Advancing the management of water resources

Columbia, MD Raleigh, NC

Andrew D. Dehoff, P.E and

Daniel P. Sheer, Ph.D., P.E.

October 22, 2004

Position Analysis and Forecast-Based Water Supply Operations

Reducing Risk and Saving Money by Operating Smarter

River

Con

o win

go C

ree k

Susquehanna

W. Con

ewag

o C

reek

Creek

Peque

a

Y O R K C O U N T Y

L A N C A S T E R C O U N T Y

H A R F O R D C O U N T Y C E C I L C O U N T Y

PENNSYLVANIA

MARYLAND

York

LancasterMarietta

Columbia

LakeWilliams

LakeRedman

OctoraroLake

ChesapeakeBay

Riv

er

Conest

ogaC

odor

us Cree

k

E. Branch

S . Branch

Muddy Creek

Creek

Deer

Cre

ek

O ctor

aro

Marietta Gage

Conowingo Gage

City of LancasterPipeline

Chester WaterAuthority Pipelines

Muddy Run PumpedStorage Facility

Safe HarborDam

Holtwood Dam

Conowingo Dam

City of BaltimorePipeline

Peach BottomAtomic Power Station

York WaterCo. Pipeline

Conowingo Pool and surrounding area

Conflicting Objectives

FERC minimum pool elevation FERC minimum flow requirements FERC recreation requirements Municipal withdrawals Habitat/fishery needs Salinity intrusion

ConowingoDam

108.5 ft. Normal Pool

104.5 ft. Normal Minimum Level due to hydroelectric operations

102.5 ft. Critical Level for Peach Bottom

100.5 ft. Minimum Conowingo Pool

97.0 ft. Peach Bottom shuts down

Limited Range of Operations

No reservoir system is fail safe

There is always some risk of running out of water.

Water supply planning and operations can reduce, but not eliminate that risk.

Position Analysis and forecast-based operations can allow you to manage risk explicitly and effectively

Reducing Risk

Capacity expansion reduces the risk of water shortage

Smart operating policies can also reduce the risk of water shortage

What are you willing to do or pay to achieve an acceptable level of reliability?

Traditional Approach

Safe Yield:

The amount of water that can be safely withdrawn from a reservoir (system) during some specified drought.

Implied Reliability:

Safe yield of record

50-year safe yield

20-year safe yield

Etc.

Traditional Approach, cont’d.

Sup

ply/

Dem

and

Time

Safe Yield

Supply

Demand

Traditional Approach, cont’d.

Sup

ply/

Dem

and

Time

Safe Yield

Supply

Demand

Safe Yield Isn’t a measure of reliability

Safe yield in all but the simplest systems is more than the sum of the safe yields of individual facilities

How the system is operated is very important Safe Yield isn’t as Safe as It Sounds

A system operated at historical safe yield will run out of water in any worse drought

A system operated at 20 year safe yield will run out of water an average of 1 in 21 years

Operations are Crucial to Supply

Increasing water available during droughts increases reliability

Conjoint operations of all facilities can substantially increase the water available during droughts

The objective of such operations is to minimize water loss through spill or seepage

Conjoint Use Simple Example

Reservoir A - minimum spring inflow = 25% of storage

Reservoir B - minimum spring inflow = 125% of storage

Rule - empty reservoir B before drawing Reservoir A down more than 25%

If the drought is 2 years long, the system will benefit from a complete refill of Reservoir B in the spring of the second year.

Examples of Conjoint Operations

Upper and Lower Delaware Basin reservoirs

Conowingo Pool and Baltimore City reservoirs

Kansas River Basin

Potomac River Basin

G52G53

G54

G56

G57

G58 G59

G60

G61G62

Baltimore

Susq-17

Conod'Ck-1

LetterKenny

Hogestown

445

Susq-18

450Harrisburg

455

H'burgDem

460CampHill

YellowBrCkSusq-19

475 YorkHaven

480 TMI

485 BrunnerIsPP

490

Pinchot

492

Manchester

ConewagoCk

Susq-20

495 Marietta

496

Mar'Local

500

EDonegalPP

505

Marburg

CodorusCk-1

510SpringGrove

515Glatfelter 520IndianRock

525Redman

ECod'Ck-1

530Williams

535YorkPS

SBrCod'Ck

536 York

538

YorkWTP

York'Ret

540

YorkDem 545Lancaster

ConestogaR

548Lanc'Ret

Lanc'Dem

550

555

SafeHarbor

PumpStorage

556

SHLocal

560Holtwood

561

HoltLocal

565

MuddyRun

570Conowingo

BigInch-1

571

ConoLocal575

PeachBPP

580

DeltaPP OctoraroCk

585

OctoraroLk590

Chester

595SusqMouth

596

MouthLocal

600

Rocks

DeerCk-1605

OtherDem

Darlington

610

615

Aberdeen

620

Perryvil le

625

HavreDeGrace

630

CecilCo

700

Harford

BigInch-2

705

Harf'Dem

PrettyBoy

GunPow.R-1

735

MontebelloPipe

740LochRav en

GunPow.R-2

742

MontebelloFP

745

Montebello-2

750

HillenVernonPS

760

EBaltimore

765

Liberty

PatapscoR

AshburtonFS

770

Joint

998 CheBay-1

999

CheBay-2

Measuring Reliability

In actual operations, emergency measures will prevent a system from running out of water Short-term conservation Alternative higher cost or lower quality

sources Emergency measures will be undertaken in many

droughts, even if they are less severe than the drought used to calculate safe yield

Expected frequency, duration, and severity of drought emergencies are good measures

0

5

10

15

20

25

run1 run2Run

Phase 4

Phase 3

Phase 2

Phase 1

Water Shortage PhasesPercent of Years in Each Phase

Frequency, Severity and Duration

Evaluated with a simulation model like OASIS Model includes facilities, demands, facility

operating policies and drought management policies

Model runs for a long (~50 year or longer) record, usually of historical flows

Model outputs include when emergency measures are imposed

Model also produces output that can be used for environmental and economic evaluations

Operations to Increase Reliability

Low cost conservation measures implemented early can avert high cost draconian measures later.

Water saved = #days * demand_reduction/day Costs are more than monetary costs Costs of reducing 50% may be orders of

magnitude larger than the cost of reducing 10%

Operating Rules

Rules will also include provisions that minimize the negative effects of implementing emergency measures

Rules will include operations to increase water available

Rules should be adequate to handle any historical drought with a reasonable margin of safety.

Rules will NOT be traditional safe yield rules

Operating Rules Can Impact

Demands

Environmental requirements

Balancing of supplies

Operating Rules Can Consider

Present state of the resource Storage Groundwater levels

Inflow forecasts Time of year Demand forecasts Other factors

We Can’t Forecast Weather, But

Streamflow depends on both weather and the “dryness” or “wetness” of the drainage basin

The snow, soil and aquifers in the basin are reservoirs and we can get some useful information about how much water they will or can contribute to a stream or to wells over the next several months

That information can be used to help make better operating decisions

NOT Particularly Useful Forecasts

Expected value of flow over the next x months -on average, there will always be enough water

Flows are likely to be higher than normal or lower than normal - OK, worry a little less or a little more

Worst case scenario - wonderful if you’re paranoid, otherwise useless unless you know its probability (most are arbitrary)

Useful Forecasts

Flow time series and their probabilities

Flow totals for various periods into the future and their probabilities

timet0

Historical Time Series

Str

eam

flow

Conditional Time Series Forecasts

All start from today’s flows Can be produced using statistical methods or

rainfall runoff models that represent surficial and deep aquifer interactions (and snow, if appropriate

Usually produce “equally likely” traces based on the assumption that any year’s historical weather is equally likely to repeat

timet0

Time series conditioned upon today’s basin conditions

Str

eam

flow

Forecasts to Useful Information

Run the simulation model for each of the equally likely forecasts

If there are 50 equally likely traces, and the reservoir falls below a given level in only 5 of the simulations, then the probability of falling below that level is 10% given the rules used in the simulation

Probabilities of other consequences, including environmental consequences and utility revenue can also be evaluated

0 10 20 30 40 50 60 70 80 90 100

Probability of Non-exceedance

100

105

110

115

120

125

130

Stag

e in

feet

Reservoir Stage930911, Tue Aug 17 2004 17:15

2 week 4 week 8 week 12 week 16 week

0 10 20 30 40 50 60 70 80 90 100

Probability of Non-exceedance

100

105

110

115

120

125

130

Stag

e in

feet

Reservoir Stage930731, Tue Aug 17 2004 17:18

2 week 4 week 8 week 12 week 16 week

0 10 20 30 40 50 60 70 80 90 100

Probability of Non-exceedance

100

105

110

115

120

125

130

Stag

e in

feet

Reservoir Stage930814, Tue Aug 17 2004 17:17

2 week 4 week 8 week 12 week 16 week

0 10 20 30 40 50 60 70 80 90 100

Probability of Non-exceedance

100

105

110

115

120

125

130

Stag

e in

feet

Reservoir Stage930828, Tue Aug 17 2004 17:16

2 week 4 week 8 week 12 week 16 week

0 10 20 30 40 50 60 70 80 90 100

Probability of Non-exceedance

100

105

110

115

120

125

130

Stag

e in

feet

Reservoir Stage930911, Tue Aug 17 2004 17:15

2 week 4 week 8 week 12 week 16 week

How to respond

If the probabilities of untoward events are too high, change the rule and test again. If it works and you’re happy with the rule, do it.

Find a rule that works for the long-term so you don’t have to change rules “on the fly”

Creating a Rule That Uses Forecasts

Many forms are possible The Rocky Mount rule

If the probability of the reservoir falling below 25% in the next 8 weeks is > 20% institute water use restrictions and reduce instream flow

A possible (and untested) Conowingo Rule If the probability of Marietta flow falling

below 2500 cfs in the next summer is > 20% begin Baltimore pumping

Effective Rules

Can NOT miss any droughts Must provide enough advance notice for

remedial action to be effective Minimize the number of false alarms

Developed by trial and error through the use of simulation models.

Evaluating Rules

Scenario Number of

Days in Water Restriction

Number of Years with Water Restrictions

Volume of Water Not Delivered

(million gallons)

1 10 1 25

2 16 3 30

3 5 5 5

4 25 3 140

5 30 6 130

6 18 2 65

Checking that the Rule is Sufficient

PERFORMANCE SUMMARY(15 mgd Demand; 80 cfs Minimum Release)

DROUGHTS OFRECORD

REDUCTION TRIGGERED INLONG-TERM SIMULATION(20%, 2 months, Elev. 115)

REDUCTION BASED ON REAL-TIMEEVALUATION (weeks)

(3) 1932 Yes 4

(5) 1933-34 No 0

(2) 1968 Yes 5

(4) 1980-81 Yes 16

1986 Yes 2

1987 Yes 1

1988 Yes 0

(1) 1993 Yes 12

Rocky Mount Results

Change in operating rule replaced $70M pipeline Required emergency measures infrequent and

not severe NCDNR agreed to tie reductions in instream

flow requirements to imposition of demand restrictions

Environmental impacts of infrequent reductions in low flow (still well above “natural flows”) judged less than impacts of building and operating the pipeline

Advancing the management of water resources

Columbia, MD Raleigh, NC

Thank You