The Full Cost of Electricity (FCe-)
Impact of Renewable Generation on
Operational Reserves Requirements:
When More Could be Less
PART OF A SERIES OF WHITE PAPERS
is an interdisciplinary initiative of
the Energy Institute of the University of Texas to identify and quantify
the full-system cost of electric power generation and delivery – from
the power plant to the wall socket. The purpose is to inform public
policy discourse with comprehensive, rigorous and impartial analysis.
THE FULL COST OF ELECTRICITY
The generation of electric power and the infrastructure that delivers it is
in the midst of dramatic and rapid change. Since 2000, declining renewable energy costs, stringent emissions standards, low-priced natural gas
(post-2008), competitive electricity markets, and a host of technological
innovations promise to forever change the landscape of an industry that
has remained static for decades. Heightened awareness of newfound options available to consumers has injected yet another element to the policy
debate surrounding these transformative changes, moving it beyond utility boardrooms and legislative hearing rooms to everyday living rooms.
This paper is one in
a series of Full Cost
of Electricity white
papers that examine
particular aspects of
the electricity system.
Other white papers produced through the study
can be accessed at the
University of Texas Energy
Institute website:
energy.utexas.edu
The Full Cost of Electricity (FCe-) study employs a holistic approach to
thoroughly examine the key factors affecting the total direct and indirect costs
of generating and delivering electricity. As an interdisciplinary project, the
FCe- synthesizes the expert analysis and different perspectives of faculty
across the UT Austin campus, from engineering, economics, law, and policy.
In addition to producing authoritative white papers that provide comprehensive assessment and analysis of various electric power system
options, the study team developed online calculators that allow policymakers and other stakeholders, including the public, to estimate the
cost implications of potential policy actions. A framework of the research initiative, and a list of research participants and project sponsors
are also available on the Energy Institute website: energy.utexas.edu
All authors abide by the disclosure policies of the University of Texas at
Austin. The University of Texas at Austin is committed to transparency and
disclosure of all potential conflicts of interest. All UT investigators involved
with this research have filed their required financial disclosure forms with
the university. Through this process the university has determined that
there are neither conflicts of interest nor the appearance of such conflicts.
Impact of Renewable Generation on
Operational Reserves Requirements:
When More Could be Less
Juan Andrade,
Department of Electrical and Computer Engineering
Yingzhang Dong, Department of Electrical and Computer Engineering
Ross Baldick, Department of Electrical and Computer Engineering
Andrade, Juan, Dong, Yingzhang, Baldrick Ross, “Impact of Renewable Generation on Operational Reserves Requirements: When More Could be Less,” White
Paper UTEI/2016-11-2, 2017, available at http://energy.utexas.edu/the-full-cost-of-electricity-fce/.
ABSTRACT
This report presents a qualitative and quantitative
description of the impact of renewable generation
on the requirements for ancillary services.
The fundamental concepts related to ancillary
services are presented in the report. First, the
need for ancillary services is described, and
then the quantification of those needs to satisfy
reliability requirements. Then pricing of those
services is described. The quantitative part of
the report presents an analysis of the impact
of nodal protocol revisions as well as installed
generation on the procured ancillary services
in ERCOT using a statistical approach. This
approach allowed correlations to be identified
between procured reserves, installed power, and
demand. In addition, the approach allowed a
ranking of nodal protocol revisions according
to their impact on reserves procurements.
LIST OF ACRONYMS
ACE
Area-control error
AS
Ancillary Services
CAISO
California Independent System Operator
CPS
Control Performance Standard
CREZ
Competitive Renewable Energy Zones
DCS
Disturbance Control Standard
ERCOT
Electric Reliability Council of Texas
FACTS
Flexible Alternating Current Transmission Systems
FERC
Federal Energy Regulatory Commission
ISO
Independent System Operator
NPRR
Nodal Protocols Revision Request
RTO
Regional Transmission Organization
RDD
Regression Discontinuity Design
TSP
Transmission Service Provider
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
3
1
| INTRODUCTION
BACKGROUND
The Full Cost of Electricity project aims to
understand and explain the cost causation in
the whole electricity supply chain. Given the
undesired consequences of interrupting electricity
supply, reliability considerations play a key role in
how electric systems are operated and planned.
These reliability considerations are relevant
for the full cost of electricity because they may
constrain the operation of the power system,
increasing its operational costs. This report is
focused on short-term reliability considerations, in
particular the ones covered by ancillary services.
Depending on the time horizon from which
reliability is considered, different assumptions
can be made about the capability of the system
to ensure reliability. Traditionally, in the short
term the concern is about being able to support
imbalances between generation and demand at
all times. The underlying assumption in such
analysis is a fixed generation fleet. In particular,
operational reserves compensate for increases or
decreases in net demand that are not tracked by
market or dispatch operations. A more detailed
description of them is provided in Chapter 2.
On the other hand, in the long term the concern is
whether there is enough installed capacity to satisfy
an increasing peak demand, that is, the concern
is with so-called resource adequacy. A traditional
metric used in this context is the reserve margin,
which is the quotient between peak power demand
(forecast, in case of forward estimates) divided
by the installed power (again, including forecast
new resources in case of forward estimates).
Based on the descriptions presented above, the
traditional time separation to analyze generation
reliability is as presented in Figure 1, which is
based on [1]. In the picture it is illustrated that
this report is only focused in a part of the system
security problem. It is important to keep in
mind that some terminology used in the security
context might appear in the context of resource
adequacy, but with a somewhat or completely
different meaning. For example, reserves in the
context of operational reserves is something
different to reserves for reserve margin.
FIGURE 1
Concepts related to system reliability.
System Reliability
System Security
How to mitigate generation-demand
imbalances at any time?
• Ancillary Services
• Operational Reserves
• ...
Short term (Minutes-Months)
Report focus
System Adequacy
Will there be enough capacity to
reliably satisfy the demand?
• Generation reserve margin targets
• ...
Long term (Months-Years)
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
4
Nowadays there are concerns about the capability
to maintain reliability under different disruptive
technology changes. One of those changes is the
massive development of utility scale renewable
generation, whose fluctuating nature is of concern
for maintaining security and generation adequacy
for power systems. It is important to note that
“system security” or “security” in the context of
this report are related to maintaining reliable
delivery of power to the loads and avoiding
disruptions of electricity service to customers.
Cyber-Security discussions related to information
technology security from computer hacking are
certainly important but separate and distinct
topics outside the scope of this analysis.
second issue is about the meaning of the results
obtained. From the regression models constructed,
it was possible to identify significant correlations
between variables such as installed generation,
demand, and operational reserves. However, no
conclusions about causation can be inferred from
these results alone. A third issue is the quality of
the data used. For some data sets, it is possible to
use finer time resolution, which might eventually
improve the quality of the results obtained. For
example, the updates in installed power can
be performed with monthly resolution rather
than the annual data used in this analysis.
PURPOSE OF THE REPORT
The main content of this report is about ancillary
services, which is covered in Chapter 2. It starts
by explaining the needs for providing ancillary
services. Then the concept of operational reserves,
and the different types of reserves, is presented.
After describing the concepts, the methodology
for calculating requirements for operational
reserves is described, with a particular emphasis
on the case of ERCOT. Next is presented a
brief discussion about the capability to provide
ancillary services, continuing with how a market
oriented approach can assign a price to a reserve.
Chapter 3 presents the results obtained from the
analysis of historical procured ancillary services.
This analysis is focused on the identification
of the effect on procured quantities of reserves
due to the changes in protocols that define
them, as well as due to the capacity of installed
wind power. The details about the calculations
performed are presented in an appendix at the
end of the document. Finally, conclusions for
the overall report are presented in Chapter 4.
The purpose of this report is to describe and
identify the consequences of the development
of utility scale renewable generation (principally
wind power) for the Electric Reliability Council of
Texas (ERCOT). This analysis is only focused on
security implications, which is evaluated through
the impact of utility scale renewable generation
on operational system requirements such as
procurements of particular ancillary services.
REPORT SCOPE
The analysis presented in this report has limitations
in terms of its capability to predict counterfactual
scenarios outside certain study periods. This is
because the statistical methodologies used only
allow for a meaningful result for predictions
confined around a narrow temporal range.
Examples of counterfactual scenarios that cannot
be explained are the implications had CREZ
not been developed and the possible massive
development of solar power in West Texas. A
REPORT STRUCTURE
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
5
2
| ANCILLARY SERVICES DESCRIPTION
BACKGROUND
Secure operation of a power system means that
no component should function outside its safe
operating range even in the event of disturbances
[2]. For example, it is required that system electric
frequency remains in very narrow ranges, despite
variation in supply and demand, because of
generator design characteristics. Another example
is the maintaining of voltage levels across the
transmission network, fulfilling the requirement of
maintaining load voltages within a narrow range to
avoid damage to equipment, among other issues.
In principle, security could be maintained
through interruption of loads. However, it
is generally accepted that consequences of
interrupting the electricity supply to loads have
significant consequences, which justifies the
need for provisions in the system to maintain
security without significant resort to load
interruptions even despite disturbances, such
as failures of generators or, in recent years,
significant changes in renewable production.
Based on the above ideas, and the ongoing market
restructuring in U.S. in the 90’s, the Federal
Energy Regulatory Commission (FERC) defined
in its Order 888 ancillary services as “those
services necessary to support the transmission
of electric power from seller to purchaser given
the obligations of control areas and transmitting
utilities within those control areas to maintain
reliable operations of the interconnected
transmission system” [3]. Traditionally, ancillary
services have been provided by generators, but
they can be provided by other types of technologies
such as transmission equipment and load.
Ancillary services as defined by FERC, and their
respective time scales are presented in Table 1.
Scheduling, System Control, and Dispatch are
tasks that an Independent System Operator
(ISO) or Regional Transmission Organization
(RTO) has to perform. It includes the schedule
of generating units, transmission resources, and
transactions before the fact and the monitoring and
control of transmission resources and generation
units in real-time to maintain reliability.
Reactive supply and voltage control is the capability
to provide the resources required in the system
to keep the voltage at certain locations within
desired ranges. The transference of real power
in an alternating current network requires the
establishment and sustainment of electric and
magnetic fields. An abstraction used to represent
this requirement is the so-called reactive power,
which has direct implications on the voltage in
the network. Generators can absorb or inject
reactive power to maintain voltage levels within
normal ranges. However, this action limits the
capability to inject real power into the network.
Therefore, it is a service that has an opportunity
cost associated. Moreover, the impact of voltage
regulation has a more local effect than system
TABLE 1
Ancillary services defined by FERC and their associated time scales [4].
Service
Time scale
Scheduling, System Control and Dispatch
Seconds to hours
Reactive supply and voltage control
Seconds
Regulation and frequency response
~ 1 minute
Energy imbalance
Hourly
Operational reserves – Synchronized reserve
Seconds to < 10 minutes
Operational reserves - Supplemental reserves
> 10 minutes
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
6
wide, therefore, there are few opportunities for
system-wide procurement of reactive power.
The regulation and frequency service is the
capacity to take corrective actions to balance
mismatches between generation and demand.
Under the normal operation of power systems,
these imbalances commonly happen because
of reasonable and inevitable forecasting errors,
the sudden failure of generators, lines, and
load disconnections. As a consequence of
these imbalances, the system frequency drops
or increases. This change in frequency may
be harmful for some elements in the system.
The above justifies the need for a service to
help regulate frequency, by controlling shortterm production of resources to restore
frequency towards its nominal value.
Energy imbalance is a service provided by
transmission service providers (TSPs) associated
with the discrepancy between scheduled and actual
delivery of energy to a load located in a control
area. In RTO and ISO markets it is provided by
the resources procured in the real-time market.
Synchronized reserves (called responsive reserve
in ERCOT) is the use of generating equipment and
interruptible load that can essentially immediately
respond to changes in frequency and that can
be dispatched to correct for generation/load
imbalances, typically with a requirement that it is
fully available within 10 minutes. Supplemental
reserve is the use of generating equipment and
interruptible load that can be fully available
within a somewhat longer period to correct for
generation/load imbalances caused by generation
or transmission outages. Supplemental reserve
differs from spinning reserve primarily in that
supplemental reserve need not begin responding
to an outage immediately. This service involves
the provision of additional generating capacity
that can commence after 10 minutes but must be
fully operational within thirty to sixty minutes.
It is important to notice that these FERC Order
888 definitions were promulgated in 1996, a
time prior to the massive development of utility
scale renewable generation. However, as is
pointed out in [5], these definitions are still
useful to some extent. However, the radical
change due to much greater levels of intermittent
renewables has required a revisiting of the
impact on ancillary services, in particular those
related with operational reserves. Therefore,
the rest of the report is focused on operating
reserves: its definitions and classifications,
requirements, and provision. The next section
describes operational reserves in more detail.
OPERATIONAL RESERVES
The balance between generation and demand has to
be maintained at all times due to the lack of costeffective massive-scale grid storage at the timescales
required. To do so, different strategies are used
for different timescales, which are illustrated
in Figure 2. Forward scheduling of the power
system includes schedules and unit commitment
directions to meet the general load pattern of the
day. Load following is the action to follow the
general trending load pattern within the day. This
is usually performed by economic dispatch and
sometimes involves the starting and stopping of
quick-start combustion turbines or hydro facilities.
Regulation is the balancing of fast second-tosecond and minute-to-minute random variations
in load or generation. This is done by centralized
control centers sending out control signals to
generating units (and some responsive loads) that
have the capability to rapidly adjust their dispatch
set points. These strategies represent the balancing
during normal conditions of the power system.
The above description does not consider sudden
demand-generation imbalance perturbations
which naturally happen during the operation
of power systems, typically due to the failure of
a generator. The existence of these imbalances
justify the need for additional operational
reserves besides the regulating reserves
described above. These reserves stabilize system
frequency within a prescribed amount of time,
and subsequent actions restore the system to
being secure with respect to another failure.
The supply-demand imbalances can be related
to the presence or not of an event in the system.
Events include severe and rare occurrences, but in
addition there are effectively continuous changes in
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
7
FIGURE 2
Power system operation time-frames [6].
supply-demand due to the addition of a large
number of individual consumption changes, which
we will refer to as “non-event considerations.”
This separation between event and non-event
considerations has been used to categorize different
types of operational reserves commonly used
in different systems interconnections. Based on
this, [6] created the classification presented in
Figure 3, which will be used in this report. It is
important to mention that there is no a single
classification for operational reserves, however
the classification selected here is general enough
to explain most of the issues involved in operation
reserves. Later on, the equivalent ERCOT
reserves will be associated with this description.
Regulating reserves covers the continuous fast and
frequent changes in load and generation that create
energy imbalance resulting in changes in system
frequency. It is the finest scale of balancing done
during normal conditions. It is used to correct the
current imbalance caused by load or generation
variation within the shortest applicable market
or economic dispatch interval, and correct shortterm load and renewable forecast errors. In some
areas, the shortest scheduling interval may be up
to an hour and in others this interval may be as
short as 5 minutes. What this means is that if the
system operator dispatched units thinking the net
load was moving in a certain direction, and the
magnitude or direction is different than anticipated,
the Regulating Reserve must be used to help
correct the discrepancy until the next economic
dispatch cycle can update. Required Regulating
Reserves would tend to increase with increasing
forecast error in the short-term wind forecast.
Following Reserve is analogous to Regulating
Reserve, but on a slower time scale. It is needed to
accommodate the variability and uncertainty that
FIGURE 3
Illustration of concepts
involved in categorization
of operating reserves [6].
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
8
FIGURE 4
Example contingency (loss of supply)
event and typical responses [6].
occur during normal conditions. The definition
used in [6] represents the movements that are
reflected in the economic dispatch to correct an
imbalance that is forecast to occur in the future.
Issues that impact the need for Following Reserve
are much less random than those for Regulating
Reserve, but larger in magnitude than Regulating
Reserve. Following reserve covers both typical
load and variable generation patterns and interschedule interval variability. In the case of ERCOT,
the load and generation short-term trend forecasts
are covered by real-time economic dispatch. On the
other hand, the 95% of the remaining generation
and load uncertainty is covered by Regulationup, Regulation-down, and Non-spinning reserve.
In addition, Non-spinning reserve can be used
for loss of generating units. The requirement for
participation in providing this service is to be either
on-line (for Regulation-up and Regulation-down)
or off-line (for Non-Spinning Reserve), with full
response required within thirty minutes [6].
Unlike Regulating and Following Reserves,
Contingency Reserves are called upon during rare
sudden events. The events usually considered are a
large sudden loss of supply either from generating
resources or large transmission lines carrying
imports, but more generally can consider loss of
large blocks of load as well. Contingencies occur
quickly and much of the reserves must start acting
essentially immediately. Figure 4 presents a typical
response to a sudden loss of a large generating unit.
Immediately following the event, the generator
rotating masses will supply kinetic energy, partly
mitigating the generation-demand imbalance.
As a consequence, system frequency experiences
changes due to deceleration of generator rotors.
It is important to mention that there are current
advances in power electronics that allow nonsynchronous generators (e.g. wind and solar
power) to provide synthetic inertial response [7].
Ramping reserve is probably the least well defined
category of the list presented by [6] and is only
explicitly procured currently in some markets, such
as flexi-ramp in CAISO (California Independent
System Operator) [8]. This type of reserve is used
for rare severe events that are not instantaneous
in nature. Large load variations occur every
day, are predictable, and are met with Following
Reserve and the action of the energy markets
rather than Ramping Reserve. Due to the greater
unpredictability of wind and solar, infrequent
large magnitude events may occur that require
additional Operating Reserves. The separation
between Following Reserves and Ramping Reserves
is that the first may cover most of the possible
deviations, and ramping reserves the remaining.
Primary reserves correspond to a certain portion
of contingency reserves that must be automatically
responsive to changes in frequency. Primary
reserve is needed to stop frequency deviations from
becoming too large. This protects generators from
excessive frequency deviations which can create
conditions that may cause damage to the generators
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
9
TABLE 2
Similarities between concepts in [6] and ERCOT.
Ela, Milligan, and Kirby AS definition
ERCOT similarities
Regulating reserves
Regulation-up and regulation-down
Following Reserve
Real-time dispatch, and Non-spinning reserve
Contingency reserve – Primary
Responsive reserve
or set off under-or over-frequency relays which can
shed system load or disconnect generators, which
would further worsen the effect of the contingency.
Secondary reserve including both Contingency
Reserve and Ramping Reserve are used to return
the frequency back to its nominal value and
reduce ACE (Area Control Error) back to zero.
Tertiary reserve is unique because it is the only
reserve category that is not deployed for energy
imbalance, but is instead deployed for reserve
insufficiency. In other words, it is held in some
manner so that when certain operating reserve
types are used to correct the energy imbalance
and converted into energy, it is used to restore that
form of Operating Reserve. Tertiary Reserves do
not need to be as fast as the secondary reserves,
and are used to allow the system to recover reserves
by introducing new economic generation.
Based on the previous review of reserves, and
the definitions for ERCOT reserves provided in
[9], Table 2 presents some similarities between
the definitions of operational reserves used in
this chapter, and the current ERCOT reserves
recognized. The way in which these reserves are
determined will be described in the next section.
QUANTIFICATION OF OPERATIONAL
RESERVES REQUIREMENTS
Ancillary services requirements that ISOs and
RTOs procure in their markets are directed
by reliability requirements, which vary among
different interconnections. In North America,
the reliability requirements are specified by
North American Electric Reliability Corporation
(NERC). Based on these requirements, and
the variances associated to demand and
renewable generation, procured operational
reserves are determined to ensure certain
reliability performance indices satisfy required
specifications. Some of the specifications are
statistical. However, for certain types of reserves,
such as contingency reserves, deterministic
reserves requirements based on conservative
considerations (such as the largest creditable
generation contingency) have been traditionally
used, although it is also possible to define them
economically based on probabilistic terms [10].
Examples of the application of these reliability
requirements in the quantification of operational
reserves requirements can be found in the Eastern
Wind Integration and Transmission study [11],
and in Western wind and solar integration study
[12], which determine different operational
reserve requirements in terms of renewable
generation production, demands, as well as the
variances associated to them, among other things.
In this section, the requirements in ERCOT
for each operational reserve are described.
REQUIREMENTS FOR REGULATING
RESERVES IN ERCOT
For the purpose of regulating reserves NERC
established the NERC standard BAL-005 [13]
which states that “Each balancing authority shall
maintain regulating reserve that can be controlled
by automatic generation controllers to meet the
Control Performance Standard (CPS),” where the
metric CPS is in terms of the frequency excursion
under and above the system frequency. This means
that each ISO/RTO has to find a way to define
its operational reserves in such way to fulfill CPS
criteria. In the case of ERCOT, which is a single
balancing authority, it has to fulfill that its CPS1 is
greater than 100%. (The maximum possible value
is 200%). As can be seen in Figure 5, ERCOT has
easily fulfilled this standard. In order to procure
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
10
FIGURE 5
Historical ERCOT CPS1 [15].
regulation, ERCOT splits the regulation into
regulation-up and regulation-down. The amount
of reserves procured are based on 98.9th percentile
of regulation reserve utilized in previous 30 days
and on the same percentile of regulation utilized
in the previous year and adjusted by installed
wind penetrations, as discussed below. This
requirement is differentiated on an hourly basis.
To adjust the amount of Ancillary Services
procured by installed wind penetration, ERCOT
calculates the increased amount of wind capacity
compared to the previous months and uses
a look-up table to add increased amounts of
regulation based on a study performed for the
region [14]. In addition, if during the course
of the past 30 days, the average CPS1 score was
less than the 100% target, an additional 10%
of regulation will be procured during those
hours where it was less than 100%. If the CPS1
score was less than 90%, an additional 20% of
regulation service will be added [6], [9].
FIGURE 7
Historical procured
regulation-up reserves
in ERCOT.
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
11
FIGURE 8
Historical procured responsive reserve in ERCOT.
As a result of these correction mechanisms,
CPS1 compliance has been maintained over
time. The consequence of that compliance can
be seen in the change over time of the procured
regulation-up and regulation-down as presented
in Figure 6 and Figure 7, respectively.
From both figures it can be appreciated that there
have been changes over time for these quantifies,
with CPS1 scores generally improving, while
regulation procurement quantities have remained
approximately static or reduced, despite significant
increases in the installed wind capacity.
REQUIREMENTS FOR RESPONSIVE
AND NON-SPINNING RESERVES
In the same way as for regulating reserves, NERC
Disturbance Control Standard (DCS) Standard
BAL-002 [16] establishes that “as a minimum,
the balancing authority or reserve sharing groups
shall carry at least enough contingency reserve
to cover the most severe single contingency”.
In the case of ERCOT, the contingency reserve
considered have been the trip of nuclear
generation units for at least 2300 MW as can
FIGURE 9
Historical procured non-spinning reserve in ERCOT.
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
12
be seen in Figure 8. However, based on the
conclusions of a study performed for ERCOT
[14], an additional 500 MW of responsive reserves
was added to the requirement in order to ensure
that reserve regulation, responsive reserve, and
non-spinning reserve have a capacity to respond
to the 95% of the uncertainty in the net load. [9].
This underpins the increase in responsive reserve
requirement from 2300 MW to 2800 MW in 2012,
and the corresponding decrease in non-spinning
reserve requirement.
Regarding Non-Spinning Reserves, historically
the need for it has occurred during hot weather,
during cold weather, during unexpected
changes in weather, or during large unit trips
when large amounts of spinning reserve have
not been on line. That can explain in part
the sudden changes that can be seen for the
procurement of this reserve in Figure 9 [9].
ERCOT FUTURE NEEDS FOR
ANCILLARY SERVICES
Finally, it has to be mentioned that the massive
development of utility scale non-synchronous
generation (e.g. wind power) has raised concerns
about the capability of the system to maintain
security. In particular, the replacement of
synchronous generation by non-synchronous
generation may produce significant reduction
in system inertia, which makes it less resilient to
contingencies. In order to overcome this situation,
ERCOT has proposed an unbundling of ancillary
services associated with frequency regulation,
in order to incentivize market participants to
provide inertia (instead of being a compulsory
service like it is nowadays) [17]. According to
the reference, the implementation of these future
ancillary services has a benefit cost ratio of 10.
PROVISION OF ANCILLARY SERVICES
Once the requirement of ancillary services is
identified, a way is needed to procure it. First, the
technical capability to provide the service must be
identified, and second is the need for a mechanism
to compulsorily or voluntarily obtain these
requirements. Traditionally, ancillary services have
been provided to a great extent by conventional
generators (e.g. thermal and hydro units),
and to a lesser extent by other devices such as
capacitor banks, FACTS, synchronous motors, etc.
Nowadays, there are more technical alternatives
such as non-traditional generation ancillary
services, demand response, and energy storage.
COMPULSORY PROVISION OF ANCILLARY
SERVICES
This way to provide ancillary service is the closest
to the vertical integration case. Typically, for a
new generator to enter into the power system, it
is required to provide certain types of ancillary
service. A typical example in North America is
the contribution of primary frequency response.
In [2] several reasons are given for why this
is not necessarily a good economic policy:
• It may cause unnecessary investments
compared to what is needed. For
example, not all generating units need
to take part in frequency control to
maintain the security of the system.
• It does not give room for technological
or commercial innovation because there
is no differentiation amongst provision.
• It generates a negative perception among
providers because they feel that they are
forced to supply a service that adds to
their costs without being remunerated.
MARKET ORIENTED PROVISION OF
ANCILLARY SERVICES
The individual ancillary services differ substantially
in their features, competitiveness, provision,
and pricing. Operating reserves, for example,
can likely be provided by competitive markets.
The primary supplier cost for this service is the
opportunity cost associated with foregone energy
sales; significant fuel costs are incurred only when
these reserves are called upon to respond to the
loss of a major generation transmission outage.
Injection and absorption of reactive power, on
the other hand, must be provided close to the
location where the voltage control is needed.
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
13
FIGURE 10
Comparison of all-in Prices
across U.S. Markets [19].
Therefore, it may not be feasible to create
competitive markets for this service. Rather,
the pricing and provision of voltage control may
continue to be regulated. Capital costs are the
dominant costs for this service for both generators
and transmission equipment. Opportunity costs
arise only when generators are operating at or
near full real-power output and are called upon to
increase reactive-power output beyond the level
associated with the unit’s rated power factor. [4]
in a sequence determined by the speed of the
reserves required. The experiences obtained from
this approach were unsatisfactory in terms of
efficient price formation for reserves. For example,
it is expected that there should be a higher price
for a faster response, which was not the case in this
original approach [18]. From these experiences,
the now commonly used co-optimization approach
for valuation of ancillary services and energy was
developed. Co-optimization is used in ERCOT.
In a market oriented approach to provide ancillary
services, market participants make offers to provide
their services, which are cleared by obtaining
a price for these services. Figure 10 shows the
prices paid for different electricity product across
different interconnections in U.S. from 2011 to
2014. It can be appreciated that the price associated
with ancillary services is small in comparison
with energy prices. In particular, average ERCOT
ancillary services price represents in average
between 3% and 4% of the annual energy prices.
CO-OPTIMIZED PROVISION OF ENERGY
AND RESERVES IN A CENTRALIZED
ELECTRICITY MARKET1
There are different ways to obtain a price for
ancillary services. In early years of competitive
electricity markets, energy and each type of
operational reserve were treated in separated
markets. These markets were cleared successively
In a co-optimized energy and reserves markets,
the price of the ancillary services is obtained
as Lagrange multipliers on the reserve demand
constraints, in an economic dispatch.
To fix ideas, consider a power system with four
generators, in which the transmission network
is neglected like the one presented in Figure 11.
Table 3 presents the parameters of the generators.
In this system only the requirements of spinning
reserves and energy will be considered.
1 The example presented here was extracted and paraphrased from [2].
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
14
11 with four generators, in which the transmission network is neglected
FIGURE 12
To fix ideas, consider aFIGURE
power system
like the one presented in Figure 11. Table 3 presents the parameters of the generators. In this system only
System used to illustrate the co-optimization
Amount of reserves that generating units 2 and 3 can provide
the requirements of spinning reserves and energy will be considered.
of energy and ancillary services.
G1
G2
as a function of the amount of electrical energy production.
G3
Reserve
capacity
(MW)
G4
Reserve
capacity
(MW)
Unit 2
160
Unit 3
190
D
Notice in this example that some generators
cannot provide reserves. A reason for this could
be that they are so slow to change generation as
Table 3. Generators
data for of
illustrative
system. reserves,
to be considered
incapable
providing
or that
they simply
to provide
Generating
unit
Marginaldon’t
cost ofwant Pmax
[MW] them
Rmax [MW]
into the market
for[$/MWh]
strategic reasons. Another
energy
Reserve
Reserve
1
250a generator
0
important2issue is that
although
capacity
capacity
2
17
230
160
might be willing to offer
(MW)up to certain amount
(MW)
3
20
240
190
of
reserves,
the
generation
unit
capacity
limits
4
28
250
0
Unit 2 of
the total amount of reserves plus generation
160
190
energy at every time. This situation is illustrated
Notice in this example that some generators cannot provide reserves. A reason for this could be that they
Figure as
12,towhere
generation
units
2 and reserves, or that they simply
are so slow to changein
generation
be considered
incapable
of providing
areinto
limited
in the
of reserves
they
don’t want to provide3them
the market
foramount
strategic reasons.
Another
important issue is that although
a generator might be can
willing
to offer up
to a
certain
of reserves,
the generation unit capacity limits
provide,
for
givenamount
generation
level.
Figure 11. System used to illustrate the co-optimization of energy and ancillary services.
70
230
50
Energy
produced
(MW)
Unit 3
the total amount of reserves plus generation of energy at every time. This situation is illustrated in Figure
12, where generationIn
units
3 are limited
in thethe
amount
of reserves
they can provide, for a given
this2 and
example,
we seek
generation
levels
generation level.
70 the total230 Energy
50
240 Energy
for the four generators that minimize
produced
produced
generation
cost,levels
assuming
thatgenerators
they arethat
already
In this example, we seek
the generation
for the four
minimize the(MW)
total generation
(MW)
cost, assuming that they
are already
generate.
This objective
is expressed in (1). This
committed
to committed
generate.toThis
objective
is expressed
problem is called economic
Notice that
it is a different
problem
than the unit commitment
in (1).dispatch.
This problem
is called
economic
dispatch.
consumption
such that the net fuel consumption
problem, in which apart from the generation levels of generators, the commitment status
is also
Notice that it is a different problem than the unit
difference
is
assumed
to be near zero. Responsive
considered. Also notice that non-spinning reserves don’t have a cost by themselves given a generator is
commitment
problem,
in
which
apart
from
the
not consuming fuel to produce them. Regulation up or down reserves fuel consumption increases
reservesand
(spinning reserves) do require thermal
generation
levels
of generators,
thedifference
commitment
decreases fuel consumption
such that
the net
fuel consumption
is assumed to be
near zero. to be synchronously on-line and
generations
Responsive reserves (spinning
do require thermal
to be
synchronously
on-line andsome modest amount of fuel.
status isreserves)
also considered.
Alsogenerations
notice that
nonconsume
consume some modest
amount
of
fuel.
spinning reserves don’t have a cost by themselves
The demand balance is presented in equation
(2). Notice that this balance doesn’t correspond
necessarily to the real conditions that might occur
given a generator is not consuming fuel to
18 up or down reserves
produce them. Regulation
fuel consumption increases and decreases fuel
TABLE 3
Generators data for illustrative system.
Generating unit
Marginal cost of energy [$/MWh]
Pmax [MW]
Rmax [MW]
1
2
250
0
2
17
230
160
3
20
240
190
4
28
250
0
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
15
Figure 12. Amount of reserves that generating units 2 and 3 can provide as a function of the amount of electrical energy
Figure 12. Amount of reserves that generating unitsEnergy
2 and
3 can provide as a function of the amountEnergy
of electrical energy
production.
70
230 (MW)
50
240 (MW)
production.
produced
produced
Figure 12. Amount of reserves that generating units(MW)
2 and 3 can provide as a function of the amount(MW)
of electrical energy
production.
12. Amount
of reserves
that generating
units (2).
2 andNotice
3 can provide
as abalance
function of
the amount
of electrical
energy
TheFigure
demand
balance
is presented
in equation
that this
doesn’t
correspond
necessarily
production.
The demand balance is presented in equation (2).
Notice that this balance doesn’t correspond necessarily
to the real conditions that might occur in the system at any time. But it is a reference for the production
to
the
real conditions
that But
might
occur
in the system at
any reserves
time. Butthan
it isthose
a reference
forand
theso
production
in the
the
system
at any
time.
a reference
more
required,
the
of
generators.
Moreover,
ititisisused
for pricing
The
energy
prices
correspond
to the
Lagrange
The
demand
balance
is presented
in
equation
(2).purposes.
Notice
that
this
balance
doesn’t
correspond
necessarily
of
it is usedMoreover,
for pricing purposes.
The energy
prices correspond
to the The
Lagrange
forthe
thegenerators.
productionMoreover,
of the generators.
constraint
is represented
as an inequality.
to
the realassociated
conditionsto
that
occurEquation
in the system
at any time.
But it is a reference
for the
multiplier
thismight
equation.
(3) represents
the requirement
for reserves
forproduction
the whole
it
is
used
for
pricing
purposes.
The
energy
prices
price
associated
to
these
reserves
will
be
obtained
multiplier
associated
to
this
equation.
Equation
(3)
represents
the
requirement
for
reserves
for
the
whole
The demand balance is presented in equation (2). Notice that this balance doesn’t correspond necessarily
of
the generators.
Moreover,
itthat
is used
for
pricing to
purposes.
The
energy prices
correspond
to the
Lagrange
system.
It can
be appreciated
it isassociated
possible
have
more
reserves
than
thoseassociated
required,
and
so the
correspond
to
the
Lagrange
multiplier
from
the
Lagrange
multiplier
to
this
system.
It can
be appreciated
that
it is in
possible
to have
more
reserves
those required,
and so the
to
the real
conditions
that might
occur
the system
at any
time.
But it isthan
a reference
for the production
constraint
is represented
as(3)
anrepresents
inequality.
The (3)
price
associated
torequirement
theseinreserves
will
be(5),
obtained
from
multiplier
associated
to this
equation.
Equation
represents
theFinally,
for reserves
for
the(6),
whole
to
this
equation.
Equation
the
equation.
equations
(4),
and
are
constraint
is
represented
as
an
inequality.
The
price
associated
to
these
reserves
will
be
obtained
from
of the generators. Moreover, it is used for pricing purposes. The energy prices correspond to the Lagrange
the
Lagrange
multiplier
associated
to
this
equation.
Finally,
in
equations
(4),
(5),
and
(6),
are
presented
system.
It
can
be
appreciated
that
it
is
possible
to
have
more
reserves
than
those
required,
and
so
the
requirement
for
reserves
for
the
whole
system.
presented
the
bounds
for
generation
and
reserves.
the Lagrange
multiplier
associated
to Equation
this equation.
Finally, inthe
equations
(4), (5),
(6), are
multiplier
associated
to this
equation.
(3) represents
requirement
for and
reserves
for presented
the whole
constraint
isfor
represented
inequality.
The price associated to these reserves will be obtained from
the
bounds
generation
and
reserves.
It can
be appreciated
that itas
isan
possible
to have
the bounds
forbe
generation
and that
reserves.
system.
It can
appreciated
it is possible to have more reserves than those required, and so the
the Lagrange multiplier associated to this equation. Finally, in equations (4), (5), and (6), are presented
(1)
constraint
is represented as an inequality. The price associated to these reserves will be obtained from
the bounds for generation and reserves.
the Lagrange multiplier associated to this equation. Finally, in equations (4), (5), and (6), are presented
𝑍𝑍 = min 2𝑃𝑃1 + 17 𝑃𝑃2 + 20𝑃𝑃3 + 28𝑃𝑃4
(1)
the bounds for generation and reserves.
𝑍𝑍 = 𝑃𝑃min
(1)
1 ,𝑃𝑃2 2𝑃𝑃1 + 17 𝑃𝑃2 + 20𝑃𝑃3 + 28𝑃𝑃4
𝑃𝑃1 ,𝑃𝑃2
𝑍𝑍 = min 2𝑃𝑃1 + 17 𝑃𝑃2 + 20𝑃𝑃3 + 28𝑃𝑃4
(1)
𝑃𝑃 + 𝑃𝑃 + 𝑃𝑃 + 𝑃𝑃 = 𝐷𝐷
𝑅𝑅1 1+ 𝑅𝑅2 2+ 𝑅𝑅33+ 𝑅𝑅44 ≥ 250
𝑅𝑅1 + 𝑅𝑅2 + 𝑅𝑅3 + 𝑅𝑅4 ≥ 250
𝑃𝑃1 + 𝑃𝑃2 + 𝑃𝑃3 + 𝑃𝑃4 = 𝐷𝐷
𝑅𝑅1 + 𝑅𝑅2 + 𝑅𝑅3 + 𝑅𝑅4 ≥ 250
0 ≤ 𝑃𝑃1 ≤ 250
0 ≤ 𝑃𝑃1 ≤ 250
0
≤
𝑅𝑅1 + 𝑅𝑅02 ≤
+ 𝑃𝑃
+ 230
𝑅𝑅4 ≥ 250
≤
𝑃𝑃𝑅𝑅223 ≤
230
0
≤
𝑃𝑃
≤
240
3 ≤ 250
0
≤
𝑃𝑃
1
0 ≤ 𝑃𝑃3 ≤ 240
≤ 250
0
230
0≤
≤ 𝑃𝑃
𝑃𝑃442 ≤
250
0
≤
𝑃𝑃
≤
250
1
0
≤
𝑃𝑃
≤
240
3
0
≤
𝑃𝑃
≤
230
2
0 ≤ 𝑃𝑃419≤ 250
0 ≤ 𝑃𝑃319≤ 240
0 ≤ 𝑃𝑃419≤ 250
(2)
(3)
(3)
(2)
(3)
(4)
(4)
(3)
(4)
𝑃𝑃1 ,𝑃𝑃2
(2)
(1)
(2)
(2)
𝑍𝑍 = min
17𝑃𝑃𝑃𝑃2++𝑃𝑃20𝑃𝑃
𝑃𝑃12𝑃𝑃
+1𝑃𝑃++
= 3𝐷𝐷+ 28𝑃𝑃4
𝑃𝑃1 ,𝑃𝑃
𝑃𝑃21 + 𝑃𝑃22 + 𝑃𝑃33 + 𝑃𝑃44 = 𝐷𝐷
(3)
(4)
(5)
(4)
19
𝑅𝑅1 = 0
𝑅𝑅 = 0
0 ≤ 𝑅𝑅12 ≤ 160
0 ≤ 𝑅𝑅2 ≤ 160
0 ≤ 𝑅𝑅3 ≤ 190
0 ≤ 𝑅𝑅3 ≤ 190
𝑅𝑅4 = 0
𝑅𝑅4 = 0
(6)
(5)
(5)
𝑃𝑃1 + 𝑅𝑅1 ≤ 250
𝑃𝑃 + 𝑅𝑅 ≤ 250
𝑃𝑃12 + 𝑅𝑅12 ≤ 230
𝑃𝑃2 + 𝑅𝑅2 ≤ 230
𝑃𝑃3 + 𝑅𝑅3 ≤ 240
𝑃𝑃3 + 𝑅𝑅3 ≤ 240
𝑃𝑃4 + 𝑅𝑅4 ≤ 250
𝑃𝑃4 + 𝑅𝑅4 ≤ 250
(6)
(6)
TABLE 4
Table 4.Solution
Solutionofofthethe
optimization
problem
for different
demand
optimization
problem
for different
demand
levels. levels.
Table 4. Solution of the optimization problem for different demand levels.
Demand
Demand
Demand
[MW]
[MW]
[MW]
300-420
300-420
300-420
420-470
420-470
420-470
470-720
470-720
470-720
𝑃𝑃P11
𝑃𝑃1
[MW]
[MW]
[MW]
250250
250
250250
250
250
250250
𝑅𝑅R11
𝑅𝑅1
[MW]
[MW]
[MW]
0 0
0
0 0
0
0
0 0
𝑃𝑃22
P
𝑃𝑃2
[MW]
[MW]
[MW]
50-170
50-170
50-170
170170
170
170
170170
R𝑅𝑅2 2
𝑅𝑅2
[MW]
[MW]
[MW]
60 60
60
60 60
60
60
60 60
P𝑃𝑃
𝑃𝑃3 3
[MW]3
[MW]
[MW]
0 0
0
0-50
0-50
0-50
50
5050
R𝑅𝑅
3 3
𝑅𝑅
[MW]3
[MW]
[MW]
190
190
190
190
190
190
190
190
190
P4𝑃𝑃4
𝑃𝑃
[MW]4
[MW]
[MW]
0 0
0
0 0
0
0-250
0-250
0-250
R4𝑅𝑅4
𝑅𝑅
[MW] 4
[MW]
[MW]
00
0
00
0
0
00
In Table 4 are presented solutions for different demand levels. The checking of the results by hand
In Table 4 are presented solutions for different demand levels. The checking of the results by hand
provides
insights
about(FCe-)
how ancillary
services
are priced
under Costs
a co-optimization
simplified
The Full Cost
of Electricity
in Transmission,
Distribution,
and Administration
for U.S. Investor Ownedapproach.
Electric Utilities,A
2017 | 16
provides
insights
about howTrends
ancillary
services
are priced
under a co-optimization
approach.
AJanuary
simplified
approach to obtain the prices for energy and ancillary services is to ask what would be the overall increase
approach to obtain the prices for energy and ancillary services is to ask what would be the overall increase
in the operating cost of the system if an infinitesimal increase in demand or reserves requirements were
In Table 4 are presented solutions for different
demand levels. The checking of the results by hand
provides insights about how ancillary services
are priced under a co-optimization approach.
A simplified approach to obtain the prices for
energy and ancillary services is to ask what
would be the overall increase in the operating
cost of the system if an infinitesimal increase in
demand or reserves requirements were to occur.
For a demand level between 300 to 420 MW, it is
clear that generator 1 has to operate at full load
because it is the cheapest and it cannot provide
reserves. The next cheapest generator is generator
2, which can provide energy and reserves. Notice
that it is most economical to generate as much
of the remaining required energy as possible
with this generator, or in other terms, for it to
provide reserves as little as possible. However,
the other generator that can provide reserves,
generator 3, does not have enough capacity to
supply the full requirement of 250 MW of reserves
(it only can provide up to 190 MW of reserves).
Therefore, the minimum amount of reserves
that generator 2 has to provide is 60 MW. Thus,
generator 2 has to provide the remaining energy.
On the other hand, generator 3 has to provide as
much reserves as possible, because in this way
generator 2 can generate cheaper energy. Under
this demand scenario, maintaining the amount
of reserves required, an additional unit of energy
would be provided by generator 2 at a cost of $17.
Furthermore, maintaining the energy demand, if
an additional unit of reserves has to be provided,
only generator 2 can provide it. Given that in this
range generator 2 is not limited in its production,
the provision of an additional unit of reserves
doesn’t incur in any additional system cost.
For a demand range between 420 to 470 MW,
generator 2 has to produce as much energy as
possible. However, this is not enough to satisfy
the demand. Therefore, the next generator in
cost, generator 3 has to provide the remaining
energy. Under this scenario, and maintaining
the requirement of reserves, an additional
energy unit has to be provided by generator 3
at a cost of $20. Regarding an additional unit
of reserves, maintaining the energy production,
it is clear that only generator 2 can provide it.
In order to do that, it is required to reduce its
energy production by one unit, increase the
reserves production in one unit, and cover the
remaining demand with generator 3. So in this
exercise the overall operation cost is increased
by $20 and decreased by $17, which means that
the cost of an additional reserve unit is $3.
Finally, in the case when the demand is between
470 MW and 720 MW, the energy production of
generator 3 is not enough and generator 4 has to
contribute with energy. The reserves requirements
can be satisfied by generator 2 and generator 3.
To summarize, in a co-optimized market, the
price of the reserves is set by the opportunity
cost of foregone energy sales, as described
in the above paragraphs, or in some cases
by a non-zero offer cost of reserves.
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
17
3
| RESULTS OF IMPACT ANALYSIS
This chapter describes the results obtained for
the analysis of procured ancillary services in
ERCOT between 01/01/2007 to 04/13/2014. The
purpose of the analysis is to obtain the impact
significance of changes in nodal protocols
revisions as well as changes in installed generation.
The analysis started with a preliminary list of
nodal protocol revisions related with wind
power. The verification of the significance was
performed by a simplified statistical approach
using Regression Discontinuity Design rather
than a direct calculation method, which allowed
significant reductions in the amount of calculations
involved. The details of this analysis are presented
in the appendix of the report. From the above
analysis the following results were obtained:
• There are significant correlations
between the procured regulation-up and
regulation-down and the daily maximum
demand, daily minimum demand, and
installed wind power, and the associated
coefficients are presented in Tables Table
10 to Table 18. In several cases it was
found that non-coastal and thermal
generation installed power were positively
correlated with procured regulationup and regulation-down reserves.
However, for the zonal market period,
it was found that installed wind coastal
generation has a negative correlation.
• There are significant correlations with past
procured reserves, which is due to the
self-corrective action considered based
on past operation performed by ERCOT.
• Regarding network protocol
revisions purely associated to
wind power, it was found that the
following ones were significant:
o NPRR 352 (6/1/2011):
 Improvements in
prediction of the
maximum sustained
energy production
after curtailment.
o NPRR 361 (9/1/2011):
 Requires submission of 5
min resolution wind data
for real time purposes.
o NPRR 460 (12/1/2012):
 Increases the wind
powered generation
resource ramp rate
limitation from 10% per
minute of nameplate
rating to five minute
average of 20% per
minute of nameplate
rating with no individual
minute exceeding 25%.
• Finally from Table 19, Table 20, and Table
21, it was found that the most significant
impact was made during the introduction
of the nodal market, which are compared
against the other protocol revisions in
Figure 13 and Figure 14. In particular,
the change from 15-minutes to 5-minutes
dispatch intervals was a dramatic change
[20]. The reason for this is that as the
dispatch time is reduced, the action of the
real-time market can more quickly take
corrective actions for the perturbations in
the demand-generation balance. Therefore,
less reserves are required to cope with
the reduced uncertainty in net load over
the shorter dispatch interval used in the
nodal market. This result is in accordance
with the recommendations to increase
renewable penetration in the Western
Interconnection [21]. In addition, it was
observed that the least significant protocol
revision was NPRR352, which was related
with the wind forecasting improvements.
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
18
FIGURE 13
Impact of protocols revisions on Regulation-up reserve requirements.
FIGURE 14
Impact of protocols revisions on Regulation-down reserve requirements.
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
19
4
| CONCLUSIONS
This report has described the need, identification,
and valuation for ancillary services, with emphasis
on ERCOT. A statistical analysis of the procured
quantities of ancillary services was analyzed in the
context of changes to the market and an evaluation
was made of the most important effects on ancillary
services procurement. It was found that all of the
nodal protocol revisions considered produced
changes in terms of the procured regulation-up
and regulation-down reserves. Moreover, it was
found that installed power, regardless of its type
(e.g. coastal wind, non-coastal wind, thermal) is
positively correlated with procured reserves. An
exception for this was during the time before the
nodal market introduction, where coastal wind was
negatively correlated with reserves procurement.
The results obtained suggest that the changes in
requirements for procured reserves due to protocol
revisions performed during the transition between
zonal to nodal market have been more significant
than the changes in requirements due to increases
in installed wind power capacity for around 8,000
MW for the period within 2007 to 2013. This
observation motivates the exploration of better
ways to operate the grid allowing more renewable
integration without significant additional cost
due to its fluctuations. To illustrate this situation,
let’s consider Figure 15 where it is conceptually
represented a change in operational conditions
in the system (e.g. a new network protocol
revision). By doing this change, the reserve
requirements change from point 1 to point 1’.
Now after an increase in installed wind power, the
operational reserves requirements reductions by
the operation condition improvements are offset
by the additional operating reserves required due
to new renewable generation. Notice that without
this change the operational reserves associated
to point 2 would be obtained instead of 2’, which
may have significant economic consequences
in the operation of the power market.
FIGURE 15
Operational reserves
requirements
Illustration of operation reserves requirements change
due to operational conditions changes.
Operation conditions A
2
2'
1
1'
Operation conditions B
Installed wind power
Δ Installed wind power
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
20
APPENDIX: Impact analysis of wind related
nodal protocol revision requests on ancillary
services requirements
BACKGROUND
The requirements for ancillary services in markets
such as ERCOT are determined by policies that
consider past data, as well as forecasts for future
conditions to determine adequate quantities to
be procured. This section provides an analysis
of correlation and causation of ancillary services
requirements and other system variables and the
changes in these relationships due to changes
in the underlying “protocols” of the ERCOT
market. To facilitate the discussion, a list of
“nodal protocol revision requests” (NPPRS) is
first developed that is related to ancillary services
and then reviewed. Secondly, some variables
are identified that are correlated with ancillary
reserves requirements. Finally, a statistical
methodology is used to test whether NPRRs
have changed ancillary services requirements.
IDENTIFICATION OF ANCILLARY
SERVICES RELATED PROTOCOLS
This report covers a study period between
01/01/2007 to 04/13/2014 in ERCOT. During that
period, several major changes to ERCOT system
have occurred. After the introduction of the nodal
market, which itself is the most momentous change
to the market over this period, these changes have
been called Nodal Protocol Revision Requests
(NPRRs). Given that utility scale renewable
generation is a significant source of variability and
uncertainty in the system, it is expected that the
protocol revisions related to this type of generation
have an impact on ancillary services requirements.
It is for the above reason that this report is focused
on the analysis of NPRRs related with wind
generation. By searching for wind related protocols,
a preliminary list of protocols presented was
constructed, which is presented in Table 5.2 In this
table, the protocol revisions that are not sufficiently
related with ancillary services were discarded.
For example NPRR389, NPRR423, and NPRR424
are related with reactive compensation and were
discarded since reactive compensation is very
weakly related with operating reserves provisions,
which are mostly related with active power.
Amongst the remaining NPRRs that are related
to wind operation and forecast, there is a group
that was effective upon the Texas Nodal Market
implementation. Therefore, it is not possible to
temporally isolate their individual impact on
ancillary services requirements, nor is it possible
to isolate their impact from the impact of the
change to the nodal market itself, including the
change from 15 minute portfolio-based dispatch
to 5 minute unit-specific dispatch. Similarly,
several other groups of NPPRs were implemented
during short periods of time. Based on the
implementation dates of the selected protocols,
the whole study period was split in five periods as
presented in Table 6. This study identifies changes
in AS requirements that can be associated with
the groups of protocol revisions in Table 5.
2 The help of Walter Reid, Shams Siddiqi, and Dan Jones is gratefully
acknowledged in providing this preliminary list.
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
21
TABLE 5
Selected ERCOT nodal protocols related with ancillary services (*): Effective upon Texas Nodal Market Implementation.
N
Name
Title
Selected
Approval
date
Implementation
Effective date
Description
1
NPRR045 (*)
Wind Power Forecasting
✔
10/16/2007
11/1/2007
It clarifies the production of ERCOT
wind forecasts with a 80% of
probability of confidence.
2
NPRR050 (*)
Clarifications for HSL Values
for WGRs and WGR Values
to be Used in the RUC
Capacity Short Calculation
✔
07/17/2007
08/01/2007
It corrects an inconsistency in the
protocols and clarify the values
for wind generation resources
to be used in the reliability unit
commitment.
3
NPRR159 (*)
Resource Category Startup
Offer Generic Cap for Wind
Resources
✘
01/20/2009
02/01/2009
It stablishes that the O&M cost for
wind is zero.
4
NPRR177 (*)
Synchronization of Nodal
Protocols with PRR808,
Clean-up and Alignment
of RECs Trading Program
Language with PUCT Rules
✘
08/18/2009
09/01/2009
It has to be with renewable energy
credits under the ERCOT nodal
market operation.
5
NPRR210 (*)
Wind Forecasting Change to
P50, Synchronization with
PRR841
✔
06/15/2010
07/01/2007
IT changes the wind forecasting
methodology to use 50% of
probability of exceedance instead of
80% for Reliability Unit Commitment
considerations.
6
NPRR214 (*)
Wind-powered Generation
Resource (WGR) High
Sustained Limit (HSL)
Update Process
✔
05/18/2010
01/01/2010
It clarifies the timing for providing
the maximum sustained production
limit by wind generation resources.
7
NPRR239 (*)
Ramp Rate Limitation of
10% per minute of On-Line
Installed Capability for
Wind-powered Generation
Resources
✔
07/20/2010
08/01/2010
It limits the unit ramp rate of wind
generation resources to 10% of their
nameplate rating.
8
NPRR258 (*)
Synchronization with
PRR824 and PRR833 and
Additional Clarifications
✔
11/16/2010
12/01/2010
It aligns nodal protocols with primary
frequency requirements for wind
generation resources.
9
NPRR270 (*)
Defining the Variable Used
in the Wind Generation
Formula
✘
11/16/2010
12/01/2010
It is related with consideration of
wind generation at distributed level
for transmission and distribution
service providers.
10
NPRR281 (*)
Replace 7-Day Forecast
Requirement for QSEs
Representing WGRs
✔
11/16/2010
12/01/2010
It eliminates the requirement
that qualified scheduling entities
must provide a 7 day forecast.
It is considered that it produces
unreliable results.
11
NPRR285 (*)
Generation Resource Base
Point Deviation Charge
Corrections
✔
11/16/2010
12/01/2010
It provides a clear curtailment signal
for intermittent renewable resources.
12
NPRR352
Real-Time HSL Telemetry
for WGRs
✔
05/17/2011
6/1/2011
It is related with improvements
in the prediction of the maximum
sustained energy production
capability of a wind generator after
curtailment.
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
22
13
NPRR361
Real-Time Wind Power Production Data Transparency
✔
8/16/2011
9/1/2011
It requires the submitting of 5 min
resolution wind data for real time
purposes.
14
NPRR389
Modification of Voltage
Support Requirements to
Address Existing NonExempt WGRs
✘
10/18/2011
11/1/2011
It clarifies the reactive power
capability for wind power generation
resources.
15
NPRR423
Add Voltage Support
Requirement for IRRs
and Allow SCADA Control
of Static VAr Devices if
Approved by ERCOT
✘
2/21/2012
3/1/2012
It clarifies voltage and reactive
requirements for intermittent
renewable resources.
16
NPRR424
Reactive Capability Testing
Requirements for IRRs
✘
4/17/2012
5/1/2012
It defines the reactive testing
requirement for intermittent
renewable resources.
17
NPRR425
Creation of a WGR Group
for GREDP and Base Point
Deviation Evaluation and
Mixing Turbine Types Within
a WGR (formerly “Creation
of a WGR Group for GREDP
and Base Point Deviation
Evaluation”)
✘
11/13/2012
12/1/2012
It proposes the aggregation of
wind farms in groups for dispatch
purposes, to avoid control
limitations.
18
NPRR437
Allow Aggregation of
Multiple Generators Into A
Single Resource For Market
and Engineering Modeling
✘
02/21/2012
03/01/2012
It allows the aggregation of similar
qualified non-wind powered
generators for market and
engineering modeling purposes.
19
NPRR460
WGR Ramp Rate Limitation
✔
11/13/2012
12/1/2012
It increases the wind powered
generation resource ramp rate
limitation from 10% per minute
of nameplate rating to five minute
average of 20% per minute of
nameplate rating with no individual
minute exceeding 25%.
20
NPRR531
Clarification of IRR
Forecasting Process Posting
Requirement
✘
7/16/2013
8/1/2013
It clarifies that ERCOT has to publish
their procedures for forecasting, in
special for intermittent renewable
resources.
21
NPRR577
As-Built Clarification for
Portion of WGR Group
GREDP Evaluation
✘
2/11/2014
3/1/2014
It is a clarification of NPRR425.
22
NPRR611
Modifications to CDR Wind
Capacity Value
✘
10/14/2014
11/01/2014
It proposes modifications to the
Capacity, Demand, and Reserves
methodology for calculating capacity
value of wind during peak load
periods.
23
NPRR678
Posting of Wind Peak
Average Capacity
Percentage Data
✘
04/14/2015
05/01/2015
It is about the requirement of
information to calculate wind peak
average capacity percentages in
ERCOT webpage.
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
23
CORRELATION ANALYSIS
OF STUDY PERIODS
For each time period defined in Table 6, a
regression analysis was performed to obtain
a regression model for the required ancillary
services reserves. In this report, only regressions
for regulation-up and regulation-down were
performed since preliminary investigation
of the other AS suggests that they will not be
affected by the NPRRs. The outcome variables
are the ones presented in Table 7. The data
used for regression were the hourly procured
regulation-up and regulation-down in ERCOT.
The proposed regressors are the ones presented
in Table 8, and their data is presented in Figures
Figure 16 - Figure 21. The idea behind this
selection is to represent the factors that will likely
affect the procured quantities of regulation reserves,
which depend on the uncertainty in generation and
demand.
TABLE 6
Study periods for NPRRs.
Study
period
Description
Start
End
Duration
[days]
1
Pre-Nodal market
1/1/2007
12/01/2010
1430
2
NPRR045, NPRR050, NPRR210, NPRR214, NPRR239, NPRR258, and
other zonal to nodal market changes
12/01/2010
06/01/2011
182
3
NPRR352
06/01/2011
09/01/2011
92
4
NPRR361
09/01/2011
12/01/2012
457
5
NPRR460
12/01/2012
04/13/2014
498
TABLE 7
Variables to be regressed.
Symbol
Description
Expected value for regulation-up reserve in the study period at time.
Expected value for regulation-down reserve in the study period at time.
TABLE 8
Regressors selected for ancillary services correlation analysis.
Symbol
Description
Units
Source
ERCOT non-coastal wind generation accumulated installed power at time.
MW
[22]
ERCOT coastal wind generation accumulated installed power at time.
MW
ERCOT thermal generation accumulated installed power at time at time.
MW
Daily average total load in ERCOT at time.
MW
Daily total minimum load in ERCOT at time.
MW
Daily total maximum load in ERCOT at time.
MW
[23]
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
24
FIGURE 16
ERCOT non-coastal wind generation accumulated installed power.
FIGURE 17
ERCOT coastal wind generation accumulated installed power.
FIGURE 18
ERCOT thermal generation accumulated installed power.
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
25
FIGURE 19
Daily average total load in ERCOT.
FIGURE 20
Daily total minimum load in ERCOT.
FIGURE 21
Daily total maximum load in ERCOT.
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
26
where 𝐴𝐴𝑝𝑝,𝑗𝑗 and 𝐵𝐵𝑝𝑝,𝑗𝑗 are coefficients associated with the model. After performing regre
regulation-up and regulation-down for each study period independently, the information
Figure 21. Daily total maximum
load in
residuals
is ERCOT.
the one presented in figures Figure 22 to Figure 41. It can be appreciated that altho
Figure 21. Daily total maximum load in ERCOT.
is a correlation between the regressors and outcomes, the residuals obtained are far from whit
particular,
in the
autocorrelation
plotstend
of the
residuals,
it can
be appreciated that regression res
The regression model evaluated
for each
study
operators
to adjust
reserve
requirements
correlated
with thefunctional
ones of compared
the
previous
30 days.
observation
sion model evaluated
forhas
each
period
has the following
form:
period
thestudy
following
functional
form:
to the
most This
recently
procuredsuggests
reserves,the necessity of incl
sion model evaluated for each study period in
hasthe
theregression
following functional
form:
are correlated with
theconsideration
programmed has a physical meaning.
to cope which
with
autocorrelation.
This
(7)
reserves.from
Thus,one
it ismonth
considered
to include
30 to be similar. Moreove
operational reserves requirement
to the
next tend
day lags
in the regression
modelstoby
modifying
6
operators
tend to adjust reserve
requirements
compared
the
most recently procured reserv
(7)
6 𝑉𝑉𝑗𝑗,𝑡𝑡 𝐴𝐴𝑝𝑝,𝑗𝑗 ,
𝑈𝑈̂
equations
(7)
and
(8)
in
the
following
way:
𝑝𝑝,𝑡𝑡 = 𝐴𝐴𝑝𝑝,0 + ∑
(7) Thus, it is considered to include 30 day la
are
correlated with the programmed reserves.
𝑗𝑗=1 𝑉𝑉 𝐴𝐴 ,
𝑈𝑈̂
𝑝𝑝,𝑡𝑡 = 𝐴𝐴𝑝𝑝,0 + ∑
𝑗𝑗,𝑡𝑡 𝑝𝑝,𝑗𝑗
regression
models by modifying
equations (7) and (8) in the following way:
𝑗𝑗=1
(9)
(8)
6
30
(8)
𝑈𝑈̂
=
𝐴𝐴
+
∑
𝑉𝑉
𝐴𝐴
+
∑
𝐹𝐹𝑝𝑝,𝑗𝑗 𝑈𝑈𝑝𝑝,𝑡𝑡−𝑗𝑗 ,
𝑝𝑝,𝑡𝑡
𝑝𝑝,0
(8) 𝑗𝑗,𝑡𝑡 𝑝𝑝,𝑗𝑗
6
𝐷𝐷̂
𝑝𝑝,𝑡𝑡 = 𝐵𝐵𝑝𝑝,0 + ∑6 𝑉𝑉𝑗𝑗,𝑡𝑡 𝐵𝐵𝑝𝑝,𝑗𝑗 ,
𝑗𝑗=1 𝑉𝑉 𝐵𝐵 ,
𝐷𝐷̂
𝑝𝑝,𝑡𝑡 = 𝐵𝐵𝑝𝑝,0 + ∑
𝑗𝑗,𝑡𝑡 𝑝𝑝,𝑗𝑗
𝑗𝑗=1
𝑗𝑗=1
(10)
𝑗𝑗=1
where A p,j and
B p,j are coefficients
and 𝐵𝐵𝑝𝑝,𝑗𝑗 are coefficients
associated
with the associated
model. After performing regressions6 for
30
34
with the model.
After performing
regressions
for performing
are coefficients
the independently,
model. After
regressions
for
𝐷𝐷̂
∑ the
𝑉𝑉𝑗𝑗,𝑡𝑡 𝐵𝐵𝑝𝑝,𝑗𝑗 + ∑ 𝐺𝐺𝑝𝑝,𝑗𝑗 𝐷𝐷𝑝𝑝,𝑡𝑡−𝑗𝑗 ,
upand
and𝐵𝐵𝑝𝑝,𝑗𝑗
regulation-down
forassociated
each studywith
period
the information
about
𝑝𝑝,𝑡𝑡 = 𝐵𝐵𝑝𝑝,0 +
regulation-up
and regulation-down
for each study the information about𝑗𝑗=1
𝑗𝑗=1
up
for Figure
each
study
period41.
independently,
the
theand
one regulation-down
presented
in figures
22the
to Figure
It can
bethe
appreciated that although there
period
independently,
information
about
the between
one presented
in figuresand
Figure
22 to Figure
41. It canobtained
be appreciated
that although
there
tion
theresiduals
regressors
outcomes,
the
far from
noise.
F p,jwhite
and G
p,j areIncoefficients associated with
is the one
presented
inresiduals
figures Figure
22 are where
tion
between
the
regressors
and
outcomes,
the
residuals
obtained
are
far
from
white
noise.
In lag
where
𝐹𝐹that
and
𝐺𝐺𝑝𝑝,𝑗𝑗 are
coefficients
associated
with
variables.
Tablesthe
Table 9 - Table 18 p
n the autocorrelation
plots41.
of It
the
residuals,
it can be
appreciated
that
regression
residuals
lag variables.
Tables are
Table
9 - Table
18 present
𝑝𝑝,𝑗𝑗
to Figure
can
be appreciated
although
n
the
autocorrelation
plots
of
the
residuals,
it
can
be
appreciated
that
regression
residuals
are
with the ones of the
previous
30 days.between
This observation
suggests
necessity
of including
lags
regression
coefficients
for the
regression
coefficients
the
significant
regressors.
Thesignificant
impact of regressors.
including lags in the regress
there
is a correlation
the regressors
and thefor
with
thetoones
of with
the
previous
days. This
observation
suggests
the
necessity
of
including
lags
The
impact
of
including
lags
in
the
regression
outcomes,
the30residuals
obtained
are
far
from
ession
cope
autocorrelation.
This
consideration
has
a
physical
meaning.
First,
the
appreciated in the autocorrelations of the residuals, which are much smaller. Notice that in m
to cope
with
Thisto
consideration
physical
meaning.
First,
cancase
be
appreciated
inthe
the autocorrelations
of the
whiteautocorrelation.
noise.
particular,
in
the
lession
reserves
requirement
from In
one
month
the autocorrelation
next
tendhas
to abe
similar.
Moreover,
system
cases,
in comparison
with
the
without
lags,
the
residuals autocorrelations
were decreased
residuals,
which
are
much
smaller.
Notice
that
lend
reserves
requirement
from
one
month
to
the
next
tend
to
be
similar.
Moreover,
system
plots
of
the
residuals,
it
can
be
appreciated
that
to adjust reserve requirements compared
to the
most
recently
reserves, which
in some
cases
such
as theprocured
one for regulation-Up
for study period 2, the autocorrelations are still
in most
of the
cases,
in comparison with the case
regression
residuals
are
correlated
with
the
end
adjust
requirements
compared
to
the
most
recently
procured
reserves,
which
ted to
with
the reserve
programmed
reserves.
Thus,
it
is
considered
to
include
30
day
lags
in
the
which indicates that the regression models are not complete yet. This issue will be addressed
without
lags,lags
the residuals
ones of the previous
30Thus,
days. itThis
observation to include
ted with
programmed
reserves.
considered
30 day
in the autocorrelations were
models
bythe
modifying
equations
(7) and
(8)
inreport.
theisfollowing
way:
decreased. However, in some cases such as the
suggests
the necessity
including
lags in the
models by modifying
equations
(7) andof(8)
in the following
way:
30 This
one for regulation-Up
regression to cope 6with autocorrelation.
(9)for study period 2, the
𝑈𝑈̂
∑a6 physical
𝑉𝑉𝑗𝑗,𝑡𝑡 𝐴𝐴𝑝𝑝,𝑗𝑗meaning.
+ ∑30 First,
𝐹𝐹𝑝𝑝,𝑗𝑗 𝑈𝑈the
𝑝𝑝,𝑡𝑡 = 𝐴𝐴𝑝𝑝,0 +has
𝑝𝑝,𝑡𝑡−𝑗𝑗 ,
autocorrelations are(9)
still significant, which indicates
consideration
𝑗𝑗=1 𝑉𝑉 𝐴𝐴
𝑗𝑗=1 𝐹𝐹 𝑈𝑈
𝑈𝑈̂
,
𝑝𝑝,𝑡𝑡 = 𝐴𝐴𝑝𝑝,0 + ∑
𝑗𝑗,𝑡𝑡 𝑝𝑝,𝑗𝑗 + ∑ Table
𝑝𝑝,𝑗𝑗 9.𝑝𝑝,𝑡𝑡−𝑗𝑗
Significant
regression
coeffients
for
Regulation-Up
study period
1 (with 5% of significance).
that the regression models
are notincomplete
yet.
operational reserves
requirement from
𝑗𝑗=1
𝑗𝑗=1one month
This issue will be addressed in a future report.
to the next tend to be similar. Moreover, system
Parameter
Coefficient
p-value
34
𝐴𝐴1,2
-0.002280
0.009
34
𝐴𝐴1,3
0.000717
0.000
𝐴𝐴1,5
-0.000201
0.007
𝐹𝐹1,1
0.8883
0.000
TABLE
9
𝐹𝐹1,2
0.0713
0.008
Significant regression coeffients for Regulation-Up in study period 1 (with 5% of significance).
Parameter
A 1,2
Table 10. Significant regression coefficients for Regulation-Down in study period 1 (with 5% of significanc
Coefficient
p-value
Parameter
-0.002280
𝐵𝐵1,1
𝐵𝐵1,3 0.000717
𝐵𝐵1,4 -0.000201
𝐺𝐺1,1
𝐺𝐺1,2 0.8883
𝐺𝐺1,3 0.0713
𝐺𝐺1,4
𝐺𝐺1,5
A 1,3
A 1,5
F 1,1
F 1,2
Coefficient
0.001942
0.001199
0.000335
0.2905
0.2132
0.1565
0.1227
0.1059
0.009
0.000
0.007
0.000
0.008
p-value
0.002
0.000
0.007
0.000
0.000
0.000
0.000
0.000
Table 11. Significant regression coeffients for Regulation-Up in study period 2 (with 5% of significance).
Parameter
Coefficient
p-value
0.3483
0.000
|
The Full Cost of Electricity (FCe-) Trends in Transmission,𝐴𝐴Distribution, and Administration-2691
Costs for U.S. Investor Owned Electric Utilities, January0.000
2017 27
2,0
𝐴𝐴2,1
TABLE 10
Significant regression coefficients for Regulation-Down in study period 1 (with 5% of significance).
Parameter
Coefficient
p-value
B 1,1
0.001942
0.002
B 1,3
0.001199
0.000
B 1,4
0.000335
0.007
G1,1
0.2905
0.000
G1,2
0.2132
0.000
G1,3
0.1565
0.000
G1,4
0.1227
0.000
G1,5
0.1059
0.000
TABLE 11
Significant regression coeffients for Regulation-Up in study period 2 (with 5% of significance).
Parameter
Coefficient
p-value
A 2,0
-2691
0.000
A 2,1
0.3483
0.000
A 2,4
0.002031
0.009
A 2,5
-0.003305
0.000
F2,1
0.4511
0.000
F2,30
0.1791
0.000
TABLE 12
Significant regression coefficients for Regulation-Down in study period 2 (with 5% of significance).
Parameter
Coefficient
p-value
B 2,0
-3129
0.000
B 2,3
0.05739
0.000
B 2,4
0.00472
0.000
B 2,5
-0.004022
0.000
B 2,6
-0.001094
0.047
G 2,1
0.1236
0.000
G 2,30
0.1151
0.000
TABLE 13
Significant regression coeffients for Regulation-Up in study period 3 (with 5% of significance).
Parameter
Coefficient
p-value
A3,6
0.000222
0.034
F3,1
0.9774
0.000
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
28
TABLE 14
Significant regression coefficients for Regulation-Down in study period 3 (with 5% of significance).
Parameter
Coefficient
p-value
131.5
0.003
0.9117
0.000
-0.1672
0.004
TABLE 15
Significant regression coeffients for Regulation-Up in study period 4 (with 5.3% of significance).
Parameter
Coefficient
p-value
0.002209
0.000
-0.000193
0.053
0.000117
0.010
0.96276
0.000
TABLE 16
Significant regression coefficients for Regulation-Down in study period 4 (with 5% of significance).
Parameter
Coefficient
p-value
0.999663
0.000
TABLE 17
Significant regression coeffients for Regulation-Up in study period 5 (with 5.1% of significance).
Parameter
Coefficient
p-value
0.000207
0.002
0.000198
0.032
-0.000238
0.051
0.97127
0.000
TABLE 18
Significant regression coefficients for Regulation-Down in study period 5 (with 5.4% of significance).
Parameter
Coefficient
p-value
8.75
0.014
0.000173
0.037
-0.000207
0.054
0.97788
0.000
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
29
FIGURE 22
Residual Plots for Regulation-Up in study period 1(without lags).
R esidual Plots for R egulation- Up in study period 1
N ormal Probability Plot
Versus Fits
99.99
100
99
50
Residual
Percent
90
50
10
0
- 50
- 100
1
0.01
- 100
0
100
800
825
Residual
H istogram
900
Versus Order
150
Residual
Frequency
875
100
200
100
50
0
850
Fitted Value
50
0
- 50
- 100
- 120
- 90
- 60
- 30
0
30
60
90
1
Residual
0
0
0
0
0
0
0
0
0
0
0
0
0
0
10 20 30 4 0 50 60 70 80 90 100 110 120 130 14 0
Observation Order
FIGURE 23
Autocorrelation Function for Regulation-Up in study period 1 residuals (without lags).
Autocorrelation Function for R egulation- Up in study period 1 residuals
(with 5% significance limits for the autocorrelations)
1.0
0.8
Autocorrelation
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
2
4
6
8
10
12
14
16
18
20
22
24
26
28
30
Days lag
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
30
FIGURE 24
Residual Plots for Regulation-Down in study period 1 (without lags).
R esidual Plots for R egulation- Down in study period 1
N ormal Probability Plot
Versus Fits
99.99
800
99
Residual
Percent
90
50
10
600
400
200
1
0
0.01
0
250
500
750
1000
750
900
H istogram
Versus Order
950
800
300
Residual
Frequency
850
Fitted Value
400
200
100
0
800
Residual
600
400
200
0
0
150
300
450
600
750
1
Residual
0 0 0 0 0 0 0 0 0 0 0 0 0 0
10 2 0 30 4 0 50 60 70 80 90 100 110 120 13 0 140
Observation Order
FIGURE 25
Autocorrelation Function for Regulation-Down in study period 1 residuals (without lags).
Autocorrelation Function for R egulation- Down in study period 1 residuals
(with 5% significance limits for the autocorrelations)
1.0
0.8
Autocorrelation
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
2
4
6
8
10
12
14
16
18
20
22
24
26
28
30
Lag
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
31
FIGURE 26
Residual Plots for Regulation-Up in study period 2 (without lags).
R esidual Plots for R egulation- Up in study period 2
N ormal Probability Plot
Versus Fits
99.9
50
90
Residual
Percent
99
50
10
1
0.1
0
- 50
- 100
- 100
- 50
0
50
100
400
500
550
Fitted Value
H istogram
Versus Order
40
600
50
30
Residual
Frequency
450
Residual
20
0
- 50
10
- 100
0
- 90
- 60
- 30
0
30
60
1
20
40
Residual
60
80
100
120
140
160
180
Observation Order
FIGURE 27
Autocorrelation Function for Regulation-Up in study period 2 residuals (without lags).
Autocorrelation Function for R egulation- Up in study period 2 residuals
(with 5% significance limits for the autocorrelations)
1.0
0.8
Autocorrelation
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
2
4
6
8
10
12
14
16
18
20
22
24
26
28
30
Lag
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
32
FIGURE 28
Residual Plots for Regulation-Down in study period 2 (without lags).
R esidual Plots for R egulation- Down in study period 2
N ormal Probability Plot
Versus Fits
99.9
20
90
Residual
Percent
99
50
10
1
0.1
0
- 20
- 40
- 40
- 20
0
20
40
420
440
460
480
Residual
Fitted Value
H istogram
Versus Order
500
20
30
Residual
Frequency
40
20
10
0
0
- 20
- 40
- 40
- 30
- 20
- 10
0
10
20
30
1
20
40
Residual
60
80
100
120
140
160
180
Observation Order
FIGURE 29
Autocorrelation Function for Regulation-Down in study period 2 residuals (without lags).
Autocorrelation Function for R egulation- Down in study period 2 residuals
(with 5% significance limits for the autocorrelations)
1.0
0.8
Autocorrelation
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
2
4
6
8
10
12
14
16
18
20
22
24
26
28
30
Days lag
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
33
FIGURE 30
Residual Plots for Regulation-Up in study period 3 (without lags).
R esidual Plots for R egulation- Up in study period 3
N ormal Probability Plot
Versus Fits
99.9
40
99
Residual
Percent
90
50
10
1
20
0
- 20
0.1
- 40
- 20
0
20
40
580
590
Residual
600
610
620
Fitted Value
H istogram
Versus Order
40
15
Residual
Frequency
20
10
20
0
5
- 20
0
- 20
- 10
0
10
20
30
1
10
20
Residual
30
40
50
60
70
80
90
Observation Order
FIGURE 31
Autocorrelation Function for Regulation-Up in study period 3 residuals (without lags).
Autocorrelation Function for R egulation- Up in study period 3 residuals
(with 5% significance limits for the autocorrelations)
1.0
0.8
Autocorrelation
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
2
4
6
8
10
12
14
16
18
20
22
24
26
28
30
Days Lag
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
34
FIGURE 32
Residual Plots for Regulation-Down in study period 3 (without lags).
R esidual Plots for R egulation- Down in study period 3
N ormal Probability Plot
Versus Fits
99.9
40
90
20
Residual
Percent
99
50
10
1
0
- 20
0.1
- 40
- 20
0
20
40
580
590
Residual
600
610
620
Fitted Value
H istogram
Versus Order
40
15
20
Residual
Frequency
20
10
0
5
- 20
0
- 20
- 10
0
10
20
30
1
10
20
Residual
30
40
50
60
70
80
90
Observation Order
FIGURE 33
Autocorrelation Function for Regulation-Down in study period 3 residuals (without lags).
Autocorrelation Function for R egulation- Down in study period 3 residuals
(with 5% significance limits for the autocorrelations)
1.0
0.8
Autocorrelation
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
2
4
6
8
10
12
14
16
18
20
22
24
26
28
30
Days Lag
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
35
FIGURE 34
Residual Plots for Regulation-Up in study period 4 (without lags).
R esidual Plots for R egulation- Up in study period 4
N ormal Probability Plot
Versus Fits
99.9
50
90
Residual
Percent
99
50
10
25
0
- 25
1
0.1
- 50
- 50
- 25
0
25
50
500
520
Residual
540
560
580
Fitted Value
H istogram
Versus Order
50
Residual
Frequency
60
45
30
15
0
25
0
- 25
- 50
- 45
- 30
- 15
0
15
30
45
1
50
100
Residual
150
200
250
300
350
400
450
Observation Order
FIGURE 35
Autocorrelation Function for Regulation-Up in study period 4 residuals (without lags).
Autocorrelation Function for R egulation- Down in study period 4 residuals
(with 5% significance limits for the autocorrelations)
1.0
0.8
Autocorrelation
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
2
4
6
8
10
12
14
16
18
20
22
24
26
28
30
Days lag
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
36
Days Lag
Figure 35. Autocorrelation Function for Regulation-Up in study period 4 residuals (without lags
FIGURE 36
Residual Plots for Regulation-Down in study period 4 (without lags).
R esidual Plots for R egulation- Down in study period 4
N ormal Probability Plot
Versus Fits
99.9
60
90
30
Residual
Percent
99
50
10
0
- 30
1
0.1
- 50
0
50
- 60
100
400
425
450
475
Residual
Fitted Value
H istogram
Versus Order
48
60
36
30
Residual
Frequency
- 100
24
500
0
- 30
12
0
- 40
- 20
0
20
40
- 60
60
1
50
100
Residual
150
200
250
300
350
400
450
Observation Order
Figure 36. Residual Plots for Regulation-Down in study period 4 (without lags).
FIGURE 37
Autocorrelation Function for Regulation-Down in study period 4 residuals (without lags).
Autocorrelation Function for Regulation-Down in study period 4 residuals
(with 5% significance limits for the autocorrelations)
1.0
0.8
Autocorrelation
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
2
4
6
8
10
12
14
16
18
20
22
24
26
28
30
Days lag
Figure 37. Autocorrelation Function for Regulation-Down in study period 4 residuals (without lag
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
37
FIGURE 38
Residual Plots for Regulation-Up in study period 5 (without lags).
R esidual Plots for R egulation- Up in study period 5
N ormal Probability Plot
Versus Fits
99.9
50
90
Residual
Percent
99
50
10
0
- 50
1
0.1
- 100
- 50
0
- 100
50
460
480
H istogram
Versus Order
500
50
60
Residual
Frequency
440
Fitted Value
80
40
20
0
420
Residual
- 80
- 60
- 40
- 20
0
20
40
0
- 50
- 100
60
1
50
100 150 200 250 300 350 400 450
Residual
Observation Order
FIGURE 39
Autocorrelation Function for Regulation-Up in study period 5 residuals (without lags).
Autocorrelation Function for R egulation- Up in study period 5 residuals
(with 5% significance limits for the autocorrelations)
1.0
0.8
Autocorrelation
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
2
4
6
8
10
12
14
16
18
20
22
24
26
28
30
Days Lag
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
38
FIGURE 40
Residual Plots for Regulation-Down in study period 5 (without lags).
R esidual Plots for R egulation- Down in study period 5
N ormal Probability Plot
Versus Fits
99.9
50
90
Residual
Percent
99
50
10
25
0
- 25
1
0.1
- 50
- 25
0
25
- 50
50
400
420
440
H istogram
Versus Order
50
45
Residual
Frequency
380
Fitted Value
60
30
15
0
360
Residual
25
0
- 25
- 45
- 30
- 15
0
15
30
45
- 50
60
1
50
100
Residual
150 200 250 300
350 400 450
Observation Order
FIGURE 41
Autocorrelation Function for Regulation-Down in study period 5 residuals (without lags).
Autocorrelation Function for R egulation- Down in study period 5 residuals
(with 5% significance limits for the autocorrelations)
1.0
0.8
Autocorrelation
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
2
4
6
8
10
12
14
16
18
20
22
24
26
28
30
Days Lag
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
39
FIGURE 42
Residual Plotts for Regulation-Up in study period 1 (with lags).
R esidual Plots for R egulation- Up in study period 1 (with lags)
N ormal Probability Plot
Versus Fits
99.99
100
99
Residual
Percent
90
50
10
0
- 100
1
0.01
- 100
0
100
700
900
Fitted Value
H istogram
Versus Order
1200
100
900
Residual
Frequency
800
Residual
600
300
0
- 120
- 80
- 40
0
40
80
0
- 100
120
1
Residual
0 0 0 0 0 0 0 0 0 0 0 0 0
10 2 0 3 0 40 50 60 70 80 90 10 0 110 12 0 130
Observation Order
FIGURE 43
Autocorrelation Function for Regulation-Up in study period 1 residuals.
Autocorrelation Function for R egulation- Up in study period 1 residuals.
(with 5% significance limits for the autocorrelations)
1.0
0.8
Autocorrelation
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
2
4
6
8
10
12
14
16
18
20
22
24
26
28
30
Days Lag
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
40
FIGURE 44
Residual Plots for Regulation-Down in study period 1 (with lags).
R esidual Plots for R egulation- Down in study period 1 (with lags)
N ormal Probability Plot
Versus Fits
99.99
800
99
Residual
Percent
90
50
10
400
0
1
0.01
0
400
800
700
800
900
Residual
H istogram
Versus Order
900
Residual
Frequency
1100
800
1200
600
300
0
1000
Fitted Value
- 150
0
150
300
450
600
400
0
750
1
Residual
0 0 0 0 0 0 0 0 0 0 0 0 0
10 20 30 40 5 0 6 0 70 80 90 10 0 110 120 130
Observation Order
FIGURE 45
Autocorrelation Function for Regulation-Down in study period 1 residuals.
Autocorrelation Function for R egulation- Down in study period 1 residuals.
(with 5% significance limits for the autocorrelations)
1.0
0.8
Autocorrelation
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
2
4
6
8
10
12
14
16
18
20
22
24
26
28
30
Day Lags
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
41
FIGURE 46
Residual Plots for Regulation-Up in study period 2 (with lags).
R esidual Plots for R egulation- Up in study period 2 (with lags).
N ormal Probability Plot
Versus Fits
99
50
90
0
Residual
Percent
99.9
50
10
1
- 50
- 100
- 150
0.1
- 150
- 100
- 50
0
50
400
450
500
550
Residual
Fitted Value
H istogram
Versus Order
600
100
Residual
Frequency
50
75
50
25
0
- 50
- 100
- 150
0
- 160
- 120
- 80
- 40
0
40
1
20
40
Residual
60
80
100
120
140
160
180
Observation Order
FIGURE 47
Autocorrelation Function for Regulation-Up in study period 2 residuals.
Autocorrelation Function for R egulation- Up in study period 2 residuals.
(with 5% significance limits for the autocorrelations)
1.0
0.8
Autocorrelation
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
2
4
6
8
10
12
14
16
18
20
22
24
26
28
30
Days Lag
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
42
FIGURE 48
Residual Plots for Regulation-Down in study period 2 (with lags).
R esidual Plots for R egulation- Down in study period 2 (with lags)
N ormal Probability Plot
Versus Fits
99.9
50
99
Residual
Percent
90
50
10
25
0
- 25
1
0.1
- 50
- 25
0
25
- 50
50
400
425
Residual
450
475
500
Fitted Value
H istogram
Versus Order
50
25
Residual
Frequency
40
30
20
10
0
- 30
- 15
0
15
30
0
- 25
- 50
45
1
20
40
Residual
60
80
100
120
140
160
180
Observation Order
FIGURE 49
Autocorrelation Function for Regulation-Down in study period 2 residuals.
Autocorrelation Function for R egulation- Down in study period 2 residuals.
(with 5% significance limits for the autocorrelations)
1.0
0.8
Autocorrelation
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
2
4
6
8
10
12
14
16
18
20
22
24
26
28
30
Days lag
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
43
FIGURE 50
Residual Plots for Regulation-Up in study period 3 (with lags).
R esidual Plots for R egulation- Up in study period 3 (with lags).
N ormal Probability Plot
Versus Fits
99.9
30
99
Residual
Percent
90
50
10
20
10
0
1
0.1
- 10
0
10
20
- 10
30
580
590
600
Residual
610
620
Fitted Value
H istogram
Versus Order
30
20
60
Residual
Frequency
80
40
10
0
20
0
-5
0
5
10
15
20
- 10
25
1
10
20
Residual
30
40
50
60
70
80
90
Observation Order
FIGURE 51
Autocorrelation Function for Regulation-Up in study period 3 residuals.
Autocorrelation Function for R egulation- Up in study period 3 residuals.
(with 5% significance limits for the autocorrelations)
1.0
0.8
Autocorrelation
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
2
4
6
8
10
12
14
16
18
20
22
24
26
28
30
Days lag
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
44
FIGURE 52
Residual Plots for Regulation-Down in study period 3 (with lags).
R esidual Plots for R egulation- Down in study period 3 (with lags)
N ormal Probability Plot
Versus Fits
99.9
90
Residual
Percent
99
50
10
1
0.1
0
- 25
- 50
- 50
- 25
0
480
495
510
525
Residual
Fitted Value
H istogram
Versus Order
540
60
Residual
Frequency
80
40
0
- 25
20
0
- 50
- 50
- 40
- 30
- 20
- 10
0
10
20
1
10
20
Residual
30
40
50
60
70
80
90
Observation Order
FIGURE 53
Autocorrelation Function for Regulation-Down study period 3 residuals.
Autocorrelation Function for R egulation- Down study period 3 residuals.
(with 5% significance limits for the autocorrelations)
1.0
0.8
Autocorrelation
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
2
4
6
8
10
12
14
16
18
20
22
24
26
28
30
Days Lag
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
45
FIGURE 54
Residual Plots for Regulation-Up in study period 4 (with lags).
R esidual Plots for R egulation- Up in study period 4 (with lags).
N ormal Probability Plot
Versus Fits
99.9
50
25
90
Residual
Percent
99
50
10
1
0.1
0
- 25
- 50
- 50
- 25
0
25
50
500
525
Residual
H istogram
600
Versus Order
25
300
Residual
Frequency
575
50
400
200
100
0
550
Fitted Value
0
- 25
- 50
- 45
- 30
- 15
0
15
30
1
50
100
Residual
150
200
250
300
350
400
450
Observation Order
FIGURE 55
Autocorrelation Function for Regulation-Up study period residuals.
Autocorrelation Function for R egulation- Up study period 4 residuals.
(with 5% significance limits for the autocorrelations)
1.0
0.8
Autocorrelation
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
2
4
6
8
10
12
14
16
18
20
22
24
26
28
30
Lag
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
46
FIGURE 56
Residual Plots for Regulation-Down in study period 4 (with lags).
R esidual Plots for R egulation- Down in study period 4 (with lags)
N ormal Probability Plot
Versus Fits
99.9
50
99
Residual
Percent
90
50
10
1
0.1
25
0
- 25
- 50
- 50
- 25
0
25
50
420
450
Residual
H istogram
510
Versus Order
50
400
300
Residual
Frequency
480
Fitted Value
200
100
25
0
- 25
- 50
0
- 45
- 30
- 15
0
15
30
45
1
50
100
Residual
150
200
250
300
350
400
450
Observation Order
FIGURE 57
Autocorrelation Function for Regulation-Down study period 4 residuals.
Autocorrelation Function for R egulation- Down study period 4 residuals.
(with 5% significance limits for the autocorrelations)
1.0
0.8
Autocorrelation
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
2
4
6
8
10
12
14
16
18
20
22
24
26
28
30
Lag
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
47
FIGURE 58
Residual Plots for Regulation-Up in study period 5 (with lags).
R esidual Plots for R egulation- Up in study period 5 (with lags)
N ormal Probability Plot
Versus Fits
99
50
90
25
Residual
Percent
99.9
50
10
0
- 25
1
0.1
- 50
- 50
- 25
0
25
50
400
440
480
Residual
Fitted Value
H istogram
Versus Order
520
50
300
200
0
- 25
100
0
25
Residual
Frequency
400
- 50
- 45
- 30
- 15
0
15
30
45
1
50
100
Residual
150 200 250 300 350 400 450
Observation Order
FIGURE 59
Autocorrelation Function for Regulation-Up period 5 residuals.
Autocorrelation Function for R egulation- Up study period 5 residuals
(with 5% significance limits for the autocorrelations)
1.0
0.8
Autocorrelation
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
2
4
6
8
10
12
14
16
18
20
22
24
26
28
30
Lag
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
48
FIGURE 60
Residual Plots for Regulation-Down in study period 5 (with lags).
R esidual Plots for R egulation- Down in study period 5 (with lags)
N ormal Probability Plot
Versus Fits
99.9
50
99
Residual
Percent
90
50
10
25
0
- 25
1
0.1
- 50
- 25
0
25
- 50
50
380
400
Residual
360
25
240
120
- 45
- 30
- 15
440
460
Versus Order
50
Residual
Frequency
H istogram
480
0
420
Fitted Value
0
15
30
0
- 25
- 50
45
1
50
100
Residual
150 200 250 300
350 400 450
Observation Order
FIGURE 61
Autocorrelation Function for Regulation-Down study period 5 residuals.
Autocorrelation Function for R egulation- Down study period 5 residuals.
(with 5% significance limits for the autocorrelations)
1.0
0.8
Autocorrelation
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
2
4
6
8
10
12
14
16
18
20
22
24
26
28
30
Lag
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
49
CAUSATION ANALYSIS OF NETWORK
PROTOCOLS REVISIONS WITH
ANCILLARY SERVICES REQUIREMENTS
The previous section presented the development of
regression models for required regulation-up and
regulation-down ancillary services, which were
tailored for each study period. Based on them,
it is possible to support the claim that there is a
correlation between ancillary services requirements,
and demand, and installed wind power. Assuming
that no significant protocol revisions occurred
within the study periods identified, it can be
considered that no change in operating reserves
is due to NPRRs inside those periods, and all
change is due to the effect of the NPPRs at the
beginning of the respective study period.
In order to evaluate the impact of NPRRs, we
consider AS procurement in the proximity of
the effective date of a NPRR. It can be assumed
that system conditions (i.e. installed power and
demand) are practically similar just before and
just after that date. Moreover, at each period
transition, the same type and time resolution
for reserves are presented. Thus, if there is a
significant change in reserve requirements, it has
to be due to the NPPR that defines the beginning
of the study period. This observation causal
inferences to be made about short-term effects
and it is known in the literature as Regression
Discontinuity Designs (RDD) [24], [25]
To illustrate how RDD works, let’s consider the
situation represented in Figure 62. The Figure
presents the application of a policy change. Just
before the change, it can be considered that the
conditions at the point A are very similar to the
ones at point B, which is after the policy. At left
and right hand sides of the cut-off, regressions
were obtained. If the regression obtained for the
right hand side is extended somewhat to the right
hand side of a small window, a counterfactual
estimation can be obtained. In this case, if there is
a significant difference between the counterfactual
case and the real data, it means that the impact of
the policy was significant at the cut-off. Notice
that although the causation analysis is restricted
to only the data points in the proximity of the
cut-off, the distant data points impact on the
regression function definition. However, in
practical conditions the window size cannot be so
small due to lack of data. Therefore, the impact of
other variable (e.g. installed power and demand
for this case) might start to be significant inducing
a bias in the treatment effect of the policy. [25]
FIGURE 62
Outcome variable
Illustration of RDD concepts.
l case
actua
terf
Coun
Window width
Running variable (Time)
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
50
For the particular problem of this report,
of the period p+1, and τp,p+1 and β p,p+1 are the
and considering equations (9) and (10),
coefficients associated to the dummy variable
the RDD experiment can be expressed
for regulation up and regulation down.
mathematically in the following way:
TheRDD
coefficients
associated to each periodticular problem of this report, and considering equations (9) and (10), the
experiment
(11)
transition
dummy
variable and their significance
essed problem
mathematically
in the following
way: equations (9) and (10), the RDD experiment
icular
of this report,
and considering
are presented in Table 19. For window size it
essed mathematically in the
followinǵ way:
was considered 15 (11)
days before and 15 days after
𝑈𝑈̂
𝑝𝑝+1,𝑡𝑡 = 𝑈𝑈𝑝𝑝+1,𝑡𝑡 + 𝑋𝑋𝑝𝑝,𝑝𝑝+1 𝜏𝜏𝑝𝑝,𝑝𝑝+1 ,
the
cut-off.
The
significance
and coefficients for
́
(11)
𝑈𝑈̂
𝑝𝑝+1,𝑡𝑡 = 𝑈𝑈𝑝𝑝+1,𝑡𝑡 + 𝑋𝑋𝑝𝑝,𝑝𝑝+1 𝜏𝜏𝑝𝑝,𝑝𝑝+1 ,
other
window
sizes
are
also
presented.
From the
(12)
́
(12)
𝐷𝐷̂
=
𝐷𝐷
+
𝑋𝑋
𝛽𝛽
,
results
in
the
Table,
it
can
be
concluded
that all
𝑝𝑝,𝑡𝑡
𝑝𝑝+1,𝑡𝑡
𝑝𝑝,𝑝𝑝+1 𝑝𝑝,𝑝𝑝+1
the NPRRs changes
considered in each period
́
(12)
𝐷𝐷̂
𝑝𝑝,𝑡𝑡 = 𝐷𝐷𝑝𝑝+1,𝑡𝑡 + 𝑋𝑋𝑝𝑝,𝑝𝑝+1 𝛽𝛽𝑝𝑝,𝑝𝑝+1 ,
transmission produced significant changes
́
and procured just after
1,𝑡𝑡 and 𝐷𝐷𝑝𝑝+1,𝑡𝑡 represent the counterfactual estimations of the required
in theregulation
amount ofup
reserves
́
ves by using thewhere
regression
for the
period
𝑝𝑝 , 𝑋𝑋𝑝𝑝,𝑝𝑝+1
a dummy Unfortunately,
variable
Uthe
p´+1,tmodel
and Dobtained
p´+1,t represent
thetime
counterfactual
the transition.
the conclusions
counterfactual
estimations
of
the required
regulation
up and
1,𝑡𝑡 and 𝐷𝐷𝑝𝑝+1,𝑡𝑡 represent
estimations
of
the
required
regulation
up
and
obtained
directly
from
RDD
can
only establish
al
one
after
the
beginning
the
period
𝑝𝑝
+
1,
and
𝜏𝜏
and
𝛽𝛽
are
the
coefficients
ves by using the regression model obtained for the time𝑝𝑝,𝑝𝑝+1
period 𝑝𝑝 , 𝑝𝑝,𝑝𝑝+1
𝑋𝑋𝑝𝑝,𝑝𝑝+1 a dummy variable
down
reserves
by
using
the
regression
model
results
for
the
short
term.
However,
if we consider
to one
the dummy
variable
for regulation
up and
regulation
down.and 𝛽𝛽𝑝𝑝,𝑝𝑝+1 are the coefficients
al
after the
beginning
of the period
𝑝𝑝 +
1, and 𝜏𝜏𝑝𝑝,𝑝𝑝+1
obtained for the time period p, Xp,p+1 a dummy
the regression coefficients associated to lags in
o theassociated
dummy variable
regulation
regulation
down.
ents
to
eachfor
period-transition
dummy
and their significance
are presented
variable
that
is equal up
oneand
after
thevariable
beginning
the regression
models presented in Tables Table
9.
For
window
size
it
was
considered
15
days
before
and
15
days
after
the
cut-off.
The
ents associated to each period-transition dummy variable and their significance are presented
e.and
foritother
sizes15are
alsobefore
presented.
theafter
results
the Table,
For coefficients
window size
was window
considered
days
and From
15 days
thein cut-off.
Theit
TABLE 19
cluded
that all thefor
NPRRs
changes
considered
in each
period transmission
produced
and coefficients
other
window
sizes are also
presented.
From the results
in thesignificant
Table, it
Treatments
effect
for
period
transitions
for
required
Regulation-Up
and
Regulation-Down reserves.
the
amount
of
reserves
procured
just
after
the
transition.
Unfortunately,
the
conclusions
luded that all the NPRRs changes considered in each period transmission produced significant
irectly
from RDD
can onlyprocured
establishjust
results
thetransition.
short term.
However, if we
the
the amount
of reserves
afterforthe
Unfortunately,
the consider
conclusions
coefficients
to lags
in the regression
presented
in TablesifTable
10-Tablethe
18,
rectly from associated
RDD can only
establish
results for models
theRegulation
short
term.
However,
we consider
Regulation - Down
- Up
erve
that
these
coefficients
are
very
close
to
one.
This
means
that
if
there
is
a
change
in
the
coefficients associated to lags in the regression modelsτpresented
in Tables Table 10-Table 18,
β p,p+1
p,p+1
Periodbeginning
transition
eserve
value
at the
early
of close
a period
(in which
it is known
thereis was
a significant
erve that
these
coefficients
are very
to one.
This means
thatthat
if there
a change
in the
Window size
to RDD),
this
is beginning
propagated
the long-term
due
toknown
theWindow
coefficient
associated
to lags. p-value
p-value
[days]
serve
value
atchange
the early
ofinaCoefficient
period
(in which
it is
thatsize
there
was aCoefficient
significant
to RDD), this change is propagated in the long-term due to the coefficient associated to lags.
-422.932
0.000
10
-299.85
0.000
20
-425.448
le 19. Treatments effect for period
1à 2 transitions for required Regulation-Up and Regulation-Down reserves.
-272.27
0.000
30
-427.617
e 19. Treatments effect for period transitions for required Regulation-Up and Regulation-Down reserves.
Regulation - Up -256.19
Regulation
- Down -429.346
0.000
40
0.000
20
0.000
30
0.000
40
2
3
4
5
-326.12
0.000
10
[days]
𝜏𝜏𝑝𝑝,𝑝𝑝+1 - Up -1.90
𝛽𝛽10𝑝𝑝,𝑝𝑝+1
0.578
0.004
10
Regulation
Regulation
- Down 10.16
Coefficient p-value
Window
size
Coefficient
p-value
Window
size
𝜏𝜏𝑝𝑝,𝑝𝑝+1
𝛽𝛽20𝑝𝑝,𝑝𝑝+1
0.134
-3.99
11.79
0.000
20
2à 3
[days] size
[days]
Coefficient p-value
Window
Coefficient
p-value
Window
size
0.024
-4.79
30
11.83
0.000
30
-326.12
0.000
10
-422.932
0.000
10
[days]
[days]
-7.09
0.000
40
12.98
0.000
40
-299.85
0.000
20
-425.448
0.000
20
-326.12
0.000
10
-422.932
0.000
10
-55.05
0.000
10
-37.46
0.000
10
-272.27
0.000
30
-427.617
0.000
30
-299.85
0.000
20
-425.448
0.000
20
-49
0.000
20
-49.93
0.000
20
-256.19
0.000
40
-429.346
0.000
40
3à 4
-272.27
0.000
30
-427.617
0.000
30
-45.25
0.000
30
-59.35
0.000
30
-1.90
0.578
10
10.16
0.004
10
-256.19
0.000
40
-429.346
0.000
40
0.000
40
0.000
40
-3.99
0.134
20 -40.92
11.79
0.000
20 -66.56
-1.90
0.578
10
10.16
0.004
10
-45.829
0.000
10
-24.027
0.000
10
-4.79
0.024
30
11.83
0.000
30
-3.99
0.134
20
11.79
0.000
20
0.000
20
0.000
20
-7.09
0.000
40 -47.3
12.98
0.000
40 -23.672
-4.79
0.024
30
11.83
0.000
30
4à 5
0.000
30
0.000
30
-55.05
0.000
10 -48.032
-37.46
0.000
10 -23.318
-7.09
0.000
40
12.98
0.000
40
0.000
40
0.000
40
-49
0.000
20 -48.745
-49.93
0.000
20 -22.964
-55.05
0.000
10
-37.46
0.000
10
-45.25
0.000
30
-59.35
0.000
30
-49
0.000
20
-49.93
0.000
20
-40.92
0.000
40
-66.56
0.000
40
-45.25
0.000
30
-59.35
0.000
30
-45.829
0.000
10
-24.027
0.000
10
-40.92
0.000
40
-66.56
0.000
40
-47.3
0.000
20
-23.672
0.000
20
-45.829
0.000
10
-24.027
0.000
10
-48.032 The Full0.000
0.000
30 (FCe-) Trends in Transmission,
-23.318Distribution,
0.000
30Costs for U.S. Investor Owned Electric Utilities, January 2017 | 51
Cost of Electricity
and Administration
-47.3
20
-23.672
0.000
20
-48.032
0.000
30
-23.318
0.000
30
54
10-Table 18, we can observe that these coefficients
are very close to one. This means that if there
is a change in the procured reserve value at
the early beginning of a period (in which it is
known that there was a significant change due
to RDD), this change is propagated in the longterm due to the coefficient associated to lags.
In Table 19 are presented the coefficients and
significance associated to the dummy variables used
to represent the application of NPRRs. From these
results, it is possible to rank the NPRRs according
to their impact on terms of reserves requirements.
In Table 20 and Table 21 are presented the NPRRs
identified sorted according to their significance. It
can be observed from Table 20 and Table 21 that
the most significant changes in terms of regulationup and regulation-down reserves required
occurred during the transition from zonal to nodal
market. In addition, it can be observed that the
least significant was the NPRR352, which was
related with the wind forecasting improvements.
TABLE 20
Sorting of NPRRs according to their impact on Regulation-Up procured reserves.
N
NPRR
Coefficient
1
NPRR045, NPRR050, NPRR210, NPRR214, NPRR239, NPRR258, and other zonal to nodal market changes
-272.27
2
NPRR460
-48.032
3
NPRR361
-45.25
4
NPRR352
-4.79
TABLE 21
Sorting of NPRRs according to their impact on Regulation-Down procured reserves.
N
NPRR
Coefficient
1
NPRR045, NPRR050, NPRR210, NPRR214, NPRR239, NPRR258, and other zonal to nodal market changes
-427.617
2
NPRR361
-59.35
3
NPRR460
-23.318
4
NPRR352
11.83
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
52
BIBLIOGRAPHY
[1] R. Billington and W. Li, Reliability Assessment
of Electric Power Systems Using Monte Carlo
Methods, Plenum Press, 1994.
[2] D. Kirschen and G. Strbac, Fundamentals of Power
System Economics, John Wiley & Sons, 2004.
[3] U.S. Federal Energy Regulatory Commission,
Promoting Wholesale Competition Through
Open Access Non-discriminatory Transmission
Services by Public Utilities, Docket RM95-8-000,
Washington, DC, 1995.
[4] E. Hirst and B. Kirby, “Unbundling Generation and
Transmission Services for Competitive Electricity
Markets,” Oak Ridge National Laboratory,
Tennessee, 1998.
[5] E. Ela, B. Kirby, N. Navid and J. C. Smith, “Effective
Ancillary Services Market Designs on High Wind
Power Penetration Systems,” in IEEE Power and
Energy Society General Meeting, San Diego, CA,
2012.
[6] E. Ela, M. Milligan and B. Kirby, “Operating
Reserves and Variable Generation,” 2011.
[7] Y. Zhang, V. Gevorgian, E. Ela, V. Singhvi and
P. Pourbeik, “Role of Wind Power in Primary
Frequency Response of an Interconnection,” in
International Workshop on Large-Scale Integration
of Wind Power Into Power Systems as Well as on
Transmission Networks for Offshore Wind Plants,
London, UK, 2013.
[8] L. Xu and D. Treatheway, “Flexible Ramping
Products: Revised Draft Final Proposal,” CAISO,
2014.
[9] J. Dumas, “2013 Ancillary Services Methodology:
Issue for the ERCOT Board of Directors,” 2013.
[10] D. Chattopadhyay and R. Baldick, “Unit
Commitment with probabilistic reserve,” Proc.
IEEE Power Engineering Society Winter Meeting,
pp. 280-285, 2002.
[11] NREL, “Eastern Wind Integration and
Transmission Study,” 2011.
[12] General Electric, “Western Wind and Solar
Integration Study,” NREL, 2010.
[13] NERC, “Standard BAL-0.005-0b Automatic
Generation Control,” 2007.
[15] ERCOT, “Monthly Operation Overview-January
2016,” 2016.
[16] NERC, “Standard BAL-002-0-Disturbance Control
Performance,” 2005.
[17] S. Newell, R. Carroll, P. Ruiz and W. Gorman,
“Cost-Benefit Analysis of ERCOT’s Future
Ancillary Services (FAS) Proposal,” 2015.
[18] S. S. Oren, “Auction design for ancillary reserve
products,” IEEE Prower Engineering Society
Summer Meeting, vol. 3, pp. 1238-1239, 2002.
[19] Potomac Economics, “2014 State of the Market
Report for the ERCOT Wholesale Electricity
Markets,” 2015.
[20] J. Dumas and D. Maggio, “Electric Reliability
Council of Texas Case Study: Reserve Management
for Integrating Renewable Generation in Electricity
Markets,” in Renewable Energy Integration
Practical Management of variability, uncertainty,
and flexibility in power grids, L. E. Jones, Ed.,
Academic Press, 2014, pp. 125-133.
[21] National Renewable Energy Laboratory, “Western
Wind and Solar Integration Study,” 2010.
[22] ERCOT, “Capacity, Demand and Reserves Report
May 2016,” 2016.
[23] ERCOT, “Hourly Load Data Archives,” 2016.
[Online]. Available: http://www.ercot.com/
gridinfo/load/load_hist/.
[24] D. S. Lee and T. Lemiux, “Regression Discontinuity
Designs in Economics,” Journal of Economic
Literature, pp. 281-355, 2010.
[25] M. Percoco, Regional Perspectives on Policy
Evaluation, SpringerBriefs in Regional Science,
2014.
[26] O. Ma, N. Alkadi, P. Cappers, P. Denholm, J.
Dudley, S. Goli, M. Hummon, S. Kiliccote, J.
MacDonald, N. Matson, D. Olsen, C. Rose, M.
D. Sohn, M. Starke, B. Kirby and M. O’Malley,
“Demand Response for Ancillary Services,” IEEE
Transactions on smart grid, pp. 1988-1995, 2013.
[27] U.S. Energy Information Administration, “Form
EIA-860 detailed data,” 2016. [Online]. Available:
http://www.eia.gov/electricity/data/eia860/.
[14] GE Energy, “Analysis of Wind Generation Impact
on ERCOT Ancillary Services Requirements,”
2008.
The Full Cost of Electricity (FCe-) Trends in Transmission, Distribution, and Administration Costs for U.S. Investor Owned Electric Utilities, January 2017
|
53