An Empirical Study of Electric Power Demand Control by Real-time Feedback of Consumption Levels: Case of Nushima Island Households Academic research paper on "Economics and business"

Available online at www.sciencedirect.com

ScienceDirect

Procedía Technology 18 (2014) 53 - 57

International workshop on Innovations in Information and Communication Science and Technology, IICST 2014, 3-5 September 2014, Warsaw, Poland

An Empirical Study of Electric Power Demand Control by RealTime Feedback of Consumption Levels: Case of Nushima Island

households

Koji Shimadaa*, Yuki Ochib, Takuya Matsumotoc, Hiroshi Matsugid and Takao Awatae

aRitsumeikan University, 1-1-1, Nojihigashi, Kusatsu, Shiga, 525-8577, Japan bE-konzal, #304 Maison Toan, 541, Asakura-cho, Nakagyo-ku, Kyoto, 604-8074, Japan cGraduate School of System Informatics, Kobe University, 1-1 Rokkodai-cho, Nada-ku, Kobe, 657-8501, Japan dHyogo Prefectural Institute of Technology, 3-1-12, Yukihira-cho, Suma-ku, Kobe, 654-0037, Japan eKei Communication Technology Inc., 8F Shinko Building, 8, Kaigandori, Chuo-ku, Kobe, 650-0024, Japan

Abstract

Electric power demand management will play an important role in the creation of smart-energy communities. We are conducting

a field experiment on the real-time feedback of electric power consumption via smart meters and tablet PCs with the participation

of 51 households on Nushima Island, one of Japan's remote islands. Our estimate of the effect of feedback in reducing power

demand by panel data analysis has revealed that such feedback achieves a saving of 7.6 percent in electric power demand.

© 2014 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license

(http://creativecommons.Org/licenses/by-nc-nd/3.0/).

Peer-review under responsibility of the Scientific Committee of IICST 2014

Keywords: electric power consumption; real-time feedback; field experiment, demand management, panel data analysis

1. Introduction

In order to create smart-energy communities that are climate-friendly and resilient to disasters, energy demand management through feedback of power consumption will play a significant role.

* Corresponding author. Tel.: +81-77-561-5183; fax: +81-77-561-3947 E-mail address: shimada@ec.ritsumei.ac.jp

2212-0173 © 2014 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license

(http://creativecommons.Org/licenses/by-nc-nd/3.0/).

Peer-review under responsibility of the Scientific Committee of IICST 2014

doi: 10.1016/j .protcy.2014.11.012

Nomenclature

CDH CF

Number of air conditioners [units] Constant

Cooling degree-hours [degree-hours] Number of commercial freezers [units]

Dummy variable for households where all energy is supplied by electric power Dummy variable for use of non-electrical heating equipment such as a kerosene heater Dummy variable for living in a timber house Dummy variable for summer vacation season

Dummy variable for feedback period (X: pattern of feedback, Y: month)

Daily electric power consumption [WMlay]

Heating degree-hours [degree-hours]

Number of household members [persons]

Number of refrigerators [units]

Daily frequency of viewing electric power consumption on the tablet PC [times/day]

Daily mean wind speed [m/s]

Partial regression coefficient

Individual effect of each household

ID number of each household

DUME DUMK DUMT DUMV DUMX Y

Faruqui et al. [1] reviewed a dozen utility pilot programs that focused on either the energy conservation impact of in-home displays (IHDs) or demand-side management technologies. The study showed that the power demand of consumers who actively used an IHD decreased on average by about 7 percent when prepayment for electric power was not involved. Houde et al. [2] found that access to real-time feedback led to an average reduction in household electric power consumption of 5.7 percent and that significant declines continued for up to four weeks.

On the other hand, real-time information feedback through technology has been reported to produce reductions of up to 20 percent in residential energy consumption [1], with the level of reduction depending on the experimental conditions. In this context, we have been focusing our attention on the development of an effective feedback method for the management of power demand that is suited to regional conditions. We have been conducting a field experiment for this purpose on Nushima Island, located in Hyogo Prefecture, Japan. This paper describes the design of the experiment and reports the findings obtained from an empirical analysis.

2. Methodology

2.1. Outline of the experiment

This experimental study being conducted on Nushima Island is a three-year project that commenced in 2012. In 2012, smart meters were installed in the residences of 51 households. In May 2013, tablet PCs were distributed to the participants that presented feedback on their electric power consumption. The contents of the feedback are rotated monthly from Pattern 1 to Pattern 3. As shown in Fig. 1, each household can view its electric power consumption and per-capita consumption in real time via the tablet PC in Pattern 1. They are also able to compare their consumption with the average consumption of participating households in Pattern 2. Moreover, an electric power consumption ranking among the participating households is displayed in Pattern 3. Table 1 shows the schedule of the feedback from Pattern 1 to Pattern 3 in 2013. We are assessing the impacts of the feedback on the actual power demand by using data collected via smart meters.

Only in Patterns 2 and 3

Only in Pattern 3

Electric power consumptions in your home

(W/person) (Wh/person) Yesterday (Wh/person) This month (Wh/person)

450 3,000 7,500 27,000

Per-capita electric power consumption

(W/person) (Wh/person) Yesterday (Wh/person) This month (Wh/person)

Your house 150 1,000 2,500 9,000

Average of 3,625 10,756

Lowest electric power consumption ranking

Your rank: 14th (yesterday)

Fig. 1. Information displayed on a tablet PC.

Table 1. Schedule of feedback.

Pattern

Jun. o

Jul. o

Aug. o

Sep. o

Oct. o

Nov. o

Dec. o

2.2. Analysis method

In this study, we are implementing panel data analysis to assess the effect of feedback on the management of electric power demand. Power demand depends on a variety of factors, including temperature, household size, and types of electrical appliances used. Hence, it is significant to estimate the impact of feedback separately from other factors that may have an influence on electric power consumption taking the heterogeneity of the participating households into consideration.

We analyzed the daily electric power consumption of the 51 households from May to December 2013. Equation (1) is the estimating equation used to determine the impact of feedback on daily power demand. The equation defines daily electric power consumption in terms of an explained variable and three types of factors, namely, factors related to the external environment, internal environment, and feedback, as explanatory variables. The formula was originally developed for this study based on the authors' investigations.

First, daily electric power consumption values were observed from the smart meters. The external environmental factors were then obtained; namely, the cooling degree-hours, heating degree-hours, and daily mean wind speed. These data were collected from the Climate Statistics provided by Japan Meteorological Agency [3]. The cooling/heating degree-hours describe to what extent (in degrees) and for how long (in hours) the outside air

temperature is higher/lower than a specific base temperature. The base temperature for cooling degree-hours was set at 24°C and that for heating degree-hours was set at 18°C.

The internal factors were composed of the number of household members, air conditioners, refrigerators, and commercial freezers for fishery; whether all of the energy in a household was supplied by electricity; whether nonelectrical heating equipment was used; whether the household was living in a timber house; and whether the targeted period was the summer vacation season. Finally, the factors related to feedback consisted of whether feedback was introduced and the frequency of viewing electric power consumption on the tablet PC.

From the results of a Hausman test, we adopted a random effect model in which an individual effect was treated as a stochastic variable. The number of samples obtained excluding missing values amounted to 12,011, which were collected from the 51 households over a period of 275 days from April 1 to December 31, 2013.

= C + a HHi4 + a2 HHi4 2 + a3CDHid + a4CDHi, d + a5 HDHi d + a6 HDHid + a7WSij + a8 ACid + a9 RFU d + a10CFw + anDUMEhd + a DUMKi4

+ a23 ( x DUM 1_ 5id ) + a24 (( x DUM1 _ 6, d ) ♦a,(( xDUM 1_9id)(( xDUM1_12i4) + a27((Cid xDUM2_7w)(vC^ xDUM2_10 J + a29 (d x DUM3 _ 8i4 ) + a30 (vCi4 x DUM3 _ 11i4 )

+ a13DUMT d +a14DUMVi d + a15DUM1_5i d + a16DUM 1_6i d

+ a17 DUM1_9i d + a18 DUM1_12 i d + a19 DUM 2_7 i d + a20 DUM 2_10 i d

3. Results and Discussion

The estimation results obtained from the daily panel data analysis are shown in Tab. 2. The coefficients of DUM15, DUM1_6, DUM1_9, DUM2_10, DUM3_8, and DUM3_11 are statistically significant at the 1 percent level. This indicates that the feedback had the effect of reducing electric power demand, except in August and December. The effect is estimated as a 7.6 percent reduction compared with the mean daily consumption of the participating households in May, June, September, October, and November 2013.

Table 2. Results of daily data analysis.

Variable Coefficient t value Variable Coefficient t value

C -6.515X104 ** -2.061 DUM1_5 -1,048 *** -5.758

HHin 7.401 x104 ** 1.994 DUM1_9 -1,126 *** -6.395 -5.220

CDH2 0.2177 *** 6.764 DUM2_7 -265.8 1.100 -1.065

HDH1 5.696x10-2 *** 8.378 DUM3_8 1,981 *** -8.586 7.559

AC 302.2 0.1308 VCxDUM1_5 29.34 -3.097 -0.5105

CF 1.005x104 *** 3.889 VCxDUM1_9 -48.44 1.107 -0.7291

DUMK -1,495 *** -10.23 VCxDUM2_7 -66.46 *** 8.876 -3.198

DUMV 2,184 *** 9.240 VCxDUM3_8 VCxDUM3 11 -192.8 364.6 *** *** 1.009 -2.993 4.904

Adjusted R-s quared | 0.3608 Durbin-Watson statistic 0.7139

*, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

Additionally, the coefficients of (VC*DUM1_12), (VC*DUM2_7), (VC*DUM3_8), and (VC*DUM3_11) are also statistically significant at the 1 percent level. These results indicate that viewing frequency has an effect on electric power demand in both directions: a decreasing effect in July and August and an increasing effect in November and December. The differing effects may be related to consumers' perception gap between subjective savings and real savings, as reported by Attari et al. [4].

4. Conclusion

We found that real-time information feedback reduced electric power consumption by 7.6 percent on average, although the effects included both positive and negative factors in terms of energy savings. A further experimental study is to be implemented in the summer and winter of 2014 in order to estimate the additional effect of dynamic pricing.

Acknowledgements

This study was supported by the technological development and demonstration research project for global warming countermeasures of the Ministry of the Environment, Japan.

References

[1] Faruqui A, Sergici S, Sharif A. The Impact of Informational Feedback on Energy Consumption - A Survey of the Experimental Evidence. Energy 2010; 35(4): 1598-1608.

[2] Houde S, Todd A, Sudarshan A, Flora JA, Armet KC. Real-time Feedback and Electricity Consumption: A Field Experiment Assessing the Potential for Savings and Persistence. The Energy Journal 2013; 34(1): 87-102.

[3] Japan Meteorological Agency. Climate Statistics. Retrieved June 17, 2014, from the JMA website: http://www.data.jma.go.jp/obd/stats/etrn/index.php?prec_no=63&block_no=1337&year=&month=&day=&view=

[4] Attari S, Dekay ML, Davidson CI, de Bruin WB. Public perception of energy consumption and savings. Proceedings of the National Academy of Science of the United States of America 2010; 107(37): 16054-16059.