integration of plug-in electric vehicles in renewable...
TRANSCRIPT
Integration of Plug-In Electric Vehicles in Renewable-Dominated Power
Systems Miguel Carrión
Escuela de Ingeniería Industrial, Toledo, Spain
University College Dublin, 2016
Motivation
It is expected a high increment of the number of PEVs
2
University College Dublin, 2016
Drawbacks of using Electrical Vehicles
Consumers acceptance (uncertainty)Reduced autonomySmall diversity of electrical vehicles (nowadays)Technological evolution (uncertainty)Need of charging point at homeReduced number of outdoor charging pointsHigh purchase cost
4
University College Dublin, 2016
Advantages of Using Electrical Vehicles
Increase of the energetic independenceReduction of the effects of the environmental
issue (reduction of contaminant emissions)Increase of the energetic efficiency (well to
wheel)Reduction of noise pollution
5
University College Dublin, 2016
Impact in Power Systems
A large increment in the number of electrical vehicles:Is it affordable by a power system?
What are the consequences? (electricity prices, CO2 emissions, generating mix,…)
6
University College Dublin, 2016
Plug-in Electric Vehicles (EVs) in power systems
Active research topic W. Kempton, S. E. Letendre, 1997 J. Tomic, W. Kempton, 2007 H. Lund and W. Kempton, 2008 C. K. Ekman, 2011 D. Dallinger and M. Wietshel, 2011 M. A. Ortega, F. Bouffard and V. Silva, 2013 K. Seddig, P. Jochem and W. Fitchner, 2014
7
University College Dublin, 2016
Objective
Analysis of the incorporation of a significant number of EVs in a power system with a large penetration of renewable energy
Controlled/Uncontrolled charge of EVs Stochastic dispatch model
8
University College Dublin, 2016
PEVs modeling
PEVs are equipped with a large battery that is charged from the grid
Definition of several groups of PEVs with similar behavior patterns
The time interval in which PEVs can be charged is defined for each group
Initial battery status of each group is estimated (daily distance driven, consumption per km, etc.)
9
University College Dublin, 2016
Charge modeling
Uncontrolled charge Charging process is not controlled by the ISO
Grid-to-vehicle control (G2V) Charging process is controlled by the ISO
Vehicle-to-grid control (V2G) Charging and discharging processes are controlled by
the ISO
10
University College Dublin, 2016
Stochastic dispatch modelStochastic dispatch model that co-optimizes
simultaneously energy and reserve (Pritchard et al., 2010, Morales et al., 2012, Domínguez et al., 2014)
Energy scheduling (Day-ahead schedule)Pool prices
Real-time operation (Real-time dispatch)Balancing prices
11
University College Dublin, 2016
Decision stages
Two-stage stochastic programming model First stage:
• Energy schedule • Uncontrolled charge of PEVs (additional inelastic load)
Second stage:• Reserve deployment (changes in the energy schedule)• Controlled charge/discharge of PEVs (decided by ISO)
o Charge: Down reserveo Discharge: Up reserve
12
University College Dublin, 2016
Model assumptions
Smart grid frameworkInelastic loadsLinear generating-side offersMinimum power outputs equal to zeroDC model of the transmission networkReserve capacity bids are not consideredEquipment failures are not considered
14
University College Dublin, 2016
Uncertainty
Uncertain parameters Demand Renewable resources (wind, PV) PEV demand
Modelled using a set of scenarios
15
University College Dublin, 2016
Optimization problem
Minimize Expected generation costs
Subject to:Technical constraints
Reserve deployment constraints
Power flow constraints
Power balance on day-ahead schedule
Power balance for deviations in the real-time dispatch
State-of-charge of PEVs
16
University College Dublin, 2016
Objective function
17
Day-ahead energy costs
Reserve deployment energy costs
Load shedding costs
University College Dublin, 2016
Technical constraints
18
Thermal units
Hydro units
Renewable units
University College Dublin, 2016
Reserve constraints
19
Thermal units
Hydro units
University College Dublin, 2016
Power balance for deviations in the real-time dispatch
22
University College Dublin, 2016
Input data
25
Transmission system based on the equivalent of the European System (Zhou and Bialek, 2005)
226 buses 61 conventional generating units 77 hydro units 90 wind units Solar PV distributed per node 390 transmission lines
University College Dublin, 2016
Input data
26
Technology Capacity (GW)
Nuclear 7.655
Coal 12.519
Gas 5.532
Combined Cycle 31.914
Hydro 21.077
Wind 23.456
PV 3.677
Total 105.8300 20 40 60 80 100
0
10
20
30
40
50
60
70
Power (GW)
Off
erin
g pr
ice
(€/M
Wh)
PVWindHydroNuclearCoalCCGTOCGT
University College Dublin, 2016
Input data
PEVs demand
Battery capacity: 22 kWh
Efficiency of charging/discharging: 0.88
Consumption: 0.18 kW/km
Charge rate: 7.2 kW (SAE J1772 low)
Number of vehicles: 30 millions
Percentage of PEVs: 30 %
27
University College Dublin, 2016
Input data
PEVs demand 3 groups (from mobility studies)
28
Type Beginning hour Ending hour Duration Percentage of PEVs
1 16:00 7:00 (next day) 16 hours 45 %
2 21:00 7:00 (next day) 11 hours 40 %
3 2:00 21:00 19 hours 15 %
University College Dublin, 2016
Input data Daily driven distance: N(35,9.6)
(G. Tal, M. A. Nicholas, J. Davies, J. Woodjack, 2014)
29
0 20 40 60 800
0.01
0.02
0.03
0.04
0.05PDF
Daily distance driven (km)0 20 40 60 80
0
0.2
0.4
0.6
0.8
1CDF
Daily distance driven (km)
University College Dublin, 2016
Input dataPEVs demand
30
0 5 10 15 20 25 300
1000
2000
Hours
Dem
and
(MW
h) NC-1
0 5 10 15 20 25 300
500
1000
1500
2000
Hours
Dem
and
(MW
h) NC-2
NC-1: Charging starts when PEVs are available to be charged
Demand peaks
NC-2: Charging is evenly distributed during all periods in which PEVs are available to be charged
Smooth demand
University College Dublin, 2016
Input dataUncertain parameters (Morales et al., 2009)
31
Initial set (600 scenarios) Final set (15 scenarios)
0 5 10 15 20 25 301520253035
Period
Dem
and
(GW
h)0 5 10 15 20 25 30
0
0.5
1
Period
PV
(p.
u.)
0 5 10 15 20 25 300
0.5
1
Period
Win
d (p
.u.)
0 5 10 15 20 25 301520253035
Period
Dem
and
(GW
h)
0 5 10 15 20 25 300
0.5
1
Period
PV
(p.
u.)
0 5 10 15 20 25 300
0.5
1
Period
Win
d (p
.u.)
University College Dublin, 2016
Solution
Computational size Constraints: 1.8 millions Continuous variables: 1.6 millions
CPLEX 12.2.0.1 Server with four 2.9 GHz and 250 GB of RAM
Solution time Between 1.3 and 4.5 hours
32
University College Dublin, 2016
Increment of 7.6% in the total demand (66.4 GWh)
The expected cost increases between 16.2 and 20.5%
The coordination G2V reduces the cost between 3.4 and 2.0%
V2G does not outperform G2V significantly (0.1%)
Expected generation costs
33
Non-PEV NC-1 NC-2 G2V V2G
Expected cost (M€)
11.480 13.830 13.638 13.359 13.342
University College Dublin, 2016
Expected demand profile
34
0 10 20 300
20
40 NC-1E
nerg
y (G
Wh)
Hours0 10 20 30
0
20
40 NC-2
Hours
Ene
rgy
(GW
h)
0 10 20 300
20
40 G2V
Hours
Ene
rgy
(GW
h)
0 10 20 300
20
40 V2G
Hours
Ene
rgy
(GW
h)
University College Dublin, 2016
Day-ahead schedule
35
0 10 20 300
10
20
30
40
Hours
Ene
rgy
(GW
h)
NVE
PV Wind Hydro CCGT+OCGT Coal Nuclear
0 10 20 300
10
20
30
40
NC-1
Hours
Ene
rgy
(GW
h)
0 10 20 300
10
20
30
40
G2V
Hours
Ene
rgy
(GW
h)
University College Dublin, 2016
Reserve deployment
36
0 10 20 30-5
0
5 NC-1
Hours
Dep
loye
d re
serv
e (G
W)
0 10 20 30-5
0
5 NC-2
Hours
Dep
loye
d re
serv
e (G
W)
0 10 20 30-5
0
5 G2V
Hours
Dep
loye
d re
serv
e (G
W)
0 10 20 30-5
0
5 V2G
Hours
Dep
loye
d re
serv
e (G
W)
University College Dublin, 2016
Wind spillage
37
0 10 20 300
1000
2000
3000
NC-1
Hours
Pow
er (
MW
)
0 10 20 300
1000
2000
3000
NC-2
Hours
Pow
er (
MW
)
0 10 20 300
1000
2000
3000
G2V
Hours
Pow
er (
MW
)
0 10 20 300
1000
2000
3000
V2G
Hours
Pow
er (
MW
)
University College Dublin, 2016
Day-ahead prices
38
0 5 10 15 20 25 3010
15
20
25
30
35
40
45
Hours
Pri
ce (E
uro/
MW
h)
NC-1NC-2G2VV2G
University College Dublin, 2016
Real-time prices
39
0 10 20 30
-20
0
20
40
NC-1
Hours
Rea
l Tim
e P
rice
(€/
MW
h)
0 10 20 30
-20
0
20
40
NC-2
Hours
Rea
l Tim
e P
rice
(€/
MW
h)
0 10 20 30
-20
0
20
40
G2V
Hours
Rea
l Tim
e P
rice
(€/
MW
h)
0 10 20 30
-20
0
20
40
V2G
Hours
Rea
l Tim
e P
rice
(€/
MW
h)
University College Dublin, 2016
Consumers payments
40
Non-PEV NC-1 NC-2 G2V V2G
Total 27.202 31.613 31.150 31.095 31.123
PEVs 0.000 2.539 2.255 2.214 2.221
Non PEVs 27.202 29.074 28.894 28.881 28.902
C Expected payment (M€)
University College Dublin, 2016
Sensitivity analysis
41
Peak power charge rate(G2V versus V2G)
Renewable capacity(G2V versus NC-2)
100 200 300 400 500-0.24
-0.22
-0.2
-0.18
-0.16
-0.14
-0.12
-0.1
-0.08
Charge rate (%)Cos
t var
iati
on o
f G2V
wit
h re
spec
t to
V2G
(%
)
0 50 100 150 200-5
-4
-3
-2
-1
0
Renewable capacity (%)Cos
t var
iati
on o
f G
2V w
ith
resp
ect t
o N
C-2
(%
)
University College Dublin, 2016
Sensitivity analysis
42
0 50 100 150 200-2.5
-2
-1.5
-1
Hydro offer price (%)
Cos
t var
iati
on w
ith
resp
ect t
o N
C-2
(%
)Hydro offer cost
(G2V versus NC-2)
University College Dublin, 2016
Number of scenarios
43
0 5 10 15
11.2
11.3
11.4
11.5
Number of scenarios
Exp
ecte
d co
st (M
€)Non PEVs
0 5 10 15
13.4
13.5
13.6
13.7
Number of scenarios
Exp
ecte
d co
st (M
€)
NC-2
0 5 10 15
13.2
13.3
13.4
Number of scenarios
Exp
ecte
d co
st (M
€)
G2V
University College Dublin, 2016
Conclusions For the studied day, the coordination between the Independent System
Operator and the PEVs: Reduces the expected costs Reduces the wind spillage and the usage of deployed reserves Reduces the difference between peak and valley day-ahead prices Reduces the volatility of real-time prices Is appropriated for those power systems with a significant amount of
renewable resources
The coordination V2G does not outperform G2V significantly
Solution times are high if coordination is taken into account
44
University College Dublin, 2016
Input dataUncertainty
47
0 5 10 15 20 25 301520253035
Period
Dem
and
(GW
h)
0 5 10 15 20 25 300
0.5
1
Period
PV
(p.
u.)
0 5 10 15 20 25 300
0.5
1
Period
Win
d (p
.u.)
Initial set: 600 (3 x 200) scenarios
Final set: 15 scenarios
(Morales et al., 2009)
University College Dublin, 2016
Input dataInitial set of 600 (3 x 200) scenarios
Scenario reduction (Morales et al., 2009)
48
12.5 13 13.5 14 14.5 15 15.5 160
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Pro
babi
lity
Cost (M€)
15 scen.600 scen.
University College Dublin, 2016
Example for one scenario
49
0 5 10 15 20 25 300
1000
2000
3000
4000
Up reserve
Hours
Ene
rgy
(MW
h)
0 5 10 15 20 25 30
-8000
-6000
-4000
-2000
0
Hours
Ene
rgy
(MW
h)
Down Reserve
V2G/G2V Hydro CCGT+OCGT Coal Nuclear
0 5 10 15 20 25 305000
0
5000
Hours
Dem
and
(MW
)
0 5 10 15 20 25 30500
0
500
Hours
PV
(M
W)
0 5 10 15 20 25 301
0
1x 10
4
Hours
Win
d (
MW
)
0 5 10 15 20 25 30
-20
0
20
40
NC2
Hours
Rea
l Tim
e P
rice
(€/
MW
h)
0 5 10 15 20 25 30
-20
0
20
40
V2G
Hours
Rea
l Tim
e P
rice
(€/
MW
h)
0 5 10 15 20 25 300
1000
2000
3000
4000
Up reserve
Hours
Ene
rgy
(MW
h)
0 5 10 15 20 25 30
-5000
-4000
-3000
-2000
-1000
0
Hours
Ene
rgy
(MW
h)
Down Reserve
V2G/G2V Hydro CCGT+OCGT Coal Nuclear
V2G
V2G
NC-2
NC-2