HomeMy WebLinkAboutElectric Power Systems Railbelt Wind Regulation Study JohnHiebDavidBurlingame 03-07-2014-RB
Alaska Energy Authority
Regulation Resource Study
Technology Recommendation and Cost Estimates
March 7, 2014
John DL Hieb
David W. Burlingame, P.E.
March 7, 2014 Page ii
Summary of Changes
Revision Revision Date Revision Description
0 August 6, 2012 Initial Rough Draft
1 March 7, 2014 Report Revision Based on Comments Received from AEA
Table of Contents
EXECUTIVE SUMMARY ................................................................................................. 1
1 INTRODUCTION ...................................................................................................... 3
2 REGULATION RESOURCES .................................................................................. 4
3 AVAILABLE ENERGY STORAGE TECHNOLOGIES .............................................. 5
3.1 Lead-Acid .......................................................................................................................... 5
3.2 Nickel-Cadmium ................................................................................................................ 6
3.3 Nickel Metal Hydride ......................................................................................................... 6
3.4 Lithium-Ion ........................................................................................................................ 7
3.5 Sodium-Sulfur ................................................................................................................... 7
3.6 Vanadium-Redox .............................................................................................................. 7
3.7 Zinc-Bromine ..................................................................................................................... 8
3.8 Advanced Flywheels ......................................................................................................... 8
3.9 Applicable Technologies for Railbelt Regulation Application ............................................ 9
4 PRELIMINARY WIND ANALYSIS ............................................................................ 9
4.1 Regulation Resource Power Requirement ........................................................................ 9
4.2 One Hour Regulation Resource Energy Requirement .................................................... 10
4.3 Battery Life Evaluation .................................................................................................... 17
5 TECHNOLOGY RECOMMENDATION .................................................................. 19
5.1 Financial Considerations ................................................................................................. 19
5.2 Economic Analysis – Advanced Lead-Acid vs. Lithium-Ion ............................................. 20
6 SIX HOUR ENERGY NEEDS ................................................................................ 22
6.1 Wind Regulation .............................................................................................................. 22
6.2 Loss of Kenai Tie ............................................................................................................ 28
6.3 Gas Storage Description and Costs ................................................................................ 33
7 CONCLUSIONS AND RECOMMENDATIONS ...................................................... 34
8 REFERENCES ....................................................................................................... 36
March 7, 2014 Page iii
List of Tables
Table 1: Regulation Shortfall and Feathering Analysis Results ................................................... 13
Table 2: Regulation Shortfall and Feathering Analysis Results ................................................... 16
Table 3: Energy Storage Systems Cost Update ............................................................................ 20
Table 4: Battery Life Based on Battery Capacity ......................................................................... 21
Table 5: Battery Initial Installation Cost and 20 Year Project Cost ............................................. 22
Table 6: 52 MW Wind Schedules for September 1 through September 2 .................................... 25
Table 7: Six-Hour Regulation Simulation Results for a 52 MW Wind Farm .............................. 27
Table of Figures
Figure 1: Desired Regulation Characteristic ................................................................................. 11
Figure 2: Desired Regulation Characteristic ................................................................................. 15
Figure 3: Battery Cycle-Life vs. Depth of Discharge ................................................................... 17
Figure 4: Cycle Counting Method ................................................................................................ 18
Figure 5: Six Hour Energy Needs Based on Wind Schedule ........................................................ 23
Figure 6: Wind Power for First Provided Days ............................................................................ 25
Figure 7: Summer Valley Loss of Kenai Tie at Maximum Flow ................................................. 30
March 7, 2014 Page 1
Executive Summary
The intent of this phase of the study is to provide a recommendation on the technology and
develop a budgetary cost estimate of regulation technology for the Railbelt. The selected
regulation resource should enable the electrical system to accept more renewable energy by
alleviating the gas constraints on utility generation that are currently prohibiting its development
and secondarily provide contingency reserves for loss of generation or transmission resources
in the Railbelt.
The study evaluated the impact of the single transmission line from the Kenai and the changing
generation characteristics of the Railbelt. These factors were included in the selection and
sizing of the regulation resources evaluated in the study.
Due to both gas and electrical system constraints faced by the Railbelt utilities, the ability to
regulate an intermittent resource such as wind generation is limited. In order to deal with these
system constraints a regulation resource that could use energy storage to regulate an
intermittent wind resource is required prior to developing a renewable energy portfolio for the
Railbelt.
The changing generation technology of the Railbelt has a dramatic impact on the regulation
capability of the Railbelt. As the Railbelt utilities move towards smaller, more efficient units that
more closely matches their capacity requirements, the chances of having “excess” regulation
capability on the system to respond to unexpected events decreases dramatically. Even if
sufficient gas supplies were available, operating capacity to respond to unexpected changes in
non-dispatchable renewable energy or the loss of the Anchorage – Kenai Intertie will be
minimal. Consequently, flexible regulation resources must be developed to allow additional
renewables to be incorporated into the system, protect against sudden loss of generation or
transmission resources, and to optimize the use of the new generation of high-efficiency gas
generation.
This study evaluated three major technologies and their applicability in the Railbelt. These three
technologies were: Battery Energy Storage Systems (BESS), Flywheel or rotating inertia
technology, and Flexible Gas Storage (FGS). The selected technologies would augment the
regulation capability of the Railbelt hydro resources and be used in conjunction with other
regulation capabilities of the Railbelt.
The criteria used for the regulation evaluation were that no single event should result in the loss
of load in the Railbelt and that the regulation system must work with any single regulation
resource out of service or unavailable. For example, if the Kenai intertie was out service and
hydro regulation was unavailable or if hydro was not scheduled for generation, the remaining
regulation resources in the Southcentral Railbelt must be capable of providing the required
regulation.
The driving force for determining the regulation requirement in the existing system is the loss of
the single Anchorage – Kenai Intertie under maximum import conditions. Following the
retirement of the large gas turbines, this contingency will be the largest resource loss for the
Southcentral area.
The regulation requirements of the Railbelt are divided into short-term regulation requirements
caused by variations in load and variable generation and long-term regulation required by
sustained wind ramp events, loss of transmission interconnections, or the loss of a generation
unit. The long-term regulation requirements exceed the capabilities of flywheel technology,
consequently, this technology was dropped from consideration.
March 7, 2014 Page 2
BESS and FGS technologies are ideally suited for the Railbelt and can be used in
complimentary fashions to provide the optimum system performance. During normal operation,
regulation would be provided by a combination of BESS, FGS and hydro resources. The system
is capable of providing the required regulation following the loss of any one regulation resource.
By utilizing complimentary regulation resources, the costs and sizes of each resource can be
optimized to meet the Railbelt needs.
The BESS was sized to be the primary resource for short-term variable energy deviations from
renewable projects such as wind or solar or from the instantaneous loss of generation. The goal
for sizing the BESS was to provide regulation such that that the reliability of the Railbelt would
be maintained at approximately its current levels following the addition of variable generation to
the grid. The target is to provide enough regulation energy such that on average only twelve
events which exceed the regulation capabilities of the BESS are expected during an average
year.
The requirements for FGS were developed to enable the Southcentral utilities sufficient storage
at gas generation plants to provide fuel for thermal regulation following the loss of the largest
contingency (Anchorage-Kenai Intertie). The thermal regulation capacity must be capable of
allowing the utilities to schedule gas from in-ground storage at the next scheduling interval,
estimated at 6 hours.
Based on these criteria, we recommend the utilities use a BESS and on-site gas storage
systems to provide the required regulation during both the short-term and long-term events.
With the construction of the Beluga – Bernice tie, the BESS should be constructed with a
capacity/energy rating of 25 MW/ 14 MWH and the FGS should be constructed to provide 1.91
MCF (262.5 MWh) to cover the small wind farm and the loss of one of the Kenai ties under
maximum import conditions. Both the proposed BESS system and the FGS systems can be
constructed in blocks either simultaneously or independently. However, construction of the
facilities in blocks will increase the total costs over the duration of the project.
If the HVDC line is not constructed, the size of the BESS and FGS will increase significantly.
The final size will be determined by the largest expected transfer of the single Kenai –
Anchorage Intertie. To maintain reliability equivalent to the HVDC system, the BESS capacity
would need to be increased to approximately 100 MW. It is unlikely that this BESS could be
economically installed; therefore, lower import or lower reliability measures would need to be
adopted. For the larger regulation requirement we recommend the FGS be located at two
different locations with thermal generation. Two different locations are recommended to
maximize the availability of on-line and off-line regulation resources.
The cost of the recommended alternative is as follows:
Description Wind Farm Size Capacity Energy Costs
BESS – HVDC 17 MW 25 MW 14 MWh $26.7 M
FGS – HVDC 17 MW NA 1.91 MCF (262.5 MWh) $18.2 M
March 7, 2014 Page 3
1 Introduction
The purpose of this report is to provide the results of a regulation resource technology
evaluation and a preliminary cost estimate for the recommended alternative. EPS will
recommend a regulation technology to provide the best-fit for the regulation application. EPS
will then provide the life-cycle costs of different storage technologies based on their ability to
meet the regulation needs.
The full breadth of the study includes the evaluation of BESS, FGS and Flywheel technologies
in the South-central Railbelt area. Each of these technologies was evaluated independently and
in combination with each other in order to provide the optimum solution for the Railbelt utilities.
The primary goal of the regulation resource is to provide the Railbelt utilities the ability provide
regulation capability for renewable energy resources in the region that cannot be regulated by
the current generation’s fuel supply system.
In addition to providing regulation for renewable energy projects, the proposed system’s
secondary goal is to provide response to the loss of the Anchorage-Kenai intertie which will
soon be the largest operating contingency in the Southcentral area. It is assumed that the
Railbelt transmission planning study will recommend a second line connecting the Kenai
Peninsula with the Southcentral transmission system (the Beluga – Bernice Lake HVDC
Intertie). The regulation requirements were studied with and without this second transmission
line to determine the impact on the regulation size and technology. However, due to the impact
on transfer capability of the second transmission line between the Kenai and Anchorage, this
study reduced the maximum import level into Anchorage from 125 MW to 75 MW.
The regulation resource must be capable of relieving the gas constraints placed on the
Southcentral utilities by both gas transportation and gas producing entities in providing
regulation for both non-dispatchable resources and system contingencies.
The system should be flexible in its response and implementation. No single failure of a
regulation resource should result in the lack of regulation capability in the system. The
regulation resource must also be designed to not place scheduling requirements on the Railbelt
generation, it must be available if no hydro is scheduled to meet system load. The system
should also be flexible in terms of its implementation and construction, allowing for modular
implementation if budgetary constraints require it.
The regulation technologies to be evaluated are as follows:
Battery Energy Storage System (BESS) – A BESS consists of large battery systems
designed to provide both energy input to the system during generation shortfall and
absorb energy during generation excess conditions
Flexible Gas Storage (FGS) – FGS consists of compressed gas storage facilities located
at or near thermal generation resources. FGS can provide stored gas to generation
during energy shortfalls or absorb scheduled gas during excess energy production.
Flywheel Technology – Flywheel technology consists of using the inertia of a rotating
mass to provide or absorb energy stored in the rotating mass to the power system.
Earlier flywheels were directly connected to the power system and their discharge or
absorption was determined by frequency fluctuations of the power system. Modern
inertia systems are connected by an inverter system, allowing the characteristics of the
flywheel to be manipulated by the inverter controller.
March 7, 2014 Page 4
2 Regulation Resources
As this study is highly sensitive to initial assumptions, it is important that the various project
assumptions be understood when evaluating the results of the study. The following sections
highlight the major assumptions used in the study, along with the expected impact of the
assumptions. In order to determine the energy and power requirements provided by the
proposed storage devices, the current and expected future regulation capabilities must be
defined. EPS assumed the following regulation capabilities for the various Railbelt resources:
2015 Cases:
o Natural Gas Turbines
Due to gas scheduling constraints, the Railbelt natural gas turbines will
provide no regulation power or energy other than the requirements to
meet the scheduled load ramp (absent the installation of flexible gas
storage).
The gas turbines are set on a six-hour schedule that should not be
revised except for emergency conditions. If the severe wind ramp events
occur as infrequently as a couple of times per year, then the capability of
changing the gas schedules at the hour will be analyzed as it pertains to
the storage capabilities. However, excursions greater than one time per
month (one average) must be compensated by other means.
o Hydro Turbines
The hydro turbines at Cooper and Eklutna will provide no regulation
power or energy during the hour.
The hydro resource schedules will be fully dispatchable at the hour from
the maximum to minimum capabilities of the units. This will result in
“ponding” water during those times when wind is available and hydro is
scheduled, but is being displaced by wind energy.
The hydro resources may not be scheduled 24 hours/day for energy
delivery to the utilities.
o Wind Turbines
The study will assume no capability to forecast wind power output or
ramps other than utilizing the same day patterns to predict the next few
hours. This will provide a solution that will be capable of responding to the
most likely, unconstrained wind changes.
Self-regulation by feathering the wind turbine blades will be evaluated as
part of the storage solutions.
Both large and small wind farm sizes will be evaluated. EPS has
projected wind power outputs for both wind farm sizes.
It is assumed that the northern Railbelt system will provide the regulation
for the wind generation at Eva Creek. Additionally, since there is only one
tie to the northern system, the two areas must be able to be operated so
as to not impact the other, or in the extreme case, operated islanded from
each other.
2025 Cases:
March 7, 2014 Page 5
o Watana Hydro Addition
The proposed Watana hydro plant will not provide any sub-hour
regulation power or energy during the hour due to downstream flow
restrictions.
Time Frames:
o Since the utilities must account for all wind variations in order to maintain
frequency stability, the required power and energy will be analyzed for several
different time frames including 20, 30, and 60 minutes for electrical energy
requirements and up to six hours for the flexible gas storage options.
o Since the hydro resource schedules are able to change at the hour, the 60-
minute time frame will take precedence over the other time frames, however, it is
recognized that the majority of hydro resources are only available through a
single contingency transmission line unless the Beluga-Bernice Lake HVDC line
is constructed and that hydro resources are not typically scheduled 24-hours/day.
In addition to the regulation requirements, the capacity and energy requirements
for the loss of the largest intertie and largest unit will be evaluated.
o The installation of a second Anchorage – Kenai transmission line will reduce the
maximum capacity lost in the Anchorage area to 30 MW for the loss of the
existing Anchorage – Kenai Intertie.
Criteria
o Due to the critical nature of the regulation requirements, the regulation system
must be capable of operation during the loss of any single regulation source, i.e.
loss of stored energy, loss of hydro, loss of gas storage. The system will not be
required to operate during an N-2 condition.
3 Available Energy Storage Technologies
This section gives a basic summary of the battery and flywheel technologies that are currently
available. The applicability for each technology for use as a regulation resource is determined.
For each technology that is deemed applicable, an economic analysis will be performed to
determine the lowest-cost option for the Railbelt regulation resource.
3.1 Lead-Acid
The lead-acid battery is the most mature battery technology with well over 100 years of service.
Currently, there are three types of lead-acid batteries. The first of which is the flooded cell lead-
acid battery. This technology is the most common form of the lead-acid battery. This technology
uses lead/ lead alloy plates that will react with a sulfuric acid electrolyte to produce the
movement of charge.
The flooded cell lead-acid battery has the advantage of being the lowest cost battery option with
excellent shelf life, and good efficiency. The main problems with the flooded cell lead-acid
battery are the numerous environmental concerns and the low cycle-life (only a couple hundred
cycles for deep discharges). Since the regulation application will require thousands of cycles per
year, the flooded cell lead-acid battery should not be considered for a regulation application.
The second type of lead-acid battery is the valve regulated lead-acid battery (VRLA). The VRLA
battery was designed to reduce some of the maintenance concerns with the flooded cell lead-
March 7, 2014 Page 6
acid battery. Unfortunately, the changes required to reduce the maintenance needs further
reduced the cycle-life of the battery, as such, should not be considered for a regulation
application.
The third type of lead-acid battery is the advanced lead-acid. Due to continuing research into
the lead-acid technology, some breakthroughs in the electrode materials have resulted in
drastically improved battery cycle-life. With the cycle-life improvement, the advanced lead-acid
batteries could be a potential solution for providing regulation services for the intermittent wind
resource and should be further investigated, and included in the economic analysis. The two
main competing companies using advanced lead-acid batteries are Axion Power and Xtreme
Power. The Xtreme Power dynamic power resource (DPR) has been used in conjunction with
several wind farm applications in Hawaii, and has recently been proposed as the battery
technology to provide 36 MW, 24 MWh in conjunction with a large wind farm in Texas. This
Texas installation represents one of the largest battery installations in the world, and is on the
same scale as would be required for a Railbelt regulation resource. At a much smaller scale,
Axion Power has recently connected to the PJM regulation market. This connection is significant
in that it is also a regulation application that requires many charge/discharge cycles. The
advanced lead-acid battery is recommended for further consideration as an option for the
Railbelt regulation resource.
3.2 Nickel-Cadmium
The nickel-cadmium battery technology is the most common nickel-electrode battery in the utility
industry. The nickel-cadmium battery is a favored alternative to the traditional lead-acid batteries
due to the advantages of 1) greater depth of discharge, 2) greater tolerance of extreme
temperature variation, 3) greater tolerance to over/under charging, and 4) lower maintenance
requirements. This battery technology does have some setbacks that include 1) lower efficiency
than lead-acid and, 2) environmental concerns due to the cadmium.
Although the nickel-cadmium battery is superior to the traditional lead-acid battery in
performance, it does have a higher rate of self-discharge and requires continuous charge
maintenance. The Railbelt system has experience with a nickel-cadmium battery system since
the GVEA BESS uses the nickel-cadmium technology. The GVEA BESS was designed for VAr
support, spinning reserve, and power system stabilization, but it was not designed for
regulation. Due to the relatively limited cycle-life of nickel-cadmium batteries and the maturation
of the nickel metal hydride battery, this technology should not be considered for a regulation
application.
3.3 Nickel Metal Hydride
The Nickel Metal Hydride battery (NiMH) has basically displaced the nickel-cadmium battery
since it has better energy density, better cycle-life, and no heavy metals (fewer environmental
concerns). This battery technology was used in the early Toyota Prius Hybrid vehicles (the
newest plug-in model uses lithium-ion). Due to the use in the plug-in hybrid vehicle market,
these batteries are among the most field-tested solutions. The NiMH battery technology does
not have the same discharge capabilities that a Ni-Cd battery has. Hence, NiMH batteries are
used more often for low-current applications such as portable computers and cell phones, while
the Ni-Cd batteries are used for high current applications such as portable power tools [2]. The
Ni-MH batteries have slightly worse charge retention than their Ni-Cd counterparts and would
require continuous charge maintenance.
Currently, there are no large format NiMH batteries. Large format NiMH cells would be better
suited to a large-scale stationary battery system for utility use. The NiMH battery has largely
March 7, 2014 Page 7
been replaced by the lithium-ion technologies in consumer electronics, and does not have the
same level of investment that it once had. Due to these factors, the Nickel Metal Hydride battery
would not be a good selection for the Railbelt regulation application.
3.4 Lithium-Ion
The lithium-ion battery technology has rapidly taken over the consumer electronics industry due
to its energy density advantage over the nickel metal hydride battery technology. This battery
technology comes in several flavors based on the specific chemistry of the cathode. The
different types include lithium-ion cobalt, lithium-ion manganese, lithium-ion phosphate, and
lithium-ion titanate. The different chemistries offer differing specific power (charge/discharge
rate), safety characteristics, and cycle-life [1].
The Chevrolet Volt and the newest Toyota Prius vehicles use lithium-ion battery packs. The
selection of the lithium-ion technology for the transportation sector suggests that the regulation
market might be an acceptable utility application for this technology since the frequent battery
usage associated with a hybrid vehicle is similar usage that would be seen in utility regulation
applications. Additionally, the new manufacturing capacity required by the electric vehicle
industry will have a price reduction effect due to economies of scale.
The lithium-ion batteries have several desirable characteristics such as long-cycle lives, good
energy density, and high power density. The lithium-ion batteries, however, are more expensive
than many of the competing battery technologies, but due to their superior performance,
particularly the excellent cycle-life, this battery technology should be considered for the
regulation resource project.
3.5 Sodium-Sulfur
The sodium-sulfur battery is currently the most widely used utility-scale battery technology. It
has been heavily used in Japan by TEPCO (Tokyo Electric Power Company) and is produced
by NGK. There are several installations in the United States.
The sodium-sulfur battery must maintain high operating temperatures (> 250°C). As such, the
batteries must be heavily insulated to maintain the temperature, and when the batteries are not
providing power, must be heated via resistor banks. These batteries are primarily used for
uninterruptible power supplies in Japan, but are beginning to see applications such as load
shifting and wind smoothing here in the United States.
There was a sodium-sulfur battery fire on September 21, 2011 which has brought some scrutiny
toward the battery safety. The cause has not been identified, and the production of these
batteries has been put on hold until the safety concerns are resolved.
The sodium-sulfur batteries advantages are that the technology has a high round-trip efficiency.
It has good energy density and cycle-life for large discharge depths (>5,000 at 90%), but poor
cycle-life for smaller discharge depths (45,000 at 10%). The sodium-sulfur technology is not
well-suited to frequent charge/discharge as would be expected with a regulation application.
The sodium-sulfur battery technology should not be considered for the Railbelt regulation
application.
3.6 Vanadium-Redox
The Vanadium-redox battery is a flow type battery. Flow batteries store their energy in liquid
electrolytes, and pump the liquid to a fuel cell where the electro-chemical reactions occur. The
vanadium-redox battery basically stores the energy in different ionic forms of vanadium. One of
March 7, 2014 Page 8
the advantages of this flow battery system is that the energy capacity (MWh) and the power
capability (MW) can be sized separately based on the application. For example, if more energy
is needed, simply adding electrolyte storage tanks will increase the battery energy. This is a
desirable attribute for matching a vanadium-redox battery to an application that may require
additional capacity at a later date.
The vanadium-redox battery technology is being developed by Prudent Energy. This battery
technology is currently being tested at the University of Alaska Fairbanks. The vanadium-redox
battery has decent AC-to-AC efficiency of 70% to 75%, good cycle-life, and good reliability. The
vanadium-redox battery has some disadvantages such as have high cost, low energy density,
and a limited number of installations in the field. The vanadium-redox battery technology is
better suited to applications requiring several hours of stored energy such as peak shaving or
energy arbitrage. Due to these disadvantages, the vanadium-redox battery is not recommended
for the Railbelt regulation application.
3.7 Zinc-Bromine
The zinc-bromine battery is also a flow type battery. This technology has a significant promise,
but has very limited field applications. During charging, metallic zinc is plated from the
electrolyte onto the negative electrode and bromide is converted to bromine at the positive
electrode. During discharge, the metallic zinc dissolves into the electrolyte.
The zinc-bromine technology has several advantages over the vanadium-redox battery. Zinc-
bromine batteries have better energy density, lower cost, and fewer environmental concerns
since zinc-bromine technology uses less toxic materials. However, the zinc-bromine batteries do
not have independent sizing like the vanadium-redox battery. Also, the power capacity of the
zinc-bromine battery is low which limits the charge/discharge rate.
The zinc-bromine battery technology is being developed and manufactured by ZBB Energy
Corp. and Premium Power Corp. ZBB Energy has more utility scale projects online, but still has
limited experience in the utility sector. Due to the poor power capability of this technology, a 50
MW system would require at least 150 MWh of storage. This would add to the cost of such a
system compared to other technologies that could have a 50 MW / 50MWh configuration.
Another disadvantage of this battery technology is that the battery maintenance requires
“stripping”. Stripping is performed by discharging the battery cell down to zero volts. This will
remove all zinc from the negative electrode. This process is performed to increase efficiency,
and ensure consistent operation of all battery cells. Due to the poor power capability, the need
for ‘stripping”, and the minimal field applications the zinc-bromine technology should not be
considered for the regulation application.
3.8 Advanced Flywheels
Flywheels convert the electrical energy from the grid and convert it into rotating kinetic energy.
The advanced flywheels spin at high speeds. In order to reduce the frictional losses, these
flywheels operate with magnetic bearings in a vacuum. In order to maintain structural integrity at
high rotational speeds, these flywheels are made of high-tech composite materials.
These advanced flywheels can charge and discharge without performance degradation which
makes them ideally suited to regulation applications. Unfortunately, the advanced flywheel
systems are quite expensive. The flywheel technology is primarily used in uninterruptible power
supply applications. There are several flywheel manufacturers, but only Beacon Power is
marketing towards utility applications. All the other companies are marketing toward
uninterruptible power supplies. Beacon Power has a 20MW, 5 MWh flywheel system used for
March 7, 2014 Page 9
the New York regulation market. This project cost a reported $69 million. The Railbelt regulation
application will need at least five times the storage, and that would make the flywheel option too
expensive for the hour-long energy needs. Advanced flywheels are not recommended for the
Railbelt regulation application.
3.9 Applicable Technologies for Railbelt Regulation Application
The need for near-constant charging and discharging characteristics of a regulation application
removes several technologies from consideration based on limited cycle-lives (nickel-cadmium,
sodium-sulfur, traditional lead-acid). Limited field experience and high costs also removes some
technologies from consideration (vanadium-redox, advanced flywheels). The two technologies
that should be further investigated are advanced lead-acid batteries, and lithium-ion battery
technology.
4 Preliminary Wind Analysis
4.1 Regulation Resource Power Requirement
4.1.1 52 MW Wind Farm
With the assumption that the regulation resource must provide all the regulation within the hour,
the wind data was analyzed to determine the power capacity needed to fully regulate a large
wind farm with a capacity of 52 MW. The wind data was provided by Clarity Analytical in a one-
minute time series showing wind farm power output. The maximum power required by the
regulation resource was determined by the maximum inter-hour power change.
For each minute in the two years of analyzed wind data, the inter-hour maximum and minimum
wind power output were found and compared. Using this method, the maximum inter-hour
power change for the wind farm was approximately a net of 48.25 MW for the 52 MW wind
Farm. Therefore, the estimated wind farm output can almost go from maximum power output to
zero within one hour. In order for the regulation resource to prevent the rest of the Railbelt from
seeing power fluctuations from the wind farm, the regulation resource, including the option of
curtailment must compensate for the full net power of 48 MW. EPS recommends a regulation
resource with at least 50 MW power capability in order to fully regulate the wind farm.
4.1.2 17 MW Wind Farm
With the assumption that the regulation resource must provide all the regulation within the hour,
the wind data was analyzed to determine the power capacity needed to fully regulate a smaller
wind farm with a capacity of 17 MW. The wind data was provided by Clarity Analytical in a one-
minute time series showing wind farm power output. The maximum power required by the
regulation resource was determined by the maximum inter-hour power change.
For each minute in the two years of analyzed wind data, the inter-hour maximum and minimum
wind power outputs were found and compared. Using this method, the maximum inter-hour
power change for the wind farm was approximately a net of 17 MW for the 17 MW wind Farm.
Therefore, the estimated wind farm output can almost go from maximum power output to zero
within one hour. In order for the regulation resource to prevent the rest of the Railbelt from
seeing power fluctuations from the wind farm, the regulation resource, including the option of
curtailment must compensate for the full net power of 17 MW. EPS recommends a regulation
resource with at least 17 MW power capability in order to fully regulate the wind farm.
March 7, 2014 Page 10
4.2 One Hour Regulation Resource Energy Requirement
4.2.1 52 MW Wind Farm
One way of determining the regulation energy requirement was based on the worst case one-
hour need. This need is based on the upward regulation requirement to compensate for the
wind output. The worst case scenario is the one-hour interval that represents the maximum
amount of regulation energy that the utilities must provide for in the regulation scenario.
Conditions where the wind turbines are operating near cut-out or are experiencing severe
fluctuations are assumed to be curtailed by the operating utility.
The wind was analyzed for each year of data available. For each minute of wind data, the
change in wind output was integrated over a one-hour time period to provide the necessary
energy for that hour. The worst-case hour would require 36 MWh from the regulation resource.
Sizing the regulation resource to provide for the worst case scenario would result in an
expensive, over-sized regulation resource. A simple control method was developed in an
attempt to minimize the battery energy sizing using two years’ worth of wind data. This method
and its results are described in the next few paragraphs.
First, the worst-case hourly regulation requirement based on the initial wind power was
calculated from the two years’ worth of wind data. This worst case regulation was calculated
using a very simple method. The next hour wind schedule was set to the value of the wind plant
at the beginning of the hour. Any downward movement in the wind would be balanced out by
the regulation resource such that the net power out of the wind farm plus the regulation
resource would stay flat for the entire hour. The worst case energy requirement was tabulated
for each 2 MW range of wind starting power. The resulting energy need was calculated and put
on a graph to visualize the results. The following two examples should provide some assistance
in understanding the blue curve shown below in Figure 1. The blue curve represents the largest
amount of regulation energy required to fully regulate the wind farm output based on the initial
power output at the beginning of the hour.
For a starting wind power output of 0 to 2 MW, the worst case energy needs for a wind
down ramp is 1.86 MWh. This would occur if the wind started at 2 MW and quickly
ramped down to 0 MW and stayed there for the rest of the hour.
For a starting wind power output between 32 and 34 MW, the worst case energy needs
for a wind down ramp is 27.1 MWh. Again, this would occur if the wind quickly ramps
down from the starting value to near zero and stays there for the rest of the hour.
A linear characteristic was selected to represent the necessary battery regulation based on the
starting wind power output. The slope of this line is approximately 0.83 MWh required per MW
initial power output. The characteristic is shown in red in Figure 1. It is assumed that the battery
power rating is such that the battery can provide a power output equal to the large wind farm
plant output (50 MW). The red characteristic curve shows a desired regulation resource charge
level. A control strategy should attempt to keep the regulation resource energy at the desired
energy in real time. By controlling the regulation resource to the resource characteristic, two
benefits are realized:
1. For lower wind power outputs (0 – 30 MW), the regulation resource would provide
enough energy for all wind down ramps. In order to minimize the regulation resource
energy requirement, the loss of a large amount of wind power will require grid regulation
resources to survive. This can occur when the blue curve is above the red curve for
large starting power outputs (30 – 50 MW). The regulation resource should be sized to
keep the occurrences of this shortfall to fewer than once per month.
March 7, 2014 Page 11
2. By maintaining a minimum charge level that will survive all wind down ramps, the
regulation resource will have a maximum amount of room to absorb energy for wind up
ramps. This will minimize the need to feather the wind turbine blades and maximize the
amount of energy captured from the wind farm.
Figure 1: Desired Regulation Characteristic
Stated again, the characteristic shown in Figure 1 would represent the desired regulation from
the regulation resource up to its energy limit (25MWh in this example). There will be instances, if
the wind is near its maximum power output, when there can be a regulation resource shortfall.
When the blue curve is above the red curve, it is possible to run into these shortfall conditions.
The battery energy management system should try to keep the battery state of charge near the
red regulation characteristic. It is not prudent to always keep the battery charged near its
maximum output since this would mean that the battery could not absorb the positive changes
in wind power. So in order to maximize the battery’s usefulness, the battery should be kept near
the desired regulation characteristic. This way the battery has the maximum ability to absorb the
wind energy when its power increases while always maintaining enough energy to survive
severe wind down ramps.
Due to the limitation of battery sizing, there will be times when the battery will have insufficient
energy to fully regulate all wind down ramps. Additionally, there will be times when the battery
does not have sufficient room to absorb the wind up ramps. The wind plant can be controlled to
limit the up ramps to prevent battery overcharging, but results in unused wind energy. For the
extreme wind down ramps, the grid will need to supply for the regulation shortfall. To determine
the number of hours, and amount of shortfall and feathered energy, a simulation was run for the
two years’ worth of wind data. This simulation can determine the effect that battery sizing and
0
5
10
15
20
25
30
35
0 102030405060Regulation Resource Required Energy (MWh)Initial Wind Power (MW)
Desired Regulation Resource Characteristic (25 MWh Capacity)
DesReg
Up Reg
March 7, 2014 Page 12
control strategy have on the amount and frequency of regulation shortfall, and wind turbine
feathering.
The following battery control strategy was implemented to keep the battery near the desired
regulation characteristic using the following equation.
ܹ݅݊݀௦ௗ௨ ൌ ܣݒ݁ݎܽ݃݁௦௦ ൫ܧ݊݁ݎ݃ݕ௧௧௬ െ ܧ݊݁ݎ݃ݕ௦ௗ ൯ ∗ ܥݎݎ݁ܿݐ݅݊ ܨܽܿݐݎ
Equation 1: Basic Wind Scheduling Method
The Average term is equal to the average value of the wind power output for x number of
samples before the hour schedule begins. The battery energy is the available energy in the
battery at the start of the hour. The desired energy is the energy value taken from the
characteristic from Figure 1 using the average power as a look-up value. The correction factor is
a value used to determine how quickly the controls will adjust the battery to the desired energy
level.
For example, let’s assume that the battery has an energy level of 20 MWh, and the wind has
been steady at 20 MW. The desired energy is approximately 17 MWh. In order to maximize the
ability of the battery to absorb energy if the wind increases, the battery charge should be
reduced to the ideal value of 17 MWh. Therefore, the wind schedule will be adjusted for the next
hour so that the battery will discharge its excess energy. Going through the calculation Wind
schedule = 20 MW + (20 MWh – 17 MWh) * (1/5hours) = 20.6 MW. The wind schedule for the
next hour would be 20.6 MW. Again, if the wind holds steady at 20 MW, the battery will
discharge 0.6 MW for the entire hour to maintain the wind schedule of 20.6 MW. By providing
this energy, the battery will be closer to the desired energy value for the next hour. A larger
correction factor will move the battery charge level to the desired level more quickly. This control
strategy was implemented for the two years’ worth of wind data using different battery energy
ratings and correction factors. Table 1 shows the results of the analysis.
March 7, 2014 Page 13
Table 1: Regulation Shortfall and Feathering Analysis Results
Average Shortfall
Case Battery Size Correction Factor Samples Total Wind Feathered Shortfall % Feathered Hours
1 35 0.2 5 152666 233 0.4 0.2% 5
2 35 0.5 5 152666 625 0.0 0.4% 0
3 30 0.2 5 152666 870 5.5 0.6% 6
4 30 0.5 5 152666 2008 5.4 1.3% 2
5 25 0.2 5 152666 1830 21.1 1.2% 11
6 25 0.5 5 152666 3892 18.5 2.5% 5
7 20 0.2 5 152666 3100 82.8 2.0% 48
8 20 0.5 5 152666 6280 52.0 4.1% 16
1a 35 0.2 5 162228 369 1.2 0.2% 3
2a 35 0.5 5 162228 773 0.0 0.5% 0
3a 30 0.2 5 162228 1122 1.2 0.7% 3
4a 30 0.5 5 162228 2319 0.0 1.4% 0
5a 25 0.2 5 162228 2203 21.7 1.4% 19
6a 25 0.5 5 162228 4424 16.2 2.7% 9
7a 20 0.2 5 162228 3613 120.5 2.2% 70
8a 20 0.5 5 162228 7018 67.5 4.3% 25
Energy (MWh)
Cases 1-8 show the results for the first year of wind data, whereas 1a – 8a represent the
second year. Battery sizes were selected from 35 MWh to 20 MWh in increments of 5 MWh.
Correction factors of 0.2 and 0.5 were used for each battery size using the 5 minute wind
average to determine the wind scheduling. Case 1 resulted in 233 MWh of lost energy due to
the need to feather the blades when the battery could not absorb the entire wind increase which
corresponds to 0.2% of the annual wind energy. Case 1 resulted in shortfall of only 0.4 MWh
that occurred over 5 separate hours throughout the year. The grid would need to supply this
additional energy.
The general observations from the results shown in Table 1 are that as the battery energy level
decreases, the amount and frequency of feathering and shortfalls increases. Also, the smaller
correction factor results in less energy lost due to feathering, but more regulation shortfall. The
recommended regulation resource should only rely on the grid for regulation for emergency
conditions. For this reason, a 20 MWh battery that would rely on the grid to supply shortfall
energy more than once per month should not be considered. The 25 MWh battery should be the
smallest battery considered for further analysis for a 52 MW wind project since it would have
between 5 and 19 shortfall hours per year. Economic analysis will determine the appropriate
amount of battery storage as it compares to the value of the unused wind energy, and frequency
of battery pack replacement.
4.2.2 17 MW Wind Farm
Similar to the 52 MW scenario, the wind output was analyzed for each year of data available.
For each minute of wind data, the change in wind output was integrated over a one-hour time
period to provide the necessary energy for that hour. The worst-case hour would require 13
MWh from the regulation resource. Sizing the regulation resource to provide for the worst case
scenario would result in an expensive, over-sized regulation resource. The same control method
used for the 52 MW wind farm was used for the 17 MW wind farm. The worst case energy
March 7, 2014 Page 14
requirement was tabulated for each 1 MW range of wind starting power. The resulting energy
need was calculated and put on a graph to visualize the results. The following two examples
should provide some assistance in understanding the blue curve shown below in Figure 2. The
blue curve represents the largest amount of regulation energy required to fully regulate the wind
farm output based on the initial power output at the beginning of the hour.
For a starting wind power output of 0 to 1 MW, the worst case energy needs for a wind
down ramp is 0.95 MWh. This would occur if the wind started at 1 MW and quickly
ramped down to 0 MW and stayed there for the rest of the hour.
For a starting wind power output between 14 and 15 MW, the worst case energy needs
for a wind down ramp is 12.1 MWh. Again, this would occur if the wind quickly ramps
down from the starting value to near zero and stays there for the rest of the hour.
A linear characteristic was selected to represent the necessary battery regulation based on the
starting wind power output. The slope of this line is approximately 0.83 MWh required per MW
initial power output. The characteristic is shown in red in Figure 2. It is assumed that the battery
power rating is such that the battery can provide a power output equal to the wind farm plant
output (17 MW). The red characteristic curve shows a desired regulation resource charge level.
A control strategy should attempt to keep the regulation resource energy at the desired energy
in real time. By controlling the regulation resource to the resource characteristic, two benefits
are realized:
3. For lower wind power outputs (0 – 12 MW), the regulation resource would provide
enough energy for all wind down ramps. In order to minimize the regulation resource
energy requirement, the loss of a large amount of wind power will require grid regulation
resources to survive. This can occur when the blue curve is above the red curve for
large starting power outputs (12 – 17 MW). The regulation resource should be sized to
keep the occurrences of this shortfall to fewer than once per month.
4. By maintaining a minimum charge level that will survive all wind down ramps, the
regulation resource will have a maximum amount of room to absorb energy for wind up
ramps. This will minimize the need to feather the wind turbine blades and maximize the
amount of energy captured from the wind farm.
March 7, 2014 Page 15
Figure 2: Desired Regulation Characteristic
Stated again, the characteristic shown in Figure 2 would represent the desired regulation from
the regulation resource up to its energy limit (10 MWh in this example). There will be instances if
the wind is near its maximum power output, when there can be a regulation resource shortfall.
When the blue curve is above the red curve, it is possible to run into these shortfall conditions.
The battery energy management system should try to keep the battery state of charge near the
red regulation characteristic. It is not prudent to always keep the battery charged near its
maximum output since this would mean that the battery could not absorb the positive changes
in wind power. So in order to maximize the battery’s usefulness, the battery should be kept near
the desired regulation characteristic. This way the battery has the maximum ability to absorb the
wind energy when its power increases while always maintaining enough energy to survive
severe wind down ramps.
Due to the limitation of battery sizing, there will be times when the battery will have insufficient
energy to fully regulate all wind down ramps. Additionally, there will be times when the battery
does not have sufficient room to absorb the wind up ramps. The wind plant can be controlled to
limit the up ramps to prevent battery overcharging, but results in unused wind energy. For the
extreme wind down ramps, the grid will need to supply for the regulation shortfall. To determine
the number of hours, and amount of shortfall and feathered energy, a simulation was run for the
two years’ worth of wind data. This simulation can determine the effect that battery sizing and
control strategy have on the amount and frequency of regulation shortfall, and wind turbine
feathering.
The following battery control strategy was implemented to keep the battery near the desired
regulation characteristic using the following equation.
0
2
4
6
8
10
12
14
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19Regulation Required (MWh)Initial Power (MW)
Desired Regulation Resource Characteristic with
10 MWh Capacity (17 MW Wind)
Worst Case Reg
Desired Characteristic
March 7, 2014 Page 16
ܹ݅݊݀௦ௗ௨ ൌ ܣݒ݁ݎܽ݃݁௦௦ ൫ܧ݊݁ݎ݃ݕ௧௧௬ െ ܧ݊݁ݎ݃ݕ௦ௗ ൯ ∗ ܥݎݎ݁ܿݐ݅݊ ܨܽܿݐݎ
Equation 2: Basic Wind Scheduling Method
The Average term is equal to the average value of the wind power output for x number of
samples before the hour schedule begins. The battery energy is the available energy in the
battery at the start of the hour. The desired energy is the energy value taken from the
characteristic from Figure 2 using the average power as a look-up value. The correction factor is
a value used to determine how quickly the controls will adjust the battery to the desired energy
level.
For example, let’s assume that the battery has an energy level of 8 MWh, and the wind has
been steady at 7 MW. The desired energy is approximately 7 MWh. In order to maximize the
ability of the battery to absorb energy if the wind increases, the battery charge should be
reduced to the ideal value of 7 MWh. Therefore, the wind schedule will be adjusted for the next
hour so that the battery will discharge its excess energy. Going through the calculation Wind
schedule = 7 MW + (8 MWh – 7 MWh) * (1/5hours) = 7.2 MW. The wind schedule for the next
hour would be 7.2 MW. Again, if the wind holds steady at 7 MW, the battery will discharge 0.2
MW for the entire hour to maintain the wind schedule of 7.2 MW. By providing this energy, the
battery will be closer to the desired energy value for the next hour. A larger correction factor will
move the battery charge level to the desired level more quickly. This control strategy was
implemented for the two years’ worth of wind data using different battery energy ratings and
correction factors. Figure 2 shows the results of the analysis.
Table 2: Regulation Shortfall and Feathering Analysis Results
Average Shortfall
Case Battery Size Correction Factor Samples Total Wind Feathered Shortfall % Feathered Hours
1 8 0.2 5 56320 1035 13.75 1.8% 20
2 8 0.5 5 56320 1933 7.2 3.4% 9
3 10 0.2 5 56320 545 1.2 1.0% 5
4 10 0.5 5 56320 1053 0.5 1.9% 2
5 12 0.2 5 56320 176 0.2 0.3% 2
6 12 0.5 5 56320 402 0 0.7% 0
7 14 0.2 5 56320 15.1 0 0.0% 0
8 14 0.5 5 56320 402 0 0.7% 0
1a 8 0.2 5 59083 1101 13 1.9% 31
2a 8 0.5 5 59083 2097 5 3.5% 12
3a 10 0.2 5 59083 589 0.3 1.0% 4
4a 10 0.5 5 59083 1154 0 2.0% 0
5a 12 0.2 5 59083 198 0 0.3% 0
6a 12 0.5 5 59083 444 0 0.8% 0
7a 14 0.2 5 59083 3.7 0 0.0% 0
8a 14 0.5 5 59083 14.6 0 0.0% 0
Energy (MWh)
Cases 1-8 show the results for the first year of wind data, whereas 1a – 8a represent the
second year. Battery sizes were selected from 8 MWh to 14 MWh in increments of 2 MWh.
Correction factors of 0.2 and 0.5 were used for each battery size using the 5 minute wind
average to determine the wind scheduling. Case 3 resulted in 545 MWh of lost energy due to
the need to feather the blades when the battery could not absorb the entire wind increase which
March 7, 2014 Page 17
corresponds to 1.0% of the annual wind energy. Case 1 resulted in shortfall of only 1.2 MWh
that occurred over 5 separate hours throughout the year. The grid would need to supply this
additional energy.
The general observations from the results shown in Figure 2 are that as the battery energy level
decreases, the amount and frequency of feathering and shortfalls increases. Also, the smaller
correction factor results in less energy lost due to feathering, but more regulation shortfall. The
recommended regulation resource should only rely on the grid for regulation for emergency
conditions. For this reason, a 8 MWh battery that would rely on the grid to supply shortfall
energy more than once per month should not be considered. The 10 MWh battery should be the
smallest battery considered for further analysis for a 17 MW wind project since it would have
between 0 and 5 shortfall hours per year. Economic analysis will determine the appropriate
amount of battery storage as it compares to the value of the unused wind energy, and frequency
of battery pack replacement.
4.3 Battery Life Evaluation
It is well understood that as batteries go through charge and discharge cycles, their effective life
is reduced. Additionally, large charge/discharge cycles degrade the battery life more quickly
than the small cycles. Many battery manufacturers provide curves that show the expected
number of charge/discharge cycles based on the depth of discharge. The newer battery
technologies can have a million or more cycles at low discharge depths, but approximately
3,000 cycles at 80% depth of discharge. A curve showing a SAFT Li-ion battery characteristic is
shown in Figure 3.
Figure 3: Battery Cycle‐Life vs. Depth of Discharge
In order to determine the length of battery life, the wind data was analyzed to give a count of the
different depths of discharge. Figure 4 is shown to explain the method behind the cycle
March 7, 2014 Page 18
counting. The blue trace shows a fictional wind power output over the course of 5 hours. The
red trace represents the wind schedule. The area between the two curves would be where the
battery would either charge or discharge to keep the total wind plus battery output equal to the
schedule.
0
5
10
15
20
25
30
35
40
45
50
0123456Power (MW)Time (hours)
Cycle Counting Methodology
Wind Output
Wind Schedule
Figure 4: Cycle Counting Method
In the first hour, there are several very small wind fluctuations. The analysis assumed that
deviations less than 500kW away from the schedule would not cause the battery to charge or
discharge. As such, the first hour has no charge/discharge cycles. The second hour, the wind
increases, therefore the battery would charge. The total energy absorbed by the battery (the
area between the curves represents the energy). During the third hour, the wind decreases, and
the battery would discharge. At the beginning of the fourth hour, the wind is still decreasing.
However, since the battery did not switch from charging to discharging, the beginning of the
fourth hour counts as a continuation of the third hour discharge. This five hour example results
in two large charge/discharge cycles followed by three smaller cycles.
Analysis was performed using the described cycle counting method and the simulated wind data
for a large wind farm. The charge/discharge cycles were tabulated for the entire year and
resulted in approximately 21,000 cycles. The majority of the cycles occur at small discharge
depths of less than 10%. Using an Excel curve fit equation to describe the battery cycle-life
characteristic shown in Figure 3, the expected battery life was calculated at 8.7 years. An
example of the calculation is shown below:
∑௧௨ ௬௦ ௧ % ை
ோ௧ௗ ௬௦ ௧ % ை
ଵ%ୀ%,ଶ%… for n = 6% ହଶ
ସ ൌ 0.143% of total battery life
There were 572 charge/discharge cycles between 4 and 6 percent for the 35 MWh battery
control strategy shown in Table 1 as case 1. At 6% depth of discharge, the SAFT Li-Ion battery
could withstand 400,000 cycles. Therefore, the 4-6% discharges account for 0.1% yearly battery
life degradation. This was added to all the other depth of discharge ranges, and resulted in an
annual battery degradation of 11.5%, or a battery life of 8.7 years for the first year’s data set,
and 8 years for the second year’s data set. Using the same controls, a battery with a 25 MWh
size would last for 6.3 years and 5.7 years respectively. When combined with the expected
need for feathering and regulation shortfall the economic impact of battery size can be
March 7, 2014 Page 19
determined. This analysis was performed for both the 17 MW wind farm and the 52 MW wind
farm options.
5 Technology Recommendation
When combined with the mature technology and lowest installation price, the recent
breakthroughs in the lead-acid battery technology make the advanced lead-acid battery
technology a front-runner for the stationary utility application market. The Sandia report further
reinforces the market trend toward lead-acid batteries with carbon enhanced electrodes such as
those provided by Xtreme Power, and Axion Power.
However, with lithium-ion’s dominance in the consumer electronics industry and its move into
the hybrid electric vehicle market, the lithium-ion battery technology should be considered. The
lithium-ion battery technology does provide superior performance when compared to the
advanced lead-acid battery technology. The high initial price of lithium-ion systems could be
offset by its superior cycle-life which would mean fewer replacement battery packs. Therefore, it
is important to study the impact of battery pack replacement costs when determining the best-fit
battery technology.
5.1 Financial Considerations
The Sandia National Laboratories recently updated an Energy Storage Systems Cost report
[2,3]. This report compared the different storage technologies and application types. The energy
storage types studied included lead-acid batteries, sodium-sulfur (Na/S), zinc-bromine (Zn/Br),
vanadium-redox (V-redox), Lithium-ion, compressed air (CAES), Pumped Hydro, High-speed
flywheels, and super capacitors. The analysis studied the 10-year ownership of the storage
device using the following factors:
Efficiency
Cycle-life
Initial Capital Costs
Operations and Maintenance
Storage-device Replacement
Of course, the storage system cycle-life, and replacement costs are dependent on the
application. The Railbelt regulation application is most closely represented in the Sandia report
[3] by frequent, short-duration discharges. Table 3 which was taken from the Sandia report
shows the costs in $/kW of the different technologies and applications. The most applicable data
set is the row inside the gold box. The results of this study give an idea of the cheapest
technological selection. The flywheel and super-capacitors are not suited for the Railbelt
regulation application due to their limited storage capacities. The cheapest choices are the
Carbon-enhanced electrode Lead-acid batteries, and the zinc-bromine batteries. While the cost
analysis used for the Sandia report did not have the level of detail that will be used to determine
the battery cycle-life for the Railbelt application, but it does provide a good baseline.
March 7, 2014 Page 20
Table 3: Energy Storage Systems Cost Update
The battery technologies that should be evaluated in greater depth for the Railbelt regulation
application are the advanced lead-acid battery technology and the lithium-ion technology. By
combining the battery sizing, expected regulation shortfall, expected wind feathering, battery
efficiency, and battery pack replacement frequency, the battery lifetime costs can be estimated.
The battery size and replacement frequency are closely related. If a large battery is purchased,
it will have a large initial capital cost. Since the battery is large, the same charge/discharge
cycles would result in a lower depth of discharge. Both the advanced lead-acid and the lithium-
ion battery technologies can withstand orders of magnitude more charge/discharge cycles at
low discharge depths. The result is that a larger battery will last longer and may need fewer
battery pack replacements during the battery system design life as was shown in the Battery
Life Evaluation section.
5.2 Economic Analysis – Advanced Lead-Acid vs. Lithium-Ion
A preliminary economic analysis was performed to compare the advanced lead-acid technology
against the lithium-ion technology. For the 52 MW and 17 MW wind farms’ various battery
energy capacities, the economic analysis took into account the following costs for assuming a
project life of 20 years and a discount rate of 5%:
Initial battery cost
Cost of battery losses (lithium-Ion batteries have better round-trip efficiency)
Battery pack replacement(s)
The battery life was calculated using the method discussed in the Battery Life Evaluation
section. The analysis determined the expected time between battery pack replacements. The
results of this analysis are shown below in Table 4.
March 7, 2014 Page 21
Table 4: Battery Life Based on Battery Capacity
10 6.2 6.6 26.5
12 8.3 8 30.4
14 11.2 9.8 34.8
25 5.2 6.3 22.4
30 6.4 7.4 24.7
35 8 8.7 27
42 11.4 11.3 31.3
17 MW Wind Farm
Battery
Size (MWh)
Xtreme
Power Saft Li ‐Ion Altair Nano
Li ‐Tritanate
Xtreme
Power Saft Li ‐Ion Altair Nano
Li ‐Tritanate
Battery
Size (MWh)
52 MW Wind Farm
The following assumptions were made for the economic analysis:
A 5% discount rate was used
Cost for energy lost due to battery inefficiency is assumed to be 100 $/MWh
Interconnection costs including a building to house the battery, the step-up transformer,
and circuit breakers will cost a total of $2.25M. The same interconnection costs will be
used for all energy storage capabilities, even though a smaller battery will need a
smaller building.
The Xtreme Power Battery costs that are based on a smaller-scale battery quote are:
o $500,000 per MW of power conditioning system
o $850,000 per MWh of initial battery pack installation
o $300,000 per MWh of replacement battery packs
The Saft Li-Ion battery costs that are based on smaller-scale battery quote are:
o $500,000 per MW of power conditioning system
o $2,500,000 per MWh of initial battery pack installation
o $1,250,000 per MWh of replacement battery packs (No quote received, assumed
½ initial cost)
The Altair Nanotechnolgies Li-Titanate battery costs that are based on smaller-scale
battery quote are:
o $1,500,000 per MW of power conditioning system
o $2,417,000 per MWh of initial battery pack installation
o $1,208,500 per MWh of replacement battery packs (No quote received, assumed
½ initial cost)
The results of the basic net present cost economic analysis are shown below in Table 5.
March 7, 2014 Page 22
Table 5: Battery Initial Installation Cost and 20 Year Project Cost
Case
(MWh) Initial $20‐year $Initial $20‐year $Initial $20‐year $
Case 10 19.25$ 26.41$ 35.75$ 58.84$ 51.92$ 53.53$
Case 12 20.95$ 27.04$ 40.75$ 59.38$ 56.75$ 58.36$
Case 14 22.65$ 27.11$ 45.75$ 65.91$ 61.59$ 63.19$
Case
(MWh) Initial $20‐year $Initial $20‐year $Initial $20‐year $
Case 25 48.50$ 66.95$ 89.75$ 133.96$ 137.68$ 141.16$
Case 30 52.75$ 72.40$ 102.25$ 152.66$ 149.76$ 153.25$
Case 35 57.00$ 73.28$ 114.75$ 166.94$ 161.85$ 165.33$
Case 42 62.95$ 74.68$ 132.25$ 166.44$ 178.76$ 182.25$
17 MW Wind Farm
Xtreme Power Saft Li ‐Ion Altair Nano Li ‐Titanate
Xtreme Power Saft Li ‐Ion Altair Nano Li ‐Titanate
52 MW Wind Farm
The economic results show that the initial cost is the dominant term of the 20-year project cost.
Also, the high cost of the lithium-ion battery technologies is not offset by its superior cycle-life.
This basic analysis shows that the lithium-ion technology is not as cost effective as the Xtreme
Power even though the Case 25 Xtreme Power battery packs needed three sets of replacement
battery packs over the 20-year project life. Due to the large price differential, additional factors
that would have slightly improve the lithium economics such as: smaller building and lower
shipping costs due to better energy density, lower maintenance costs, and a better
environmental image would likely not make up for the significant price differential. EPS
recommends the advanced lead-acid technology be used as the regulation resource.
When selecting a battery energy storage size, the frequency of shortfall hours should be
considered. Shortfall hours are the hours that the battery runs out of energy during the hour,
and the grid must supply for the shortfall. If a 25 MWh battery is selected to regulate the 52 MW
wind farm, the number of shortfall hours would be between 5 and 19 hours per year. It would be
up to the utilities to determine the best mix of battery size and frequency of shortfall. The 42
MWh case has been included since it is a combination of 25 and 16.67 MWh. This battery size
would mimic the control characteristics of the 25 MWh, but would leave 16.67 MWh as a
reserve for transient response to the loss of a generation unit, or the Kenai tie. The shortfall
hours would be eliminated since the 16.67 reserve capacity could be used for severe wind ramp
events, but during typical wind conditions, the 16.67 MWh would be reserved for a trip event.
The 42 MWh battery system would be more expensive, but would require fewer replacement
battery packs, and would provide the additional system benefit of transient event response to
the loss of a unit or Kenai tie.
6 Six Hour Energy Needs
6.1 Wind Regulation
The one hour analysis assumes the ability to change the unit schedules each hour. The single
contingency outage of the Kenai tie can island the Cooper Lake, and Bradley Lake regulation
resources. There are also many hours during a typical day when hydro resources are not
scheduled to meet the utility’s energy demands. The loss of these regulation resources severely
March 7, 2014 Page 23
limits the Anchorage area utilities’ ability to deal with the intermittent wind resource. This is due,
in part, to the current gas delivery contracts which are scheduled every six hours. So, when
either the Kenai tie is not energized or hydro is not scheduled, additional storage is necessary to
regulate the wind farm output or the wind farm must be curtailed. Due to the amount of energy
required for a six hour window (up to 300 MWh for a 52 MW wind farm), the battery and flywheel
technologies will not be economical at this scale. Therefore, the addition of flexible fuel storage
will be investigated as a means to regulate the intermittent wind resource while the Kenai
regulation resources are not available. Even with a second transmission line connecting the
Southcentral transmission system to the Kenai Peninsula, the loss of either transmission line
can reduce the transfer capacity by approximately 30 MW. Flexible fuel storage would be
needed to make up this shortfall until new gas schedules can be implemented.
Again, a power and energy requirement must be determined before any economic analysis can
be performed. The six-hour energy requirement for wind regulation will be determined in much
the same way that the one hour requirement was evaluated. The wind data was analyzed and
the largest wind down ramps were sorted by the initial wind power output were plotted in blue on
Figure 5. As an example, let’s assume the wind starts out at 25 MW. The worst possible case
would be an immediate ramp down to zero followed by six hours at 0 MW. This would result in
an energy need of 150 MWh (25 MW* 6 hours). However, in the field, the wind never ramps
down immediately, so the curve actually shows the worst case at 25 MW initial wind power to be
145 MWh. A linear curve was created to represent the six-hour energy needs against the wind
starting power, and is shown in red.
Figure 5: Six Hour Energy Needs Based on Wind Schedule
0
50
100
150
200
250
300
0 1020304050606 Hour Energy Needs (MWh)Starting Power (MW)
6 Hour Energy Needs Assuming No Capability to
Forecast Wind Ramps (52 MW Wind Farm)
Worst Energy Needs
Characteristic
March 7, 2014 Page 24
The years of wind data were analyzed by setting up a six-hour scheduling method. This
assumes that the Kenai intertie is out of service which isolates the hourly regulation resources
on the Kenai. Several different methods for creating a six-hour wind schedule were tested. Each
method that was tested assumed there is no ability to forecast wind production. Each of these
schedules was set and maintained for the entire six-hour time frame. For example, if the wind
schedule is 25 MW, and the wind stops at the beginning of the first hour, the rest of the six hour
scheduling period, the 25 MW wind schedule will be provided by the regulation resource.
The first method of setting a wind schedule uses the average wind output from the last 5
minutes of the previous hour as the basis for a wind schedule for the next six hours. This
method simply averages the 5 minutes before the hour begins and uses the average as the
schedule. The basic assumption made by the first method is “whatever the wind is doing now, it
will continue in the future.”
Second, the average wind power output of the previous six hours was used to create the
schedule for the next six hours. Again, the assumption is that the wind will continue what it did in
the previous six hours, but by using a longer time-frame, will not be influenced by short-term
wind fluctuations.
Third, the six-hour time frame from the previous day was used. This method assumes that the
wind will follow a daily cycle, and the six hours from the previous day are a good indication of
what will occur today.
Finally, a six-hour weighted average wind power was used. The weighting was assigned as
(hour-1)*0.5 + (hour-2)*0.2 + (hour-3)*0.1 + (hour-4)*0.1 + (hour-5)*0.05 + (hour-6)*0.05. This
method puts extra weight on the most recent hour, but would help remove some of the shorter
term volatility from the wind scheduling.
A simulation was run for a 17 MW and a 52 MW wind farm with each of the scheduling methods
discussed above. During this simulation, the wind data was used to determine the impact of
different scheduling philosophies on the amount of wind spilled, and the amount of regulation
shortfall. Based on the wind schedule, the regulation resource would either supply or absorb
power to maintain the wind schedule using the same formula used for the one hour regulation
analysis shown as Equation 1. The energy provided by the regulation resource during the six
hour schedule was calculated. For the six hour periods where more energy to regulate a
downward wind ramp is required than was available at the beginning of the six-hour schedule,
the time frame is listed as a shortfall. For the six-hour periods where more energy to regulate an
upward wind ramp is required than was available at the beginning of the six-hour schedule, the
time frame is listed as feathered, and the energy difference would be “spilled”. The first few days
of data is shown in Figure 6 and Table 6 below.
March 7, 2014 Page 25
Figure 6: Wind Power for First Provided Days
Table 6: 52 MW Wind Schedules for September 1 through September 2
Hour Schedule Type Initial StorageScheduleEnergy Used Schedule Type Initial Storage Schedule Energy Used
05 Min Avg 123 MWh 18.4 58.8 6 Hr Avg 123 MWh 18.4 58.8
65 Min Avg 64.2MWh 8.4 48.6 6 Hr Avg 64.2 MWh 10.1 58.5
12 5 Min Avg 15.7 MWh 1.5 8.9 6 Hr Avg 5.8 MWh 0.3 1.4
18 5 Min Avg 6.8MWh 0.3 ‐115.8 6 Hr Avg 4.4 MWh 0.0 ‐117.5
24 5 Min Avg 122.6 MWh 27.2 ‐107.0 6 Hr Avg 121.9 MWh 20.4 ‐147.5
30 5 Min Avg 229.5 MWh 41.1 ‐26.7 6 Hr Avg 269.5 MWh 46.2 3.5
36 5 Min Avg 256.2 MWh 42.5 197.5 6 Hr Avg 266.0 MWh 45.8 217.4
42 5 Min Avg 58.7 MWh 7.3 ‐49.5 6 Hr Avg 48.6 MWh 8.2 ‐43.8
0Prev Day 6hr Avg 123 MWh 18.4 58.8 6hr Weighted 123 MWh 18.4 58.8
6Prev Day 6hr Avg 64.2 MWh 10.1 58.5 6hr Weighted 64.2 MWh 9.1 52.8
12 Prev Day 6hr Avg 5.8 MWh 0.3 1.4 6hr Weighted 11.3 MWh 1.0 5.7
18 Prev Day 6hr Avg 4.4 MWh 0.0 ‐117.5 6hr Weighted 5.6 MWh 24.2 ‐116.6
24 Prev Day 6hr Avg 121.9 MWh 20.4 ‐147.5 6hr Weighted 123 MWh 43.3 ‐124.8
30 Prev Day 6hr Avg 269.5 MWh 37.4 ‐49.0 6hr Weighted 64.2 MWh 44.5 ‐13.8
36 Prev Day 6hr Avg 300 MWh 0.3 ‐55.8 6hr Weighted 123 MWh 7.0 209.5
42 Prev Day 6hr Avg 300 MWh 0.1 ‐92.9 6hr Weighted 64.2 MWh 20.9 ‐50.9
Figure 6 shows the first four days from the first year of wind power data with each vertical axis
line representing a six-hour period. During these 4 days there are three spikes of full/near full
0
10000
20000
30000
40000
50000
60000
0 1440 2880 4320 5760kW
Minute
Sept 1‐4
Series1
March 7, 2014 Page 26
wind power output. The first spike lasts for a full 12 hours. There are several ramps of the full
wind output within the six-hour schedules. Again, without the ability to forecast the wind, these
ramps must either be mitigated by the regulation resource, curtailing the wind, or a combination
of the two. An improved forecasting system could reduce the energy needs for regulating the
wind resources, and should be evaluated by the utilities. However, the impact of a state-of-the-
art wind forecasting system was not evaluated as part of this study.
Table 6 shows the energy storage usage for the first two days based on the different scheduling
methods described above. The top left quadrant shows the five minute averaging method. The
top right quadrant shows the six-hour average method. The bottom left quadrant shows the
results for the previous day six-hour average method. Finally, the bottom right quadrant shows
the results for using a six-hour weighted average scheduling method. The initial storage column
lists the amount of gas energy in storage at the beginning of each six-hour time frame. The
schedule lists the wind schedule used for the next six-hour time frame. The energy used column
lists the amount of gas storage energy that was used or saved during the six-hour time frame.
The five-minute average scheduling method example is explained below:
At hour zero, the gas storage has 123 MWh of energy. And the wind power output over five
minutes preceding the zero hour was 18.4 MW (schedule). During the next six hours, the wind
output steadily drops. In order to make up for the shortfall from the schedule, the gas storage
supplies the difference between the actual wind power, and the scheduled wind power. The total
energy used to maintain the wind schedule was 58.8 MWh. At hour six, the initial storage is 123
MWh – 58.8 MWh = 64.2 MWh. The new schedule is 8.4 MW, and again, the wind power drops
to zero over the next six hours, and the gas storage uses another 48.6 MWh. This process was
repeated for the entire year.
This analysis clearly showed that the best method for creating a schedule in terms of minimizing
feathered energy, minimizing shortfall energy, and minimizing total regulation usage was to use
the first method of averaging the last five minutes of the previous hour to create a wind
schedule. This means that the previous five minutes of wind data did the best job forecasting
the next six hours of wind power. This result is not surprising since there is a weak correlation of
wind power from day to day. This is easily observed by reviewing minutes 2880, and 4320
which are one day apart and vary by the full wind output.
Several year-long operational simulations using the five minute average wind scheduling
method were run to determine the frequency of regulation shortfall and wind feathering based
on the six-hour regulation resource and correction factor. Table 7 shows the results for a six-
hour regulation resource designed for the regulation of wind farm output.
March 7, 2014 Page 27
Table 7: Six‐Hour Regulation Simulation Results for a 52 MW Wind Farm
1 300 0.8 152666 0.0 0.0 0.0% 0
2 300 0.5 152666 8.6 32.0 0.0% 3
3 250 0.8 152666 461.7 0.0 0.3% 0
4 250 0.5 152666 289.1 32.8 0.2% 3
5 225 0.8 152666 2361.0 0.0 1.5% 0
6 225 0.5 152666 1397.4 34.5 0.9% 3
7 200 0.8 152666 5767.0 45.5 3.8% 2
8 200 0.5 152666 3329.0 56.7 2.2% 4
1a 300 0.8 162228 0.0 12.0 0.0% 1
2a 300 0.5 162228 0.0 91.3 0.0% 4
3a 250 0.8 162228 592.0 12.0 0.4% 1
4a 250 0.5 162228 380.0 91.9 0.2% 4
5a 225 0.8 162228 2477.8 28.6 1.5% 3
6a 225 0.5 162228 1435.6 98.2 0.9% 4
7a 200 0.8 162228 5406.9 70.8 3.3% 3
8a 200 0.5 162228 3276.0 170.2 2.0% 6
%feathered 6‐hr Schedule
Shortfall CountCaseEnergy
MWh
Correction
Factor
Total
wind
Feathered
MWh
Shortfall
MWh
The Energy MWh column lists the size of the gas storage facilities. The Feathered MWh lists the
energy that the gas storage facility would not be able to store during the simulation year, and
would force curtailment of the wind. The Shortfall MWh column lists the amount of energy that
the gas storage facility is unable to supply during the simulation. The 6-hr Schedule Shortfall
Count lists the number of six-hour schedules during which the gas storage is insufficient to
cover a wind down ramp. Based on these results, a 300 MWh gas storage facility would be
capable of providing the storage to fully regulate all wind up and down ramps for a year as
shown in case 1. However, in order to minimize the project cost, a storage facility could regulate
the large wind farm with as little as 200 MWh. It is recommended that the six-hour energy
storage be at least 200 MWh for the purpose of regulating the wind farm output.
March 7, 2014 Page 28
Table 8: Six‐Hour Regulation Simulation Results 17 MW Wind Farm
1 100 0.8 56320 0.0 0.0 0.0% 0
2 100 0.5 56320 8.6 18.7 0.0% 3
3 85 0.8 56320 556.0 0.0 1.0% 0
4 85 0.5 56320 351.0 6.0 0.6% 4
5 75 0.8 56320 1694.6 2.0 3.0% 2
6 75 0.5 56320 1085.6 6.7 1.9% 5
7 65 0.8 56320 3229.0 55.0 5.7% 10
8 65 0.5 56320 2124.3 51.2 3.8% 13
1a 100 0.8 59083 0.0 0.1 0.0% 1
2a 100 0.5 59083 8.6 18.7 0.0% 3
3a 85 0.8 59083 589.9 0.5 1.0% 1
4a 85 0.5 59083 366.5 18.7 0.6% 3
5a 75 0.8 59083 1631.6 16.7 2.8% 4
6a 75 0.5 59083 1029.8 32.0 1.7% 6
7a 65 0.8 59083 3354.3 73.2 5.7% 12
8a 65 0.5 59083 2061.3 66.7 3.5% 13
%feathered 6‐hr Schedule
Shortfall CountCaseEnergy
MWh
Correction
Factor
Total
wind
Feathered
MWh
Shortfall
MWh
The same analysis was performed to determine the regulation requirements for a 17 MW wind
farm as opposed to a 52 MW wind farm. The results of this analysis are shown above in Table
8. The storage sizes were selected to be approximately one third of the sizes studied for the 52
MW wind farm. However, this analysis shows that the percentage of feathered energy is greater
for the 17 MW wind farm for a storage facility of proportional size. This suggests that the 17 MW
wind farm size could be more volatile and may require more storage as a percentage of the
power than a larger wind farm. EPS would not recommend a gas storage facility smaller than 65
MWh for a 17 MW wind farm due to the frequency of energy shortfall, but anything 75 MWh or
bigger would be acceptable. The costs of the feathered energy would need to be weighed
against the cost of gas storage installation.
6.2 Loss of Kenai Tie
A secondary storage system sizing requirement is to compensate for the loss of the largest unit,
or the Kenai tie. Since the largest unit on the system in 2015 is expected to be 61 MW, the
largest single contingency in the existing transmission system will be the loss of the Kenai tie at
its maximum import into the Anchorage area. The line’s existing limit is 75 MW leaving Dave’s
Creek substation. When subtracting the loads along the line, this 75 MW import is less than 60
MW in the winter peak conditions, and as much as 68 MW in the summer valley condition.
However the loss of the Quartz Creek – Daves line section results in a loss of generation of
approximately 86 MW in the winter and 80 MW in the summer.
There are a few issues when considering energy storage for the loss of the Kenai tie. First, the
ability to reschedule the hydro resources to compensate for the wind ramps is removed since
these resources are islanded from the wind. Based on current gas scheduling contracts, the gas
turbines are scheduled for six hours at a time. This could result in up to six hours of schedule
mismatch. Second, the loss of the power import into the Anchorage area could result in load
shedding for cases that have minimal spinning reserve in the Anchorage area. The
recommended battery system for this secondary criterion could be used to supply for the lost
March 7, 2014 Page 29
import until the balancing authority has sufficient time to start a unit and prevent loadshedding
following the loss of the tie.
Following the construction of the recommended HVDC Beluga-Bernice Lake transmission line,
the outage of the existing Daves Creek – University line would result in a loss of approximately
30 MW of power during the maximum power transfer of 130 MW due to the 100 MW transfer
limit of the HVDC Intertie. This could result in the need for 180 MWh (30 MW * 6 hours) of
energy storage.
Prior to the construction of the new HVDC Intertie, or if the HVDC line is not constructed, the
maximum import capability into Anchorage is assumed to be 75 MW.
In order to give the balancing authority sufficient time to start a unit, the battery must be sized to
cover for the loss. PSS/E dynamics simulations were run with the Kenai tie importing 75 MW
into the Anchorage area. The case was created with minimal spinning reserve. Setting the case
up with minimal spinning reserve will give a worst-case simulation for any loss of generation or
import into the Anchorage area. A 50 MW battery system was added to the Railbelt database.
The battery was setup with a droop value that would force the battery to full output before the
first stage of load shed. The Kenai tie was then tripped. The result of a PSS/E simulation is
shown below in Figure 7.
It should be noted that the 75 MW import is not the worst case, single contingency event under
this import condition. The loss of the Quartz Creek – Daves Creek Line section results in a loss
of generation into the Anchorage area of 85-95 MW depending on the loads at Seward and
along the University – Daves Creek transmission line. However, the loss of this line is extremely
rare and it is unknown if the Railbelt utilities would limit the imports to cover the loss of this line
or accept limited load shedding should it occur. For purposes of this study, we have assumed
the utilities will accept limited load shedding for this contingency.
March 7, 2014 Page 30
Figure 7: Summer Valley Loss of Kenai Tie at Maximum Flow
The top set of traces show the frequencies at various places in the Railbelt system. The bottom
traces show the battery power outputs in red (MW) and blue (MVAR). The battery system
prevented load shed by quickly ramping up to its maximum power output of 50 MW. This
simulation resulted in the continued low frequency 30 seconds after the initial trip since there is
no additional room on any of the units to restore the frequency to 60 Hz. Without operator
intervention, the system would remain in this state until a unit could be started to restore
frequency and off-load the battery. It is assumed that a unit could be started in 20 minutes after
the loss of the Kenai tie. Therefore, a minimum of 16.67 MWh of battery storage is needed in
March 7, 2014 Page 31
order to supply 50 MW between the time when the tie is tripped, and a unit is started. The unit
would then run using the gas storage system for the remainder of the six-hour schedule.
After starting a gas turbine using gas storage and restoring the load along the Anchorage –
Kenai intertie, approximately 60 MW would be required from the gas turbine until the gas
schedules can be changed. In order to provide this energy for six hours, the energy requirement
would be approximately 60 MW for six hours or 360 MWh. This energy storage would be
sufficient to prevent scheduling conflicts with the gas delivery companies for the worst case
conditions of a) maximum import into the Anchorage area from the Kenai tie and, b) the loss of
the tie immediately after the current gas schedule begins. The 360 MWh of energy storage
would not be able to provide for any additional wind down ramps during the six-hour schedule.
Therefore, in order to provide some margin, EPS recommends that the storage requirement be
increased by 25% to 450 MWh to allow for additional gas storage to provide for some regulation
for wind up or down ramps after the loss of the Kenai tie. While the 450 MWh of storage is not
enough energy to deal with both the loss of the full wind output and the Kenai tie, it will be able
to fully handle the loss of the Kenai tie along with a moderate wind down ramp. The loss of the
full wind plant output coupled with the loss of the tie at its maximum import should be
considered an N-2 contingency, and should not be part of the requirements of the storage
system.
With the Kenai tie open, the utilities can change the gas schedules every six hours. With the
450 MWh of gas energy storage, the wind output could be fully regulated. In fact, the 52 MW
wind could be regulated with at least 200 MWh of gas storage as was shown in case 1 in Table
7.
Assuming the second Anchorage – Kenai intertie is built, the loss of the existing intertie would
result in the loss of 30 MW of capacity for the Anchorage utilities. In order to regulate the 17 MW
wind farm, and provide for the 30 MW lost capacity 250 MWh (70 MWh + 180 MWh) would be
needed. In order to regulate the 52 MW wind farm and provide for the 30 MW lost capacity, 380
MWh (200 MWh + 180 MWh) would be needed.
March 7, 2014 Page 32
Figure 8: Loss of AC Anchorage ‐ Kenai Intertie, 17 MW BESS, 30 MW Lost Import
Figure 8 shows the simulation result for the loss of the AC Anchorage – Kenai intertie with a 17
MW battery. It can be seen that the 17 MW battery can prevent the load shedding. The top left
set of traces show the frequencies at various places in the Railbelt system. The top left traces
show the battery signals such as power output (blue), per-unit energy (red), reactive power
(green). The bottom left set of traces show the line flows in (MW). The bottom right traces show
the bus voltages in per-unit at various places throughout the Railbelt system. This simulation
resulted in the continued low frequency 30 seconds after the initial trip since there is no
additional room on any of the units to restore the frequency to 60 Hz. Without operator
intervention, the system would remain in this state until a unit could be started to restore
frequency and off-load the battery. It is assumed that a unit could be started in 20 minutes after
the loss of the AC Anchorage – Kenai tie.
In order to provide some margin for the condition where the wind is decreasing, and the AC
Anchorage – Kenai Intertie is tripped, the recommended power and energy ratings are 25 MW
and 14 MWh respectively. This would allow the capability of regulating a smaller wind ramp
down coupled with the loss of the tie, but would not be able to respond to the full loss of the
wind farm and the tie.
This study assumes that the recommended HVDC tie is being constructed. However, another
option is to upgrade the existing line to allow a transfer capability of 125 MW. This case was not
March 7, 2014 Page 33
evaluated in this study. The minimum battery size and energy would be based on the trip of the
upgraded transmission line and the loss of the 125 MW import into the Anchorage area. For this
transmission configuration, the BESS should be sized as part of the overall transmission plan.
6.3 Gas Storage Description and Costs
In order to provide the energy required for longer term regulation of a six hour schedule, a
battery system is not financially reasonable. Therefore, EPS recommends the use of a
compressed natural gas storage system.
EPS recommends the use of containerized storage modules which store the natural gas at high
pressure in trailer-sized transportable modules. For a design capacity of 360 MWh, eleven
storage modules would be required. These storage modules would have approximately 25% of
“emergency” capacity that could be used to supply regulation energy for extreme wind ramp
events. Each storage module has 4 tanks that contain a total of 355,440 SCF of natural gas
compressed to 3,600 psig. Eleven storage modules would results in a total storage capacity of
3,909,840 SCF. This storage could supply approximately 450 MWh of energy.
The gas storage facility would be placed immediately adjacent to an existing power plant. The
storage facility consists of the gas storage modules, a compressor, and an electric driver motor,
and associated piping etc. The compressor requires a 1250 hp motor and would take the gas
from the pipeline which operates at 100 psig and compress it to 3,600 psig for storage.
The compressor and motor driver will be inside a pre-engineered metal building with concrete
floor slab and foundation which will protect the compressor and driver motor from the elements
and provide comfortable working conditions for maintenance. The storage modules will be
located outdoors, anchored to concrete slabs. The compressor building will incorporate electric
unit heaters for periods when the compressor is shut down for maintenance or repair. The
building would also have a ventilation system capable of discharging the heat rejected from the
motor and compressor. The ventilation system would also provide adequate air movement to
prevent the buildup of flammable gas within the building.
The facility would tie in to the existing natural gas pipeline serving the power plant. The natural
gas would be piped to the storage facility compressed and stored when the wind turbines are
producing excess energy. When the wind turbines are providing less power than scheduled, the
generators would ramp up and draw natural gas from the storage modules. During the storage
discharge, a pressure regulating station will knock down the gas pressure from its storage
pressure of 3,600 psig to the generator input pressure of 100 psig. The total cost for a 360 MWh
gas storage facility would be approximately $22.8 million.
The cost analysis assumed a two gas storage facilities with associated compressors, buildings,
site piping, storage modules, and labor expenses. A compressor building is included for the
ML&P power plant facility. The cost analysis assumes that the Southcentral Power Plant has
available room to house the natural gas compressor. One facility would have five gas storage
modules while the other would have six. Two storage facilities would provide increased flexibility
for maximizing the availability of on-line and off-line regulation resources.
Five different gas storage sizes were evaluated depending on the design criteria and the size of
the wind farm.
For a 17 MW Wind Farm
o A 70 MWh gas storage facility with 25% “emergency” capacity (87.5 MWh)
March 7, 2014 Page 34
o Gas storage located at one generation station
o Total system cost of $9.3 million
For a 17 MW Wind Farm with capability to pick up 30 MW import reduction for 6 hours
o A second Anchorage – Kenai intertie line is built, but the loss of the existing line
would result in a 30 MW reduction in the Anchorage import capacity
o A 250 MWh gas storage facility would be needed
o Gas storage located at two generation stations
o Total system cost of $18.2 million
For a 52 MW wind Farm
o A 210 MWh gas storage facility with 25% “emergency” capacity (262.5 MWh)
o Gas storage located at two generation stations for improved availability
o Total system cost of $18.2 million
For a 52 MW Wind Farm with capability to pick up 30 MW import reduction for 6 hours
o A second Anchorage – Kenai intertie line is built, but the loss of the existing line
would result in a 30 MW reduction in the Anchorage import capacity
o A 380 MWh gas storage facility would be needed
o Gas storage located at two generation stations
o Total system cost of $23.5 million
For capacity restoration for loss of largest unit/Kenai Tie
o No new Anchorage – Kenai Intertie
o A 360 MWh gas storage facility with 25% “emergency” capacity (450 MWh)
o Gas storage located at two generation stations for improved availability
o Total system cost of $23.5 million
7 Conclusions and Recommendations
To provide Railbelt utilities with the ability to regulate both variable generation resources and the
loss of the largest contingency in the Southcentral Railbelt, additional regulation resources are
required in the Railbelt system. Regulation resources utilizing batteries, flywheels and
compressed natural gas were evaluated in this study. Different battery and flywheel
technologies were evaluated first by their suitability to a regulation application, and secondarily
by their relative cost-effectiveness. Based on the long-term energy requirements, flywheel
technology was considered infeasible for the Railbelt system. Based on the suitability and cost-
effectiveness, EPS recommends the advanced lead-acid technology. The two main
manufacturers of suitable lead-acid technology include Xtreme Power Inc., and Axion Power
Inc. This report has focused on the Xtreme Power Inc. specific battery, but it is assumed that the
Axion Power manufacturer would provide a system with similar capabilities and costs.
The regulation resource sizing has been evaluated using primary and secondary criteria.
Primarily, the regulation resource should provide adequate regulation for the intermittent wind
resource. Secondarily, the regulation resource could be able to provide adequate response to
March 7, 2014 Page 35
prevent load shedding in the Anchorage area for the loss of either a large unit, or the Kenai tie
(with or without a HVDC intertie).
The power and energy requirements were evaluated separately. For a 17 MW wind farm
configuration EPS recommends a power capability of at least 17 MW. For a 52 MW wind farm,
EPS recommends a power capability of at least 50 MW. This power capability would be
sufficient to regulate the full range of net power from the 52 MW wind farm. Additionally, the 50
MW is the minimum power capability to prevent load-shedding for the loss of the Kenai tie while
operating at a maximum import into the Anchorage area.
For a 17 MW wind farm, to prevent an excessive number of hours during which the battery
cannot account for wind down ramps, the minimum battery energy that should be considered is
10 MWh. The one-hour regulation resource energy capability was evaluated using economic
analysis based on several factors including initial purchase price, battery pack replacement
frequency, and losses due to battery inefficiency. The lowest installation cost was for a 17 MW,
10 MWh battery energy storage system.
For a 52 MW wind farm, to prevent an excessive number of hours during which the battery
cannot account for wind down ramps, the minimum battery energy that should be considered is
25 MWh. The one-hour regulation resource energy capability was evaluated using economic
analysis based on several factors including initial purchase price, battery pack replacement
frequency, and losses due to battery inefficiency. The lowest installation cost was for a 50 MW,
25 MWh battery system.
However, a battery system with 42 MWh of energy capacity would be a reasonable alternative.
It would increase the original purchase price by 30%, but would provide the additional system
benefit of carrying enough storage capacity to provide 50 MW for 20 minutes. The reserve
capacity would be enough to provide enough energy to survive the loss of the Kenai tie at 75
MW import in to the Anchorage area without load shedding, and provide enough time to start an
additional gas turbine. The 42 MWh battery would only cost 11.5% more over the 20 year life of
the project since the battery packs would need less frequent replacement. EPS recommends a
battery energy capacity of 25 MWh solely for the regulation of the wind farm. However, with the
additional system benefits of a 42 MWh battery, the larger battery energy capacity should be
considered as an alternative in the final regulation resource decision process. Additionally, if the
DC tie is not built and the utilities move forward with an upgrade of the AC tie, the battery
MW/MWH capabilities should be determined as part of the coordinated transmission plan.
Similarly, for a 17 MW wind farm and the addition of a HVDC Beluga-Bernice Lake intertie, a
BESS capacity of 25 MW and 14 MWH of energy is recommended to prevent load shedding in
the Southcentral area following the largest contingency and to provide regulation of the wind
resource.
In order to minimize costs, it could be possible to implement a battery system with a smaller
power capability. An example would be to use the 25 MW battery system to regulate a 52 MW
wind farm. During a system configuration with both minimal online reserves, and a risk of losing
the full wind output, the wind farm could be curtailed prior to the loss of the total wind farm to
prevent an energy shortfall. This would result in a slight reduction of renewable energy over the
year, but would provide significant savings in the form of a smaller battery inverter system.
Due to the current gas contract schedule, the natural gas delivery amounts are set with six-hour
schedules. With the loss of the Kenai tie and the associated hydro regulation resources, the
Anchorage area utilities have very little capability to provide regulation for an intermittent
resource. With the installation of gas storage facilities, the utilities will have the ability to regulate
an intermittent resource such as wind. Again, the sizing of this resource was evaluated using
March 7, 2014 Page 36
the primary design criteria of providing wind regulation and the secondary criteria to cover the
loss of the Kenai tie. In order to fully regulate the full wind output over a six hour schedule for a
17 MW or a 52 MW wind farm, approximately 100 MWh or 270 MWh of energy is needed,
respectively. In order to minimize the costs, the gas storage capacity was reduced to a level that
would minimize costs while still minimizing the frequency of storage energy shortfall.
The secondary criteria of covering the loss of the Kenai tie would need 360 MWh without a
Beluga-Bernice HVDC tie or 180 MWH if the HVDC tie is constructed. In order to survive the
worst case loss of the Kenai tie and provide minimal wind regulation, EPS recommends a 25%
reserve margin of 90 MWh above the 360 MWh or 450 MWh for the no HVDC option and 45
MWH if the HVDC tie is constructed.
8 References
1. National Rural Electric Cooperative Association, Cooperative Research Network, “Energy
Storage for Renewable Energy and Transmission and Distribution Asset Deferral,” November
2009.
2. EPRI-DOE Handbook of Energy Storage for Transmission & Distribution Applications, EPRI,
Palo Alto, CA, and the U.S. Depatrment of Energy, Washington, DC: 2003. 1001834.
3. Schoenung, Susan and Eyer, James. Benefit/Cost Framework for Evaluating Modular Energy
Storage. SAND2008-0978. 2008.