Handout: #:1

Ir. Edy Anto Soentoro, MASc., PhD.

SIMULATION of Seasonal Reservoir Operation

Reservoir Schematic Diagram

Where Vuc = Upper Critical Volume (this can vary with time)
Vlc = Lower Critical Volume (this can vary with time)
Vmin = Minimum Reservoir Volume (Dead Storage)
Vmax = Maximum Reservoir Volume

Active Storage Volume is the volume between Vmax and Vmin

SYSTEM EQUATION (WATER BALANCE)

Vt+1 = Vt + It - Et - Rt - Lt
or
Vt+1 = Vt + It - Et - (Rt + Shortt - Surplus t ) - Lt , inside bracket = Demand t
where
Vt = Volume (mcm) at the beginning of period t
It = Inflow volume (mcm) during period t
Et = Net Evaporation {Evaporation - Precipitation} volume (mcm) during period t
Lt = Other Loss volume (mcm) during period t. This is actually be a loss or a gain from
sources such as leakage, bank storage, and groundwater interactions.
Rt = Release volume (mcm) during period t

**Derive the Water Balance Equation from the Linear Reservoir Storage Equation:

dV
 ( I  O)
dT

1
EVAPORATION

Et = Eratet * Areat

where
Eratet = Evaporation depth (m) during time period t
Areat = Reservoir surface area (m2) during period t

Surface area of the reservoir is related to reservoir volume via the elevation-area-volume
relationship based upon the topography of the reservoir site. The nature of this relationship is
such that a power function will usually fit the data for area-volume, that is:

Area = aVb

The area during some time period should most accurately be an average of the surface area at
the beginning of the period based on Vt and the surface area at the end of the period based on
Vt+1.

Areat = f(0.5[aVtb + aVt+1b])

Usually in simulation models we assume that Areat = f(Vt) ; that is we only consider the
beginning period volume. This assumption then leads to:

Et = Eratet * (aVtb)

What computational advantages does this assumption provide and what errors will result?
What could we do to overcome this problem?

RELEASE

The release for a given time period must be established by an operational policy. This operational
policy relates the release, Rt, to the demand for the period, Dt. There are three basic options:
1. Release the demand. This is usually the operational policy of choice for normal operations.
2. Release less than the demand. This is usually the operational policy employed during drought
operations.
3. Release more than the demand. This is usually the operational policy employed during
periods of high stream flows (flood operations).

Lets define three operational zones:

* Zone 1 – For Vt < Vlct and Vt  Vmin denote this as the Deficit Operation Zone
When Vt < Vlct, Then Rt = DSP * Dt, where DSP = Demand Satisfaction Proportion (0 – 1)
Actually Rt = Minimum of : [DSP * Dt] or [(Vt + It – Et) – Vmin]

The Demand Satisfaction Proportion can either be established by the modeler or computed, such
as by using the following linear relationship:
DSP = {(Vt – Vmin) / (Vlct – Vmin)}

2
* Zone 2 – For Vt  Vlct and Vt  Vuct denote this as the Normal Operation Zone
where Rt = Dt

* Zone 3 – For Vt+1 > Vuct and Vt  Vmax denote this as the Excess Operation Zone
When Vt > Vuct, Then Rt = Dt + Extrat, where Extrat = Additional Release which
might be calculated by, Extrat = Vt+1 - Vuct

WATER SUPPLY DEMAND REPRESENTATION

It is often convenient to represent the demand for each period as a proportion of an annual
demand. This allows us to study both the magnitude of the total demand and the time
distribution of the demand over a year. Lets define our demand per period as:

Dt = t * TAD, where t = proportion of the annual demand in period t;


t 1
t 1

For t = 1 month and t representing monthly “water use coefficients”

SIMULATION PROCESS

We wish to simulate the operation of a reservoir using the equations and concepts developed
previously. We must begin with an initial boundary condition for the volume in the reservoir at
the beginning of the first period of simulation, V1. We assume that the inflows, demands per
period and evaporation rates for the period are known. Using these values we can calculate the
amount of evaporation volume and release volume. We then use our system equation to
calculate the end of period volume by:

Vt+1 = Vt + It - Et - Lt - Rt for t = 1, 2, ….., n

Actually, we have to make calculation check each period to insure that our simulated volumes are
within the bounds of Vmin to Vmax. We do this by computing our system equation and then
adjusting the values as needed. Lets define:

Vt+1calc = Vt + It - Et - Lt - Rt

3
SPILLS

If Vt+1calc > Vmax Then Spillt = Vt+1calc - Vmax

And Vt+1adj = Vt+1calc – Spillt = Vmax

SHORTAGES OR DEFICITS

We also need to account for any failures to meet the demand. Lets define a shortage (or deficit)
as:

Shortaget = Dt – Rt when Rt < Dt otherwise Shortaget = 0.

Lets also denote a “Failure Period” as any period in which we fail to deliver the full demand. If we
count the number of failure periods, we can use this information to determine the frequency of
failure periods within the period of analysis of the simulation model.

CRITICAL PERIOD

The critical period within a period of analysis is the longest period required to go from a full
condition (Vt = Vmax) to almost empty (Vt  Vmin) and return to a full condition.

Storage

1200.0

Critical Period
1000.0

800.0
Volume in MCM

600.0

400.0

200.0

Vmin

0.0
0 20 40 60 80 100 120 140
Months

4
HYDROPOWER CALCULATIONS

The potential power from releasing water can be calculated by:

P = 9.8*q*h*

Where
P = power output in KW
q = turbine flow in m3/s
h = head on the turbine in m
 = turbine efficiency (0 – 1)

The head on the turbine is based upon the elevation of the water in the reservoir. This can be
calculated in a similar manner to calculating the reservoir surface area. Elevation of the reservoir
is related to reservoir volume via the elevation-area-volume relationship based upon the
topography of the reservoir site. The nature of this relationship is such that a power function will
usually fit the data for elevation-volume, that is:

Elevation or Head = cVd

The area during some time period should most accurately be an average of the head at the
beginning of the period based on Vt and the head at the end of the period based on Vt+1.

ht = f(0.5[cVtd + cVt+1d])

Usually in simulation models we assume that ht = f(Vt) ; that is we only consider the beginning
period volume. This assumption then leads to:

ht = cVtd

What computational advantages does this assumption provide and what errors will result? What
could we do to overcome this problem?

If we produce this power for some period of time, we are producing energy. Lets assume that we
are using a monthly time step in our reservoir model. Lets further assume that a “typical” month
has 730 hours (24 * [365/12]). Then Q(mcm) = q(m3/s)*2.628 and

Energyt = (9.8*730/2.628)*Qt*ht*

Where : Energyt = Monthly hydro-electric energy in KWH

Converting Energyt to MWH gives : Energyt = 2.72*Qt*ht*

It is often useful to evaluate the time of generation at something other than full generation (730
hours per month). Lets define HGR = proportion of time (hours) of generation per month (0 – 1).
Then our equation for generation energy becomes:

Energyt = 2.72*HGR*Qt*ht*

5
Where
Energyt = Monthly hydro-electric energy in MWH
HGR = proportion of time for generation (0 – 1)
Qt = monthly release volume in mcm
ht = head on the turbine in m
 = turbine efficiency (0 – 1)

We can also invert the above equation to solve for the required monthly release to produce the
desired monthly energy:

Qtreq = Energyt / [2.72*HGR*ht*]

ENERGY DEMAND REPRESENTATION

We will use a similar concept to that described for representing the period demand for water in
terms of an annual demand and proportions of demand per period. We will define our annual
energy requirement as annual “firm” energy. Lets define our energy demand per period as:
Energyt = t * AFE, where t = proportion of the annual firm energy in
period t;
AFE = annual firm energy.
Also
12


t 1
t 1

For t = 1 month and t representing monthly “energy use coefficients”

ADDITIONAL ENERGY CONSIDERATIONS

There may be restrictions on the maximum flow that the turbine can handle and the minimum
head that must be present to produce energy. These restrictions could be added as constraints
on the head; ht  hmin and Qt  Qturbine_max.

COMBINING WATER SUPPLY & HYDROPOWER

If we are simulating the operation of a reservoir to produce water supply, we can compute the
release, Rt, required to meet the demand. If we are simulating the operation of a reservoir to
produce hydropower, we can compute the release, Qtreq, to meet the energy demand. Many
reservoirs, however, consider both water supply and hydropower. The release policy must then
be based upon the appropriate priorities for operation. For example, if both purposes are equally
important we could calculate a combined release as: Reltcomb = Max[Rt ; Qtreq ]

6
A. Application of LINEAR/NON-LINEAR PROGRAMMING Optimization
FOR: RESERVOIR OPERATION

OBJECTIVES FOR THE OPERATION/SIMULATION ANALYSIS

There are many objectives of reservoir analysis that we can use reservoir simulation
models to evaluate. The following are some of the more common objectives:

1. Minimize the amount of active storage volume to meet the annual water supply
demand
2. Maximize the annual water supply (also called Firm Water Yield) given the available
active storage volume and demand pattern.
3. Minimize the amount of active storage volume to meet the annual energy demand
4. Maximize the annual energy supply (also called Firm Energy) given the available active
storage volume and energy demand pattern.
5. Maximize the total annual energy without regard to an energy demand pattern and
the available active storage volume.
6. Max the annual Firm Energy constrained by meeting a given minimum level of annual
water supply
7. Max the annual water supply constrained by meeting a given minimum level of annual
energy production
8. Minimize the Maximum Water Shortage that occurs in any of the periods
9. Minimize the Total Squared Water Shortages
10. Minimize the Maximum Energy Shortage that occurs in any of the periods
11. Minimize the Total Squared Energy Shortages
12. Minimize a weighted combination of the total squared energy shortages and total
squared water shortages