Abstract
This paper evaluates the effectiveness of combining direct load control with a residential zoned-cooling technology in meeting the objectives of reducing peak demand and maintaining home comfort level. In contrast, the traditional approach has been for utilities to smooth summer peak cooling loads, by controlling the cooling load of the whole house. While accounting for weather, dwelling characteristics and demographics, with detailed field data, we are able to develop empirical models to evaluate the benefits of utility control of cooling loads for a residential zoned cooling system during summer peak-demand periods and to compare with non-zoned systems. A zoned house allows for an upper floor cooling interruption without affecting the comfort on the main floor. An upper floor interruption for a full 4 h during the day leads to an average peak air conditioning change of −0.52 kW, approximately 1.6 times the reduction from the curtailment of cooling by cycling the air conditioning serving the whole house.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
In parts of North America, there are examples of non-forced air systems delivering cooling and heating requirements, such as window air conditioners and baseboard heating (e.g., in Quebec). However, the focus of this paper is on combining direct load control with zoned cooling delivered by a forced air system.
See Mountain et al. (2010) for a more comprehensive discussion of the delivery of heating to multiple zones in the home in the context of a zoned system.
The incremental cost of a zoned versus a non-zoned system can be broken into two components, the HVAC equipment and the ducts. For the equipment, the incremental cost is about $1,600 above a standard $2,150 cost of a high efficiency furnace. The ductwork (with no duct seals) of a standard non-zoned system is about $3,098 versus $3,372 for a (duct seal (worth about $250)) a zoned system. Holding the level of duct sealing constant for both systems, the incremental cost of ductwork is $24 for a zoned system.
A ‘hot day’ was defined as one with a humidex value of 28 °C or higher in July and August, and 24 °C or higher in September.
For Kitchener-Wilmot, the weather station is located at the Region of Waterloo International Airport and for Chatham-Kent the weather station is at Windsor Airport.
While the models are estimated separately, we recognize by way of some common explanatory variables in all three models of the interrelationship of these three components in the assessment of the impact of combining utility-controlled demand response with zoning.
The formula for the humidex is presented in "Field data description". Because of the lags in responsiveness of an insulated house to the external temperature and humidity, an average humidex over the previous 8 h is used as an explanatory variable. We also tried averages over the previous 12 and 24 h, but found the 8-h average to provide the best fit.
In treatments 1, 2, and 3, the main floor air conditioning is curtailed over night (MN it = 1 for t representing overnight hours). For treatments 1 and 2, the main floor air conditioning is interrupted for some hours during the day (MD it = 1). For treatments 1, 2, and 3 there are hours during the day that the upper floor is interrupted (U it = 1). There are three possible combinations indicating the interruption of the main and the upper floor air conditioning during the day. The interaction terms U it (1 − MD it ), MD it (1 − U it ), and MD it U it discriminate between interruption of only the upper floor, only the main floor, and the simultaneous interruption of both floors during the day. exp{−UD it }, exp{−MDD it } and exp{−MND it } refer to the corresponding exponential duration (in hours) of interruption.
The various combinations of interruption in zoned households are consistent with the air conditioning condenser model.
From the regression (Eq. 3.1), the average R2 for weekdays is 71.0 % (72.2 % for July and August; 69.7 % for May, June, and September), and for weekends is 72.1 %. For the regression (Eq. 3.4), the average R2 for weekdays is 81.8 %, and for weekend is 70.9 %. This is a good fit and the model seems to be well specified. For the regression (Eq. 3.8), the average R2 for weekdays of July and August is 89.4 %.
In the first stage air conditioning condenser weekday model estimation, for non-zoned houses, the hypothesis that all the interruption coefficients are jointly zero was rejected for 3 out of 11 models. For zoned houses, for only 3 out of 18 models were the interruption coefficients jointly zero and all three were for September.
Limited by the length of the paper, we will not report the original estimated coefficients (before calculating the splined coefficients). Because not enough weekend days in May, June, and September had days with positive CDHM, we were not able to reliably estimate the parameters of the first stage air conditioning condenser model for these boundary months. The estimated parameters for the July and August model during weekend days and for the May, June, and September model during weekdays are not displayed here but are included in Mountain et al. (2011).
In Figure 4, we make an assumption that \( \frac{\overline{{\left. DOC U\right|}_{\left( z\kern0.5em =\kern0.5em 1\right) t}}}{\overline{{\left. DOC\right|}_{\left( z\kern0.5em =\kern0.5em 1\right), t}}}\kern0.5em =\kern0.5em 1 \) during the night (from 1900 to 0700 h) and during the day (from 0800 to 1800 h) the value is 0.5.
After dropping the coefficients which were statistically insignificant, the utility interruption terms in the final models which go to the second stage air condenser β kim model as dependent variables are as follows:
$$ \begin{array}{l}{\beta}_{1 im}{U}_{it}\cdot \left(1- M{D}_{it}\right)\cdot CDH{M}_{it m}\cdot ZC,{\beta}_{2 im}{U}_{it}\cdot \left(1- M{D}_{it}\right)\cdot \exp \left\{- U{D}_{it}\right\}\cdot CDH{M}_{it m}\cdot ZC;\hfill \\ {}{\beta}_{5 im} MN\cdot CDH{M}_{it m}\cdot ZC;{\beta}_{7 im}{U}_{it}\cdot M{D}_{it}\cdot CDH{M}_{it m}\cdot ZC;{\beta}_{9 im} A{U}_{it}\cdot \exp \left\{- AU{D}_{it}\right\}\cdot CDH{M}_{it m}\cdot ZC;\hfill \\ {}{\beta}_{10 im} A{U}_{it}\cdot \exp \left\{- AMD{D}_{it}\right\}\cdot CDH{M}_{it m}\cdot ZC;\mathrm{and}\kern1em {\beta}_{12 im}{H}_{it},{\beta}_{14 im} A{H}_{it}\cdot \exp \left\{- AH{D}_{it}\right\}\cdot CDH{M}_{it m}\cdot ZC.\hfill \end{array} $$The other two delta models for weekend days in July and August as well as weekdays in May, June, and September are separately estimated similarly. Both of them have good fits and most of the coefficients seem to have significant impacts. Similar to the air conditioning second stage model, they are not shown here.
The normalized house is a house with the mean characteristics of the average sample house for this study.
After dropping the coefficients which were statistically insignificant, the utility interruption variables in final models which go to the second stage comfort τ kim model as dependent variables are as follows:
$$ \begin{array}{l}{\tau}_{1 im}\left({U}_{it}+ A{U}_{it}\right)\cdot \left(1- M{D}_{it}\right)\cdot CDH{M}_{it m}\cdot ZC,{\tau}_{2 im}\left( M{D}_{it}+ A M{D}_{it}\right)\cdot \left(1-{U}_{it}\right)\cdot CDH{M}_{it m}\cdot ZC,\hfill \\ {}{\tau}_{3 im}\left( M{N}_{it}+ A M{N}_{it}\right)\cdot CDH{M}_{it m}\cdot ZC,{\tau}_{5 im} A{U}_{it}\cdot AU{D}_{it}\cdot CDH{M}_{it m}\cdot ZC,{\tau}_{6 im} A M{D}_{it}\cdot AMD{D}_{it}\cdot CDH{M}_{it m}\cdot ZC.\hfill \end{array} $$Tables for non-interruption days of boundary months (May, June and September) are not shown here. The air conditioning loads show increase and the fan loads show significant reductions. Overall the reductions in the total of the air conditioning and fan loads are moderate relative to July and August. For the weekend loads in July and August, the air conditioning load shows significant reductions in August (19.0 %) and including the fan load, the total reduction in August is 31.0 %. For July, the overall load reduction is 11.9 %.
For the three treatments for zoned houses, the results continue to be reported for the same normalized house as discussed for non-interruption days. However, we normalize the houses to specific occupancy and indoor temperature settings corresponding to the respective treatment because the original treatments were assigned to specific households with particular lifestyles with respect to work and occupancy patterns.
Interruptions which occurred in September cause less saving, maybe because September tends to be cooler than July and August, and residents may already be doing their own load control which leaves little room for utility interruption. The impacts of interruption for September tended to be negligible for treatments 1 and 2.
Past Ontario studies examining the impacts of whole house AC interruption show reductions from 0.1 to 0.8 kW (for a temperature range of 28 to 33 °C) for a 15 min: 45 min on/off cycling (Mountain 2007) and reductions from 0.2 to 0.9 kW, where the thermostat is set up by 2 °C for a 4-hour period (Newsham et al. 2011).
An examination of the weather data shows the hottest day in July, 2010 had an average reading of five degrees above the average CDHM of 10.72 during the interruption days.
The shapes across different treatments are similar, so here we show treatments 1 and 2.
The shapes across different treatments are similar, so here we show treatments 1 and 2.
Both treatments share the characteristics of our average house described earlier, including the weather of an average interruption day in July and August, with the exception of occupancy and indoor temperature settings which correspond to the average of treatments 1, 2, and 3.
References
American Meteorological Society (2014). Meteorology Glossary, http://amsglossary.allenpress.com/glossary/.
Ericson, T. (2009). Direct load control of residential water heaters. Energy Policy, 37, 3502–3512.
Hamlet, A. F., Lee, S.-Y., Mickelson, E. B., & Elsner, M. M. (2010). Effects of projected climate change on energy supply and demand in the Pacific Northwest and Washington State. Climatic Change, 102(1–2), 103–128.
Hsiao, C., Chan, M. W. L., Mountain, D. C., & Tsui, K. Y. (1987). An integrated monthly and hourly regional electricity model for Ontario, Canada. Resources and Energy, 9, 275–299.
Kirby, B. (2003). Spinning reserves from responsive loads, January. Oak Ridge: Oak Ridge National Laboratory.
Miller, N., Hayhoe, K., Jin, J., & Auffhammer, M. (2008). Climate, extreme heat, and electricity demand in California. Journal of Applied Meteorology and Climatology, 47, 1834–1844.
Mountain, D.C. (2007). Residential Load Control: The Hydro One Pilot, March, Hydro One: Toronto, Canada
Mountain, D. C., Strack, T., Sager, J. (2010). Advanced Residential Load Reduction Pilot Project, March, Natural Resources Canada: Ottawa, Canada.
Mountain, D. C., Strack, T., Zhou, W., Lomanowski, B. (2011). The Zone-Saver Field Trial: Utility Controlled Demand Response with Residential Zoned Cooling, December, Natural Resources Canada and Ontario Power Authority: Ottawa and Toronto, Canada.
Newsham, G. R., Birt, B. J., & Rowlands, I. H. (2011). Comparison of four methods to evaluate the effect of a utility residential air-conditioner load control program on peak electricity use. Energy Policy, 39(10), 6376–6389.
Newsham, G. R., & Bowker, B. G. (2010). The effect of utility time-varying pricing and load control strategies on residential summer peak electricity use: a review. Energy Policy, 38(7), 3289–3296.
Platt, H. D., Einhorn, M. A., Ignelzi, P. C., & Poirier, D. J. (1981). Regional load curve models: Specification and estimation of the DRI Model (Vol. 1). Palo Alto: Electric Power Research Institute.
Suits, D. B., Mason, A., & Chan, L. (1978). Spline functions fitted by standard regression methods. Review of Economics and Statistics, 60, 132–139.
Acknowledgment
Research funding was provided by Natural Resources Canada, Ontario Power Authority and Social Sciences and Humanities Research Council of Canada. The analysis in this paper owes much to Terry Strack who oversaw the fieldwork portion of the pilot and to Jeremy Sager and Dan Carr who provided advice throughout the duration of the research. Thanks are also extended to three anonymous reviewers for their very helpful suggestions and comments.
Author information
Authors and Affiliations
Corresponding author
Appendix
Appendix
Rights and permissions
About this article
Cite this article
Zhou, W., Mountain, D.C. The benefits of combining utility-controlled demand response with residential zoned cooling. Energy Efficiency 7, 1067–1099 (2014). https://doi.org/10.1007/s12053-014-9265-7
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s12053-014-9265-7