Optimal Pump Scheduling For Peak Load Management: Abstract-In Water Pumping Systems, There Is

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

228

Optimal Pump Scheduling for Peak Load Management

S. Ashok and R. Banerjee

Abstract- In water pumping systems, there is pumping, about 10% of the total expenditure of a
often a flexibility of scheduling the pumps. The typical Indian city [1]. Presently the pumping
optimal schedule minimizes the peak electricity schedules or storage capacity are not determined by
demand and shifts the load to off peak period the electricity tariff.
based on the electricity tariff. The customer gets Many state owned electricity boards (SEBs) in India
the benefit of reduction in electricity cost. Peak are facing peak demand deficit and energy shortage.
saving by shifting the load to off peak with time Scheduling of industrial or process loads has been
of use (TOU) rates is the benefit to utility. This identified as a tool for peak load management [2,3].
paper presents a methodology for determining Reports on potential of TOU tariff for load
the optimal schedule of pumps satisfying management (LM) and corresponding load response
pumping, storage and water consumption are also found in literature [4,5]. In order to induce
constraints. The formulation is applied with the consumers to reschedule their loads from peak to
objective of minimizing the total electricity cost off- peak periods, many SEBs are introducing TOU
under different tariff structures. A case study tariff for large consumers instead of flat tariff. Water
conducted for a typical campus shows 27 % pumping load is significant and affords a cost
reduction in peak period coincident demand with effective option for peak load management. Pumps
the optimal schedule under TOU tariff. Pump can be scheduled to take advantage of the off-peak
schedule with optimal storage results in 14 % rates, minimising the electricity cost. This also
saving in the electricity cost and 48 % reduction reduces the peak coincident demand resulting in
in peak demand under TOU tariff. The peak saving for the utility.
corresponding saving in peak coincident demand In the case of pumping with storage like municipal
for the utility is 53 %. water pumping, pumps can be operated at its best
Index terms- Load management, Pumping, TOU efficiency point, matching with system
rates characteristics subject to storage limits and
1. INTRODUCTION consumption requirement. Then the control is only
Water pumping is an energy intensive service. As on or off the pumps at an optimal schedule to
water needs grow and electricity cost increases, minimize the peak demand and hence the electricity
water service related energy use and associated costs cost, satisfying the constraints. Flexibility in
become more difficult to afford. On an average 70% selecting the pumps as per the load and efficiency
of typical municipal electricity bill goes to water characteristics in a multi-pumping system will also
Ashok S. (corresponding author, e-mail:ashoks@me.iitb.ac.in) reduce the energy consumption.
and Rangan Banerjee, (e-mail:rangan@me.iitb.ac.in, Fax:022-
Case studies of scheduling of water, oil and milk
5726875) are with Energy Systems Engineering, Indian Institute
pumping plants are reported [6,7]. The formulations
of technology Bombay, Mumbai, 400076, India.
used in these case studies are restricted to the

Energy Automation
229

specific configurations and hence cannot be (3) Decision variable -The decision variable tmi
extended to general applications. A general represents the operating time (in hours) of the pump
formulation for pump scheduling incorporating m with in an interval i such that
pumps and system characteristics is useful for both tmi = 0 the pump m is off in the interval
the consumer and utility. As optimal pump schedule tmi =Di the pump m is on for the entire interval
directly relates to storage capacity limit, the model 0 < tmi < Di the pump m is on for a fraction of
can provide the optimal storage capacity for a given the interval.
system considering the cost parameters. This paper (4) Storage constraint - With M pumps inflows and
presents a formulation to determine the optimal O distribution out flows for the storage tank, the net
scheduling of pumps satisfying the constraints, inflow, Fi to the tank for the interval i is
under specified tariff structure to minimise the M O

electricity cost. It also gives the optimal storage Fi = (Q mi * t mi ) ( q oi ) (5)


m =1 o =1
capacity. The model is illustrated with a case study. Total water available in the tank at the end of ih
interval,
11. FORMULATION
i

(1) Pump power, Pp is given in terms of S i = S 0 + Fi ( k 1) ( 6)


k =1
(specific gravity), Q (discharge) and H (differential where qoi is the outflow for consumption o in the
head) at the operating point, as interval i and S0 is the initial water level in the tank

Pp = ( * Q * g * H ) (1) at the start of the planning horizon.


Maximum storage capacity (Sm) constraint is then
Corresponding motor input power, Pm
modeled as
Pm = Pp /( e p * em ) ( 2)
Si S m (7)
where ep and em are the efficiencies of pump and
If the suction is from a sump tank or sump well,
motor respectively at the operating point. When Pm
incorporating a similar constraint can provide a
is in kW, electrical demand in kVA of the pump
minimum water level for safe pumping.
with power factor of operation pfm is
(5) Consumption constraint-The required
MDm = Pm / pf m (3)
consumption for the interval i must be met.
(2) Time intervals - As pumps can be operated under M O

best efficiency point matching with the system Qmi * tmi + Si1 qoi
m =1 o =1
(8)
characteristics at all times, the on-off control
(8) Availability of machines -During the
formulation is based on discrete time representation
maintenance interval, the pump is not available for
of the entire time horizon, H of interest (e.g. one day
operation. So the total availability of pump m in the
or one month). Taking equal time duration for all the
planning horizon is
intervals, with D as the time duration (in hours) of
N
the interval i and total number of intervals N, such t
i 1
mi Tm ( 9)
that
where Tm is the total time available in the planning
N

D
i =1
i =H ( 4) horizon for the pump m.

Energy Automation
230

(9)Objective function- The objective is the Three pumps are installed with specifications
minimization of the total electricity cost, which (collected from the manufacturers characteristics)
consists of charges for energy consumed and charges as shown in Table 1. The water tank storage capacity
for the registered maximum demand (MD). Energy is 1000 kl. The availability of pumps is maximum 18
charge is either flat or TOU while the demand hours in a day. Static capacitors are installed in the
charge is not time varying, viz., constant irrespective control panel maintaining a power factor of 0.98.
of the time when the MD is registered in a month. Load management actions considered are (1)
Extending the daily schedule to a planning horizon optimal schedule of pumps with the existing storage
of one month, we have monthly energy cost (Eg), (2) optimal storage size with schedule of operation.
N M SEBs have various tariff structures for water
E g = ( Pmi * t mi ) * Ci (10)
pumping depending on customer class. The optimal
i =1 m =1
response has been studied under the MSEB
when Ci is the cost of energy (Rs./kWh) for the
(Maharashtra SEB) HT industrial flat tariff and the
interval i . The objective function is to minimize
recently introduced TOU tariff as given in Table 2.
monthly electricity cost (Ec ),
Table 1. Rated specifications of Pumps
E c = Min.[ E g + C d * MD ] (11)
where Cd is the MD charge (Rs./kVA) and MD is the
registered maximum demand of the pumping station.
The additional cost incurred for load management
actions (like capital cost due to increasing the
storage capacity) if any, is added to the above
Table. 2. Electricity tariff rates
objective function to minimize the total operating
cost. Monthly additional cost for LM action,
N
E L = C ai * I ai * Di (12)
i =1

Where Cai -the additional cost (Rs./hour -additional


operating cost or annualized cost of investment) of
LM actions for the interval i., and Iai - the selection IV. Results
variable 1, if the LM action corresponding to Cai Optimal schedule for the existing storage- Load

is taking place or 0 otherwise. curves of optimal schedule for the existing storage

Addition of MD charge or cost of LM actions makes are shown in Fig.2 for the flat and TOU tariff rates.

the formulation non linear. It is solved by conjugate Optimal responses of both flat and TOU tariff results

gradient method. in 196 kVA peak demand. Average demand during

III. Case study the utility peak period is reduced by 21 % (from 113

To illustrate the model a case study of water kVA to 90 kVA) under flat tariff and 42 % (from

pumping for a typical institute campus is done. Daily 113 kVA to 65 kVA) under TOU tariff as compared

water consumption of the campus is estimated as to load following strategy. Daily energy

10230 kl with hourly variation shown in Fig.1. consumption gets reduced by 4.2 and 3.5 % under

Energy Automation
231

flat and TOU tariffs respectively from 1986 kWh. the optimal strategy under flat tariff reduces
The annual electricity cost with load following electricity bill by 3.4 % and 4.8 % as compared to
strategy is 30.9 and 32.1 lakhs with flat and TOU load following strategy. The average demand during
tariff rates. The saving in electricity cost with the the utility peak is reduced by 42 % with TOU tariff.
optimal schedule is 3.4 and 4.8 % respectively with The model is also used to determine the optimal
flat and TOU tariff rates. storage size. With optimal storage, operating
(2) Optimal storage size with scheduling- strategy reduces the peak demand by 48 % and the
Scheduling is done with the optimal storage capacity average demand during the utility peak by 53 % with
instead of the available storage. The additional cost TOU tariff. Pump scheduling provides a cost
of the storage capacity is annualised (with 30 % effective tool for LM.
discount rate and 20 years of life) and added to the VI. References
1. Report on Municipal Water Efficiency Program
objective function. The cost of the construction of
by The Alliance to Save Energy, NW,
the concrete storage tank is estimated with a fixed
Washington DC, USA, Report No. 20036,1996
cost of Rs. 50,000 and a variable cost proportional to
(www.asc.org).
the surface area at the rate of Rs. 4000/m3 for
2. S.I. Gustaffson, Load Management in Municipal
concrete.
Electricity Systems, Int. Journal of Energy
Flat tariff results in the optimal storage of 1338 kl
Research, Vol.21 (9),1997, pp 787-791.
with an investment of Rs.9.34 lakhs while TOU
3. Zonghui Iou ,R.Kumar, J.Sottle and J.C.Yingling,
tariff results in 1770 kl with an investment of Rs.
"An MILP formulation for Load side Demand
12.2 lakhs. Under flat tariff, the trade off is between
Control", EM&PS., Vol. 26, 1998, pp 935-949.
the annualised storage cost and MD charge, while
4. J.N.Sheen, C.S. Chen, J.K. Yang, Time of Use
under TOU tariff, trade off is between MD charge
pricing of Load Management program in Taiwan
and differential energy rates with annualised storage
Power Company, IEEE Trans. on PS, Vol. 9, (1)
cost. MD gets reduced to 102 kVA with a peak
Feb. 1994, pp 388-396.
coincident demand of 53 kVA under TOU tariff.
5. R.Banerjee, Load management in the Indian
Reduction in peak coincident demand is 41 % with
power sector using US experience, Energy,
the TOU tariff compared with that under flat tariff.
Volume 23, (11), Nov. 1998, pp 961-972
Table 3 gives the results of optimal storage and the
6. Beckwith SF, K.P.Wong, A GA approach for
corresponding load curves are shown in Fig. 3.
electric Pump scheduling in water supply
Table 3. Optimal response with optimal storage
system, IEEE Int. Conference on Evolutionary
Computing, Perth, Nov. 1995, pp 21-26.
7. C.Basnet, L.R.Foulds, A Decision Aid for milk
tanker Pump Scheduling, Research project
report series-1996-(1), Dept. of Management

V. Conclusion Systems, University of Waikato, NZ.


An optimization model is presented to determine the
optimal schedule for water pumping system for a
specified electricity tariff. A case study shows that

Energy Automation
232

Fig. 1 Water consumption pattern

Fig. 2 Load curves of optimal schedules with existing storage

Fig. 3 Load curves of optimal schedules with optimal storage

Energy Automation

You might also like