Multi-microgrid impact assessment


By Julija Vasiljevska, Joao A. Peças Lopes and Manuel A. Matos


Fig 1: Active Network Management in period
of typical electricity market prices.

Microgeneration technologies have progressed from isolated operation, when used in remote areas, to a grid interconnected mode that may involve its active management under the concept of a Microgrid. A microgrid can be defined as an active cell within the Low Voltage (LV) network, according to the European concept of microgrid, which consists of several microgrid sources, storage devices and controllable loads, having total installed capacity of a few kW up to a few hundreds of kW. Microgrid mostly operates interconnected to the main distribution grid, but also islanded, in case of external faults. The management and control of a microgrid involves a hierarchical control system architecture comprising the following three control levels:

  1. Local Microsource Controllers (MC) and Load Controllers (LC)
  2. Microgrid System Central Controller (MGCC)
  3. Central Autonomous Management Controller (CAMC).

Each MC follows requests from the central controller, when connected to the power grid, and performs local optimisation of the microgrid active and reactive power production, and fast load tracking following an islanding situation. LCs installed at the controllable loads provide load control capabilities following orders from the MGCC for load management. The MGCC is responsible for optimisation of the microgrid operation. It uses the market prices of electricity and gas and grid security concerns to determine the amount of power that the microgrid should draw from the distribution system, thus optimising the local production capabilities. Regulators are one of the players that need to address the benefits microgrid can bring to distribution networks in order to find out the right incentives to encourage Distribution System Operators (DNO) and microgrid owners to be involved in the deployment of the microgrid concept. Furthermore, potential conclusions can be drawn regarding the share of costs and benefits among the entities involved in the deployment of the microgrid concept. The development of microgrid solutions requires the adoption of individual microgrid optimisation procedures, it being possible to evaluate the overall benefits one can get from a microgrid, namely when operated in a real market environment. For this purpose, an active local management is adopted at each microgrid level, in terms of microgrid optimisation and load management performed through Demand Side Bidding (DSB).


Fig 2: Active Network Management in period
of high electricity market prices.

In this research several microgrid technologies have been considered to be present within a microgrid, such as Micro- Turbine (MT), micro Wind Turbine (WT) and Photovoltaics (PVs). In order to evaluate the benefits that can be obtained from the presence of microgrid, different study case scenarios are considered regarding different levels of microgrid installed capacity within the Microgrids and different electricity market prices. What is expected to take place is an increase of microgrid production in periods of high electricity market prices. The microgrid optimisation procedure – active network management – is performed at a Micro Grid Central Controller (MGCC), housed in the local MV/LV substation, requiring the execution of the following steps:

  1. Each unit within the microgrid bids for production for the next hour in n-minutes intervals, according to the electricity market prices, operating costs of the unit and the profit for the unit’s owner
  2. Each consumer within the microgrid bids for their load supply for the next hour in the same n-minutes intervals, where each bid reflects the amount of energy he is willing to pay for at that time interval taking into account the possibility of shifting to the next time interval, where the electricity price is lower, a certain percentage of its load (considered as “low priority” load)
  3. Solving the optimisation procedure by defining the microgrid units being committed and the consumers’ bids being accepted.

Fig 3: Medium-voltage rural network

The rest of the demand is served from the upstream network at real market prices. In this optimisation procedure the Renewable Energy Sources (RES) within the microgrid are not considered as competitive units, i.e. they are always dispatched once their primary energy resource is available. Therefore the only units that bid within the microgrid are the fuel-consuming units, defined as controllable units, since their production is in correlation with the electricity market prices. Moreover, we are assuming that all consumers within the microgrid are price-sensitive, i.e. they may respond to the high prices at the load peak period, by shifting 10% of their demand to the next time interval of lower electricity price.

Since the main idea of this work is the impact assessment that several Microgrids may have at upper network level (MV level), several scenarios are created regarding the Microgrid installed capacity at the LV grid, ranging from 10% to 30% of the peak power of the corresponding MV/LV substation. Since two independent analyses, regarding the type of the distribution network, rural and urban, take place, the percentage of microgrid technology regarding the total microgrid installed capacity is different for these two types of networks.

In the case of a rural network, most of the Microgrids include mainly RES (with 80% of PV and 20% of WT, from the total percentage of RES being installed), gradually increasing its installed capacity from 10% to 30%. Since the installation of MTs is not feasible in rural areas, once natural gas networks are not easily available, only 4 Microgrids comprise MTs, with installed capacity of 30kW each, with efficiency of 26% for burning a natural gas.


Fig 4: Minimum voltage profiles at ten worst
nodes due to voltage drop.

The fuel price is assumed to be 10 €ct/m3. Regarding the RES units, its generation depends on the availability of the primary energy resources (wind and sun). Typical wind speed and daily sun radiation data for Portugal have been used to define the generation levels for these sources. For the micro WT, an average capacity factor of 40% was assumed. The daily load consumption profiles, seen from the microgrid LV/MV transformer, with and without the microgrid active management are presented in the following figures for the case of maximum microgrid installed capacity, i.e. 30% of the peak power.

Figure 1 presents the outcome of the local active network management for a given load scenario in periods of typical electricity market prices, whereas Figure 2 introduces additional value of the active network management in periods of high electricity market prices when adopting the local optimum management procedure described before. The area below the bottom curve indicates the amount of energy needed to be bought from the upstream network at open market prices. As expected, high electricity prices yields higher microgrid production, i.e. dispatch of the fuel-consuming units (MT in our case) and lower amount of energy to be bought from the upstream network.


Fig 5: Total active losses in a typical MV rural
network for a day of typical and high electricity
market prices.

After having locally optimised the operation of a single microgrid, the next step is moving towards higher network level. Namely, the idea was setting up several Microgrids and identifying the impact at MV level. Since the whole application is taking place in a real market environment, reflecting realistic market prices, several scenarios are created in respect to four evaluation alternatives, namely, without microgrid (alternative A), with microgrid and 10% microgrid installed capacity within the microgrid (alternative B), with microgrid and 20% microgrid installed capacity 3 within the microgrid (alternative C) and with microgrid and 30% microgrid installed capacity within the microgrid (alternative D).

Figure 3 shows a typical rural network, which was used in our study, with several Microgrids (described in the figure by a generator and a load). The ten worst nodes regarding the voltage drop are designated in black. Figure 4 presents the minimum voltage levels reached in the network after a load increase in every node during x years, assuming an annual load rate increase of 3%. The results presented in Figure 4 consider the case of 30% microgrid installed capacity within the microgrid of the local peak load, for a typical day of high electricity market prices, when the microgrid units’ production reaches it’s the higher value. The investment deferral time is defined by the period of time between two moments: a) when the minimum voltage level is reached in the most critical bus for a scenario with certain percentage of microgrid installed capacity within the microgrid b) regarding the one without microgrid.

Moreover, it is clearly evident that strategically located Microgrids may systematically reduce the distribution network losses, beneficial to the Distribution Utility (DU), as an entity responsible or mandated to keep losses at low levels. Nevertheless, this “technical” benefit of loss reduction resulting from the Microgrids presence can be translated into an economic benefit and considered as one of the criteria of the multi-attribute problem described in the next section. Figure 5 presents the total active losses for a typical day of high electricity market prices for the case of 30% microgrid installed regarding the microgrid peak load.


Fig 6: Typical Medium Voltage urban network

What can be observed from the above figure is a significant active power losses reduction at peak hour, regarding the case of not having Microgrids installed, reaching 19.7% of reduction for a day with typical electricity market prices and 24.6% in a day of high electricity prices, since the controllable generation units within each microgrid are capable of providing energy in more favourable economic conditions than buying it from the upstream network. Similar analyses are applied for a typical urban network, whereas in this case the line congestion levels define the criteria through which it is possible to identify the period of time that investment in reinforcements can be deferred.

Similarly, several Microgrids, with a total installed capacity limited to 20% of the microgrid peak load were considered for this study case. The mix of microgrid technologies in each microgrid, placed as depicted in Figure6, is 25% RES (mainly PVs, since it is an urban area) and 75% of controllable units, namely MT. The ten most congested lines are shown in bold.

What is evident from Figure 7 is a significant congestion level reduction at the peak hour in the ten most congested lines, especially for period of high electricity prices when the MT units are expected to be fully dispatched. The recorded values reach 7.9% congestion level reduction in the most congested line in respect to the case with no Microgrids being installed, for typical electricity prices, and reduction of 15.7% in the most congested line for a period of high electricity market prices. As expectable, a large active losses reduction has been achieved, reaching value of 23.7% at the peak hour, for a period of high electricity prices in respect of the case without Microgrids being installed. (Fig.8).


Fig 7: Congestion level at the ten most congested
lines at the year of investment

All the technical issues, in terms of voltage drop (rural networks), congestion level (urban networks) and total active losses, are addressed through simulation, namely power flow studies, defining the attributes of the multiattribute problem considered in the next section.

Since the problem is modelled as a multi-criteria, more accurately, multi-attribute problem, the attributes are explicitly defined, addressed in the previous section through simulation. Moreover, the criteria of the problem are defined through the attributes, recognising three main criteria in our case: total annualised cost for putting in place microgrid, investment deferral and active losses. Therefore, the installation cost is considered as an a priori cost, annualised, as well as the microgrid operation net cost, subject to careful examination with regards to avoiding duplications of cost or benefits. The cost of putting in place the microgrid, in terms of microgrid communication and control infrastructure that is essential for the coordinated control of the microgrid units in microgrid operation, is considered as installation cost. The multi-criteria analysis presented in this section are applied for the case of the rural MV network, having the possibility of application at the urban MV network as well.

Four main alternatives are considered from the DM perspective, namely no-microgrid (alternative A), with microgrid and 10% microgrid installed capacity within each microgrid (alternative B), with microgrid and 20% microgrid installed capacity within the each microgrid (alternative C) and with microgrid and 30% microgrid installed capacity within each microgrid (alternative D). Having concluded in the first part of the research, that market prices influence the microgrid production, two scenarios regarding the prices are created, corresponding to days with high and typical electricity market prices.

Furthermore, considering the investment deferral, two scenarios regarding the load growth are added. Therefore, four scenarios have been created: typical electricity prices and 3% load growth (scenario 1), typical electricity prices and 4% load growth (scenario 2), high electricity prices and 3% load growth (scenario 3), and high electricity prices and 4% load growth (scenario 4). The microgrid installation cost includes the cost of the MGCC in each of the microgrids, the cost of the MC for each type of source considered within each microgrid, as well as the LC for each of the consumers within each microgrid being part of the microgrid.

The cost for putting in place the MGCC is covered by the DNO. Indicative values for the cost of each local controller were used, namely €300 for each micro wind generator and PVs local controller, €500 for each MGCC local controller and U100 for each LV load local controller. The cost of the microgrid CC is assumed to be €500. The microgrid net operation cost comprises the fuel cost for the fuel-consuming units within each microgrid, being dispatched at the peak hour.


Fig 8: Total active losses in a typical MV urban network for
a day of typical and high electricity market prices.

This article deals with the evaluation of potential costs and benefits by deployment of the microgrid concept using multicriteria decision aid methods. Within the first part of the work, the potential technical and economic benefits have been assessed using optimisation techniques at a microgrid level, under different scenarios of microgrid installed capacity and electricity market prices. Furthermore, extending the microgrid concept into microgrid concept, potential cost and benefits coming out of the microgrid deployment have been evaluated at the MV level, following different load growth scenarios and electricity market prices. The results lead to more favourable microgrid deployment in periods of high electricity prices. The assessment is made within a multicriteria framework, using different decision aid techniques. Starting with the trade-off analysis, we have shown how different trade-offs lead normally to different evaluations/ rankings in each scenario. In our analysis, we have used typical MV rural network, leading to some conclusions that the microgrid concept deployment is not that favourable, unless we give high importance to the benefits that can come out of the investment deferral. Moreover, we have assumed an equal share of microgrid installation costs, in terms of communication and control infrastructure cost, between the consumers and the DNO. If the consumers within the microgrid recognise and value the overall benefits coming out, thereby having higher share of microgrid installation cost, it would assumingly lead to different results in the multi-criteria framework. What can be captured from the studies, performed as a generalised conclusion, is that the problem is rather case sensitive and network dependent. Nevertheless, we may conclude that the development of the microgrid solution becomes interesting in high energy market price scenarios within a specific range of trade-offs.