You are on page 1of 7

Optimal Scheduling of DER Generation using MultiObjective

Optimization Algorithms
Srinka Basu1 , Avishek Banerjee2 , Soumyadip Ghosh3

Abstract The advancement of the DERs in modern smart One of the most significant green house gas has been CO2 .
grid systems has made a large impact on the energy market.A Carbon dioxide (CO2) and other similar greenhouse gases
lot of extensive research has been done on the efficient schedul- act like a covering over the earths atmosphere absorbing IR
ing of such renewable energy sources to reduce costs and also
to meet market demand efficiently. A prime aim of the smart radiation and not letting them escape in the outer atmosphere.
grid paradigm is to reduce emissions levels in the electricity The direct consequences is the rising of earths temperature
generation and distribution systems. Distributed Generation due to gradual heating. Most of the household use fossil
scheduling algorithms must also give adequate importance to fuels and other non renewable sources as distributed energy
emission levels while scheduling different DGs. In this paper,a resources which can produce electricity at a minimal cost.
comparative view is presented over a number of optimization
techniques which are adopted to model the DER system to But the amount of CO2 emission has been alarming. The
minimize both the energy cost and the environmental impact environmental constraints thus have become a major role
i.e. mainly the amount of CO2 emission. in this energy market. Thus the main objective here is to
minimize the c02 emission along with the cost minimization.
I. INTRODUCTION However these two objectives are competitive in nature. to
The utility industry across the world is trying to address solve such a problem a trade off is required among the two
numerous challenges, including generation diversification, objective functions and we get a pareto optimal solution to
optimal deployment of expensive assets, demand response, such two objectives. This paper gives an extensive solution
energy conservation, and reduction of the overall carbon to such biobjective optimization. The paper is organised as
footprint. It is evident that such critical issues cannot be follows. There is a problem formulation section which deals
addressed within the confines of the existing electricity grid. with the modeling of the Objective equations both the cost
The next-generation electricity grid is expected to address and emission and load determination. The next part called the
the major shortcomings of the existing grid.Smart grid gen- System models mainly concerns about two type of System
erally refers to a class of technology people are using to Modeling. Next comes the Solution section which mainly
bring utility electricity delivery systems into the 21st cen- deals with the various Multiobjective algorithms. The result
tury, using computer-based remote control and automation. sections gives the comparative view of various algorithms
These systems are made possible by two-way communication described.
technology and computer processing that has been used for
decades in other industries. They are beginning to be used II. PROBLEM FORMULATION
on electricity networks, from the power plants and wind
The main challenge of the modern DER system is to
farms all the way to the consumers of electricity in homes
optimize both the energy cost and the CO2 emissions.Thus
and businesses. They offer many benefits to utilities and
operational scheduling of DER systems inherently involve
consumers mostly seen in big improvements in energy
multiple and conflicting objectives. Therefore, mathematical
efficiency on the electricity grid and in the energy users
models become more realistic if distinct evaluation aspects,
homes and offices.
such as cost and environmental concerns, are explicitly
A smart grid is an evolved grid system that manages
considered by giving them an explicit role as objective
electricity demand in a sustainable, reliable and economic
functions rather than aggregating them in a single economic
manner, built on advanced infrastructure and tuned to facil-
indicator objective function.[]
itate the integration of all involved. [?]ABB. Now the main
The operation of Smart Grid Energy Market can be
utilization of the smart grid lies in the energy market. Using
broadly explained as follows. The energy market typically
the data from the Smart grid the optimization of the cost
has the following participants.
function of the energy market can be carried out which is
a major issue in the present world economy. Optimizing the A Central Power Station which is run by the grid
generation of DERs to minimize the overall market price operator.
has been a challenge and has been solved by many ways. DERs which generate and sell energy.
But the emission of the green house gases while producing Buyers who demand and bid for the energy
electricity has also been a major concern. Gas molecules The auction and trading occur at the predefined intervals
that absorb infrared radiation,from the sun in a significant and the total energy required for the next operational time
amount can result in various type of environmental changes. period is traded. The steps in a typical double auction are as
These type of gas are mainly known as greenhouse gases. follows.
1) The grid operator announces the price for buying or by fossil-fueled generating units can be modeled
selling energy to the grid. separately.i , i , i , i , i reprsents the emission coef-
2) The buyers submit their bids that constitute of energy ficients.
and price to the system operator.
3) The sellers submit their asks which constitute of energy B. Constraints
and price to the system operator. The constraints for this optimization problem are as
4) The system operator depending on the auction strategy follows.
determines the energy trade. 1) Each DG has some minimum and maximum produc-
5) The buyers and sellers participate in an iterative nego- tion levels and it can only produce energy within those
tiation to determine the buying and selling price. limits. Mathematically,
Any auction strategy can be used to trade energy between
the buyers and the sellers. Risk based, risk aversion are Pmin xi Pmax (3)
some of the trading algorithms which have been used in this
respect. Typically, the optimal DER generation routine if run 2) The total load or demand in the market at that point
after step (2) and before step (3) the auction process would of time must be satisfied by the combined production
become lot more efficient and fair of DERs and the Main Grid. Mathematically,
N K
A. Objective functions X X
xi + X = lit t (4)
This paper considers two objective functions, which i=1 i=i
quantify the total energy cost and the environmental impact,
where li is the individual loads of the various con-
both to be minimized. There are some operational constraints
sumers.
on the generator units as well.
1) Economic objective function C. Load Determination
A Cost function is defined which is the quantity to The load determination of the households is done with the
be minimized. The Cost represents the total economic help of previous universal market price and the Elasticity of
cost of producing electricity in the microgrid. It is demand factor or P ED. Price elasticity of demand (PED)
defined as the sum of a quadratic and a sinusoidal is a measure used in economics to show the responsiveness,
function :- or elasticity, of the quantity demanded of a good or service
T
N X to a change in its price. It has been formulated as follows
X
Cost = g(xji ) (1)
i=1 j=1 (Dreal )tj = (Dori )tj + P ricet P ED (5)

where g(xi ) = ai x2i +bi xi +ci +di sin ei (Pimin xi ) where Dori is the desired load and Dreal is the final load
xi represents the Generation of Each DER. g(xi ) rep- using the Elasticity of demand factor under consideration.
resents the cost function of each DER. ai , bi , ci , di , ei
represents the cost coefficients of ith unit. It is a sum- III. SYSTEM MODEL
mation of quadratic and sinusoidal function //Citation 24 hour market A whole day market simulation using
needed. N is the number of DG currently participating the price of the previous day to optimised outputs and
in the market. T denotes the 24 hour slot. the new price that can be used for the next day.
2) Environmental Objective Function Hourly market An hourly simulation using the price of
The environmental mainly takes into account the the last hour to determine the demand and simulating
environmental pollutants CO2 , N Ox and S0x emis- using the objective function getting the new optimum
sions of the various Distributed Generators. However, cost and the new price for the next hour ( Total
for comparison purposes, the total emission of these cost/demand). This model can be executed for every
pollutants is taken as the sum of a quadratic and an hour separately to get the simulation for the 24 hour
exponential function as follows. slot.
N X
X T IV. PROBLEM SOLUTION
Cost = f (xji ) (2)
Linear Weighted Sum Method
i=1 j=1
In this method the two objective function are linearly
2 combined into a single objective function using a vari-
where f(xji ) = i xji + i xji + i + exp(i xji )
represents the net summation of all pollutants emission able . This new Objective function is minimized using
of each DER. xji is the amount of energy produced by SQP. The value varies from 0.1 to 1 to give varying
DER i at time period j. relative importance to the economic and emission costs.
The atmospheric pollutants such as sulphur ox- Qbjectivef unction = ()Cost + (1 )Fe
ides (SOx) and nitrogen oxides (NOx) caused Differential Evolution
TABLE II
In evolutionary computation, differential evolution (DE)
24 HOUR DER GENERATION
is a method that optimizes a problem by iteratively
trying to improve a candidate solution with regard to Hour DER1 DER2 DER3 DER4 DER5 D
a given measure of quality. Such methods are com- 1 152.755522 135.421256 92.803898 93.711270 73.261494 60.2
2 164.664305 138.193275 75.313876 63.996530 160.969500 91.5
monly known as metaheuristics as they make few or 3 163.800223 152.854025 76.212069 83.850487 103.317886 67.0
no assumptions about the problem being optimized and 4 151.128449 138.565739 74.831676 78.481963 74.088584 66.6
can search very large spaces of candidate solutions. 5 154.950182 146.927749 75.072591 60.978782 111.997731 64.6
However, metaheuristics such as DE do not guarantee 6 152.532761 137.166274 120.962957 93.460384 102.293199 117.
7 150.784761 144.132157 78.983035 60.236574 113.968810 100.
an optimal solution is ever found. 8 165.464741 141.642310 107.119717 91.466790 76.395307 68.4
DE is used for multidimensional real-valued functions 9 168.193742 136.393585 82.375417 67.154104 81.982058 93.9
but does not use the gradient of the problem being 10 152.698205 144.973390 127.217186 66.108099 75.078400 90.4
11 151.512471 135.049240 77.770271 78.548821 89.523672 61.7
optimized, which means DE does not require for the 12 170.134093 138.930404 76.203664 60.931487 82.298209 65.5
optimization problem to be differentiable as is required 13 165.751537 136.191863 82.815656 74.207506 79.192845 59.1
by classic optimization methods such as gradient de- 14 150.379961 136.224804 103.162613 104.948627 91.280288 66.5
scent and quasi-newton methods. DE can therefore also 15 152.315557 136.609565 80.296944 60.176439 112.841533 102.
16 153.553471 140.519444 84.094702 79.313099 75.580903 105.
be used on optimization problems that are not even 17 154.782290 135.772906 82.070783 65.288188 75.608858 108.
continuous, are noisy, change over time, etc.[wiki] 18 151.350977 139.136316 132.791453 110.644261 93.096476 64.7
Multi Objective Optimization Methods 19 150.315291 137.838326 93.276878 87.873419 88.115434 58.2
In this method the cost and the environmental impact 20 157.294521 135.455086 74.293712 126.525732 74.635802 58.3
21 151.470629 135.229053 73.272744 97.640883 103.952922 57.3
functions are taken as two separate objective functions 22 165.384786 136.351706 103.172569 61.644554 106.505684 72.1
and are optimized using the following optimization 23 152.518158 148.678513 73.245042 62.374953 120.652123 57.9
techniques: 24 151.331081 145.738522 80.392025 82.727843 106.251333 74.3
1. NSGAII
2. eMOEA
3. eNSGAII
4. OMOPSO
5. PESAII
The various parameters used to compare the MultiOb-
jective Optimization Techniques are :-
V. E XPERIMENTAL SETUP AND DATA DESCRIPTION
BEFORE R ESULTS SECTION
VI. R ESULTS AND D ISCUSSION
In this section we present the results of the two models
that we have discussed above.
A. Model 1
This is the 24 hour model.The simulation is done with
the help of NSGA II algorithm. Table 1 gives the whole
day simulation result for the two objective functions.Table
II gives the overall DER generation for the whole day.The
Fig. 1. 24 hour DER generation
Figure 1 gives a comparative view of the market price
before(original) and after(new) simulation
TABLE I 1) NSGAII: The table III gives the Objective I and Ob-
NSGAII jective II result of the objective function using NSGA II
algorithm.
Objective1 ObjectiveII
2) PESA: The table V and table VI
9.08 105 5.01 104
3) eOMEA: Table VII and table VIII
4) OMOPSO: Table IX and X
B. Model 2 In case of NSGA II a detailed comparison on the price is
This is the 2nd model which is mentioned earlier.It is a performed on consecutive 5 runs. Figure II shows that
hourly model where every hour the generation of DERs are Here the R1,R2,R3,R4,R5 are the 5 computational runs
calculated and the price for every hour is calculated which is done on the basis of model 2 as described in the System
used as a market price for the next hour. Different algorithms Model Section where every hour the price is calculated which
are used to have a comparative view of the model. is utilized for the next hour demand calculation.
TABLE IV
NSGAII

TABLE III
Hour Objective1 ObjectiveII
ACTUAL 24 HOUR D EMAND
1 42725.1246 2816.3329
2 46067.5694 3366.0408
Hour Load
3 58516.0881 5650.5919
1 1036.000000
4 81059.7353 8678.6689
2 1110.000000
5 80948.1831 8593.6856
3 1258.000000
6 105064.2365 14771.6493
4 1406.000000
7 113078.8380 20140.1208
5 1480.000000
8 107393.2663 17179.2145
6 1628.000000
9 129036.0152 29581.5133
7 1702.000000
10 110468.6714 17308.5340
8 1776.000000
11 128147.0217 26692.6679
9 1924.000000
12 120897.0207 20911.2966
10 1942.000000
13 110685.2142 17065.6889
11 1978.000000
14 109797.6994 16566.3234
12 1984.000000
15 100778.7524 14766.0277
13 1942.000000
16 86920.5190 9811.8528
14 1924.000000
17 69929.6191 8537.2800
15 1776.000000
18 91667.2599 11061.5173
16 1554.000000
19 89313.9469 10790.6418
17 1480.000000
20 115882.3206 19128.6696
18 1628.000000
21 121090.1102 20845.6998
19 1776.000000
22 99350.5750 19188.1686
20 1972.000000
23 66486.0778 7611.3280
21 1924.000000
24 52168.5706 5598.5330
22 1628.000000
23 1332.000000
TABLE V
24 1184.000000
NSGAII PRICE

Hour Price
1 58.049808
2 57.015258
3 58.563275
4 67.186268
5 64.679302
6 77.000561
7 80.303332
8 75.279505
9 81.958281
10 69.865739
11 79.824208
12 74.603123
13 69.952168
14 69.533380
15 68.897875
16 69.126254
17 60.611740
18 72.263737
19 64.834528
20 73.154085
21 76.630951
Fig. 2. Market Price Comparison 22 74.657647
23 64.865072
24 57.428280
TABLE VII
PESA

Hour Price
1 58.637367
2 59.408526
3 59.611566
4 69.208577
5 54.874538
6 76.519063
7 76.666306
8 68.943800
9 65.623058
10 82.991302
11 76.347539
12 76.088738
13 72.279272
14 71.348862
Fig. 3. Market Price Comparison 15 72.673913
16 78.442858
17 62.815603
18 74.582912
19 76.254697
20 81.486253
TABLE VI 21 73.499370
PESA 22 68.037354
23 70.885233
Hour Objective1 ObjectiveII 24 57.969224
1 43157.5721 2609.6762
2 48001.2978 3109.0221
3 59563.5344 6625.5869
4 83499.6358 10756.0742
5 68677.2119 10088.0519
6 104407.2515 18368.2063
7 107957.3733 16902.1842
8 98354.7894 14483.6065
9 103317.6625 14988.5979
10 131222.2403 32398.1557
11 122565.6976 23476.5291
12 123304.5141 22513.0017
13 114367.3877 23828.9779
14 112664.4632 17283.7480
15 106302.0627 15625.3244
16 98635.3747 12958.3721
17 72472.2832 7704.3415
18 94609.1566 11505.7802
19 105046.0011 15201.9226
20 129081.1869 28496.5470
21 116141.6717 19789.3280
22 90540.6284 12278.2197
23 72656.6858 7795.9111
24 52659.9708 4004.2514
Fig. 4. Market Price Comparison
TABLE VIII
E MOEA

Hour Objective1 ObjectiveII


1 42519.9293 2772.8864
2 49782.8339 3232.4009
3 60088.9944 4941.2496
4 87783.7723 10752.5290
5 76781.3588 11857.6371
6 105464.5226 18284.1793
7 105929.9246 19477.2601
8 104171.4964 15837.2055
9 119593.8226 19993.3450
10 115780.6890 18598.1492
11 111699.3821 16828.7496
12 114553.3945 18909.1046
13 113789.4833 18598.6429
14 112627.8476 17181.4133
15 118008.8762 23304.5465
16 87849.9863 11110.0704
17 69053.4944 6999.7264 Fig. 5. MOEA
18 87535.9467 11812.6150
19 89714.3808 10105.7470
20 109117.0058 18831.0061
21 109118.1634 16187.3187
22 79967.9161 12035.5055
23 61591.5443 6585.1362
TABLE X
24 59714.3880 5007.1288
MOPSO
TABLE IX
Hour Objective1 ObjectiveII
E MOEA
1 41640.0788 2696.2012
Hour Price 2 47878.8893 3241.4238
1 57.771013 3 59097.2296 7556.6531
2 61.613434 4 76694.7527 7855.8991
3 60.137450 5 82036.7981 12297.9969
4 72.759478 6 87528.3058 12851.6705
5 61.349922 7 96944.0250 15322.8563
6 77.293927 8 103350.8685 14890.7392
7 75.226507 9 126178.2416 24735.5819
8 73.021140 10 126671.6213 28217.8133
9 75.960995 11 129270.8514 27974.7347
10 73.225317 12 130532.1986 30399.9587
11 69.578790 13 116564.8959 20786.1937
12 70.688598 14 124205.8447 24685.1087
13 71.914041 15 105422.8692 19047.9763
14 71.325674 16 108012.9163 18793.6952
15 80.677332 17 69827.1928 6822.3859
16 69.865442 18 83346.5387 11461.7851
17 59.852356 19 100133.4848 14249.4889
18 69.006913 20 124015.5442 25409.6523
19 65.125210 21 124553.0393 25448.8893
20 68.883283 22 86422.6071 10702.4683
21 69.054596 23 63753.7584 5819.6777
22 60.092419 24 53090.4640 4349.6039
23 60.089873
24 65.734877
TABLE XI
MOP

Hour Price
1 56.575577
2 59.257028
3 59.144886
4 63.568357
5 65.549129
6 64.148647
7 68.845139
8 72.445904
9 80.143142
10 80.113270
11 80.524254
12 80.548797
Fig. 7. Comparison of 5 runs
13 73.668080
14 78.657861 TABLE XII
15 72.072848 DATA
16 85.900641
17 60.522962 unit P maxi P mini ai bi ci di ei i
18 65.704291 1 150 470 786.7988 38.5397 0.1524 450 0.041 103.3
19 72.688617 2 135 470 451.3251 46.1591 0.1058 600 0.036 103.3
20 78.288419 3 73 340 1049.9977 40.3965 0.0280 320 0.028 300.3
21 78.822439 4 60 300 1243.5311 38.3055 0.0354 260 0.052 300.3
22 64.942840 5 73 243 1658.5696 36.3278 0.0211 280 0.063 320.0
23 62.199369 6 57 160 1356.6592 38.2704 0.0179 310 0.048 320.0
24 58.443120

VII. CONCLUSIONS
Thus along with the minimization of the cost function the
emission of CO2 minimization makes the system of DER
generation much more challenging and realistic. eNSGAII
gives farmost the best result for optimizing this bi objective
function. However more more than two objective functions
the result obtained may not be the best one

Fig. 6. OMOPSO

You might also like