P2P Assign3

You might also like

Download as pptx, pdf, or txt
Download as pptx, pdf, or txt
You are on page 1of 21

Peer-to-Peer Control

System for DC
Microgrids

VANEEZA AHMED
02-133202-003
BEE-8A
Introduction
Current smart grids, including nanogrids and microgrids, heavily rely on the traditional AC grid for reliability,
governed by standards like IEEE 1547 and IEEE 2030. However, IEEE 1547 focuses on large generation and adapting
systems to connect to the existing grid.
An emerging approach promotes bottom-up, grid-independent, and decentralized designs to enhance self-sufficiency
and resilience. Despite their potential, few studies offer comprehensive control systems for microgrid interconnection
due to the complexity of achieving both system automation and power stability, and the need to redesign multiple
infrastructure layers.

Advancements in Information Communication Technologies (ICT) are enabling a shift towards decentralized grid
systems, which are more dynamic, local, resilient, flexible, and scalable. This presentation proposes and implements a
decentralized control system for a DC microgrid, using peer-to-peer (P2P) networking concepts. The system
eliminates single points of failure by allowing each household to autonomously negotiate and execute energy deals
with its neighbors, without requiring centralized control or global knowledge. By leveraging DC power and modular
software, the system achieves high flexibility and resilience, as demonstrated through real-world testing and
simulations.
Overlay Network as Control Structure
This presentation leverages overlay networks, which are communication structures built on top of existing
networks (like the internet), for decentralized control in smart grids.

Benefits: Facilitate intelligent communication, improve scalability and flexibility.

Control Structures via Overlay Networks:

• Client/Server: Centralized control entity communicates with distributed devices.


• Pure P2P: All devices communicate directly with each other.
• Hybrid P2P: Combines centralized coordination with direct device communication.

This approach aligns with Rohbogner's control structure classification based on information processing location.
Overlay Network as Control Structure
A. Central-Hierarchical B. Distributed- C. Decentralized
Hierarchical
A central control entity Local control entities have Distributed physical systems are
communicates with all local some autonomy, but still report fully controlled and optimized by
control entities to collect data, to a higher-level coordinator. local CEs.
calculate optimal control No dedicated central control
strategies, and issue entity, allowing for dynamic
commands. election of coordination roles.
Corresponds to Pure P2P, where
any single entity can be removed
without loss of network service.

D. Control Structures in Practice:


Approaches for realizing P2P control structures:
1. Combination of TCP/UDP messages lacks clear middleware, scalability.
2. Agent-based middleware like JADE supports "hybrid" P2P.
3. Generic middleware for decentralized systems, such as Distributed Hash Table (DHT), adapted for
Pure P2P.
Classification of Control Structures

Table 1: classification of control structures


Decentralized Control Structures via Overlay
Networks
This paper explores using overlay networks for decentralized control in smart grids. Here, we detail three
control structures mapped to these networks:

Central-Hierarchical (Client/Server): 1
• A central entity monitors,
optimizes, and controls all Distributed-Hierarchical (Hybrid P2P)
devices. 2
• Resembles a traditional client- • Local control entities have some
server architecture. autonomy but rely on a central
entity for system-wide control.
• Similar to a hybrid P2P network
Decentralized (Pure P2P) with central coordination.
3
• Local control entities fully
manage their systems and
communicate directly with each Implementation Approaches:
other.
• Equivalent to a pure P2P • Limited P2P (TCP/UDP): Scalability
network with no single point of challenges.
failure. • Agent-based middleware (JADE): Supports
hybrid P2P only.
• Generic middleware (DHT): Suitable for pure
P2P due to scalability and robustness.
Autonomous Energy Exchange
A. Controlled Physical System

We consider an Open Energy System (OES) where autonomous DC subsystems are interconnected via a DC power
bus, as illustrated in Figure 1. Each house in the system is equipped with a DC nanogrid that includes batteries,
energy sources (such as PV panels), and loads, forming a basic standalone subsystem. These subsystems are
interconnected within a community-wide DC microgrid using bidirectional DC-DC converters. These converters
actively control power flow, allowing DC power exchange within the community to balance demand-response
requirements, enhancing self-sufficiency and resilience without requiring global knowledge or control.

Fig. 1. General architecture of a heterogenous Open Energy Systems (OES).


Autonomous Energy Exchange
B. Control Logic
The control logic for autonomous energy exchange involves peers (units) making power deals to either give or
receive power. The process is divided into deal negotiation and deal execution, each governed by distinct
procedures.

1. Flow Charts:

• Deal Negotiation: Each unit uses an individual scenario to negotiate deals. Requests for power exchange
are generated based on the unit's status (e.g., battery status, solar input) and personal scenario, then forwarded to
other units. Responses are analyzed to choose the most favorable offer. If a deal is agreed upon, confirmation is
sent to the chosen responder.

• Deal Execution: A Grid Master executes the agreed deals according to a community-wide policy, ensuring
safety and adherence to rules. The Grid Master executes and monitors multiple power deals and, upon
completion of its own deal, initiates the Grid Master release procedure. If other deals are ongoing, a new Grid
Master is selected through a switching procedure.
Autonomous Energy Exchange

Fig. 2. Flow chart to negotiate and execute power exchange deals


using scenario and policy files for decision making (decisions
affected by scenario / policy are highlighted).
Autonomous Energy Exchange
2. Exchange Scenario:

Instead of system-wide optimization, local optimization scenarios are implemented without requiring global
knowledge. Two basic strategies are presented based on battery State of Charge (SoC):

• Trigger-only Strategy: This strategy uses start and stop triggers based on SoC levels. It is stateless, as both
participants can stop the deal anytime, with the Grid Master enforcing end-of-deal for safety.

• Amount-based Strategy: The energy amount to be exchanged is negotiated based on the scenario, with
deals limited to small quantities to reduce uncertainties. The Grid Master monitors the transfer until the agreed
energy amount is exchanged, then stops the deal.

These control logic frameworks allow decentralized, autonomous energy exchanges within a DC microgrid,
promoting efficiency and resilience in community energy management systems.
Autonomous Energy Exchange
2. Exchange Scenario:

Fig. 3. Examples of SoC based strategies. Left: both start and stop are determined
according to triggers. Right: maximum amount is determined at deal agreement
according to four states depending on the stored energy
Autonomous Energy Exchange with Decentralized
Control
A. Controlled Physical System (Open Energy System):*

 Houses interconnected via DC power bus to form a microgrid.


 Bidirectional DC-DC converters facilitate power flow control.

B. Control Logic:

1. Deal Negotiation and Execution:


 Negotiation: Units exchange requests and responses based on individual scenarios and a community-
wide policy.
 Execution: A designated Grid Master executes deals according to the policy and monitors safety.

2. Exchange Scenarios (based on Battery State of Charge):


 Trigger-only strategy: Simple, stateless approach using start/stop triggers.
 Amount-based strategy: Units negotiate the amount of energy to be exchanged.
Control Software Layering/Implementation:
TESTING ENVIRONMENT
The proposed concepts were verified using both a controlled testing environment and a full-scale physical system,
complemented by simulations to evaluate performance.

A. Verifying the Control Structure


To test the control structure and message exchange, we used 35 physical controllers (BeagleBone Black running
Debian 7.0) communicating over a LAN in a peer-to-peer manner (Fig. 5). The physical system (L0-L1) was
emulated centrally, calculating power flows for all controllers similar to the simulator

Fig. 4. Testing environment for Control Structure using 35 individual


TESTING ENVIRONMENT
B. Verification on Full-Scale Platform
The control structure was implemented in Okinawa, Japan, on a full-scale platform involving 19 inhabited family
houses. Each house's DC subsystem included:

• 2.8kWp or 4.2kWp PV panels with PV chargers (DC-DC converters with MPPT control).
• 4.8kWh Olivine Lithium-Ion Iron Phosphate batteries with a Battery Management Unit (BMU).
• An Uninterruptible Power Supply (UPS) for AC power input and supply.
• An internal controller for internal submodules (excluding the DC-DC converter).
• A bidirectional DC-DC converter (resonant-type, 2.5kW output) for voltage and current regulation.
• A Linux-based controller connected to Ethernet for autonomous exchange software and internal module
interface.

The physical control procedure was first tested on a lab prototype with three subsystems and then deployed in 19
houses in December 2014. From December 2014 to October 2015, we tested the hybrid P2P approach with trigger-
level-based control. Since October 2015, a pure P2P control scheme with an amount-based exchange strategy has
been in use. During six utility blackouts, the system provided uninterrupted power, and the autonomous exchange
algorithm continued to operate, allowing power-sharing between houses.
Evaluation and Simulation Results:

A. Input Data B. Simulator Setup


We used annual demand data from 19 A Python-based simulator emulated the
houses and solar radiation data measured by power flows for all subsystems at each time
a local weather station. Missing demand step (52560 t/year). The simulation
data was interpolated using daily patterns included linear approximations of losses
and community-wide averages. and was tuned to match hardware
limitations, achieving a 4-6% discrepancy
with recorded annual SSR. Optimizations
were applied by removing limitations and
neglecting non-DC system losses.
Evaluation and Simulation Results:

C. Simulation Results
Performance Indicators:
Self-Sufficiency Ratio (SSR):

Solar Operation Ratio (SOR):

Collaborative Resilience Factor (CRF):



Verification and Testing
1 Control Structure 2 Full-Scale Platform 3 Simulation and
Verification Evaluation
The peer-to-peer control The complete control A simulator was used to
structure was tested using system was deployed in compare the self-
35 physical controllers Okinawa, Japan on a sufficiency, solar
communicating over a microgrid of 19 utilization, and resilience
local network, with a interconnected houses, of the decentralized DC
central emulator simulating demonstrating real-world exchange approach against
the physical system. operation and resilience standalone and centralized
during grid outages. systems.
Self-Sufficiency Improvements

Self-Sufficiency Ratio Solar Operation Ratio Collaborative


(SSR) (SOR) Resilience Factor
(CRF)
The decentralized DC The SOR, which measures
exchange approach can solar energy utilization, is In emergency situations, the
significantly improve the also improved by the decentralized architecture
SSR compared to standalone autonomous exchange, as it allows neighboring houses
systems, by enabling better helps minimize curtailment to share their stored energy,
balancing of supply and of excess solar generation. increasing the resilience of
demand across the critical loads by a factor of
community. over 16 on average.
Flexibility and Resilience

Plug-and-Play Resilience to Outages Emergency Scenarios


The decentralized architecture During grid outages, the In extreme situations, the
allows new subsystems to be autonomous exchange decentralized control structure
dynamically added to the algorithm continued to can be used to prioritize
microgrid, demonstrating the function, enabling houses critical loads, with
flexibility of the peer-to-peer with low batteries to obtain neighboring houses able to
control approach. power from neighbors with share their stored energy to
surplus energy. increase resilience.
Conclusion
This presentation presented a fully decentralized control system for microgrids using a pure P2P network overlay. This
system allows autonomous DC power exchange, verified through simulations and testing environments, demonstrating
improved self-sufficiency and resilience.

Advancements in ICT have made decentralized designs feasible, offering enhanced resilience against blackouts and
network failures. Despite the challenges and longer development times of fully decentralized systems, they provide
significant advantages in unstable environments compared to hybrid systems.

Future research should focus on scenario optimization, user interaction, and cloud-based optimization, with the
potential for advanced control schemes like auction-based energy negotiation.

You might also like