MicroGrid Renewable Integration Dispatch and Sizing (MiGRIDS) Analysis of Spinning and Regulating Reserve Options for Wind in an Alaskan Diesel Microgrid

St. Mary’s is a remote, isolated, Alaska microgrid that relies on diesel generation. Recently, a wind turbine generator (WTG) with a rated capacity just below system peak load was installed to help reduce diesel fuel consumption. With this WTG, St. Mary’s could significantly reduce their fuel consumption and run significant amounts of the year without any diesel generators (DG) online (“diesels-off”). However, wind power needs to be covered with spinning reserve capacity (SRC) and regulating reserves. Since there are no other sources, this needs to come from the DG, which does not allow them to turn offline and reduces the amount of potential fuel savings. The Alaska Village Electric Cooperative (AVEC) has invested in a “Grid Bridging System” (GBS), an energy storage system meant primarily to provide SRC and regulating reserves to allow the DG to go offline, scheduled to be installed in 2023. This paper uses the open source Python-based software tool Micro Grid Renewable Integration Dispatch and Sizing (MiGRIDS) to investigate the benefits and impacts of several methods to reduce or replace SRC and regulating requirements on the DG. These include optimizing SRC requirements, adding a GBS to the system, using short-term wind forecasting and using a fast acting controllable thermal energy storage system (TESS). The GBS resulted in the greatest fuel savings. Optimizing SRC and short-term wind forecasting resulted in a small decrease in fuel consumption on their own, but in combination with a GBS resulted in a much greater reduction. Finally, the TESS was able to regulate a poorly controllable WTG so that it behaved similar to a highly controllable WTG, at some expense to wind import.


I. INTRODUCTION
St. Mary's and Mountain Village, located in the eastern region of Alaska, are two interconnected villages with a combined population of 1,221 [1] whose power systems are operated by the Alaska Village Electric Cooperative The associate editor coordinating the review of this manuscript and approving it for publication was Pratyasa Bhui.
(AVEC).Their combined electric peak and minimum loads are 1,000 kW and 400 kW respectively.As with most other remote microgrids, diesel generators (DG) have historically been the main sources of power.In January 2019, a 900 kW wind turbine generator (WTG) was installed in St. Mary's, providing enough power to potentially allow up to 100% power penetration for significant portions of the year [2].However, WTG generation needs to be covered with spinning reserve capacity (SRC) and regulating reserves.Without other sources, this needs to provided by the DG, which does not allow them to turn offline (''diesel-off'' operation) and reduces the amount of potential fuel savings.The community goals include a reduction in the reliance on DGs while maintaining or improving power quality.Towards this goal, AVEC has installed the WTG (2018), built the intertie connecting St. Mary's and Mountain Village (2019) and procured a specialized energy storage system that will be installed in 2023.The energy storage system has been called a Grid Bridging System (GBS) and is meant primarily to provide SRC and regulating reserves to allow the DG to go offline [3], [4].The St. Mary's GBS will utilize a lithium iron phosphate (LFP) battery.
This paper investigates several methods to incorporate a greater amount of the wind energy while maintaining or improving power quality by ensuring sufficient SRC and regulating capacity including: optimizing SRC requirements, adding a GBS to the system, using short-term wind forecasting and using a fast acting controllable thermal energy storage system (TESS).This paper uses the open source Python-based software tool MiGRIDS (Micro Grid Renewable Integration Dispatch and Sizing) to run detailed energy balance simulations at a 10 s time step to investigate these methods.
Previous transient analysis of the St. Mary's grid have modeled the GBS providing spinning reserves and regulation and showed that it can reduce voltage deviations and the frequency nadir of the system [5], [6] and that a WTG can be used to provide frequency regulation and damping support in wind-diesel microgrids [7], [8], [9].
There is a research focus on short-term wind forecasting that can predict several minutes into the future in order to help mitigate against the impacts of large ramps in wind power.The use of lidar, sodar and radar as well as advanced models using meteorological data have all be studied [10], [11].Specific forecasting techniques for microgrids have also been studied [10], [12] as well as simulations that incorporate wind forecasts into the microgrid control [13], [14].However, the focus is on complex microgrids that use the forecasts to optimally schedule multiple sources of generation and load.Most remote wind-diesel microgrids do not have complex controls and do not have multiple sources of generation and controllable loads that would benefit from day-ahead scheduling [3].Instead, they respond in real-time to renewable generation and load.This paper looks into the ability of short-term wind power forecasts to directly reduce spinning reserve requirements in those systems.
AVEC commonly uses a fast acting TESS to regulate wind energy in their communities.A techno-economic analysis in the literature has shown its viability to provide both frequency regulation and spinning reserves [15].

II. DISTRIBUTION SYSTEM
The village of St. Mary's, located on the Yukon River on the western edge of Alaska, is an islanded community power system with a peak load of 600 kW and minimum load of 150 kW.Historically, power has been provided by three diesel engines, a 499 kW Cummins QSX15, a 611 kW Caterpillar 3512, and a 908 kW Caterpillar 3508.In order to displace diesel generation, an EWT 900 kW Type IV (model DW54*900HH50) pitch-controlled WTG was installed in 2018, significantly increasing the instantaneous renewable energy penetration.In 2019, an intertie was installed connecting St. Mary's with the neighboring village of Mountain Village.Combined they have a peak load of 1000 kW and a minimum load of 400 kW.The DG and WTG at St. Mary's are the primary source of generation for both communities.The St. Mary's power system is shown in Figure 1 [5].

III. METHODOLOGY
Four cases are evaluated: The first case evaluates the amount of SRC required without any other changes to microgrid components or controls.The second case examines the impact of adding various capacity GBS.The third case looks at how a two-minute wind resource forecast with varying degrees of accuracy can be used to reduce SRC requirements.The fourth and final case examines the effect of a 250 kW TESS used as a dispatchable load to help regulate wind power.Cases 1, 3 and 4 were analyzed using a full year of data, while Case 2 was analyzed using the month of January.

A. CASE 1: SRC OPTIMIZATION
The standard practice in Alaska electric grids at the moment is to cover 100% of wind imports with SRC.This paper delves into various ways to provide SRC for wind energy, or to reduce the amount of spinning reserve required.This first step calculated how much spinning reserve backup is required without additional grid components or services.
While it takes around 30 seconds for the automatic switchgear to start a DG and bring it online, AVEC specifies that 2 minutes of spinning reserves are required in order to include a safety margin.Thus, the system needs to be able to handle any drop in wind within a 2 minute window without any additional DG coming online.
One year of measured wind speeds were used to calculate available wind power from the 900 kW EWT WTG.The measured wind speeds had a temporal resolution of 1 Sample/minute.A Langevin upsampling method was used to increase the resolution to 1 Sample/second.This data was binned into 10 kW bins.For each bin, the maximum drops in wind within a moving 2 minute window were calculated.The 99, 99.9, 99.99 and 100th percentiles were calculated from the results.The percentiles show what spinning reserve backup is required for wind in order to avoid system overloads for a certain percentage of the year.For example, 99.99% of the time the maximum drop in wind power within a two minute window will be less than the 99.99th percentile.So an SRC backup equal to the 99.99th percentile should avoid system overloads 99.99% of the time.This provides information on the trade off between maximizing reliability and minimizing SRC coverage, which is directly linked to costs and DG utilization.This analysis does not take into account drops in wind power as a result of maintenance, line outages, control issues or any reason other than wind resource availability.Thus, depending on the situation, there may be need for higher levels of spinning reserve coverage.

B. CASE 2: GRID BRIDGING SYSTEM
MIGRIDS includes several options for dispatching generating units.For the simulations presented in this paper, a dispatch curve was used to dispatch the DG.The curve identifies which combination of DG to run at which loading ranges, based on their fuel curves.When the loading on the DG goes outside of their specified range a timer is initiated after which a switch to a new combination will be initiated.
In addition to loading, MIGRIDS tracks whether there is sufficient SRC in the system.A lack of SRC will also initiate a timer after which a switch to a new combination of DG will be initiated.
A GBS is used to help integrate renewable generation onto a diesel microgrid.This is achieved by allowing smaller or no DG run online when there is sufficient wind power to cover the load.The GBS provides SRC, replacing or reducing this requirement on the DG and even allowing them to turn completely off if there is sufficient wind power.As discussed in the previous section, the GBS is required to have sufficient energy to provide 2 minutes at the specified SRC power to allow sufficient time to switch a DG online.The GBS also helps to delay diesel switching by charging or discharging to keep the online combination of DG within their operating range.If the load drops below the minimum loading of the online combination of DG, the GBS will charge to maintain it at the minimum loading till it is fully charged to delay switching to a smaller DG combination.If the loading rises above the maximum loading, the GBS will discharge to maintain it at the maximum loading until it no longer has sufficient energy to provide SRC to delay switching to a higher DG combination.This replicates the controls that AVEC intends to use at St. Mary's.
A range of GBS power and energy capacities were simulated to identify their relative benefits to the grid in terms of reduction in fuel consumption.Other impacts to system operation including DG switching and loading were also investigated.The simulations include GBS power capacities of 1000 kW, 500 kW, and 0 kW (to represent the system with no GBS).Each power capacity was simulated with sufficient energy storage system capable of maintaining rated output for 600 s, 3,600 s, and 14,000 s (10 minutes, 1 hour, and 4 hours).

C. CASE 3: WIND FORECASTING
Sufficient SRC is required on the system to cover a rise in load or drop in wind for at least 2 minutes to allow a DG to start and come online.While DG have the capability to supply SRC, it is desirable to reduce their utilization as much as possible to reduce fuel consumption and operation and maintenance (O&M) costs.Removing SRC requirements from DG can be achieved by providing it with other components (for example a GBS) or by reducing the requirement for SRC.One alternative to reducing the SRC requirement is to use short-term wind forecasting.In this case, we are interested in a 2 minute forecast.Forecasting methods are not perfect and provide certain confidence ranges with their prediction.For example, a P90 range would provide the maximum and minimum values that the wind is expected to stay within at a 90% confidence level.A certain level of uncertainty, and thus risk, is an inherent part of relying on forecasts.An informed trade-off must be made regarding reliability and cost.
For this analysis, short-term wind forecasts were modeled that predicted the minimum wind value in the following 2 minute period with varying degrees of accuracy.The forecasts were modeled by taking the actual minimum value and adding a normally distributed error with a specified standard deviation.Since forecasts were only used to predict a downturn in wind power, the forecasts were capped at the current wind power.Before running the simulations, appropriate SRC levels for different wind forecast accuracies were identified using the percentiles of actual wind being lower than forecast wind, binned by wind power, similar to what was done in the first case scenario that considered SRC Optimization.

D. CASE 4: DISPATCHABLE LOAD COORDINATION
In the previous cases, the WTG was modeled as being perfectly controllable up to the available wind power.This is actually a reasonably accurate assumption based on operating experience with the Type IV model of WTG that is currently installed at St. Mary's.However, older and smaller WTG (similar to many installed in other AVEC villages) typically have reduced controllability, which impacts their ability to be integrated into a diesel microgrid.This case investigates a hypothetical scenario where the WTG is poorly controllable and with no GBS or other form of energy storage to regulate it.Fast acting controllable loads can be used to provide wind regulation, and this case models a TESS being used in this way.TESS are energy storage devices with the ability to store energy in the form of heat for later use [16].TESS have been used in combination with WTGs and other renewable energy sources [17], [18].The use of TESS has been a solution for remote communities in Alaska [16], [19].
The WTG outputs were controlled as a pecentage of available wind power.Thus, without a change in control setpoint, the WTG output would fluctuate proportional to wind speed, even at a reduced power setpoint.WTG were assigned a controllability parameter that specified the rate at which they were able to change that percentage setpoint.Poorly controllable WTG had a low value and their output tended to fluctuate with the wind speed, despite attempting to hold a power setpoint.
A maximum desired ramp rate for the WTG output was specified.The TESS was controlled to maintain the DG within their maximum and minimum loading as well as compensate for fluctuations in the WTG output and maintain the WTG + TESS net output within the maximum allowed ramp rate.The TESS had a 50% charging setpoint which the WTG was controlled to maintain.Since the TESS was controlled to maintain limits on DG, wind power supplied to the TESS did not need to be covered with SRC.The TESS did not reduce the SRC requirements for wind power supplied to the grid.This implementation is modeled after installed TESS used by AVEC in their villages.

IV. RESULTS
A. CASE 1: SRC OPTIMIZATION Figure 2 shows the curves for the percentile maximum drops in wind power within a moving 2 minute window at different starting wind powers.The 100th percentile line (shown in purple) illustrates the largest drop in wind power observed over the entire year.This line represents the maximum SRC coverage that would have been needed to prevent all overloads as a result of a drop in wind.Every year is different, and the green line was added as an example of a rule curve that could be used to calculate the amount of spinning reserve required at each wind power.Results show 100% SRC coverage required for wind power is between 0 and 400 kW, and linearly decreasing to 70% coverage at 900 kW.Other rule curves could be selected as well, for example along the 99.99th percentile, if that is deemed to be sufficiently reliable.

B. CASE 2: GRID BRIDGING SYSTEM
Adding a GBS reduced the amount of SRC required from the DG.This allowed smaller, or at times no, DG to run online, allowing more wind to be imported, DG to run at a higher and more efficient loading and less diesel fuel consumption.Adding GBS to the system also resulted in a higher number of times the DG switched on and offline.This is because unlike DG, a GBS is energy limited and after excessive discharging a DG needs to be brought online to supplement it.
As can be seen in Figure 3, the power capacity of the GBS has a much greater impact on the system operation than its energy capacity (displayed here as rated duration in seconds).Increasing the energy capacity only slightly decreases system fuel consumption.GBS energy capacity does have a significant impact on generator switching and loading: decreasing switching and increasing loading.
The results from Figure 3 were for a fixed SRC coverage requirement for wind of 100%.As discussed in the SRC Optimization section and shown in Figure 2, a dynamic SRC coverage of wind based on wind power could be employed to avoid having to run 100% SRC coverage for wind at all times.Figures 4 and 5 compare generator fuel consumption and system overload for a fixed SRC of 100% (red line) and the variable SRC rule curve identified in the SRC Optimization section (blue line).Results are plotted for different GBS power capacities and for a rated duration of 3600 seconds.
As expected, the variable (lower) SRC requirement on wind power resulted in less fuel consumption.It also resulted in a slight increase in system overloads.Requiring less SRC naturally leads to more overloads as there is less backup against drops in wind power.Figure 5 also shows that adding a GBS to the system results in more overloads.With no GBS, the large and discrete nature of the DG capacities resulted in there often being significantly more SRC than required.With a GBS, the system can operate much closer to the minimum required SRC.Note that the amount of system overload shown is on the order of 1/1000th of a percent of total system generation and would not be of serious consequence.

C. CASE 3: WIND FORECASTING
In this case multiple levels of wind forecasting accuracy are tested by specifying the Root Mean Square Error (RMSE)  of the prediction: 0, 0.1, and 0.25.For each level of accuracy a different minimum SRC curve is utilized.This was  determined by plotting the 99th, 99.9th, 99.99th, and 100th percentiles of the cumulative distributions of over prediction of wind power for a two minute window across the whole time series.At an RMSE of 0.25, Figure 6, a linearly decreasing minimum SRC from 0.55 pu at 0 kW of wind output to 0.4 pu at 900 kW was selected to ensure 100% of wind overpredictions could be compensated with no overload.At an RMSE of 0.1, Figure 7, a linearly decreasing minimum SRC from 0.35 pu at 0 kW to 0.3 pu at 900 kW was used.For an RMSE of 0, the prediction is perfect and no SRC is necessary for the WTG.
Simulation results with and without a 1000 kW / 1000 kWh GBS for the different levels of wind forecasting accuracy can be seen in Figure 8.As RMSE increases (wind forecasting accuracy decreases) generator fuel consumption increases as well due to the more conservative SRC requirements.This impact is greater in simulations with a GBS than those without.
Fractional loading of the DG decreases as RSME increased because the simulations with more accurate predictions operated with lower SRC, allowing a lower loading on the DG.Again, a greater difference occured with a GBS compared to without.
Simulations without a GBS saw no diesel-off time or system overload, but for simulations with a GBS, diesel-off time and system overload each decrease as the RMSE increased.Lower RMSE allow lower SRC requirements which allow more diesel-off time.The lower SRC requirements did also result in some small increase in overloads.
The amount of generator switching underwent opposite trends with and without a GBS as RMSE increased.The higher SRC requirements impact the system with a GBS differently because the GBS is able to provide smaller increments of SRC than the DG.In the no GBS scenario, the high SRC requirement leads to more frequent switching between one and two generators, while the low SRC requirement allows a single generator to provide the necessary SRC more often.With a GBS, however, the system can go diesel-off more frequently with a lower SRC requirement, which also requires the DG to switch back online more often.

D. CASE 4: DISTPATCHABLE LOAD COORDINATION
In all previous cases, it has been assumed the WTG is controllable up to the instantaneously available wind power.Many smaller WTG are not.With large wind penetrations, a controllable load can be used to control ramp rates on the DG as well as prevent under loading of the DG.
A 250 kW dispatchable TESS was simulated using MiGRIDS in place of the 1 MW GBS system.The maximum control ramp rate of the WTG was varied to examine the effects of a TESS on systems with different levels of WTG controllability.This control ramp rate is defined as the rate by which the WTG is able to change its output in response to a change in its set point, given in per unit of available wind power per second (PU/s).Additionally, the wind farm and the TESS are controlled to attempt to limit the absolute ramp rate of their combined output below 0.1 kW/s.While there is sufficient wind, the TESS is charged at 50% of its rated capacity.When there is a significant deviation in the load or increase in available wind, the TESS reacts quickly to allow for a more gradual reaction by the DG and WTG.
Simulations were conducted with and without a TESS under three levels of controllability: 0.005 PU/s to test a system with low controllabilty, 0.025 PU/s for a system with  moderate controllability, and 0.1 PU/s for a system with high controllability.Figure 9 shows an eight hour segment of the simulation without a TESS and with a low controllability WTG and figure 10 shows the same time period with the addition of a TESS.The blue area represents power imported from the DG, the red area represents the power supplied by the WTG, the yellow area is the power consumed by the TESS, and the purple line is the available wind power.
The addition of a TESS reduces the variability of WTG imports to the grid, thereby reducing variability of the DG generation as well.Furthermore, a significant underloading TABLE 1.The 90 th , 99 th , 99.9 th percentiles of wind farm import ramp rates for each contorllablity and TESS scenario.TABLE 2. The 90 th , 99 th , 99.9 th percentiles of DG underloading for each contorllablity and TESS scenario.event slightly before 16:00 was avoided by using a TESS.There was a sharp increase in available wind, and due to the limited controllablity of the WTG output, the DG had to operate below its minimum optimal loading of 400 kW.With the TESS, however, it was able compensate for rapid swing in WTG output rather than forcing the DG to be underloaded.
Table 1 shows the net WTG and TESS import ramp rate that falls within the 90 th , 99 th , 99.9 th percentiles.The TESS significantly reduces ramp rates in systems with WTG that are difficult or slow to control.The TESS has less effect with WTG that have higher controllability.
The TESS also reduced underloading of the DG.Table 2 shows the amount of underloading within the 90 th , 99 th , 99.9 th percentiles.As with the import ramp rates, the addition of a TESS significantly reduces the amount of underloading for wind farms with low controllablity, while the effect is lessened in systems with high controllability.
While the WTG import ramp rates and DG underloading decrease with the presence of a TESS, the amount of wind energy imported to the grid does as well.The charging of the TESS with wind energy results in less wind being imported but it also results in less wind that needs to be curtailed, resulting in a higher utilization of wind.Table 3 shows the amount of wind energy imported, the TESS throughput, and the unutilized wind energy available for each scenario.The higher wind imports without a TESS come at the expense of higher ramp rates and underloading on the DG.

V. DISCUSSION
The goal of adding wind power to a diesel microgrid is to displace diesel fuel consumption.Wind is a variable resource and needs to be backed up with SRC and regulated with a firm power source to maintain the balance of power on the grid.Normally, these services would be provided by the DG.However, that limits the ability to turn DG offline and the possible fuel savings.This paper presented several methods to either replace or reduce the amount of SRC and regulation reserves required from the DG.
In the SRC Optimization Case, no additional components were added to the grid.The amount of SRC required to cover the WTG was investigated and the tradeoff between reliability and system cost was explored.A rule curve created for minimum SRC coverage as a function of wind power, allowing lower SRC coverage at higher wind powers.Using this optimized SRC rule curve, as opposed to a fixed value, resulted in a small amount of diesel fuel savings and did not result in system overloads.
When a GBS was added to the system, it replaced the DG in supplying much of the SRC.This allowed smaller (or at times no) DG to run online resulting in less fuel consumption, and higher DG loading and diesel-off time.Increasing the GBS power capacity had a much greater impact than increasing its energy capacity.This is because the bulk of the benefit of the GBS comes from providing SRC, for which it only needs enough energy to discharge for 2 minutes.Any additional energy capacity is effectively used to time-shift wind energy from periods with excess to periods where it can be used, providing benefit, but at a reduced rate.Adding a GBS also results in a very slight increase in system overloads, on the order of 1/1000th of a percent of total generation, as a result of the GBS allowing the system to operate closer to its SRC limits.
Short-term wind forecasting could be used to reduce the amount of SRC that is required.The accuracy of the forecast will impact the amount by which SRC can safely be reduced.As a result, a more accurate forecast means less SRC can be required which means less fuel consumption.Several wind forecast accuracies and corresponding SRC requirements were modeled.As expected, lower SRC requirements resulted in a lower diesel fuel consumption.Similar to results from the SRC Optimization and GBS sections, a reduced SRC requirement had a relatively much larger benefit with a GBS than without a GBS.Again, with a GBS, a very small amount of system overload was observed.
Finally, the use of a TESS to help regulate WTG that have low controllabilty was investigated.Previous cases have provided SRC to backup against a drop in wind power.The assumption has been that the WTG can perfectly hold a power setpoint up to its available wind power at that time.This is not always a good assumption, especially for older or smaller WTG.In those cases, a source of regulation is needed to balance out the fluctuations in wind power.GBS are an excellent source of regulation as they are fast responding and can charge and discharge at full rated power.DG can also provide regulation, but need to maintain a minimum loading.Depending on the size and controllability of the WTG, the DG may be at risk of being under, or even reverse, loaded.A fast responding controllable load can be used to regulate the WTG and prevent the DG from being underloaded.A TESS was modelled as a controllable load, similar to how AVEC uses them in some of their communities.WTG with different levels of controllability were modeled.Without a TESS, poorly controllable WTG resulted in high output ramp rates and underloading of the DG.With a TESS, poorly controllable WTG performed similarly to highly controllable WTG.With a TESS, less wind was imported to the grid and less wind was curtailed, with the difference used to charge the TESS.While a lower monetary value than electric sales, supplying heating loads is vital and valuable in Alaska.Without a GBS (or other form of energy storage) on the system to provide regulation, the amount of poorly controllable WTG that can be integrated into a diesel microgrid is much more limited than for highly controllable WTG.Adding a controllable load such as a TESS can allow a much higher penetration of poorly controllable WTG.

VI. CONCLUSION
Various methods of reducing or replacing SRC and regulating requirements on DG for the integration of wind energy were investigated in this paper.The analysis was performed using the open source Python-based software tool MiGRIDS.Replacing SRC requirements with a GBS resulted in the greatest fuel savings.Reducing SRC requirements resulted in small fuel savings on their own.However, when combined with a GBS resulted in much more substantial fuel savings.Finally, the ability of a controllable load to regulate poorly controllable WTG was demonstrated, allowing a higher wind power penetration.

FIGURE 1 .
FIGURE 1. Diagram of the St. Mary's Power System, copied with permission from [5].

FIGURE 2 .
FIGURE 2. Cumulative distribution of the maximum drop in available wind power within a 2 minute window.

FIGURE 3 .
FIGURE 3. Diesel fuel consumption (first plot), diesel-off time (second plot), DG switching (third plot) and diesel fuel consumption (fourth plot) for different GBS power capacities and rated durations, using a fixed SRC coverage.Results are for the month of January.

FIGURE 4 .
FIGURE 4. Diesel fuel consumption for different GBS power capacities and levels of SRC coverage for wind.Results are for the month of January.

FIGURE 5 .
FIGURE 5. System overloading for different GBS power capacities and levels of SRC coverage for wind.

FIGURE 6 .
FIGURE 6. Cumulative distribution of the wind power within a 2 minute window for a 0.25 RMSE of the wind prediction.Results are for a full year.

FIGURE 7 .
FIGURE 7. Cumulative distribution of the maximum drop in available wind power within a 2 minute window for a 0.1 RMSE of the wind prediction.

FIGURE 8 .
FIGURE 8. Diesel fuel consumption (first plot), (second plot), DG switching (third plot) and diesel fuel consumption (fourth plot) for different GBS power capacities and forecasting RMSE.A variable amount of SRC coverage was used for each value of RMSE.

FIGURE 9 .
FIGURE 9. Time series of power imported from the wind farm and DG along with the available wind power for a low controllablity scenario with no TESS.

FIGURE 10 .
FIGURE 10.Time series of power imported from the wind farm and DG along with the TESS consumption and available wind power for a low controllablity scenario with a 250 kW TESS.

TABLE 3 .
Wind energy statistics including the amount imported, the TESS throughput, and the amount curtailed.