An analysis of thermostat cycling vs. temperature offset
by John Rossi, Comverge Inc.
Today’s power system is plagued by demand-side and supply-side problems. While these problems have been well-documented, solutions are often oversimplified and in some cases not understood.
An example is peak-demand control, a proven tool utilities have used for many years to address energy management challenges and improve power system reliability. Peak-demand control involves reducing system load during high usage. This reduces supply costs, which are particularly high during peak usage, and increases reliability by preventing any possible system overload.
Peak-demand control programs for residential customers typically use switches or programmable controllable thermostats (PCTs) to control the energy consumed by heating, ventilation and air conditioning (HVAC) systems.
To help determine how PCTs can deliver the optimum balance between the amount of load shed and customers’ comfort, we must evaluate the two ways they can be used in residential demand control programs.
The first, temperature offset, uses a temperature offset to remotely raise the current set point during the period a utility initiates its demand response program to reduce load. This period also is called a control event. While this simplifies temperature control and optimizes load shed for short periods, it maximizes customer discomfort and provides a highly variable load for the utility to manage.
The offset amounts to 100 percent off cycling strategy of the compressor plus curtailment of the circulating fan until the offset temperature is reached. The circulating fan will be off when the compressor is off if the thermostat is set in auto mode. The fan would remain on if the fan is set on at the thermostat. The auto mode is the most widely observed method.
The second method, thermostat cycling, involves sending a command that cycles the HVAC system to lower the duty cycle and reduce energy consumption during a control event. In contrast to temperature offset, thermostat cycling curtails the compressor and circulating fan if required for a period and allows both to run for a second period. If the cycling strategy includes an adaptive algorithm to handle oversized compressors, thermostat cycling can improve customer comfort and load shape for control events longer than several hours.
To demonstrate how thermostat cycling and temperature offset impact load control and customer comfort, we conducted a controlled study within a residential peak-demand control program in a hot, southwestern climate.
Temperature offset raises the thermostat set point by a fixed amount, typically 4 degrees. The compressor and circulating fan are subsequently switched off until the new set point is reached. At that time, the thermostat returns to normal operation at the new, higher temperature.
To determine how temperature offset impacts load shed performance and customer comfort, we examined many homes at a temperature equal to the average summer peak temperature for three control events, each lasting three hours. The data from each site includes the interior temperature at the end of each hour and duration of compressor run time during that hour.
For these events, the thermostat set point was raised 4 degrees at 3 p.m. As the thermostat settings were raised, the compressor was held off until the interior temperature reached the higher set point. The dashed blue line in Figure 1 shows homes that rose 4 degrees within the first hour of control. Because compressors in these homes came on during the first hour of the event, it can be determined that the rate of rise was at least 4 degrees per hour. It was impossible to calculate the exact rate of temperature rise, however, because readings were available only at the end of each hour.
This was repeated for homes where the compressor did not come on until the second hour. For homes where the compressor came on during the first hour of the event, the average temperature rise was 2.85 degrees per hour. We look at the rise in the first hour because this data is unaffected by compressor run time and it gives us an unaffected end temperature.
The homes where the compressor did not come on until the third hour experienced a rise of 1.975 degrees per hour. For some homes, the compressor did not come on at all during the event. These homes showed a temperature rise of 1.21 degrees per hour.
The data shows when different groups would reach the 4-degree offset value where the air conditioning would turn on.
To demonstrate what this means for today’s power system, Figure 2 shows how the population can be divided into these groups.
Most residences–61.2 percent–rose 4 degrees within the first hour of a control event, and then remained at that new set point for the event’s remainder. About 10 percent of compressors came on in the next hour, and an additional 6 percent started in the third hour. The remaining 23.9 percent of the population had compressors that did not run in the hour prior to control and were eliminated from this analysis (estimated at 14 percent of the population) or did not experience a significant enough temperature rise for the compressor to be activated during the three-hour test. While the rate of rise might be less when peak temperatures are not reached, the impact on the load would be the same with a large, initial load drop followed by a small load contribution after the new offset period is reached.
Cycling air conditioner compressors to achieve demand response is well-established and the standard method for compressor-mounted load control switches.
Thermostat controls, however, deliver the flexibility to offer different operating modes (e.g., varying the programmed operation as a function of price and displaying the price on the thermostat), direct programmability to respond to price signals and the potential for controlling the HVAC system circulation fan. These capabilities, combined with the potential to operate in a mode that cycles the compressor and circulating fan off for selected intervals in response to a load control initiation signal, give thermostat controls significant advantages over compressor-mounted, switch-based programs.
Specifically, in relation to temperature offset, this mode offers the following advantages:
- Increased Customer Comfort. Running the system periodically is more comfortable for occupants because it decreases the rate of temperature rise.
- Reduced Humidity. Because the system is run periodically, humidity does not increase as much as with temperature offset.
- Predictable, Evenly Distributed Load Shed. The load shed provided by this control method is predictable and more evenly distributed over the control interval when compared with temperature offset, which is weighted for the initial period after control.
- Accurate Forecasts. With near real-time data from a monitoring and verification system, the hourly load sheds for temperature cycling can be forecasted more accurately than with temperature offset.
Cycling With Adaptive Algorithm
An adaptive algorithm allows cycling modifications based on historical run time data and is more effective than applying a simple, time-based algorithm for cycling (e.g., turn the HVAC off for 30 minutes and let it run 30 minutes).
To accurately modify cycling times, an adaptive algorithm learns the compressor run time pattern and then compensates for it during a control event. This provides equitable load shed contribution from all participants for oversized compressors and operations at temperatures below the maximum design temperature in an area. In both of these real-world scenarios, compressors will not run constantly, and with a simple, 50 percent control strategy, the compressor will be held off for 30 minutes an hour. If the compressor can maintain temperature during that hour by running less than 30 minutes, the compressor will not contribute any load to the control event, and the temperature will not rise in the home.
With an adaptive algorithm, the switch or thermostat monitors prior run time and compensates if the run time is less than the full hour. So if a compressor only runs 30 minutes in an hour to maintain temperature, the algorithm for 50 percent cycling will take half of that run time and allow the compressor to run only 15 minutes per hour during the control event.
The problem with oversized compressors at a given temperature is illustrated in Figure 3, which shows data on compressor run time for an adaptive algorithm compared with a simple, time-based solution.
This figure shows that until the compressor runs 50 percent of the time, a nonadaptive solution provides no kilowatt reduction, whereas the adaptive algorithm solution provides immediate kilowatt savings.
Load Shed for Offset
Having evaluated some of the differences between temperature offset and thermostat cycling, we will examine the load shed for each during peak-demand control events.
Figure 4 shows the load shed for temperature offset over the three-hour events while Figure 5 shows a four-hour load drop for 50 percent adaptive cycling.
The drop is plotted relative to a baseline of the load in the hour prior to the control event. The load shed for the temperature cycling is uniform throughout the event. This is important when the load shed is required over an extended period because this flat load shape enables the load drop to be estimated accurately over a period if the load in the prior hour is known.
Stated another way, with temperature offset, the load drop for the population will vary by hour and the extent will be difficult to estimate because it depends on factors such as the ambient temperature and insulation level.
In contrast, with adaptive cycling, if the load in the prior hour is known (via monitoring and verification, for example), then the load shed for the event’s duration is not strongly dependent on temperature and remains nearly constant throughout the event.
Temperature offset provides a large initial load shed that decreases over time and is likely to maximize customer discomfort by rapidly increasing the average household’s interior temperature.
In contrast, thermostat cycling delivers a lower, more predictable load shed that improves customer comfort for a longer period. The data also indicates that harder cycling would be possible for this population at the temperature analyzed.
Thermostat cycling without an adaptive algorithm allows for a high percentage of free riders who are unaffected by peak-demand control programs and contribute little or no load.
John Rossi is senior vice president of business development with Comverge Inc.