressure management also has other important benefits in addition to the reduction of existing leakage.

Figure 3. schematic representation of leakage reducing options

Pump scheduling in water distribution systems Optimal pump scheduling has been investigated by researchers over many years and different approaches have been developed. A number of methods have been proposed or developed to establish pump-scheduling processes in order to optimize the scheduling of pump run times. These methods are intended primarily to leverage timeof-use tariff structures in order to minimize energy costs. Time-of-use tariffs reflect the real-time supply-and-demand market in energy pricing. When demand is high, the tariff is also high. This typically occurs when heating or air-conditioning loads are high. At nighttimes, when demand for electricity is low, the tariff is also low. Water utilities are in an advantageous position to maximize pumping to storage during off-

8

Introduction peak hours when energy rates are lowest, and then the utilities draw on water in the storage tanks when energy costs are high. Due to the complexity of most water distribution systems and the multiple production requirements and operational constraints, it is generally considered very difficult to explicitly solve the scheduling problem using mathematical techniques. A brief review of the developed methodologies and techniques used to solve a typical pumps scheduling problem will be provided in the following section. (Miyaoka and Funabashi 1984) introduced a modelling technique and an optimal control scheme for water distribution networks. To overcome the large scale and nonlinearity of the network, a network aggregation method and a two-level control scheme are developed. The first level of the scheme decides operating pints using a nonlinear optimization method, where the pressure/flow equations are solwd using a high-speed technique derived from network flow theory. The second level is a feedback control around the operating points, which absorbs estimate error and small variations in consumption. (Lansey and Awumah 1994; Ormsbee and Lansey 1994) evaluated the state of the art to that date in optimization techniques for pumping systems. These researchers used relatively well-known linear and nonlinear programming and dynamic programming tools. (Klernpous et al. 1997) discussed a multilevel algorithm for finding optimal control in a static distribution system based on the idea of aggregation technique. The mathematical model of this system was presented with its elements as well as two basic algorithms. The first is a simulation algorithm of the pipeline network and the other is an algorithm for finding an optimal control at the pumping station. (Savic et al. 1997; Sotelo et al. 2002) presentd several improvements of the single objective genetic algorithm (GA) and the results of the further investigation into the use of multiobjective GAs for solving the pump scheduling problem. A multiobjective approach used in this work deals with both the energy cost and the pump switching criterion, at the same time. The performance of the algorithm is tested for different demand profiles and additional requirements and compared to that of the single objective GA Also, (Barán et al. 2005) used multiobjective Evolutionary Algorithms (MOEAs) to solve an optimal pump-scheduling problem with four objectives to be minimized: electric energy cost, maintenance cost, maximum power peak, and level variation in a reservoir. Six different MOEAs were implemented and compared. In order to consider hydraulic and technical constraints, a heuristic algorithm was developed and combined with each implemented MOEA. Evaluation of experimental results of a set of metrics shows that the Strength Pareto Evolutionary Algorithm achieves better overall performance than other MOEAs for the parameters considered in the test problem, providing a wide range of optimal pump schedules to chose from. (López-Ibáñez et al. 2005a; López-Ibáñez et al. 2005b) considerd the pump scheduling problem using a multi-objective approach and showed the viability of a multi-objective approach for solving the Pump Scheduling problem, which allows the system operator to examine a range of Pareto-optimal solutions and choose one solution with regard to additional criteria. (López-Ibáñez et al. 2007) applied Max-Min Ant System to solve the pump scheduling problem. Instead of the typical binary representation, a representation based on time-controlled triggers was used. Ant colony optimization was introduced and applied to water distribution system design and operations by (Maier et al. 2003; Ostfeld and Tubaltzev 2008; Zecchin et al. 2007; Zecchin et al. 2005). A method for near-optimal real-time on-line operation of urban water distribution system was presented and demonstrated by (Shamir and Salomons 2008). It uses a reduced model of the network which reproduces its performance over time with high fidelity with optimization by a

9

Introduction genetic algorithm. (Wang et al. 2009) proposed a genetic algorithm-based pump scheduling method for not only cost reduction but also environment protection. The proposed method can achieve lower pumping cost and provide a wider range of ecoaware schedules. The experimental results also suggest that the proposed method may be extended to other similar optimization problems. Optimal scheduling is a complex task as it includes the extended period hydraulic model represented by differential algebraic equations and mixed-integer decision variables. Obtaining a strictly optimal solution involves excessive computational effort. (Bounds et al. 2006; Ulanicki et al. 2007) presented a new dynamic optimization approach to solve large scale optimal scheduling problems for water distribution networks, which is significantly faster than existing approaches. The proposed method progresses in two stages initially a relaxed continuous problem is solved and in the second stage, a mixed-integer solution is found which tracks the optimal reservoir trajectories by time decomposition and application of a local branch and bound method.

Scheduling multi-source, multi reservoir systems Water companies practise one or more of leakage management strategies, which are general pipe rehabilitation, direct detection and repair of existing leaks, and operational pressure management. General pipe rehabilitation is the most costly and long term action, but is undertaken to improve a number of different factors including leakage and water quality (Clark et al. 2002; Engelhardt et al. 2000). Direct detection and repair of existing leaks is one of the most powerful policies, that is used to prevent the high level leakage from burst, but it is not economically efficient in background leakage. the expense of detecting and reducing leakages is an attractive solution, and many algorithms were developed to predict, detect the location and quantify the leakage in WDSs (Koppel et al. 2007; Mounce et al. 2003; Wu and Sage 2006). Operational pressure management is a cost-effective method for leakage reduction over whole DMAs, and for minimizing the risk of further leaks by smoothing pressure variations. Many researchers ware presented, developed and implemented various methods and algorithms to optimise the operational pressure, and the results showed that, the leakage can be reduced by up to 60%. (Miyaoka and Funabashi 1984) developed and implemented an optimal pressure regulation control and achieved an average rate of the leakage reduction is about 22%, but the leakage model was not considered. (Jowitt and Xu 1990) developed a linear algorithm for the determination of control valve settings to minimize leakage. The numerical results exhibited that the overall reduction in leakage is about 20%. (Vairavamoorthy and Lumbers 1998) described an optimization method to minimize leakage in water distribution systems through the most effective settings of flow reduction valves and reduced the water leakage substantially. (Alonso et al. 2000) presented a parallel computing using the discrete volume element method for the leakage minimisation, which was reduced by 25% to 60% of its original values. (Burn et al. 2002) analysed the effect of employing pressure management techniques on the cost of WDSs, which increases the savings by a further 20-55%. (Girard and Stewart 2007) implemented the pressure and leakage management strategies on the Gold Coast, Australia, and the results revealed that a good opportunity to achieve significant water savings. (Marunga et al. 2006) implemented a pressure management as a leakage reduction, in Mutare, Zimbabwe. The results showed that an operating pressure reduction from 77 m to 50 m resulted in 25% reduction in the total leakage. (Ulanicki et al. 2008) developed a fast and efficient method to calculate time schedules and flow modulation 10

Introduction curves for boundary and internal PRVs, and the results showed a potential leakage reduction of 30% and 45% for different two case studies. Reservoirs play an important role within the water resources management framework. Finding optimal operating policies for a reservoir system has been a major area of study in water resources systems. An operating policy is a set of rules for determining the quantities of water to be stored and to be released under various conditions. Among the available techniques for reservoir optimal operation, the most well known one is stochastic dynamic programming (SDP). In recent years, artificial intelligence techniques such as genetic algorithms (GA) and artificial neural networks have arisen as an alternative to overcome some of the limitations of traditional methods. Some of these limitations are related to the difficulty in combining SDP with other simulation and prediction models, due to the increase in the number of decision and state variables, and the error resulting from the rough discretization of these variables. In the following paragraph a optimization methodologies and techniques of multi-source, multi-reservoir water system is discussed. (Brdys et al. 1988) developed an optimization approach to cover multi-source, multireservoir systems is based on Lagrangean relaxation of the hydraulic constraints which couple the pump stations to the network. The method was applied to a practical multi-source, multi-reservoir systems. (Nalbantis and Koutsoyiannis 1997) introduced and formulated a parametric rule for multi-reservoir system operation. It is a generalization of the well-known space rule to simultaneously account for various system operating goals in addition to the standard goal of avoiding unnecessary spills, including: avoidance of leakage losses, avoidance of conveyance problems, the impact of the reservoir system topology, and assurance of satisfying secondary uses. (Bessler et al. 2003) described and developed a general operating policy for a water supply system using the methodology of data mining. To define an operating policy using this approach, both a single-reservoir and a multi-reservoir water system were modeled and optimized for a set of historical inflows data. (Cervellera et al. 2006) presented a numerical solution to a 30-dimensional water reservoir network optimization problem, based on stochastic dynamic programming. In such problems the amount of water to be released from each reservoir is chosen to minimize a nonlinear cost (or maximize benefit) function while satisfying proper physical constraints. (Momtahen and Dariane 2007) proposed the direct search approach to determine optimal reservoir operating policies with a real coded genetic algorithm as the optimization method. (Chaves and Kojiri 2007) introduced a new approach for system optimization and operation, named stochastic fuzzy neural network (SFNN), which can be defined as a neuro-fuzzy system that is stochastically trained optimized by a GA model to represent the system operational strategy. (Kim et al. 2007) presented optimization techniques for enhancing reservoir operations, which use sampling stochastic dynamic programming (SSDP) with ensemble stream flow prediction (ESP). SSDP used with historical inflow scenarios (SSDP/Hist) derives an off-line optimal operating policy through a backward-moving solution procedure. In contrast, SSDP used with monthly forecasts of ESP (SSSDP/ESP) re-optimizes the off-line policy. These stochastic models were used to derive a monthly joint operating policy.

Aim and objectives Optimal pressure control, pump scheduling, and operating policies of the multisource, multi-reservoir system have been investigated by researchers over many years and different approaches have been developed. The scheduling problem is nonlinear, 11

Introduction dynamic and mixed-integer and is numerically hard to solve. However, it lends itself well to different simplifying assumptions to calculate near-optimal solutions. The current solutions are applicable to grid systems, but significant research is required to link these to pressure control aspects (i.e., valve scheduling and leakage reduction) which can be selectively applied, especially in those parts of the system with direct pumping. The main objective of energy and pressure managements in water supply systems is to minimise the operating cost in terms of electricity charges for pumping, treatment costs and charges for imported water while satisfying operating constraints. The relevant operating constraints typically include the following: ¾ Normal maximum and minimum water levels in reservoirs ¾ Maximum and minimum allowable pressures in the mains or distribution pipes ¾ Patterns of demand which must be satisfied ¾ Power supply limitations ¾ Constraints on the combinations of pumps, which can be run together ¾ Maximum and minimum time of the pump operation, ON/OFF time ¾ Maximum and minimum speeds for the variable speed pumps

Aim ¾ To develop innovation in pressure management to deliver key leakage and energy savings ¾ To provide options to make significant savings in energy e.g. through pump schedule optimisation ¾ To develop a tool facilitating the off-line energy and pressure management in the grid part and in the distribution part of the water system.

Objectives ¾ To develop the concept of integrated energy and pressure management through optimal scheduling ¾ To develop a methodology for and to implement a general purpose optimal scheduling module.

Methodology In this report, integration algorithm between pump scheduling for energy management and DMA pressure control for leakage reduction will be provided. Traditionally these tasks are considered separately but if the pressure reducing valve (PRV) inlet pressure is higher than required, in the cases when this is possible (i.e. when there is not an intermediate distribution reservoir at the upstream side) it can be reduced by adjusting pumping schedules in the upstream part of the network. Many modern pumps are equipped with variable speed drives and manipulating speed is a very efficient method of reducing energy use - the power consumption changes with the cube of the speed. The proposed approach is based on nonlinear mixed-integer programming and novel local search ideas supported by heuristics derived from numerous industrial casestudies. The new challenge in this algorithm will be to include simplified models of DMAs and also to reduce the calculation time and improve robustness of the algorithms in order to satisfy real-time requirements. The module will calculate time schedules for treatment works, pumps, valves and reservoirs. It can allow control rules to be derived for the transmission system by running different hypothetical scenarios and synthesising these rules.

12

Introduction

The proposed method for combined energy and pressure management, based on formulating and solving an optimisation problem, is an extension of the pump scheduling algorithms described in (Bounds et al. 2006). The method involves utilisation of a hydraulic model of the network with pressure dependent leakage and inclusion of a PRV model with the PRV set points included in a set of decision variables. The cost function represents the total cost of water treatment and pumping. An excessive pumping contributes to a high total cost in two ways. Firstly, it leads to high energy usage. Secondly, it induces high pressure, hence increased leakage, which means that more water needs to be pumped and taken from sources. Therefore, the optimizer, by minimising the total cost, attempts to optimize the energy usage, reduce its cost and minimise the leakage. In the optimisation problem considered some of the decision variables are continuous (e.g. water production, pump speed, and valve position) and some are integer (e.g. number of pumps switched on). Problems containing both continuous and integer variables are called mixedinteger problems and are hard to solve numerically. Continuous relaxation of integer variables (e.g. allowing 2.5 pumps on) enables network scheduling to be treated initially as a continuous optimisation problem solved by a non-linear programming algorithm. Subsequently, the continuous solution can be transformed into an integer solution by manual post-processing, or by further optimisation, see (Bounds et al. 2006). For example, the result “2.5 pumps on” can be realised by a combination of 2 and 3 pumps switched over the time step. 0describes continuous optimisation problem solved by a non-linear programming algorithm. In Section 0 a schedule discretisation algorithm that allows the user to interact with the discretisation process is described. Remark 1. An experienced network operator is able to manually transform continuous pump schedules into equivalent discrete schedules (Ulanicki et al. 2007). Optimisation methods described in this report are model-based and, as such, require hydraulic model of the network to be optimised. Such hydraulic model is usually developed in a modelling environment such as EPANET, FINESSE etc. and consists of three main components: boundary conditions (sources and exports), a hydraulic nonlinear network made up of pipes, pumps, valves, and reservoir dynamics. In order to reduce the size of the optimisation problem the full hydraulic model is simplified using module reduction algorithm. In the simplified model all reservoirs and all control elements, such as pumps and valves, remain unchanged, but the number of pipes and nodes is significantly reduced. It should be noted that the connections (pipes) generated by module reduction algorithm may not represent actual physical pipes. However, parameters of these connections are computed such that the simplified and full models are equivalent mathematically. Details about model reduction are given in Appendix D of this report. Both the non-linear programming algorithm employed to compute the continuous schedule, and also schedule discretisation method, require a simulator of the hydraulic network. Simulators are part of modelling software such as EPANET or FINESSE. In this work a FINESSE simulator called Ginas has been employed, however, other simulators could also be used and in future versions of the network scheduler EPANET simulator may be utilised.

United Utilities’ expectations United Utilities would like to know the savings over current operation that can be achieved by:

13

Introduction 1. Using local time based control rules to encourage refill overnight. This would require a simulation of these rules which could be derived by optimisation as part of the study or by using predefined rules. Rules are expected to be seasonal to reflect Summer and Winter tariffs. 2. Global optimisation. Options are: a. Daily schedule (reran if needed by exception) b. Half-hourly schedule The benefit of (b) over (a) will depend on accuracy of the demand forecaster and likelihood of system operation needing to change e.g. due to asset failure. United Utilities suggest to ignore the latter for the study. An indication of the cost difference between the 2 methods would be appreciated. This information will be used by UU in cost benefit analysis for monitoring, optimisation and control options. WTW productions should preferably remain unchanged from the baseline scenario. Clarification of baseline for the study is required, United Utilities suggest using summer and winter baselines as originally proposed. WTW limits and marginal costs provided. If there is a difference then savings against WTWs and PSs should be quantified. United Utilities expect future reservoir operation to alter to make more use of the storage: 1. Quantify the cost benefit of various minimum tank levels e.g. at 10% increments from 80% to 40% depth. 2. Quantify the associated increase in risk to supplies i.e. if reservoir levels are dropped to 40% what is likelihood of dropping to a dangerously low level (say 25%) due to demands higher than forecast (say 20%)? The benefit on water age could also be described for WQ improvements. United Utilities would like to know how the system could be applied in UU i.e. what modelling application, solution speed, how telemetry data could be linked to the model, how the model could be linked to control, safeguards for comms failure and missing/erroneous data. An understanding of how the schedule would be implemented is needed. United Utilities would like details of the demand forecaster – method, historical data required, accuracy. The results should preferably be broken down by site as below: Table 1. United Utilities suggested table to summarise the results Baseline scenario Site

Total Ml

kWh

Optimisation scenarios 1,2,3, etc. £

Total Ml

kWh

£

PS1 PS2 WTW1 Total

United Utilities would like to understand how the savings are made i.e. distinguish between 1. £ only (not kWh) savings due to managing pump times against tariffs 2. kWh (and subsequent £) savings due to more efficient pumping.

14

Introduction

ABB’s expectations 1. Have a common understanding of scope and expectations to UU case study between project partners (combined energy, flow & pressure management, source flow and reservoir level optimization, …) 2. Understand network topology, understand hydraulic model used as a base and understand how network is operated today at UU (à clear problem description and definition of starting point). 3. From documentation in report, be able to assess and understand modelling and solution approach chosen for optimizing the network as well as optimization algorithm and work steps needed (engineering complexity, appropriateness of algorithm to problem) 4. From documented results be able to understand and assess which benefits / major improvements DMU’s algorithm would bring to what UU does today (quality of results) 5. From result discussion see how much results vary with changing input parameters 6. Understand where difficulties are and where additional work might be needed

Requirements To apply and evaluate the proposed methodology, a proper case study includes the both aspects of energy and pressure management should be provided. To be able to handle the case study, there are a number of the requirements should be satisfied. A strategic hydraulic model of the network area containing the major schedulable components and representing the actual network in terms of flows, trajectories of the reservoirs, demands and pressures, have to be provided of the minimum required data as following 1. Pipe (lengths, diameters and Hazen William coefficient) 2. Node elevations 3. Node demands (domestic and fire flow, and the leakage level) 4. Nodes of water imports and exports (importing and exporting flow patterns) 5. Water treatment work plant (Elevations and capacities) 6. Reservoirs ,storage facilities (dimensions, capacities ,elevations, initial, bottom and top water levels) 7. Supply pumps and booster pumps (hydraulic characteristics) 8. Control valves (pressure and flow control valves) 9. Pressure regulating valves Other operational data needs include: 1. Water treatment work plant (capacity “maximum daily production” and elevation) 2. Pumping station capacity and design head (pump curves, i.e., head-flow curve and power/ efficiency -flow curves, control rules) 3. Electricity tariff(s) for all pumps / pumping stations 4. Reservoirs and storage tank (initial water levels, control rules) 5. Control valves (control rules) Potential data sources for the above include: 1. Water utility schematics / maps (control valve locations, pressure zones) 2. Operation personnel (institutional/operational knowledge) 15

Introduction 3. Supervisory control and data acquisition information (flow rates/demands, pump stations flow, reservoirs and tanks trajectories) Operating constraints 1. Normal maximum and minimum water levels in reservoirs 2. Maximum daily production of water treatment work plant 3. Maximum and minimum allowable pressures in the mains or distribution pipes 4. Patterns of demand which must be satisfied 5. Power supply limitations 6. Constraints on the combinations of pumps, which can be run together 7. Maximum and minimum time of the pump operation, ON/OFF time 8. Maximum and minimum speeds for the variable speed pumps

16

Conclusions

Conclusions In this report a method was described for combined energy and pressure management via integration and coordination of pump scheduling with pressure control aspects. The method is based on formulating and solving an optimisation problem and involves utilisation of a hydraulic model of the network with pressure dependent leakage. The cost function of the optimisation problem represents the total cost of water treatment and pumping. Developed network scheduling algorithm consists of two stages. First stage involves solving of a continuous problem, where operation of each pump is described by continuous variable, and utilises GAMS modelling language and CONOPT3 non-linear programming solver. Subsequently, in second stage continuous pump schedules are discretised, such that operation of each pump is described by binary variable (on/off). The case study selected was Oldham water supply system, being part of United Utilities (UU). Network topology and current operation has been described, followed by description of the process of obtaining the network model for scheduling. Developed models, both full and simplified, showed good agreement with the reference EPAnet model provided by UU. Network scheduling studies considered different leakage levels, initial reservoir levels and lower and upper reservoir constraints. A comparison has been made between the cost of the current network operation and the optimised operation. Taking into account only the costs associated with KWh usage of scheduled pumping stations and assuming summer tariff the optimised operation reduced the daily electricity cost from 11 to 35% of the current operation for different scenarios (Computed from the provided EPAnet model).

Overview of optimisation results Initially it was agreed to solve five scheduling scenarios. Following presentation of results for these scenarios and extensive discussions between DMU, UU and ABB, additional variations of the existing scenarios and also new scenarios were introduced. In total 19 different scenarios were optimised: 9 for continuous solution only and 10 for both continuous and discrete solutions. Scenarios optimised for 24h horizon, despite different lower constraints of reservoir level, resulted in comparable savings: between 58% and 66% when the flow from water treatment works (WTW) was optimised together with pump operation. It was found that in these scenarios the bottom constraint was never hit for most reservoirs, due to the constraint on final reservoir level. It is considered that the results obtained for two scenarios (further denoted as S1 and S2) optimised over one week horizon are the most relevant, for two reasons: • The optimiser could utilise full allowed capacity of the reservoirs and indeed the constraints on lower reservoir levels were hit for some reservoirs. • Reservoir mass balance was similar for the current and optimised operation. For scenario S1 the mass balance for optimised operation was smaller than that for current operation by 7 Ml, which corresponds to 1.7% of total demand. For scenario S2 the mass balance of optimised operation was larger than that of current operation by 10 Ml. The savings, compared to the current operation, for scenarios S1 and S2 were 45% and 35%, respectively, which correspond to weekly savings of £1730.84 and £1333.66, respectively. Note that in these scenarios flows from water treatment works

374

Conclusions (WTW) were not optimised. It has been demonstrated during solving of other scenarios that optimisation of flow from WTW considerably increases the savings; hence it is believed that, if WTW were scheduled for S1 and S2, the saving would be even more significant. For all scenarios not only the cost, but also the actual use of electrical energy was significantly reduced, since the tariffs are nearly flat. Furthermore, optimised operation resulted in savings in leakage between 0% and 4.5%, which correspond to up to 0.28 Ml per day, depending on scenario. Note that in the considered scenarios the cost of electrical energy is high compared to the leakage level, therefore reducing the energy cost was of higher importance during the optimisation process, compared to reducing the leakage. The savings are due to operating the network optimally, or close to being optimal. From hydraulic point of view, when compared to the current operation, the savings come mainly from: • • •

•

Running pumps at lower speed, which is more efficient for the considered water distribution system. Pumping more intensively during cheap tariff period. Use of pumps that are not used in current operation. Due to this the flow from Aqueduct is split into two paths, which results in decreased head drop, hence less energy is wasted. Optimised flow from sources. In scenarios where the sources were optimised the savings were significantly increased compared to corresponding scenarios without sources optimisation.

Three scenarios with different lower reservoirs level and the same initial reservoirs level have been solved to study the effect of changing LRC on the continuous solution. The results show that, changing the lower reservoirs constraints have a minor impact on the continuous solution of the optimization problem. Only the Pilsworth PS and Hatters PS flows have slight differences which affect the trajectory of Hatters reservoir. And there is no expected saving in both pumping energy and cost in the discrete solution By reviewing and analysing the results of 7 days optimal schedules, it has been noticed that the pumps flows and the reservoirs trajectories take acyclic patterns especially after the second day. Hence the initial reservoirs level have been modified and taken from these cyclic patterns, and the optimisation problem has been solved again but in continuous solution only to study the effect of this changing of initial reservoirs level on the cyclic or periodic stability behaviour of the pumps flows and the reservoirs trajectories. The results show that, the initial reservoirs level taken from these cyclic patterns has a big impact, and the continuous solution of the optimisation problem shows more periodic stability most of the time for both pumps flows and reservoirs trajectories.

375

References

References Alonso, J. M., Alvarruiz, F., Guerrero, D., Hernández, V., Ruiz, P. A., Vidal, A. M., Martínez, F., Vercher, J., and Ulanicki, B. (2000). "Parallel computing in water network analysis and leakage minimization." Journal of Water Resources Planning and Management, 126(4), 251-260. Barán, B., Von-Lücken, C., and Sotelo, A. (2005). "Multi-objective pump scheduling optimisation using evolutionary strategies." Advances in Engineering Software, 36(1), 39-47. Bergkamp, G., Burton, I., and Orlando, B. (2003). Change: Adaptation of Water Resources Management to Climate Change, International Union for Conservation of Nature and Natural Resources, Gland, Switzerland. Bessler, F. T., Savic, D. A., and Walters, G. A. (2003). "Water Reservoir Control with Data Mining." Journal of Water Resources Planning and Management, 129(1), 26-34. Bounds, P., Kahler, J., and Ulanicki, B. (2006). "Efficient energy management of a large-scale water supply system." Civil Engineering and Environmental Systems, 23(3), 209-220. Brdys, M., Coulbeck, B., and Orr, C. H. (1988). "A method for scheduling of multiSource, multi-Reservoir water supply systems containing only fixed speed pumps " International Conference on Control 88, Oxford, UK Brown, K. (2007). "Modeling Leakage in Water Distribution Systems," M.Sc thesis, The Florida State University, Florida. Burn, L. S., Silva, D. D., and Shipton, R. J. (2002). "Effect of demand management and system operation on potable water infrastructure costs." Urban Water, 4(1), 229-236. Cervellera, C., Chen, V. C. P., and Wen, A. (2006). "Optimization of a large-scale water reservoir network by stochastic dynamic programming with efficient state space discretization." European Journal of Operational Research, 171 (3), 139-151. Chaves, P., and Kojiri, T. (2007). "Stochastic fuzzy neural network: case study of optimal reservoir operation." Journal of Water Resources Planning and Management, 133(6), 509-518. Clark, R. M., Sivaganesan, M., Selvakumar, A., and Sethi, V. (2002). "Cost Models for Water Supply Distribution Systems." Journal of Water Resources Planning and Management, 128(5), 312-321. Desalegn, W. d. B. (2005). "Water supply coverage and water loss in distribution systems the case of Addis Ababa," M.Sc thesis, International Institute for GeoInformation Science and Earth Observation, Enschede, The Netherlands.

376

References Engelhardt, M. O., Skipworth, P. J., Savic, D. A., Saul, A. J., and Walters, G. A. (2000). "Rehabilitation strategies for water distribution networks: a literature review with a UK perspective." Urban Water, 2(2), 135-152. Germanopoulos, G. (1995). "Valve control regulation for reducing leakage." Improving Efficiency and Reliability in Water Distribution Systems, E. Cabrera and A. F. Vela, eds., Kluwer, Dordrecht , Boston , London, 165-188. Girard, M., and Stewart, R. A. (2007). "Implementation of pressure and leakage management strategies on the Gold Coast, Australia: case study." Journal of Water Resources Planning and Management, 133(3), 210-217. Giustolisi, O., Savic, D., and Kapelan, Z. (2008). "Pressure-driven demand and leakage simulation for water distribution networks." Journal of Hydraulic Engineering, 134(5), 626-635. Houérou, H. N. L. (1996). "Climate change, drought and desertification." Journal of Arid Environments, 34(2), 133-185. Jowitt, P. W., and Xu, C. (1990). "Optimal valve control in water distribution networks." Journal of Water Resources Planning and Management, 116(4), 445-462. Kim, Y.-O., Eum, H.-I., Lee, E.-G., and Ko, I. H. (2007). "Optimizing operational policies of a korean multireservoir system using sampling stochastic dynamic programming with ensemble streamflow prediction." Journal of Water Resources Planning and Management, 133(1), 4-14. Klernpous, R., Kotowski, J., Nikodem, J., and Ultasiewicz, J. (1997). "Optimization algorithms of operative control in water distribution systems." Journal of Computational and Applied Mathematics, 84(1), 81-97. Koppel, T., Ainola, L., and Puust, R. (2007). "A mathematical model for the determination of leakage in mains and water distribution networks." Estonian Journal of Engineering, 13(1), 3-16. Lansey, K. E., and Awumah, K. (1994). "Optimal pump operations considering pump switches." Journal of Water Resources Planning and Management, 120(1), 17-35. López-Ibáñez, M., Prasad, T. D., and Paechter, B. (2005a). "Multi-objective optimisation of the pump scheduling problem using SPEA2." The 2005 IEEE Congress on Evolutionary Computation, IEEE, Edinburgh, UK. López-Ibáñez, M., Prasad, T. D., and Paechter, B. (2005b). "Optimal pump scheduling: Representation and multiple objectives." 8th International Conference on Computing and Control for the Water Industry (CCWI2005), D. A. Savic, G. A. Walters, R. King, and S. Thiam-Khu, eds., University of Exeter, UK.

377

References López-Ibáñez, M., Prasad, T. D., and Paechter, B. (2007). "Solving optimal pumpcontrol problem using max-min ant system." 9th annual conference on Genetic and evolutionary computation, ACM, London, England. Maier, H. R., Simpson, A. R., Zecchin, A. C., Foong, W. K., Phang, K. Y., Seah, H. Y., and Tan, a. C. L. (2003). "Ant Colony Optimization for Design of Water Distribution Systems." Journal of Water Resources Planning and Management, 129(3), 200-209. Marunga, A., Hoko, Z., and Kaseke, E. (2006). "Pressure management as a leakage reduction and water demand management tool: The case of the City of Mutare, Zimbabwe." Physics and Chemistry of the Earth, 31 (15-16), 763-770. Miyaoka, S. I., and Funabashi, M. (1984). "Optimal control of water distribution systems by network flow theory." IEEE Transactions On Automatic Control, Ac-29(4), 303-311. Momtahen, S., and Dariane, A. B. (2007). "Direct search approaches using genetic algorithms for optimization of water reservoir operating policies." Journal of Water Resources Planning and Management, 133(3), 202-209. Mounce, S. R., Khan, A., Wood, A. S., Day, A. J., Widdop, P. D., and Machell, J. (2003). "Sensor-fusion of hydraulic data for burst detection and location in a treated water distribution system." Information Fusion, 4(3), 217-229. Nalbantis, I., and Koutsoyiannis, D. (1997). "A parametric rule for planning and management of multiple reservoir systems." Water Resources Research, 33(9), 2165-2177. OFWAT. (2006). "Security of supply, leakage and water efficiency." 2005-06 Report, Ofwat, Birmingham, UK. Ormsbee, L. E., and Lansey, K. E. (1994). "OPTIMAL CONTROL OF WATER SUPPLY PUMPING SYSTEMS." Journal of Water Resources Planning and Management, 120(2), 237-252. Ostfeld, A., and Tubaltzev, A. (2008). "Ant colony optimization for least-cost design and operation of pumping water distribution systems." Journal of Water Resources Planning and Management, 134(2), 107-118. Rogers, D. (2005). "Reducing leakage in Jakarta, Indonesia." Leakage 2005, Nova Scotia, Canada. Savic, D. A., Walters, G. A., and Schwab, M. (1997). "Multiobjective genetic algorithms for pump scheduling in water supply." Evolutionary Computing, AISB International Workshop, Manchester, UK, April 7-8, 1997, Selected Papers, D. Corne and J. L. Shapiro, eds., Springer-Verlag, London, UK, 227236. Shamir, U., and Salomons, E. (2008). "Optimal real-time operation of urban water distribution systems using reduced models." Journal of Water Resources Planning and Management, 134(2), 181-185. 378

References Sotelo, A., Lücken, C. v., and Bar, B. (2002). "Multiobjective evolutionary algorithms in pump scheduling optimisation." Proceedings of the third international conference on Engineering computational technology, Civil-Comp press, Stirling, Scotland. Ulanicki, B., AbdelMeguid, H., Bounds, P., and Patel, R. (2008). "Pressure control in district metering areas with boundary and internal pressure reducing valves." 10th International Water Distribution System Analysis conference, WDSA2008, The Kruger National Park, South Africa. Ulanicki, B., Bounds, P. L. M., Rance, J. P., and Reynolds, L. (2000). "Open and closed loop pressure control for leakage reduction." Urban Water, 2 (2), 105114. Ulanicki, B., Kahler, J., and See, H. (2007). "Dynamic optimization approach for solving an optimal scheduling problem in water distribution systems." Journal of Water Resources Planning and Management, 133(1), 23-32. Vairavamoorthy, K., and Lumbers, J. (1998). "Leakage reduction in water distribution systems: optimal valve control." Journal of Hydraulic Engineering, 124(11), 1146-1154. Wang, J.-Y., Chang, T.-P., and Chen, J.-S. (2009). "An enhanced genetic algorithm for bi-objective pump scheduling in water supply." Expert Systems with Applications, 36, 10249–10258. Wu, Z. Y., and Sage, P. (2006). "Water loss detection via genetic algorithm optimization-based model calibration." ASCE 8th Annual International Symposium on Water Distribution System Ananlysis, Cincinnati, Ohio, USA. Zecchin, A. C., Maier, H. R., Simpson, A. R., Leonard, M., and Nixon, J. B. (2007). "Ant colony optimization applied to water distribution system design: comparative study of five algorithms." Journal of Water Resources Planning and Management, 133(1), 87-92. Zecchin, A. C., Simpson, A. R., Maier, H. R., and Nixon, J. B. (2005). "Parametric Study for an Ant Algorithm Applied to Water Distribution System Optimization." IEEE Transactions On Evolutionary Computation, 9(2), 175191.

379