5.1. Project Overview
Monash University has committed to transitioning to net zero emissions by 2030. As a part of this transition, a microgrid is being developed to demonstrate how a 100% renewable electricity system can operate reliably, and the value it can provide to customers and the broader energy network. This Microgrid provides a realistic and useful platform for research into technological, business and customer behavioral features of the deployment of distributed resources and their coordination through microgrid operations. The microgrid system is intended to be a fully functioning local electricity network and trading market with dynamic optimization of resources interacting with an external energy market.
Monash Microgrid includes a range of customers and DER assets and is located at the Clayton campus, 20 km southeast of Melbourne’s central business district. The microgrid network indicating different assets and their location is presented in Figure 3
. The campus microgrid is composed of controllable loads totaling 3.5 MW, 1 MW of solar generation, two 22 kW EV chargers, and 1 MWh of battery storage. The energy management of these DERs and the microgrid as a whole will be achieved through the deployment of a smart microgrid platform. The microgrid’s 3.5 MW controllable loads are the combination of the 20 mixed use and multi-tenanted buildings. The microgrid connected buildings closely replicate a real community with a variety of old and new buildings and a wide diversity of load profiles that emulate high occupancy spaces (arts theatre, laboratories and lecture theatres), light industrial (swimming pool complex), commercial (retail and staff offices) and residential (self-contained apartments).
In addition to the buildings, the microgrid also has a number of EV chargers, which also act as controllable loads. The microgrid also includes 1 MW of solar generation through 8 separate PV systems installed on the rooftops of the buildings. While they are physically integrated with individual buildings, they can be considered as independent DERs when being managed and not as a generator of the building. The 1 MWh of battery storage is an integrated storage system consisting of a 120 kW/120 kWh lithium-ion battery and a 180 kW/900 kWh vanadium flow machine. As with the PV, the system has been installed and integrated with one of the buildings but can be managed as an independent DER.
5.2. Smart Microgrid Platform
The efficient energy management of the microgrid is achieved through a framework of progressive enablement layers, as illustrated in Figure 4
, that introduces the compatibility with transactive energy governance. The foundational layers of DER Integration
, Active Grid Management
, and Smart Energy Management
provide the functional and technological abstractions required to establish fundamental transactive microgrid behaviors. The formative completion with a governing TEM affords the managed operational autonomy required to classify as the framework as a smart microgrid
The detailed illustration of the enablement layers of transactive energy management is given in Figure 5
. The DER Integration
layer (Figure 5
a) is a combined hardware and software deployment and integrated with each DER. An IoT Edge device is deployed on each DER with direct connections to the DER control systems and metering. A containerized IoT platform is installed on each device to provide a base for extending DER with data and control capabilities. Fundamental to the IoT platform is the decentralized transport of messages between nodes using a secure IoT message bus. It is crucial that DER can securely communicate with other DERs in the microgrid.
In preparation for the deployment of a smart microgrid platform, an enabling IoT hardware installation has been performed on all DERs in the Monash Microgrid. Ensuring the standards and maintenance requirements of the campus electrical infrastructure are met in a compliant and management manner necessitates a common method of physical integration, and installation of IoT functionality is applied to all DERs. This has been achieved through a uniform integration design that is implemented and installed using a common cabinet based hardware form factor. Connections to the DER’s metering is through indirect Modbus TCP connections, where a gateway proxy communication between the meter and the IoT device. The PV systems are also connected by Modbus TCP through a gateway. Although the gateways do not provide further functionality, they do enforce a security precaution by keeping the DER and IoT systems on separate networks.
The integration design for buildings differ slightly in that while the specified building controller also enforces a network separation, the controller also performs additional functionality beyond just proxying communication. There are a variety of different building automation systems (BASs) deployed across the 20 buildings, and the functionality to map a common control interface to the individual BAS control actions should be implemented as extensions to the building control and not as IoT functions. By adding smart building controllers as the entry point to all BAS systems, a platform can be deployed encapsulating the complexities of exposing BAS control under a common campus standard for building control. Figure 6
shows the physical installation of the integration design for a building DER. The common cabinet form can be seen to house the IoT device (a), the building controller (b), along with their supporting electrical switches (c), and internal networking interface (d).
The Active Grid Management
layer (Figure 5
b) introduces data monitoring and control capabilities to the microgrid DERs. These capabilities are deployed as extensions to the installed DER IoT platforms and allow the energy data and, where relevant, power quality data and control to be made available on the IoT message bus. These higher level DER capabilities allow the microgrid DERs to managed under the control of two active grid management cloud systems, a monitoring portal, and a power quality management system. The monitoring portal is a web based reporting and analytical application that combines and presents all available DER data in a single environment to ensure human operators can gain transparency and insight into the microgrid’s operation. The power quality management system runs a live model of the microgrid using the latest data available from DERs to actively manage the microgrid power quality through detecting or predicting quality events and dispatching corrective or preventative actions to DERs with exposed power quality control capabilities.
The Smart Energy Management
layer, presented in Figure 5
d, adds two further sets of functionalities to the microgrid: DER energy management capabilities and a market-based solution for the aggregation, planning and dispatch of microgrid’s DERs. The identified DER energy management functionality required for transactive energy is achieved through the addition of capabilities for demand forecasting, flexibility forecasting, and flexibility control. As illustrated in Figure 5
c, each agent has a Flexibility Manager
to indicate the range of flexibility which can be provided based on the demand and flexibility forecasts, and to control flexible assets in response to the flexibility requests from the market. Building on the preceding layers of DER Integration
and Active Grid Management
, these forecasting and control capabilities complete the functional foundation required for the market-based energy management solution of TEM.
While the TEM energy management functionality is a component of the Smart Energy Management layer, its complexity and decoupled foundation of DER flexibility capabilities warrant consideration as the final TEM governing layer of the Smart Microgrid Platform. This governing architecture is derived from the TEM design principles introduced in this paper with particular accommodation to the encapsulation of preference privacy and decision making within independent transactive DER.
5.3. TEM Implementation: Diagrams and Technical Architecture
The implementation of the TEM in the Smart Energy Management
layer requires a distributed software solution, which provides the functionalities needed in achieving an efficient energy management. The key components of the TEM implementation and their interaction diagrams are presented in Figure 7
. Figure 7
a illustrates the key components of the TEM, where a central transactive microgrid system is paired with necessary transactive DER capabilities. The separation is reflective of the design principles of delegating aggregation, planning and orchestration to a distributed market mechanism while ensuring preference privacy and decision making are retained within the control of DER.
The core transactive DER capability is the Transactive Agent
, which enables the participation in the TEM by responding to market requests (Figure 7
b). The Transactive Microgrid system is the platform and host for a TEM Agent
to orchestrate the flexible behavior of Transactive DERs in response to market signals received from both internal and external market requesters. Internal Requesters
are microgrid monitoring capabilities that issue flexibility market requests in response to their monitoring function. For instance, a peak demand monitoring capability would issue a request to reduce the demand during the period for which a risk at exceeding a peak demand level has been forecast. External Requesters
are capabilities which integrate with external systems such as energy or flexibility markets. An external demand response market would be integrated as an external requester capability, which would issue an internal flexibility market request corresponding to the external demand response signal. Signals from both internal and external requesters are resolved using the same Market
capability that triggers the creation of internal markets to plan the coordination of Transactive DERs in the collective fulfillment market requests. The market plan is a collection of Transactive DER commitments to fulfill the market requester’s flexibility request and will be recorded in Microgrid Transactive Energy Ledger
Agent’s request diagram is given in Figure 7
c. With the forecasts provided by the Flexibility Manager
in conjunction with the DER preferences provided by the Preference Manager
, the agent is equipped for considered responses to TEM signals. The Preference Manager
provides a preference representation derived from the DER core function, such as HVAC schedules, and further data sourced from related business systems, such as a customer portal to record willingness to accept flexibility discomfort. Any accepted offer of the agents would be a commitment and needs to be delivered in the market. The delivery of commitments is managed by the Commitment Manager
who manages the flexibility plan by sending flexibility signals to the Flexibility Manger
, as presented in Figure 7
d. The agent’s commitments are recorded in the DER Transactive Energy Ledger
The technical architecture of the Monash Microgrid software solution is presented in Figure 8
. The DER agents are deployed on IoT devices of all transactive agents in the microgrid. The Market components are deployed as cloud services, communicating with agents via standard web protocols such as HTTPS or similar TCP messaging mechanism, represented as Application Bus
. Multiple Docker containers are deployed on each edge device, with services supporting operational flow explained in Figure 7
. These containers use internal messaging bus such as message queuing telemetry transport (MQTT) to communicate. The programming of different containers is implemented using the Java programming language. The cloud components include market requests, market settlement, and billing services. The market and billing systems have application program interfaces (APIs) exposed to user interfaces for Microgrid manager and market participants. To support interaction of the TEM with external requesters, the market API can also be exposed over the internet, to enable communication with the wholesale market or retailers. The configuration file server is hosted on the cloud. A directory is assigned to each edge node as a mirror of its local configuration directory. Configuration manager can dynamically update files in the cloud, which will then replicate to corresponding local drives on edge nodes.
5.4. Illustrative Example
In this section, we present and discuss an application of the TEM to implement the scenario discussed in Section 4.5
within the Monash Microgrid introduced in Section 5
. This section aims to illustrate the impact of the consideration of design choices and pricing mechanisms on achieving different objectives in the market, and therefore, to highlight the value in using a framework to guide the design. We introduce two potential alternative objectives of maximizing individual DER value and minimizing DER discomfort costs. The results of simulating the pricing mechanisms are discussed in terms of both the operating objectives and the broader TEM design considerations.
It is assumed that the External Requester
sends a signal for
kW demand reduction for one hour duration to the microgrid and the rewarding price for this reduction is $
/kWh. This price is an indicator of offers that Monash Microgrid is receiving from network providers to provide flexibility. All of the 20 flexible buildings in the microgrid participate in the market as transactive agents to provide the requested flexibility, where each building is represented by an agent. So hereafter, agent refers to a flexible building in the microgrid. The amount of flexibility that each agent can provide is indicated by the forecasts provided by the Flexibility Manager
and this flexibility is bounded by a minimum and maximum limits. Each agent has a set of private preferences and chooses its flexibility cost parameters based on its preferences, obtained through the Preference Manager
. The flexibility cost parameters, maximum available flexibility, and offered price in auction-based method for all agents are represented in Table 2
. For all agents
are assumed to be zero.
, illustrates the clearing process in auction-based pricing. After receiving all offers from agents and generating the aggregated flexibility curve, the intersection of this curve with the requested flexibility line indicates dispatched/not dispatched flexibilities in the market. The last dispatched building is agent No. 17, with offered price
= 1.8 $
/kWh. Therefore, in uniform price auction, the clearing price would be equal to
, while in pay as bid auction, agents would be paid based on what they have offered.
Simulation results for distributed optimization pricing are shown in Figure 10
. The step size and the convergence criterion are set to 0.01 and 0.001 respectively. As shown in Figure 10
a, after five iterations, the offered flexibility by agents converges to the requested flexibility, and the market rewarding price for agents is 1.72 $
/kWh, which is less than
as shown in Figure 10
The commitment of different agents as a percentage of their offered flexibility in different approaches are compared in Figure 11
a. In the auction-based approach, the dispatched flexibility is based on agents’ offers, and only agents with offers lower than the last dispatched agent’s offer win the auction. While in the distributed optimization pricing, agents can iteratively offer their flexibility based on the offered price by the market operator. The total flexibility cost and total paid money to agents for different mechanisms are compared in Figure 11
b. As illustrated, the lowest flexibility cost is obtained in the optimization-based method. This is due to the fact that the presented optimization-based pricing mechanism allows agents to minimize their flexibility cost at each iteration in response to the offered price by the market operator. The total paid money to agents is the highest in uniform price auction, as market price in this approach is the highest. The auction-based approach allows DERs to maximize their individual profit instead of minimizing the total flexibility cost of agents.
In both auction-based and distributed pricing mechanisms, the aim is to compute a solution that achieves the TEM goal through the coordinated dispatch of transactive agents’ flexibility. Subject to the design of the given pricing mechanism, the degree of an individual DER’s participation in the dispatch coordination is ultimately determined through the decision authority that rests with the DER’s transactive agent. While the decision on participation rests with each agent, the balance between the maximization of value and the minimization of flexibility cost is intrinsic to the function of the pricing mechanism and, therefore, not directly controllable through an agent’s decision. Although both of the discussed pricing mechanisms are designed for the same end, it is evident in the results that they achieve this end with a distinct variation in the competitive opportunity and balance between value and discomfort cost.
The competitive opportunity afforded to individual agents through the auction-based mechanism limits the flexibility provision to only the subset of agents whose offers were successful. This results in a higher total payout and an increased share of the total flexibility value distributed to the successful agents. This increased total and individual value come not only with the increased discomfort cost but also with the inevitable risks of a competitive auction environment. When considering these risks, particular consideration must be given to potentially lucrative collusive, predatory and entry deterring behaviors when the designing auction mechanisms for energy markets [52
The iterative nature of the distributed optimization mechanism introduces a more involved method of participation, where all agents contribute to the provision of flexibility. Unlike the auction mechanism where an agent’s successful participation is primarily whether they are included in the subset of those with accepted offers, successful participation in the distributed mechanism is a more interpretable consideration. This consideration is in fact an agent’s enactment of their decision authority in the iterative re-evaluation of their offered quantity of flexibility in response to the offered price by the market operator and their own internal discomfort cost. Consequently, an agent’s competitive opportunity when participating in the distributed mechanism is largely contained within their own efficiency of decision making and discomfort cost. While the distributed mechanism does share the same risk of exploitation by untrustworthy agents, it is constrained by the technical risk of scalability where the communication and processing overhead of the iterative computation will increase with the number of participating agents and therefore be limited by the minimum time taken to compute a solution [53
With these considerations, it is evident that the outcomes of different pricing mechanisms are likely to align with some objectives for applying TEM but not others. An objective of creating a competitive environment where individual agents are able to maximize their own value is suited to an auction-based mechanism. Whereas the objective of ensuring the minimization of the total discomfort cost of all agents is best served with the application of distributed pricing mechanism. Hence, in designing an objective-driven application of TEM, it is a fundamental consideration to design the pricing mechanism in conjunction with a clear understanding of the relationship between the TEM objectives and the likely market outcomes that can be expected from the application of the pricing mechanism. In turn, these functional and technical considerations must align with broader organizational and objectives and regulatory constraints. These must be addressed in the formulation of an efficient and strategically optimized framework for the successful application of TEM within an organization and physical DER environment.