Professional Documents
Culture Documents
2014 BESS Control Strategies For Participating in Grid Frequency Regulation
2014 BESS Control Strategies For Participating in Grid Frequency Regulation
Abstract: Battery Energy Storage Systems (BESS) are very effective means of supporting
system frequency by providing fast response to power imbalances in the grid. However, BESS are
costly, and careful system design and operation strategies are needed in order to generate revenue
for the system owner. We propose control strategies which will help to maintain BESS’s State
of Charge (SoC) in the optimal range and slow down battery aging significantly. A validation of
these strategies using data from ENTSO-E (for the German regulation market) in Continental
Europe and the PJM interconnection in the USA is presented in the results section.
80
EU PFC Power
Increment
PJM RegD P_bs
step
60
Occurance (%)
Minimum
P_bm
power
40
time
t_ms t_md t_bm
Last market Market Delivery Delivery
20 clearance clearance start end
0
−1 −0.8 −0.6 −0.4 −0.2 0 0.2 0.4 0.6 0.8 1
Fig. 2. Generalized intraday market regulations.
Regulation power (P )
n
3. SOC CONTROL STRATEGIES
(a) Regulation power
100 3.1 Intraday Bidding Control
EU PFC
PJM RegD
80
To maintain BESS’s SoC and ensure the regulation per-
Occurance (%)
40
determined according to the BESS’s condition. Energy
20
is purchased from the intraday market to charge up the
battery if SoC is low, or sold if SoC is high. Upon power
0
−0.1 −0.08 −0.06 −0.04 −0.02 0 0.02 0.04 0.06 0.08 0.1
delivery, the operation point of BESS is set as
Deviation events energy throughput (Pnh)
4025
19th IFAC World Congress
Cape Town, South Africa. August 24-29, 2014
floored (or capped if the result is negative) to the nearest method averages battery’s net energy consumption over
allowed bidding power by the regulation market. an averaging period (a) and set the working point (PW P )
of the battery accordingly after a certain delay (d). For
3.2 GM(1,1) SoC Predictor time step k, this mechanism can be explained as
4026
19th IFAC World Congress
Cape Town, South Africa. August 24-29, 2014
DoD Stress Model SoC Stress Model efficiency (Voc = Vnom , Ro = 0), the simulation model can
6 2 be simplified as
fδ(×10−5)
4 1.5
Pext ts
fσ
2 1 σ(k + 1) = σ(k) − , (19)
EBESS
0 0.5
0 10 20 30 40 50 60 70 80 90100 0 10 20 30 40 50 60 70 80 90100
DoD[%] SoC[%]
where EBESS is BESS’s energy capacity.
Temperature Stress Model Time Stress Model
0.2
9 4.3 Pricing Setup
7
kcalt
5 0.1
fT
Lam et al. [2011] are used in this work. When ignoring cell operators (www.regelleistung.net)
4027
19th IFAC World Congress
Cape Town, South Africa. August 24-29, 2014
0.2 500
Regulation energy throughput(Pn*h)
400
300 0.8
Pmarket(kW)
100
0.4
0 0
−100 0.2
−200
−0.1 0
−300 60
Pmarket SoC SoCforecast 45 60
−400 30 45
30
−0.2 −500 15 15
5 5
0 1 2 3 4 5 6 7 8 9 10 11 12 Contract duration (minutes) Market delay (minutes)
Hours
Fig. 4. Example of regulation energy throughput simula- Fig. 6. BESS energy capacity with various intraday market
tion with market SoC control and GM(1,1) predictor. regulations in PFC.
100 100
80 80
60 60
SoC(%)
SoC(%)
40 40
20 20
0 0
0 5 10 15 20 25 30 0 4 8 12 16 20 24 28 32
Days Days
Fig. 5. SoC in PFC simulation with battery efficiency Fig. 7. SoC in PJM simulation with battery efficiency
(0.6Pn h size, profile: 2011/11/01 - 2011/11/30). (0.6Pn h size, profile: 2012/12/18 - 2013/01/18).
Fig. 4, in which variation of regulation energy, energy 5.2 PJM Interconnection Case
prediction and the intraday market power are included.
The minimal energy size of the battery obtained from the In the PJM RegD case, simulations are performed with the
yearly simulation is 0.45 Pn h, corresponding to 0.45 MWh battery efficiency model and constant BESS temperature
with 1 MW PFC reserve. (25◦ C). The moving average delay d is fixed at 15 minutes,
The energy throughput simulation is also performed with meaning compensating operations are made at secondary
different market regulations. It is assumed in the sim- control timescale. The averaging period a includes 5, 15,
ulations that the duration of power contracts is always 60, or 120 minutes, and for each value of a a simulation is
equal to the period of market clearance (tms ≡ tbm ). The performed with an appropriate BESS size. The regulation
round-trip battery efficiency is η 2 = 0.9, and SoC level capability is set to 1 MW. The RegD profile from Dec
is controlled between 10% to 90%. The resulting BESS 18, 2012 to Jan 18, 2013 is used, with a total duration of
sizes with respect to different market delays and contract 32 days, and the average signal mileage is 16.9. The SoC
durations is shown in Fig. 6. The fact that the BESS size result of the simulation with 0.6 MWh/MW BESS size (a
increases linearly with larger contract duration is due to = 60 minutes) is shown in Fig. 7.
that BESS is not able to quickly adjust its operation point At each simulation, the linearized capacity degradation is
and thus has to sustain longer deviations. The BESS size estimated. The performance score and the payment credit
also increases with market power delay, which is mainly are calculated hourly and their average is shown in the
due to the larger errors in the SoC prediction, while in following table as well as other parameters.
the simulation with ideal forecast the BESS size results
showed no dependences on the market delay. Table 1. PJM RegD Simulation Results (20121218-20130118)
Simulation with battery efficiency model is performed for Case 1 Case 2 Case 3 Case 4
providing 1MW reserve with the minimal battery size of
BESS Size
0.6 MWh. During the simulation, 261.6 MWh energy is
(MWh/MW) 0.20 0.33 0.60 1.00
discharged from the BESS, while 264.9 MWh is charged, Avg. period
and 3.3 MWh of heat is generated by the BESS. 144.3 (minutes) 5 15 60 120
MWh energy is purchased from the intraday market, Performance
and 136.0 MWh is sold, giving a net energy flow of 8.6 Score 0.9 0.96 0.99 0.99
MWh. All power outputs are below the 1 MW system Capacity
rating throughout all simulations. The resulting linearized Degradation(%) 2.77 1.66 1.14 0.84
capacity degradation is 3.43%, corresponding to a loss of Heating
21 kWh in battery cells. The SoC simulation result of Energy(MWh) 16.81 8.07 3.37 2.50
November, in which the largest SoC deviation happened, Average
is shown in Fig. 5. Credit($/h) 113 120 123 123
4028
19th IFAC World Congress
Cape Town, South Africa. August 24-29, 2014
8
ENTSO−E PFC revenue implies that to introduce BESS into existing ancillary
PJM RegD revenue
6 service frameworks, regulation market polices must be ad-
Total revenue (M$)
0 REFERENCES
−2 PJM Manual 12: Balancing Operations, 2012. URL http:
0 2 4 6 8 10 12 14 16 18 20
Year //pjm.com/~/media/documents/manuals/m12.ashx.
Theodor Borsche, Andreas Ulbig, Michael Koller, and
Fig. 8. 20 years NPV revenue comparison of a 0.6 MWh Göran Andersson. Power and energy capacity require-
BESS providing PFC in ENTSO-E and RegD in PJM. ments of storages providing frequency control reserves.
In Power and Energy Society (PES) General Meeting,
The results show a generally good regulation performance 2013 IEEE Vancouver, 2013.
with a score between 0.9 to 0.99, while the average per- Stephen D Downing and DF Socie. Simple rainflow
formance score in the PJM regulation market was around counting algorithms. International Journal of Fatigue,
0.7 to 0.8 4 . With a shorter averaging period, the BESS 4(1):31–40, 1982.
adjusts its operating point more aggressively and thus Erdal Kayacan, Baris Ulutas, and Okyay Kaynak. Grey
results in a lower performance score. And the degradation system theory-based models in time series predic-
is more severe due to smaller battery size and more intense tion. Expert Systems with Applications, 37(2):1784–
operation. 1789, 2010.
Muhammad Khalid and Andrey V Savkin. Model predic-
5.3 Revenue Analysis tive control based efficient operation of battery energy
storage system for primary frequency control. In Control
Net present value (NPV) revenue analysis is performed for Automation Robotics & Vision (ICARCV), 2010 11th
both simulation cases with the established pricing and a International Conference on, pages 2248–2252. IEEE,
discount rate of 6%. The BESS size is 0.6 MWh and the 2010.
regulation capability is 1 MW in both cases. Regulation Long Lam, P. Bauer, and E. Kelder. A practical circuit-
payments are assumed to be proportional to BESS’s state based model for li-ion battery cells in electric vehicle
of health (SoH), while the cost includes installation, cell applications. In Telecommunications Energy Conference
replacement, cooling and intraday market transactions. (INTELEC), 2011 IEEE 33rd International, pages 1–9,
Based on the simulation results, battery cells are replaced 2011. doi: 10.1109/INTLEC.2011.6099803.
every 7 years in the ENTSO-E PFC case, and every 2 years Pascal Mercier, Rachid Cherkaoui, and Alexandre
in the PJM RegD case. Oudalov. Optimizing a battery energy storage system
The result shows a much more profitable situation in for frequency control application in an isolated power
providing RegD regulation in PJM, with a system payback system. Power Systems, IEEE Transactions on, 24(3):
time in 3 years, while the ENTSO-E PFC case is shown 1469–1477, 2009.
to not be profitable. This is mainly due to the different A. Millner. Modeling lithium ion battery degradation
levels of reserve capacity payments. Regulation payments in electric vehicles. In Innovative Technologies for
from PJM are roughly 6 times higher compared to PFC in an Efficient and Reliable Electricity Supply (CITRES),
ENTSO-E. 2010 IEEE Conference on, pages 349–356, 2010. doi:
10.1109/CITRES.2010.5619782.
6. CONCLUSION A Oudalov, D Chartouni, C Ohler, and G Linhofer. Value
analysis of battery energy storage applications in power
systems. In Power Systems Conference and Exposition,
In this paper, control strategies aiming at maintaining
2006. PSCE’06. 2006 IEEE PES, pages 2206–2211.
BESS’s SoC level in frequency regulation applications
IEEE, 2006.
are presented. An intraday market based SoC control
Alexandre Oudalov, Daniel Chartouni, and Christian
method is proposed, with which BESS’s operation can be
Ohler. Optimizing a battery energy storage system
undisturbed when its SoC being restored, and a GM(1,1)
for primary frequency control. Power Systems, IEEE
based SoC predictor is developed to offer forecasts for the
Transactions on, 22(3):1259–1266, 2007.
controller. An averaging SoC control method is shown as
Yann G Rebours, Daniel S Kirschen, Marc Trotignon, and
an alternative choice when certain flexibilities are allowed
Sébastien Rossignol. A survey of frequency and volt-
in providing regulation.
age control ancillary servicespart i: Technical features.
Regulation simulations are performed for providing regu- Power Systems, IEEE Transactions on, 22(1):350–357,
lation in Germany and in PJM, USA. With the proposed 2007.
control strategies, a BESS size of 0.6Pn h can be enough for J. Vetter, P. Novk, M.R. Wagner, C. Veit, K.-C. Mller,
both cases. By modeling the degradation as well as other J.O. Besenhard, M. Winter, M. Wohlfahrt-Mehrens,
cost factors, NPV revenue analysis shows a much more C. Vogler, and A. Hammouche. Ageing mechanisms in
profitable case in PJM compared to Germany. This result lithium-ion batteries. Journal of Power Sources, 147
4 Source: http://www.pjm.com/markets-and-operations/
(12):269 – 281, 2005. ISSN 0378-7753.
ancillary-services/mkt-based-regulation.aspx (last accessed: Kun-Li Wen. Grey systems: modeling and prediction.
2013-08-21) Yang’s Scientific Research Institute, 2004.
4029