ARTICLE IN PRESS

Energy 32 (2007) 16171633 www.elsevier.com/locate/energy

Evaluation of city-scale impact of residential energy conservation measures using the detailed end-use simulation model
Yoshiyuki Shimoda, Takahiro Asahi, Ayako Taniguchi, Minoru Mizuno
Division of Sustainable Energy and Environmental Engineering, Graduate School of Engineering, Osaka University, 2-1, Yamada-oka, Suita, Osaka 565-0871, Japan Received 6 October 2006

Abstract Energy conservation policies for the residential sector are evaluated by a model that simulates city-scale energy consumption in the residential sector by considering the diversity of household and building types. In this model, all the households in the city are classied into 380 categories based on the household and building type. The energy consumption for each household category is simulated by the dynamic energy simulation model, which includes an energy use schedule model and a heating and cooling load calculation model. Since the energy usage of each appliance is simulated for every 5 min according to the occupants energy usage activity, this model can evaluate not only the energy conservation measures by improving the buildings and appliances but also the measures that involve changing the occupants activities. The accuracy of the model is veried by comparing its results with the statistical and the measured data on Osaka City, Japan. Various types of energy conservation measures planned by the Japanese government for the residential sector are simulated and their effects on Osaka City are evaluated quantitatively. The future effects of these combined measures on the energy consumption are also predicted. r 2007 Elsevier Ltd. All rights reserved.
Keywords: City scale evaluation; Energy conservation measures; Household type distribution; Occupants behavior

1. Introduction In Japan, energy consumption in the residential sector has been increasing continuously due to improvements in the standard of living, such as enlargement of houses, popular use of various types of home electric appliances, and an increase in the number of small families. In the last 25 years, the energy consumption of the residential sector has doubled while the population has increased by only 10% [1]. To achieve the 6% greenhouse gas reduction commitment in keeping with the Kyoto Protocol, various kinds of measures were proposed for the residential sector in Japan. The New Climate Change Policy Program, which was adopted in March 2002, aims to maintain the CO2 emission from energy usage at its levels in 1990. However, CO2 emission from energy use in the residential sector had
Corresponding author. Fax: +81 668797665.

E-mail address: shimoda@see.eng.osaka-u.ac.jp (Y. Shimoda). 0360-5442/$ - see front matter r 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.energy.2007.01.007

increased from its level in 1990 by 28.8% in 2002 due to an increase in the number of household appliances. According to this increase, the Kyoto Protocol Target Achievement Plan, which was adopted in April 2005, aims to reduce CO2 emission from the residential sector such that it is 6% greater than its level in 1990 by strengthening the energy efciency measures for buildings and appliances. As a part of these programs, the revised version of the Law Concerning Rational Use of Energy established one of the highest energy efciency standardscommonly known as the Top-runner Standardin the world for home electric appliances. According to this standard, an appliance manufacturers average energy efciency in 2004 must be higher than that of the most efcient model in 1999. This standard was revised in 2005 to extend it to more types of appliances. Besides the top-runner standard, the New Climate Change Policy Program species various kinds of energy conservation measures such as increase in energy efcient residential buildings, reduction in standby power,

ARTICLE IN PRESS
1618 Y. Shimoda et al. / Energy 32 (2007) 16171633

promotion of high-efciency water heaters. The program also species the change in occupants behavior such as easing set room air temperature, family members staying together in the living room and not in their individual rooms, and reducing the number of hours spent watching TV. The effects of these measures are interactive with other measures. For example, energy efcient appliances and change in the occupants schedule affects the cooling and heating load. Sezgan and Koomey [2] estimated the interaction between the lighting and space conditioning energy use in commercial buildings. Therefore, these measures should be evaluated by a model that can treat the combination of these measures simultaneously. Generally, it is very difcult to quantitatively estimate the city or national scale effects of these measures. The quantitative evaluation of energy conservation measures has thus far been based on the simulation results for a standard household, which implies a family comprising two adults and two children. However, the energy consumption of each household differs considerably depending on the household type (number and age of members), building type, the number and efciency of appliances, the occupants activity, and other factors. Ultimately, to quantify the city-scale effect of the various energy conservation measures, which include the dissemination of energy efcient appliances and buildings and change in the occupants behavior correctly, a virtual city model must be developed. This model should be capable of simultaneously simulating the operation of all appliances and the occupants behavior in all households within the objective region without using unit energy consumption per household/person/oor area. Clarke et al. [3] applied a building simulation program for estimating the effect of the improvement in a thermodynamic class such as window size and insulation level by considering the present distribution of house types and thermodynamic classes in Scotland. Jones et al. [4] developed a model that estimates the residential energy used in a city by considering the distribution of the building energy used based on Geographical Information System techniques. Brownsword et al. [5] developed the urban energy model which simulates spatial and diurnal variations of energy demand based on diurnal demand prole of each consumer type. However, these models could not consider the energy use by each appliance. Michalik et al. [6] developed a structure model of electricity demand in the residential sector of a region based on a bottom-up approach that sums up each appliances operation schedule for improving the electricity load curve by demand side management. However, this method does not include the heat load calculation and the fuel consumption. The authors [7] have developed a bottom-up simulation model that simulates the city-scale energy consumption in the residential sector by considering the diversity of household and building types. In this model, the energy consumption for each household category was simulated

by the appliance energy use model, hot water supply model, and heating and cooling model. In the appliance energy use model, the energy use of each appliance was simulated individually based on the schedule data of the occupants behavior. In the heating and cooling model, the cooling and heating load was simulated from the building data and weather data. The internal heat gain, which was calculated by the appliance energy use model, and the occupants behavior schedule were also used in this model. Since rooms in Japan are commonly equipped with room air conditioners and heaters, which are intermittently used for air-conditioning, consideration of the occupants behavior schedule is necessary to correctly estimate the energy consumption for heating, cooling, and lighting as well as the energy consumption of appliances such as televisions. In Japan, the time allocation of living activities is surveyed every 5 years by Broadcasting Culture Research Institute [8]. These results can be used for modeling the occupants energy use schedule [9]. In this paper, our previous model was improved in terms of the heat load calculation, simulation of the occupants behavior schedule, and so on. The new model has been applied to Osaka City (population: 2598 thousand, households: 1044 thousand). The present amount of energy consumption in the residential sector is estimated and compared with the statistical data. In the nal part of this paper, the energy conservation effects of the various kinds of measures are evaluated quantitatively. 2. Simulation model 2.1. Structure of the simulation model Fig. 1 shows the structure of the simulation model. In this simulation, the annual energy consumption of one household is calculated iteratively for 19 household categories and 20 building categories10 categories for detached houses and 10 categories for apartment houses are set depending on the oor area. In addition, ve types of building insulation levels are assumed. Each occupants time allocation for living activities, amount and temperature of hot water supply, weather data, and appliances energy efciency properties are provided as input data. The simulation of heat load and energy use is conducted in time steps of 5 min. The total energy consumption by the residential sector in the object region can be estimated by multiplying the simulated energy consumption and the number of households in each category and then summing the products. The authors have also developed a stock model, which estimates the distribution of an appliances energy efciency and a buildings insulation levels in the object region and the object year. Furthermore, this model also estimates the input data for the simulation model, such as the appliances average energy efciency ratio, which means the weighted average of the energy efciencies of the existing appliances determined by the year of manufacture.

ARTICLE IN PRESS
Y. Shimoda et al. / Energy 32 (2007) 16171633 1619

Fig. 1. Flowchart of the simulation.

2.2. Appliance energy use model 2.2.1. Schedule of living activities To determine the occupants schedule of living activities, the results of the time allocation survey of living activities, which was performed by the Broadcasting

Culture Research Institute, NHK (Japan Broadcasting Corporation), have been used. In the former model [7], the living activity of an occupant at each time step was expressed as a percentage of probability. In the new model, the living activity of each household member of each household category over 1 year is

ARTICLE IN PRESS
1620 Y. Shimoda et al. / Energy 32 (2007) 16171633 Table 1 Power consumption of home electric appliances used in this simulation Appliances Room Number of Power consumption (W) holdings (per 100 households) Operating Standby mode 88.1 22.2 72.1 100.6 79.0 238.1 122.8 100.0 109.3 26.4 133.9 100.0 148.3 102.5 127.4 88.0 86.2 47.6 47.6 43.1 39.1 138.2 53.4 116.3 116.1 1250.0 1000.0 1000.0 200.0 500.0 114.0 600.0 20.0 126.0 1300.0 450.0 30.0 200.0 500.0 21.0 100.0 100.0 62.7 20.0
b

determined by the occupants behavior simulation as follows:

All household members are classied into 8 occupant categories corresponding to the time allocation survey: employed male, employed female, housewife, elementary school student, junior high school student, high school student, aged male, aged female. For each occupant category, the behavior schedule is simulated in time steps of ve minutes for 500 days of weekdays and holidays from the result of the time allocation survey. In this simulation, simultaneous activities such as watching television while having a meal can be considered. For each household category, the behavior schedule of each household member for a day is selected at random from the abovementioned 500-day schedules on a daily basis. This procedure enables the expression of the distribution of activities among occupants in the same category and avoids the occurrence of unrealistic peaks of simulated electricity load curves for the entire city.

Rice cooker Dishwasher Thermos Microwave Toaster TV Refrigeratora Fan Washing Machine Tumble dryer Hair dryer Desk lamp Vacuum Iron VCR

Kitchen Kitchen Kitchen Kitchen Kitchen Living and bedroom Kitchen Kitchen Bathroom

35.0 3.0 45.0 3.4 2.8 2.4 No standby 0.0 0.7 0.2 0.0 0.0 0.0 0.0 3.7 14.0 14.0 3.3 34.8 13.9 20.0 5.0 35.0

2.2.2. Link between living activities and energy use Each living activity calculated by this simulation is linked with the energy use of appliances and hot water use. The probability of appliance use is also considered for each time period. From the links between each family member and a room, the room where each living activity occurs is also identied to determine the energy use for heating, cooling, and lighting. Table 1 shows the power consumption and the standby power of the home electric appliances used in this simulation. The dissemination ratios of these appliances shown in the Table 1 are also considered in this model. 2.3. Lighting energy use schedule model It is assumed that all the occupied rooms (except when occupant is asleep) and corridors are illuminated at night. In the daytime, occupied rooms are classied as rooms that are always illuminated, rooms that are never illuminated, and rooms where the lighting is dependent on the brightness of daylight. The ratios of these categories are determined from the results of the questionnaire survey. Energy consumption due to lighting is set as 5 W/m2 for all rooms. The brightness of daylight is calculated from the weather data. 2.4. Hot water energy use model Energy consumption due to hot water use is calculated from the amount of hot water used and its temperature and the city water temperature. The city water temperature is considered to be a function of the outdoor air temperature. 2.5. Heating and cooling model 2.5.1. Heat load simulation Using the Standard Weather Data of Osaka City, dynamic heat load simulation is carried out. In the former

Bathroom Bathroom Bedroom Living room Living room Living and bedroom Radio Living and bedroom CD player Bedroom PC Bedroom PC accessories Bedroom BS tuner Living and bedroom Fax Living room Telephone Living room Shower toilet Toilet Kotatsu (foot Living room warmer) Electric carpet Living room

b b b

500.0 580.0

0.0 0.1

a Power consumption of refrigerator is modeled as a function of outdoor air temperature. b Operating mode is not considered since operation time is quite small.

model [7], the heat conduction and ventilation between rooms was not considered and only one insulation level (the average insulation level) was considered. In the new model, the ventilation and heat conduction between rooms is considered by a thermal circuit network method. As shown in Fig. 2, room air, a thin (interior) wall, or a window is expressed as one node and a thick (exterior) wall is expressed as two nodes (inside and outside) in the thermal circuit network. The heat transfer between two such nodes is calculated from the thermal resistance and temperature difference. Incident solar radiation on walls and windows is also considered. In this model, all detached houses are assumed to be wooden buildings and all apartment houses are assumed to be reinforced concrete buildings. To simulate the heating and cooling load for each building category, the oor plan of houses for each of

ARTICLE IN PRESS
Y. Shimoda et al. / Energy 32 (2007) 16171633 1621

the 20 building categories is assumed, as shown in Fig. 3. The heating and cooling load and energy consumption is simulated for ve types of insulation: no insulation, insulation below the 1980 standard, insulation up to the 1980 standard, insulation up to the 1992 standard, and insulation up to the 1999 standard. The heating and cooling energy consumption of each household and building category is obtained from the weighted average of these results with the share of each insulation level estimated from our stock model. The room air temperature of cooling is 27 1C. The ventilation rate (including inltration)

and room air temperature of heating are set differently depending on the building insulation category in order to consider the air tightness and the rebound effect of insulation [10]: 3.0 ac/h and 18 1C for no insulation, 2.0 ac/h and 20 1C for insulation below the 1980 standard, 1.0 ac/h and 21 1C for insulation up to the 1980 standard, 0.5 ac/h and 22 1C for insulation up to the 1992 and 1999 standards.

Fig. 2. Thermal circuit network model.

2.5.2. Room air conditioner model The coefcient of performance (COP) of a room air conditioner is modeled as a function of the outdoor air temperature and part load ratio (compressor speed); it varies with each time step. This function is derived from the calculation method of the Seasonal Energy Efciency Ratio (SEER) for a room air conditioner by Japan Refrigeration and Air Conditioning Industry Association (JRAIA) [11]. Fig. 4 shows the COP of the room air conditioner used in the base case (averaged value in 2000) and that which conformed to the 1999 Law Concerning Rational Use of Energy (top-runner standard). The state transition probability function, which decides whether the room air conditioner is powered based on the room air temperature and time [9], is also considered in this model [12]. In Japan, most room air conditioners have a heat pump mode. The energy consumption for heating is calculated from the estimated share and efciency of the room air conditioner, electric heater, city gas heater, and kerosene heater.

Bed room 9.96 m2

Bed room 11.6 m2

Kitchen 4.98 m2 Bed room 11.6 m2 Living room 14.9 m2

9100

9.96 m

8190

Living room 19.9 m2

Bed room 9.96 m2

Bed room 9.96 m2

5460 Balcony 7280

1F

Balcony

Apartment house 68.7m2

Detached house 87.2m2

Fig. 3. Example of the plan of the house used in this model.

8190

Bed room

Kitchen 8.3 m2

ARTICLE IN PRESS
1622 Y. Shimoda et al. / Energy 32 (2007) 16171633

12 10

2.6. Classication of households

Cooling (Minimum speed) Top runner (Maximum speed) (Minimum speed)

8 COP 6 Base case 4

Using the results of the National Population Census [13], all households in Osaka City are classied into 19 household categories, 2 building categories (detached house or apartment house), and 10 oor area categories for each building category. Table 2 shows the household categories and number of households in Osaka City used in this study. 2.7. Identication of the parameters

(Maximum speed) 2 0 20 22 24 26 28 30 32 34 36 Outdoor Air Temperature [C]

10 Heating 8 Top 6 (Minimum speed)

In addition to the settings shown in the previous sections, there are some parameters that are necessary to run this simulation model, such as the share of heating equipment, frequency of bathing, dissemination ratio of appliances, and usage ratio of lighting in daytime. These parameters are determined from the results of the questionnaire survey for the residents of Osaka City [14]. 3. Simulation results for present condition and verication of the model 3.1. Simulation results for each household

COP

Base (Minimum speed) 4 (Maximum speed) 2 0 -10 -5 0 5 10 15 Outdoor Air Temperature [C] (Maximum speed)

Fig. 4. Relationship between the outdoor air temperature and COP. (Cooling and heating mode, 2.8 kW cooling capacity). Table 2 Household types and number of households in Osaka City No. of household members 1 Family type of household Male Female Aged male Aged female Couple Aged couple Mother and a childb 3 Couple and a child Mother and children 4 5 More than 5 Total
a b

In this paper, energy consumption is indicated as the primary energy consumption, and electricity consumption is calculated by the following relation: 1 kWh of electricity 9830 kJ of primary energy consumption.

No. of working personsa 1 1 0 0 2 1 0 1 0 2 1 1 0 2 1 2 1 2 1

Category 1a 1b 1c 1d 2a 2b 2c 2d 2e 3a 3b 3c 3d 4a 4b 5a 5b 6a 6b

Detached house 22,442 15,670 10,490 39,513 22,847 21,259 58,887 16,438 9739 25,279 40,808 9139 5414 30,665 39,909 10,571 13,757 4833 6487 404,146

Apartment house 178,867 125,233 22,323 40,278 43,432 40,414 38,968 25,833 15,304 31,942 51,565 11,146 6603 35,816 46,612 10,595 13,788 2639 3543 744,901

Couple and children Couple and children Couple, children, and parents

Working person means a family member who is out on business on a weekday. All children are dened as students.

ARTICLE IN PRESS
Y. Shimoda et al. / Energy 32 (2007) 16171633 1623

Fig. 5 shows the distribution of the simulated annual primary energy consumption of one household in a detached house and an apartment house. These gures show that the total energy consumption depends more on the number of family members than the total oor area. The inuence of the total oor area on the energy consumption is stronger in a large family since the number of occupied rooms and energy use for lighting, heating, and cooling increases with the number of family members. In addition, energy consumption for TVs increases with the number of rooms occupied since a TV is assumed to be available in the living room and all of the bedrooms. By this assumption, the total number of TVs owned approaches the statistical value shown in Table 1.

For the same oor area and household category, the energy consumption in a detached house is greater than that in an apartment house. Most of this difference is due to the amount of heating energy. The difference increases in a large familys house, as the number of rooms heated is greater.

3.2. Total energy consumption in the city and comparison with statistical data Fig. 6 shows the comparison between the simulated annual primary energy consumption and actual energy
1a

160.0 140.0 consumption [GJ/year/household] 120.0 100.0 80.0 60.0 40.0 20.0 0.0 0 20 40 60 80 100
2

1b 1c 1d 2a 2b 2c 2d 2e 3a 3b 3c 3d 4a 4b 5a 5b 120 140 160 180 6a 6b

Annual primary energy

Floor area [m ]

1a 160.0 140.0 consumption [GJ/year/household] 120.0 100.0 80.0 60.0 40.0 20.0 0.0 0 20 40 60 80 100 120 140 160 180 Floor area [m2] 1b 1c 1d 2a 2b 2c 2d 2e 3a 3b 3c 3d 4a 4b 5a 5b 6a 6b Annual primary energy

Fig. 5. Distribution of the total energy consumption for an apartment house [top] and detached house [bottom]. (Symbols 1a-6b in the legend correspond to the category column in Table 2).

ARTICLE IN PRESS
1624 Y. Shimoda et al. / Energy 32 (2007) 16171633

Fig. 6. Comparison between the simulated energy consumption in Osaka City and the statistical value (in primary energy [1 kWh of electricity 9830 kJ]).
3.0 Simulated 2.5 Gas Consumption[PJ/month] Actual supply data

2.0

1.5

1.0

10.2%, and 30.3%, respectively. Monthly data of city gas supply is compared in Fig. 7. The annual and monthly simulated gas consumption have good agreement with the actual data. Fig. 8 shows the electricity load curve of Osaka citys residential sector. An improvement of the energy use schedule model (introduction of occupants behavior simulation) makes this load curve more realistic than that of the previous model [7]. Even at midnight, the electricity consumption by appliances does not reduce signicantly since most of it is in the form of standby energy use. Since citys level load curve for the residential sector alone cannot be measured, the measured electricity load curve of the residential district in the vicinity of Osaka City in July 2001 and sensitivity of electricity consumption to outdoor air temperature derived from the load curve [15] are used for verication of the model. In this district, total oor areas of residential buildings and nonresidential buildings are 1,144,000 m2, and 184,000 m2, respectively. Simulation was done for Osaka City area using weather data of the same period as measurement. The differences of household and building type distribution between Osaka City and the measurement district are not considered. Fig. 9(a) shows the comparison of the normalized electricity load curve. The simulated load curve on holiday and the daily peak values have a good agreement with the actual value. On the other hand, the simulated electricity consumption at midnight and in the weekdays daytime is smaller than the actual value. The reasons for these results are supposed that the actual value contains the electricity use by end-uses other than residential buildings, and that the ratio of people at home in the daytime in the measurement district is larger than Osaka City as the proportion of single family in the district is small. Fig. 9(b) shows the percentage of increase in electricity consumption by a 1 1C increase of outdoor air temperature. Simulated sensitivity shows good agreement with the value estimated from measured data. Since sensitivity of electricity consumption to air temperature in summer has a close relationship with cooling, it is clear that the model can simulate cooling energy consumption with sufcient accuracy. 4. Evaluation of various energy-saving measures

0.5

0.0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Fig. 7. Comparison between the simulated monthly gas consumption and the actual value.

supply in the residential sector in Osaka City in 2000, when the weather data used in the model is observed. The actual supply data was estimated by utility companies. The amount of kerosene supplied is estimated from the units per household obtained from the statistical data. The simulated total energy consumption agrees with the actual primary energy supply within 1.0%. The errors in electricity, city gas and kerosene supply are 1.3%,

One signicant advantage of this model is its simulation of the heating and cooling loads precisely by coupling the dynamic heat load simulation and energy use schedule model. Further, it enables to consider the change in the energy consumption due to heat insulation of buildings, climate conditions, and schedules of living activities. Accordingly, the effect of a heat insulation standard, an energy efciency standard for room air conditioners and an introduction of daylight saving time are predicted as an example of the energy conservation policy evaluation by this simulation model. 4.1. Heat insulation of a building The annual energy consumption for heating and cooling for each household and building category

ARTICLE IN PRESS
Y. Shimoda et al. / Energy 32 (2007) 16171633 1625

Fig. 8. Simulated electricity load curve of Osaka Citys residential sector (August 6 [Sunday] to August 7 [Monday]). (Kitchen means ve appliances from the top of Table 1; Energy use of refrigerator is modeled as a function of the daily average outdoor air temperature. Hourly change is not modeled).

Fig. 9. Comparison between the simulated electricity load curve and the measured value.

is simulated under three heat insulation conditions as follows: (1) Base case: The share of each insulation level in 2000 is assumed as shown in Table 3. Since an insulation

standard for houses is not mandatory in Japan, the shares of the 1992 and 1999 standards are very small. (2) 1992 Standard case: All houses in the city conform to the 1992 standard (New Energy Conservation Standard).

ARTICLE IN PRESS
1626 Table 3 Heat insulation conditions used in this study Standard Detached house No insulation Below 1980 1980 Standard 1992 Standard 1999 Standard Apartment house No insulation Below 1980 1980 Standard 1992 Standard 1999 Standard Exterior wall Window Single glazing (6.51 W/m2K) with lace curtain Single glazing (6.51 W/m2K) with lace curtain Single glazing (6.51 W/m2K) with lace curtain Single glazing (6.51 W/m2K) with lace curtain Double glazing (4.65 W/m2K) with blind Single glazing (6.51 W/m2K) with lace curtain Single glazing (6.51 W/m2K) with lace curtain Single glazing (6.51 W/m2K) with lace curtain Single glazing (6.51 W/m2K) with lace curtain Double glazing (4.65 W/m2K) with blind Share in the base case (%) Y. Shimoda et al. / Energy 32 (2007) 16171633

None Glass Glass Glass Glass

wool wool wool wool

(0.050 W/mK) (0.050W/mK) (0.038W/mK) (0.036W/mK)

15.0 mm 30.0 mm 38.0 mm 79.2 mm

47.6 16.1 31.7 4.4 0.2 20.2 24.2 50.6 4.8 0.2

None Polystyrene Polystyrene Polystyrene Polystyrene

foam foam foam foam

(0.040W/mK) (0.040W/mK) (0.040W/mK) (0.028W/mK)

10.0 mm 20.0 mm 36.0 mm 30.8 mm

35.0 Annual primary energy reduction [GJ/year/household] 30.0 25.0 20.0 15.0 10.0 5.0 0.0 0 20 40 60 80 100
2

120

140

160

180

Floor area [m ]

1a 1b 1c 1d 2a 2b 2c 2d 2e 3a 3b 3c 3d 4a 4b 5a 5b 6a 6b

Fig. 10. Distribution of the primary energy saving from the base case by conforming to the 1999 insulation standard (detached house). (Symbols 1a6b in the legend correspond to the category column in Table 2).

(3) 1999 Standard case: All houses in the city conform to the 1999 Standard (Next Generation Energy Conservation Standard). Fig. 10 shows the distribution of the simulated annual primary energy reduction in the 1999 standard case in comparison with the base case for detached houses. Even in the same house, the amount of energy conservation increases with the number of household members. The difference of energy saving between a one-person household and a six-person household is 56 times for a large house. The energy saving in a detached house is 12 times greater than that in an apartment house with the same oor area. Fig. 11 shows the total energy consumption for heating and cooling in the Osaka City area under the three cases. If all buildings conform to the 1999 standard, the heating energy consumption is reduced by 9.1 PJ/year from the base case. The cooling energy consumption is also reduced,

Fig. 11. The effect of heat insulation on the total energy consumption for heating and cooling in Osaka City.

but the difference is small. The total amount of energy saved is equal to 13.2% of the total primary energy consumption in the residential sector in Osaka City.

ARTICLE IN PRESS
Y. Shimoda et al. / Energy 32 (2007) 16171633 1627

4.2. Energy efciency standard for room air conditioner Fig. 12 shows the distribution of the simulated primary annual energy reduction by substituting all the room air conditioners with those that conform to the 1999 Law Concerning Rational Use of Energy (top-runner standard) in comparison with the base case for a detached house. Since the difference of the COP in cooling is larger than that in heating, as shown in Fig. 4, and other heating equipment such as the city gas heater is used for heating, the energy saving in cooling is larger than that in heating. Even in the same building and room air conditioner setting, the amount of energy saving varies considerably depending on the occupants behavior. Fig. 13 shows the comparison of the primary energy consumption in Osaka City for cooling and heating under the present energy efciency and the efciency conforming to the top-runner standard. From the simulation results

shown in the center of the gure, the primary energy consumption for cooling and heating in Osaka City is reduced by 45% (2.63 PJ/year) and 6% (0.69 PJ/year), respectively. The sum of both these reductions is equal to 4.6% of the total primary energy consumption in Osaka City. The left side of Fig. 13 shows the energy consumption estimated by the JRAIAs SEER calculation method [11]. In our simulation model, room air conditioners are assumed to be installed in living rooms and all bedrooms for each of the 20 building categories, and the capacity of each room air conditioner is selected from ve categories (2.2, 2.5, 2.8, 3.5, and 4.0 kW of cooling capacity) based on the room area. By multiplying the total number of households for each building category and summing up the products, the total number of room air conditioners units in Osaka City is obtained for the respective capacities. Therefore, by multiplying the seasonal electricity
1a 1b 1c 1d 2a 2b 2c 2d 2e 3a 3b 3c 3d 4a 4b 5a 5b 6a 6b 1a 1b 1c 1d 2a 2b 2c 2d 2e 3a 3b 3c 3d 4a 4b 5a 5b 6a 6b

1
Annual cooling energy reduction

9.0 Cooling 8.0 7.0 [GJ/year/household] 6.0 5.0 4.0 3.0 2.0 1.0 0.0 0 20 40 60 80 100 [m2] 120 140 160 180 Floor area

2
Annual heating energy reduction

3.0 Heating 2.5 [GJ/year/household]

2.0

1.5

1.0

0.5

0.0 0 20 40 60 80 100 120 140 160 180 Floor area [m2]

Fig. 12. Distribution of the primary energy saving in room air conditioner from base case by conforming to the top-runner standard (detached house). (Symbols 1a-6b in the legend correspond to the category column in Table 2).

ARTICLE IN PRESS
1628
60 Primary Energy Consumption [PJ]
Heating

Y. Shimoda et al. / Energy 32 (2007) 16171633

50
Cooling

40 30 20 10 0 Base Top-runner JRAIA Base Top-runner Top-runner Simulation (All RAC Heating) Simulation

Fig. 13. Change in primary energy consumption of room air conditioner by conforming to the top-runner standard.

consumption for each capacity estimated by the JRAIA method and the total number of units, the annual energy consumption by room air conditioners in Osaka City is estimated with the same unit number condition as that in the simulation. Both the total energy consumption and energy-saving effect of the top-runner standard estimated by the JRAIA method are much larger than the simulation results shown in the center of the Fig. 13. One of the reasons is the overestimation of the operation hours by the JRAIA method. Although 18 h operation (from 6 AM to 0 AM) per day is assumed in the JRAIA SEER calculation method, room air conditioners are not operated often during the daytime in Japanese residences. As shown in Fig. 8, the result of the simulation shows that the operation of room air conditioners during the daytime on weekdays is restrained. As previously mentioned, the other reason is that other heating appliances are used in the simulation model. To make the best use of the energy efciency of the toprunner standard, a case in which all heating is provided by the room air conditioner is also simulated. The result is shown on the right-hand side of Fig. 13. In this case, the primary energy consumption is reduced by 40% (8.29 PJ) from the ordinary top-runner standard case. In this case, the total primary energy saving by conforming to the toprunner standard and changing the heating equipment is 8.8 PJ from the base case, and it is equivalent to 12.4% of the total primary energy consumption in Osaka City. 4.3. Introduction of daylight saving time Daylight saving time, which has not been adopted in Japan, intends to reduce the lighting energy by setting the clocks 1 h later than the normal local time. However, in Japan, the daylight saving time may potentially increase the cooling load in the evening. In the simulation, the term representing the daylight saving time is assumed to have limits from the rst Sunday

in April to the last Sunday in October. The daylight saving time is modeled by shifting the weather data from the other schedule data by 1 h. The electricity load curve for lighting and cooling in Osaka Citys residential sector on August 1 and 2 is shown in Fig. 14. In late afternoon, the lighting energy is reduced by the daylight saving time due to the difference of sunlight. The energy use for cooling in the morning is also slightly reduced since the outdoor air temperature becomes lower than that in the base case. On the other hand, the energy use for cooling in the evening becomes greater than that in the base case. This is because the cooling load becomes higher due to the higher outdoor air temperature and solar radiation, and the frequency of cooling use is higher in the evening than in the morning. In the seasonal total, the energy reduction of lighting becomes 0.02% of the total annual primary energy consumption in the residential sector in Osaka City. The energy use for cooling increases by 0.15% of the total energy use. Therefore, by the adoption of daylight saving time, the primary energy use is increased by 0.13% of the total primary energy consumption in Osaka Citys residential sector. This result depends on the magnitude of the cooling load and the occupants cooling behavior in Japan. In order to examine the energy-saving effect of daylight saving time, a simultaneous evaluation in non-residential buildings is necessary. 4.4. Evaluation of the other energy conservation measures In addition to the results mentioned above, major energy conservation measures in the residential sector, as stated in the New Climate Change Policy Program, were evaluated quantitatively. The energy conservation measures considered in this paper and these simulation methods are as follows:

  

 

Easing set room temperature. Set room temperature is changed from base case to 28 1C in cooling and 19 1C in heating. Conforming to top-runner standard for refrigerators. Annual energy consumption is reduced from 600 to 400 kWh. Conforming to top-runner standard for televisions. Energy consumption in the operation mode is reduced from 120 to 103 W, while standby power is reduced from 2.0 to 0.6 W. Conforming to top-runner standard for all other home appliances. All lighting devices, VCRs, shower toilets, gas water heaters, and oil water heaters are made to conform to their corresponding top-runner standard. Introduction of water saving shower head. Hot water used for showers is reduced by 50%. All the members of family watching TV together. In the energy use schedule model, the room where the activity of watching TV and resting occurs is changed from each members bedroom to the living room in order to

ARTICLE IN PRESS
Y. Shimoda et al. / Energy 32 (2007) 16171633 1629

Fig. 14. Change in the electricity load curve by adopting daylight saving time.

Fig. 15. Energy conservation effects of each measure in Osaka City.

ARTICLE IN PRESS
1630 Y. Shimoda et al. / Energy 32 (2007) 16171633

reduce the energy consumption for television, lighting, heating, and cooling in the bedroom. Fig. 15 shows the energy conservation effects in Osaka City by all measures including the heat insulation of buildings and conforming to the top-runner standards for all room air conditioners in order of the amount of the energy conservation effect. The result shows that the energy conservation standards for heat insulation, refrigerator, and room air conditioner are effective for city-scale energy conservation. On the other hand, occupants behavioral changes such as watching TV together and easing set room temperature are less effective. Fig. 16 shows the peak electricity reduction effects for various measures that are identical to those in Fig. 15. In Osaka Citys residential sector, peak electricity occurred at 18:00 in July. The top-runner standard for room air conditioners shows the maximum peak value for the electricity reduction effect. Although the annual cooling energy does not decrease signicantly, peak electricity decreases considerably due to the heat insulation standard. 4.5. The total effect by combining multiple measures Since the effect of each energy efciency measure interacts with the other measures, the combined effect of some measures cannot be estimated by summing each effect, which is shown in Fig. 15. Therefore, the simulations on the energy-saving effect of multiple measures are performed. These simulations are performed according to the following three steps: (1) Step 1: Energy efciency of the appliances: Conforming to the top-runner standard for room air conditioner, refrigerator, television, and the other home appliances.

(2) Step 2: Step 1+Energy-saving occupants behavior: Combination of easing set room temperature, introduction of water-saving shower head and all family members watching TV together. (3) Step 3: Step 2+All houses conforming to the 1999 insulation standard.

Results of these simulations are shown in Fig. 17. In Step 1, the energy consumption is reduced by 13.7% from the base case value. This value is lower than 14.2%, which is the sum of each measures effect, as shown in Fig. 15. One of the reasons for this is that the energy used for heating has not reduced, as shown in Fig. 15, since the internal heat gain is reduced by a decrease in the energy use of appliances. Since the top-runner standard has become mandatory, this reduction can be expected in the near future. In Step 2, the energy consumption is reduced by 5.1% from that in Step 1 due to the reduction in heating, cooling, hot water, television, and lighting. The effect of building insulation in Step 3 is very large. The total energy consumption for cooling and heating becomes 37.5% of that in the base case. In this case, since the set room air temperature is lowered to 19 1C, the energy reduction by building insulation is greater than that shown in Figs. 11 and 15, which consider the rebound effect of the building insulation (increase in set room temperature with the insulation level). On the other hand, the substitution of all other heating equipment with room air conditioner is not considered in this simulation. By Step 3, the total energy reduction becomes 28.5% of the base case. However, a substantial amount of time will be required to achieve this situation since the building insulation

Fig. 16. Peak electricity reduction effects of each measure in Osaka City.

ARTICLE IN PRESS
Y. Shimoda et al. / Energy 32 (2007) 16171633 1631

Fig. 17. Combined effects of all the measures.

Fig. 18 shows the distribution of the annual primary energy consumption of one household in the case of Step 3. In general, the energy-saving effect of the measures increases with the oor area of the building and the number of household members. Consequently, as is evident by comparing Fig. 18 with Fig. 5, the difference between the annual primary energy consumption in an apartment house and a detached house, the difference due to the oor area of the building, and the difference between household categories, all become smaller than that in the base case. 5. Conclusion In this paper, a new virtual city end-use model, which simultaneously simulates the operation of all appliances and occupants behavior in all households in the city, is developed. This model can quantitatively estimate the cityscale effect of various types of energy conservation measures such as energy efcient appliances, insulation of buildings, and change in occupants behavior. Using this model, various types of energy conservation measures in the residential sector proposed by the Japanese government are evaluated quantitatively for the residential sector in Osaka City. The results of the simulation are as follows:

Fig. 18. Distribution of the total energy consumption for apartment house [top] and detached house [bottom] in the combination case of Step 3. (Symbols 1a-6b in the legend correspond to the category column in Table 2).


standard is currently not mandatory and a long time will be required to substitute the present building stock with a new one.

The effect of measures with regard to heating and cooling energy use such as building insulation and introduction of energy efcient room air conditioners is large. In particular, if all the residential buildings conform to the 1999 insulation standard, the primary energy consumption in Osaka City will decrease by 13.2% even if the rebound effect on the set air temperature is considered. For estimating the total annual energy consumption of appliances in the city, the conventional estimation method, which multiplies the standard annual energy consumption of an appliance with the total number of appliances, results in an overestimation in case of room

ARTICLE IN PRESS
1632 Y. Shimoda et al. / Energy 32 (2007) 16171633

 

air conditioners. One of the reasons for this is that intermittent heating and cooling is common in Japanese residential buildings. Changes in the occupants behavior such as all family members watching television together and easing room air temperature are less effective. In this model, the combined effect of the measures can be evaluated by considering the interaction between the measures such as the heating load increase by a reduction in the energy used by appliances in the room. The simulation result shows that the top-runner standard for home electric appliances reduces the primary energy consumption in Osaka City by 13.7%. If all the measures for building insulation and occupants behavior considered in this study are adopted and the appliances conform to the top-runner standard, the primary energy consumption in Osaka City can be reduced by 28.5% of its present value. On this occasion, the difference in the primary energy consumption due to the difference in building and household types reduces. Although the authors had not adjusted the parameters in accordance with the difference between the simulation results and the measured data, a good agreement was observed between the results and the statistical data in Figs. 6 and 7. However, it is considered that many errors may be included in each household level. For example, the rebound effect of insulation (e.g., room air temperature in heating) must vary with the household, and it is difcult to dene this effect. If there is no rebound effect and the room air temperature of the entire room during heating is 18 1C, the energy consumption for heating decreases by 12.5% from the base case, and the total energy consumption decreases by 2.4%. In addition, this model does not considered the shading effect of neighboring buildings, the position of the room in the apartment house, and the building orientation since the distributions of these parameters are difcult to estimate from statistical data and the inclusion of these parameters in the model requires considerable computation time. Further, validation of the model with various types of measured data is necessary to improve it.

as weather conditions and demographic conditions. For example, this model can evaluate the change in energy use due to an increase in air temperature on account of an urban heat island effect [7,12]. In the future, this can be applied to predict the impact of climate change on the energy efciency measures [16]. Acknowledgment This work is supported by Grants-in-Aid for Scientic Research, Japan Society for the Promotion of Science, Nos. 15360310 and 18360273. References
[1] The Energy Data and Modeling Center, The Institute of Energy Economics, Japan. EDMC handbook of energy and economic statistics in Japan 2002. Tokyo: The Energy Conservation Center, Japan; 2002. [2] Sezgan O, Koomey JG. Interactions between lighting and space conditioning energy use in US commercial buildings. Energy 2000;25:793805. [3] Clarke JA, Johnstone CM, Kondratenko M, Lever M, McElroy LB, Prazeres L, et al. Using simulation to formulate domestic sector upgrading strategies for Scotland. Energy Build 2004;36:75970. [4] Jones PJ, Lannon S, Williams J. Modeling building energy use at urban scale. In: Lamberts R, Negrao C, Hensen J, editors. Proceedings of the seventh international IBPSA conference. Rio de Janeiro, Brazil: IBPSA; 2001. p. 17580. [5] Brownsword RA, Fleming PD, Powell JC, Pearsall N. Sustainablecitiesmodelling urban energy supply and demand. Appl Energy 2005;82:16780. [6] Michalik G, Khan ME, Bonwick WJ, Mielczarski W. Structural modeling of energy demand in the residential sector: 1. Development of Structural Models. Energy 1997;22(10):93747. [7] Shimoda Y, Fujii T, Morikawa T, Mizuno M. Residential end-use energy simulation at city scale. Build Environ 2004;39(8):95967. [8] Broadcasting Culture Research Institute, Japan Broadcasting Corporation. NHK data book 2000 national time use survey, National, Prefectural (Kokumin Seikatsu Jikan Chosa). Tokyo: Japan Broadcast Publishing; 2001 [in Japanese]. [9] Tanimoto J, Hagishima A, Sagara H. A methodology for peak energy requirement considering actual variation of occupants behavior schedules. Build Environ, in press, doi:10.1016/j.buildenv.2006. 06.034. [10] Mizuishi T, Murakami S, Ikaga T. LCCO2 assessment on thermal insulation of residential buildings including uorocarbon leakage. J Environ Eng (Transactions of Architectural Institute of Japan, 5-26-20 Shiba, Munato-ku, Tokyo 108-8414) 2004;579:8996 [In Japanese]. [11] Japan Refrigeration and Air Conditioning Industry Association. Calculating method of annual power consumption for room air conditioners, JRA4046. Tokyo, Japan: Refrigeration and Air Conditioning Industry Association (3-5-8, Shibakoen, Minatoku, Tokyo 105-0011); 1999 [in Japanese]. [12] Shimoda Y, Nakao T, Narumi D, Habara H, Mizuno M, Tsuji K. Verication of residential energy end-use model at city scale and effect of outdoor air temperature. Energy Resour (Enerugi- Shigen, 1-9-10 Kyomachibori, Nishi-ku, Osaka 550-0003) 2006;27(6):4317 [in Japanese]. [13] Statistics Bureau and Statistics Center. 2000 Population Census of Japan, Tokyo, Japan Statistical Association (2-4-6 Hyakunin-cho, Shinjukuku, Tokyo 169-0073), 2000 [in Japanese]. [14] Morikawa T, Shimoda Y, Mizuno M. Estimation of residential energy end-use in Osaka city considering the distribution of residence

When estimating the national-scale effect of these measures from the results of this paper, it must be noticed that the proportion of single families and small buildings in Osaka City is large in comparison to that in many other cities. For example, the proportion of single families in Osaka City is 40.3%, while the Japanese average is 27.4%. Nonetheless, the authors feel that this model can be applied to many other locations and various types of measures. For evaluating a micro-CHP system, which requires matching of electricity demand and supply, the detailed electricity load curve is indispensable. Therefore, it is assumed that this model, which uses a 5 min time step, will guarantee benecial effects for such a purpose. This model also permits the application of various types of conditions such

ARTICLE IN PRESS
Y. Shimoda et al. / Energy 32 (2007) 16171633 and appliances. J Environ Eng (Transactions of Architectural Institute of Japan, 5-26-20 Shiba, Munato-ku, Tokyo 108-8414) 2005;595:97104 [in Japanese]. [15] Narumi D, Niurao Y, Shimoda Y, Mizuno M. Effects of temperature change upon the regional electric power consumption, energy and 1633 resources (Enerugi-Shigen, 1-9-10 Kyomachibori, Nishiku, Osaka 550-0003) 2007;28(3), in press [in Japanese]. [16] Gaterell MR, McEvoy ME. The impact of climate change uncertainties on the performance of energy efciency measures applied to dwellings. Energy Build 2005;37:98295.