Canal Operations and Automation

Canal Operations Canal Operations and Automationand Automation

Bert ClemmensBert ClemmensU.S. Arid-Land Agricultural Research CenterU.S. Arid-Land Agricultural Research Center

UDSA-ARSUDSA-ARSMaricopa, Arizona, USAMaricopa, Arizona, USA

What have we learned so far?What have we learned so far?

Steady-state backwater curves are useful Steady-state backwater curves are useful for understanding how canal systems for understanding how canal systems function. Downstream influences can be function. Downstream influences can be transferred upstream.transferred upstream.

Free-flow structures and sections of Free-flow structures and sections of uniform flow (i.e., at normal depth) isolate uniform flow (i.e., at normal depth) isolate downstream influences from transferring downstream influences from transferring upstream.upstream.

Canal Operations are improved if Canal Operations are improved if operation of structure is simple.operation of structure is simple.

Things that simplify operations:Things that simplify operations: Good flow measurementGood flow measurement Free-flow structuresFree-flow structures long-crested weirslong-crested weirs

Things that complicate operationsThings that complicate operations Lack of flow measurementLack of flow measurement Uncertain structure hydraulics & operationsUncertain structure hydraulics & operations Intermediate, non-regulating structuresIntermediate, non-regulating structures

It is possible to determine wave travel It is possible to determine wave travel times for operations from steady-state times for operations from steady-state backwater curves (volume compensation)backwater curves (volume compensation)

Structure hydraulics have a big influence Structure hydraulics have a big influence on how waves travel through canal pools.on how waves travel through canal pools.

Canals under backwater and weirs provide Canals under backwater and weirs provide faster (and probably more predictable) faster (and probably more predictable) response times.response times.

Some things we didn’t coverSome things we didn’t cover

Typical operations can be considered Typical operations can be considered upstream control, since we want to keep upstream control, since we want to keep the water level on the upstream side of the water level on the upstream side of structures constant.structures constant.

Long-crested weirs are being promoted Long-crested weirs are being promoted because they reducing response times because they reducing response times and keep water levels more constantand keep water levels more constant

Automatic upstream level control is Automatic upstream level control is becoming more common.becoming more common.

Tail-Ender ProblemTail-Ender Problem

These approaches contribute to the tail-These approaches contribute to the tail-ended problem, where all the mismatches ended problem, where all the mismatches end up at the downstream end of the end up at the downstream end of the system.system.

Solving the Tail-Ender ProblemSolving the Tail-Ender Problem

Good measurement and accounting keep Good measurement and accounting keep right amount of water in canals and reduce right amount of water in canals and reduce the chance for serious mismatches.the chance for serious mismatches. This mean operators should keep track of and This mean operators should keep track of and

be accountable for mismatchesbe accountable for mismatches Remote manual operation can identify and Remote manual operation can identify and

correct the problem. (SCADA)correct the problem. (SCADA) New methods for automatic downstream New methods for automatic downstream

level control are being developed.level control are being developed.

Old canal systems were not designed for accurate measurement, control and

accounting Canal control is difficult because:

Upstream changes are delayed downstream Upstream changes arrive gradually Pool volumes change with discharge, roughness, and depth

at structure Canal operators want steady flows and rigid schedules

-- farmers want flexibility and responsiveness Common problems result

Flow rates fluctuate Flow rates may be too high or low Operations are unresponsive to needs Inadequate accounting for water entering and leaving canal

CAIDD SCADA – 109 sites

Manual Supervisory Control

Additional FeaturesAdditional Features Incremental ( Relative ) Gate Flow ChangeIncremental ( Relative ) Gate Flow Change Flow rate control at headgatesFlow rate control at headgates Check-Structure Flow Balance Showing Check-Structure Flow Balance Showing

Accumulated Downstream DemandsAccumulated Downstream Demands Routing of known demand changesRouting of known demand changes Pool-Volume Balance to Help Determine Flow Pool-Volume Balance to Help Determine Flow

MismatchesMismatches High/Low Water Level Limits Can Be Enforced High/Low Water Level Limits Can Be Enforced

Through Automatic Upstream Level ControlThrough Automatic Upstream Level Control

Flow Monitoring Standard Supervisory Control Features using Standard Supervisory Control Features using

iFix Dynamics from Intellution, Inc. (MSIDD)iFix Dynamics from Intellution, Inc. (MSIDD)

Flow Rate Control

Good Flow Measurement does not happen by Good Flow Measurement does not happen by accidentaccident

Check gates are usually not accurate Check gates are usually not accurate measurement gates – but good incremental flow measurement gates – but good incremental flow control is possiblecontrol is possible

Continuous monitoring of flows and/or levels/gate Continuous monitoring of flows and/or levels/gate position is requiredposition is required

Turnout flow monitoring usually is based on Turnout flow monitoring usually is based on assuming constant upstream water levelassuming constant upstream water level

Only applicable for headgates or for check gates Only applicable for headgates or for check gates under centralized controlunder centralized control

Incremental flow-rate control

iFix allows user-defined displays (CAIDD)iFix allows user-defined displays (CAIDD)

Software for Automated Canal Management -- SacMan

Main features include: Routing of known demand changes with

volume compensation (similar to Canal de Provence and CAP).

Automatic control of flow rate at key locations.

Distant downstream water level feedback control to account for

flow measurement errors, demand routing errors, & unknown disturbances.

Canal Automation Scheme

LocalControl

LocalControl

LocalControl

LocalControl

LocalControl

Central Observation and Control Water Demands

Water SupplyConstraints

Water SupplySchedule

Water DeliverySchedule

LocalControl

LocalControl

LocalControl

LocalControl

LocalControl

Central Observation and Control Water Demands

Water SupplyConstraints

Water SupplySchedule

Water DeliverySchedule

RTU

Operator

Personal Computer

iFix SCADASacMan

iFixModbus Driver

iFix SCADAMonitor & Control

SacManDemandDatabase

SacManOrder

SacManControl Program

iFixProcessDatabase-

Software for Automated Canal Software for Automated Canal Management -- SacManManagement -- SacMan

Components of control logicComponents of control logic

Feedforward controller

Setpoint

Flow controller

Feedback controller

Qff

Qfb

y

yy

e

PI Optimization with LQRPI Optimization with LQR

Multi-input multi-output optimizationMulti-input multi-output optimization

Retains PI format for understandabilityRetains PI format for understandability

Based on incremental flow controlBased on incremental flow control

Extensively tested through simulation and considerable real-time Extensively tested through simulation and considerable real-time

testingtesting

MPC optimization MPC optimization

Multi-input multi-output optimizationMulti-input multi-output optimization

Black box formatBlack box format

Based on incremental flow controlBased on incremental flow control

Some testing through simulation and limited real-time testingSome testing through simulation and limited real-time testing

Can handle a wider variety of constraintsCan handle a wider variety of constraints

Feedback control of Downstream Water LevelsFeedback control of Downstream Water Levels

SacMan Orders generates a schedule of flow changes

Flow changes are implemented by SacMan Control Program or printed for manual

operation

Pool Flow Balance

SCADA screen showsSCADA screen shows Actual flow at check structureActual flow at check structure Current downstream demand at check Current downstream demand at check

structurestructure Operator can examine differences Operator can examine differences

between actual and intended flows to between actual and intended flows to identify control (or measurement) issuesidentify control (or measurement) issues

Pool Volume Mismatches With steady inflow and outflow, changes in With steady inflow and outflow, changes in

pool water level indicate mismatches in net pool water level indicate mismatches in net inflow rate (inflow – outflow)inflow rate (inflow – outflow)

SCADA screen shows flow mismatchesSCADA screen shows flow mismatches Operators can use flow mismatches to Operators can use flow mismatches to

change flows between pools and route change flows between pools and route additional flow into the systemadditional flow into the system

If water level is low and declining, operator If water level is low and declining, operator needs to both increase flow and pool volume!needs to both increase flow and pool volume!

Note: these water-level based flow Note: these water-level based flow mismatches are meaningless when flow mismatches are meaningless when flow changes have recently been madechanges have recently been made

Software for Automated Canal Management

SacMan

Manual Local Central

Flow Monitoring

Flow Control

Demand Scheduling

Incremental gate flow changes

Out-of-Bounds control

Pool Volume Mismatches

Pool Flow Balance

Control start-up

Water-Level setpoint changes

Alarms

Overall Control Strategy

Reachable Goal of Current Canal Automation Product

Development Water orders are entered into computerWater orders are entered into computer SacMan routes flow changes through canal SacMan routes flow changes through canal

system based on volume compensationsystem based on volume compensation Canal operator opens turnout gate at prescribed Canal operator opens turnout gate at prescribed

timetime Water level errors are corrected with feedback Water level errors are corrected with feedback

control (new technology works!)control (new technology works!) Headgate and check gate flow controllers Headgate and check gate flow controllers

maintain flow balancesmaintain flow balances If needed, main canal pool volumes are used to If needed, main canal pool volumes are used to

balance secondary canal volume errorsbalance secondary canal volume errors

Maricopa-Stanfield Irrigation Maricopa-Stanfield Irrigation and Drainage District and Drainage District

(MSIDD)(MSIDD) Approximately 87,000 Approximately 87,000

Acres (35,000 Acres (35,000 Hectares )Hectares )

Construction Construction completed in 1987completed in 1987

Designed for Designed for supervisory control supervisory control with optional automatic with optional automatic downstream controldownstream control

WM LateralWM Lateral An excellent test facility for An excellent test facility for canal automation research!canal automation research!

90 cfs ( 2.5 m90 cfs ( 2.5 m33/s) capacity/s) capacity 5.4 miles (9 km) long5.4 miles (9 km) long 8 pools8 pools 10 turnouts (2 pumped)10 turnouts (2 pumped) 2 wells pumping in2 wells pumping in emergency spill in last poolemergency spill in last pool ultrasonic meters on turnoutsultrasonic meters on turnouts canal travel time less than 2 hourscanal travel time less than 2 hours Main canal can tolerate changes in WM inflowMain canal can tolerate changes in WM inflow

Typical Check StructureTypical Check Structure

Testing during 2004 & 2005Testing during 2004 & 2005

1)1) 30 day test period30 day test period2)2) 48 typical water orders48 typical water orders3)3) 2 atypical flow changes (power 2 atypical flow changes (power

outgages)outgages)4)4) Tested different upstream, LQR Tested different upstream, LQR

downstream control strategies and downstream control strategies and combinationscombinations

5)5) In 2005, tested Model Predictive ControlIn 2005, tested Model Predictive Control

Typical test scenarioTypical test scenario

Routine operations (full control)Routine operations (full control) Water orders entered into SacMan OrderWater orders entered into SacMan Order Schedule of gate flow changes madeSchedule of gate flow changes made Schedule posted to SacMan CPSchedule posted to SacMan CP SacMan CP changes flow setpointSacMan CP changes flow setpoint Check gate flow set every 2 minute Check gate flow set every 2 minute Operator delivers water to userOperator delivers water to user Feedback control adjustments every 10 Feedback control adjustments every 10

minutesminutes

Typical results- 2004Typical results- 20043 delivery changes routed automatically3 delivery changes routed automatically

PIPI++-1-1 downstream controller (80% cap), pool 5 under level control downstream controller (80% cap), pool 5 under level control

WM-0 July 17, 2004

0

5

10

15

20

25

30

35

40

6:00 9:00 12:00 15:00 18:00

Flo

w R

ate

(cfs

)

WM-1 July 17, 2004

2

2.5

3

3.5

6:00 9:00 12:00 15:00 18:00

Dep

th (

ft)

WM-2 July 17, 2004

2

2.5

3

3.5

6:00 9:00 12:00 15:00 18:00

Dep

th (

ft)

WM-3 July 17, 2004

2

2.5

3

3.5

6:00 9:00 12:00 15:00 18:00

Dep

th (

ft)

WM-4 July 17, 2004

2

2.5

3

3.5

6:00 9:00 12:00 15:00 18:00

Dep

th (

ft)

WM-5 July 17, 2004

2

2.5

3

3.5

6:00 9:00 12:00 15:00 18:00

Dep

th (

ft)

WM-6 July 17, 2004

2

2.5

3

3.5

6:00 9:00 12:00 15:00 18:00

Dep

th (

ft)

WM-7 July 17, 2004

2

2.5

3

3.5

6:00 9:00 12:00 15:00 18:00

Dep

th (

ft)

Feedback control of Feedback control of Downstream Water LevelsDownstream Water Levels

Multi-input multi-output optimizationMulti-input multi-output optimization

Retains PI format for understandabilityRetains PI format for understandability

Based on incremental flow controlBased on incremental flow control

Extensively tested through simulationExtensively tested through simulation

Research (2002) on WM canal identified capabilities and limitation of various downstream water-level controllers

Series of simple PI Controllers

More CentralizedPI+1

-1 Controller

Pool 2 -- Sept. 25

0.60

0.70

0.80

0.90

1.00

1.10

1.20

1.30

9.50 11.50 13.50 15.50 17.50 19.50

Time

Wat

er D

epth

(m

)

SOBEK Simulation

Field Test

Pool 2 -- Oct. 16

0.70

0.75

0.80

0.85

0.90

0.95

1.00

1.05

1.10

9.00 10.00 11.00 12.00 13.00 14.00

Time

Wat

er D

epth

(m

)SOBEK Simulation

Field Test

Feedback control tests 2004Feedback control tests 2004Simulated groundwater well failure PISimulated groundwater well failure PI++

-1-1 and PIL and PIL++--

More centralized controller spread out the disturbanceMore centralized controller spread out the disturbance

WM-0 July 30 & 31, 2004

0

5

10

15

20

25

30

18:00 21:00 0:00 3:00 6:00

Flo

w R

ate

(cfs

)

WM-1 July 30 & 31, 2004

2

2.5

3

3.5

18:00 21:00 0:00 3:00 6:00

Dep

th (

ft)

WM-2 July 30 & 31, 2004

2

2.5

3

3.5

18:00 21:00 0:00 3:00 6:00

Dep

th (

ft)

WM-3 July 30 & 31, 2004

2

2.5

3

3.5

18:00 21:00 0:00 3:00 6:00

Dep

th (

ft)

WM-4 July 30 & 31, 2004

2

2.5

3

3.5

18:00 21:00 0:00 3:00 6:00

Dep

th (

ft)

WM-5 July 30 & 31, 2004

2

2.5

3

3.5

18:00 21:00 0:00 3:00 6:00

Dep

th (

ft)

WM-6 July 30 & 31, 2004

2

2.5

3

3.5

18:00 21:00 0:00 3:00 6:00

Dep

th (

ft)

SummarySummary

SacMan can control a canal under full SacMan can control a canal under full automatic controlautomatic control

SacMan Order can effectively route SacMan Order can effectively route deliveries under manual or automatic deliveries under manual or automatic controlcontrol

Performance depends on canal Performance depends on canal propertiesproperties

Future PlansFuture Plans

Implementation at CAIDD (120 sites) is Implementation at CAIDD (120 sites) is ongoingongoing

SRP starting to test SacMan OrderSRP starting to test SacMan Order

New training tool for SCADA operators New training tool for SCADA operators and canal automation testing has been and canal automation testing has been developeddeveloped

USWCL Canal Control SystemUSWCL Canal Control System

Modbusover

RS-232 RTU

RTU

RTU

Modbusover

RS-232

Spread SpectrumRadio System

iFixUser Interface

iFixProcess

Database

SACMan

Off-line SCADA Training through SimulationOff-line SCADA Training through Simulation

Radio link sent to another computer that simulates RTU and Canal FlowRadio link sent to another computer that simulates RTU and Canal Flow

Sobek/Matlab

PhysicalSimulator

HardwareSimulator

iFix SCADA

Gate Positions/ChangesWater LevelsBattery VoltageModbus overRS-232

Gate PositionWater LevelsBattery VoltageXML-RPC overTCP/IP

Gate PositionsWaterlevelsTurnout FlowsXML-RPC overTCP/IP

Simulator Software

Sequence of Canal Automation Implementation

Step 1: Remote, manual monitoring and control, using SCADA system

Limitations: Rely on operator judgment. Wait-and-see approach. Hard to control large network. Transient flows are hard to control.

Sequence of Canal Automation Implementation

Step 2: Independent, automated control of individual gates based on local water levels

Central control within SCADA (e.g. SRP) Remote control with PLC (e.g. ITRC) or RTU.

Limitations: No coordination Failures not recorded or observed Tailender problems and spills Wrong level controlled?

Sequence of Canal Automation Implementation

Step 3:Centralized automated control, using SCADA

and computer-driven logic.

Limitations:• Accurate flow measurement or significant in-

canal storage required. • Controls average water level rather than

downstream water level (e.g. CAP).