alexandre oudalov, abb switzerland ltd., 10th...
TRANSCRIPT
Microgrid Storage IntegrationBattery modeling and advanced control
Alexandre Oudalov, ABB Switzerland Ltd., 10th Microgrid Symposium, Beijing, November 13-14, 2014
Microgrid Storage IntegrationOutline
Introduction
Energy storage technologies and applications
LIB aging model
Semi-empirical model overview
Advanced BESS control strategies
Application and SoC control
Conclusions
LIB Lithium Ion Battery
BESS Battery Energy Storage System
SOC State of Charge
Energy Storage Applications
ESS
Integration of
renewables
1-100 MW,1-10hPeak shaving
0.5-10 MW, 1h
220 kV
110 kV
20 kV ring
20 kV
Conventional
Central
Generation
Variable
Renewable
Generation
220 KV
Load leveling
for generation utilization
10-1000 MW, 1-8h
ESS
110 kV Industry/
Large Commercial
Load Center
20 kV
ESS
Spinning reserve
In case of line loss
10-500 MW, 0.25-1 h
Load leveling
for postponement of grid upgrade
1-10 MW, 1-6h
ESS
ESS
Frequency Regulation
1-50 MW, 0.25-1h
220 kV
110 kV
ESS
Solar PV time shift
1-100 kW, 2-6h
ESS
0.4 kV
Residential/Small Commercial/Nano-gridsESS
Micro-grid
0.1-5 MW, 5 min (stabilization)
0.1-50 MW, 1-10h (self-
consumption of local
renewables)
Microgrid Storage IntegrationEnergy vs Power Applications in Microgrids
Grid ON Grid OFFWeak grid ON
Energ
y
P
ow
er
Energy applications Power applications
Energy Time Shift
Capacity firming
Ramp rate ctrl Ramp rate ctrl
Spinning reserve
Energy Time Shift
Line load levelling
Frequency ctrl
Spinning reserve
Energy Time Shift
Microgrid Storage IntegrationKey Technologies
Batteries Hydrogen
Super-
conducting
(SMES)
Capacitors
Supercaps
Pumped
hydro
Compr. Air
(CAES)
Flywheel Vanadium
ZnBr
PSBr
Electro-
lyser &
Fuel Cell
Lead acid
NiCd
NaS
Lithium
Ni-MH
Flow cells
Adiabatic
CAES
Pressure
Pressure
Heat
Gravitation
Kinetic Electric
Magnetic
Mechanical Thermo-
dynamic
Electro-
magnetic
Electrochemical
Thermo
electric
Heat
Projected Cost Reductions of LIB ($/kWh)
0
200
400
600
800
1000
1200
1400
1600
2010 2015 2020 2025 2030
Bloomberg New Energy Finance Navigant Research Deutsche Bank IHS (iSuppli)
*) final prices include margins most probably in the range of 30-40%
Battery Aging ModelOverview
Battery capacity fading is a limiting
factor for BESS performance
Customers usually expect a certain
battery life in years or number of
cycles for a given application
Battery manufacturers usually over-
dimension the battery to reduce the
risk of earlier system depletion
We recommend to include battery
aging models in the design of and
operation strategies for BESS
Source:batteryuniversity.com
Source: skywriting-net
Battery Aging ModelLIB Model
We propose a semi-empirical model
Capacity fades due to battery
cycling and time elapsed
𝑙𝑟𝑒𝑚𝑎𝑖𝑛= 𝑝𝑆𝐸𝐼𝑒
−𝑟𝑆𝐸𝐼𝑙𝑓𝑎𝑑𝑒𝑑 + 1 − 𝑝𝑆𝐸𝐼 𝑒−𝑙𝑓𝑎𝑑𝑒𝑑
𝑙𝑓𝑎𝑑𝑒𝑑 = 𝑙𝑐𝑦𝑐𝑙𝑖𝑛𝑔 + 𝑙𝑐𝑎𝑙𝑒𝑛𝑑𝑎𝑟
Cycling aging is fully dependent on
battery usage and is modeled as a
sum of aging during each cycle
Calendar aging is independent of
battery usage and is modeled as a
linear function of time, average SoC
and temperature
𝑙𝑐𝑦𝑐𝑙𝑖𝑛𝑔
=
𝑖=1
𝑁
𝜎𝐷𝑜𝐷 𝐷𝑜𝐷𝒊 𝜎𝑆𝑜𝐶 𝑆𝑜𝐶𝒊 𝜎𝐼(𝐼𝒊) 𝜎𝑇(𝑇𝒊)
𝑙𝑐𝑎𝑙𝑒𝑛𝑑𝑎𝑟
= 𝑘𝑐𝑎𝑙𝑒𝑛𝑑𝑎𝑟 𝜎𝑆𝑜𝐶
𝑖=1
𝑁𝑆𝑜𝐶𝑖𝑁𝜎𝑇
𝑖=1
𝑁𝑇𝑖𝑁𝑖
𝑡
Battery Aging ModelLIB Test Data vs. Model Reconstruction
© ABB Group
December 9, 2014 | Slide 10
Grid
Su
b-S
tatio
nC
on
ve
rte
rS
tora
ge
Converter
control
System
level
control
Switching
commands
V, I, T
on/off
P, Q
status
Secondary side
V, I, f, P, Q
P, Q, status,
application
SoC, SoH, Ramp limit, faultson/off
BMS
HVAC control
speed
Status
Status, availability (SoC)
HVAC control
on/off
on/off
T
Modbus TCP/RTU
Primary side Functionality
Local applications
and ESS
protection
Charge and health
monitoring
Protection
Switching of power
electronics
BESSGeneral system architecture
Other
controllers
Microgrid Plus SystemEfficient and reliable power flow management
PV plant
Diesel Generator
Grid Stabilising
System
Distribution feeder
Power
Micro-Grid
Wind Turbines
MGC600-W M+ Operations
Local and remoteMGC600-P
MGC600-G
MGC600-E
MGC600-F
Communication
Network
Distribution feederMGC600-F
Grid connection
MGC600-N
Residential
Industrial & commercial
Advanced Battery Control StrategiesCombination of application and SoC control
Due to a battery inefficiency and in some
applications due to a none zero mean control
signal, the BESS can be totally discharged or
charged in a short time interval
It limits a use of BESS until its SoC will be
back into an acceptable range
An optimal BESS control strategy must cover
both:
Application control
State of charge control
Consideration of a battery aging model can
provide additional information in order to take
pro-active measures to fulfill the lifetime
targets
BESS PCS
Application
control
State of charge
control
SetpointsActual state
Constraints
& setpoints
Setpoints
0 2000 4000 6000 8000 10000
-1
-0.5
0
0.5
1
x 105
Time (seconds)
Pow
er
(W)
0 2000 4000 6000 8000 10000
0.45
0.5
0.55
0.6
0.65
SoC
(-)
P offset
SoC
SoCmin
SoCmax
0 2000 4000 6000 8000 10000-0.1
-0.05
0
0.05
0.1
Time (seconds)
Fre
qu
ency
De
via
tion
(H
z)
modified frequency deviation
Advanced Battery Control StrategiesFrequency Control in Microgrids
State of charge (SoC) control according to one of the
following strategies
Strategy 1:
Active when SoC exceeds adjustable thresholds
and frequency is within a regulation dead-band
Off-set value is 0-100% *𝑃𝑛𝑜𝑚 at any time step
(preferably small values)
Strategy 2:
Is continuously activated and adjusts a power
set-point using an average over the previous
usage, i.e. ctrl signal is zero-mean
Variable off-set value is taken from a secondary
reserve
Strategy 3:
Active when SoC exceeds adjustable thresholds
Fixed off-set value is taken via a ramp of fixed
slope for a fixed duration from a secondary
reserve or an intraday market
0 2000 4000 6000 8000 10000
-1
-0.5
0
0.5
1
x 105
Time (seconds)
Pow
er
(W)
0 2000 4000 6000 8000 10000
0.45
0.5
0.55
0.6
0.65
SoC
(-)
P offset
SoC
SoCmin
SoCmax
0 2000 4000 6000 8000 10000-0.1
-0.05
0
0.05
0.1
Time (seconds)
Fre
qu
ency
De
via
tion
(H
z)
modified frequency deviation
0 2000 4000 6000 8000 10000
-1
-0.5
0
0.5
x 105
Time (seconds)
Pow
er
(W)
0 2000 4000 6000 8000 100000.4
0.5
0.6
SoC
(-)
0 2000 4000 6000 8000 10000-0.1
-0.05
0
0.05
0.1
Time (seconds)
Fre
qu
ency
De
via
tion
(H
z)
modified frequency deviation
P offset
SoC
0 2000 4000 6000 8000 10000
-1
-0.5
0
0.5
1
x 105
Time (seconds)
Pow
er
(W)
0 2000 4000 6000 8000 100000.45
0.5
0.55
0.6
0.65
SoC
(-)
P offset
SoC
SoCmin
SoCmax
0 2000 4000 6000 8000 10000-0.1
-0.05
0
0.05
0.1
Time (seconds)
Fre
qu
ency
De
via
tion
(H
z)
modified frequency deviation
SoC control strategy 1Offset and SoC variations
• For Poffset = 5% of rated power, the annual capacity fading ≈ 6%
• For Poffset = 10% of rated power, the annual capacity fading ≈ 3.7%
0 2000 4000 6000 8000 10000-3
-2
-1
0
1
2
3
4x 10
5
Time (seconds)
Pow
er
(W)
P AS
P BESS
P Offset
SoC control strategy 1Variations of different power signals
• Offset is active when system frequency is within a deadband
• BESS follows exactly the requested ancillary service power when Δf≠0
Charge
Discharge
0 2000 4000 6000 8000 10000
-1
-0.5
0
0.5
x 105
Time (seconds)
Pow
er
(W)
0 2000 4000 6000 8000 100000.4
0.5
0.6
SoC
(-)
0 2000 4000 6000 8000 10000-0.1
-0.05
0
0.05
0.1
Time (seconds)
Fre
qu
ency
De
via
tion
(H
z)
modified frequency deviation
P offset
SoC
SoC control strategy 2Offset and SoC variations
• For an averaging period of 1 hour and a dispatch delay of 15 min,
the annual capacity fading ≈ 3.2%
0 2000 4000 6000 8000 10000 12000-4
-3
-2
-1
0
1
2
3
4x 10
5
Time (seconds)
Pow
er
(W)
P AS
P BESS
P Offset
SoC control strategy 2Variations of different power signals
• Offset mechanism ‘forces’ the BESS power signal to be zero-mean
• Deviations between PAS and PBESS
Charge
Discharge
0 2000 4000 6000 8000 10000
-1
-0.5
0
0.5
1
x 105
Time (seconds)
Pow
er
(W)
0 2000 4000 6000 8000 100000.45
0.5
0.55
0.6
0.65
SoC
(-)
P offset
SoC
SoCmin
SoCmax
0 2000 4000 6000 8000 10000-0.1
-0.05
0
0.05
0.1
Time (seconds)
Fre
qu
ency
De
via
tion
(H
z)
modified frequency deviation
SoC control strategy 3Offset and SoC variations
• For SoC upper threshold = 53%, SoC lower threshold = 47%, off-set level = 10% of rated power,
ramp up/down in 5 min and min offset duration = 15 min, the annual capacity fading ≈ 3%
0 2000 4000 6000 8000 10000-4
-3
-2
-1
0
1
2
3
4x 10
5
Time (seconds)
Pow
er
(W)
P AS
P BESS
P Offset
SoC control strategy 3Variations of different power signal
• Offset active when SoC hits the thresholds
• Deviations between PAS and PBESS
Charge
Discharge
Comparison of three SoC control strategies
0 2000 4000 6000 8000 100000.44
0.46
0.48
0.5
0.52
0.54
0.56
0.58
0.6
0.62
0.64
Time (sec)
SoC
(-)
SoC evolution according to the three strategies
SoC 1
SoC 2
SoC 3
Offset 1 depends on Δf → frequent threshold violations with limited control
Offset 2 depends on past values of PAS → forces BESS to reach SoCnom
Offset 3 is activated once thresholds are hit → less sensitive to SoC variation
Reduced peak
for strategy 3
Offset 3 is activated
once thresholds are hit
Less degradation
for strategy 3
Continuous adjustment of
SoC for strategy 2
(even between thresholds)
Preferred SoC Control StrategiesNeed for an adaptive approach
Depending on the operation mode grid-tied or off-grid we switch between
strategies
Based on the actual generation mix we can tune the parameters of each
strategy
Operation
modeReference Advantages
Off-grid Strategy 1• The offset does not directly cancel
the control signal
Grid-tied Strategy 3
• Less energy is cycled through
the offset than in strategy 2
• Less battery capacity fading
Conclusions
Battery based energy storage plays an important role in
microgrids with a large amount of RES
A battery model allows to quantify capacity fading and to
take corrective measures in case of deviations from the
initially planned lifetime trajectory
There are several strategies to control SoC and a preferred
strategy depends on:
status of the microgrid (grid-tied vs isolated)
available options for the off-setting part (available
generation mix, accessibility to power markets, etc.)
Availability of forecast information (RES, load, scheduled
islanding operation, etc.) can help to predict future SoC
and parameterize the control system accordingly
Battery Aging ModelStress Factors
0 10 20 3040 50 6070 80 901000
2
4
6
DoD[%]
f DoD(
10
-5)
DoD Stress Model
0 10 20 3040 50 6070 80 901000.5
1
1.5
2
SoC[%]
f SoC
SoC Stress Model
0 1 2 3 40.5
1
1.5
2
2.5
C-rate
f C
C-rate Stress Model
15 20 25 30 35 40 45 50 55
1
3
5
7
T[C]
f T
Temperature Stress Model
0 0.5 1 1.5 2 2.5 3
x 106
0
10
20
30
40
50
60
70
80
90
100SoC evolution - Model 1
Time (seconds)
SoC
(%
)
SoC evolution
SoC limit violations
SoC limits
SoC Control StrategiesModel 1 – SoC evolution for one month
December 9, 2014 | Slide 25
© ABB
BESS degradation after 1 year: = 3.67 %
𝑃𝑜𝑓𝑓𝑠𝑒𝑡 = 10% 𝑃𝑛𝑜𝑚
0 0.5 1 1.5 2 2.5 3
x 106
0
10
20
30
40
50
60
70
80
90
100SoC evolution - Model 2
Time (seconds)
SoC
(%
)
SoC evolution
SoC limit violations
SoC limits
SoC Control StrategiesModel 2 – SoC evolution for one month
December 9, 2014 | Slide 26
© ABB
BESS degradation after 1 year: = 3.22 %
Averaging period:
1 hour
Dispatch delay:
15 min
0 0.5 1 1.5 2 2.5 3
x 106
0
10
20
30
40
50
60
70
80
90
100SoC evolution - Model 3
Time (seconds)
SoC
(%
)
SoC evolution
SoC limit violations
SoC limits
SoC Control StrategiesModel 3 – SoC evolution for one month
BESS degradation after 1 year: = 3.02 %
SoC upper threshold:
53 %
SoC lower threshold:
47 %
Ramp duration:
5 min
Min offset time:
15 min
References – Ancillary power system servicesSP AusNet Grid Energy Storage System
Project name
SP AusNet GESSCountry
Victoria, AustraliaCustomer
SP AusNetCompletion date
Due to be completed in 2014
ABB solution
Design, engineering, installation and testing of PowerStore-Battery, transformer and diesel generator
Microgrid Plus System for overall system management
Based on transportable containerized solution
Customer benefits
Active and reactive power support during high demand periods
Transition into isolated/Off-grid operation on command or in emergency cases without supply interruption
Delay of power line investments
First grid-tied microgrid with Battery Energy Storage for
distribution network support in Australia
About the project