Distribution Grid Planning Enhancement Using Profiling ...
Post on 05-Jun-2022
3 Views
Preview:
Transcript
HAL Id: hal-01392912https://hal.archives-ouvertes.fr/hal-01392912
Submitted on 5 Nov 2016
HAL is a multi-disciplinary open accessarchive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come fromteaching and research institutions in France orabroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, estdestinée au dépôt et à la diffusion de documentsscientifiques de niveau recherche, publiés ou non,émanant des établissements d’enseignement et derecherche français ou étrangers, des laboratoirespublics ou privés.
Distribution Grid Planning Enhancement UsingProfiling Estimation Technic
Siyamak Sarabi, Arnaud Davigny, Vincent Courtecuisse, Léo Coutard, BenoitRobyns
To cite this version:Siyamak Sarabi, Arnaud Davigny, Vincent Courtecuisse, Léo Coutard, Benoit Robyns. DistributionGrid Planning Enhancement Using Profiling Estimation Technic. CIRED Workshop 2016, Jun 2016,Helsinki, Finland. �hal-01392912�
CIRED Workshop - Helsinki 14-15 June 2016Paper 0171
Distribution Grid Planning Enhancement Using ProfilingEstimation Technic
Siyamak Sarabi∗, Arnaud Davigny∗, Vincent Courtecuisse†, Leo Coutard†, Benoit Robyns∗
∗L2EP-HEI, France, siyamak.sarabi@hei.fr, †GEREDIS Deux-Sevres, France
Keywords: Distribution grid, Profiling estimation, HV/MVsubstation, MV/LV substation, Artificial neural networks,
Abstract
In this study, the load profile estimation is done using theprofiling technic, and its performance is enhanced using thereal consumption values of the consumers. The results of es-timation are evaluated using the measured data at the levelof HV/MV (High Voltage/Low Voltage) substation and alsoMV feeders containing industrial and residential consumersand wind power productions. Based on the gap between ac-tual measured data and estimated profiles, a correction factoris introduced that is applicable at the level of single MV/LVsubstations. These factors will be used for grid simulation andplanning studies where the results will be more reliable com-pared to the previous estimation technics.
1 Introduction
Planning of power distribution system is one the main respon-sibilities of Distribution System Operators (DSO) which willprovide them the appropriate decision making strategies re-lated to the distribution grid’s state and reinforcement neces-sities [1]. The growing nature of electricity demand enforcesthe DSO to have a perspective view on grid reinforcement ca-pacity. This perspective should be in different temporal termfrom short term to long term perspective (typically a ten yearinvestment plan is considered) [2]. Historically, distributiongrids were sized to ensure the consumption peak, the massiveintegration of distributed generation requires today to take itinto account in the design of the distribution grid [3].
The mastery of the short-term consumption (week/month)allows optimizing operations and driving patterns of grids(joules limit losses and minimize routing costs, operating planin case of accident). In other side, the future smart grid con-tributions to the distribution grid need massive flow of infor-mation through the communication infrastructure. Advancedmetering infrastructure are growing to prepare actual distri-bution grid for future smart grid [4]. The metering infras-tructure is supposed to be available for each single consumerpoint. While currently, the actual measuring units provide onlythe consumed energy of the consumers but not the real power.Hence having a load profile of the single consumer is not pos-sible.
Knowing that, the real power consumption of the network isnecessary for planning and reinforcement studies, estimativetechnics provide possible solution to the problem prior to com-munication infrastructure development. As the actual reliablemeasurement of the distribution grid state is at the level ofHV/MV substation (230-63 kV/20-15 kV), an estimation tech-nic can calculate the actual load profile of the downstreamconsumers and evaluate its precision using measurement atHV/MV substation level.
In this paper, the aim is to reconstruct the load profile of theMV/LV substations using profiling technics and enhancementstrategies presented in next sections. In section 2, data prepro-cessing are explained which prepares the appropriate data forthe algorithm. After that, in section 3, the load profile at thehead of MV feeder will be reconstructed then in section 4 theprofiling technic is explained and applied in one feeder as casestudy.
2 Data pre-processing
In order to do the profiling calculations the necessary data mustbe prepared. Each type of data comes from the different database. As manager of the distribution grid of Deux-Sevres de-partment, GEREDIS, has access to certain information at dif-ferent points of the electricity grid. The structure of data basepreparation are depicted in Fig.1.
To calculate the profiles the following values are required;
1. The annual consumption of each client from billingserver.
2. Information of the contract and price of each customer onbilling server.
3. The reference of each customer to find out the position ofthat client on the network (available on billing server).
4. The distribution substation, the MV feeder and HV/MVsubstation position of each client on Geographic Infor-mation System (GIS) server.
3 MV feeder profile correction
The substation presented in Fig. 2 is located at the border be-tween TSO (Transmission System Operator) and DSO. Eachsubstation normally contains 3 HV/MV transformers. The
1
CIRED Workshop - Helsinki 14-15 June 2016Paper 0171
Fig. 1: The structure of profiling estimation using GIS and billing servers’ data bases.
C.B.
36 MVA90/15 kV
TR1TR2TR3 36 MVA90/15 kV
36 MVA90/15 kV
RES1RES2RES3
Feeder1
Feeder2
Feeder3
Feeder4
Feeder5
Feeder6
Feeder7
Feeder8
Feeder9
Feeder10
Feeder11
Feeder12
Feeder13
Feeder14
Feeder15
Feeder16
Feeder17
Fig. 2: The HV/MV substation diagram with its MV feeders.
measurement of power at 10 minutes sample time is providedat the level of each transformer. These measurements are usedfor billing the energy transmission cost and so they are accurateand available continuously. Second level of measurement arelocated at the head of each MV feeder containing both con-sumption and production feeders. The measurements at thislevel are not reliable and there are a lot of communicationslost. Hence, the first step is to find the complete data using thevalues of measurement of first level at transformer. This step isdone using an Artificial Neural Network (Fig. 3). The resultsare brought in Fig. 4. The algorithm tries to learn from avail-able data to estimated the missed data. As the inputs of ANN,energy price (Eprice), day type vectors (Day1−7type, Day1−365type ,Day1−144type , Day0−1type) and temperature are used. Afterwards,using the corrected profile of the feeder, the profiling estima-tion will be examined. The correction procedure should followthe rules described below.
PTR1 = Pfeeder1 + ...+ Pfeeder5 − PRES1 (1)
PTR2 = Pfeeder6 + ...+ Pfeeder12 − PRES2 (2)
PTR3 = Pfeeder13 + ...+ Pfeeder17 − PRES3 (3)
Where PTRx is the power of transformer TRx which includesthe summation of all consumer feeders (Pfeeders) and sub-straction of local production in PRESx.
4 Enhanced Profiling technic
Profiling is a statistical technic for modeling of consumption(or production) applied to a group of clients [5]. As lead man-ager of the distribution network, ERDF, is responsible for es-tablishing and making available the national profiles of marketplayers’ consumption and production contract applied to re-plenish the half-hourly energy flows through responsible bal-ance.
The methodology of profiling based on annual consumptionvalues of each customer for the various sub profiles in eachprofile type is explained on forms of below equations.
2
CIRED Workshop - Helsinki 14-15 June 2016Paper 0171
l1n2
l1n3
l1n4
l1n5
l1n6
l1n7
l1n1
l1n8
l1w1
l1w2
l1w3
l1w4
l1w5
l1w6
l1w7
l1w8
PTR
PRES
Temp
Day1−7type
Eprice
Day1−365type
Day0−1type
Day1−144hour
l2n2
l2n3
l2n4
l2n5
l2n6
l2n7
l2n1
l2n48
l2n8
Hidden 1
l3n2
l3n3
l3n4
l3n5
l3n12
l3n1
Hidden 2
l4n2
l4n3
l4n4
l4n5
l4n1
Output
Input
Pfeeder1
Pfeeder2
Pfeeder3
Pfeeder4
Pfeeder5
Fig. 3: ANN model for missed data estimation at MV feeder.
P jprofiling =
k∑i=1
(HPi
2Hy× Ej
y
)× P j
pf (t) (4)
P jprofiling =
(Ej
y
2Hy× P j
pf (t)
)(HP1 +HP2 + · · ·+HPk)
(5)(HP1 +HP2 + · · ·+HPk) = Hy (6)
P jprofiling =
(Ej
y
Hy× P j
pf (t)
)(7)
Where P jprofiling is the annual load profile of the consumer j
containing the k typical consumption periods. This consumerhas a typical consumption profile defined by P j
pf (t), with itsannual consumption of Ej
y in kWh.
Afterwards, the sum of profiles of all consumers under theMV/LV substation z will give the profile of this substation.For MV feeder w, the profile is given by (9).
P zprofiling =
J∑j=1
P jprofiling (8)
Pwprofiling =
Z∑z=1
P zprofiling (9)
The estimated profile of the MV feeder w will be comparedwith the corrected measured data of the feeder calculated inprevious section. The results are shown in Fig. 5. The differ-ence is presented as an error factor as follows:
Errorw(t) = Pwmeasured(t)− Pw
profiling(t) (10)
Fig. 4: MV feeders’ profile correction for transformer 1, TR1.
Using the error profile, the modified profile of MV/LV subsa-tion z is calculated in (11).
P zMprofiling(t) =
[Errorw(t)×
P zprofiling(t)
Pwprofiling(t)
]+P z
profiling(t)
(11)In fact, the difference between measured profile and estimatedprofile at the level of MV feeder is propagated in the profiles ofthe MV/LV substations under this feeder as shown in Fig. 6. Inother word, the estimated profiles’ accuracy is enhanced usingmeasured data at MV feeder level. This algorithm is applied oneach single MV/LV substation to calculate its profile. An illus-trative example of some substations’ profiles are given in Fig.7. These profiles are now more reliable than only estimationtechnics as their accuracy are enhanced using real measureddata at upper levels.
5 Conclusion
In this paper, an estimation technic for load profile reconstruc-tion was presented. As there is no communication infrastruc-ture at the level of MV/LV substations, the profile of these sub-
3
CIRED Workshop - Helsinki 14-15 June 2016Paper 0171
1 2 3 4 5
x 104
500
1000
1500
2000
1 year
Pow
er k
W
Corrected real measurementProfiling
Fig. 5: Comparison of profiling output with corrected realmeasurement profile of MV feeder. Upper subplot: oneyear profile for feeder 13, Middle subplot: one week ofsummer, Lower subplot: one week of winter.
Day 1 Day 2 Day 3 Day 4 Day 5 Day 6 Day 7
2
3
4
5
Pow
er k
W
MV/LV substation #1
Day 1 Day 2 Day 3 Day 4 Day 5 Day 6 Day 7
3
4
5
6
7
8
Pow
er k
W
MV/LV substation #2
Day 1 Day 2 Day 3 Day 4 Day 5 Day 6 Day 7
6
8
10
12
14
Pow
er k
W
MV/LV substation #3
Modified profile Estimated profile
Fig. 6: Comparison of profiling output with modified profileof 3 MV/LV substations.
Fig. 7: Illustration of some profiles at MV/LV substation level.
stations are valuable in term of distribution grid planning anddimensioning. Load profiles of MV/LV substations were es-timated using profiling estimation technics. In addition, theiraccuracy were enhanced using real measurement data firstly atHV/MV substation level then at MV feeder level. The recon-structed MV/LV substations’ profiles can be used for distribu-tion grid planning and simulation purposes. The condition ofeach substation under its enhanced estimated profiles can beevaluated from distribution grid operator point of view.
References
[1] S. Barsali, G. Celli, M. Ceraolo, R. Giglioli, P. Pelac-chi, F. Pilo, Operating and planning issues of distribu-tion grids containing diffuse generation, in: Electric-ity Distribution, 2001. Part 1: Contributions. CIRED.16th International Conference and Exhibition on (IEEConf. Publ No. 482), Vol. 4, 2001, pp. 5 pp. vol.4–.doi:10.1049/cp:20010846.
[2] B. Robyns, B. Francois, A. Davigny, J. Sprooten, A. Hen-neton, Electricity production from renewables energies,Wiley, 2012.
[3] S. Sarabi, A. Davigny, Y. Riffonneau, V. Courtecuisse,B. ROBYNS, Contribution and Impacts of Grid Inte-grated Electric Vehicles to the Distribution Networksand Railway Station Parking Lots, in: 23rd InternationalConference and Exhibition on Electricity Distribution(CIRED 2015), CIRED 2015, Lyon, France, 2015, pp.1–5.
[4] A. Bouallaga, R. Kadri, V. Albinet, A. Davigny, F. Colas,V. Courtecuisse, A. Merdassi, X. Guillaud, B. Robyns,Advanced metering infrastructure for real-time coordina-tion of renewable energy and electric vehicles chargingin distribution grid, in: CIRED Workshop, 2014.
[5] Profiles and profiling, ERDF, http://www.erdf.fr/profiles-and-profiling (2014).
4
top related