Baltimore Gas & Electric Co. has had a Long and Successful History Running Demand-Side Management Programs. In particular, BGE's direct load-control program for residential and commercial customers began in the late 1980s and still exists today.

BGE (Baltimore, Maryland, U.S.) currently pays seasonal credits to more than 218,000 residential customers to cycle their air conditioners during high peak demand periods in the summer. In addition, BGE has a water-heater cycling program for residential customers with approximately 72,000 participants and a commercial air-conditioning switch program with more than 5500 participants. Combined, these programs provide more than 210 MW of peak load relief for the central Maryland area.


The transition to higher market-based generation costs over the last several years in the area reinforced the need for a renewed focus on energy-saving programs. In addition, the Energy Policy Act of 2005 and the U.S. Department of Energy's and the Federal Energy Regulatory Commission's reports during 2006 provided a favorable national perspective on demand response. With demand-side management programs regaining momentum, BGE decided to help solve some of the most urgent energy needs of its community. For BGE, the benefits of a demand-response infrastructure (DRI) program include improved system reliability, reduced cost of service, less infrastructure, increased customer satisfaction and reduced emissions.

So, after many years of use and with BGE's legacy demand-response programs nearing the end of their life cycles, BGE conducted a DRI pilot program during the summer of 2007. The first priority was to test several assumptions and design elements regarding a next-generation demand-response program, including customer acceptance, load impacts, temperature impacts, technology preference and program benefits. Two residential cycling devices were deployed to approximately 1000 eligible residential customers in the Baltimore area. Customers with one electric air conditioner or heat pump and living in a single-family home or townhouse were eligible to participate.

BGE tested customer acceptance of two new technologies — an advanced load-control switch (Smart Switch) and an advanced programmable communicating thermostat (Smart Thermostat) — as well as various cycling strategies. After the customer agreed to participate, BGE installed either a Smart Thermostat in the customer's home or a Smart Switch on or near the outside compressor unit of the customer's air-conditioning system. Each technology enables cycling (turning the air conditioner off and on at prescribed intervals) that is activated by BGE and transmitted from its private paging VHF radio frequency.


During the pilot, three cycling strategies were tested: 30%, 50% and 75%. These strategies controlled the amount of time an air conditioner was allowed to run during each load-control event.

BGE's legacy program operates at a 50% cycling strategy that turns the air conditioner off and on at 15-minute intervals, regardless of how long the unit normally runs. The pilot program system uses an advanced algorithm that sends a paging signal to the devices to memorize the amount of time the air conditioner typically runs every hour. In essence, a run-time curve is developed for each device that determines how long each customer's air conditioner will be controlled during each hour of a control event. For example, if BGE cycled at 50% and the customer's air conditioner normally runs for 50 minutes and is off the other 10 minutes in a specific hour, then the air conditioner was allowed to run for 25 minutes and was cycled off for 35 minutes during that hour of the event.

In addition, customers who chose a thermostat had the capability of remotely setting their thermostat via the Internet. Internet-programming functionality included heating, cooling and temporary (vacation hold) setting options. Also, both switch and thermostat customers could override two events during the pilot by calling a customer service number provided in the program literature or through their Internet account. The override limit was tested to gauge customer reactions to this feature if BGE were to allow it for control events during nonemergency periods.


In order for BGE to estimate the impact of DRI on hourly demand, a subset of pilot accounts also was simultaneously recruited for sample meter installations. Approximately 200 accounts were recruited for the installation of end-use and whole-house meters to collect load data. The end-use meters measured air-conditioner compressor load in 5-minute intervals. The whole-house meters were able to measure load on the entire house in 15-minute intervals.

For the development of average impacts for DRI, nonevent days are compared to event days using linear regression analysis. Reductions are defined as the difference between these two-day types. Average reduction estimates are modeled by weather conditions and cycling strategy.

The table provides reduction estimates by cycling strategy for BGE's 2007 system peak day. The weighted temperature humidity index (WTHI) for each hour is shown in the table. The 2007 BGE system peak occurred at the hour ending 5 p.m. on Aug. 8. The estimated reduction in load at this time was 1.22 kW per device under a 50% cycling strategy. Per-device load reductions for DRI participants ranged from 0.76 kW at the hour ending 7 p.m. (30%) to 2.24 kW at the hour ending 3 p.m. (100%). Note that 100% cycling estimates were implied based on total average load measured at the compressor by the end-use metering.

The load-impact estimates presented in the table were derived from end-use metering. BGE also measured load-impact estimates from whole-house meters. There was no statistically significant difference found between these two load-reduction estimates for the hours ending 4 p.m. through 7 p.m.


To measure changes in indoor air temperature and humidity during cycling events, data loggers were installed in 104 homes of DRI pilot participants. These loggers captured temperature and humidity in the homes at 15-minute intervals. At the end of the DRI pilot, customers were asked to mail the data loggers back to BGE. Of the 98 returned, 89 were from homes that had unambiguous evidence of event impacts, 36 were from homes with switches and 53 were from homes with thermostats.

To develop average changes in temperature and relative humidity, nonevent-day changes were compared to event-day changes. Nonevent days were used to temper event-day temperature changes. Average indoor temperature and relative humidity profiles were developed for each account for the nonevent days of July 9, Aug. 1 and Aug. 7. For each event day, the indoor temperature and relative humidity readings were captured at the start and end of an event. Average changes in event-day indoor temperature and relative humidity were then developed.

When reviewing all event days by cycling strategy, duration and outside temperature, it was clear that events impact the temperature and humidity levels in the home. Events with 30% cycling had an average increase in temperature of 0.9°F (0.5°C), while 50% and 75% cycling saw average increases of 1.8°F (1.0°C) and 1.9°F (1.06°C), respectively. For 50% strategies, the 4-hour events revealed an average temperature increase of 0.96°F (0.53°C), while the 5-hour events revealed an average temperature increase of 2.2°F (1.22°C). Even with these averages around 2°F (1.11°C), there were some customers who experienced very large changes in indoor temperature. Records show that 27 customers had increases in temperature of over 4.5°F (2.5°C) on specific event days, which translates into 4.5% of the observations.

The chart on page 41 compares indoor temperature changes during events. The first bar depicts all of the event days and shows that 76% of the events resulted in temperature increases of 2°F (1.11°C) or less. The second bar depicts the events that occurred using a 50% or 75% cycling strategy. This bar shows that 71% of the events resulted in temperature increases of 2°F (1.11°C) or less. It is evident from the chart that when 30% of events are removed, indoor temperatures increase from the application of higher cycling strategies. On the hottest day of the year, Aug. 8 (maximum WTHI of 86.4°F [30.2°C]), 52% of the participants experienced a 2°F (1.11°C) or less temperature increase, while 17% experienced an increase of 5°F (2.78°C) or more. This reveals that, although the average increase is around 2°F (1.11°C), cycling strategy and outdoor temperature play key roles in how much the indoor temperature rises.

No statistically significant difference was found in temperature changes between customers with switches and customers with thermostats. Both showed similar increases. However, relative humidity increases for the customers with switches were statistically larger than for those with thermostats. This may be attributed to the fact that fans automatically run during an event for those customers with switches, while customers with thermostats must turn the fan on for it to run during an event. The running of the fan during an event can cause unconditioned humid air to circulate throughout the home, thus raising the relative humidity level in the home for the customers with switches.


Overall, customer acceptance of the DRI pilot program was overwhelmingly positive. Market research showed customers who were more familiar with switch technology preferred the installation of a Smart Switch (BGE legacy program participants). New program customers had an equal preference between choosing the Smart Thermostat or the Smart Switch.

At the completion of the DRI pilot, BGE surveyed 386 of the 1000 participants. Pilot participants were overwhelmingly motivated by bill savings, with 73% ranking this the most important reason for participating, followed by conservation at 20%. Other findings included:

  • Comfort issues due to cycling were not a major concern

  • The vast majority of participants were satisfied with their pilot experience and felt the program met their expectations

  • The likelihood of future participation was strong

  • Future technology preference was slightly biased toward the Smart Thermostat

  • Participants were amenable to varying bill credits for varying degrees of control.

BGE viewed the results of the DRI pilot as extremely encouraging to promote a similar program on a larger scale. BGE filed with the Maryland Public Service Commission to deploy the DRI program to all eligible BGE residential customers. The filing was approved, and BGE will be rolling out DRI to central Maryland during 2008. It's the DRI project team's goal that the program will serve as a model for similar demand-response programs throughout the country.

David Greenberg is principal electric supply analyst with Baltimore Gas & Electric's Supply & Forecasting division.

Mary Straub is principal load analyst with Baltimore Gas & Electric's Customer Load & Settlement division.

Estimated premise level impacts by cycling strategy
Hour ending Weighted temperature humidity index 30% (kW) 50% (kW) 75% (kW) 100% (kW)
1 p.m. 84.2°F 0.92 1.15 1.43 1.71
2 p.m 84.9°F 0.75 1.06 1.43 1.81
3 p.m. 85.9°F 1.14 1.45 1.85 2.24
4 p.m. 85.7°F 0.99 1.32 1.73 2.13
5 p.m. 85.3°F 0.84 1.22 1.69 2.16
6 p.m. 84.8°F 0.79 1.16 1.63 2.09
7 p.m. 83.4°F 0.76 1.1 1.53 1.96