TODAY’S STUDY: Hawaii’s Utility Answers Its Regulators
Transforming to Deliver a More Affordable Clean Energy Future for Hawai‘i; HECO’s 2016 Power Supply Improvement Plan (Part 2)
April 2016 (Hawaiian Electric Companies)
Eight Observations and Concerns
The Commission noted eight Observations and Concerns [in a plan submitted earlier this year that was rejected], each of which encompasses a wide swath of areas under analysis in developing our 2016 updated PSIP. None of these eight Observations and Concerns can be considered in isolation. As such, we have integrated them throughout our planning, modeling, analyses, and decision-making.
#1. CUSTOMER RATE AND BILL IMPACTS
Chapter 3 fully describes the overall planning process, plan development, and iterative optimization process from the 1st iteration, which was included in the PSIP Interim Status Report filed February 16, 2016, through the development of the Final Plans and selection of the Preferred Plans. Financial analysis and “all-in” results are presented in Chapter 4. The Net Present Value of cumulative revenue requirements, under both 2015 EIA Annual Energy Outlook Reference and February 2016 EIA Short Term Energy Outlook fuel price forecasts, have been calculated for the best evaluated resource plan for each theme. Residential customer rates and monthly bill impacts, in nominal and real (2016) $/kWh, are provided for both fuel price forecasts. It should be noted that all finalist and Preferred Plans meet or exceed all statutory RPS requirements.
To maximize the accuracy of our analyses, we updated all input assumptions, including resource costs, fuel costs, and resource availability assumptions. We also shared all relevant assumptions with the Parties to solicit feedback. In addition, we engaged NREL to independently assess resource cost assumptions and provide an analysis of wind and PV availability. NREL’s reports can be found in Appendix F.
Theme 2, which uses the LNG fuel price forecasts included in Appendix J, produced significant cost savings and has the largest beneficial impact to customer bills. To address the uncertainty in future fuel prices, sensitivity analyses were completed for both the 2015 EIA Annual Energy Outlook Reference fuel price forecast and February 2016 EIA Short Term Energy Outlook fuel price forecast for each case. While there is no way to accurately predict future fuel prices, results from Ascend Analytics’ stochastic modeling of all-in delivered LNG and oil indicate that oil prices are characterized by “higher levels of volatility and slower rates of mean reversion as compared to natural gas. Higher volatility in oil prices translates to more uncertainty in future oil prices and a wider 90- percent confidence band in comparison to LNG.” Figure 2-1 depicts these results.
To address the capital expenditure constraints, revenue requirement projections which included capital expenditure projections for power supply, smart grid, ERP, and all other utility capital expenditures (referred to as “balance of utility business capital expenditures”) were considered. As described in detail in Appendix I, the balance of utility business capital expenditures have been calculated using a top down approach for the high fuel price scenario. Chapter 4 summarizes the capital expenditures by category for each Theme.
#2. TECHNICAL COSTS AND RESOURCE AVAILABILITY
Utility-scale resources are a key decision variable in the Decision Framework, which assesses the cost-effectiveness of various resource types.
We started by updating all resource costs, including capital costs, interconnection costs, fuel costs, O&M costs, and resource availability assumptions. Virtually all deployable technologies were considered. Though found not to be cost-effective at this time, new concepts such as accelerating alternative fuel vehicle adoptions (electric vehicles and hydrogen vehicles) and flexible electrification where electric vehicles could be used for load balancing were evaluated by E3. We retained NREL to independently assess our new resource cost assumptions and made appropriate adjustments to our assumptions as a result. We also commissioned NREL to develop independent assessments of the utility– scale solar PV and wind levels that could be developed on each island based on topographic, land-use restrictions, proximity to urban areas, and renewable energy production potentials in specific locations. NREL’s reports can be found in Appendix F.
Although adjustments were made to O‘ahu for utility scale PV and onshore wind to be consistent with NREL’s resource potential estimates, cases including high levels of PV were developed and analyzed. We compared case results of varying levels of energy storage and biofuels, and developed an optimized-mix of these dispatchable resources. In addition, we included community-based renewable energy (CBRE), DER and DR resources, utility scale PV, geothermal, onshore and offshore wind, biomass, biofuels, pumped storage hydro, and battery energy storage systems. (After this filing, we will complete our analysis of an inter-island transmission system, including estimated costs and benefits relative to offshore renewable energy serving O‘ahu and benefits of combined grid operations.)
Chapter 3 fully describes the planning process and Appendix K provides all of the cases considered. Both high DG-PV and market DG-PV cases were evaluated. Integration requirements for DG-PV are discussed in detail in Appendix N. Identification and consideration of integration costs for DG-PV was included in all of the analyses. In addition, accelerating renewables (Theme 1) which achieves 100% RE on the neighbor islands (including Lana‘i and Moloka‘i) by 2030 were developed and optimized for cost.
As noted in Chapter 3 and Appendix C, the overarching objective of the planning process was to optimize and find the lowest cost mix of resources and plan to achieve the statutory RPS requirements. The resulting near-term actions to acquire cost-effective RE projects are described in Chapter 8.
#3. DISTRIBUTED ENERGY RESOURCES INTEGRATION
DER is one of three key resource-types that were optimized as part of the Decision Framework, and we evaluated the full spectrum of DER. Energy efficiency attainment and electric vehicle adoption were forecast and incorporated in system net load for all PSIP cases. Demand response, distributed storage, and DG-PV were optimized through iterative cycles to achieve lowest system cost while enabling customers to provide costeffective and reliable grid services. Self-consumption economics were based on retail rates; grid export economics were based on the value the DER provides the system (utility-scale PV LCOE for DG-PV, value of storage to the system for distributed storage, value to the system for DR).
Multiple options were developed to integrate DG-PV on over-hosting capacity circuits and the lowest cost integration option was selected for explicit consideration in the economics for those DG-PV systems forecast to be installed on an over-hosting capacity circuit. The DG-PV integration strategies and costs are more fully described in Appendix N.
We determined high-value system-level use cases for DER in 2016 - 2020 as follows. Robust DG-PV adoption compensated at utility-scale PV LCOE reduces the need to procure utility-scale PV and helps meet near-term RPS targets cost-effectively. Storage was analyzed as a decision variable in the various PSIP cases, and was found to be cost effective for selected use cases in DR programs.
We sought cost effective solutions by weighing the costs and benefits of (full or partial) inverter retrofit against alternative ones when addressing either circuit or system-level interconnection barriers. For instance, we are currently considering the cost and benefits of legacy inverters without ride-through capabilities in our contingency battery analysis. We considered retrofit of inverters to ones that have reactive power capabilities for voltage mitigation in the DG-PV integration analysis (see, Appendix N).
A cornerstone of the DR program portfolio is the aggregation of DR resources. All of the proposed DR services utilize various DER technologies to achieve this aggregation philosophy. Furthermore, the demand response management system that will be used to deliver the DR services through the intelligent management and optimization of groups of DERs has been specified to allow for the attribution, selection and dispatch of these resources across various zones. These zones map to the physical topography of the various islands’ systems and span from the system level at the highest level down to the individual circuit at the lowest level. As such, the current architecture and system design of the DR portfolio implementation allows for targeted deployment of DERs, which is suitable and appropriate as a tool for helping to address distribution or transmission level constraints such as those being considered by non-transmission alternatives in South Maui.
We varied RPS attainment in the analysis cases and, through iterative cycles, optimized DER amounts across islands and across cases to determine the role and contribution of DER in high-RPS attainment scenarios. In addition to the DG-PV adoption forecast optimized for the system, we analyzed a "high DG-PV" forecast to further characterize the role and contribution of DER in aggressive RPS attainment scenarios. DER plays a significant role in the preferred plans. Further work on how to achieve the sustainable DER adoption as envisioned by the preferred plans will be covered in the DER 2.0 proceedings.
#4. FOSSIL-FUEL PLANT DISPATCH AND RETIREMENTS
Chapter 3 outlines the breadth of cases considered in the three iterations completed, around three Themes: Theme 1–Accelerate Renewables, Theme 2–Renewables With LNG, and Theme 3–Renewables Without LNG. Cases considered various mixes and amounts of resources. The multiple cases were specifically designed to iterate towards a low-cost objective, and address risks associated with changes in fuel price by analyzing both LNG and oil, and analyzing various fuel price forecasts. We refined those cases to incorporate results from preceding runs of DER, DR, and utility-scale resources iterations to determine low cost potential with minimized risks, and analyzed grid modernization to characterize the tradeoffs and risks of modernizing our generating fleet versus other resource options. We identified potential dates for displacement of fossil generation, then updated our Fossil Generation Retirement Plans. Additional details for the Fossil Generation Retirement Plan can be found in Chapter 8 and the Component Plans included in Appendix M.
Theme 2 included LNG as a transitional fuel on O‘ahu, Maui, and Hawai‘i Island and modernization of the generation fleet on O‘ahu with efficient, flexible replacement generation selected to support the growing renewable fleet on O‘ahu. Additional details of LNG as a transitional fuel are described below. For all cases, both high and low fuel price forecasts were evaluated to understand the respective cost impact. The analyses suggest that the most significant savings can be achieved with LNG and modernization of the generation fleet with market DG-PV. Details of the Preferred Plan are provided in Chapters, 5, 6, and 7, and the financial results are provided in Chapter 4. It should be noted that all cases comply with statutory RPS requirements.
As part of our analysis, we reviewed and clarified our environmental compliance strategies, and updated our Environmental Compliance Plan and Key Generator Utilization Plan. Finally, we updated our Generation Commitment and Economic Dispatch Review. All of these plans are included in Appendix M, Component Plans. LNG as a Transitional Fuel
We have highlighted the need for modernized and flexible generation resources in order to minimize costs, reduce emissions and facilitate the increased integration of variable renewable resources. Even with these new resources in place, the Companies’ current fuel source for its dispatchable generation during the transition period to a 100% RE will be petroleum-based fuels.
As a result, customers will be exposed to a petroleum-based fuel which is: ■ Forecasted to cost more than LNG. ■ Significantly more volatile in price than LNG. ■ Subject to increasing restrictions under tightening federal environmental standards.
With LNG as a transition fuel, the Companies see an opportunity to lower the cost to customers, reduce pricing volatility, and accelerate the reduction in air emissions. An LNG plan has been designed specifically as a transition solution for Hawai‘i that seeks to limit the amount of investment in permanent island infrastructure. Further, the Companies’ plan contemplates that the LNG seller will have the ability to remarket excess LNG, which will reduce the risk for potential variability in the demand for LNG as the integration of renewable resources increases. Hawaiian Electric does not view LNG as substituting for, or competing with, new renewable resources on the islands. Rather LNG represents a complementary solution which can help achieve the Companies’ goals of keeping costs to the customers as low as possible while mitigating impacts to the environment and flexibility integrating intermittent renewable resources. LNG represents a good value proposition to customers under a wide range of potential renewable penetration scenarios, especially when combined with the flexible, efficient, modernized generation described in the previous section.
Overview of the LNG Delivery System: In initially evaluating an LNG delivery solution for Hawai‘i, the Companies looked at (1) land based LNG import terminals and (2) Floating Storage and Regasification Units (FSRU), both of which entailed installation of permanent infrastructure on and offshore, new gas pipelines, and long permitting processes. Therefore, the Companies opted to issue a request for proposal (RFP) for a containerized LNG solution to land LNG in Hawai‘i and distribute it to its generation fleet across the State. This solution would use International Standards Organization (ISO) containers, metal vessels that can be loaded and transported on a conventional truck, to transport LNG locally and, maximize flexibility and reduce requirements for dedicated land based infrastructure.
A possible LNG supply chain would consist of the following components: ■ Natural gas sourced from some of the most prolific gas reserves located in Northeast British Columbia. The gas would be transported from the gas reserves to Fortis BC’s Tilbury liquefaction plant on the Fraser River by pipeline where it would be liquefied. ■ The LNG would be loaded onboard ships for transport to Hawai‘i. Upon arrival in Hawai‘i, the LNG would be delivered in ISO containers to points of use on O‘ahu, Maui, and Hawai‘i Island. ■ Multiple ships, owned and operated by the seller, would be employed to ensure a steady rate of LNG delivery to the various generating stations.
The containerized supply chain was selected as the option with the greatest congruence with the following evaluation criteria set forth by the Companies.
Flexibility with Minimal Permanent Infrastructure: To be consistent with achieving the RPS goals, the Companies required any fuel supply to have flexibility to accommodate a dynamic energy environment and generation from renewable resources. The fuel supply system should have minimal permanent infrastructure that could limit flexibility and increase the risk of stranded assets.
Neighbor Island Coverage: The Companies required a cost-effective solution that could supply fuel to Maui and Hawai‘i Island just as easily as to O‘ahu without making substantial modifications to the overall supply chain.
Minimal Permitting: To expedite adoption of cheaper natural gas in the fuel portfolio, the Companies required non-permanent infrastructure for the LNG supply system to avoid extensive and time-consuming permitting processes associated with developing an LNG terminal.
Security of Supply: To mitigate geo-political risk and ensure continuity of supply, the Companies sought a fuel supply from a North America as opposed to gas sourced from politically sensitive global locations.
Lower Price Volatility to Customers-Gas vs. Oil Indexed Pricing: Globally, LNG is typically priced off a formula which is indexed to oil prices. To reduce dependence on oil-linked, fuel pricing (current fuel portfolio) and minimize commodity pricing volatility, the Companies required LNG to be indexed off of North American natural gas prices.
Ability to Serve Other Customers in Hawai‘i: The Companies wanted the LNG seller to have the ability to sell excess volumes to third party off-takers and/or for the Companies to take additional spot volumes if available.
Unit Conversions Under a merged scenario between the Hawaiian Electric Companies and NextEra Energy, the Companies intend to enter into an agreement to acquire approximately 800,000 metric tons of LNG annually from the Fortis LNG facility in Vancouver, BC. Deliveries could start in 2021 and coincide with the commencement of commercial operations of modernized combined cycle units at Kahe. In addition to the modernized units, the Companies would convert five of their existing generation units (six including HEP if its purchase by the Companies is approved by the Commission) to allow them to use LNG in addition to petroleum-based fuels. This involves installation of new equipment to receive, store and regasify the LNG, and conversion of the existing generating units to allow for gas utilization (with total estimated cost of the conversions at approximately $340 million). Although not yet negotiated, it is assumed that the two combustion turbines at the Kalaeloa Partners LP Generating Station would also be modified to use LNG. After the completion of the modernization and conversions, the Companies would have approximately 1,100 MW of generation capacity capable of using LNG-based fuel during the transition period to 100% RPS (as outlined in Table 2-1).
#5. SYSTEM SECURITY REQUIREMENTS
Selected resource cases from each of the three Themes for each island grid were screened for system security with a focus on loss of generator and electrical transmission fault disturbances. These selected resource plans formed the basis for performing a limited system security analyses that defined in a technology neutral manner the fast frequency response (FFR) and primary frequency response (PFR) requirements for selected years of a plan. The results of the security analysis are presented in Appendix O.
Since filing our 2014 PSIPs, we have updated and revised our system security requirements and focused this analysis on single contingency loss of generation events to determine acceptable under frequency load shedding (UFLS)14 capacities. Loss of generation contingencies have a greater impact on resource plans because it dictates online reserve requirements which in turn, establish FFR and PFR requirements. A full system security analysis that includes voltage stability, rotor angle stability and fault current protection coordination on for all islands will be performed for the preferred plans.
For O‘ahu, HI-TPL-001 was revised to allow no UFLS for single generator contingency events (previous criteria allowed 12% customer loss) while Maui and Hawai‘i Island allow 15% loss of system load (previous criteria allowed 15% customer loss). The Moloka‘i and Lana‘i systems were removed from HI-TPL-001 since these systems are unique island distribution systems that do not qualify as transmission systems. Further revisions to HI-TPL-001 are required for multiple contingency events, both loss of generation and/or loss of transmission elements.
The more stringent HI-TPL-001 criteria for O‘ahu is designed to minimize the risk of deep load shed events, and potential island-wide blackouts with an appropriately sized FFR resource such as a BESS which become more likely in the future with even more distributed PV. Under high levels of distributed PV penetration, the residential load net of PV is reduced so UFLS schemes are less effective, compromising system security. UFLS is designed to shed low impact loads and avoid critical load like hospitals, emergency responders, military bases, schools, etc. The proliferation of distributed PV is primarily on residential distribution circuits so the daytime UFLS capacities continue to degrade and it is becoming more difficult to find sufficient load to shed during a single contingency event. Additionally, the more stringent criterion support the use of distributed resources to supply fast frequency response. Load shedding of the distribution system, as allowed under the previous criteria, would be counterproductive since it would disconnect demand response resources from the system.
The limited system security analysis for Hawai‘i Island was expanded to simulate the impacts of transmission faults that cause loss of generation contingency events for selected resource plans. Hawai‘i Island's transmission infrastructure covers a very large territory that increases its exposure to electrical faults that can cause large capacities of DG-PV to disconnect from the system. Additional analyses were performed to determine FFR and PFR requirements to ensure system security for Hawai‘i Island and should be indicative findings when these analyses are conducted for the preferred plan.
Fundamentally, distributed generation (primarily PV) poses one of the biggest challenges to system security because it imposes conflicting requirements on the electrical system: 1) the reduction of system load displaces synchronous generators and 2) distributed resources increases regulating and frequency response reserve requirements that are traditionally provided by synchronous generators.
More specifically, transformation of the electrical system must address the following system security issues: ■ DG-PV displaces synchronous generators that provide essential grid services like inertia, regulating reserves, and system fault current. ■ DG-PV reduces the capacity of the system’s under frequency load shed scheme (UFLS). ■ Legacy DG-PV and their less flexible frequency ride through ability increases the magnitude of a loss of generation or fault contingency. ■ DG-PV is currently not controllable by and is invisible to the system operator.
The process of identifying needs and designing solutions follows a several-step process that we believe addresses the Commission’s concerns regarding the prior PSIP filing. (Note that this process was outlined as six steps in the Companies’ February 2016 filing. The revised process is equivalent, but reorganized to complement the rest of the PSIP more clearly.) The five steps are: 1. Establish operational reliability criteria. 2. Define technology-neutral ancillary services for meeting reliability criteria. 3. Determine the amount of ancillary services needed to support the resource plan. 4. Find the lowest reasonable cost solution, considering all types of qualified resources. 5. Identify flexible planning and future analyses to optimize over time.
The amounts of each type of ancillary service needed to meet system security vary by island, resource plan, and time period. That is because Frequency Response needs are driven by the size of the largest contingency event, which is generally the loss of the largest unit online at the time (combined with potential sympathetic loss of legacy DG-PV). Regulation needs are driven by the variability of net load (that is, load minus variable generation output), which depends especially on the amount of PV and wind, and Replacement Reserve needs are driven by the amounts of Frequency Response and Regulation needed after an event.
The Companies defined fast frequency response and primary frequency response requirements in technology-neutral terms so any qualified resource can meet them, whether traditional generation, advanced features of inverter-interfaced generation and storage, or demand response. Our objective is to identify the lowest reasonable cost combination that ensures system security for a given resource plan and in subsequent iterations, let the market and specific resource applications determine available resources.
To do so, we break the analysis into three steps: 1. Construct an initial pre-DR solution that meets system security needs; 2. Substitute DR to the full extent it is cost-effective, producing a revised resource strategy; 3. Consider whether the solution would affect system conditions (especially unit commitment and dispatch, affecting inertia and the amount of Primary Frequency Reserves available) to warrant another iteration of analysis.
There was not sufficient time to complete these three steps for the preferred plans. These steps will be done in conjunction with development of the Demand Response Programs.
#6. ANCILLARY SERVICES
As part of this filing, the Companies’ analyses began with the establishment of operational reliability criteria and the refinement of grid service definitions sufficient to meet these reliability criteria. This refinement of ancillary services was grounded in the definitions of grid services found in the Supplemental Report filed under Docket No. 2007-0341, filed November 30, 2015.
In particular, Fast Frequency Response (FFR) was refined into several sub-categories of FFR, including: Instantaneous Inertia (II), Primary Frequency Reserves (PFR), Fast Frequency Reserves 1 Up (FFR1Up) and 2 Up (FFR2Up), and Fast Frequency Reserves Down (FFRDown). Further, Supplemental Reserves was recast to Replacement Reserves (RR) and Regulating Reserves was refined to Regulation Reserves Up (RegUp) and Regulating Reserves Down (RegDown). The Companies then revised these ancillary services needs for the O‘ahu cases.
These revised ancillary service needs for O‘ahu were coupled with the existing needs defined for the other island systems and a set of resources that are capable of costeffectively meeting the ancillary service needs were identified. Included in this resource pool was utility-scale, centralized energy storage resource options as well as a DR portfolio that included the use of distributed, behind-the-meter storage options. As part of the DR optimization effort, the Companies developed respective optimal and most cost-effective implementation of the combination of these resources. The final optimized potential of distributed storage will be iterated and refined prior to filing the Final DR Program Portfolio application.
Consistent with the previous methodology applied during the development of the Interim DR Program Portfolio application (Docket No. 2015-0412), the Companies assessed the quantities of these service needs over a 30-year horizon and developed the value of these services by virtue of the costs associated with delivering them. With these values defined, the Companies were then positioned to assess substitution opportunities for delivering these services via the most cost-effective means possible.
The DR portfolio, utilizing a growing population of DERs, was considered as a cost effective substitution option for delivering these ancillary services. The Companies refined the DR portfolio based on previous feedback in an attempt to find the lowest reasonable cost solution considering all types of qualified resources for all islands. The Companies then identified flexible planning and future analyses to optimize the DR portfolio over time. This process is not complete, but will continue until the Final DR Program Portfolio application is filed in mid-2016. Finally, the Companies updated our Must-Run Generation Reduction Plans and Generation Flexibility Plans to include these ancillary service refinements.
#7. INTER-ISLAND TRANSMISSION
Our PSIP analyses show that, for O‘ahu to achieve 100% renewable energy in 2045, significantly greater off–island renewable resources will be required (if found to be more cost effective than biofuels). Analysis performed in this updated PSIP has shown that O‘ahu would require more offshore capacity than was included in our 2014 PSIP assumptions. Because of this, we plan to further analyze an array of inter-island transmission options after April 1, 2016. A plan for addressing the interisland transmission analysis is discussed in Chapter 9: Next Steps. In conjunction with the analysis, we also plan to further investigate offshore wind.
#8. IMPLEMENTATION RISKS AND CONTINGENCIES
Our Decision Framework contains nine risks and uncertainties that we used as part of our assessment to develop our 2016 updated PSIP. The risks identified in the Decision Framework were used as parameters in the selection of representative resource plans for each Theme on each island and ultimately to select each island’s Preferred Plan.
Chapter 3 describes the multiple initial cases, which were specifically designed to iterate toward a low-cost objective. The impact of accelerating the implementation of renewable energy resources, LNG and generation modernization, while accounting for risks attributed to changes in fuel prices for both LNG and oil, were evaluated. We refined these cases to incorporate results from preceding runs of DER, DR, and utility-scale resource iterative cycles, iterated to achieve low-cost and minimized risk objectives, and analyzed grid modernization to characterize tradeoffs and risks of capital investments. We ran production simulation using different modeling software (via consultants) for comparative purposes, conducted stochastic analysis to characterize risks associated with fuel price forecasts (through Ascend Analytics as described above), and ran sensitivity analyses using high and low fuel price forecasts.
We calculated present values of revenue requirements, and the relative difference in revenue requirements between cases for initial cases. Capital expenditure constraints were considered as described above. Using the Decision Framework, the Preferred Plans were selected and five-year action plans to implement the Preferred Plans were developed. It should be noted that with the exception of Theme 2 which requires LNG, generation modernization, and unit conversions, the near term actions for all final plans are very similar.