Wind-Diesel Hybrid System Testing at the Alaska Center for Energy And

ENERGY EFFICIENCY EVALUATIONS

Wind-Diesel Hybrid System Testing at the Alaska Center for Energy and Power A Review of Project Activities under the Denali Commission Emerging Energy Technology Grant Award #01233-00

Dennis Witmer 10/31/2013

Final Report Introduction As the cost of diesel fuel has risen sharply in the past few years, the incentive to replace expensive diesel electric power generation with less costly alternatives has also increased. Many remote Alaska communities have excellent wind resources, but the cost of installing utility-scale wind turbines in these locations is high. Even more challenging is the stochastic (random) nature of wind energy, which makes it difficult to provide utility-grade electricity from this resource.

High-penetration wind-diesel hybrid systems that use energy storage are one option, but they require several components: diesel electric generators (DEGs), wind turbines, batteries or flywheels for energy storage, and a control system. Inverters—devices that convert alternating current (AC) to direct current (DC)—are usually required for transferring energy in and out of batteries as well as converting “wild AC” from wind turbines and flywheels into utility-grade 60-cycle AC. Often the inverter is at the center of the control strategy, allowing the system to collect excess energy when available (from high wind events), then store that energy (in the battery or flywheel) and release it later. Since fuel savings can be maximized if DEGs are off during wind events, in an ideal system the diesel would operate during calm periods but be turned off when sufficient wind energy is available. Conventional utilities depend on rotating generators to provide AC power for both energy and reactive power support, tasks that must be done by the inverter in the new hybrid systems.

In 2009, the Denali Commission, an independent federal agency in Alaska, released a public solicitation for proposals to be funded by its Emerging Energy Technology Grant program (EETG). The EETG targeted (1) research, development, or demonstration projects designed to test new energy technologies or methods of conserving energy or improve an existing energy technology, and (2) applied research projects that employ energy technology with a reasonable expectation that the technology will be commercially viable in Alaska in not more than five years. The Denali Commission selected “Evaluating NW100B Inverter to Support Diesel-Off Operation in Alaskan Wind-Diesel Systems,” a proposal submitted by the Alaska Center for Energy and Power (ACEP), as one its project. This report is a review of that project and includes an overview of the technology, summarizes project activities, and provides results and next steps.

Background Today, much of Alaska remains wild and undeveloped. Approximately 175 small communities are scattered across the vast landscape, where electric power is provided by DEGS, each serving a small, local power gird connected to a few dozen to a thousand houses. There are only a few roads and electrical transmission lines connecting the largest communities.

The absence of roads, electrical grids, and pipelines means that much of rural Alaska experiences very high energy costs. While DEGs are efficient (35% and higher for new generators), diesel fuel is

1 considerably more expensive than the natural gas and coal available for electricity generation along the Alaska Railbelt1.

Diesel fuel sells for about $3.50 per gallon at the pump along the Railbelt at current world oil prices. This fuel must be then shipped to rural communities by barge, which adds approximately $1 per gallon to the delivered cost of fuel.2 At $4.50 a gallon and 14 kWh per gallon3, fuel costs $0.32 per kWh, much higher than the cost of several cents per kWh for natural gas or coal, which is used in power plants along the Railbelt.

Fuel costs along with capital costs for the power plant, operations and maintenance costs, local utility line costs, and administrative and overhead costs add up to a total cost of electricity between $0.40 and $1.20 per kWh to the customer, as reported to the Regulatory Commission of Alaska under the Power Cost Equalization (PCE) program.4

The high cost of diesel fuel has led to increased interest in the use of local energy sources, in particular renewable energy sources. Because many rural Alaska communities have significant wind resources, there has been substantial interest in developing wind projects over the past few years. Alaska has been implementing utility-scale wind power as an energy source for rural communities for the past 15 years, beginning with the installation of wind turbines in Kotzebue and Wales. Since that time, wind turbines have been installed in more than 20 rural communities, and several installations have been made along the Railbelt.

1 A term referring to the broad geographic area served by the Alaska Railroad from the Kenai Peninsula to Fairbanks, which is also connected by the state’s largest electricity grid. 2 In current prices, natural gas is currently sold in bulk at $3.50 per MMBTU, but the same chemical energy costs about $32.60 when delivered to a rural village as diesel fuel at $4.50 per gallon. 3 Diesel fuel contains about 39 kWh of chemical energy, so 14 kWh per gallon represents a DEG efficiency of about 36%, typical of well-maintained modern diesel generators. The kWh per gallon statistic is reported by utilities to the PCE program and is one simple metric for utilities. 4 The PCE program is intended to reduce residential power bills by providing a subsidized rate for the first 500 kWh per month of an electric bill, with payments made directly to the utility to compensate for lost revenue, but it requires utilities to provide documentation to justify payments. http://www.akenergyauthority.org/PDF%20files/pcereports/fy12statisticalrptcomt.pdf.

Figure 1: Low-penetration wind system, where the wind power remains well below the load. System stability is not threatened, and fuel savings are realized. Data represents a one-day wind event (24 hours) presented in seconds (86,400 seconds/day).

Figure 2: High wind event, simulated for modeling purposes. Note that the wind output sometimes rises above the load (suggesting the possibility of a diesel-off mode) but there are also events during which the wind falls rapidly from above to below the load. Event simulated from wind event shown in Figure 1, above.

The challenge with wind (and other renewable sources such as solar) is that the energy is intermittent, available only when the wind blows, and not necessarily when people want it. In remote Alaska communities, the stochastic nature of wind is a significant problem. The only other source of power is the DEG, which must provide power when the wind is not available.

The simplest wind-diesel hybrid system uses an appropriately sized wind turbine to provide a portion of the energy needed, similar to the larger utility wind farms along the Railbelt. If the wind energy contribution is relatively low (less than 50% of the local load), the diesel engine can act as the bridge between the wind and the load, providing the necessary power to make up the difference between the two and providing some diesel fuel savings as well. This type of system is referred to as a low- penetration system. The advantage of this type of system is its relative simplicity (although integrating wind turbines and diesel generators is not without its challenges), but the disadvantage is that relatively little wind energy can be harvested for use compared to the overall energy requirement, and the costs per installed kilowatt are often quite high compared to larger utility wind farms.

Some systems have attempted to incorporate additional wind turbines—referred to as medium- penetration systems (the wind output rises to equal or exceed the load)—but unfortunately the diesel engines must be kept running to cover the breaks in the wind. This means keeping the diesel engine on at some reasonable minimum load—most often between 20% and 40% (depending on engine size) of the maximum—to minimize maintenance issues and diverting the excess electricity to “dump loads” or curtailing the output from the wind turbines. Either way, less energy is delivered to the load from the wind turbine than it is capable of providing. The total diesel displacement at the power plant is greater than with a low-penetration system, but capital costs are higher. There may be some value for the heat energy provided, but this a lower-value use of the electricity. Stable operation of medium-penetration systems has been demonstrated in several locations in Alaska.

High-penetration systems are the most complex, but they also provide the potential for additional fuel savings by enabling a “diesel-off” mode of operation. The typical components of a high-penetration wind-diesel hybrid system are (1) a conventional diesel plant large enough to supply the entire village load when wind is not available, (2) wind turbines that are capable of providing significantly more power that the grid demands, (3) energy storage capability, and (4) a control system to manage the operation of each of the individual components. The energy storage system can contain capacitors, batteries, or flywheels, usually matched with an inverter. In addition, a synchronous condenser (essentially a rotating generator with no engine) can be used for frequency stabilization during diesel-off operation, although some systems have been proposed to eliminate the need for this component.

Of the components of a high-penetration system, the two that currently require the most attention are the energy storage system and control system. While it is possible to devise energy storage systems that use only utility-grade AC electricity (for example, pumped hydro can use AC electricity to pump water uphill and generate AC electricity on the return), many energy storage systems use either DC electrical components (batteries) or devices that provide AC at highly variable frequencies and then convert it to DC power and back to utility-grade AC (often used for flywheel storage systems). The conversion of AC to DC is relatively easy, although not necessarily efficient, and is done with a “rectifier” circuit,

4 frequently seen in consumer devices as “power supply.” The conversion from DC to AC is the job of an “inverter,” although many devices called inverters also operate in the opposite direction, allowing bi- directional flow of energy between AC and DC. To date, inverters have two major proposed applications in wind-diesel hybrids: (1) enabling the use of batteries for energy storage (all batteries store and deliver DC power) and (2) the conversion of “wild AC” from wind turbines or flywheels into grid-quality 60-cycle AC.

In the U.S., utilities deliver electricity to their customers as AC power at 60 cycles and 120 volts; maintaining a constant frequency and voltage within very narrow bands is critical to proper performance of many devices, including clocks and motors. Deviation from either the voltage or frequency is referred to as poor “power quality.”

In diesel-only systems (and low-penetration wind systems), maintaining frequency and voltage is completely the job of the DEG and is achieved by maintaining the generator at constant rotational speed. In a medium-penetration wind-diesel hybrid system, a control circuit must be added to control a dump load or to curtail output from the wind turbines, but frequency and voltage control remain the function of the DEG. In high-penetration systems intended to operate in a diesel-off mode, the DEG is no longer rotating, so some other method must be used to maintain power quality.

In grid-connected battery storage systems, the battery can be used to inject additional power into the grid when the frequency drops (a sign that some other part of power distribution system has failed), but the battery is not intended to provide energy for the entire grid.

Since the goal of a high-penetration wind system is to be able to operate in a diesel-off mode, the diesel generator is not able to provide the voltage and frequency control for the system. Because the wind resource is erratic, and always eventually stops, the wind turbines alone cannot provide dependable energy for the system.5

More reliable operation can be achieved by adding energy storage to the system, either electrical storage (a battery) or mechanical storage (a flywheel), both of which require an inverter. However, operation of such a system has two requirements: (1) the battery/inverter must be sufficiently sized to cover the load for at least as long as it takes to start a diesel generator, and (2) the inverter must be able to maintain a stable power output during all phases of the operation. The inverter must be large enough to meet the entire village load for at least a few minutes in case of a sudden and complete loss of wind power (most likely under high wind speeds, when turbines go from steady state, high output to zero

5 The original funded project was to demonstrate that the a new inverter design to be incorporated with a wind turbine could provide grid-quality AC and enable “wind-only” diesel-off operation—a claim made by Northern Power Systems. The immediately apparent drawback to this configuration is that it does not include any energy storage beyond the mechanical energy stored in the rotating wind turbine, and would likely lead to frequent power interruptions whenever the wind dropped below the load for more than a very brief time. It appears that this new design was actually a “low-voltage-ride-through” strategy, intended to keep wind turbines on line when low-voltage events occur on grid-connected power, something that keeps the wind turbine on line, but does nothing to provide continuous power for the load.

5 output after cutout), and it must also be capable of providing continuous power in diesel shutdown, diesel-off operation, and diesel start-up, and maintain adequate power quality during all these stages.

In addition to the inverter, some form of energy storage must be incorporated, and batteries and flywheels are the two most commonly used storage methods in village applications. Conventional lead- acid batteries such as those used in automobiles work well for their intended application (long, low- current charging and infrequent, high-current partial discharge) but fail rapidly under the high-discharge conditions encountered in village power systems. New lead-acid batteries are currently being marketed that claim to provide better performance under deep cycling conditions, and they should be tested. Lithium-ion batteries are also being developed for the electric car and might also prove useful, although they do tend to be of higher initial cost. Flywheels have been used for energy storage for a long time, but they have issues with mechanical losses (some energy is lost to friction over time) and maintainability as well as control and power management systems.

To operate remote microgrids in diesel-off mode, inverters connected to energy storage systems are required to replace the diesel generators as the grid-forming machine on the grid, (i.e., the inverter has to provide voltage and frequency stability for the grid.) Generally, inverters as energy sources on a grid with other synchronous energy sources operate in real/reactive (P/Q) control mode. When the inverter “forms” the grid, voltage/frequency (V/Hz) control mode is required. In addition, the inverter must be capable of transitioning from one control mode to the other while online; when in V/Hz control mode, it must provide “inertial” qualities similar to a diesel.

Project Overview The original project proposal submitted by ACEP, “Evaluating NW100B Inverter to Support Diesel-Off Operation in Alaskan Wind-Diesel Systems,” was to investigate the use of advanced inverter technology (specifically, the Northern Power Systems Northwind 100 inverter) to control wind-diesel hybrid systems in diesel-off mode. Funding was to be spent primarily on purchasing two Northwind 100 nacelles to construct a wind turbine simulator that would be supported by existing UAF diesel test bed infrastructure. The total project request was for $860,000.

The Denali Commission selected the project for funding, but at only half the requested funding level, i.e., about $430,000.6 The reduced funding level unfortunately precluded the purchase of the Northwind 100 nacelles, originally quoted at approximately $500,000. The project instead pursued an alternate approach developed by Sustainable Automation that proposed using an inverter and energy storage system (battery bank) to manage a village-scale power system, with the claim that this could be done with diesels off and without the need for a synchronous condenser. The heart of this system was a grid- forming energy storage power inverter, or GRIDFORM Power Converter, which Sustainable Automation represented as a near-commercial product. The following components were purchased through Sustainable Automation under this alternative project approach:

 A 100 kW wind simulator with LabView control interface ($100,000)

6 Due to EETG program funding constraints, several selected projects were awarded at reduced levels.

 A commercial valve-regulated lead-acid battery bank ($75,000)  A GRIDFORM Power Converter ($160,000)  An isolation transformer ($7,500)  A Matlab/Simulink/PowerSim model ($10,000)

While most of the hardware purchased functioned as expected, including the wind turbine simulator and the isolation transformer, the grid forming inverter turned out to be less well developed than anticipated.7 The core of the Sustainable Automation GRIDFORM Power Converter is the American Superconductor (AMSC) PM3000, intended for wind turbine systems. The PM3000 appears to be a low- voltage ride-through wind turbine controller, designed to keep turbines on line during low voltage events in a grid-connected environment. Based on specifications available online, it appears that it is designed to handle at least 300 kW of power, with a 1150 V DC bus connection and a 690 V x 750 A AC connection. These devices were manufactured in China, at considerably lower cost than inverters assembled in other parts of the world.

An initial functional test of the hardware was conducted at Marsh Creek in November 2011, but a more complete series of tests was conducted at the ACEP wind-diesel test bed facility at UAF between June and October 2012. During testing at the UAF test bed, several deficiencies were noted in the assembly of the GRIDFORM Power Converter that caused concern about the safety of this device. These deficiencies were identified and conveyed to the individuals associated with the former Sustainable Automation company, but they have not been corrected or resolved because the company is no longer in business. During the same testing event it was observed that the inverter did not produce the rated power (200 kW); this information was also conveyed to the same individuals, who claimed that the inverter had performed at rated power during testing in the fall of 2011 and suggested that either the battery settings or the battery condition was at fault. Sustainable Automation also noted that the AMSC PM3000 had “inherent limitations” that the company “could not overcome” and that this conclusion led to the demise of the company in December 2011. It is apparent that the Sustainable Automation GRIDFORM Power Converter not a commercially available product since the company is no longer in business to support the unit or to provide other units to customers (see Appendix B for comprehensive reporting on testing activities and results).

Despite the fact that the Sustainable Automation GRIDFORM Power Converter is not a finished product, the October 2012 test did successfully demonstrate many of the desired functions needed for an inverter in a high-penetration wind system. First, the unit smoothly transitioned between a diesel-on to a diesel-off state with acceptable transient performance, maintained good power quality during the diesel-off period, including frequency, voltage, and VARS support, and then transitioned back to a diesel- on state, once again with acceptable transients. The unit functioned well under unbalanced load conditions. The major performance deficiencies noted were that the unit did not provide the rated power level of 200 kW and it was not a reliable source of power under changing supply scenarios (i.e., a shut-down of wind generation during diesel-off would cause the unit to fault).

7 Sustainable Automation ceased operations as a company in December 2011, shortly after delivery of the hardware and an initial test of the equipment, which means that there is no warranty support for the device.

Based on these results, it appears that a high-penetration wind-diesel hybrid battery system using an inverter configured for VARS support without the use of a synchronous condenser is technically achievable. However, it is recommended that the Sustainable Automation inverter be replaced with a commercially available unit constructed to industry specifications and supported by the supplier.

Results and Next Steps For all high-penetration wind-diesel hybrid systems, a major issue is system control, especially in the diesel-off mode; the issue of maintaining frequency and voltage within acceptable limits while dispatching heat loads and energy in and out of the energy storage device is not trivial. In these systems, the inverter controls the flow of energy, collecting excess energy from the wind when available and sending it to storage, dispatching it to the load when needed, and maintaining power quality at all times. Demonstrating that such an inverter exists, can be purchased at reasonable cost, and is sufficiently robust for operation in remote utilities is critical. The GRIDFORM Power Converter project was undertaken to achieve these goals, and it succeeded in showing that energy storage and inverter technology can be used for diesel-off operation in high-penetration wind-diesel hybrid systems, but it did not demonstrate that such hardware is commercially available, cost effective, or sufficiently dependable to justify being incorporated into rural utilities.

A related issue is the availability, reliability, and cost of energy storage systems. Available conventional battery systems have proved to be insufficiently reliable and too costly (mostly because conventional lead-acid batteries do not survive long under deep-cycling conditions), but several new technologies, including advanced lead-acid batteries and lithium-ion battery packs being developed for the electric car market, may prove more suitable. Other energy storage ideas include large capacitor banks (and “super- capacitors”), mechanical flywheels, and compressed air.

While there are many new energy storage systems being proposed, it is not clear if any of these systems are suitable for use in wind-diesel hybrid systems. Adding storage to these systems adds capital and operation and maintenance (O&M) costs. Nearly all electrochemical storage systems degrade over time and thus have a finite lifetime, so the question is this: Does the energy storage system provide sufficient benefit during its operational lifetime to pay for the cost of replacing it?

Modeling can and should be used to address these economic questions, especially whether high- penetration systems can be designed and operated so that they result in a lower cost of energy for rural residents. There is little point in increasing system complexity and capital cost if the resulting energy costs are higher than those of a conventional diesel plant. The exact costs and benefits depend on many factors: the cost of fuel, the capital costs of the system, the quality of the local wind resource, the nature of the load being served, the value of any heat that can be delivered to a user, and the O&M costs and replacement intervals needed for each component.

Economic models such as HOMER—which was developed to help project developers quickly screen various installation configurations to determine which might prove the most cost-effective—have been accepted by the renewable energy community and by bankers alike. However, HOMER has limitations, especially with regard to high-penetration wind-diesel hybrid systems. One problem is that HOMER

8 provides estimates of fuel savings and wind energy production over a one-year period based on a historic wind pattern that has been fed into the model, making it difficult to compare results from a short-term test period with model results. It is possible to compare annual wind production to a value predicted by HOMER, but the actual wind characteristics will be different than the wind pattern used in the model, and HOMER calculates only one data point a year. Several critics have noted that none of the wind installations in rural Alaska have ever saved as much fuel as predicted by HOMER, and many fall short by very significant margins. Given the stochastic nature of wind, one might expect some random variation in comparing wind output to HOMER’s predictions, but results from existing wind systems are consistently lower than projected by HOMER.

The wind-diesel hybrid test bed at ACEP presents a new opportunity since it can be used as a “hardware model” for performance of wind-diesel hybrid systems for shorter-term tests and could provide modelers with improved performance parameters. For example, diesel engine efficiency curves are provided by engine manufacturers and have been verified in steady-state testing, but in wind-diesel systems without storage, the diesel engines can be “flogged” by the need to follow the wind and become less efficient. Adding short-term storage (capacitors, flywheels, or high-current battery systems) would allow gentler ramp rates on the diesel engine and enable the diesel engine to operate closer to the steady-state efficiency curve. During testing, diesel consumption in the two cases could be compared and some estimate made of the value of the short-term storage. To use the wind-diesel hybrid test bed as a hardware model, the system would need to be operated under (as near as possible) the same patterns—the same load, the same wind pattern, the same diesel engine—with different hardware configurations.

High-penetration wind-diesel hybrid systems for energy storage offer hope for reducing the cost of energy delivered to residents of rural Alaska communities, but additional development is required before these systems are suitable for mass deployment. Both hardware testing and computer modeling are needed to understand the potential benefits of these systems. ACEP and the Denali Commission are continuing to work with technology suppliers, utilities, and rural residents to evaluate and demonstrate currently available hardware.

Appendix A: Current ACEP Test Bed Facility Configuration and Status A “test bed” is defined broadly in hardware development as “any device, facility, or means for testing something in development.” In the computer world, a test bed is “an environment that is created for testing purposes.” At ACEP, the wind-diesel hybrid test bed under development is intended to test both the hardware and software components used in wind-diesel hybrid systems. The intention is to use the test bed for multiple projects, each requiring multiple test runs. The test bed is intended to serve the needs of a variety customers, including equipment suppliers, electric utilities, state agencies, and university researchers.

Systems under development must be tested against specific criteria, including the expectation that the new system provide some tangible benefit, such as lower capital costs, lower operating costs, or environmental benefits. There are two basic sources of testing criteria: (1) specifications given by technology suppliers and defined by purchase contracts and (2) system performance expectations based on comparing performance data to existing systems. Testing a device against its own specifications is critical in assessing new technologies, as spec sheets are sometimes based more on aspiration than performance and are intended more as bait for investors than engineering data for customers. Modeling is useful for comparing the diesel-only status quo option (low capital investment, high fuel costs) with the new hybrids, which have higher initial capital costs but lower fuel costs. Well-calibrated instruments, standardized testing procedures, and standardized data collection systems are all required.

The University of Alaska Fairbanks (UAF) began developing a technology test bed in 1998, testing fuel cells and diesel reformers. Sandia National Laboratories provided critical assistance in developing the test bed by helping with the planning and construction of fuel cell test stations, purchasing sensors and software, and mentoring UAF staff in dealing with technology suppliers, data collection and analysis, and report generation.

In 2004, UAF purchased a diesel generator to test new fuels. A test bed was built around an operating diesel engine instrumented so that operational data could be collected on the system, especially engine efficiency and emissions data, and modified to test heat recovery systems. This existing diesel test bed was referenced in the 2009 proposal to the Denali Commission. This test bed was installed in a 40-foot conex box and had been subjected to numerous experiments by graduate students. Given the limited access provided by the conex box installation, UAF elected to construct a new test bed facility, where the focus is on wind-diesel hybrid issues. This current test bed includes the following:

 All hardware purchased under the Denali Commission program o Sustainable Automation GRIDFORM Power Converter, battery bank, wind turbine simulator and control system, and isolation transformer  A 320-kW Caterpillar diesel generator and WoodWard EasyGen controller  A Detroit Diesel Series 50 125-kW generator (discussed above, purchased with Department of Energy funds in 2004)

 Manual switchgear (large manual breaker switches) for all major AC components, assembled at ACEP  Two LoadTech 250-kW AC load banks (one from the previous generation diesel test bed, one new)  ABB control system, including two diesel generator modules, a feeder control (to monitor loads), a dispachable load controller, and a wind turbine controller  A Fluke 435 Power and Energy Analyzer for evaluating power quality associated with both continuous operation and power quality “events”  Additional supplies and equipment, including cable trays, cabling, control wire, power meters, cabinets, protective and safety equipment, tools, etc.

In addition to grant funding from the Denali Commission, other contributions were made to the project by ACEP and UAF:

 A significant fraction of the staff time of the Alaska Wind-Diesel Applications Center (WIDAC) Director Kat Keith, Research Faculty Billy Muhndo, former postdoc and now Research Faculty Marc Muller-Stoffels, and Research Engineer David Light, as well the non-negligible effort on the behalf of ACEP director Gwen Holdman and Research Director Brent Sheets and UAF’s procurement office. At this time, a dollar figure for this additional effort has not been calculated.  Construction of a new energy test facility at UAF, at a total estimated cost of $4 million for an approximately 5,000-square-foot facility.  In addition, various pieces of hardware have either been constructed or purchased to augment the hardware purchased under the Denali Commission funding, as described above.

Several comments on the current state of the test bed facility are as follows, based on a site visit and interviews conducted during the summer of 2013:

 The test bed installation is being assembled with meticulous attention to the quality of workmanship with respect to appearance, safety, and long-term serviceability. The test bed has the look and feel of a utility (in fact, considerably better than most operational utility plants in Alaska). Much of the credit for this goes to the hard work and dedication of ACEP Research Engineer David Light.  Core pieces of equipment installed and operational: o The wind turbine simulator is capable of generating a simulated wind pattern from a computer file. o Load banks are capable of providing simulated village load from a computer file. o Caterpillar generator appears to be functional and communicating with the Woodward easYgen controller. o Fluke power quality meter can provide National Institute of Standards and Technology (NIST) traceable data for power quality events. o Revenue-quality power meters are installed on all three phase legs of the test bed.  Pieces of equipment physically present in the test bay but not installed or operational:

o ABB controller is installed in an electronic cabinet but is currently not able to control the wind-diesel system or provide data acquisition functions o Sustainable Automation GRIDFORM Power Converter does not perform to specifications and should be replaced o Smaller diesel generator is currently disconnected  Additional equipment purchases are envisioned, including: o Automatic switchgear (allows automatic connection and disconnection of various components, diesel engine dispatch, etc.) typical of new rural Alaska power plants o Fault generator to simulate power quality events (“line slap”, shorts, turbine faults) o Inverter to replace Sustainable Automation GRIDFORM Power Converter  The diesel generators need to be relocated to the diesel engine test bay. This will require the purchase of some hardware (radiators for the cooling system, piping for the glycol lines) and significant staff effort.

Discussions with ACEP staff indicate that the test bed facility is expected to be fully operational by summer 2013.

The test bed at ACEP is well suited for addressing the following issues:

 Economics: Can a wind-diesel hybrid system provide energy at lower cost than a conventional diesel plant?  Technical issues associated with high-penetration wind systems, including the following: o The need for a supervisory controller or some other control strategy to manage the various energy sources (wind, diesel engine, and energy storage system) and various loads (village load, energy storage system, dump loads) and curtail wind output. Commercial solutions, such as the ABB controller, may be available. o The need to manage power quality (frequency, voltage, and phase balance), which is influenced by the rapid fluctuation of the wind turbines, and sometimes the quality of energy delivered from the wind turbines. o The increased system costs associated with hybrid systems, including: . Cost of supervisory control system, associated component controllers, and software . Costs associated with maintaining the hybrid configuration, most notably the possible cost of replacing the batteries (especially if they are not used properly and fail quickly) . Costs of developing workforce able to deal with additional complications of hybrid systems, especially at the village level o The need to develop a “standardized” system approach, to allow both costs and results to be accurately predicted o Product, system, and methodology testing, including the following: . Testing and vetting new inverters  Sustainable Automation testing (completed)

III

 Need for “fault testing” system  Strategies for improving power quality . Testing of new battery systems  VRB battery (test completed)  Testing new lead-acid battery technology

Other needs of the wind-diesel-hybrid community that do not require a physical test facility:

 A nuclear industry-style “deficiency” reporting system—database and information distribution system for wind and wind diesel hybrid systems  Accurate cost analysis of conventional diesel plants, low-penetration wind systems, and wind- diesel-battery high-penetration systems.  Modeling, including economic modeling (models can be verified in laboratory test bed or utility installation)  “Procurement experiment” data—what is available to the industry, what does it cost, who can you buy it from, and does it work  Procurement standards to be used in state-funded projects (standards exist for wind turbines but are less well developed for battery and inverter systems)  Data quality control (all the way from sensor calibration to verification of database integrity) for utility demonstrations

The following figures were made June 21, 2013, during a visit to the ACEP wind-diesel hybrid test facility and illustrate the test bed installation at that point in time.

Figure 1: Single line drawing of test bus.

Figure 2: Wind simulator motors. One acts as a drive powered from a separate source; the other acts like a wind turbine and provides power to the experimental wind-diesel hybrid test bed. Denial Commission funded.

Figure 3: 336-V 1000-Ah valve-regulated lead-acid battery bank purchased as an energy storage system for the wind-diesel hybrid test bed. Installation includes 168 cells, each nominally 2 V, connected in series. Denali Commission funded.

Figure 4: Electrical panel for wind simulator provided by Sustainable Automation. Denali Commission funded.

Figure 5: Sustainable Automation GRIDFORM Power Converter, with AC panel door open. Denali Commission funded.

Figure 6: Central cabinet from Sustainable Automation GRIDFORM Power Converter. AMSC PM3000 power management unit designed for wind turbine applications in center. Denali Commission funded.

Figure 7: Manual switchgear for connecting test bed components to electrical bus.

VII

Figure 8: Isolation transformer installed between inverter and test grid. Denali Commission funded.

Figure 9: Switchgear associated with wind turbine simulator. Denali Commission funded.

VIII

Figure 10: One of two programmable AC load banks installed in test bed.

Figure 11: Diesel generator (in yellow) — 320 kW Caterpillar.

Figure 12: Diesel engine control and fire suppression tank installation.

Figure 13: Woodward easYgen controller (on left) and Caterpillar 320-kW engine controller.

Figure 14: 2000-gallon bulk fuel tank for storage of diesel fuel. Tank appears to be double-walled, with a spill prevention fill system. Tank does not appear to have fuel transfer pump (to move fuel to day tank) installed.

Figure 15: Detroit Diesel 125-kW Gen Set, currently in need of installation and repair.

Figure 16: Control computers for test bed. Wind simulator, manual switchgear, battery bank, and inverter in background.

Figure 17: Fluke 435 Power and Energy Analyzer, with current probes.

XII

Figure 18: PowerCorp (now ABB) control modules on top row. From the left to the right, generator module 1, generator module 2, feeder controller, wind turbine controller, battery controller, and secondary load controller. Bottom row is Schneider revenue-quality power meters.

Figure 19: Smaller load banks—one on left is a 55-kW AC load bank, top right is a 5-kW DC load bank, bottom is a 5-kW programmable DC load bank, useful for small scale battery testing.

XIII

Figure 20: View of physical test bay. Note the space available for placement of test equipment, the high bay doors, and the shiny floors.

XIV

Appendix B: ACEP Report

GRIDFORM INVERTER TESTS AND ASSESSMENT

Marc Mueller-Stoffels David Light Gwen Holdmann Brent Sheets

Funded under the Denali Commission’s Emerging Energy Technology Grant Fund Grant number: Denali Commission: 01233; UAF: G6249.

Independent review by Philip Maker

Alaska Center for Energy and Power Institute of Northern Engineering University of Alaska Fairbanks PO Box 755910 Fairbanks, AK 99775-5910

Gwen Holdmann, Director [email protected]

Dr. Marc Mueller-Stoﬀels, Research Assistant Professor mmuellerstoﬀ[email protected]

Revision history: Dec. 13, 2012 v 0.8 For review by Phil Maker and Jason Meyer Dec. 16, 2012 v 0.9 For second review by Phil Maker and the author team Dec. 18, 2012 v 0.95 For ﬁnal review and approval Jan. 16, 2013 v 0.96 Final edits. Pre-release version Mar. 19, 2013 v 0.97 Pre-release version for stakeholder review. Oct. 18, 2013 v 1.00 Final edits. Release version. Executive Summary

An inverter-battery system manufactured by Sustainable Automation Inc. (Sec. 3) un- derwent testing at the Alaska Center for Energy and Power’s Power Systems Integra- tion Laboratory (Sec. 2). The aim of the tests was to demonstrate that inverter-battery systems are a viable strategy for diesel-off mode operation of wind-diesel grids, and to investigate whether this particular inverter-battery system is ready to be deployed in rural Alaska. The system tests showed that diesel-off mode is attainable with the grid-forming inverter tested here (GRIDFORM inverter by Sustainable Automation Inc.). The GRIDFORM inverter provided high quality power and grid stability in diesel-off operation. However, the tests and the design review also revealed several shortcomings of the equipment. It is recommended that these shortcomings be addressed before the GRIDFORM inverter-battery system is deployed to rural Alaska. In addition, the need for extensive operator training with these new technologies is significant and should be factored into a purchase decision.

3 4 Acknowledgement of Funding

The GRIDFORM inverter testing project was funded by the Denali Commission under the Emerging Energy Technology Grant Fund, Grant number: Denali Commis- sion: 01233; UAF: G6249. Additional resources for ACEP’s Energy Technology Laboratory testbed development have been contributed by the Department of Energy through the EPSCoR program grant ‘Making Wind Work for Alaska’. The Alaska Center for Energy and Power thanks all funding sources for their con- tinued generous support.

5 6 Contents

1 Project Overview 9 References...... 10

2 ACEP Power Systems Integration Laboratory 13

3 GRIDFORMInverter 15 3.1 Inverter Design Review ...... 17 3.2 Inverter Battery Interaction...... 18 3.3 Data Acquisition Equipment Utilized ...... 18 3.4 Grid-Forming Inverter Performance ...... 19 3.4.1 Power Quality ...... 19 3.4.2 Diesel-oﬀ Mode ...... 20 3.4.3 Ride-through Capabilities ...... 20 3.4.4 Grid-Forming Inverter Energy Eﬃciency ...... 23 3.4.5 Limitations ...... 23

4 Conclusion and Recommendations 25

A Independent Review by Philip Maker 27

7 CONTENTS CONTENTS

8 1

Project Overview

At a glance: • Diesel-oﬀ mode in wind-diesel systems requires additional technology beyond what is required to integrate wind into a diesel grid.

• Sustainable Automation Inc. provided an inverter-battery system (grid-forming inverter) designed to operate an isolated grid in diesel-oﬀ mode.

• The grid-forming inverter system was integrated into a laboratory wind-diesel power grid and its performance to provide stable power in diesel-oﬀ mode was evaluated.

Utilities in rural Alaska typically produce electricity using diesel generators in isolated micro-grids. Since the 1980s several utilities have integrated wind power generators into their grids to supplement power production and mitigate high fuel costs. Wind power integration below 50% instantaneous power contribution is well understood. In this case the diesel gensets remain the prime power producer in the grid, and fulfill the function of keeping the grid stable by conditioning the power generated from wind turbines. More recent efforts are aimed at increasing the contribution of wind power to levels up to 100% thereby enabling to operate without diesel gensets being online. However, due to the typical nature of wind generators and the variabil- ity of the resource, this approach requires strategies to stabilize power quality and to manage drops of wind power production below load levels. One strategy to achieve diesel-off mode is to include energy storage systems (ESS) into the grid. An ESS needs to be sized such that it can supply enough power to mitigate temporary reductions in wind power, at least long enough to bring a diesel genset back online. At the same time, an ESS should be capable of improving power

9 REFERENCES 1. PROJECT OVERVIEW quality by stabilizing frequency and voltage in the grid, and by providing sufficient reactive power support for the wind generators and loads. The ESS tested in this case consists of an inverter-battery system. The function of the inverter is to convert power between alternating current (AC) on the grid side and direct current (DC) on the battery side. Furthermore, the inverter is capable of stabilizing the grid frequency and voltage, and can provide reactive power support. The system can operate standalone as an inverter-battery system with wind generators, or in conjunction with a diesel genset. ACEP tested the grid-forming ability of a newly developed inverter-battery system, called GRIDFORM inverter by the manufacturer (Sustainable Automation Inc.). In order to determine whether the GRIDFORM inverter was capable of supporting an isolated micro-grid in diesel-off mode, it was integrated into ACEP’s Power Systems Integration Laboratory (see Chap. 2) and subjected to test scenarios. In addition, the engineering of the GRIDFORM inverter was reviewed with respect to its utility to be deployed in rural Alaska. ‘Grid-forming’ is described in the literature as the capability of an inverter to operate in frequency and voltage control mode as an islanded grid. Distinctions are made between ‘grid-forming’ and ‘grid-following’, where grid-following is described as an inverter operating under real and reactive power set-points. Some authors define a third category (grid-supporting), best suited for micro-grids, meaning an inverter in voltage and frequency mode, that is sharing real and reactive loads via frequency and voltage control [1–9].

References [1] A Engler.“Control of parallel operating battery inverters”. In: Photovoltaic Hybrid Power Systems …49 (2000),pp. 1–4. url: http://renknownet2.iwes.fraunhofer. de/pages/hybird\_system/data/2000aix- en- provence\_engler\_pv- hybrid.pdf. [2] Micah J. Erickson, T. M. Jahns, and Robert H. Lasseter. “Comparison of PV inverter controller conﬁgurations for CERTS microgrid applications”. In: 2011 IEEE Energy Conversion Congress and Exposition. Ieee, Sept. 2011, pp. 659–666. isbn: 978-1-4577-0542-7. doi: 10.1109/ECCE.2011.6063832. url: http://ieeexplore. ieee.org/lpdocs/epic03/wrapper.htm?arnumber=6063832. [3] Luisa Frosio. Control Systems for RES-based systems and Smart Grids. 2012. url: http: //www.ics.trieste.it/media/1197308/18Frosiominigridscontrol.pdf.

10 1. PROJECT OVERVIEW REFERENCES

[4] F Katiraei and R Iravani. “Microgrids Management”. In: Power and Energy … June (2008), pp. 54–65. url: http://ieeexplore.ieee.org/xpls/abs\_all.jsp? arnumber=4505827. [5] A Mohd, E Ortjohann, W Sinsukthavorn, M Lingemann, N Hamsic, and D Mor- ton. “Supervisory control and energy management of an inverter-based modular smart grid”. In: Power Systems Conference and Exposition,2009. 2009, pp. 1–6. isbn: 9781424438112. url: http://ieeexplore.ieee.org/xpls/abs\_all.jsp? arnumber=4840000. [6] NA Ninad and LAC Lopes. “Per-phase vector (dq) controlled three-phase grid- forming inverter for stand-alone systems”. In: Industrial Electronics (ISIE), 2011 IEEE … (2011), pp. 1626–1631. url: http://ieeexplore.ieee.org/xpls/abs\ _all.jsp?arnumber=5984404. [7] E Ortjohann, A Arias, and D Morton. “Grid-Forming Three-Phase Inverters for Unbalanced Loads in Hybrid Power Systems”. In: Conference Record of the 2006 IEEE 4th World Conference on Photovoltaic Energy Conversion (2006), pp. 2396–2399. url: http://ieeexplore.ieee.org/xpls/abs\_all.jsp?arnumber=4060160. [8] Ph Strauss and A Engler. “AC coupled PV hybrid systems and microgrids-state of the art and future trends”. In: Proceedings of 3rd World Conference on Photovoltaic Energy Conversion,2003. (2003). url: http://ieeexplore.ieee.org/xpls/abs\ _all.jsp?arnumber=1305005. [9] Tao Zhou and Bruno François. “Energy management and power control of a hybrid active wind generator for distributed power generation and grid integration”. In: Industrial Electronics, IEEE Transactions … 58.1 (2011), pp. 95–104. url: http : //ieeexplore.ieee.org/xpls/abs\_all.jsp?arnumber=5439778.

11 REFERENCES 1. PROJECT OVERVIEW

12 2

ACEP Power Systems Integration Laboratory

At a glance: • ACEP operates an isolated micro-grid laboratory at a similar scale as rural Alaskan power systems.

• The grid consists of a 320 kWe diesel genset, a 100 kW wind turbine simulator, and a 250 kW load bank.

• The system is setup for 480 VAC operation. ACEP has developed a Power Systems Integration Laboratory to emulate typical rural Alaskan wind-diesel systems up to power levels of 500 kW. The purpose of the lab is to test technologies meant to increase power plant eﬃciency, as measured by the amount of diesel consumed per unit of energy produced, in a controlled setting which is readily accessible by road. Through this, the risk of acquiring sub-optimal equipment for rural Alaskan utilities is to be reduced by demonstrating and testing equipment designed/destined for rural Alaska before it is deployed. The testbed setup utilized for the test of a particular piece of equipment can be adjusted depending on the equipment’s typical and rated power levels, or based on the power levels of a given rural power plant. For the test regiment described here the testbed was conﬁgured with a 320 kWe Caterpillar diesel genset, a 100 kW wind turbine simulator and a 250 kW/187.5 kvar variable load bank. The nominal grid voltage is 480 VAC, three-phase. The wind turbine simulator, also a product of Sustainable Automation Inc., consists of two mechanically coupled induction machines, a motor and a generator. The

13 2. ACEP POWER SYSTEMS INTEGRATION LABORATORY motor is controlled by a variable frequency drive (VFD) and its output torque can be controlled via torque, power, or wind speed and wind turbine power curve time series, or set point inputs. The generator connects directly to the main grid bus. The power output is not conditioned. The 250 kW/187.5 kvar variable load bank is a product of Load Technology, Inc. Load can be controlled in 5 kW and 3.5 kvar steps, independently of each other. The nominal voltage of the load bank is three phase 208 VAC, and it is connected to the grid through a Delta-Wye-connected transformer (Delta on 480 VAC, Wye on 208 VAC, 300 kVA). The lab setup for this test is shown in Fig. 2.1. WTS

VFD IM IG L2 L1 T1 DC Battery AC SG T2

Figure 2.1: Single line drawing of Power Systems Integration Laboratory setup for the GRIDFORM inverter test. The GRIDFORM inverter is shown in green, connected to the battery and the isolation transformer (T2). The components of the wind turbine simulator (WTS) are shown on gray background. The induction motor (IM) drives the induction generator (IG) based on control signals transmitted to the variable frequency drive (VFD). Only the IG of the WTS is electrically connected to the grid, the VFD and IM receive external grid power. The main load bank (L1) is connected to the hybrid grid through a voltage transformer (T1). The load bank (L2) was used to simulate slight phase imbalances. The load bank connection and the connection to the synchronous generator (SG, diesel genset) were instrumented with WattsOn meters (blue dots). A Fluke 435 II Power Quality and Energy Analyzer (red dot) was connected to the load bank transformer for most tests. The exception is the inverter eﬃciency test, where the Fluke 435 II was connected to the grid side of the isolation transformer and the DC link of the GRIDFORM inverter (see Section 3.4.4).

14 3

GRIDFORM Inverter

At a glance: • GRIDFORM Inverter is IGBT-based, rated at 200 kVA/160 kW.

• As per manufacturer, additionally, a synchronous condenser is required for full functionality. The inverter-battery setup tested is a product of Sustainable Automation Inc., Boulder, CO1 (SAI). The GRIDFORM inverter is a new product developed by SAI, with an American Superconductor PM3000 IGBT-based inverter module at its core. The inverter tested is accompanied by an Absolyte® GP valve regulated lead-acid battery bank, sized by SAI. The grid connection is made through a Delta-Wye isolation transformer (Wye on grid-side, Delta on inverter-side). The GRIDFORM inverter supplied for the lab is serial number 2, with serial number 1 being deployed at Kokhanok, AK, and a smaller development unit exiting at SAI facilities. The specifications given by SAI for the GRIDFORM inverter are shown in Ta- ble 3.1. The GRIDFORM inverter cannot support a grid without an inertial machine (genset, wind turbine, synchronous condenser, or similar) being online. The manufacturer sug- gests a combination of GRIDFORM inverter and a synchronous condenser if diesel- off mode is desired. In this case, the synchronous condenser can provide reactive power support, and be the back-up inertial machine, should the wind turbine(s) sud- denly drop offline. This requirement is not an concern in the laboratory setting where

1Sustainable Automation Inc. ceased operation in December 2011. Final commissioning was performed by Sustainable Power Systems LLC. Sustainable Power Systems LLC does no longer oﬀer Sus- tainable Automation Inc.’s line of products, but pursues diﬀerent approaches to remote micro grid operation.

15 3. GRIDFORM INVERTER

Table 3.1: GRIDFORM inverter speciﬁcations as per ’Energy Storage Inverter User Manual’, Rev. 1, June 2012, provided by SAI.

Electrical Characterisitics Rated AC power 200 kVA Rated AC current 240 A AC line-to-line voltage 480 VAC Nominal battery voltage 336 VDC Rated DC current 500 A Nominal DC link voltage 750 VDC Switching frequency 3 kHz

Environmental Characteristics ◦ ◦ Storage temperature -40 C to 85 C ◦ ◦ Ambient operating temperature -25 C to 40 C Humidity 0 to 95% non-condensing Altitude <1000 m without derating

permanent grid stability is not an issue; a synchronous condenser was not part of the test setup. By some deﬁnitions (see Sec. 1), this renders the GRIDFROM inverter a grid-supporting inverter, as the synchronous condenser can be considered a source of reactive power on the grid with which the GRIDFORM inverter shares load via voltage control.

16 3. GRIDFORM INVERTER 3.1. INVERTER DESIGN REVIEW

3.1 Inverter Design Review At a glance: • Electrocution Hazard: Placement of DC meter/relay is extremely dangerous. • Placement of cable entry points vs connection points can be improved.

• Routing of internal connections and plumbing can be improved. In the course of using the GRIDFORM inverter several design issues impacting installation and safe operation became apparent. The cabinet in which the inverter was supplied is designed as a bottom entry for all electrical connections. The connection points within the cabinet do not lend them- selves for this connection scheme, though. In a future iteration of the packaging, it is suggested to move all connection points to one location close to the design entry point for electrical connections. Similarly, the control system’s customer connections are placed on the center door of the three door cabinet. Again it is suggested to move these connections to a more convenient location within the cabinet. It is also suggested to group all AC control breakers and DC control breakers in two easily accessible panels. Several internal power connections are routed directly across metal bars and sharp edges, and are held in place with 1/8th inch zip ties and self-adhesive cable holders. After transport and low-duty operation, some of the connectors have already failed and, in some places, the cables show wear to their insulation. The inverter can be controlled through a touchscreen interface at the center door. This interface allows the review of fault messages and resetting of faults. The only ex- ceptions to this are DC bus over/under-voltage faults, which are generated through a meter/relay in the back of the cabinet. To clear these faults, the user has to reach over the DC bus link and capacitor bank (exposed bus bar; does not de-energize immediately upon opening breaker)2SAI. This creates an unnecessary electrocution hazard. Ideally, this fault could be reset at the front touch panel as is true for all other faults. DC connector blocks and capacitor banks should be covered. This is especially true for the capacitor bank, which does not de-energize immediately upon DC breaker opening. In addition, AC ﬁltering equipment should be contained; capacitors should have covers contain blow-outs that could destroy other equipment; resistors should have covers to protect technicians from potential burn hazards.

2The manufacturer notes that this relay can be set to non-latching in order to avoid this hazard. However, the unit was supplied as is and not modiﬁed by ACEP personnel, neither in software nor hardware, post commissioning.

17 3.2. INVERTER BATTERY INTERACTION 3. GRIDFORM INVERTER

Control power for the GRIDFORM inverter is supplied from the AC side of the inverter. Since this particular inverter cannot act as an uninterruptible power supply, this is not an issue here. However, should a future iteration of the product allow for standalone (no inertial machines) operation, control power should be provided from the energy storage side to allow for black-start capability. The GRIDFORM inverter is liquid cooled, with a 50/50 glycol-water mix. The coolant has to be drained for transportation and long-term storage. Small amounts of coolant left in the cooling system can dry out in this case and form a flaky residue when new coolant is introduced into the cooling system. This residue caused a filter in the system to clog and subsequently a shut-down due to overheating. It is suggested that, when the cooling system is drained, it be flushed with clean water to avoid such issues.

3.2 Inverter Battery Interaction The interaction between a bi-directional inverter and energy storage is not trivial. The GRIDFORM inverter employs a lead-acid battery bank to manage power import and export. It is the nature of lead-acid battery technology that maximum import/export power levels are a function of state of charge (SOC). SAI has divided the charge levels of the battery into ﬁve quadrants - high-high, high, mid-range, low, and low-low. Grid-forming mode is not available in high-high and low-low state of charge; that is, the battery SOC needs to be kept between 79% and 24% for full operability of the system. Optimal management of the battery system requires an additional level of control system. This control system would have to be adjusted to the particular battery system used, e.g., the system in Kokhanok, AK employs a diﬀerent battery. SAI supplies a top-level control system, which was not available for the test performed here.

3.3 Data Acquisition Equipment Utilized The testbed grid was monitored with several power meters during all tests. Two sta- tionary WattsOn® 1100 meters by Elkor Technologies, Inc., were used to monitor voltage and current at the load bank transformer and the diesel genset to bus connection (Blue dots in Fig. 2.1). These meters allow power ﬂow and frequency metering at 5s intervals. A Fluke® 435 Series II Power Quality and Energy Analyzer was used to monitor power quality as seen at the load bank transformer (T1 in Fig.2.1). This meter is capable

18 3. GRIDFORM INVERTER 3.4. GRID-FORMING INVERTER PERFORMANCE of high frequency measurements. However, the recorded data is reduced to 1 sample/s and minima, maxima, amid averages are recorded for these intervals. The meter was placed at the load bank transformer as this location is the closest emulation of a station feeder. Additionally, a Fluke® 434 Series I Energy Analyzer was deployed to measured energy ﬂow at the grid-side of the GRIDFORM inverter isolation transformer.

3.4 Grid-Forming Inverter Performance At a glance: • Power quality is good, with acceptable voltage dips under load changes. • Diesel-off mode was demonstrated. • Ride-through capabilities/fault hardiness could not be tested due to lack of equipment. • Inverter efficiency is inversely proportional to loading and within 68 to 95%. • Power level given in manufacturer’s specifications could not be reached. The focus of the testing described below is concerned with the diesel-off mode of the GRIDFORM inverter. In this mode, the GRIDFORM inverter provides support to variable generation sources, such as a wind turbine. It can sense whether additional power is needed to meet demand, or whether the variable power source is exceeding demand. Accordingly, the GRIDFORM inverter will provide power from/absorb power into the battery. In addition, the GRIDFORM inverter manages the power quality of the grid by adjusting voltages and frequency.

3.4.1 Power Quality The power quality in the grid in diesel-oﬀ mode was monitored against ITIC/CBEMA power acceptability curves. The Fluke® 435 Series II Power Quality and Energy Ana- lyzer at the load bank transformer (T1 in Fig. 2.1) was setup to record voltage dip and swell triggered-events slightly within the aforementioned acceptability curves. Only two voltage dip events were recorded in diesel-oﬀ mode: One during a sudden change in kvar loading, i.e., power factor change from 0.83 to 0.98 with demand and WTS both set to 50 kW; and one event during a series of rapid WTS output changes. Dur- ing both events power quality remained within both the ITIC and CBEMA power acceptability curves, with voltage deviations no more than 150 V. Both events were shorter than 70 ms.

19 3.4. GRID-FORMING INVERTER PERFORMANCE 3. GRIDFORM INVERTER

3.4.2 Diesel-off Mode When the diesel genset is online, it usually regulates frequency and voltage in the grid. In this setting the GRIDFORM inverter can be used to provide kW/kvar support at given set-points . In diesel-off mode the GRIDFORM inverter provides voltage and frequency support to the grid. The transition to diesel-off mode was demonstrated with the wind turbine simulator providing parts of the power demanded by the load bank (and inertia). The transition to diesel-off mode is quite smooth both in the voltage and frequency pic- ture (Figs.3.1 and 3.2). Variable wind or variable loads do not have a significant effect on frequency stability. Only sudden load changes exceeding 100 kW resulted in frequency deviations larger than 0.03 Hz. Under the same conditions, voltages remained stable, with the exception of the dips discussed in Sec. 3.4.1. The GRIDFORM inverter is designed to be operated within a Hybrid Supervi- sory Control System (HSCS) developed by SAI. This system would generally control transitions into and out of diesel-off mode, i.e., diesel synchronization and change of operating mode of the GRIDFORM inverter. This system requires that a Wood- ward EasyGen controller (or other automatically synchronizing tool) be in place on the diesel to control diesel-to-grid synchronization when bringing the diesel back online. This controller was not available at the time of testing. Nonetheless, manual synchronization to bring the diesel back online was successfully performed. Even with this crude method, no detrimental effect on power quality during the transition was observed.

3.4.3 Ride-through Capabilities System robustness in fault and unusual load situations, such as phase imbalances, is a major concern for Alaskan utilities. To test the GRIDFORM inverter performance under imbalanced load, a smaller resistive load bank was connected to the main load bank transformer. With this, phase to phase imbalances up to 19 kW could be simulated. The GRIDFORM inverter performed well under these conditions. The lack of a fault simulator at the time of testing did not allow for investigation of the general fault-hardiness of the system. However, since the manufacturer recom- mends the parallel use of a synchronous condenser, at least a moderate level of fault hardiness of the GRIDFORM inverter-synchronous condenser combination can be expected.

20 3. GRIDFORM INVERTER 3.4. GRID-FORMING INVERTER PERFORMANCE

280

279

278

277

276 Line−Neutral Voltage [V] 275

274

273 10:26:24 10:27:50 10:29:16 10:30:43 10:32:09 10:33:36 10:35:02 Time [min]

Figure 3.1: Voltage response during transition to diesel-oﬀ mode. Shown are average line to neutral rms voltages (blue - L1N, red - L2N, black - L3N). The transition occurs at 30 min. While the diesel provides a tighter spread of voltages on all lines, the GRIDFORM inverter voltage control keeps the line to neutral voltage on all three phases well within allowable bounds. The dips and swells in diesel-oﬀ mode are due to load changes. Data shown was taken with the Fluke® 435 Series II Power Quality and Energy Analyzer at the load bank transformer.

21 3.4. GRID-FORMING INVERTER PERFORMANCE 3. GRIDFORM INVERTER

60.12

60.1

60.08

60.06

60.04

Frequency [Hz] 60.02

59.98

59.96 26:24 27:50 29:16 30:43 32:09 33:36 35:02 Time [min]

Figure 3.2: Frequency response during transition to diesel-oﬀ mode. The transition occurs at 30 min and the GRIDFORM inverter is able to hold frequency very close to the nominal 60 Hz. The deviations in diesel-oﬀ mode are due to load changes. Data shown was taken with the Fluke® 435 Series II Power Quality and Energy Analyzer at the load bank transformer.

22 3. GRIDFORM INVERTER 3.4. GRID-FORMING INVERTER PERFORMANCE

3.4.4 Grid-Forming Inverter Energy Efficiency The GRIDFORM inverter efficiency was assessed both for importing power to and exporting power from the battery. This assessment was performed with the GRID- FORM inverter in kW/kvar mode and the diesel being online. The reason for this was that power levels could exceed the maximum power output of the WTS. This did not occur due to other reasons (see Sec.3.4.5). Efficiency was assessed using the Fluke® 435 Series II Power Quality and Energy Analyzer with the DC measurement performed at the battery to GRIDFORM inverter connection and the AC measurement taken on the grid-side of the isolation transformer3 (Fig. 2.1, red line from the right of T2 to inverter-battery connection depicts the meter connections). The meter calculates energy efficiency internally. Due to the nature of the measurement, the given efficiency is only for the inverter, including the required isolation transformer. It does not include roundtrip efficiency of the battery. The inverter system efficiency is logarithmically dependent on power levels (Fig.3.3). At low power levels (< 30 kW), efficiencies between 68% and 88% are observed with the efficiency of charging the battery being slightly lower below 20 kW than the efficiency for discharging the battery. At about 20 kW this trend is reversed. At high power levels, importing power into the battery is between 90% and 95% efficient. While the efficiency of exporting power is about 5% lower for any given power level. The lower efficiency at low power levels does not necessarily pose a problem as total energy throughput at these levels will be low as well, that is, total energy loss will be low.

3.4.5 Limitations During the eﬃciency test it was noticed that nominal power levels (160 kW, 200 kVA) could not be reached. The inverter went into a DC under/over-voltage fault at 106 kVA (discharging) and 87 kVA (charging) respectively. The test was performed between 63% and 60% SOC. Power levels where increased to the fault levels in 5 kW steps. Block loading of the inverter led to over-/under-voltage faults at even lower power levels. It is suspected that the buﬀering (programming of PID controllers) for voltage control require more development4.

3The transformer was included in the eﬃciency assessment as it is required for the grid-connection of the inverter. 4After consulting with the manufacturer about this problem, the battery was exercised (two full battery capacity tests and equalization cycles) in order to determine if this would improve performance. This had no eﬀect on the maximum available power level, but more faults due to DC over-voltage, and DC over-current were encountered.

23 3.4. GRID-FORMING INVERTER PERFORMANCE 3. GRIDFORM INVERTER

100 GFI Supplying Energy GFI Absorbing Energy 95

80 Efficiency [%]

65 10 20 30 40 50 60 70 80 90 100 Active Power [kW]

Figure 3.3: Efficiency vs active power. The GRIDFORM inverter exhibits up to 95% efficiency when absorbing power into the battery (black dots), and up to 91% efficiency when supplying power to the grid (blue dots). At low power levels efficiency is generally much reduced. Efficiency was assed using the inverter efficiency mode of the Fluke® 435 Series II Power Quality and Energy Analyzer with the DC measurement performed at the battery to GRIDFORM inverter connection and the AC measurement taken on the grid-side of the isolation transformer. The GRIDFORM inverter was in kW/kvar mode during this test with the kvar set-point constant at 35 kvar. The diesel was online.

24 4

Conclusion and Recommendations

At a glance: • The operation of the grid-forming inverter as a means of operating in diesel-oﬀ mode has been demonstrated.

• Power quality and grid stability are within acceptable bounds.

• Further development is recommended before deployment to rural Alaska.

• Signiﬁcant operator training is necessary for successful deployment of high-contribution renewable energy systems.

The GRIDFORM inverter-battery system supplied by SAI has been shown to be capable to act as energy storage solution that allows the operation in diesel-off mode in a wind-diesel grid. The power quality and grid stability in diesel-off mode is managed by the GRID- FORM inverter and has been shown to be within acceptable bounds for smooth grid operation. The inverter efficiency at low power levels is rather low. This would be further compounded by the roundtrip efficiency of the battery employed. While an economic and power flow study was not performed as part of this project, it is suspected that much of the operation of the GRIDFORM inverter would occur at fairly low power levels and, thus, the energy losses observed would have a significant impact during operation in a utility setting. Power levels observed were well below the levels specified by the manufacturer. The reason for this is likely insufficient adjustment of the inverter controls to the battery dynamics. This, in combination with several design deficiencies (one of them

25 4. CONCLUSION AND RECOMMENDATIONS dangerous) leads to the conclusion that the GRIDFORM inverter in its current form should not be deployed in rural Alaska. This is compounded by the fact that additional, costly equipment and controls are required to make full use of the diesel-off capabilities of the GRIDFORM inverter. By no means is the above conclusion to be understood as a condemnation of the approach of using an inverter-battery system to achieve diesel-off operation in high contribution renewable energy situation. However, it is to be understood that the interactions between the inverter and the grid and the inverter and the energy storage system are by no means trivial. Thus, the design and control of such systems requires further development. There are several suppliers of technology similar to the GRIDFORM inverter tested here. Some of these suppliers claim the capability to operate their inverter in diesel-off mode without the need of further equipment, e.g., a synchronous condenser. However, the total number of deployed systems is too small to conclude that any of them are fully proven and mature technology. It is recommended that utilities, considering the deployment of a high-contribution renewable energy system with inverter-energy storage technology, require manufacturers to demonstrate their technology in full before a purchase is made. This would include, but is not limited to: 1) operation at advertised power levels both importing and exporting, 2) smooth transition from grid-following to grid-forming mode, 3) acceptable power quality in all operating modes, especially under block (un-)loading, 4) stable operation in grid-forming mode. Additional consideration has to be given to control systems. Most common SCADA systems will not be able to manage diesel-off mode without being adjusted. Furthermore, utilities should be aware that their operating staff will require significant training to efficiently operate and service such technology, i.e., inverter and battery trouble shooting skills, network trouble shooting skills, knowledge of save operation of DC power systems are required, in addition to the common skills of operating a wind-diesel power system.

26 Appendix A

Independent Review by Philip Maker

Is inverter-based energy storage a viable option for high-contribution renewable energy systems? Yes: its been installed in a variety of sites but the number of sites running at high contribution remains small. ABB and SMA are two larger companies that can provide systems at a variety of sizes in this area that have been demonstrated to work in the long term. The critical issues are:

• Fault behavior particularly as compared to a diesel generator.

• Integration into a complete control system.

What are the general technological road blocks to implementing successful inverter-based high-contribution systems (ESS, and inverter control)? • System and device integration.

• Front inverter design and control.

• Reliability and system longevity.

Is this particular inverter (plus synchronous condenser) ready to be deployed in real-life remote energy systems? The current system is for various reasons outlined in the report not suitable but it is a fair distance down the path.

27 APPENDIX A. INDEPENDENT REVIEW BY PHILIP MAKER

Has the testing been suﬃcient to investigate performance and limitations of the GRIDFORM inverter? Yes, within the limits of an initial investigation and not a full blown engineering test. Missing fault simulation is a key feature to be added to the lab. Deeply rooted bugs can only found through longer term testing. 60% of problems were probably found.

Are the conclusions given based on testing and review correct and adequate? Yes.

What other testing is recommended to ensure viability of the GRIDFORM inverter in a real-life remote energy system? • Development and review of a fault/failure model.

• Environmental testing.

• Testing under faults.

• Testing within a particular control system framework.