An Interoperable Communication Framework for Grid Frequency Regulation Support from Microgrids

Renewable energy sources, which are controllable under the management of the microgrids with the contribution of energy storage systems and smart inverters, can support power system frequency regulation along with traditionally frequency control providers. This issue will not be viable without a robust communication architecture that meets all communication specification requirements of frequency regulation, including latency, reliability, and security. Therefore, this paper focuses on providing a communication framework of interacting between the power grid management system and microgrid central controller. In this scenario, the microgrid control center is integrated into the utility grid as a frequency regulation supporter for the main grid. This communication structure emulates the information model of the IEC 61850 protocol to meet interoperability. By employing IoT’s transmission protocol data distribution services, the structure satisfies the communication requirements for interacting in the wide-area network. This paper represents an interoperable information model for the microgrid central controller and power system management sectors’ interactions based on the IEC 61850–8–2 standard. Furthermore, we evaluate our scenario by measuring the latency, reliability, and security performance of data distribution services on a real communication testbed.


Introduction
Microgrids (MG) are known as the main blocks of a smart grid [1]. Based on the IEEE Std 2030.7-2017 MG is defined as "A group of interconnected loads and distributed energy resources with clearly defined electrical boundaries that acts as a single controllable entity with respect to the grid and can connect and disconnect from the grid to enable it to operate in both grid-connected or island modes" [2]. MG penetration has increased due to the reduction in the cost of electricity transmission infrastructure from utilizing distributed energy resources (DER) [3,4]. Employing renewable energy sources (RES) in MG due to their uncountable profits will dominate other kinds of DER. However, the nature of RES is intermittent. Therefore, the primary role of inverters, i.e., power conversion and energy feeding, has been developed to provide dispatchable sources, able to overcome RES randomness as well as meet tied power system frequency and voltage deviations. Smart inverters offer services such as harmonics, voltage, and frequency control to achieve cooperation between RES and the utility grid [5][6][7][8][9]. Along with smart inverters, the presence of energy storage systems (ESS) and responsive loads (RL) allows the MGs to participate in the electricity market as a prosumer, selling overproduced electricity and buying in case of resources unavailability or system failure in grid-connected mode [10]. These characteristics have brought the opportunity of employing MG as an ancillary service (AS) provider of utility grid, including voltage control, black-start aid, and especially frequency regulation (FR) The before-mentioned practices applied the DDS with a dedicated communication network infrastructure for MG internal interactions. Moreover, their platforms were mostly implemented in the LAN environment. In a recent effort, Aftab et al. [32] planned an information model based on IEC 61850 for an islanded MG based on IEC 61850 and examined load frequency control (LFC) inside the islanded MG. The authors in this paper arranged WAN on a simulator environment and evaluated latency for their LFC scenarios. In the same line of thought, Hong et al. [33] applied IEC 61850 to provide a fault detection and system restoration in the Multi-MG system with the assistance of hardware-in-loop (HIL) testbed. The authors deployed an Ethernet-based infrastructure for communication which represents a LAN arrangement. However, to prove the implementation possibility of an interoperable, fast, cost-effective, reliable, and secure interaction for MG in the real WAN environment, we utilized DDS as a robust IoT protocol to transport IEC 61850 messages format for the scenario of MG as the utility grid FR support on the Internet. Hence, the contribution of this paper will be: The remainder of this paper is organized as follows: Section 2 provides the MG structure and control methods to act as an FR-support of utility grid and presents its set of variables of interest. Section 3 investigates the communication model and services of our scenario based on IEC 61850 and maps onto the DDS protocol to obtain communication in the Internet infrastructure. Section 4 is devoted to the experimental setup and its results to prove the effectiveness of our proposed model. Ultimately, this paper ceases in Section 5 as a conclusion.

Power System FR with the MG Presence
Managing reliability and stability of the power system make it necessary to keep frequency in a defined boundary. Equilibrium between active power generation and consumption provides FR. Frequency control in the power grid has been divided into three levels: primary, secondary, and tertiary control. Primary control is automatic and responded to in seconds. The elements cooperating in this level of frequency control are synchronous generators (SG). Traditionally, governors in SG apply the droop control method to accommodate primary control. Furthermore, large scale RES power plants (LRES) and ESS (LESS) equipped with smart inverters as well as responsive loads (RL) through demand response scheduling can take part in this fast required response control level. Secondary control that is called load frequency control (LFC), which is automatic and centralized as the main function of automatic generation control (AGC), should be responded to in a few seconds to minutes. Secondary control, implemented by the control loop, compensates deviations of frequency beyond the coverage of primary control level and determines the participation rate of each generation unit in this process. Active power tie-line deviation is added as a tertiary control level to the secondary level since the power grid is an integrated multi-area. The tertiary level is manual and should be controlled in minutes. Active power deviation in power system with high penetration of MG, LRES, and LESS can be calculated as follows [35,36].
where; ∆P SG : deviation in power generated by SG; ∆P LESS : deviation in power delivered or consumed by large scale ESS; ∆P LRES : deviation in power generated by large sclae RES; ∆P RL : deviation in the power demanded by RL; ∆P MG : deviation in power delivered or consumed by MG; ∆P d : disturbances in loads power demand; H: system inertia provided by synchronous generators; D: power system damping coefficient.
MG in our proposal will be a collection of micro-SG (mSG), small scale ESS, and RES equipped with smart inverters, noncritical loads can participate in demand response known as RL, and critical loads, which should be served continuously. Therefore, MG's output power, which can be positive in generative mode of MG and negative in consumer mode, w.r.t. its elements states is calculated as shown in (2).  Figure 1 visualizes the basic frequency control levels in the power system with the MG presence. The TSO monitors the frequency of grid and determines the share of each generation unit in the FR. R is the droop control characteristic of each generation unit. K i (s), which is the transfer function of the secondary control level for each participant, with assistance from α i , β i , γ i , ξ i , and θ i coefficients show the participation factor of each unit in frequency control. Regarding this definition, we can estimate each generation unit portion shown below: where; i: number of actors in the FR; α: participation factor of ESS in FR; β: participation factor of SG in FR; γ: participation factor of LRES in FR; ξ: participation factor of RL in FR; θ: participation factor of MG in FR.
Equations (3) to (7) describes the amount of active power that should be delivered or consumed by ESS, SG, LRES, RL, and MG, respectively to contribute to LFC while the number of participants in this schema is according to Equation (9). Equation (8) shows the necessity of equilibrium in power consumption and generation in the power system. Table 2 shows time constraints in frequency control based on the time of processing T p and resources participation detection T d according to the European network of transmission system operators (ENTSO) regulations [37].

MG Structure and Control Methodologies
According to the initial objectives of MG advent in version 2003 of the IEEE 1547 standard, the MG works in grid-connected mode and supplies consumers by the grid. However, in the case of any failure, which could result in deviation of frequency and the voltage of the main grid, MG should disconnect from the grid and supply whole or part of loads autonomously by the ESS, which has been charged by RES during the grid-connected mode [38]. In the 2018 version of IEEE 1547 standard, this strategy improved by adding voltage and frequency ride-through capabilities to the RES interfaces in abnormal situations during the grid-connected mode [39,40]. These characteristics will be served by smart inverters. In addition to their typical responsibility of converting resources output power to grid desired level, smart inverters control voltage and frequency of sources while adjusting to grid set points. Applying this enhancement in the MG control methodology accelerates the pace of MG integration to the utility grid as a service provider. As can be seen in Figure 2, MG control methodology contains four levels: local, secondary, central/emergency, and global control levels [41]. While local control level ensures autonomous stability of each MG's power source through voltage and current internal loops, the secondary control level using communication adjusts power sources voltage and frequency in response to any fluctuation in loads or power sources. Smart inverters serve local and secondary control levels of MG. Central/emergency control level determines islanded or grid-connected mode of MG by implementing EMS and protection schemas in both normal situation and contingencies. The MG central controller (MGCC) represents central/emergency control level of MG. The MGCC interacts with a higher level of power system, where global control level is located. Global control level, which can be considered as DMS/DSO or power distribution system's EMS, is responsible for coordinating the performance of MG with neighbor MGs and the utility grid. To contribute in FR support of the main grid scenario, MG should provide stable frequency and voltage and a specified amount of output power according to the grid requirements. This statement means each MGCC receives set points from the DSO to adjust its actions in the grid-connected mode [14]. The IEC 61850--90--7 provides the communication framework for MG, which interacts with its supervisory level and considers smart inverters deployment for the RES to manage their intermittency and local and secondary control level implementation. Furthermore, the IEC 61850--90--7 determine predefined control modes for the DER and ESS. These modes can be addressed as the central level control. These schedules will result in a reduction in communication bandwidth requirements. In Algorithm 1, we determine which variables of interest to communicate with for the FR support of the utility grid by the MG. Additionally, since MG can import or export power to the main grid at the PCC, MG can communicate with DSO and participate in the FR provision for the main grid through the MGCC based on Algorithm 1. To control frequency deviation, TSO first sends a request for active power adjustment to regulate frequency and determines set points for all parties that can assist FR. Furthermore, DSO calculates the desired power level of each available unit according to their predefined cost functions and requests those set points. The MGCC receives commands, estimates which predefined schedule is suitable for DER, and issues related commands. This estimation is taken from information received by DSO and MGCC through reports sent by DER according to DS93 function of IEC 61850-90--7. MGCC by analyzing received information issues different functions of IEC 61850 to DER, ESS, and RL including: connecting to or disconnecting from the grid (Function INV1), changing their desired level of active power generation (INV2), and utilizing ESS (INV4).   IEC 61850 is a communication standard that provides an infrastructure for information exchange in the power grid by definition of information model. In addition to this benefit, the IEC 61850 also supplies an assortment of services for interaction of model elements and mapping services to the specific communication protocol [42]. Therefore, the configuration of robust communication, which assures MG as FR supporter, requires a comprehensive information model based on predefined control methodology of grid which is introduced in Section 2 and displayed in Figure 1. Physical devices, called intelligent electronic devices (IED), include the functions in IEC 61850. Furthermore, logical nodes demonstrate each function of IED and include several data objects, which delineate information interests of each IED's logical nodes. Therefore, in this paper, we consider battery management system (BMS) and DER controllers as IED and servers of IEC 61850 interacting with the MGCC. Since this interaction happens in LAN, they use manufacturing message specification (MMS) protocol and all messages are according to IEC 61850-8-1. In this communication framework, we assume that MGCC has its own logical nodes and interacts with DSO over the WAN. The MGCC and DSO apply IEC 61850-8-2 message format and DDS protocol for communication through the Internet in a WAN environment. Since the implementation of the scenario utilizes the functions in part 90-7 of the standard, logical nodes of those functions should be determined. Figure 3 shows a sequence diagram of the proposed scenario and Table 3 depicts logical nodes of this diagram actors with the sequences of interactions and service requirements. Table 2 also delineates the required set points for each step of communication.

Service Mapping Requirements Based on DDS
The information model and service requirements of our FR scenario needs an infrastructure that ensures real-time, reliable, and secure communication in the WAN. The DDS middleware, standardized by the object management group (OMG), provides these characteristics by the publish-subscribe message pattern. The DDS derives a benefit that participants do not require to know each other from applying discovery methods, including data-centric publisher-subscriber or real-time publisher-subscriber. Therefore, the DDS is a brokerless information exchange protocol without the risk of bottleneck failure. Data-centric publisher-subscriber includes the domain in which all communication entities are placed and interact. Domain participants, including Data Readers, Data Writers, Publishers, and Subscribers gain access to data, based on domain topic and type. As shown in Table 3, the implementation of the FR scenario requires the mapping IEC 61850 services on the DDS protocol. In our scenario, topics are the requests, responses, and reports of IEC 61850 services. Each Data Writer and Data Reader is dedicated to a special topic and is responsible for the marshaling and demarshaling of information. To satisfy interoperability, information exchange based on IEC 61850-8-2 message format utilizes XML parser for marshaling and demarshaling information. Publisher and Subscriber can interact with Data Writer and Reader respectively, corresponding with several topics. The DDS offers 22 levels of QoS that guarantees the quality of each domain entity interaction, even with UDP as a transport layer. It also supports TCP transport layer [43]. In step one, each entity of our proposal reports its condition to the supervisory levels. This information is time-dependent and should be updated; therefore, QoS should set Deadline and History for Data Writers and Data Readers. Other steps have a request-response pattern, requiring reliability in information exchange, then QoS of Data Writer and Reader of those steps set to Reliable mode. Interaction in WAN is on the Internet to avoid the necessity of providing the dedicated communication infrastructure. Therefore, it is vital to make this communication secure against illegitimate publications, disguised subscriptions, data theft, replay attack, and unauthorized access. To address this problem in this paper, we applied three service plugin interfaces, including authentication, access control, and cryptographic implemented by the DDS security model [44]. Figure 4 provides the experimental setup for the evaluation of the time constraint specification requirements of FR. As can be seen in this figure, LAN, the communication structure inside MG, is implemented in the LAN of the next-generation network (NGN) laboratory in Sejong university. Additionally, WAN, the Internet connection between the NGN laboratory and the Sejong University server room, accommodated by the Korean Telecommunication (KT) line. To develop the server and client of IEC 61850, we use MMS-Lite API which is an IEC 61850 data model and protocol stack implementation by systems integration specialists company (SISCO) with C language. The server of the IEC 61850 is a raspberry-pi and considered as the DER controller that connects to the client of the IEC 61850 which is MGCC. The DDS server acts on the laptop computer, which the MGCC is implemented on. Table 4 depicts our experimental network equipment and their software and hardware infrastructures in detail. It is noted that in the experimental setup arrangement we postpone TSO interaction with DSO, since it has the same arrangement of interaction between DSO and MGCC based on DDS protocol. To apply its effect on the latency of interaction in our scenario, we added one step request and reply between DSO and MGCC in our case studies. Additionally, the number of hops between DSO and MGCC is 11.

Experimental Results Investigation
To evaluate the scenario of MG as an FR supporter, we define two cases to examine latency, reliabiliy, and security performance. In Case I, the MGCC power output follows the DSO requirements since the interaction is just on the DDS protocol side of our experimental setup. In Case I I, the MGCC power output does not meet DSO requirements. Therefore, after using its optimization algorithm it issues commands to DER controller or BMS to justify their power output. Interaction in this case study is on both DDS and MMS protocol sides. Latency calculation for each case study is calculated as below: where; T 1 : time when DSO receives FR request from TSO; According to the mentioned experimental setup, we should consider delay time related to TSO interactions that are hidden in the DSO interactions. We also set T 6 , smart inverter ramp rate, to zero since FR scenario needs immediate action [18]. Both case studies are tested for 1,000,000 times on the public Internet with different interval times between each test to examine the reliability of our communication testbed. The size of request messages is 850 bytes and 514 bytes for response messages. As discussed before, reliability is necessary for the request and response messages; therefore, to compare the performance of the system, it is set to both BEST_EFFORT and RELIABLE for each participant as QoS. In each packet loss situation, there is no attention to ensure the message gets to the destination in BEST_EFFORT mode. On the contrary, in RELIABLE state, retransmission of the data happens. The deadline of QoS for assuring messages are delivered to the DSO and MGCC is about 1000 ms. Since the Internet facilitates WAN communication, DDS security mode for each type of QoS is also applied to provide a secure infrastructure and comprehensive performance comparison. Figure 5 depicts the average and maximum latency, interarrival time variance, and message loss rate of test results. Overall, this figure shows that in RELIABLE and secure communication, there is the cost of rising latency. As expected, in Case I, pure DDS system, latency is less than Case I I where the MGCC needs to negotiate with its elements on the MMS protocol. Looking at the detail, Figure 5a shows in the BEST_EFFORT mode, the average latencies of Case I are 4.6 and 4.8 ms for nonsecure and secure mode, respectively, which increased to 5.2 and 5.5 ms in RELIABLE mode. This increment in latency is due to the retransmission in any case of packet loss in RELIABLE mode from. Average latency in Case I I doubled to 10.6 and 10.7 ms in nonsecure mode for BEST_EFFORT and RELIABLE modes respectively and these values change to 10.8 and 11 ms for secure mode. However, Figure 5c clarifies the importance of setting the RELIABLE mode of QoS for DDS as it has the effect of eliminating packet loss of the system. Latency is not the only characteristic that is altered by reliable communication. Figure 5d reveals that interarrival time has also been affected by latency. Additionally noted, the variance of interarrival time in DSO is higher than MGCC. For example, in Case I I, while the variance of interarrival time of MGCC for BEST_EFFORT and RELIABLE modes are 0.845 and 21.93 ms in secure mode, these variables for DSO, racketed to 1.534 and 22.66 ms. Due to DSO being on the client side, it should wait to receive a response from the server-side; however, any loss in request message will result in an increasing loss rate of DSO. If we look at the performance of the system in secure and nonsecure modes, the security provision of interaction will not cause the failure of the system achievement. Even in the strict situation, i.e., in Case I I with Secure and RELIABLE mode, as shown in Figure 5b, the maximum latency is 1699 ms, which still abides by the time constraint requirement of FR.
In general, this experiment proves DDS can fulfill the time constraint requirements of the FR support scenario. However, we should not ignore the portion of processing time for optimization algorithms of both MGCC and DSO sides in the scenario success. Although the excellent latency performance of the pure DDS-based system in Case I validates the importance of scheduling MG performance to participate in AS provision of the utility grid, our results confirm the feasibility of MG deployment as an active element for the power grid frequency regulation.

Conclusions
From the introduction of smart inverters, the RESs, which are threads of power system stability due to their intermittence characteristics have a significant potential of turning into stability assistants. This unique plan can be viable through the supervision of MG equipped with ESS and RL when MG interacting with utility grid as an active element. Reliable control methodologies and especially communication infrastructure help this interaction happen. In this paper, we provided a scenario for MG as an FR support of utility grid. Regarding interoperability, the IEC 61850 information model and services applied along with the deployment of the DDS protocol for communication on the Internet to avoid employing dedicated infrastructure. Different situations of MG in response to main grid FR requirements implemented by deploying use cases of part 90-7 of IEC 61850.
As proof of our scenario, we provided an experimental setup and investigated time constraint representation in practice. The numerical results indicate DDS performance could meet the time constraints of FR. Although the excellent latency performance of the pure DDS-based system validates the importance of scheduling MG performance to participate in AS provision of the utility grid, our results confirm the feasibility of MG deployment as an active element of the power grid, which assists in contingency situations. As we implemented WAN on the Internet, RELIABLE applied as the QoS index in security mode. Although RELIABLE mode costs in longer delay, this parameter still stayed in the time boundaries that we require.

Conflicts of Interest:
The authors declare no conflict of interest.

Nomenclatures
The following Nomenclatures are used in this manuscript: