Valuing Energy Storage as a Flexible Resource...Valuing Energy Storage as a Flexible Resource –...

Valuing Energy Storage as a Flexible Resource

Final Phase 1 Report for Consideration in CPUC A. 14-02-006 June 19, 2014

© 2010 Copyright. All Rights Reserved.

Energy and Environmental Economics, Inc.

101 Montgomery Street, Suite 1600

San Francisco, CA 94104

415.391.5100

www.ethree.com

Valuing Energy Storage as a Flexible Resource Final Phase 1 Report for Consideration in CPUC A. 14-02-006 June 19, 2014

Table of Contents

1 Introduction ............................................................................................ 1

2 Overgeneration is a key challenge ..................................................... 3

2.1 Flexibility challenges............................................................................. 3

2.2 How soon will overgeneration occur? ............................................... 5

2.3 Magnitude of overgeneration .............................................................. 7

2.4 Increase in overgeneration under high RPS scenarios ................. 9

2.5 Frequency of overgeneration (under 40% RPS) .......................... 10

2.6 Duration of overgeneration................................................................ 12

3 Renewable Curtailment........................................................................ 13

3.1 Managing overgeneration has significant value ............................ 14

3.2 Energy storage can reduce renewable curtailment ...................... 16

4 Overgeneration requires long-duration solutions ........................... 17

5 System level cost-benefit analysis is crucial ................................... 19

5.1 Comparing value of short- and long-duration ................................ 19

5.2 Small market for frequency regulation ............................................ 20

5.3 Serial dispatch of short-duration ...................................................... 22

5.4 Production simulation understates flexibility need and value ..... 22

6 Benefits of a diverse portfolio ............................................................ 24

7 Conclusions .......................................................................................... 25

Appendix A: Cost of Renewable Curtailment .......................................... 27

7.1 Avoided RPS Costs ............................................................................ 27

Interim Report

P a g e | 1 |

Introduction

© 2010 Energy and Environmental Economics, Inc.

1 Introduction

Energy and Environmental Economics (E3) produced this report to briefly summarize key

issues that are critical to consider when analyzing the potential benefits of energy

storage. E3 is pleased to provide this report for parties to consider in the California

Public Utilities Commission (CPUC) preceding considering the applications of PG&E, SCE

and SDG&E (A. 14-04-006) for approval their respective energy storage procurement

framework and programs as required by D. 13-10-040. This document, dated June 19,

2014, contains some revisions and added material to the interim report filed on June

12th. The overall conclusions remain unchanged.

The impetus for this report grew out of discussions between E3 and several parties to

the CPUC proceeding after the June 2, 2014 Workshop on IOU Energy Storage

Procurement Applications and during the 2014 Annual Energy Storage Association

Conference in Washington D.C. June 3-6, 2014. We discussed the methods described for

utility evaluation of energy storage project proposals and for the Common Evaluation

Protocols. Throughout the week we shared our thoughts about how those methods may

not fully capture the value of energy storage with the higher penetrations of renewable

generation anticipated and planned for in California.

In January 2014 E3 published the report “Investigating a Higher Renewables Portfolio

Standard in California” on behalf of PG&E, SCE, SDG&E, SMUD and LADWP (Utility High

RPS Report).1 Using E3’s stochastic production simulation model REFLEX, E3 quantified

the flexibility needs of the California grid under 40 and 50% RPS scenarios.2 REFLEX is

specifically designed to investigate flexible capacity needs and value with variable

renewable resources (VER). REFLEX performs random draws of weather-correlated

1 “Investigating a Higher Renewables Portfolio Standard in California”, Energy and Environmental Economics, January 2014. https://ethree.com/documents/E3_Final_RPS_Report_2014_01_06_with_appendices.pdf 2 See https://ethree.com/public_projects/reflex.php

Valuing Energy Storage as a Flexible Resource – Interim Report for Consideration in CPUC A. 14-02-006

P a g e | 2 |

load, wind, solar and hydro conditions taken from a very large sample of historical and

simulated data. It characterizes the need for system ramping capability through

stochastic treatment of load, wind and solar generation, hydropower conditions,

dispatchable generator outages and other random variables on multiple time scales:

annual, monthly, diurnal, hourly and sub-hourly. The model uses optimal unit

commitment and economic dispatch to model the ability of the system’s dispatchable

resources to respond to a full range of conditions. Flexibility violations such as

shortages in upward or downward ramping capability are characterized according to

their likelihood, duration and depth, using metrics that are analogous to conventional

reliability metrics such as LOLP, Loss of Load Probability Expectation (LOLE), and

Expected Unserved Energy (EUE).

REFLEX employs an economic framework to evaluate the costs and benefits of

investments in flexible resources. In order to determine whether investments in new

flexible resources are cost-effective, flexibility violations are assigned cost penalties

based on their economic value. New resources are then tested for their ability to

prevent flexibility violations and avoid the associated cost penalties. The capital and

operating costs of new flexible resources are compared to the value of flexibility

violations they avoid, ultimately identifying a least-cost portfolio of new resources. This

framework can be used to evaluate flexible resources such as combustion turbines,

reciprocating engines, energy storage, or demand response, as well as changes in

operating procedures such as improved forecasting, participation in a regional market,

or renewable curtailment.

The Utility High RPS Report and subsequent work using the REFLEX model have

produced results demonstrating how specific costs, benefits and input assumptions can

dramatically impact the valuation of energy storage as a flexible resource. This report

presents several of the more important findings for your consideration as the CPUC

Interim Report

P a g e | 3 |

Overgeneration is a key challenge


reviews the IOU energy storage applications in A.14-04-006. The overarching themes of

this report are that a cost-effectiveness framework for energy storage must:

Include not just existing markets and avoided costs, but also the future benefits

of reducing renewable curtailment under higher RPS levels.

Describe how the relative costs, benefits and tradeoffs of short- vs. long-

duration solutions will be quantified and evaluated.

Include system level and portfolio costs and benefits.

2 Overgeneration is a key challenge

2.1 Flexibility challenges

The Utility High RPS Report models CAISO system flexibility needs under 33%, 40% and

50% RPS levels The report describes five distinct types of flexibility challenges that the

system will face under high renewable penetration and shows a sample operating day in

January that illustrates these five challenges:

1. Downward ramp: as solar generation increases in the morning, flexible resources will be needed to ramp generation down (or ramp load up).

2. Minimum generation: to accommodate solar generation during the day, fossil generation will need to turn off, or operate at minimum levels, but still be ready to increase generation in the late afternoon and early evening.

3. Upward Ramp: in the evening, as solar generation declines, other generating resources will need to ramp up (or load ramp down).

4. Peaking Capacity: sufficient resources will be needed to meet peak loads with sufficient reserve margins.

5. Sub-hourly Flexibility: flexible resources will be required to provide both existing and new types of ancillary services including frequency regulation, flexi-ramp and load following.


P a g e | 4 |

Figure 1: Renewable Integration Challenges

The Utility High RPS study models flexibility needs in high RPS scenarios in 2022 and

finds that the largest renewable integration challenge is “overgeneration”.

Overgeneration occurs when “must-run” generation—non-dispatchable renewables,

combined-heat-and-power (CHP), nuclear generation, run-of-river hydro and thermal

generation that is needed for grid stability—is greater than loads plus exports.

Overgeneration can occur even in a highly flexible power system if there is simply not

enough load to absorb the available quantity of renewable energy during a given hour.

However, additional overgeneration or curtailment of renewable output may occur due

to lack of power system flexibility. As an example of this, consider a situation where

limited upward ramping capability would prevent the system operator from meeting the

steep upward ramp that occurs after sundown on the day depicted above. One strategy

for addressing this situation is to curtail renewable resource output during the middle of

the day, thus reducing the magnitude of the upward ramp to a manageable level and

avoiding firm load curtailment.

Interim Report

P a g e | 5 |



2.2 How soon will overgeneration occur?

While there is currently no legislated RPS requirement above 33%, there are several

reasons overgeneration is likely to occur at significant levels before 2020:

Renewable procurement is on a trajectory to hit 40% levels: Even absent a

legislative requirement, procurement is on track to exceed 33% in 2020. Project

failure in recent solicitations has been much lower than anticipated based on

prior experience. Large declines in PV prices have also accelerated procurement

outside of IOU RPS solicitations.

Statewide model without transmission constraints: The production simulation

case modeled in REFLEX did not include transmission and associated constraints

that would increase overgeneration challenges.

Solar development is concentrated in Southern California: Solar project

development is heavily weighted to Southern California. The South of Path 15

(SP15) zone will reach 40% RPS generation levels and experience overgeneration

much sooner than the state as a whole.

Investment Tax Credit: Most of the solar projects planned are endeavoring to

begin operation before the end of 2016 to ensure their eligibility for the Federal

Investment Tax Credit.

Production simulation tends to overstate system flexibility: The specific ways

in which flexibility can be overstated are described below. E3 took steps to

constrain hydro generation and imports to realistic levels. However, the model

does assume all fossil generation can be dispatched by the CAISO within

operating constraints. In reality, self-scheduled generation may not be readily

available for flexible dispatch by the CAISO.

Indeed, negative prices due to overgeneration have already occurred in California, in

advance of even 33% RPS. Figures 2-4 show total generation, renewable generation and

SP-15 prices for March 6, 2014. Figure 2 shows that the thermal units are ramped down


P a g e | 6 |

in the middle of the day to accommodate ~3,000 MW of solar generation (Figure 3). This

leads to several intervals with negative prices between HE 11 and HE 17 (Figure 4).

Figure 2: CAISO March 6, 2014 – Generation by resource type

Figure 3: CAISO March 6, 2014 – Renewable generation

Interim Report

P a g e | 7 |



Figure 4: CAISO March 6, 2014 – SP-15 locational marginal price (LMP)

2.3 Magnitude of overgeneration

E3’s Higher RPS Study finds that overgeneration is pervasive at RPS levels above 33%,

particularly when the renewable portfolio is dominated by solar resources. This occurs

even after thermal generation is reduced to the minimum levels necessary to maintain

reliable operations. Overgeneration is most pronounced in the spring, when loads are

relatively low, hydro generation is peaking and solar generation is substantial.

Figure 5 shows an April day in 2030 under the 33% RPS, 40% RPS, and the 50% RPS Large

Solar Scenarios on which the system experiences both low load conditions and high

solar output. A very small amount of overgeneration is observed at 33% RPS. The 40%

RPS Scenario experiences over 5,000 MW of overgeneration, while the 50% RPS Large

Solar Scenario experiences over 20,000 MW of overgeneration.


P a g e | 8 |

Figure 5: Generation mix calculated for an April day in 2030 with the (a) 33% RPS, (b) 40% RPS, and (c) 50% RPS Large Solar portfolios showing overgeneration

Interim Report

P a g e | 9 |



2.4 Increase in overgeneration under high RPS scenarios

Table 1 and Figure 6 show overgeneration statistics for the 33%, 40% and 50% RPS Large

Solar Scenarios. In the 33% RPS scenario, overgeneration occurs during 1.6% of all

hours, amounting to 0.2% of available RPS energy.3 In the 50% RPS Large Solar case,

overgeneration must be mitigated in over 20% of all hours, amounting to 9% of available

RPS energy, and reaches 25,000 MW in the highest hour.

Table 1: 2030 Overgeneration statistics for the 33%, 40% and 50% RPS Large Solar Scenarios

Overgeneration Statistics 33% RPS 40% RPS 50% RPS

Large Solar

Total Overgeneration

GWh/yr. 190 2,000 12,000

% of available RPS energy 0.2% 1.8% 8.9%

Overgeneration frequency

Hours/yr. 140 750 2,000

Percent of hours 1.6% 8.6% 23%

Extreme Overgeneration Events

99th Percentile (MW) 610 5,600 15,000

Maximum Observed (MW) 6,300 14,000 25,000

3 Curtailment as a percentage of available RPS energy is calculated as: overgeneration divided by the amount of renewable energy that is needed to meet a given RPS target.


P a g e | 10 |

Figure 6: Increase in overgeneration with increasing renewables penetration

2.5 Frequency of overgeneration (under 40% RPS)

E3 analysis performed since the Utility High RPS Report was published shows that

overgeneration is persistent throughout the year under 40% RPS scenarios. Figure 7

shows an illustrative dispatch of resources, including ~1,500 MWs of storage, on a

flexibility constrained spring day in the REFLEX model. The model dispatches all available

resources within their defined operating parameters to meet load at the lowest possible

cost. In this example day, the storage is charging to reduce overgeneration during the

day and discharging in the early evening to reduce the dispatch of a combustion turbine

or demand response (DR) to meet peak needs. Fossil resources (Steam Turbine (ST), Gas

Turbine (GT) and combined cycle gas turbines (CCGT)) are reduced to minimum load

during the day, but must be online to meet the evening ramp and peak net load.4 Other

resources, including hydro and imports are also reduced to minimum loads during the

day. Combined heat and power (CHP) and nuclear are assumed to be non-dispatchable

base load resources.

4 Net load is the total system load minus the non-dispatchable variable energy resources, including wind and solar generation.

Interim Report

P a g e | 11 |



Figure 7: Illustrative dispatch energy storage on flexibility constrained day

Figure 8 shows that overgeneration occurs throughout the year. As described above,

overgeneration is most prevalent in the spring, with average overgeneration of over

4,000 MW near solar noon in March. Overgeneration occurs in significant quantities in

the fall and winter as well. Only when loads peak in the summer is overgeneration

minimal. Peak net loads occur in the evening and either DR or combustion turbines (CTs)

are dispatched during peak net load hours in the early evening. These resources are

dispatched not to meet peak loads, but to respond to the substantial ramp in net load

that occurs between HE 15 and HE 17.


P a g e | 12 |

Figure 8: Average renewable curtailment and demand response/CT dispatch throughout the year

2.6 Duration of overgeneration

Many prior studies have focused on the hourly and sub-hourly variability in renewable

generation to quantify operational renewable integration needs. The Utility High RPS

Study examines the integration challenge from both a long-term planning and short-

term operational perspective. With this approach, overgeneration emerges as a primary

challenge requiring longer-duration solutions. Referring back to Figure 5, on the

particular spring day presented, overgeneration occurs from hour ending (HE) 09 to 17

under 40% RPS and HE 08 to 18 under 50% RPS – durations of 8 and 10 hours

respectively.

Figure 8 shows that significant quantities of overgeneration occur from hour ending (HE)

9 to 16 throughout the spring, a duration of 7 hours.

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

1 9

2 13

3 0

4

5

6

7

8 24 76 161 106 5 8

9 6 518 801 823 354 69 6 157 74 32

10 52 305 1,963 1,860 1,489 805 151 30 252 546 395 174

11 514 928 3,272 2,846 2,154 1,238 211 48 332 1,068 897 807

12 1,124 1,471 4,168 3,493 2,618 1,439 220 79 342 1,264 1,358 1,521

13 1,465 1,750 4,298 3,533 2,667 1,421 191 59 237 1,050 1,395 1,885

14 1,333 1,567 3,914 3,035 2,245 1,142 90 11 138 654 892 1,552

15 641 847 2,726 2,074 1,578 655 49 33 259 259 593

16 27 120 1,129 902 785 271 0 20

17 2 16 54 90 47 3 42 7

18 479 68 6 26 4 21 62 58 710 725

19 267 194 347 139 67 44 46 20 219 435

20 17 61 55 11 34 27 8 91

21

22

23

24Avg. Dispatched Demand Response (MW)

Avg. Renewable Curtailment (MW)

Interim Report

P a g e | 13 |

Renewable Curtailment


If curtailment of renewable generation at such high levels is to be avoided, new

strategies, resources and market designs must be developed to provide flexibility and

absorb overgeneration. These potential solutions must be available for periods lasting

up to 10 hours, during large portions of the year and must comprise a large total

capacity.

3 Renewable Curtailment

One solution to overgeneration is to curtail renewable generation. However,

curtailment may be an expensive strategy. The immediate cost of curtailment is that the

utility cannot use zero emission and marginal cost generation that has already been

contracted and paid for. Curtailing renewable generation can also make it more difficult

for utilities to achieve RPS and GHG emission reduction goals, which can impose

additional costs on the utility.

If utilities have procured resources to meet the RPS with the expectation that a certain

level of renewable energy will be delivered from these resources, frequent renewable

curtailment may increase the risk of being out of compliance in a given year. There are

two strategies for minimizing this risk: 1) the utility can procure additional renewable

resources to comply meet RPS targets; or 2) the utility can procure resources that

provide enough flexibility to ensure that energy from their renewable resources can be

delivered (such as energy storage). For a utility, the choice between these two options

will depend on the cost of procuring additional renewables versus the cost of procuring

flexible resources, as well as the incremental fuel and operating costs associated with

each option.

E3 has developed a low and high avoided curtailment value scenario to illustrate the

impact of curtailment on system costs and flexible resource value (using methods

further described in Appendix A). The low case reflects a scenario where utilities have


P a g e | 14 |

procured sufficient renewable generation to meet RPS targets even with anticipated

curtailment levels and do not need to procure additional renewables. Hence there is no

cost to the utility for replacement renewable generation. The high case presumes that

utilities must procure additional renewables to meet required RPS targets when

curtailment occurs. In the high case, the replacement cost for renewable generation is

$125/MWh, reflecting a higher levelized cost for PV that has a lower capacity factor due

to its being curtailed on a regular basis. A high cost of curtailment leads to negative

values for energy when overgeneration occurs (Figure 9). We refer here to energy value

rather than prices because the wholesale market prices for energy will not necessarily

reflect the cost of curtailment to the utility.

Figure 9: Average hourly energy value in April under 40% RPS scenario with low and high cost of curtailment

3.1 Managing overgeneration has significant value

Avoiding curtailment of renewable generation with flexible resources can provide

significant system value. Here we consider the same 40% RPS scenario as above.

Without any additional flexible resources the default strategy to prevent overgeneration

is curtailing renewable generation. Recall that the REFLEX model dispatches available

resources to meet system load with the minimal cost of energy production. Figure 10

Interim Report

P a g e | 15 |

Renewable Curtailment


shows the relative cost of two specific strategies to meet system needs - renewable

curtailment to avoid overgeneration and dispatching flexible resources such as DR or

CTs to meet peak net loads in the evening. Reducing peak load is often considered one

of the highest potential values for energy storage. In this case the total annual value of

DR or CT capacity costs that could be avoided with a flexible resource such as energy

storage is just under $100 million.5 In comparison, the cost of curtailment that can be

avoided with flexible resources is over $300 million, more than three times the capacity

value. This illustrates that avoiding renewable curtailment is a potentially large and

quantifiable value that should be included in the cost-effectiveness evaluation of flexible

resources.

Figure 10: Relative value of avoided curtailment, avoided DR/CT dispatch with high curtailment value for a 40% RPS scenario

5 With the assumptions used in this analysis, using DR or CT’s to meet peak net loads resulted in similar total annual costs.


P a g e | 16 |

3.2 Energy storage can reduce renewable curtailment

The market value for energy storage in capacity, energy and ancillary service markets

increases on a $/kW basis with an increase in duration (Figure 11). These values are

calculated using production simulation values from REFLEX model in the EPRI Energy

Storage Valuation Tool (ESVT) developed by E3. We used the ESVT to co-optimize the

dispatch of energy storage across capacity, energy and ancillary service markets and

calculate the total value provided, on a $/kW-Yr. basis, by each respective system

configuration.

With high curtailment value there are negative values for energy when overgeneration

occurs (Figure 9). Thus the energy used to charge storage doesn’t represent a cost (e.g.

the marginal cost of generating electricity), but actually provides a positive value to grid

in reducing curtailment. This increases the value by the amount shown in the avoided

curtailment bars on the right.

Figure 11: Illustrative annual revenues for energy storage of increasing duration with low and high values for avoided curtailment for a 40% RPS scenario

In this example, the hourly values from production simulation runs are used in ESVT to

calculate annual revenues with co-optimized dispatch of energy storage. However,

Interim Report

P a g e | 17 |

Overgeneration requires long-duration solutions


evaluating each storage project individually as a price taker using production simulation

results provides an incomplete evaluation of cost-effectiveness. This approach alone

does not provide a full and accurate comparison of the system level benefits that energy

storage of different durations can provide - an issue we explore further in Section 5.

Reducing renewable curtailment provides measureable value, but is not incorporated in

current market prices. Neither is curtailment value (yet) included in the Distributed

Energy Resources (DER) Avoided Cost Framework developed by E3 and adopted by the

CPUC for cost-effectiveness evaluation. The value of reducing curtailment can be

reflected in production simulation results, but only when explicitly incorporated in the

input assumptions and modeling approach.

4 Overgeneration requires long-duration solutions

Here we consider the relative ability of short-duration (2 hour) and long-duration (4

hour) storage to avoid renewable curtailment on the same flexibility constrained day

shown above. This example, for the sake of illustration, presumes that energy storage is

cost-effective when compared to alternative strategies that can reduce curtailment.

In the first example, 4,000 MW of 2-hour storage (8,000 MWh) is shown to reduce

overgeneration and DR/CT dispatch from the top down for the peak solar generation

and net load hours respectively (Figure 12). The storage is charged at full capacity in HE

12 and at partial capacity in HE 11 and HE 13. The storage is discharged during the

evening peak to eliminate DR/CT dispatch for peak net loads, and reduce some of the

fossil generation that is online to meet the evening ramp. The 2 hour duration is not

able to address the “shoulders” of the overgeneration in the morning and late

afternoon. On this spring day, 4,000 MWs of 2-hour storage can absorb ~47% of the

excess renewable generation.


P a g e | 18 |

With longer duration, a much larger portion of the overgeneration can be absorbed with

energy storage. Here 4,000 MW of 4-hour storage (16,000 MWh) is shown addressing

the challenge from the bottom up (Figure 13). The long-duration storage absorbs

overgeneration from HE 09 to HE 15, charging over 6 hours, but at the full 4,000 MW

nameplate capacity for only 3 hours. The storage discharges and displaces fossil

generation over from HE 17 to HE 22 in the evening. The 4-hour storage is able to

absorb ~90% of the excess renewable generation. With the longer duration, the storage

is also able to displace more of the fossil generation that is needed to provide upward

ramping capacity and meet peak net loads in the evening.

Figure 12: Illustrative short-duration storage dispatch

2 Hr Storage

Interim Report

P a g e | 19 |

System level cost-benefit analysis is crucial


Figure 13: Illustrative long-duration storage dispatch

5 System level cost-benefit analysis is crucial

5.1 Comparing value of short- and long-duration

Figure 8 and Figure 13 suggest that long-duration solutions will be needed to address

the majority of the overgeneration that occurs under 40% RPS levels. A portfolio level

cost-effectiveness framework will describe how the metrics and methods can be used to

evaluate the relative costs, benefits and trade-offs between short- and long-duration

storage.

While short-duration storage does have value under this scenario, our work finds that

the dominant flexibility need is for long-duration. Moreover, long-duration storage can

provide short-duration services but not vice versa. The flexible resources procured to

meet long-duration needs might also fully satisfy the short-term needs at little or no

4 Hr Storage


P a g e | 20 |

extra cost. Thus, the quantitative work that has been done to date on this topic

provides a strong value proposition for long-duration storage.

There is an additional question with respect to the impact duration will have on GHG

emissions. Flexible resources can enable reliable operation of the grid with fewer fossil

plants required to remain online at minimum load to meet evening ramps. Reducing the

number of start-ups and minimum load hours of fossil generation helps to reduce GHG

emissions from the residual fossil fleet. It is reasonable to hypothesize that longer-

duration solutions will avoid a larger number of start-ups and minimum fossil generation

hours throughout the year (as shown for the example spring day in Figure 13). On days

when fossil generation is required over periods of 4-5 hours in the evening to meet peak

net loads, it can only be avoided with longer duration storage. The magnitude of this

benefit, however, is not yet determined. The benefits of reduced fossil start-ups and

minimum operating hours is a crucial factor in evaluating the relative costs and benefits

of short- and long-duration solutions.

A cost-effectiveness framework for energy storage should specify how it will enable a

robust, portfolio level evaluation of the relative benefits and trade-offs between short-

and long-duration flexibility solutions.

5.2 Small market for frequency regulation

Frequency regulation is currently one of the most remunerative services that energy

storage can provide, but is a relatively small market. The CAISO procures on average

roughly 350 MW of regulation down and 330 MW of regulation up services.6 Even if the

frequency regulation market doubles in size with increased renewables, a modest

amount of energy storage (or flexible loads) would be sufficient to cause market clearing

prices to decline. Figure 7 above shows that the size of a 600 MW regulation up and

6 CAISO 2012 Annual Report on Market Issues and Performance, p. 118

Interim Report

P a g e | 21 |



regulation down market is small relative to overgeneration on a flexibility constrained

spring day. Furthermore, if, as the Utility High RPS Report suggests, significant quantities

of fossil generation will be operating at minimum load throughout the day to provide

ramping in the afternoon, those units could provide frequency regulation with no

opportunity cost of lost revenue in the energy market. This would depress frequency

regulation prices in the future. This finding runs counter to popular notions that the

increased demand for short-duration services such as frequency regulation will lead to

significant economic opportunity for storage resources. However, such narratives

ignore the supply side of the equation. E3’s (admittedly early) work in this area suggests

that there will be plenty of resources available to meet this increased demand, even

before considering long-duration storage procured to meet the needs of a diurnal

energy cycle.

Evaluating projects individually, without accounting for saturation effects, will overvalue

the frequency regulation benefits that can be realized with energy storage. The

potential for this result is illustrated by the EPRI “Cost-Effectiveness of Energy Storage in

California” Report (EPRI 3002001162). The study evaluated 31 energy storage use cases

and found most use cases (with several caveats) cost-effective with the storage costs

and assumptions used. Of those use-cases, 20 relied on frequency regulation for 40% or

more of their total revenue. Attributing frequency regulation benefits to each storage

project individually will overstate the total value for the system as a whole. Given the

small market size and the potential for competition to reduce market prices, the value

that can be realized for the system as a whole will, in this case, be less than the sum of

each project evaluated individually. It is therefore important to include a system level

quantification of both costs and benefits in the cost-effectiveness analysis of energy

storage to account for the effect of market saturation, which begins to significantly

affect value at relatively low levels of penetration. A system level approach would

either, a) limit the quantity of storage that is assumed to provide frequency regulation,

or b) model the impact on market prices of all storage participating.


P a g e | 22 |

5.3 Serial dispatch of short-duration

It is feasible for three 1 MW, 2-hour batteries be dispatched in serial to effectively

provide the same duration as a single 1 MW, 6-hour battery. However, this is unlikely to

be the least-cost solution. Storage systems include costs for MW of power delivery (e.g.

inverters, power electronics) and for MWh of energy storage capacity (e.g. cells,

electrolytes). The three 2-hour systems have 3 MW of power deliver costs as compared

to 1 MW for the 6-hour system.

An assessment of how the portfolio of energy storage will be dispatched for the system

as a whole is therefore crucial. If short-duration batteries are dispatched in series to

manage longer duration overgeneration, only 1 MW of power delivery is used in any

given hour, while 2 MW sit idle. There will also be associated costs, potentially borne by

the utilities and CAISO rather than the storage project developer, to interconnect,

integrate and manage the dispatch of multiple systems. This would result in a higher

system level cost that is not reflected when evaluating storage projects individually.

5.4 Production simulation understates flexibility need and value

When quantifying system level costs and benefits, it is important to bear in mind the

limitations of production simulation, which tends to overstate the flexibility of the grid

and therefore understate the need for and value of new flexible resources such as

energy storage. System level cost and benefit calculations should account for how such

limitations will impact the valuation of flexible resources and employ techniques to

counteract these biases.

WECC wide dispatch: production simulation of the WECC will optimally dispatch

all resources to minimize costs for the region as a whole. In reality, the 20+

balancing authorities in the WECC are operated independently, reducing

Interim Report

P a g e | 23 |



operational flexibility and efficiency relative to the production simulation

model.

Unrealistic hourly variations in imports: Production simulation will tend to rely

on imports and exports for flexibility to a much greater degree than is typically

observed in practice.

Unrealistic dispatch of hydro: Similarly, hydro resourced tend to be dispatched

with much greater variation that is allowed in practice, given constraints for

recreation, environmental and water supply objectives. In addition, production

simulation will count the full upward capability of hydro for operating reserves

with actual upward ramping capabilities are constrained by other factors. When

analyzing the benefits of the Imbalance Energy Market for PacifiCorp, E3 limited

hydro contributions to flexibility reserves to 12 – 25% of nameplate capacity.

Full ISO dispatch: Production simulation assumes ISO dispatch of all fossil units.

In reality the majority of fossil units are self-scheduled and not fully available or

visible to the ISO operators.

Single “snapshot” year of market conditions. Due to data processing and run-

time limitations, conventional production simulation models such as GridView

and PLEXOS only consider a single year of load and resource conditions. Most

analyses of California and the Western Interconnection use 2005 conditions as

the test year, since 2005 represents relatively average load and hydro

conditions. However, modeling a single, average year does not consider

combinations of load and hydro conditions that can result in the highest market

prices for energy and capacity (low hydro, high load) or the largest amount of

renewable curtailment (high hydro, low load).

Taken together, these issues can significantly understate the value of adding flexible

resources such as energy storage to the resource portfolio modeled in production

simulation.


P a g e | 24 |

6 Benefits of a diverse portfolio

The Utility High RPS Study provides results for a large solar case and a diverse renewable

portfolio. The large solar case has the highest levels of renewable curtailment – the

marginal PV generation will have more than half of its generation curtailed at the 50%

RPS level. With a more diverse mix of renewable resources, marginal curtailment levels

are reduced (Figure 12).

Table 2: Marginal overgeneration of different RPS scenarios

Technology 33% RPS 40% RPS 50% RPS Large Solar

50% RPS Diverse

Geothermal 2% 9% 23% 15%

Wind 2% 10% 22% 15%

Solar PV 5% 26% 65% 42%

Figure 13: Average overgeneration under 50% RPS scenarios

Interim Report

P a g e | 25 |

Conclusions


This finding illustrates how a diverse portfolio provides system level benefits relative to

one that is dominated by a single technology. This diversity benefit is also illustrated by

the history of renewable procurement. In the early years of RPS procurement, wind was

the least cost resource. In subsequent years solar thermal was thought to have the

technological cost advantage. Finally, in recent years, the precipitous drop in panel

prices has led PV to lead the pack. Procuring all three technologies throughout the

procurement process helped spur competition and innovation and positioned California

to more quickly pivot to lower cost renewable technologies. In energy efficiency too we

have seen the perils of becoming over reliant on a dominant technology, CFLs, for the

majority of efficiency program savings. When building codes and legislation phased out

incandescent bulbs, energy efficiency program managers had to scramble to find new

strategies and technologies to meet efficiency goals.

The magnitude of the potential overgeneration under the 40% RPS scenario (5,600 MW

at the 99th percentile) suggests no single strategy or resource type will fully address the

challenge. Furthermore, if renewables penetration is to increase to support long-term

GHG reductions, even larger quantities and longer durations of curtailment will need to

be managed. This suggests that we should cast a wide net in our search for flexibility

and energy storage solutions.

7 Conclusions

In this report we have summarized the results of prior work, including the Utility High

RPS Study to illustrate why it is important to perform a system level evaluation of costs

and benefits when analyzing the cost-effectiveness of energy storage as a flexible

resource. The primary conclusions of the paper are:

Overgeneration will be a significant challenge under expected levels of RPS

before 2020.


P a g e | 26 |

Overgeneration will require flexibility solutions with durations of 7-10 hours.

Managing overgeneration with flexible resources reduces expected levels of

renewable curtailment, providing a significant and quantifiable value to utilities

and ratepayers.

Long-duration flexibility solutions provide quantifiable benefits that are not

reflected in existing markets and DER Avoided Costs. Evaluation protocols that

rely predominately on these benefits alone will understate the value of long-

duration storage.

Future and system level costs and benefits, such as reduced renewable

curtailment, are essential to fully quantify value storage and to evaluate the

relative benefits of both short- and long-duration storage.

Evaluating storage projects in an individual basis will, at a system level,

overstate the value of short-duration storage and understate the value of longer

durations. This holds true even if prices from system-wide production

simulation are used.

Our experience in RPS procurement and energy efficiency illustrates the

importance of procuring a diverse portfolio of resources. Ensuring that the cost-

effectiveness and procurement frameworks support diverse strategies is crucial

to achieve ambitious GHG reduction goals provides a long-term benefit to

ratepayers.

Interim Report

P a g e | 27 |

Appendix A: Cost of Renewable Curtailment



In the recent E3 study, “Investigating a Higher Renewables Portfolio Standard in

California,” it was determined that a significant challenge to integrating renewables

under a 40% or 50% RPS in California is the avoidance of renewable curtailment.

Renewable curtailment was identified as an operational solution to avoid

overgeneration events, which are marked by high frequency conditions and negative

prices to incentivize generators to shut down. While renewable curtailment may be a

valuable tool in maintaining reliability and avoiding volatility in the system under a

higher RPS, this renewable curtailment does not come without a cost. If utilities have

procured resources to meet the RPS with the expectation that a certain level of

renewable energy will be delivered from these resources, frequent renewable

curtailment may increase the risk of being out of compliance in a given year. There are

two strategies for minimizing this risk: 1) the utility can over procure renewable

resources; or 2) the utility can procure resources that provide enough flexibility to

ensure that energy from their renewable resources can be delivered. For a utility, the

choice between these two options will depend on the cost of procuring additional

renewables versus the cost of procuring flexible resources and the cost savings

associated with avoided renewable procurement that is afforded by these resources.

7.1 Avoided RPS Costs

The incremental cost of RPS compliance for a utility that is anticipated to be short due

to curtailment will depend on a number of factors. First and most intuitively, it will

depend on technology and installation costs, which are rolled into renewable PPA price

forecasts. In addition, the RPS costs will depend on whether utilities are in under-

procurement over over-procurement positions with respect to the RPS targets.

Consider the following scenarios:


P a g e | 28 |

Utilities are generally long on renewables – this scenario reflects the situation

in California today. When the utilities are long on renewable resource

procurement with respect to the RPS, renewable energy in excess of RPS

targets. In this scenario, there is no immediate cost for renewable curtailment.

A utility is short on renewables – in this scenario the utility must procure

additional renewable resources. If utilities are generally short on renewables,

then curtailed renewable energy must be replaced at or near the cost of new

construction for renewable generation.

Marginal Curtailment Adjustment

When renewable curtailment forces utilities to procure additional resources to ensure

RPS compliance, utilities must also account for curtailment of these incremental

renewable resources in assessing RPS compliance costs. Consider for example, that a

utility requires an additional 5,000GWh/yr. of renewables to meet its RPS requirements.

Based on operational simulations, the utility expects that 5% of the energy produced by

incremental renewable resources might be curtailed. The utility therefore sets its

procurement target at 5,000/(1-0.05) = 5,263MWh. At $70/MWh, the utility must pay

$368,000 per year to deliver the 5,000GWh for RPS compliance. The effective new build

cost is therefore $73.68/MWh. This logic leads to a PPA price multiplier that depends

on the expected curtailment of incremental renewable resources, which will be referred

to as the marginal curtailment rate for the remainder of this document. The PPA price

multiplier is equal to 1/(1-[marginal curtailment rate]).

Renewable-Based Fuel Savings Adjustment

There is an additional fuel savings adjustment that must be made when using

production simulations to quantify the net costs of new flexible resources. Production

cost simulations are typically run with portfolios of renewable resources designed to

meet RPS requirements assuming that no renewable curtailment occurs. In scenarios

Interim Report

P a g e | 29 |



when renewable curtailment occurs we assume that the utilities meet their RPS

requirements either through additional renewable procurement or through

procurement of flexible resources to ensure the delivery of their renewable resources.

This additional renewable generation offsets an equal amount of thermal production to

account for the energy value of the additional renewable resources. The cost impact of

avoided renewable curtailment therefore includes both the cost of new build and the

cost savings of avoided thermal generation.

Returning to the example described above, the utility that determined that it would be

5,000MWh short due to expected curtailment in the operational year, could expect to

see 5,000MWh of fuel savings if it were to procure the 5,263MWh of renewables to

meet its RPS requirements. If these resources offset thermal resources at $30/MWh,

then the net cost of meeting its RPS requirements with incremental resources is $73.68-

$30 = 43.68/MWh. The full build-up of this cost is shown schematically below for two

scenarios. In Scenario 1, utilities have generally over procured renewable resources,

leading to low net costs for any incremental purchases or REC’s needed by a utility to

meet its RPS requirements. With the purchase of additional renewable generation,

there is an associated reduction in fuel costs. In Scenario 2, utilities are generally short,

so the incremental compliance net costs reflect the full costs of new renewable build.


P a g e | 30 |

Figure 13: Value of avoided curtailment under scenarios when the utilities are long vs. short on renewables

Valuing Energy Storage as a Flexible Resource...Valuing Energy Storage as a Flexible Resource –...

Documents

Transcript of Valuing Energy Storage as a Flexible Resource...Valuing Energy Storage as a Flexible Resource –...