Hawaii Energy and Environmental Technologies (HEET) Initiative Office of Naval Research Grant Award Number N0014-11-1-0391 Task 5. STORAGE TECHNOLOGY 5.2 Grid Scale Storage Systems Testing Prepared by: University of Hawaii at Manoa, Hawaii Natural Energy Institute September 2016
Grid Scale Storage Systems Testing Research conducted under HEET10 included continued testing to evaluate the effectiveness of three grid-connected Battery Energy Storage Systems (BESS) for ancillary services under different operational conditions on three islands. To date, three grid-scale Battery Energy Storage Systems (BESS) have been procured and commissioned. The three BESS operating on three different islands will ultimately provide unique ancillary services, and help determine the roles BESS can play on isolated electrical grids under significant penetration from variable renewable energy resources. The integration of large amounts of variable renewable power, such as photovoltaic (PV) and wind into a power system, can lead to increased generation reserves and costs, lower power quality and reliability, or in extreme cases, can cause blackouts if not implemented properly. Rapid changes in power generation or consumption have a greater effect on small, isolated grid systems. Because the response times of BESS are relatively short, they can provide services to address some issues related to renewable integration and provide a mitigation strategy to address some of these issues. Under the current award, research efforts for the Hawaii Island grid primarily focused on continued evaluation of the BESS for frequency regulation. The Big Island BESS, procured under previous ONR funding (Award N00014-10-1-0310), has operated for over three years in both a wind smoothing and frequency regulation mode. The new work illustrated how local BESS support (power smoothing from the 10MW Hawi wind farm) can cause grid-wide issues. The current work also demonstrated that the use of expanded deadband in the BESS control algorithm greatly reduced cycling, with an expected extension of lifetime, while still providing a significant portion of the grid-wide benefit. The second Altairnano 1MW, 250kWh BESS was procured and installed on Oahu to simultaneously provide power smoothing and voltage regulation within an electric substation serving large industrial loads. Under the current funding, commissioning was completed and operation of the control algorithms demonstrated. Detailed testing and evaluation of this BESS will continue other ongoing ONR funding (Award N00014-12-1-0496). A third BESS, an Altairnano 2MW, 397kWh system also procured under Award N00014-10-1-0310 was installed, commissioned and accepted for testing on the island of Molokai. Hawaii Island BESS The Hawaii Island BESS has been in operation since December, 2012. Over that time it has cycled more than 3.3 GWh. The objective of research and analytical efforts under this award 1
were to determine how to best support the grid on Hawaii Island while minimizing battery degradation. As discussed in the HEET 09 final technical report (Award N00014-10-1-0310), this system can provide one of two ancillary services at any given time: wind smoothing or grid frequency response. The system is almost always set to frequency response because it was learned that wind smoothing very often countered the needs of the e grid system. i.e. the BESS may inject energy into the system to counter loss of power from the wind farm at time that the system requires less energy to balance frequency. While running in frequency response mode, it was learned that there are settings that serve grid needs while minimizing cycling. In HEET10 in one case, the grid frequency appeared to become marginally stable when some highly periodic wind gusts were encountered. It was learned that we could achieve 2/3 of the grid benefit (reduction of grid frequency variability) while cycling only 1/3 of the energy throughput found in our initial algorithm settings. Details about these results will be discussed in forthcoming papers. HNEI s planned data collection for grid interaction with this BESS has been completed under HEET 10. However, there is further interest in long term battery cell degradation. Hawaiian Electric Light Company (HELCO) has agreed to continue operation of the BESS and to provide operational data to HNEI for further analysis. Oahu BESS Under HEET 10, a second 1MW, 250kWh BESS was procured and installed at the Hawaiian Electric Company (HECO) substation at Campbell Industrial park on the island of Oahu. Unlike the first BESS, which primarily regulated grid frequency, this BESS will be simultaneously providing power smoothing as well as voltage regulation within an electric substation serving large industrial loads. The industrial park served by the circuit served by this substation includes a high penetration of large PV systems, which can disrupt voltage and power quality when clouds suddenly occlude the solar panels. Like the Hawaii Island BESS, the project team for the Oahu BESS consisted of both public and private partners. Table 5.2.1 lists the key partners involved in the project. 2
Table 5.2.1: Project team and roles. Partners (Public and Private) Office of Naval Research (ONR) University of Hawaii (UH), Hawaii Natural Energy Institute (HNEI) Hawaiian Electric Company (HECO) Altairnano Inc. Parker-Hannifin Inc. Sun Edison Inc. (SEI) Integrated Dynamics Inc. (IDI) Northern Plains Power Technologies (NPPT) Role Funding source Project lead, technical oversight, coordination algorithm development, performance verification, analysis, and reporting Utility grid owner, electrical design, planning, coordination, installation, communications design, safety, and metering design Battery module supplier, BESS supplier, system integration, electrical design, planning, project management coordination, and hardware installation Inverter manufacturer Site construction Algorithm development, software development, communications design, and simulations Circuit simulations, inverter simulations, and algorithm development HNEI and HECO again executed a Memoranda of Agreement stating that the ownership of the BESS procured by HNEI would be transferred to the utility after installation and acceptance of the commissioning.. After transfer of ownership, HNEI will collect data and conduct research (under other funding) for an initial period of two years. HNEI s interest is to examine the role a fast response BESS can play in supporting a feeder with a high penetration of PV capacity. This, along with battery degradation can inform the utility industry on the costs and benefit over time that are related to BESS implantations in support of renewable integration. Oahu s electric grid is much larger than any of the other Hawaiian Islands (nearly 1.2GW peak, still small compared with most power grids across the continental US) with a number of grid issues at the distribution level.. Therefore, site selection was narrowed to substations serving areas with high renewable penetration levels and/or large loads. HNEI and HECO mutually agreed to choose the Campbell Industrial Park (CIP) distribution substation near the southwestern region of the island. This substation provides power to large pumps, and is also under heavy penetration from PV. When these large pumps run, current drawn by the circuit changes abruptly, leading to voltage changes, flicker, and reduced power quality. HNEI procured a BESS for the CIP substation that is nearly identical to the one installed on Hawaii Island. a 1MW, 250kWh Altairnano system based on Lithium-ion Titanate cell chemistry. Major milestones in this development effort are documented (and available on the HNEI website) as follows: 3
Facility Acceptance Test (FAT) performed in December, 2012 [2] Algorithm development as well as substation simulation and modeling [3] Laboratory Acceptance Test Plan [4] Laboratory Acceptance Test Results [5] Site preparation (including planning, communications, construction, and electrical work) Shipping, logistics and installation Site Acceptance Test (SAT) performed in August, 2016 [6] Algorithm Acceptance Test Plan [7] Algorithm Acceptance Test Results [8] The SAT was conducted on 8/2/2016. Figure 5.2.1 shows the test in progress as a HECO lead engineer coordinates with HECO System Operators (foreground) and an Altairnano engineer discusses readings with Altairnano system experts (background). The tests were successful as the system did meet hardware performance specifications. Figure 5.2.1: Hawaiian Electric Company (HECO)/Campbell Industrial Park (CIP) Site Acceptance Test. The HECO/CIP BESS is the most heavily metered of the three HEET systems. Figure 5.2.2 shows a one-line diagram indicating the locations of 6 of the meters. Some meters are used for real-time control, while others are used for post-processing (data analysis) or for HECO monitoring. In the figure, meters M1 and M5 measure essentially the same thing: the low side of the output of the inverter. M5 is used by the inverter for control purposes. M1 is only used in post processing. M2 is the power realized by the BESS on the circuit. This is used for post processing. M3 is the power at the substation, which is used in post processing. M4 is the power 4
consumption of the circuit. This is used for both post processing and real-time control. M6 is used in post processing. This meter measures the power consumption of the BESS itself. Figure 5.2.2: One-line diagram of HECO/CIP BESS. The SAT, discussed above, determined HNEI s acceptance of the system. HECO developed its own tests to verify system operations. Part of HECO acceptance process included verification of meter readings. An HNEI representative aided HECO in this process through analysis (see Figure 5.2.3 which shows the meters are providing reasonable values). It was found that all meters used in control and post processing were functioning properly [9]. M3 and M6 track closely as they should, unless there is a power disruption between the two (the BESS). M1, M2, and M3 track closely in terms of real and reactive power. The only difference can be attributed to polling delays and transformer loss. M6 is providing reasonable readings given the HVAC power requirements. However, it was found that a meter HECO System Operation uses for monitoring was not functioning properly and will be replaced by HECO. This did not affect the acceptance of the system since only the reactive power reading was inaccurate. 5
Figure 5.2.3: Analysis of meters. (See Figure 5.2.2 for meters M1 through M6). The role envisioned for the CIP BESS is different than that of the Hawaii Island BESS. Here, the real-power of the BESS is used to smooth power transients caused by large pumps and PV while the reactive power capability of the inverter is used to regulate voltage. The algorithm development effort was performed jointly by HNEI and NPPT. The algorithm was designed to be flexible enough so that trade-off studies could inform other algorithm developers. See Chapter 2 in [3] for a detailed description of the algorithm. The general research plan HNEI envisions for this system under other funding sources is to exercise the CIP BESS differently than the Hawaii Island BESS. As discussed above, the Hawaii Island BESS maintains a SoC very close to 50% at all times. It is anticipated that the CIP BESS will cycle more slowly, yet more deeply than the Hawaii Island BESS. A future research effort will be to determine differences in battery degradation between the two, with tests to be initiated in the fall of 2016 under other funding sources. 6
Molokai BESS The third BESS, procured under prior funding, was installed on the island of Molokai. This system is intended to provide a high speed response to system disturbances such as the sudden loss of generation or load to this, the smallest of the three electrical grids under test. In fact, the grid is so small and dynamic that the speed of the original design of the BESS control system was not sufficient to serve its intended purpose. Research and development of an improved control and communications architecture has been started and is currently on-going. Of the three BESS installations, the Molokai site provided the most challenges. The inertia (in essence, a resistance to change) of the Molokai grid is low enough that relatively small imbalances between power generation and consumption can cause chain reactions, leading to island-wide blackouts. Technologies proposed to serve such a grid must be carefully vetted so that latent responses to these imbalances (manifested as grid frequency deviations) do not exacerbate the situation. Even a 200ms delay between a loss of generation (or load) and realization of a real power correction, can cause the entire grid to become classically unstable and can cause island-wide blackouts (that is, a latent correction is worse than no correction at all). The project partners and their roles are shown in Table 5.2.2. Table 5.2.2: Project team and roles. Partners (Public and Private) Office of Naval Research (ONR) University of Hawaii (UH), Hawaii Natural Energy Institute (HNEI) Maui Electric Company (MECO) Altairnano Inc. Parker-Hannifin Inc. Sun Edison Inc. (SEI) Integrated Dynamics Inc. (IDI) Northern Plains Power Technologies (NPPT) Role Funding source Project lead, technical oversight, coordination, planning, performance verification, analysis, and reporting Utility grid owner, electrical design, planning, coordination, installation, communications design, safety, and metering design Battery module supplier, BESS supplier, system integration, electrical design, planning, project management coordination, and hardware installation Inverter manufacturer Site construction Algorithm development, control system infrastructure redesign, software development, communications design, and simulations Grid simulations, inverter simulations, and algorithm development 7
Of the approximately 5MW load, on Molokai over 2MW is serviced by PV during the daytime. It was determined that an Altairnano 2MW, 397kWh BESS would be more suitable for this grid than the 1MW, 250kWh systems used on the other two islands. Although Factory Acceptance was completed and deemed acceptable, HNEI and its partners discovered that the existing control and communications technology was insufficient for application on Molokai because of the low grid inertia. HNEI is currently preparing additional contract partnerships to address this problem during future awards. The following was completed during this current award: Algorithm (controller) design [10] Simulation of BESS and grid [11] Exploratory research and documentation on the latency issue (attempt to reduce correction to 50ms or less) [12] Site preparation (including planning, communications, construction, and electrical work) Shipping, logistics, and installation SAT Results [13] For the Molokai grid, the SAT had to be more carefully designed and executed since, as mentioned, a sudden excess generation of 2MW on a 5MW grid would cause island-wide blackouts. For this reason, HNEI issued a memorandum [14] to the project team proposing a stepped approach: several small power increments were to be used to demonstrate both the power and energy performance of the BESS. When the tests were conducted, it was found that 500kW steps could be taken instead of the 100kW steps originally proposed. This simplified the test process. Power data from all active diesel generators as well as the BESS is shown in Figure 5.2.4: Generation data from the Site Acceptance Test (SAT). 8
Figure 5.2.4: Generation data from the Site Acceptance Test (SAT). An expansion of the power steps, as well as grid frequency is shown in Figure 5.2.5. While the BESS was close to 0kW (offline), 500kW incremental steps were taken. As the power was increased (away from the 0kW reference), 250kW incremental steps were taken. The 500kW steps caused 0.55 Hz frequency events, while the 250kW steps caused 0.33 Hz frequency events. These events were not large enough to cause grid-wide problems. 9
Figure 5.2.5: Power and grid frequency during SAT. During some State-of-Health tests (part of the SAT), it was found that the BESS had lost about 3-4% energy capacity between the FAT and the SAT. Review of the individual cell group data showed that the degradation occurred across all cell groups, so no single cell group was limiting the entire system. It is possible that some significant thermal events in May 2016 caused the degradation. The HVAC system was damaged during shipping. As a result, a new HVAC was installed. It was later found that the new HVAC was tuned to respond to humidity, not temperature. Even with the loss in capacity, the BESS did perform very close to specification and was deemed to be acceptable. The Molokai BESS is intended to provide fault prevention through high speed frequency regulation. Demonstration of this use will be of future value to DOD as the use of renewablebased microgrids is incorporated into base designs. Specifically, the desire is to demonstrate performance on a low inertia grid with high penetration levels of PV. 10
References 1. Altairnano Inc. ALTI-ESS 1MW System for HECO / HNEI November, 2012. PowerPoint Presentation. 2. M. Ropp, S. Perlenfein, M. Tun, Control Algorithm Requirements and Functional Specifications (Report 3), December 11, 2013. 3. K. Musser, HECO Lab Acceptance Test Plan, October 31, 2014. 4. K. Musser, HECO Lab Acceptance Test Results Report, November 17, 2014 5. Altairnano Inc., Site Acceptance Test Project HECO 1.0MW ALTI-ESS, August 11, 2016 6. K. Musser (2016), HECO CIP Control Algorithm Site Acceptance Test Plan, May 3, 2016 7. HECO/CIP Algorithm Acceptance Test Results, September 9, 2016 8. M. Tun, HECO BESS Metering Analysis, August 26, 2016 9. NPPT report, Report on a proposed controller for a frequency-stabilizing battery inverter system for the Molokai power system, January 23, 2015 10. NPPT report, Behavior of a proposed battery-inverter system on Molokai during faults, May 06, 2015. 11. K. Musser, A. Mitchell, MECO Moloka i ʻ Algorithm Investigative Project Report, January 20, 2016. 12. Altairnano Inc., Site Acceptance Test Project MECO 2.0MW ALTI-ESS, June 23, 2016. 13. M. Tun, Coastal III SAT Proposed Power Test Design, May 4, 2016. 11