Case Studies in Advanced Thermostat Control for Demand Response

Case Studies in Advanced Thermostat Control for Demand Response Joseph S. Lopes Senior Vice President Applied Energy Group, Inc. Hauppauge, NY www.appliedenergygroup.com J. Lopes; AEIC Load Research Conference St. Louis, MO; July 2004 1

Introduction More Demand for Demand Response Programs! Utilities no longer control supply in many states due to deregulation Need for more flexibility and distributed options like demand-side peak reduction Regulators, consumers concerned about shortages and price spikes Transmission constraints J. Lopes; AEIC Load Research Conference St. Louis, MO; July 2004 2

Introduction What are the best options for proven demand response? In Summer-peaking systems, air conditioning loads are the primary driver of peaks Air conditioning is somewhat discretionary New homes have a high degree of air conditioning Nearly all businesses have air conditioning J. Lopes; AEIC Load Research Conference St. Louis, MO; July 2004 3

Air Conditioning Demand Response Historically, implemented with one-way switches on central A/C systems New generation of electronic thermostats now available and used by many utilities: Two-way communications, which ensures verification Monitoring and control capability Internet access Interval data (runtime and temperature) available for virtually all sites Customer overrides can be tracked J. Lopes; AEIC Load Research Conference St. Louis, MO; July 2004 4

Case Studies AEG experience with four utilities: LIPA (NY), Southern California Edison, Consolidated Edison, Aquila (new for 2004) Use the same technology electronic programmable thermostats with override Thermostat with two-way communications and data access via public pager networks Control of either duty cycle or temperature Internet access Hourly runtime and temperature data available for virtually all sites J. Lopes; AEIC Load Research Conference St. Louis, MO; July 2004 5

Controllable Thermostat Programmable thermostat with 2-way pager access J. Lopes; AEIC Load Research Conference St. Louis, MO; July 2004 6

Case Studies Long Island Power Authority (LIPA) Since 2001, Central A/C units in over 20,000 residential, 3,000 small commercial Free thermostat and $25 (one-time) Customers have thermostat access over Internet LIPA can control up to 7 days from 2-6 pm Customer can override without penalty J. Lopes; AEIC Load Research Conference St. Louis, MO; July 2004 7

Case Studies Consolidated Edison Company of NY Since 2002, now over 10,000 residential sites Free thermostat and $25 (one-time) Thermostat access over Internet Controls when NY ISO requests (typically peak summer days 1-6 pm) Customer can override without penalty Small Commercial Pilot Program (2004) Same terms except scalable one-time incentive ($25 per 3 ton increment) J. Lopes; AEIC Load Research Conference St. Louis, MO; July 2004 8

Case Studies Southern California Edison (SCE) Since 2002; target of 5,000 small commercial sites (achieved in 2003); approved for 2004 Free thermostat plus $300 annual incentive, with $5 penalty for each override Control varies from 1-6 pm; in 2003 up to 20 control days allowed Logger data on sample confirmed runtime data accuracy (within 3%) J. Lopes; AEIC Load Research Conference St. Louis, MO; July 2004 9

Thermostat Control Options Duty Cycle Control Limits runtime to a fixed percentage (e.g. 50% control limits to 15 minutes off per half-hour) LIPA and Con Edison typically used 50% duty cycle control Setpoint Temperature Control Increase current A/C thermostat setpoint by a specific value (e.g. 4 degrees) SCE typically used 4 degree temperature control J. Lopes; AEIC Load Research Conference St. Louis, MO; July 2004 10

DLC Impact Evaluation Hybrid Comparison Day Analysis Use best comparison day, based on closest day with similar weather Temperature, humidity, heat build-up Patterns change over course of summer Not cost-effective to collect every day Compare baseline and control day 2-3 hours before control hour should match closely Small Adjustment may be needed to match up J. Lopes; AEIC Load Research Conference St. Louis, MO; July 2004 11

Duty Cycle Control Case LIPA (NY) and Con Edison Use 50% Duty Cycle Control Both residential and commercial sites Typically control during afternoon utility peak period (1-6 pm or 2-6 pm) Allow overrides without penalty Can confirm control and monitor overrides J. Lopes; AEIC Load Research Conference St. Louis, MO; July 2004 12

Duty Cycle Control Case LIPA EDGE 2003 DLC RUNTIME ANALYSIS SEGMENT R - RESIDENTIAL % of all data - Average - Hr 16-18 60% 50% 40% 30% 20% 10% 0% 48.2% 27.2% 18.5% 15.7% 8.1% 8.4% 8.4% 7.4% 8.1% 5.6% 6.0% 5.8% 5.7% 3.9% 4.3% 3.6% 2.6% 3.4% 1.0% 1.3% 2.3% 1.9% 1.6% 1.1% 0.0% 5.0% 15.0% 25.0% 35.0% 45.0% 55.0% 65.0% 75.0% 85.0% 95.0% 100.0% BASELINE - TAB 1 CONTROL DAY - TAB 11 15-18% off; 25% of baseline @100%; only overrides over 55% J. Lopes; AEIC Load Research Conference St. Louis, MO; July 2004 13

Duty Cycle Control Case 80.0% 70.0% LIPA EDGE 2003 DLC RUNTIME ANALYSIS SEGMENT R - RESIDENTIAL Duty Cycle % 60.0% 50.0% 40.0% 30.0% 20.0% 10.0% 0.0% 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 Tab 1 Adj. Baseline Tab 1 Control Day Control 3-6 pm; Some payback after control ends J. Lopes; AEIC Load Research Conference St. Louis, MO; July 2004 14

Duty Cycle Control Case LIPA EDGE 2003 DLC RUNTIME ANALYSIS SEGMENT R - RESIDENTIAL kw Load Impacts 1.00 0.90 0.80 0.70 0.60 0.50 0.40 0.30 0.20 0.10 0.00 0.91 0.89 0.75 0.00 0.00 0.00 0.00 0.00 0.00 12 13 14 15 16 17 18 19 20 Hour Ending Impacts reduced over time mainly from increased overrides J. Lopes; AEIC Load Research Conference St. Louis, MO; July 2004 15

Duty Cycle Control Case CON EDISON DLC RUNTIME ANALYSIS SEGMENT C - CONFIRMED ACCOUNTS % of all Data 1-6 pm 60% 50% 40% 30% 20% 10% 0% 55.1% 43.8% 15.3% 15.0% 8.8% 4.6% 5.5% 6.6% 7.4% 6.9% 5.4% 1.0% 1.9% 2.7% 3.1% 3.6% 3.6% 0.7% 1.7% 2.0% 1.3% 1.5% 1.2% 1.3% 0.0% 5.0% 15.0% 25.0% 35.0% 45.0% 55.0% 65.0% 75.0% 85.0% 95.0% 100.0% BASELINE - TAB 2 CONTROL DAY - TAB 21 Residential: 15% off; 44% of baseline @100% J. Lopes; AEIC Load Research Conference St. Louis, MO; July 2004 16

Duty Cycle Control Case 90.0% 80.0% 70.0% 60.0% CON EDISON DLC RUNTIME ANALYSIS SEGMENT C - CONFIRMED ACCOUNTS kw 50.0% 40.0% 30.0% 20.0% 10.0% 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 Tab 2 Baseline Tab 2 Control Day Residential 1-6 pm control - some payback after control ends J. Lopes; AEIC Load Research Conference St. Louis, MO; July 2004 17

Duty Cycle Control Case 1.20 CON EDISON DLC RUNTIME ANALYSIS SEGMENT C - CONFIRMED ACCOUNTS kw Load Impacts 1.00 0.80 0.60 0.40 0.20 1.02 0.98 0.98 0.89 0.79 0.00 12 13 14 15 16 17 18 19 20 Tab 2 Impacts (97 deg day) Impacts reduced over time mainly from increased overrides J. Lopes; AEIC Load Research Conference St. Louis, MO; July 2004 18

Duty Cycle Control Case LIPA EDGE 2003 DLC RUNTIME ANALYSIS SEGMENT C - COMMERCIAL % of all data - Average - Hr 16-18 50% 40% 30% 20% 10% 0% 40.1% 31.2% 29.7% 28.5% 10.2% 7.9% 5.1% 3.5% 4.0% 3.9% 3.9% 4.0% 4.3% 1.5% 2.4% 2.6% 2.6% 3.3% 2.6% 2.8% 1.0% 1.7% 1.9% 1.2% 0.0% 5.0% 15.0% 25.0% 35.0% 45.0% 55.0% 65.0% 75.0% 85.0% 95.0% 100.0% ADJUSTED BASELINE - TAB 1 CONTROL DAY - TAB 11 Commercial: About 40% @100%, nearly 30% of units were off J. Lopes; AEIC Load Research Conference St. Louis, MO; July 2004 19

Duty Cycle Control Case 80% LIPA EDGE 2003 DLC RUNTIME ANALYSIS SEGMENT C - COMMERCIAL 70% Duty Cycle % 60% 50% 40% 30% 20% 10% 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 Tab 1 Adj. Baseline Tab 1 Control Day Control 3-6pm; No observed payback J. Lopes; AEIC Load Research Conference St. Louis, MO; July 2004 20

Duty Cycle Control Case 2.00 LIPA EDGE 2003 DLC RUNTIME ANALYSIS SEGMENT C - COMMERCIAL kw Load Impacts 1.50 1.00 0.50 1.32 1.43 1.01 0.00 0.00 0.00 0.00 0.00 0.00 0.00 12 13 14 15 16 17 18 19 20 Hour Ending Impacts reduced over time mainly from declining loads J. Lopes; AEIC Load Research Conference St. Louis, MO; July 2004 21

Duty Cycle Control Only effective on customers whose base duty cycle (runtime) exceeds limit (e.g. 50%) Impacts are larger on more severe days as more customers exceed the duty cycle limit more suited to emergency operation Impacts are more consistent and maintained for a longer period Potentially more severe on some customers than others, such as those with undersized systems Easy to identify overrides from runtime data J. Lopes; AEIC Load Research Conference St. Louis, MO; July 2004 22

Setpoint Temperature Control Cases Southern California Edison (2003) Thermostat setpoint increased #1: Aug 12 (3-5, 2 deg.) 96 degrees max #2: Aug 15 (2-6, 4 deg.) 98 degrees max #3: Aug 18 (2-4, 4 deg.) 88 degrees max About 4,400 runtime data points each day (all commercial sites) J. Lopes; AEIC Load Research Conference St. Louis, MO; July 2004 23

Setpoint Temperature Control Cases 80% 70% SCE 2003 RUNTIME ANALYSIS Impact Analysis - All Participants Duty Cycle % 60% 50% 40% 30% 20% 10% 0% 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 Tab 2 Base - Aug 14 Tab 2 Control Day - Aug 15: 2-6p -4D 15.1dd 98 max day; 4 setpoint change 2-6pm, no payback J. Lopes; AEIC Load Research Conference St. Louis, MO; July 2004 24

Setpoint Temperature Control Cases SCE 2003 RUNTIME ANALYSIS Load Impacts by Temperature Tab - All Participants 1.50 1.40 1.28 1.00 kw 0.50 0.59 0.66 0.52 0.54 0.39 0.19 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 13 14 15 16 17 18 Tab 1 Control Day - Aug 12: 3-5p -2D 17.2dd Tab 2 Control Day - Aug 15: 2-6p -4D 15.1dd Tab 3 Control Day - Aug 18: 2-4p -4D 12.2dd RED (96 max) 3-5pm, 2 control; BLUE (98 ) 2-6pm, 4 control; GREEN (88 max) 2-4pm, 4 control J. Lopes; AEIC Load Research Conference St. Louis, MO; July 2004 25

Setpoint Temperature Control Cases 82.0 SOCAL ED 2003 TEMPERATURE ANALYSIS Impact Analysis - All Participants 81.0 Temperature 80.0 79.0 78.0 77.0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 Tab 2 Base - Aug 14 Tab 2 Control Day - Aug 15: 2-6p -4D 15.1dd 98 Day (2-6pm, 4 control) hourly indoor temperature impacts J. Lopes; AEIC Load Research Conference St. Louis, MO; July 2004 26

Setpoint Temperature Control Cases SOCAL ED 2003 TEMPERATURE ANALYSIS Impacts by Temperature Tab - All Participants 1.20 1.12 1.10 Temperature Impact 1.00 0.80 0.60 0.40 0.20 0.41 0.48 0.99 0.94 0.00 0.00 0.00 0.00 0.00 0.00 0.00 13 14 15 16 17 18 Tab 1 Control Day - Aug 12: 3-5p -2D 17.2dd Tab 2 Control Day - Aug 15: 2-6p -4D 15.1dd RED (96 max) 3-5pm, 2 control; BLUE (98 ) 2-6pm, 4 control Indoor Temperature Impacts J. Lopes; AEIC Load Research Conference St. Louis, MO; July 2004 27

Setpoint Temperature Control Affects all customers equally in terms of relative comfort Impacts are consistent across a range of weather conditions more suited to frequent use as a load reduction option Impacts are more pronounced in first hour and decline in subsequent hours Potential for customers to pre-cool and reduce impact achieved Could penalize customers already conserving Difficult to identify overrides from runtime data, but should see fewer overrides J. Lopes; AEIC Load Research Conference St. Louis, MO; July 2004 28

Duty Cycle vs. Setpoint Control Duty Cycle Control More effective for longer periods More suited to residential Temperature Setpoint Control More consistent across participants Higher initial hour impacts More suited to commercial (declining PM loads) J. Lopes; AEIC Load Research Conference St. Louis, MO; July 2004 29

Free Riders Some Participants off during control period: App. 15-18% of residential units off on hot days Mostly people not home, on vacation or prefer not using their A/C Some multiple units (e.g. 2 nd Floors) are 31% of LIPA; 81% of all units are single/1st App. 22-30% of comm. units off on hot days Multiple units are 60% of all units in SCE; 60% of units are single/1 st ; 21% are 2 nd units J. Lopes; AEIC Load Research Conference St. Louis, MO; July 2004 30

Free Riders: Multiple Units Commercial: 2 nd Units use only 79% of Primary Units; 3 rd Units only 65%, 4 th units 55%, etc. 60 SCE Runtime Data - 9/5/2003 Runtime vs. # of Thermostats Daily kwh 50 40 30 20 100.0% 79.4% 64.6% 54.7% Sum of Daily kwh (with %Max label) 49.5% 45.3% 47.5% 42.5% 37.4% 37.6% 10 0 1 2 3 4 5 6 7 8 9 10 Unit Number (declining daily use order) J. Lopes; AEIC Load Research Conference St. Louis, MO; July 2004 31

Free Riders Could only be reduced by pre-qualifying units at a site to increase average impacts Potential Discrimination Issue Minimum summer use billing increment level Residential sites could exclude 2 nd floor if only bedrooms (lower use and less coincident) Commercial sites would require evaluation of likelihood of little-used units Free Riders are generally unavoidable, but must be factored into any assessment of potential impacts and costs J. Lopes; AEIC Load Research Conference St. Louis, MO; July 2004 32

SUMMARY For Residential programs Duty cycle control works best, especially for only a few control days per season Could reduce free riders by target marketing to higher summer use customers Temperature setpoint control would work best for energy conservation improvements (by utility or customer) Most customers (77%) use programmable features of thermostats themselves J. Lopes; AEIC Load Research Conference St. Louis, MO; July 2004 33

SUMMARY For Commercial programs Temperature setpoint control works best if many control days or short-duration curtailments should also reduce overrides Duty Cycle control works best for more hours of sustained load reduction Runtime data could provide good source of baseline A/C load profiles by business type Targeting specific business types would be advisable J. Lopes; AEIC Load Research Conference St. Louis, MO; July 2004 34

NOTES J. Lopes; AEIC Load Research Conference St. Louis, MO; July 2004 35

Notes J. Lopes; AEIC Load Research Conference St. Louis, MO; July 2004 36