workshop on pv applications in power networks
TRANSCRIPT
The project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 810809
U
U
WORKSHOP ON PV APPLICATIONS
IN POWER NETWORKS
1st June 2021
Workshop
DISTRIBUTION BATTERY ENERGY STORAGE COORDINATION
FOR IMPROVED POWER NETWORK SUPPORT
Dr Alessandra Parisio
and
Dr Tianqiao Zhao, Dr Xiao Wang, Prof. Jovica V. Milanovic
http://crossbowproject.eu/
Outline
• Introduction and Motivation
• Dynamic fast frequency and voltage regulation
• Static fast frequency regulation
• Frequency regulation and congestion management
• Conclusions and future steps
Outline
• Introduction and Motivation
• Dynamic fast frequency and voltage regulation
• Static fast frequency regulation
• Frequency regulation and congestion management
• Conclusions and future steps
Challenges in Renewable Integration
Mahmud, N. and Zahedi, A. (2016). Review of control strategies for voltage regulation of the smart distribution network with high penetration of renewable distributed generation.
Renewable and Sustainable Energy Reviews, 64, 582–595.
• Distributed and renewable generation can result in significant benefits but
reduced total system inertia and reduced controllability
• The intermittency of renewable sources and its reversed power flows leads to new
balancing issues in both transmission and distribution networks
• The variability of renewable generation challenges the current practice of grid
regulation to maintain frequency and voltage stability
• At the distribution level, small photovoltaic systems generate reversed power flows
into the transmission level, resulting in significant voltage increases
Opportunities for Storage
Strategic Energy Technology Plan, European Commission and National Grid ESO, “Facilitating the transition to a flexible, low carbon energy system”, December 2019
Scully, J. (2020). Energy storage news. https://www.energy-storage.news/news/europesresidential-market-installed-745mwh-of-batterystorage-in-2019-solar/
• Storage is advocated as an excellent candidate to support safe and stable
operation of sustainable power grids and defer network investment
• System Operators (SOs) are exploring the emerging technical solutions that
storage technologies embedded within the distribution networks can afford
• Newly commissioned storage systems are mainly the small-scale ones, located at
the distribution level and close to the end-users, such as residential battery units
typically coupled with rooftop PVs, which showed a 57% annual increase in Europe
in 2020
Benefits of storage and PV integration
Strategic Energy Technology Plan, European Commission and National Grid ESO, “Facilitating the transition to a flexible, low carbon energy system”, December 2019
Scully, J. (2020). Energy storage news. https://www.energy-storage.news/news/europesresidential-market-installed-745mwh-of-batterystorage-in-2019-solar/
• Increased flexibility
• Reduction of the amount of clean electricity
curtailed at times of grid congestion or system
instability
• Reduction of imbalance charges and penalties
• Improved reactive power provision for dynamic
voltage control, even when solar is not available
• Storage participation into the balancing/energy and ancillary service markets
Motivation
• Explore the potential contribution from distributed storage technologies to efficient grid
operation and renewable integration
• In order for these flexible devices to provide an adequate service a very large number of
them must be efficiently aggregated and coordinated
Potential solution: Virtual Storage Plants
Aggregation of storage units with
same/different technologies at various
locations to provide grid support while
maximizing their performance and reducing
costs
Challenges and Contributions
• Additional uncertainties introduced by the integration of renewable generation
• New regulation services launched by system operators require dynamic and fast response capabilities
(within 1-2s)
• Huge control challenges to the coordination of a large number of geographically dispersed storage devices
and to the capability of dynamically responding to the time-varying network conditions
T. Zhao, A. Parisio, J.Milanovic, Distributed Control of Battery Energy Storage Systems for Improved Frequency Regulation, IEEE Transactions on Power Systems, 2020
T. Zhao, A. Parisio, J.Milanovic, Location-dependent Distributed Control of Battery Energy Storage Systems for Fast Frequency Response, International Journal of Electrical Power & Energy Systems, 2020
Contributions
• Scalable control frameworks to dynamically select and aggregate the most suitable combinations of
storage devices of any size that should be considered for effective service provision
• Optimal real-time coordination of thousands of storage devices accounting for both local and global
objectives
• Taking advantage of the emerging fast response capabilities (0.1-0.2s) of battery energy storage systems
Why Distributed Control
In order to be effective the control framework must
• have real-time capabilities and light computational
burden
• have adaptive and plug-and-play capabilities
• guarantee the constraint satisfaction
• integrate a feedback mechanism
• be scalable
DGDG
DG
Control center
DG
DGDG
DG
DG
LC LC
LC
LC
ESSESS
ESS
ESS
LC LC
LC
LC
Centralized Decentralized Distributed
DGDG
DG
Control center
DG
DGDG
DG
DG
LC LC
LC
LC
ESSESS
ESS
ESS
LC LC
LC
LC
Centralized Decentralized Distributed
DGDG
DG
Control center
DG
DGDG
DG
DG
LC LC
LC
LC
ESSESS
ESS
ESS
LC LC
LC
LC
Centralized Decentralized Distributed
Prone to failures and
communication
issues, less cost-
effective and robust,
lack of scalability
Suffer from
instability and sub-
optimality issues
Distributed vs. Centralized:
less communication, less
computation
Distributed vs. Decentralized:
more coordination, improved
solution
Distributed Optimisation-based Control
Dynamic fast frequency and voltage regulation• Pre-fault services and fast delivery
• Technique: distributed Online Convex Optimisation (OCO)
• Pros: computationally cheaper and asymptotic convergence to the optimum
Static fast frequency regulation• Post-fault service, fast delivery and short duration
• Technique: Alternating Direction Method of Multipliers (ADMM)
• Pros: fast practical convergence properties
Frequency regulation and congestion management• Simultaneous frequency regulation and congestion in multi-area networks
• Technique: Consensus and distributed primal-dual algorithm
• Pros: power network and VSPs dynamics embedded to drive the system to an equilibrium at
minimum cost
Outline
• Introduction and Motivation
• Dynamic fast frequency and voltage regulation
• Static fast frequency regulation
• Frequency regulation and congestion management
• Conclusions and future steps
Online Convex Optimisation (OCO)
• At time t, the player chooses a strategy
without the knowledge of the current cost
• The player observes the revealed cost
function and incurs cost
• Regret: the difference between the cost
incurred and the best fixed point chosen
offline
• Goal: sub-linear regret function (on average
the algorithm performs as the best strategy
in hindsight)
Gordon GJ. Regret bounds for prediction problems. InCOLT 1999 http://www.cs.princeton.edu/ ehazan/tutorial/OCO-tutorial-part1.pdf
OCO-based control for ESS coordination
• Multi-agent system framework where each
ESS is an agent
• ESS setpoints calculated in less than 1s
• Plug-and-play functionality
• Suitable criteria optimised
Minimise ESS costs and maximise reward
Minimise tracking errors and deviations
• Global and local constraints satisfied
ESS Local Controller
OCO Problem Formulation
Fast dynamic frequency regulation Voltage regulation (reactive power)
subject to:
Min. Costs/Max. Reward
• Storage and frequency dynamics
• Power balance
• Operational and capacity constraints
• Service technical requirement
subject to:
Min. Tracking + Local voltage deviations
+ Costs
• Storage dynamics
• Network and inverters modelling
• Operational and capacity local
constraints
• Voltage-related local constraints
T. Zhao, A. Parisio, J.Milanovic, Distributed Control of Battery Energy Storage Systems for Improved Frequency Regulation, IEEE Transactions on Power Systems, 2020
T. Zhao, A. Parisio, J.Milanovic, Distributed Control of Battery Energy Storage Systems in Distribution Networks for Voltage Regulation at Transmission-Distribution Network
Interconnection Points, Control Engineering Practice, submitted
Distributed OCO Algorithm
1. Receive information from neighbouring agents
(e.g., voltage magnitudes, dual variables, local
estimation of the global information)
2. Consensus-based local estimation of global
information (e.g., supply-demand mismatch)
3. Local updates of the control inputs (e.g., active
and reactive power setpoints) and the
Lagrangian multipliers
Solution is based on the information at the last time
step (e.g., local voltage and SoC measurement)
Frequency Regulation – Plug and Play
IEEE 33-bus system. Ten BESS with maximum power ratings ranging from 4.8 to 5.5 MW. Unexpected
and sustained supply-demand mismatch at t=0, modelled as a random variable with a uniform
distribution U(22.5,27.5)MW. BESS7 fails at t = 4s and is recovered at t = 23s
Storage power outputs Frequency response
The system frequency is regulated to the nominal value within 30 seconds, as required, and the service
provision is sustained under unexpected plug-and-play operation
Frequency Regulation - Results
Total BESS power output
Computation times
Evolution of the regret function over time
Even for 2000 BESS, BESS
power setpoints are calculated
within 0.5s
Voltage Regulation - Results
IEEE 123-bus test feeder with 100 residential ESS (1.2 kW
rating each) and 3 commercial PV generators (60 kW rating
each) . Three TN-DN interconnection points (Nodes 31, 60 and
150)
Voltage profiles kept within the limits –TSO time-varying voltage
profiles satisfactorily tracked (maximum error of 0.8e−3)
Voltage profiles at all buses:
distributed approach
Voltage profiles at all buses:
decentralized approach
Outline
• Introduction and Motivation
• Dynamic fast frequency and voltage regulation
• Static fast frequency regulation
• Frequency regulation and congestion management
• Conclusions and future steps
Problem Formulation
Fast static frequency regulation
subject to:
Min. Costs/Max. Reward
• Storage and frequency dynamics
• Power balance
• Operational and capacity constraints
• RoCoF and frequency nadir
requirements
T. Zhao, A. Parisio, J.Milanovic, Location-dependent Distributed Control of Battery Energy Storage Systems for Fast Frequency Response, International Journal of Electrical Power
& Energy Systems, 2020
Static Frequency Response - Results
IEEE 14-bus system. A total capacity of the installed RES in the system is 120 MW. Ten BESS with
maximum power ratings ranging from -4 to 4 MW.
Unexpected and sustained supply-demand mismatch of 12
MW occurs at Bus 11 at t=0.
Magnitudes of Z-bus matrix
With respect to PI-based control, RoCoF is reduced by 53%, the frequency nadir is improved from 49.18
Hz to 49.58 Hz and cost is reduced by 7%
Heat map of BESS power outputs
Static Frequency Response - Results
A modified IEEE 118-bus system with 100 BESS randomly distributed around the network
The computational time of the algorithm
is 0.18 s, which meets the fast-FR
service criterion of response time (1 s)
and is comparable with PI-based control
computational time (0.112s)
Convergence rates under different communication topologies
The topology has an impact on the
convergence rate, which is very fast
for all the topologies
Outline
• Introduction and Motivation
• Dynamic fast frequency and voltage regulation
• Static fast frequency regulation
• Frequency regulation and congestion management
• Conclusions and future steps
Coordination of Multiple VSPs
X. Wang, T. Zhao, A. Parisio, Frequency regulation and congestion management by Virtual Storage Plants, SEGAN, to be submitted
power setpoint
Area 1
Tie line
ESS
ES
ESS
ES ES ES
ESESES
ES ES
DC DC
DC DC
DC
DC DC
DCDC
VA
Transmission level(top layer)
Virtual Storage Plant (VSP)
Area 2
Area 3
GG
G
1
2
3
4 5
6
7
8 9
G
GG
G
1
2
3
4 5
6
7
8 9
G
VSP
VSP
VSP
VSP
VA
VA
VA
VA
VSP Aggregator
VSP Aggregator
Tie line
Area 4 𝒩𝑝
G
Communication line Transmission line
Distribution level(lower layer)
L
RES
ሶ𝜃 = 2𝜋𝑓0𝜔
𝑀 ሶ𝜔 = 𝐶s𝑃𝑠 − 𝑃𝑑 − 𝐷𝜔 − 𝐶𝑃𝑒
𝑃𝑒 = 𝐵𝐶𝑇𝜃
ሶ𝑆𝑜𝐶𝑘𝑠 = −
1
𝐶𝐴𝑘𝑠 ∙ 𝜂𝑘
𝑠 𝑃𝑘𝑠, ∀𝑘 ∈ 𝒮
𝑇𝑘𝑠 ሶ𝑃𝑘
𝑠 = −𝑃𝑘𝑠 + 𝑢𝑘
𝑠 , ∀𝑘 ∈ 𝒮,
subject to:
𝐶s𝑃s − 𝑃d − 𝐷𝜔 = 𝐶𝑃𝑒
𝑃𝑒 ≤ 𝐵𝐶𝑇𝜃 ≤ 𝑃𝑒
𝑃𝑠 ≤ 𝑃𝑠 ≤ 𝑃𝑠
𝑆𝑜𝐶𝑠 ≤ SoC𝑠 ≤ 𝑆𝑜𝐶𝑠
min.𝑃s,𝜃,𝜔
𝑘∈𝒮
𝑐𝑘 𝑃𝑘𝑠
Distribution level:
Driving the storage devices within
a VSP to track the power setpoints
calculated by the VSP aggregators
and to achieve an agreed state of
charge (SoC)
Transmission level:
Coordinating VSPs to improve the
frequency profile and the
congestion management through
the tie lines of the multi-area
system
Outline
• Introduction and Motivation
• Dynamic fast frequency and voltage regulation
• Static fast frequency regulation
• Frequency regulation and congestion management
• Conclusions and future steps
Conclusions and Future Works
- Distributed control framework for an optimal coordination of an arbitrary number of
storage devices for efficient grid support
- Storages coordinate each other without the need of any central entity and without sharing
any local private information with the TSO, the DSO or the storage aggregator
- Control framework to be adopted by an aggregator, such as a Virtual Storage Plant, or the
DSO
Future Works
- Extension to robustness against several other sources of uncertainty, including the
communication delays
- Analysis of the communication network design
- Extension to include demand technologies and the simultaneous provision of other
system services, e.g., new frequency services
Coordination of Multiple VSPs - Results
Four-area networks with152 buses, 30 generators, 10% RES penetration, 17 cross-border tie lines. 5
VSP assets, each with 500 storage units (Power rating: 50 – 100 kW; Energy rating: 30 – 50 kWh). Flow
limit of 110MW. The penetration of renewables is around 10% at the rated load condition. 30% step-load
increase occurs in one area at 0.1 second.
Diagram of the testbed using DigSILENT and MATLAB
• VSP-DC: proposed control framework
• VSP-LFC: a dedicated LFC for VSPs
• Baseline: storage stays idle and the system
frequency is supported by the synchronous
generators
Coordination of Multiple VSPs - Results
Improved frequency regulation and
cross-border power transfer,
congestion issues avoided
0 50 100
Time (s)
-20
0
20
40
60
80
100
VS
P p
ow
er
(MW
)
0 50 100
Time (s)
-5
0
5
10
15
20
25
0 50 100
Time (s)
-5
0
5
10
15
20
25
(c) (d) (e)
Baseline VSP-LFCVSP-DC
no droopVSP-DC droop