result management system report for college project
Peak shaving of an EV Aggregator Using Quadratic Programming
1. Peak shaving of an EV Aggregator
Using Quadratic Programming
Mingyu Seo
mgseo@knu.ac.kr
1
2. contents
2
Agenda
Introduction
Research background
V2X project in korea
Project facilities
Systems
Frame work
Sequence
Flow chart
Algorithms
Frame work
Structure
Case study
Simulation
Result
KPI
Conclusion
3. Introduction: Research Background
• The number of EV is
increasing rapidly every year.
• What will happen?
3
year Accumulated
Source: Ministry of Environment, Korea
Source: US Energy Information Administration
TSO
- cost reduction
DSO
- Install additional facilities
4. Introduction: V2X project in korea
• Why Korea Gov(Daegu) do this project?
• Why in daegu?
– Daegu is filled with electric car chargers and commercial vehicle manufacturers.
– The above conditions are good requirements for pilot projects.
Build smart city & Develop technology
(maximize the EV usage not for only drive)
ex) V2B,V2G...
Reduction Co2, electric cost
7. Introduction: I-SMART data(KEPCO)
7
2013 2014 2015 2016 2017 2018 2019
KW 186.72 332.16 357.6 378.72 299.76 360.72 372.72
186.72
332.16
357.6
378.72
299.76
360.72 372.72
KW
yearly peak trend
2019.
04
2019.
03
2019.
02
2019.
01
2018.
12
2018.
11
2018.
10
2018.
09
2018.
08
2018.
07
2018.
06
2018.
05
KW 299.3 300.5 372.7 313.9 350.4 316.1 315.1 339.1 360.7 280.6 308.6 329
299.28300.48
372.72
313.92
350.4
316.08315.12
339.12
360.72
280.56
308.64
329.04
KW
Monthly peak trend
• I-SMART contains 2 company (KIAPI, KARTECH)
– Contract power = 1750 kW (30% is 525kw)
– In data, maximum peak is around 370 kw => Peak load control(PLC) is no meaning
• Install meter for main building
• Constitution of virtual billing system based on electricity rate
8. V2X Project in Korea
• whole structure(contains installed meter and )
8
M2
(Our meter) EVSE1
ESS1
PV1
EVSE2
ESS2
PV2
PCS2
M1
(KEPCO ISMART)
EV1
Other load (test equipment, etc.)
M0~8:
Measurement
point
Target
building
PCS1
Virtual charge application office load
24kwh
50kw
50kw
50kw
250kwh
250kwh
50kw
9. Systems: Flow chart
9
• Load and EV Prediction
- How much is the peak tomorrow?
- When will the peak comes tomorrow?
- When EV will come to the office?
• EV scheduling
- What is the best schedule of EVs?
• Evaluation & Analysis
- How much we shaved the peak?
10. 10
TEMS.jar/dll Logs
CSV files
Flags / Operation Codes
gridOS
demandModelDev.dll
EVModelDev.dll
PVModelDev.dll
Model Dev
demandForecast.dll
EVForecast.dll
PVForecast.dll
Model
Forecast
objV2X.dll
Model
Optimize
HMI
HMI side
Algorithm side
Input
Data CSV,
OP codes
Output
Result CSV,
Flags
ⓐ HMI
ⓑ Input/Output
Interface
ⓒ Shell java scriptⓓ API
ⓔ Logs
Model Development
- LTData.csv
- LTV.csv
- LTP.csv
Forecast and Optimize
- STD.csv
- forecastDemandData.csv
- STV.csv
- forecastEVData.csv
- STP.csv
- forecastPVData.csv
- EV_Config_xxx.csv
Operation
- EV_Config.csv
- EV_Results.csv
- PV_Results.csv
- Demand_Results.csv
System: Frame work
Result files
Forecast result
- Demand forecast result.csv
- PV forecast result.csv
- EV forecast result.csv
Scheduling result
- VFR
- % Peak reduction
- % Cost reduction
15. Algorithms: Frame work
15
EV-EMS
EV Config.
Tariffs
Base Load
Renewable energy
Demand load
EV
PLC
TOU
PLC+TOU
Operation type Dispatch control
Operation Summery
SOC
Power
EV Schedule
16. Algorithms: structure
• Objective function
– PLC+TOU : arg min 𝑜𝑢𝑡=1
24
𝑐𝑖=𝑖𝑛
𝑜𝑢𝑡
𝐷 + 𝑃 2
+ 𝑋ℎ × 𝑇𝑎𝑟𝑖𝑓𝑓
• Constraints
– 𝑆𝑂𝐶 𝑚𝑖𝑛 ≤ 𝑆𝑂𝐶 ≤ 𝑆𝑂𝐶 𝑚𝑎𝑥
– −10 ≤ 𝑃𝑡 ≤ 20
– EEV
max
∗ 0.8 ≤ 𝐸 𝐸𝑉
𝑓𝑖𝑎𝑛𝑙
16
Charge/discharge
Power [kW]
𝑃0,2
2
Estimated
Plug in time
EV control
time
Estimated Plug
out time
Total charge / discharge required at
that time [kW]
where D : demand [kw]
P : EV charge/discharge [kw]
𝐸 𝐸𝑉 : EV SOC[kwh]
22. Algorithms: KPI
(Key Performance Indicator)
22
TEMS 224,204 -8,7704 -4%
MACH
Energy
239,021 -4,770 -2%
Ice Energy 227,302 -5,962 -2.5%
Manual
Schedule
230,589 -2,385 -1%*
No
Schedule
232,974 0 0
Total Cost Reduced Cost %Reduced
23. Conclusion: future work
• Simulation results show that the algorithm applied to the Korean V2X
project shows a cost reduction of about 60%.
• The algorithms applied to the current V2X project are deterministic.
=> The algorithm will be upgraded considering the probabilistic concept.
• 3 charging modes are under development.
– Cost saving mode for grid
– User experience mode
– Intermediate mode
23
User
Experiences
Cost Saving