Management of Water Systemsunder Uncertainty

Uri ShamirTechnion – Israel Institute of Technology

andThe Interdisciplinary Center (IDC)

Consultant to the Israeli Water AuthorityChair, Kinneret and Watershed Monitoring Team

UCOWR/NIWR/CUAHSI Annual ConferenceWater Systems, Science, and Society Under Global Change

Tufts, June 2014

A Management Model is a Platform for Disciplined Discourse

In fact, any model: conceptual, verbal, mathematical, ecological, deterministic or stochastic, simulation or optimization, single-objective or (always!) multi-objective

Discourse: in formulation of the model, examination of intermediate results, evaluation of final results and of re-formulation and re-run

A well-structured model enables/enforces a disciplined discourse

A Management Model is a Platform for Disciplined Discourse

Today: management under uncertainty of the Israeli national water system

Uncertainty on the supply and demand sides, leaving out uncertainties in the distribution system, and in economics, finances, management, politics

Brief mention of the methodology we use for robust optimization under uncertainties

Average Annual PotentialMed-Jordan R.: ~1,700 mcm/yrIsraeli system: ~1,200 mcm/yr

Western Galilee Aquifer

Carmel Aquifer

Coastal Aquifer

N. EastMountain AquiferEastWest

Lake

Tel Aviv

Jerusalem

HaifaWATER SOURCES

Negev Aq.

Arava Aq.

Highly integrated national and regional

water systems

Kinneret Watershed

~100 km

Gaza Aquifer

1930 1940 1950 1960 1970 1980 1990 2000 2010-3

-2

-1

0

1

2

3

Israel, Time Scale: 2 yearsS P

I

‐2.00

‐1.50

‐1.00

‐0.50

0.00

0.50

1.00

1.50

2.00

2.50

3.00

3.50

1962 1967 1972 1977 1982 1987 1992 1997 2002 2007 2012

SPI

שנים

Standard Precipitation Index (SPI) for the National System2-year backward moving average of the total precipitation

(Amir Givati, Meir Rom & Uri Shamir, IHS, March 2012)

SPI

SPI

6

Standard Precipitation Index (SPI)

Longer, more frequent & deeper droughts

ForecastHistory

Natural Replenishment (mcm/year)All Sources from Mediterranean Sea to Jordan River (exc. Gaza)

Data: “The Natural Water Resources Between the Mediterranean Sea and the Jordan River”, IHS 2012

1973-1992Av = 1,848SD = 684

1993-2009Av = 1,643SD = 465

1973-2009Av = 1,748SD = 584

Cum. Deficit~1,770 mcm

Cum. Deficit 1,530 mcm

1991/1992 - 3,839 mcm (Pinatubo?)

?

צריכת מים לנפש, מ"ק

0

100

200

300

400

500

600

700

1950 1960 1970 1980 1990 2000 2010

סך הצריכה משקי בית תעשייה חקלאות

Water (incl. effluents) consumption, m3/cap/year

Total

Agriculture(including effluents)

Urban

Industry

Source: Prof. Yoav Kislev

~60% effluents

Since 1960: 7-fold reduction in per capita fresh water use for agriculture, and 3.5-fold reduction is total (water+effluents) use per capita for agriculture

Urban Consumption (m3/cap/year)1996-2011 & 2012 est.

Private homes & gardens

Municipal areas(w/o Ag. & Ind.)

Total settled sector

From: Sharon Nussbaum, IWA

~15%

From: Sharon Nussbaum, IWA

Delta = (105-90) x14 million = 210 mcm/year

Total Urban Consumption (m3/cap/year)1996-2011 & 2012 est.

m3 /c

ap/y

ear 15 m3

Long‐Term Master Plan for the National Water Sector

Part A ‐ Policy DocumentVersion 4

August 2012

Max Storage

Desalination Shutdown

Water table‘Pink Line’‘Red Line’

Flows min Level

Reference Line

Replenishment

Desalination

Demand

Deficit

Deficit

Flows & Spills

Max Storage

Desalination Shutdown

Water table‘Pink Line’‘Red Line’

Flows min Level

Reference Line

Replenishment

Desalination

Demand

Deficit

Deficit

Flows & Spills

2012 Master Plan for the Israeli National Water Sector:

Stochastic simulation with an aggregate model of the national water system

The Stephen and Nancy Grand Water Research InstituteThe Stephen and Nancy Grand Water Research Institute Technion – Israel Institute of TechnologyTechnion – Israel Institute of Technology

התפלה נדרשת כתלות במדיניות אמינות אספקה*

0100200300400500600700800900

1,0001,1001,2001,3001,4001,5001,6001,700

2015 2020 2025 2030 2035 2040 2045 2050שנה

" ק למ

מה ,פל

הת

75% 90%95% 100%מחסור מקסימאלי 250 מלמ"ש תוכנית מאושרתתוכנית מומלצת

* נבחנה על בסיס תרחישים שהוגדרו

Development of Desalination Capacity as a function of Required Reliability

Des

alin

atio

n C

apac

ity, m

cm/y

ear

Ashkelon: 120 mcm/y

started 2006

Palmachim: 90 mcm/y

started 2007

We offered the Palestinians to locate a 50-100 mcm plant at Hadera for direct supply to the WB

Hadera: 127 mcm/y started end of 2009

Sorek 150 mcm/y in 2013/4

Sea-Water desalination program started in 20002014: Existing desalination capacity 587 mcm/y

50% of the average natural replenishment

Ashdod: 100 mcm/y in 2014

2050 forecast: ~1,100

+ 54 mcm/y desalinated brackish

תוחלת האוגר כתלות באמינות

0

500

1,000

1,500

2,000

2,500

3,000

2010 2015 2020 2025 2030 2035 2040 2045 2050

שנה

"ק למ מ

ר,וגא

75% 90% 95% מחסור 250 מלמ"ש 100% אוגר רצוי

Stor

age

MC

M

Year

Development of Expected Storage (above “Red Lines”)

MEDIUM

90צריכה לנפש

מילוי ממוצע רב שנתי: בינוניס "אוכלוסיה למ

.2050-מיליון ב 14.7, 1.4%קצב

LOW

80צריכה לנפש

מילוי ממוצע רב שנתי:נמוךס "אוכלוסיה למ

.2050-במיליון 12.8, 0.7%קצב

HIGH

100צריכה לנפש

15%מילוי ממוצע פחות : ס גבוה"אוכלוסיה למ

.2050-מיליון ב 16.6, 2.1%קצב W1=30%

W2=60%

W3=10%

∑Wi=100%

תכנית התפלה

תכנית התפלה

תכנית התפלה

SUGGESTED PLAN

ש"מלמ 560' מחסור מקסש"מלמ 550חקלאות

ש"מלמ 640' מחסור מקסש"מלמ 500חקלאות

ש"מלמ 0' מחסור מקסש"מלמ 400חקלאות

ש"מלמ 550' מחסור מקס

Weighted scenarios of: hydrology, population growth, consumption: per capita, agriculture

In October 2013 the Gov’t decided to reduce the purchase of desalinated water this year from 585 to 345 mcm

We used an aggregate optimization model for determining the optimal reduction plan

And are now making it monthly, for preparing next year’s plan

18

D S<Smax

3 demand zones, 8 natural sources, 5 desalination plants

New developments inOptimization under UncertaintyBased on the 2011 PhD of Mashor Housh

Highly efficient solution of the deterministic model for solving (many) scenarios

Efficient Stochastic programming, “wait and see” “here and now”, two-stage and Multi-stage (MSP)

Limited Multi-stage Stochastic Programming (LMSP)

Info-Gap model

Robust Optimization: Robust Counterpart (RC), Affine Robust Counterpart (ARC), Affine Adjustable Robust Counterpart (AARC)

Robust Counterpart.

Recharge 1

Rec

harg

e 2

The size of the ellipse is set by a user defined parameter ϴ

1,maxR1,minR

2,maxR

2,minR

1 2ˆ ˆ( , )R R

Central part of the INWSS: 15 year operation

3 aquifers, 5 desalination plants, 9 consumer zones, 14 network nodes

Aqu

ifer 1

Aqu

ifer 2

Aqu

ifer 3

Reliability vs. Mean Cost

A good compromise solution

Reliability(%)

Mean Cost (M$/year)

CPRP(Θ=3)RP(Θ=2)

RP(Θ=1)

950 1000 1050 1100 1150 1200

100

90

80

70

60

50

A Management Model is a Platform for Disciplined Discourse

In fact, any model: conceptual, verbal, mathematical, ecological, deterministic or stochastic, simulation or optimization, single-objective or (always!) multi-objective

Discourse: in formulation of the model, examination of intermediate results, evaluation of final results and of re-formulation and re-run

A well-structured model enables/enforces a disciplined discourse

Management of Water Systemsunder Uncertainty

Uri ShamirTechnion – Israel Institute of Technology

andThe Interdisciplinary Center (IDC)

Consultant to the Israeli Water AuthorityChair, Kinneret and Watershed Monitoring Team

UCOWR/NIWR/CUAHSI Annual ConferenceWater Systems, Science, and Society Under Global Change

Tufts, June 2014