An economic assessment of the benefits of storage in South Africa by Jeremy Hargreaves, Energy & Environmental Economics and Paul Frink, Energy Solutions, Parsons Energy storage technologies have many benefits including displacing the need for new capacity, energy and ancillary services from thermal generation, and new transmission infrastructure, as well as easing integration of intermittent power sources such as solar and wind. In addition to these broad system benefits, energy storage can provide a host of local distribution or customer benefits when strategically placed. The Industrial Development Corporation (IDC) is interested in evaluating the potential of energy storage technologies to increase access to reliable, affordable electricity in South Africa, encouraging policies to support the adoption of energy storage technologies, and exploring opportunities to invest in energy storage projects. IDC is currently leading a steering committee of stakeholders in the South African electricity industry to help guide and promote the adoption of energy storage technologies in South Africa. IDC commissioned the South Africa Energy Storage Technology and Market Assessment through the United States Trade and Development Agency (USTDA) to provide support to the IDC and Steering Committee in guiding and promoting the adoption of energy storage technologies in South Africa [1]. This effort included market research; technical, economic and financing assessments; development, environmental, and legal/regulatory assessments; and a roadmap that recommends steps, milestones and timelines for the adoption of energy storage technologies in South Africa through 2030. This paper highlights the findings of the economic assessment, which focused on calculating the broad system benefits of storage. These go some way to covering the costs of storage, however additional local or customer benefits through strategic placement of storage systems on the grid would be required to reach cost effectiveness. We present a funding gap analysis, comparing system benefits to storage costs, and present qualitatively how the funding gap can be covered through strategic placement of storage systems. The full spectrum of geographies on the electrical system that determine the types of service that storage can offer include bulk transmission, local congestion zones, sub-transmission, distribution, and behind-the-meter storage (Fig. 1). The benefits that we capture in the bulk system analysis presented in this report are the benefits that would be attributed to storage in current integrated resource planning (IRP) processes where least cost system operations are being solved for. These benefits include those bulk system revenue streams in the generation and bulk transmission portion of the system. Fig. 1: Potential revenue streams for storage.
The question asked from a bulk system level planning perspective is when and how much storage should be built as part of a least cost portfolio of resources, given the energy, capacity, and ancillary service needs of the system as loads grow, the existing generator fleet changes, and new renewables are installed. This is the question we have answered in this paper. However, there are other use cases and storage revenue streams potentially realizable that were not included in the quantitative part of our analysis: In local congestion zones, storage can realise the benefits of congestion price arbitrage, or local resource adequacy value in the form of avoided local generation. At the distribution level, these benefits include deferral of capital investments and improvements in reliability and power quality. Behind the meter, customers can benefit from additional back-up power. One benefit to the storage owner not mentioned above is the potential retail rate savings from demand charge reduction and arbitrage. These savings can potentially be high enough for some customers to invest in storage. However, unless the customer tariff incentivises usage of the storage device that aligns with valuable grid scale uses of the battery, this type of storage investment can lead to a more expensive electricity system, and can shift costs from one group of ratepayers to another rather than reducing costs for ratepayers as a whole. In the final section of the report, we summarise options for cost effective, customer owned storage. In our quantitative analysis of the bulk system benefits of storage, we identify the funding gap necessary to incentivize storage build. The funding gap is the difference between the expected cost of storage in future years and the cost at which bulk system level storage is a part of the least cost resource portfolio. The funding gap is therefore equal to the additional benefits storage would have to capture through the distribution and customer revenue streams described above to become cost effective against other alternative resources. This article will discuss the methodology and findings of calculating the funding gap, and qualitatively discuss how additional local storage benefits, beyond those at the bulk system calculated in this paper, could cover the funding gap. These additional benefits depend on the specific location of the resource, the markets or programs available to realise the value the device offers to the grid, the control systems available to the utility, and the ownership or business model of the installed storage device. These benefits are harder to realise than those at the bulk system, where Eskom can control and dispatch storage devices. Part of the larger Energy Storage Technology and Market Assessment was to identify implementable steps to realise the full value of storage. The funding gap To find the funding gap, we determine at what cost storage would be part of a least cost future resource portfolio for South Africa. The difference between the cost at which storage would be part of the economic portfolio and the projected future storage cost is the funding gap. Determining least cost future resource portfolios requires an integrated resource planning modeling approach. Resource planning model As renewable generation increases on the system, the types of integration solution that can be used to integrate them expands. These include conventional generation, flexible loads, storage, renewable diversity, expanded transmission, and new market products and energy policy. There are many different combinations of resources that can be included in the resource portfolio to meet reliability needs, so determining the least cost portfolio is best done through an optimisation framework. Storage competes with other resource options. By optimally selecting resources, we are effectively asking whether there are cheaper ways of offering the same grid services as storage, or is storage the most cost-effective way of meeting system needs? The lowest cost portfolio of renewables and integration solutions at any point in time will be a mix of resources that minimises both operating costs and capacity expenditures over the planning horizon. The value of each
integration solution will change over time depending on the evolving needs of the system. Those selected in an optimal resource portfolio will offer the greatest net value over their lifetime in combination with the other resources selected. Some technologies may be stepping stones to longer term portfolios. Fig. 2 depicts an optimal tradeoff between renewable overbuilding and other integration solutions in meeting a renewable sales target. The optimal point for each resource will be where the benefit of the marginal unit of any resource to the system is equal to its marginal cost. Each type of resource adds a dimension to the optimisation and each combination of resources will have complex operational interactions. Finding the least cost solution requires a sophisticated optimisation model that treats operational and investment costs while satisfying operational and reliability constraints. Fig. 2: Tradeoff between different integration solutions how to find the least cost portfolio. We use E3 s Renewable Energy Solutions model (RESOLVE), an optimal investment and operational model designed to inform long-term resource planning questions in systems with growing penetration levels of renewable energy. RESOLVE is the model used for IRP by the California Public Utilities Commission [2], the Hawaiian Electric Companies [3], and by other jurisdictions interested in least cost future investment options. RESOLVE co-optimizes investment and dispatch over a multi-year horizon with one-hour dispatch resolution for a study area, in this case the South African electricity grid. RESOLVE solves for the optimal investments in potential new resources, such as renewable resources, various energy storage technologies, and new coal, gas, and nuclear plants, subject to various different constraints on the system. These can include annual constraints on delivered renewable energy that reflects an RPS policy, a capacity adequacy constraint to maintain reliability, constraints on operations that are based on a linearised version of the classic zonal unit commitment problem as well as customised constraints by scenario or sensitivity, such as modeling the efficiency loss of the coal fleet when offering instantaneous reserves. We use the RESOLVE model in this study to investigate sets of potential future conditions in the South African system and their favorability for storage as an economic resource in offering grid services. RESOLVE finds the optimal deployment of storage depending on its price point against other resources that can offer the same services. The grid services considered within the RESOLVE framework are included in Table 1.
Table 1: Grid services that storage can offer in RESOLVE. Use Case Description Bulk Energy Services Energy arbitrage Peaking capacity Ancillary Services Instantaneous reserves Regulating 10-minute Storage costs Our storage cost projections are from a survey of the literature, including low, mid, and high price forecasts. We include a 15% adder on top of the capital costs (shown in Table 2 for the mid case) for engineering, procurement and construction (EPC) costs, and interconnection. We model replacement of the lithium ion battery pack in year eight and replacement of the flow battery and lithium ion battery power conversion system in year 10. Inverter replacement for both types of system is modeled in year 10. Replacement costs are assumed to be equal to the capital costs of the replacement item in the year of replacement (not including the 15% adder). Table 2: Energy storage mid case cost assumptions by technology [4]. Type Cost Metric 2015 2030 Lithium Ion Battery Storage Cost ($/kwh) 375 183 Power Conversion System Cost ($/kw) 300 204 (16 year lifetime, Fixed O&M Battery/Reservoir ($/kwh-yr) 7,5 3,7 92% efficiency) Fixed O&M PCS ($/kw-yr) 6,0 4,1 Flow Battery (20 year lifetime, 84% efficiency) Storage Cost ($/kwh) 700 315 Power Conversion System Cost ($/kw) 300 204 Fixed O&M Battery/Reservoir ($/kwh-yr) 14,0 6,3 Fixed O&M PCS ($/kw-yr) 6,0 4,1 Cases Our analysis looks at three different potential future IRP cases agreed upon by the study steering committee. These include the 2010 IRP base case [5], the updated 2010 IRP base case [6] and the updated 2010 IRP Rooftop PV case. The assumptions for each case are shown in Table 3. Table 3: Different cases modeled on funding gap analysis. Input 2010 IRP Base Case Updated 2010 IRP Base Case Load 2010 hourly IRP base case Hourly adjusted load forecasted out to 2030 Greenshoots load forecasted out to 2030 Renewable generation data Forecasted renewable generation capacity Generator fleet characteristics CSIR REDZ and EIA scenario wind and solar production shapes 2010 IRP build out of renewables by 2030 2010 IRP generator fleet, including additions and retirments, out to 2030. CSIR REDZ and EIA scenario wind and solar production shapes 2013 IRP update build out of renewables by 2030 2013 IRP update generator fleet, including additions and retirments, out to Updated 2010 IRP Rooftop PV Hourly adjusted Greenshoots load forecasted out to 2030 CSIR REDZ and EIA scenario wind and solar production shapes 2013 IRP update Rooftop PV case additional 21600 MW of rooftop PV by 2030 2013 IRP update generator fleet, including additions and retirments, out to
New generator characteristics Intertie characteristics System ancillary service requirements Instantaneous reserves value Fuel prices Generic operating characteristics by technology New generator operating constraints and costs from EPRI IRP report from 2012. We assume that there is no interchange with surrounding countries SA is treated as an island. Using 2016/17-2020/21 Eskom projected ancillary services in 2021. Holding 2021 reserves constant through 2030. Assumes a 0.2% efficiency loss on coal generation offering instantaneous reserves. Assumed coal plants offer 3% capacity towards instantaneous reserves. Fuel prices from 2016 CSIR LCOE model 2030. Generic operating characteristics by technology New generator operating constraints and costs from EPRI IRP report from 2012. We assume that there is no interchange with surrounding countries SA is treated as an island. Using 2016/17-2020/21 Eskom projected ancillary services in 2021. Holding 2021 reserves constant through 2030. Assumes a 0.2% efficiency loss on coal generation offering instantaneous reserves. Assumed coal plants offer 3% capacity towards instantaneous reserves. Fuel prices from 2016 CSIR LCOE model 2030. Generic operating characteristics by technology New generator operating constraints and costs from EPRI IRP report from 2012. We assume that there is no interchange with surrounding countries SA is treated as an island. Using 2016/17-2020/21 Eskom projected ancillary services in 2021. Holding 2021 reserves constant through 2030. Assumes a 0.2% efficiency loss on coal generation offering instantaneous reserves. Assumed coal plants offer 3% capacity towards instantaneous reserves. Fuel prices from 2016 CSIR LCOE model The RESOLVE model will select storage as an economic investment if the total cost of storage, including capital and operating losses, is less than the alternatives for offering the same services. The resources selected in the IRP and IRP update are sufficient to operate the system reliably to the reliability criteria defined in the IRP process. If all IRP planned resources are built through 2030, the benefits that storage can offer through different grid services are purely operational avoided investment in capital infrastructure is not possible unless the IRP were re-studied with storage as a selectable resource. To investigate the potentially higher benefits of storage when capital investment in other technologies can be avoided, we run a fourth case where the RESOLVE model can optimally select resources through 2030 to meet load while also building out the same renewable portfolio as the updated IRP rooftop PV case. In this case, the selection of storage can displace the services offered by other new resources and potentially avoid investment in them. Results The funding gap is the difference between the forecasted price in 2030 and the price at which storage adoption becomes part of the least cost resource portfolio, determined in RESOLVE. The analysis was performed for the periods 2016, 2020, 2025, and 2030, however, we focus here on 2030 given that system conditions are most favourable for storage in 2030 and forecasted storage prices are lowest. One significant driver of storage value was instantaneous reserves. The need for instantaneous reserves in South Africa varies by time of day between 500 MW and 800 MW. We model the median 650 MW for simplicity. The recent addition of the Ingula pumped hydro facility brings total pumped hydro capacity to nearly 3000 MW. It is unclear how much pumped hydro can contribute towards instantaneous reserve, but it will displace some, or all, of the instantaneous reserve opportunity that other forms of storage could take advantage of. In the following funding gap results we have bracketed this opportunity, firstly by presenting results where pumped hydro does not offer any instantaneous reserve, and secondly by presenting the funding gap where the entire instantaneous reserve need is covered by pumped storage.
The funding gap presented in the following subsections is based only on bulk system level benefits of storage. The funding gap at different levels of installed storage therefore represents the storage opportunity in South Africa: if the funding gap can be filled with benefits from use cases other than on the bulk system, storage can be a cost-effective grid solution. Additional benefits include those at the distribution and customer geographies on the system in Fig. 1. Funding gap with full instantaneous reserve opportunity Fig. 3 shows the levels of adoption in 3 different cases: The 2010 IRP case with distribution of renewable resources by REDZ The rooftop PV case using the REDZ distribution The modified rooftop PV case - the rooftop PV case where future resources do not adhere to the IRP but instead are selected by RESOLVE, allowing storage to defer capacity investments if economical. For each case, eight different price points for storage are investigated. The RESOLVE model is run at each price point to determine how many MWs of storage are part of the least cost resource portfolio to meet system needs in 2030. These prices are an estimate of the total system benefits from the marginal MW of storage installed. For example, for storage cost 1 (close to the low-price estimate for lithium ion in 2030), no storage is installed, so there is no storage that is part of the least cost resource plan, based on bulk system benefits alone, at this price. At storage cost 2, 400 MW of storage is built in the modified IRP case the cost of the marginal MW of storage built in this case is an estimate of the benefits of storage. The difference between the forecasted 2030 prices for lithium ion and storage cost 2 is the funding gap. This represents the additional benefits the 400 MW of storage would need to find from other use cases to become cost effective at 2030 forecasted prices. At storage cost 5, close to 2000 MW of storage are cost effective. The value of the additional benefits from storage, outside of bulk system use cases, will determine the size of the storage market in 2030. Fig. 3: Funding gap at different levels of storage penetration by case with instantaneous reserve opportunity.
In the following 2 subsections, we examine the bulk system use cases that storage is participating in for Case 2 - a modeled storage cost of approx. 2000 R/MWh in 2030, and Case 6 a modeled storage cost of approx. 1300 R/kWh in 2030. Case 2: Displacement of coal in offering instantaneous reserves Case 2 is the first time we see storage adoption by RESOLVE. This occurs in the modified IRP case where RESOLVE can select how to meet future system needs. The amount of renewables and pumped storage are maintained at the same level of additions as the updated IRP rooftop PV case. The RESOLVE model selects to build gas instead of coal or nuclear. This is subject to the pricing used in the model, and other constraints on gas availability not captured here. However, storage is part of the optimal portfolio, with approximately 400 MW of storage being built. Battery additions are made in 2030. The build decisions are shown in Fig. 4, where retirements are shown on the negative axis in the textured areas. Fig. 4: Additions and retirements 2016 to 2030. Fig. 5 shows an example day s dispatch. Pumped hydro is being used for energy arbitrage, pumping during high solar production in the middle of the day and discharging in the evening. However, pumped hydro is prevented from offering instantaneous reserves in these model runs. The cheap capital gas generation is used in the evenings to meet peak loads, reducing commitment and cycling of the coal fleet. Fig. 5: Case 2 Example daily dispatch.
The battery energy charge and discharge cycle is so minimal that it is not visible in the above chart. Fig. 6 shows the average battery behaviour over all modeled days. Clearly, the predominant use case of the battery is to displace the coal fleet offering instantaneous reserves. In addition to offering reserves, the batteries are used to displace peak capacity in the morning and evening. Because this case allows resource selection, the battery realizes the value of avoided capital expenditure, thus batteries are built at a higher price point than in the IRP cases (as shown in Fig. 3). Fig. 6: Case 2 Battery storage average applications. In price point cases 3 and 4 batteries offer a similar service to case 2 in the modified IRP case. This is evident from the funding gap chart where the batteries built in these two cases are capped at 650 MW the total market size for instantaneous reserves. Case 6 is the first time that storage is economically selected in the 2010 IRP case. It builds up to 650 MW in 2030 to displace coal offering instantaneous reserves, combined with energy arbitrage. Selection of storage for instantaneous reserves is at a lower price point for the 2010 IRP case because storage does not have the opportunity to displace capital investments, benefits instead come from operating cost reductions of the thermal fleet only. In the modified IRP, we see the price point becoming low enough that storage can compete with the costs of curtailing renewable resources. Curtailment in 2030 drops from 4,1% to 3,3% as economic energy arbitrage becomes a cost-effective use case. The example daily dispatch (Fig. 7) shows battery storage operating with pumped hydro to provide energy arbitrage and avoiding additional peak capacity. Fig. 7: Case 6 Example daily dispatch.
The storage average dispatch chart (Fig. 8) shows that batteries are being used for both upward reserves and energy arbitrage. The higher amount of upward reserves, above the 650 MW of instantaneous reserves, shows that storage is being used for regulating reserves as well. Fig. 8: Case 6 Battery storage average applications. Funding gap with no instantaneous reserve opportunity The same analysis as above is presented below for the case where pumped hydro is assumed to be flexible enough to offer the entire instantaneous reserve need of the system. In this case we find that storage previously found economical at relatively small funding gaps is no longer selected by the model. Given the number of MWs of pumped storage in South Africa, the opportunity to realise instantaneous reserve benefits for other forms of storage is likely to be significantly diminished. Fig. 9: Funding gap at different levels of storage penetration by case without instantaneous reserve opportunity. The first price point at which storage becomes economical in the modified IRP case is storage cost 5, where storage can now competitively displace energy and capacity services. Again, the other two cases see very limited adoption of storage, showing that the cost effectiveness of storage depends on the ability to displace capacity
investments in other types of generation. Economic storage is therefore best identified during the IRP process as a candidate for a least cost portfolio resource solution. These results show that offering instantaneous reserves drives cost effective storage at higher storage price points. The ability of the hydro fleet to offer instantaneous reserves is therefore of significant importance. The instantaneous reserve requirement varies between 500 MW and 800 MW depending on the time of day. If pumped hydro can offer at least some of this requirement, it will displace cost-effective storage driven by this use case. Closing the funding gap Closing the funding gap to reach economic storage adoption can happen in three ways: The first is that the price of storage drops to levels where adoption is economic based on bulk system benefits only. This is unlikely given that the prices triggering cost-effective storage in our analysis are lower than the low-price forecast scenario. One caveat to this is if the levels of bulk system benefits increase relative to what is expected with the modeled IRP. The second is that system conditions are different from our base assumptions, or policy changes in the future to favour storage, i.e. raising the bulk system benefits of storage. We present an investigation of a limited set of these changes in the sensitivity case results in the main report, but none of them gave enough value to storage for economic adoption with bulk system benefits only. We did not test significantly higher renewable penetrations, however a driver of storage adoption in other regions. Given the precipitous decline in renewable costs and the revision of the South African IRP both now and in the future, this could be a possibility. Renewable integration needs, and the resulting larger diurnal energy price differentials are a significant driver of storage value in regions with greater renewable penetrations. The third is that storage can realise other benefits not captured at the system level. These include local use cases that can only be realised with careful placement of the devices in high value areas. This is the most promising near-term. The funding gap at different price points and system assumptions can be thought of as the opportunity for storage: if additional benefits from other use cases can be found that equal or exceed the funding gap, storage can be cost effective. Another way of showing the funding gap analysis from Fig. 3 is shown below in Fig. 10. The MWs of storage in the least cost IRP portfolio would increase with increasing levels of additional benefits. Fig. 10: MWs of storage in least cost portfolio at different levels of additional realised benefits.
This opportunity can be determined by asking several questions: Where are the highest value locations for storage on the distribution system? How much benefit can be realised from local storage placement? What are the barriers to realising these benefits? In the next section of this paper we discuss qualitatively the additional benefits that storage can realize when placed on the distribution system and how they might be determined. Additional benefits of storage Overview of local and customer storage benefits To describe the potential local and customer storage benefits, we start with the generalised distribution system schematic in Fig. 11. The distribution system depicted is served by a high voltage transmission system, a medium voltage sub-transmission system that serves multiple distribution substations, and distribution feeders serving loads. The distribution system is radial (as opposed to networked) in this example, and uses switching to provide for contingencies and maintenance. Fig. 11: Distribution system schematic with local benefits. Within this framework, distributed energy storage systems can potentially provide local and customer benefits if the appropriate telemetry and controls are in place. These benefits could include: Local capacity investment savings (transmission, sub-transmission, and distribution) Distribution voltage/var regulation savings Customer retail rate savings, including arbitrage and demand charge peak shaving Customer back-up services There may be conflicts between the needs to operate storage for each potential value stream (system, local, and customer, as shown in Fig. 1) that would result in value reductions for one or more of the value streams. The use case for each storage system, that is, how the battery is used, will therefore determine the total benefits that the storage can provide. Table 4 lists those benefits that are potentially compatible, or mostly compatible within the same storage system, for different storage system owner and operators and storage interconnection locations. The owner and operators fall into four cases: Eskom as an owner and operator for distribution system interconnected storage. A public distribution utility that purchases wholesale energy and capacity. Customer owned storage under existing retail rates.
Customer owned storage under reformed rates. In each case the primary use of the storage system is identified along with an indication of whether additional benefits might be provided by storage. Table 4. List of compatible benefits by owner and operator/location Storage System Owner & Operator ESKOM Vertical Utility Public Distribution Utility Customer owned (existing rates) Customer owned (reformed rates) Location Distribution Distribution Customer Customer Production costs Primary Yes Generation capacity Primary Yes Ancillary services Yes, with limits Yes, with limits Transmission capacity Yes Yes Distribution wholesale utility Primary Sub-transmission capacity Yes Yes Yes Distribution capacity Yes Yes Yes Voltage / VAR regulation Yes Yes Yes, with limits Yes Customer retail savings Primary Primary Customer back-up No No Yes As the results in the previous section show, the primary bulk system benefit for Eskom from storage would be the efficiency gains from taking coal generation off instantaneous reserve duty, combined with energy arbitrage. The potential for this service is capped at the requirements for instantaneous reserves. Storage is also physically limited in offering reserves depending on state of charge. In the future at higher renewable penetrations, renewable integration may be the primary system value. At the local level, the same system can also provide transmission, sub-transmission, and distribution capacity. These services generally require less than 100 hours per year, freeing the battery to provide bulk system benefits for the remaining majority of annual hours. To the extent that the peak loads are not coincident, there may be additional hours required to be reserved for local capacity. In some situations, it may be necessary to choose between providing distribution capacity and sub-transmission or transmission capacity which will make it impossible to capture all of the capacity benefits. If equipped with the proper inverter and controls, this storage system could also help regulate local voltage and power factor which can reduce losses, improve power quality, and reduce wear and tear on distribution equipment. Public distribution utility owned and operated battery storage can provide the local utility the same local capacity benefits as well as voltage and VAR regulation as a vertical utility. The difference between the value of storage between this structure and a vertically integrated utility is that, to the extent that the wholesale pricing or utility tariffs do not align with the system renewable integration needs or regulation, then there is no way to capture all of the bulk system level benefits. The primary benefit of customer owned and operated energy storage is reducing the utility bills by arbitraging the retail rate. For storage on existing retail rates, which are only somewhat aligned with system and local needs, the operation of the storage system is unlikely to provide much system benefit. Under reformed rates that reflect system and local capacity benefits and time-dependent marginal costs of system operation, the customer
optimal storage dispatch pattern and the utility value can be more closely aligned and thereby provide additional benefits. Grid infrastructure benefits Integrating storage and the necessary enabling technologies and systems into the set of options available to planners to address their local area needs offers the potential to lower total utility costs. Using the distribution system schematic in Fig. 12, we show three possible upgrades to the capacity of a local distribution system in response to load growth. At [A] it is possible for the distribution planner to select from one or more capital investments in T&D infrastructure which is the standard industry practice. At [B], the utility could install utility owned DER interconnected to the distribution system, such as fixed or mobile storage system. At [C], the utility could encourage customer side energy storage through incentives, demand response style capacity programs and/or time-varying retail pricing. Fig. 12: Distribution system schematic with different storage locations. Value of deferral Deferral of new distribution infrastructure projects is often a high value use case for storage. Deferral value is the difference in net present value between the revenue requirement of the local area investments before and after storage is installed. After the storage is deployed, the capital investments needed to meet load growth may be delayed to a future date. Since the cost of borrowing is generally lower than inflation, a delayed investment will result in a lower net present value cost to ratepayers which results in a savings. Storage devices sited in local load pockets where load growth is expected to cause significant investment costs in new infrastructure can realise high benefits from being dispatched for local load reduction. Dispatch of storage for local deferral value can often be added as a use case without significantly reducing the bulk system level benefits of storage operated for energy, capacity and reserves. An example local load shape from California below shows why this can be the case. Peak load conditions happen in relatively few hours over the course of the year. In this particular area, peak loads can be reduced 10% by action in only 30 hours over the course of the year. Significant peak load reduction, and thus potential deferral of load growth related infrastructure can be realised with storage operating for load reduction in only a small fraction of total annual operating hours.
Fig. 13: 400 highest load hours for an example local area in California. Numerous studies have been done in the US and abroad that have characterised the distribution marginal cost using this approach for the purposes of local integrated resource planning. These use utility distribution capacity investment plans, and associated local load growth, to compute deferral value. Fig. 14 shows the result of a 2012 study in California for the three largest investor owned utilities (PG&E, SCE, and SDG&E) [7]. The marginal costs of investment deferral by distribution substation in California are sorted from highest to lowest. Other similar studies show similar patterns. In this case, there are a few areas that show very high distribution marginal costs, a fair number that are moderate, and many that are very low. The highest cost distribution areas tend to be those that require near term expensive upgrades but have relatively low levels of load growth. The moderate areas are those that have necessary upgrades but also have significant planned growth. The low areas are those without capacity constraints in the near future. Fig. 14: Range of distribution marginal capacity costs in california. Depending on the combination of projected load growth and required capital investments by local area in South Africa, there may be a number of high distribution avoided cost opportunities for storage. A two-hour duration mid-case lithium ion battery is priced at $240/kW-yr in our study for 2015. Even at this high value, there are two identified pockets in California where deferral value is high enough to justify investment, assuming the battery can continue to receive the same deferral value over its lifetime. If the battery can recover an additional $100/kW-yr from bulk energy services, several more areas in the above chart would become potentially economic storage investment locations. Searching for these high value storage locations where storage can offer both bulk energy services and local value is the key to developing economic storage.
One caveat is analogous to the bulk system level analysis presented in the first half of this paper. While storage may offer large benefits to the grid in the local pockets above, it competes against many other technologies, just as on the bulk system, to offer the same services. For example, the most expensive local capacity investments in the above chart could be deferred more cheaply using demand response programs than storage investments used for the single use case of local deferral. Economic storage is therefore best identified through integrated resource planning that incorporates both bulk system and local level resource selection. In the demand response example above, for instance, storage may be competitive because it can defer capacity, avoiding the investment in demand response programs, while also realizing bulk system benefits. Where are the high value areas? To capture the distribution capacity value, it is necessary to reform the distribution planning process to screen for the potential to defer and save on distribution capital investments. For those areas of consideration, there are generally four screening steps that must be considered; engineering solution, timeline available for deployment, potential capital expenditures that can be avoided, and cost-effectiveness. Each of these are described in more detail below. Engineering solution: does the energy storage device solve the problem that is driving the need for capital expenditure? Many distribution capital projects are designed to replace aging equipment, improve safety, improve reliability, or accomplish other goals. Generally, a good test is to ask if peak load reduction would eliminate the need for the investment. If it would, then the area is a positive candidate for deferral. Timeline available for deployment: is there enough time to deploy sufficient numbers of energy storage devices and enough storage capacity before the investment must be committed to in order to construct or order transformers or other equipment? Often, there is a very short period of time to deploy storage which presents a significant challenge. Capital expenditures: how large are the capital expenditures that storage can defer? This drives the value of deferral, so areas with inexpensive upgrades do not provide significant benefits. Cost-effectiveness: is the value of distribution deferral, when combined with other benefits that can be captured, enough to justify the costs of the storage system? Beyond identifying the opportunities for local deferral, planning tools and processes are necessary to: determine what the type and location of least cost resource investments are on the distribution system, and to determine how to pay for the services they provide, either through long term contracts, market products, or incentive design or tariff reform on the customer side. Key findings Part of the South Africa Energy Storage Technology and Market Assessment was to identify actions to support the development of energy storage in South Africa. This section summarises various strategies to do so. These are discussed in detail in the full report [8]. The categories and subcategories of storage supporting actions are shown in Fig. 15.
1 Global and SSA Energy Storage Market 2 SA Value Chain Opportunities 3 Level Playing Field 4 Transformative Policy 5 Demo Projects 6 Build Infrastructure 1.1 Analysis of global energy storage market 2.1 Analysis of HP/ ID technology value chain 3.1 Internal Value Grid Service 4.1 Procurement Targets 5.1 Tools & analysis to ID high value sites 6.1 Standardize Energy Storage Requirments 1.2 Analysis of SSA energy storage market 2.2 Match value to SA Capabilities 3.2 Market Products for Grid Services 4.2 Create Incentives 5.2 FTM Substation Demo 6.2 Encourage IT Development 1.3 Identify High Potential (HP) technologies 2.3 ID SA industry development priorities 3.3 IRP Reform to incl full value of storage 4.3 Provide Tax Credits 5.3 Behind the Meter Demo 6.3 Allow Utility Control 3.4 Tariff setting and cost recovery 5.4 Community Storage w/ Aggregator Demo 6.4 Communication Network Fig. 15: Roadmap of actions that would support storage market development in South Africa. Those categories related to the economics of storage as discussed in this paper are summarised below. Level the playing field The value proposition for storage relies on stacking benefits from offering multiple grid services. Some of those benefits are not explicitly valued by Eskom for third party providers to offer. These include ancillary service products that Eskom identifies as required for grid reliability. By not valuing ancillary services, Eskom cannot evaluate different resources against one another when procuring for grid needs. Third parties cannot receive benefits for offering them, missing out on a potentially lucrative income for storage, or at least have significant uncertainty about the value they will receive for their services. Uncertainty in value can dampen market interest. Reforming the IRP planning process to consider storage as an alternative resource to thermal generation will increase the opportunity for storage and ensure lowest cost resource procurement. The IRP should factor in the additional benefits of storage from distribution and customer levels of the system, and consider the customerlevel adoption of storage if bulk system-level and distribution-level benefits were available to customers. All grid resources should be evaluated on a level playing field to ensure least cost procurement. Transformative policy The adoption of energy storage in South Africa may have benefits and value beyond the economic case. These benefits might be in the education and experience gained in various industries associated with energy storage (manufacturing, installation, operation). In this case, South Africa may want to develop policy in the form of incentives for procurement targets to build this experience base prior to when energy storage becomes independently cost effective. These activities would be analogous to programs adopted in other countries to promote/incentivize energy storage adoption. The education and experience gained may result in better storage value over time as installers gain in efficiency, control systems and programs to offer storage services evolve to offer the grid higher value, and local storage manufacturing, if part of the transformational program, gains economies of scale.
Demonstration projects Identifying locations where energy storage demonstration projects would be effective in demonstrating the value of energy storage for South Africa and sub-saharan Africa. Demonstration projects can be designed to serve two main purposes: highlighting the challenges and solutions to siting, constructing, interconnecting, controlling, and compensating storage for the services it provides in different settings and business models on the electrical system; and identifying opportunities that are economically viable or close to economically viable in the near and medium term for storage. The highest value locations for storage will be where transmission and distribution infrastructure investments can be deferred or even avoided through use of the storage device, as described in Section 0. Business cases where storage will make the most economic sense will combine multiple different use cases from the same device. These use cases include for example, system energy and capacity services, ancillary services, both locally and at a system level, and local infrastructure deferral. It is up to the entity controlling the storage device to dispatch it such that the maximum value can be realized. However, depending on regulatory structures, tariff designs, control systems, and electricity market product availability, these entities may not have enough information to dispatch the device for highest value, or their incentives to do so may conflict with those of the system as a whole. Demonstration projects are way of testing different control regimes, infrastructure, incentives and business models to tailor an approach to cost effective storage development. Next steps The near-term priority for identifying storage value in South Africa is to identify the highest value sites for storage on the distribution system, taking advantage of the full value stack of use cases that storage can offer. These locations are the best near-term opportunities for storage, and ideal for demonstration projects that will achieve cost effectiveness. Identifying these sites will require a set of distribution system analysis tools paired with bulk system level value stream analysis. These can be applied locally by Eskom or municipalities for tailored storage implementation. References [1] Parsons Corporation for the United States Trade and Development Agency, 2016, South Africa Energy Storage Technology and Market Assessment, TDA-IE201511210, 2015-11032A. [2] http://www.cpuc.ca.gov/irp/prelimresults2017/ [3] https://www.hawaiianelectric.com/about-us/our-vision [4] We used a variety of sources to develop mid, high and low case cost projections for storage, including Lazard s Levelized Cost of Storage Analysis version 1.0 and others. Lazard, (2015), Levelized Cost of Storage Analysis - Version 1.0. https://www.lazard.com/media/2391/lazards-levelized-cost-of-storageanalysis-10.pdf [5] Integrated Resource Plan for Electricity 2010-2030. Government Gazette Staatskoerant No. 9531, Vol. 551. 6 May 2011. [6] Integrated Resource Plan for Electricity (IRP) 2010-2030, Updated Report 2013. 21 November 2013. [7] Energy and Environmental Economics, 2012. Technical Potential for Local Distributed Photovoltaics in California Preliminary Assessment, www.cpuc.ca.gov/workarea/downloadasset.aspx?id=5912 [8] Parsons Corporation for the United States Trade and Development Agency, 2016, South Africa Energy Storage Technology and Market Assessment, TDA-IE201511210, 2015-11032A. Contact Jeremy Hargreaves, E-Three, jeremy@ethree.com