Académique Documents
Professionnel Documents
Culture Documents
Abstract This paper presents a planning model for distribution quality and reliability such as improve bus voltage profile,
systems considering various energy supply options such as reduce line losses and network congestion.
distributed generation (DG), substations, and feeders. In There is a large body of literature that focus on traditional
addition, the impact of Plug-in-Electric Vehicle (PEV)
uncontrolled and smart charging loads on the plan outcome is distribution system planning which involves optimal feeder
evaluated. A new optimal power flow (OPF) based optimization design, configuration, substation design, etc. [1], while many
model is proposed to schedule PEV uncontrolled and smart recent works consider the problem of optimal DG sizing and
charging loads. Test results are presented to demonstrate the siting in distribution systems [2]. In addition, a large and
effectiveness of the proposed model. The results show that PEV growing body of literature have been devoted to integrate
charging loads significantly affects the plan outcomes.
PEVs into power system problems. For example, in [3] [4]
Index Terms-- Plug-in electric vehicles, distributed generation, models are proposed for siting and sizing DG units to
distribution system planning. mitigate the effect of PEV charging loads. While in [5], an
integrated planning model is proposed considering PEV
I. INTRODUCTION charging loads to optimally allocate substation and feeders.
So far, planning of distribution systems considering
Major changes in planning paradigms have taken place in various energy resources, namely, DGs, feeders and
power systems in recent years because of deregulation of the substation, and including PEV charging loads (both
power industry, environmental policy changes, advancements uncontrolled and smart), has not been reported in the
in technology, and the transformation of the grid to intelligent literature. It is therefore important to investigate the effect of
systems, referred to as the smart grid. These changes will PEV uncontrolled and smart charging loads on the plan
continue to drive the distribution systems planning function outcomes considering all these energy resources. In this
to evolve in the coming years. work, a planning model for distribution systems is proposed
In addition, with the increase in gas prices driven by a considering distributed generation (DG), substations, and
foreseeable fossil fuel depletion in the future, development in feeders as plan options while investigating the effect of PEV
the automotive sector, and environmental concerns, uncontrolled and smart charging loads.
penetration of Plug-in-Electric Vehicles (PEVs) has been The structure of the paper is as follows. In Section-II, the
increasing in recent times. The charging load of PEVs will mathematical model of PEV allocation is presented, which is
definitely impact the distribution grid. Therefore, the impact followed by formulation of the distribution system planning
of PEV charging loads on distribution system planning need model in Section-III. In Section-IV, the proposed framework
be investigated and included in the planning process. is applied to a 33-bus radial distribution system and the
Distributed generation (DG) is expected to be a part of the results are presented. Conclusions are drawn in Section-V.
solution of distribution system planning challenges. From an
environmental prospective, use of renewable energy based
DG will reduce emissions as well as help avoid or defer the II. PEV LOAD MODEL
construction of new transmission lines and large power The proposed model for PEV charging load allocation is
plants. DG units can have a beneficial impact on power presented in this section. A detailed drivers and transportation
data are required to develop PEV charging load profiles. The Ni,g ,ch = NoHi PEV housePEVg%arr PEVch
%e
PEV %Pen i, g, ch (3)
National Household Travel Survey (NHTS) in 2009 is used in house
where NoHi is the number of houses at bus i, while PEV
this work. The data reflects daily trips over a 24-hour period,
is the average number of vehicles assumed per a house.
and were reportedly collected for different type of vehicles,
trips, purposes, seasons, weekdays, trip lengths. PEVg%arr is the percentage of vehicles arriving at hour g,
As shown in Fig.1, the proposed framework starts with the %e
PEVch is the percentage of vehicles that need ech, and
estimation of seasonal weekdays and weekend parameters PEV%pen is the percentage of PEV penetration.
such as daily mileage driven, charging availability times, and It is to be noted that all these parameters are calculated for
last arrival time to home probabilities, which will be then fed
weekday and weekend for each season (summer, fall, winter,
into PEV demand allocation optimization model with a spring).
predetermined percentage of PEV penetration and PEV
controlled (smart) charging to develop PEV charging load.
After that, PEV charging profiles will be used as an input to
the distribution system planning model to arrive to the
optimal upgrade plan.
G {V }
of PEVs (Ni,g,ch) arriving at hour g at bus i and requiring ech 1 2
J1 = + V j2, h 2Vi , h V j , h cos( j , h i , h ) (4)
energy. It is to be noted that the mileage driven has been 2
i, j i, h
i j h
divided into a number of discrete intervals, ch, and Ei,g,ch
denotes the energy required by a PEV, in range ch, given as where Gi,j is the conductance of line i-j, Vi,h is the voltage
follows: magnitude at bus i and hour h, and i,h is the voltage angle at
E i , g ,ch = N i , g ,ch * ech / i, g , ch (1) bus i and hour h.
The associated operational and PEV charging constraints
where is the charging efficiency while ech is the energy are discussed next.
required by the fleet in range ch and calculated as follows.
BC Power Flow Equations
ech = aem
M ch ch (2) The injected power at a bus is the power from the
M
substation, net of the load, and uncontrolled and smart PEV
where BC is the battery capacity, Maem is the total mileage of charging loads; and is governed by the traditional power flow
the vehicle in all electric mode and Mch is the mileage driven equations:
in range ch. The number of PEVs for each arriving group on a
bus is calculated as follows:
2016 IEEE Electrical Power and Energy Conference (EPEC)
PiSS
,h P
g ch
PEV
i ,h , g ,ch P
g ch
PEV _ SCh
i ,h , g ,ch Pd i ,h = line in (12) is the capital and operating cost of the candidate
DG units. The second line in (12) includes the engineering,
(5) procurement, and construction (EPC) cost and the variable
jN
Vi ,hV j ,h Yi , j cos( i , j + j ,h i ,h ) i N , h
component of the capital cost to upgrade the substation,
payment toward purchased power by the LDC. The third line
in (12) is the EPC cost and the variable component of the
QiSS
, h Qdi , h = capital cost to upgrade the feeders.
(6)
Vi,hV j ,hYi,h sin(i, j + j ,h i,h ) i N , h
C C
DG . F DG Cap DG .O
jN J2 = Pi + Pi ,DG
b Hrb
where PiSS SS
,h and Qi ,h are active and reactive power imported
i N b B
and Pi,PEV
_ SCh
by LDC via substation, Pi ,PEV are + C SS . F z SS + C SS .V S NSS + b Pi SS (12)
h , g ,ch h, g ,ch
i i , b Hrb
uncontrolled and smart PEV charging loads, Pd i,h and Qd i,h i SS b B
are active, reactive power demand, Yi,j is the magnitude of bus
admittance matrix element i-j, and i,j is the angle of the bus
+ (C Fdr . F
i , j N : ( i , j )
Ge i , j Le i , j z iFdr
,j +C
Fdr .V NFdr
S i, j )
admittance element i- j.
PEV Charging Constraints where b is the index for load block (bB), C DG . F and
Constraint (7) ensures that the total energy required by C DG .O are capital and operating cost of DG unit, $/MW and
uncontrolled charging PEVs be equal to the daily energy $/MWh, C SS .F , C SS .V are capital and operating cost of DG
needed to charge the battery. In addition, this constraint unit, $/MW and $/MWh, C Fdr .F , C Fdr .V are Fixed and
ensures that each PEV group ch arriving at hour g will start variable component of capital cost of feeder, $ and $/MVA,
DG
charging immediately, at hour h=g and the charging duration Pi Cap is the capacity of DG unit, Pi ,DG
b is power generated
is ChDch, to fulfill the daily required energy. Constraint (9)
from DG unit, Hrb is hours per year in load block b, ziSS and
ensures that the total energy required by smart charging PEVs
is equal to the daily energy needed to charge the battery. In ziFdr
, j is binary decision on substation and feeder
addition, this constraint ensures that the charging window of upgrade, S iNSS and S iNFdr are capacity added to substation and
,j
arriving groups at hour g is constrained by its arrival time to SS
home and 8 a.m. Constraints (8) and (10) ensure that the feeder, b is electricity market price, $/MWh, Pi ,b is active
power drawn by PEVs during any hour must be within the power imported by LDC via substation Gei , j is geographic
charging level for uncontrolled and smart charging PEVs cost factor of feeder between i and j, Lei , j is length of feeder
respectively. between i and j, km.
The associated operational and planning constraints are
P PEV
i , h , g , ch (
= 1 PEV % SCh
)N i , g , ch E i , g ,ch i N , g , ch (7)
discussed next.
h:( g + ChD ch h g )
Power Flow Equations
Pi,PEV
h, g ,ch The injected power at a bus is the power from the
(8)
(1 PEV )N %SCh
i , g ,ch P ChL i N , h, g : ( g + ChDch h g ) substation and DG units, net of the load (typical, PEV
uncontrolled and smart); and is governed by the traditional
P PEV _ SCh
i ,h, g ,ch = PEV % SCh N i , g ,ch Ei , g ,ch i N , g , ch (9)
power flow equations:
h:( h > g ) PiSS DG
,b + Pi ,b Pd i ,b =
Pi,PEV _ SCh
h, g ,ch PEV % SCh N i, g ,ch P ChL i N , h, g : ( h > g ), ch V
jN
i,bV j ,bYi , j cos( i , j + j ,b i ,b ) i N , b B
(13)
(10)
Min Max
V Vi,b V i N , b B (11)
Feeder Capacity Limits
Power flow through any distribution feeder must comply
III. DISTRIBUTION PLANNING MODEL with thermal capacity limit of the feeder. This limit also takes
The generic mathematical model for distribution system into consideration the new investments in feeder upgrades.
planning is presented below.
Pi,Fdr 2
j ,b = Vi,bYi , j cos i , j +
Objective Function (15)
Vi,bV j ,bYi, j cos( i, j + j ,b i,b ) (i, j ) N : (i, j ), b B
The proposed objective function (J2) aims to minimize the
annualized capital and operation cost of the LDC. The first
2016 IEEE Electrical Power and Energy Conference (EPEC)
QiFdr 2
P Pd
SS Cap DGCap
, j ,b = Vi ,bYi , j sin i , j Pi + Pi NSS + i (1 + ) i , peak (26)
(16)
Vi ,bV j ,bYi, j sin( i, j + j ,b i,b ) (i, j ) N : (i, j ), b B iN iN
Voltage Limits
Pi Fdr S i , j Cap + S iNFdr
Fdr cos Fdr (i , j ) N : (i , j ), b B
(17) This constraint ensures that the voltage magnitude at a bus
, j ,b ,j i , j ,b
is within the allowable limits.
QiFdr FdrCap + S NFdr sin Fdr (18) V Min Vi,b V Max i N , b B
, j ,b S i , j i, j i , j ,b (i, j ) N : (i, j ), b B (27)
SiNFdr
,j M ziFdr
,j (i, j ) N : (i, j ), b B
IV. CASE STUDIES AND RESULTS
(19)
A. Test System
where Pi ,Fdr and QiFdr
j ,b , j ,b are active and reactive power flow The system under study, as shown in Fig.2, comprises 33
from i to j,
FdrCap
and S i , j is existing feeder capacity. buses in radial configuration [6]. The main substation is at
bus-0 with a capacity of 4.6 MVA and the total system peak
In (19), M is a sufficiently large number often called the demand is 4.34 MVA at year-0 and assumed to grow at 3%
big M, which renders the constraint (19) nonbinding. When annually over the ten-year planning horizon. Details of the
ziFdr NFdr
, j = 0, S i , j is zero while when ziFdr
, j 0, the large value system, including various other parameter values used for the
of M allows sufficient room for selection of new feeder studies are taken from [7] and [8]. For the present study, only
capacity, and the value of M ensures that (19) is satisfied. gas-turbine DG units are considered. Assuming a fuel
consumption of 300 m3/h and a gas price of 0.14 $/liter, the
Substation Capacity Limits operating cost of the DG units is calculated to be 42 $/MWh.
These constraints ensure that the total power delivered by Electricity price or the price at which the LDC imports power
the substation transformer is within substation capacity limit. through substation are specified in terms of seventeen load
These limits take into consideration new investments in blocks using load scaling factors (LSF). The average Hourly
substation upgrades. Ontario Electricity Price (HOEP) for the year of 2013, is
assumed in this study as the price for the plan period.
(P ) + (Q )
2
SS 2 SS 2
S i + S iNSS
SSCap
i ,b i ,b i SS , b B (20) For PEVs, it is assumed that all PEVs are of type PHEV60
with an electric range of 60 km and 15.9 kWh battery
S iNSS ziSS M i SS (21) capacity. The charging level is assumed to be 3.8 kW (Level-
2) with 90% charging efficiency. The percentage of PEV
DG Capacity Limits used in a trip is assumed to be 60%. The number of houses at
The power generated by a DG unit is limited by the DG a bus is calculated assuming that the entire load is residential
capacity (22). The DG installed capacity is limited by the and the peak load of a typical house is 2.08 kW. Also, it is
maximum allowable DG size (23). Constraint (24) limits the assumed that there are two vehicles per house, and PEV
number of allowable DG units. Maximum DG penetration is charging occurs only at home.
limited by the minimum load plus 60% of maximum
substation rating in order to limit the maximum reverse power
flow over the transformer (25).
DGCap
PiDG
, b Pi i N , b B (22)
DGCap
Pi DG CapMax z iDG i N (23)
z
iN
DG
i DG NMax (24)
Pi ,DG
b (Pd i ,b + Pi ,PEV
b ) + 0 .6 Pi SS
,b b B (25)
iN iN
B. Case Studies
The main focus of this paper is to investigate the effect of
uncontrolled and smart PEV charging load on the plan. To
achieve this target, four case studies have been carried out. In
the first case (Base Case), no PEV charging loads have
considered while, in the remaining three cases, 50% of the
fleet is assumed to be PEVs with 100% smart charging in
Case-2, 50% smart charging and 50% uncontrolled charging
in Case-3, and 100% uncontrolled charging in Case-4.