Retail

EE379K/EE394V Smart Grids:
The Retail Perspective
Paul Wattles
SENIOR ANALYST,
MARKET DESIGN &
DEVELOPMENT
ERCOT
Spring 2017
Ross Baldick,
Department of
Electrical and
Computer
Engineering
Outline


ERCOT Overview
Advanced Metering



15-Minute Data & 15-Minute Settlement
Retail Smart Grid Product Offerings
Impacts on:




ISO Grid Operations
Wholesale Market Prices
Load Forecasting
Consumers
2
The ERCOT Region
The interconnected
electrical system serving
most of Texas, with
limited external
connections
• 90% of Texas electric
load; 75% of Texas land
• 71,093 MW peak, August
11, 2016
• More than 46,500 miles
of transmission lines
• 550+ generation units
ERCOT connections to other grids are limited
to ~1250 MW of direct current (DC) ties,
which allow control over flow of electricity
220 MW with SPP
600 MW with SPP
30 MW with CFE
at Eagle Pass
100 MW with CFE
at Laredo
300 MW with CFE at Mc Allen
3
U.S./Canada ISOs and RTOs
Independent
System
Operators
and Regional
Transmission
Organizations
are the ‘air
traffic
controllers’ of
the bulk
electric power
grids
4
ERCOT Inc.


The Texas Legislature restructured the Texas electric market in 1999
and assigned ERCOT four primary responsibilities:
•
Maintain system reliability
•
Facilitate competitive wholesale market
•
Ensure open access to transmission
•
Facilitate competitive retail market
ERCOT is regulated by the Texas Public Utility Commission (PUC)
with oversight by the Texas Legislature

Because the ERCOT grid is intrastate, the ERCOT markets are
not jurisdictional to the Federal government (i.e., FERC)
ERCOT is not a market participant
and does not own generation or
transmission/distribution wires
5
Texas Competitive Model
Applies to formerly vertically-integrated investor-owned utilities
(Municipally-owned Utilities and Electric Cooperatives also own generation and T&D)
• Generators are owned
by privately owned
(merchant) companies,
who compete in the
ERCOT market to sell
to Load Serving
Entities
• T&D facilities are owned
and operated by
Transmission &
Distribution Service
Providers (TDSPs),
which are regulated by
the PUC
• Load is served by
Retail Electric
Providers, who
compete to sell power
to end-use customers
Parent companies may own both generation and retail
6
2 models within ERCOT
Municipals & Cooperatives
AKA Non-Opt in
Entities (NOIEs) are
still vertically
integrated
Many have existing
and developing smart
grid initiatives:
-- AMI
-- Smart thermostats
-- Demand response
Possible triggers:
Demand charge
avoidance, real-time
prices, congestion
management
Competitive Choice
‘Utility’ a mostly
obsolete term
26%
74%
Share of total ERCOT Load
The two worlds have very
different smart grid incentives
(more on this later)
Dozens of REPs
competing for
residential and
commercial accounts
Terms typically range
from 3-24 months
Some pre-paid,
renewable options
>99% advanced
metering
7
Competitive areas & NOIEs
Competitive Retail Area
Municipally Owned Utilities
and Electric Co-Ops
(NOIEs)
26%
74%
Source: Texas Solar Power Association
Share of total Load
8
Nodal Energy Market





ERCOT clears the real-time energy market every five minutes,
dispatching generation with the lowest offers to serve the load,
subject to transmission constraints
Locational marginal prices (LMPs) are
produced every 5 minutes at >11,000 nodes,
including >550 Generation Resource Nodes
If there is no congestion on the
system, all LMPs will be equal
(set by the marginal unit)
Generators are paid the LMP at
their specific Resource Node
Loads are billed the
weighted-average price
at the Load Zone
9
Annual Energy and Peak Demand
10
Evolving Resource Mix
100,000
Nameplate capacity by unit type, 1999 thru 2016
90,000
(Does not account for retirements)
80,000
70,000
MW
60,000
50,000
40,000
30,000
20,000
10,000
0
1999
2000
2001
2002
Nuclear
2003
Coal
2004
2005
2006
Other Renewables
2007
Gas-ST
2008
Gas-CC
2009
2010
Gas-GT/IC
2011
Wind
2012
2013
Solar
2014
2015
2016
11
Current Records
Peak Demand Record: 71,110 megawatts (MW)
 Aug. 11, 2016, 4-5 p.m.
Summer 2016 Monthly Peaks
Weekend Record: 66,921 MW
 Sunday, Aug. 7, 2016, 5-6 p.m.
June: 64,896 MW (June 15)
Winter Peak Record: 57,265 MW
 59,650 MW, Jan. 6, 2017
July:
Wind Generation Records (instantaneous)

•
15,033 MW, Nov. 27, 2016, 12:36 p.m.
- Non-Coastal Wind Output = 11,100 MW
- Coastal Wind Output = 3,800 MW
- Supplying 45% of the load
- Active Wind Capacity = 17,150 MW
June Record: 66,548 MW – 6/26/12
67,469 MW (July 14)
July Record: 67,650 MW – 7/30/15
August: 71,110 MW (Aug. 11)
August Record: 71,110 MW – 8/11/16
September: 66,853 MW (Sept. 19)
48.28% Wind Penetration, March 23, 2016, 1:10 a.m.
- Total Wind Output = 13,154 MW
- Total Load = 27,245 MW
Sept. Record: 66,853 MW – 9/19/16
12
Weather Impacts on Load by Customer Type
Thursday, Aug. 11, 2016
5:00 PM
ERCOT Load: 71,093 MW
Temperature in Dallas: 106°
Thursday, March 24, 2016
5:00 PM
ERCOT Load: 33,597 MW
Temperature in Dallas: 62°
>37,000 MW of
weather-sensitive
load -- 53% of peak
Customer class breakdown is
for competitive choice areas;
percentages are extrapolated
for municipals and co-ops to
achieve region-wide estimate
 Large C&I are IDR Meter
Required (>700kW)
 Hourly integrated demand
values

13
Fuel Mix on Those Same Days
Power Dispatch Summary by Fuel Type
March 24, 2016
August 11, 2016
Max Gen: 71,636 MW
at 03:55:19 PM
Wind (MW)
Wind (MW)
Max Gen: 34,593 MW
at 07:20:21 AM
14
Wholesale Market Price Caps

Escalating offer caps in the energy-only market








2002:
2007
2008
2011:
2012:
2013:
2014:
2015:
$1,000
$1,500
$2,250
$3,000
$4,500
$5,000
$7,000
$9,000
$10,000
$9,000
ERCOT
System-Wide Offer Caps
$8,000
$7,000
$6,000
$5,000
$4,000
$3,000
$2,000
$1,000
$0
2002


2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
Applies to offers for energy (MWh) and Ancillary Services (MW per hr)
The energy market has never cleared at the $9,000/MWh cap
15
Peaker Net Margin (PNM)
• The PNM is a calculation
designed to measure the
annual net revenue of a
hypothetical peaking unit
• If the PNM for a year
reaches a cumulative
total of $315,000 per
MW, the system-wide
offer cap is then reduced
to the higher of $2,000
per MWh or 50 times the
daily natural gas price
index
2016 PNM
= $29,991
• This threshold (defined in
PUC Subst. Rule §25.505)
has never been met
Source: Potomac Economics (ERCOT Independent Market Monitor) 2015 State of the Market Report
16
Real-Time Energy Prices
Intervals
HUB Avg 15-min. Settlement Point Prices
2015 and 2016
<$0
$0-$30
>$30-$75
>$75
2015
2016
238
569
31,287
31,562
3,110
2,605
405
400
17
Tale of two peaks
Summer peak days:
• Aug. 10, 2015
• Aug. 11, 2016
ERCOT Load was
>1,700 MW higher
on the peak day in
2016
18
Tale of two peaks
Summer peak days:
• Aug. 10, 2015
• Aug. 11, 2016
Day-Ahead Market
energy prices
19
Tale of two peaks
Summer peak days:
• Aug. 10, 2015
• Aug. 11, 2016
Real-Time Market
energy prices
20
Tale of two peaks
Summer peak days:
• Aug. 10, 2015
• Aug. 11, 2016
WIND OUTPUT:
2016 peak day was
>2,700 MW higher
than 2015 peak day
21
Tale of two peaks
Summer peak days:
• Aug. 10, 2015
• Aug. 11, 2016
NET LOAD:
Actual system Load
minus wind was
>1,000 MW lower
in 2016 than 2015
In this example, wind reduced price volatility
22
Wholesale market prices

Prices can spike for reasons other than scarcity of
generation



Unit trips, ramp rate limitations, inaccurate
forecasting of weather or Load
LSEs hedge their risk against price spikes through
forward contracts and/or Day-Ahead Market
procurement
Another tool is the ability to reduce Load in real
time (demand response)


If the LSE is short, DR can reduce exposure to the
high price
If the LSE is long, DR allows it to ‘sell the power back
at the real-time price’
23
Advanced Metering origins

House Bill 2129 (2005 Legislature) amended
Sec. 39.107 of the Utilities Code:


(h) The commission shall establish a nonbypassable
surcharge for an electric utility or transmission and
distribution utility to use to recover reasonable and
necessary costs incurred in deploying advanced metering
and meter information networks to residential customers
and nonresidential customers other than those required
by the independent system operator to have an interval
data recorder meter.
Surcharge = accelerated cost recovery
24
Advanced Metering origins


PUC Substantive Rule §25.130 -- Advanced
Metering (2007)
Purposes of the rule:





to implement the Legislation by authorizing the
surcharge;
increase the reliability of the regional electrical
network;
encourage dynamic pricing and demand response;
improve the deployment and operation of
generation, transmission and distribution assets, and
provide more choices for electric customers
25
Advanced Metering origins

Key elements of the Rule:




Applies to investor-owned TDSPs only
(NOIEs not affected)
Implementation optional for TDSPs, but
deployments eligible for accelerated cost
recovery via special surcharge
AMI meters must measure consumption
in 15-minute intervals
Interval data shall be used in wholesale
market settlement at the ESI ID level
26
15-minute metering
Pre-AMI
AMI
Energy data points per month
1
2,880
Applies to residential premises without on-site distributed generation.
27
Advanced Metering benefits

>7 million advanced meters now active in
the competitive choice areas of ERCOT


>99% of ERCOT Load is now settled on 15minute interval data (includes AMI,
competitive IDR, and NOIE IDR)
TDSP benefits:



Reduced meter-reading costs
Advanced outage detection
Faster switching and move-ins/move-outs
28
Advanced Metering benefits

Retail Electric Provider (REP) benefits:



ISO benefits:


Settlement accuracy (no more load profiles; see next
slide)
Real money if customers reduce load during highpriced periods
Settlement timeliness & accuracy
Customer benefits:


Access to granular energy usage data
A wider selection of REP products to choose from
29
Why settlement is important


In settlement, LSE purchases are reconciled with
generators’ sales
Prior to the AMI implementation, REP obligations for
residential and small commercial customers were
based on Load Profiles



Profiles are estimates of average individual usage, based
on statistical samples of data from ‘like’ customers
A Profile for each customer type was created for each
Operating Day, using weather & other inputs
Interval-level values were then assigned to each customer
according to their Profile type, scaled based on monthly
kWh usage
30
Profile example

A profile assumes
customers of this type
on average have the
profiled Load shape


The load magnitude is
adjusted (scaled) based
on the customer’s
monthly kWh
consumption
Prior to AMI, REP Load
was settled based on
this estimate
Residential High Winter Ratio Profile Type for
North Central Weather Zone, Aug. 11, 2016
31
Why settlement is important

Settlement based on Load Profiles would be
accurate at the REP aggregate load level only if
its customers were not deviating significantly
from the profiled shapes


The REP would have to accept the settlement
outcome even if its customers were making
significant changes to their load shapes
In other words…



Profiles are oblivious to intelligent load management
Profiles are hard barriers to price elasticity of demand
Profiles kill demand response
32
Why settlement is important

Settlement on actual 15-minute data cures this
problem


When customers are settled on their actual usage,
the benefits of any action taken to reduce Load
during a period of high wholesale prices will accrue
directly to the LSE
This gives the LSE (in this case, the REP) an incentive
proportional to real time prices to promote intelligent
load management and demand response for its
customers
33
New product options
34
Retail price/demand response
• If retail DR and price response penetration are a key
metric in measuring the success of the ERCOT retail
market and the AMI investment, how are we doing?
• PUC Subst. Rule §25.505(e)(5): Load serving entities (LSEs) shall
provide ERCOT with complete information on load response
capabilities that are self-arranged or pursuant to bilateral
agreements between LSEs and their customers.
• Leveraging this Rule language, ERCOT has worked with REPs since
2013 to collect data on various product offerings, including:
–
–
–
–
–
Time of Use
Peak Rebates
Real-Time Pricing
Block & Index Pricing
Other Load Control Products
35
Residential Time of Use
331,138
321,504
289,848
ESI IDs
Time of Use prices vary across different blocks of hours, with pre-defined
prices and schedules; examples: Free Nights, Free Weekends
Total residential ESI IDs ~6.17 million
ESI ID = Electric Service Identifier: “The basic identifier assigned to each Service Delivery Point used in the
registration and settlement systems managed by ERCOT ...” (In vast majority of cases, equates to a meter.)
36
C&I Indexed Pricing
40,000
‘Indexed pricing’ includes:
• Real-Time Pricing (tied to 15-minute wholesale market prices), and
• Block & Index (fixed pricing for a defined volume of usage coupled
with indexed pricing for usage exceeding the block)
This slide does not include 1,326 residential ESI IDs on RTP in 2016
30,000
7,386
5,507
10,986
8,326
15,464
22,850
5,820
10,000
14,146
28,286
16,493
ESI IDs
20,000
0
-10,000
Large drop-off
in 2014 likely
due to reporting
inconsistency
-10,673
-6,760
-22,779
Total non-residential ESI IDs ~946,000
-20,000
New
Stayed
Dropped
-30,000
2013
2014
2015
2016
37
478,243
465,527
Customers on Peak
Rebate plans are
eligible for financial
incentives for load
reductions taken
during periods
defined by the REP
and communicated
to the customer in
advance
409,434
ESI IDs
Residential Peak-Time Rebate
Total residential ESI IDs ~6.17 million
38
35,369
32,289
Total non-residential
ESI IDs ~946,000
30,185
ESI IDs
C&I Peak-Time Rebate
39
Effects on system Load

Since 2011, the ERCOT real-time market has
had very few sustained high-price events



Price spikes tend to be short-term (1-2 SCED
intervals), often related to ramp rate constraints
ERCOT has analyzed data from the events we
have
ERCOT has also analyzed self-initiated events
as reported by REPs, although this data is also
sub-optimal

In many cases, REPs deployed only a fraction of
total customers in the program
40
Examples of real-time price response

Combined Electric Service Identifiers (ESI IDs) reported
on Real-Time Pricing and Block & Index offerings
Analyzed days with 4 or more consecutive intervals of prices >$200
Note MW scales are different

March 3, 2014
• 10,046 ESI IDs
• Maximum reduction: 119 MW
• Maximum price: $4,991/MWh
August 13, 2016
• 13,503 ESI IDs
• Maximum reduction: 129 MW
• Maximum price: $1,089/MWh
41
Demand charges (4CP)

The Four Coincident Peaks (4CP) in ERCOT are the peak
system-wide 15-minute settlement intervals in each of
the four summer months


June, July, August, September
Simple average of energy usage
across these four intervals is the
basis of various T&D charges for
much of the ERCOT Load
100%
90%
80%
Retail Choice
Residential
37.3%
70%


NOIEs, at the boundary meter
60%
level
Retail Choice customers50%with
peak demand ≥700 kW40%
(Interval Data Recorder
30%
meter required)
Small
Commercial
18.4%
Large C&I
17.3%
Retail
choice
load
“Large C&I” =
IDR Required
20%
Combined, >44% of ERCOT Load is subject to 4CP charges
10%
NOIE
27.1%
42
4CP charges as a DR incentive


4CP was not designed as an incentive for demand
response, but…
Reducing Load during these intervals yields
considerable savings



Many Loads and NOIEs have acquired 4CP predictors
or developed 4CP prediction capability in-house


NOIEs can reduce their 4CP Load Ratio Share, lowering
their share of Transmission Cost of Service obligation
Retail Choice Loads can directly reduce charges on their
bills for the following year
Predictor services are available by subscription from LSEs
and third parties
Entities then plan demand response around potential
4CP intervals
43
4CP Tariffs: Hypothetical case study

Let’s assume an industrial customer:






Has 10 MW of Load and is capable of interrupting all of it
Is connected at transmission voltage
Is in Oncor service territory, where the Summer 2016
Transmission Cost Recovery Factor tariff for such Loads
was $3.48501 per 4CP kW
Correctly anticipated and reduced Load to zero for all four
4CP intervals in 2016
Our customer’s transmission charge line item would be
$0.00 per month for each month of 2017
If he had been consuming his usual 10 MW during
those 4 intervals, his charge would be:


$3.485 x 1000 (kW to MW) x 10 (MW) = $34,850 per
month
x 12 (months) =$418,200 in savings for the year
44
NOIE 4CP example
These are
4 of 15 Energy
Rush Hour
events from
summer of
2016




June (1)
July (5)
August (4)
Sept. (5)
(screenshot from
my cell phone)
Source: Nest.com
45
4CP incentives are accelerating
Postage Stamp Transmission Rates


Postage Stamp rate for
Transmission Cost of Service
has more than tripled since
2002
>$7B in Competitive
Renewable Energy Zone
(CREZ) transmission
investment


$4.9B of CREZ activated in
2013 alone
>$2B in projects activated in
2016
2002-2017 ($ per 4CP kW)
Source: PUCT Dockets
Currently $52.91
The postage stamp rate (per 4CP kW) is assessed
on DSPs by multiplying the rate by their average
Load across the 4CP intervals. DSPs then
reimburse TSPs for their transmission investments.
ERCOT’s analyses of numerous 4CP and near-CP days indicate:
• 300-700 MW of industrial load 4CP response
• 300-900 MW of NOIE 4CP response
Level of response varies depending on how obvious the 4CP day is
46
Transmission additions over time
Looking back….
And looking ahead….
Source:
ERCOT Report on Existing & Potential Constraints and Needs, 2016
Source:
ERCOT Transmission Project & Information Tracking report, Feb. 2017
47
Impacts on resource adequacy

How does economic DR (price and 4CP response) affect the
load forecast?

From the June 2012 Brattle Group report to the PUC on Resource
Adequacy:



In general, price and 4CP response behavior is continually
being baked into ERCOT’s load forecasts


‘Price-based load reductions were likely a major contributor to the 1,700 MW
ERCOT load forecasting error in 2011 when prices reached $3,000/MWh.
The error may also be attributable in part to 4CP response, voluntary public
response to conservation appeals, and load forecast model error.’
This was based on a rerun of the long-term load forecast using actual 2011
weather
2011 was a outlier year, to say the least
ERCOT has revised its long-term load forecast methodology



Load growth has mostly decoupled from economic growth
Previous econometric model replaced by neural networks
New methodology better incorporates demand-side behavior
48
Load forecast impacts
A couple examples of the changing landscape
THEN
NOW
55” LED
55” Plasma
(circa 2000)
690 Watts
Annual cost
to operate:
$148.63
Incandescent
100 Watts
Costs to operate based on current Energy Star standards
37 Watts
Annual cost
to operate:
$7.97
LED
100 Watt equivalent
15 Watts
49
Questions?
‘Relativity’
M.C. Escher, 1953
50
Summary

The ERCOT region consists of two types of ‘utilities’ with
different business models and incentives:







Competitive choice areas (unbundled)
Non-Opt In Entities (muni’s and coops)
Peak demand in ERCOT is driven heavily by residential air
conditioning Load
Energy prices are low, impacted by low-cost natural gas and
zero fuel-cost wind generation
Near 100% AMI deployment in competitive choice areas
Interval metering combined with settlement on actual data is
enabling growth in dynamic pricing and DR-related retail
product offerings
Largest financial incentive for DR is demand-charge
avoidance (4CP)
51
Homework Exercises:
Due Month/Day
A Retail Electric Provider with 8 MW of demand response
capability in the North Load Zone purchases power in the Day
Ahead Market for $40/MWh across the hours of 4-6 PM
(Hours ending 1700 and 1800)
1.

The next operating day, North Load Zone real-time prices at
4:15 PM rise from $40 to $2,300/MWh for three five-minute
SCED runs (SCED intervals ending at 4:20, 4:25 and 4:30) -- a
full 15-minute settlement interval

The REP accurately predicts the price spike and successfully
deploys its 8 MW of DR across the full settlement interval

What is the REP’s financial outcome (net of the Day-Ahead
purchases) for the 15-minute interval ending at 4:30 p.m.?
52
Homework Exercises:
Due Month/Day
2.


A residential customer in a retail choice area switches REPs
and signs up for a ‘Free Weekends’ time-of-use (TOU) price
offering

The new plan charges $.15/kWh (includes all T&D charges)
for weekday hours -- 6 AM Monday thru 10 PM Friday

The new plan charges $.00 for weekend hours (including $.00
for T&D)
His previous plan charged a flat $.09/kWh for all energy and
T&D
 When on the previous plan, in November 2016 he used
800kWh during the weekday hours, and 200 kWh during
weekend hours
Assuming the customer had been on the TOU plan in
November 2016, how much energy usage would he have
needed to shift to the weekend hours in order to break even
financially?
53