The Concept of MG and Multi-MG Systems

According to the definition from the EU research projects (Hatziargyriou et al., 2006; Schwaegeti, 2009) MGs comprise low voltage (LV) distribution systems with DERs (micro-turbines, fuel cells, PV, etc.), together with storage derices (flywheels, capacitors, batteries and heat storages) and flexible loads (electrical, heat). Such systems can be operated in a non-autonomous mode if interconnected to the grid, or in an autonomous mode if disconnected from the main grid. If aggregated and controlled efficiently, MGs can proride distinct benefits.

It is important to note that there are three mandatory MG features: Local loads, local DERs and intelligent control. Environmental protection schemes in terms of carbon credit will also be an essential feature of MGs if the specific country promotes that provision (Hatziargyriou, 2014). Another important point about MGs is that they are built at a specific geographic location. Therefore, generation and load aggregation ideas like YPP are not MGs as their sources are located in diverse locations and they do not have control over them. Then, it is worth mentioning that, though MGs can be operated in autonomous or islanded mode when there are disturbances in the main grid, the majority of them will be operated in the grid-connected mode for most of the tune (Hatziargyriou, 2014). A longer period of islanded operation will require the MGs to have larger storage systems. Moreover, as grid-connection offers the opportunity of participation in the larger electricity markets, the main benefits of the MG concept are seen from the grid-connected operation. In some conceptions, especially for cold countries, a heat grid is also an important part of MGs, where heat from CHP plants can be utilized to supply local heat needs and, therefore, the overall generation efficiency can be improved (Hatziargyriou, 2014). Distributed storages (DSs) and distributed heat storages (DHSs) play the vital role in MGs of balancing out generation and demand, smoothing out the ripples and shifting the peak load to a period of lesser load density.

MGs can be of various grid sizes, and, generally, they range in 100 kW to 1-2 MW of power. At large sizes, an entire LV grid can be defined as a single MG. On the other hand, an MG can be an LV feeder or even an LV house, meeting the mandatory features. In Figure 2. an MG formed of an LV feeder is shown. It can be seen that there exists an MG central controller (MGCC), assigned for central optimization, aggr egation and control. In addition, there must be local controllers (LCs) for the local control of different DERs in the MG. MGCC and LCs have bidirectional communication (as denoted by the green dashed arrows) between them and these schemes vary with the assigned control strategy.

MGs in the hierarchical structure of an SG is illustrated in Figure 3, which primarily bounds the four levels of SG hierarchical structure for control and optimization. A concise and introductory description of each level and their interaction is given here. This overview and Figure 3 will be referenced when key concepts are explained in the ensuing sections.

The top level is the level of the transmission network operator (TNO), with big generation plants and associated markets such as the capacity market, day-ahead market and, intra-day market. TNO manages

A typical MG on LY feeder

Figure 2. A typical MG on LY feeder.

MGs in the hierarchical structure of an SG

Figure 3. MGs in the hierarchical structure of an SG.

and supervises the wide-area power transmission line through the transmission supervisory control and data acquisition (SCADA) system. The second level is formed with the distribution network operator (DNO) and distribution market operator (DMO). The DNO will possess the distribution SCADA system to acquire real-time states of the distribution system. The DMO will be a platform to host an electricity market in the TNO-DNO boundary. Following this is the third level, constituted with MG laterals with MGCCs. Finally, there is the level of local controllers (LCs) for DERs and bundled loads of building and industry. Importantly, Figure 3 depicts general schemes of bidirectional data flow (denoted by dashed anows of different colors) between different components like LC, MGCC, DNO, DMO and TNO. Dashed aixows of the same color mean that they serve data between the same pair of control or optimization level. For example, the green dashed arrows represent the data exchange between LCs and a corresponding MGCC in charge. The MGCC receives bids, forecasts and states from the LCs’ sensors. Then, the MGCC utilizes these data in an optimization algorithm to send control references back to the LCs and bids up to the DNO/DMO for market participation. Similarly, the data exchanges between MGCCs and DNO/DMO as well as between the DNO/DMO and TNO is shown. At each of those interfaces, some optimization is done as the data is exchanged.

It is worth mentioning that, in some propositions, a multi-MGs level is suggested, generally applied at higher voltage levels (i.e., medium voltage), implying the coordination of interconnected but separated MGs collaborating through a multi-MG controller (Hatziargyriou. 2014). Then, if the multi-MG optimization, where several MGs can also be optimized together, is included, then there will be fir e levels in the hierarchical structure of control and optimization for an SG. There, the multi-MG level would also have data exchange and optimization. In multi-MG optimization, different DGs and controllable loads will be aggiegated and controlled by an intermediate control level between MGs and DNO/DMO. This level will optimize the operation of the multi-MG network, under a real market environment (Hatziargyriou, 2014). This level can contribute in the tertiary control level of MGs (discussed in the next section). However. multi-MGs level for SG is not a generally accepted or common scheme, so, for the sake of simplicity, this level is not illustrated in Figure 3. For the same reason, this level is not explicitly mentioned further in the discussion below. However, if this level is operational in a zone of SG, its placement in the hierarchy, data exchange, optimization, and market participation is straightforward, considering another level of optimization between MGs and DNO/DMO.

Control, optimization and market mechanisms of a single MG unit

An MG works both as an aggregator of DERs and DR resources and as a controller of the resources that it encloses. It has a complex mechanism for control, resource optimization and market participation. Hie mechanism is summarised below.

Control hierarchy of MGs

As the mechanism of microgrid control is discussed with great detail in (Bidram et al., 2017; Bidram, Member and Davoudi. 2012; Hatziargyriou, 2014), a single MG has four levels in its own hierarchical control structure, as shown in Figure 4. Zero level control with constant active power and reactive power (PQ) mode control or constant voltage and frequency (VF) mode control is the fasted level of inner loop control. Primary level control with droop characteristics is the next level of fast controls that provide voltage and fr equency stability, load sharing and plug-and-play capability. In fact, droop characteristics of any generator are the linear relationships that exists between active power versus fr equency (P-Q droop) and reactive power versus voltage (Q-V droop) (Bidram et ah, 2012). Primary and zero level controls are implemented in the LCs of the MG where different grid following topologies for the LCs can be designed

Hierarchical control structure of an MG (concepts from (Bidram et al., 2017; Hatziargyriou, 2014))

Figure 4. Hierarchical control structure of an MG (concepts from (Bidram et al., 2017; Hatziargyriou, 2014)).

by a combination of droop characteristics with constant PQ or constant VF control. Secondary control is managed by the EMS of MG, which is at the MGCC in the case of the centralized scheme and formed by a combination of MGCC and LCs in the case of the distributed scheme, explained in next subsection. At the secondary control level, references are set by a vital optimization algorithm. The optimization has other objectives, however, for the secondary control, it sets references to compensate voltage and frequency deviations by balancing demand and generation in a grid-connected or islanded market situation. This optimization keeps running in the time resolution of 5-15 minutes and the output of this is used as the references of the primary and zero level control. Above the secondary control, the tertiary control levels are managed by both the DNO/DMO operators and the EMS of MG. The mechanism of the tertiary control also varies depending on the MGs’ mode of operation (grid-connected or islanded). When an MG is operating in a grid-connected way, the tertiary control references come from the market dispatch decisions. There exist several markets with several objectives and time horizon where data of bids, forecasts and states from many agents, including MGs, are used and dispatch schedules are made. Here, MG EMSs run optimization to bid optimally to different markets. The sequence of markets is discussed in a later subsection. However, in the case where MGs are mu in isolated mode, the tertiary control references are made solely from the MG EMSs by solving the related constrained optimization problem, briefed at the next subsection.

The optimization

The vital role of MGs is to resolve the conflicting interests of all the stakeholders. The EMS of an MG plays an important role in that resolution, being in the middle of two hierarchical levels—the DNO/DMO and LCs. Firstly, there are information and requests from the LCs of each component within an MG. These requests can be in the form of bids for generation, loads and controllable loads from each stakeholder. In some alternative schemes, instead of bids, information for optimization can also be forecasts of demands and generations (Hatziargyriou and Tsikalakis, 2005). Secondly, MGs have communications with the DNO/DMO. When an MG runs in grid-connected mode, it communicates to get utility requirements, the forecast of market prices and the larger market platforms to bid from MG side. If an MG runs in islanded mode, though it does not consider the grid-side market, it still forms grid quality supply (voltage and frequency) and balance power by dispatching its generation and load. Therefore, in both grid-connected mode and islanded mode, the EMS of an MG needs to resolve many conflicting interests of stakeholders by optimization in three axes of concern—economical, technical and environmental (Hatziargyriou, 2014). A few concerns of teclmical and environmental axes also have an effect on the economical axes. For example, greenhouse gas (GHG) emissions in the environmental axes incur emission costs and power losses in the teclmical axes cause power loss costs, both on the economical axes. Finally, the outputs of this very optimization in MG EMS level are communicated to the LCs of DERs to set control references of secondary level control, which has been discussed in the previous subsection. Furthermore, optimal bidding from the MG side is also an output of the optimization which are communicated to different markets. This information flow and the optimization functions of EMS for an MG are shown in Figure 5.

For this vital optimization, centralized or distributed, either of the control schemes is applied in MG EMSs (Su and Wang, 2012). As an example, in Figure 3, MG 1 is built with a centralized scheme and MG 2 is built with a distributed scheme. These schemes are described below. The centralized scheme

In the centralized control scheme, as shown for MG 1 of Figure 3. the EMS of the MG is placed solely in a central controller, as detailed further in (Cliaouachi et al., 2013; Daniel E. Olivares and Claudio A., 2014; Su and Wang, 2012). That central controller has bi-directional communication with all the LCs of the MG and DNO/DMO level. Here, the LCs have no autonomy and intelligence. The LCs are obliged to follow the outcome references from the central controller’s optimization. This scheme is simple to implement, easy to maintain and requires a lower cost. However, having a single controller, the EMS has a huge computational burden, requires high-bandwidth links and possesses a single point of failure. This

Optimization m three axes at the MG EMS between DNO/DMO and LCs

Figure 5. Optimization m three axes at the MG EMS between DNO/DMO and LCs.

scheme is not easy to expand and has very limited plug-and-play functionality, which is required in order to deploy PHEVs and for the ever-growing installation of DERs (Su and Wang, 2012). The distributed scheme

In the distributed scheme, as shown for the MG 2 of Figure 3, each MG component is regulated by one or more LCs, rather than being governed by a central master controller. Every LC communicates with other local controllers and possesses the intelligence to make operational decisions on their own. This scheme accommodates more autonomy and intelligence for the individual stakeholders. However, it is worth noting that a central controller and EMS might still be necessary for this scheme, which exchanges energy price information, utility requirements and bids with the DNO/DMO. That central controller may take over the control of the local controller in the event of grid contingencies or equipment failure. In normal operations, the control is given back to the LCs and they operate autonomously using primary control loops, like frequency droop control, Volt-VAR control and local information (Bidram et al., 2017; Huang et al., 2011: Kantamueni et al., 2015: Logenthirau et al., 2012; Su and Wang, 2012). This scheme has better plug-and-play capabilities, system expansion ability and a lower computational burden in each controller. This scheme also avoids a single point of failure. However, this scheme requires a larger investment in the communication facility and more time for convergence between local controllers (Huang et al., 2011; Su and Wang, 2012).

Electrical market structure and MG’s participation

In a concise description, an MG’s two main functions are aggregation and control. As it aggregates DERs and DR resources, its position and participation mechanism in the electrical market is an area of great interest. The decarbonized grid of the future will run with the same three levels of regulations in a conventional power system. As the regulations and markets in power system operation are depicted in Figure 6, the primary regulation is the automatic energy variation of a unit by local proportional control to stabilize frequency. It is mandatory for every generator and not remunerated. Primary regulation is the fastest regulation and its response time is in seconds. As can be seen in Figure 4, the zero and primary level of control in the hierarchical control of MGs takes care of this primary regulation.

Errors still exist after proportional primary regulation. The secondary regulation, which is also called automatic generation control (AGC), is there to minimize remaining errors. It also returns tie-line powers

Regulations and markets in power system operating philosophy

Figure 6. Regulations and markets in power system operating philosophy.

of an area to their scheduled values. The secondary level of MGs’ hierarchical control does a similar job, as can be seen in Figure 4. However, from the main grid’s operating philosophy, AGC is optional for generators and it is allotted to generators after they bid in the capacity/reserve market. Selected generators need to adjust their generation within 5 minutes. This market is highly paid by the TNO and the penalty of failure is also high, so only technically robust generators bid here. The acceptance of aggregated renewables to bid in this market is just beginning, so whether MGs can participate in this market depends on the market regulation of a specific country. However, in the islanded mode of operation, MGs must maintain this secondaiy regulation for the network it encircles, through its secondary level of control.

Between secondaiy' and tertiary regulations, two wholesale electricity markets—the day-ahead and intraday market are operated. These are energy markets, so generators sell and consumers buy. TNO only sees who fails and who should pay penalties. Most generators initially bid in the day-ahead market. If a bid of a generator gets accepted and then there are some problems due to forecasting errors or plant failures, that generator buys the lacking generation from other generators in the intra-day market—a market that runs every 15 min in a day. Also, any generator may decide to bid only in the intra-day market, without taking part in the day-ahead market, if this strategy suits the generator. Now MGs, of course, can take part in these markets, optimizing their aggregated DERs and DR resources.

After all of these, there still remains imbalances and the capacity reserves also require refills. There comes the tertiary regulation. The flexibility' market to serve the tertiary regulation is the solution of still existing imbalances and restoration of secondaiy' regulation reserves. Flexibility market is also a market with bids where flexibility/shift from generators already assigned schedule are sold again. In this market, the response time varies from minutes to hour. With the increasing inflexibility due to the spread of RESs, it is expected that flexibility markets will be formed to run every 15 minutes. It is contrived that new flexibility market will be formed in the TNO-DNO interface where DERs, DR applications and, consequently, the MGs as aggregators can play a vital role.

Now, the day-ahead and intra-day energy market are called financial markets and the flexibility market for tertiary regulation is called a service market. As the cash flow interaction of these markets is shown in Figure 7, it can be seen that DERs, including micro-generators and storage units, will participate in wholesale financial markets mainly with the aid of MGs’ aggregation. In some cases, VPPs can be formed for aggregation if the case requires, but it would not have controllability. Then, to take part in the ancillary service market, DERs and DRs have both options—through aggregation by MGs or individually with some distributed agent-based method having blockchain-based format for privacy, as in (Guan et al., 2018; Pop et ah, 2018). In any of these markets, when the participation is through MG aggregation, there will be two modes of operation: Grid-connected or islanded and the

Cash flow interaction of financial and service markets from the distribution side (concept from (Hatziargynou

Figure 7. Cash flow interaction of financial and service markets from the distribution side (concept from (Hatziargynou,


optimization in three axes, as stated in subsection before. The optimization there can also be formed as an internal market of the MG itself. Operation of MGs in markets is described in detail in (Celli et ah, 2005; Hatziargyriou, 2014; Hatziargynou et ah, 2005; Liu et ah, 2016; Logenthiran et ah, 2008; Sinha et ah, 2008).

< Prev   CONTENTS   Source   Next >