Comparison of heating and cooling energy simulation using multi-years and typical weather data in South Korea

Abstract

Typical weather data (TWD) consists of 8760 values of various selected meteorological parameters such as ambient temperature, solar radiation, relative humidity and wind velocity, and are originally derived from long-term data. In this paper, TWDs have been generated, using three different methodologies, from hourly meteorological data measured in Seoul, Daejeon, Daegu, Gwangju, and Busan. The aims of this study were to compare the simulation results from the three methodologies (ISO Test Reference Year, CIBSE Test Reference Year, and Typical Meteorological Year) with those from individual years and their long-term means to assess the important five components of the building envelope (wall-to-window ratio, window overhang, window transmissivity, window U-value, and wall U-value) for the influence on the heating and cooling load and to study the influence of the different weather parameters (temperature and solar radiation). Results of the comparison show that predicted monthly load and energy consumption profiles from the TWD tended to follow the long-term mean quite closely. The load calculations vary almost linearly as the building envelope components varied apart from the window U-value. The yearly average values of mean bias error and root mean squared error for the heating and cooling loads were evaluated for the three TWDs and long-term measured meteorological data.

Practical application : This paper will enable engineers and building designers of buildings in South Korea to assess the heating and cooling loads with more confidence. The paper will enable designers to use the appropriate weather data when using computer simulation for the design. The designer will also be able to assess the energy efficiency better.

Keywords

Energy simulation ISO test reference year CIBSE test reference year typical meteorological year South Korea

Introduction

Computer simulation of buildings and solar energy systems is being used increasingly in energy assessments and design. Simulation often requires hourly weather data. “Typical weather data” (TWD) consists of 8760 values of various selected meteorological parameters such as ambient temperature, solar radiation, relative humidity, and wind velocity that are originally derived from long-term data. Several methods for generating typical data have been developed. The TWD methods were developed to generate statistically significant years for dynamic thermal energy analysis by computer simulation. These methods included the CIBSE Test Reference Year (CIBSE TRY), the Typical Meteorological Year (TMY) used in the USA, and the ISO 15927-4¹ TRY (ISO TRY). They are also essential for the numerical analysis of sustainable and renewable energy systems. They differ in some details, but are all based on constructing a year from 12 individually selected months of real weather data concatenated to make a complete year. In this paper the selection was made from 20 years (1986–2005) of daily average weather data for Seoul, Daejeon, Daegu, Gwangju, and Busan in South Korea.

Many authors showed the alterative design possibilities using the TWD. Lam² investigated the likely diversity in computer predictions based on different weather years in Hong Kong. Gugliernetti et al.³ investigated the influence of the stochastic component of meteorological data in evaluation office building energy performance in Mediterranean climate. Westphal and Lamberts⁴ estimated annual energy consumption of non-residential buildings based on simplified weather data in 206 Brazilian locations. Enshen⁵ compared the annual heating and cooling energy consumptions on the same building under 43 different climate conditions. Kolokotroni et al.⁶ had been used the measured air temperature data to assess the heating and cooling load of a typical air-conditioned office building positioned at 24 different locations within the London. Mui and Wong⁷ investigated the development of a new example weather year and a mathematical model to generate design occupant load profiles using Monte Carlo simulation for a subtropical climate. Karlsson et al.⁸ discussed the relative importance to the annual energy demand of different energy aspects of a Swedish low-energy house. Yong et al.⁹ compared the energy simulation results using multi-years and typical meteorological years in different climates of China. Jentsch et al.¹⁰ had suggested the generation and assessment of building simulation weather files for the potential impacts of a changing climate. Radhi¹¹ evaluated the climate variation impact on the performance of weather data from different weather periods in Bahrain. Kershaw et al.¹² compared simulations run using the reference years and the results obtained from simulation using the base data sets from multi-year. Chow and Levermore¹³ provided a comprehensive study and understanding to how various different office buildings in the UK will cope with climate change as well as the associated CO₂ emissions as a result of meeting the thermal demands.

The simulation runs were performed and building heating and cooling loads were analyzed in the five major cities using TWD (ISO TRY, CIBSE TRY, and TMY) to compare the results with those predicted with individual years within the multi-year datasets from which the TWD were developed. In the study the objectives of this paper were:

(1) To compare the impact on simulation prediction of three alternative methods for the production of TWDs from long-term measurements.

(2) To predict the annual cooling and heating load for office buildings based on the components of building envelope (wall-to-window ratio, window overhang, window transmissivity, window U-value, and wall U-value).

(3) To study the influence of the weather parameters (temperature and solar radiation) in the assessment of office building energy performance in the South Korea.

Data and TWD methods

Hourly weather data for energy simulation

The data for this study come from the five major cities in South Korea and cover 20 years (1986–2005). All meteorological and solar radiation data that are used in the study were measured by the Korea Meteorological Administration (KMA). Figure 1 and Table 1 shows key measured weather parameters and the geographical characteristics of the stations, such as latitude, longitude, and altitude above sea level. Each weather year has 8760 records of hourly data of dry bulb temperature (DBT), wet bulb temperature (WBT), global solar radiation, wind speed, atmospheric pressure, cloud cover, and wind direction. WBT data were determined from the corresponding DBT and relative humidity (RH) data measured at the respective stations. Months with missing data were rare. Specifically, the missing data as a percentage of the entire 20-year period for each parameter was 0.6% for global radiation, 0.1% for temperature and relative humidity, while there were no missing values for wind speed and wind direction. Missing values were filled by linear interpolation or by estimates. Sudden, unnatural jumps or singular solar radiation values were examined and corrected using the cloud cover and the corresponding extra-terrestrial solar component. If the raw data were available only as 3-hourly data, then the missing values were calculated by linear interpolation.

Figure 1.

Location of weather files locations in South Korea.

Table 1.

Geographical data and weather parameters for locations considered in this study.

Location	Latitude (°N)	Longitude (°N)	Altitude (m)	Weather parameters measured at the KMA
Seoul	37.34	126.58	85.5	Relative humidity (%), Dry bulb temperature (℃), Wind speed (m/s), Wind direction (°), Atmosphere pressure (Pa), Global solar radiation (W/m²), Cloud cover (fraction)
Daejeon	36.18	127.24	77.1
Daegu	35.53	128.37	57.8
Gwangju	35.08	126.55	70.9
Busan	35.06	129.02	69.2

Diffuse-to-global solar irradiation correlation

The total solar irradiations on horizontal surfaces are separated into the beam and diffuse components. Although horizontal global irradiance is a commonly measured parameter for many sites, horizontal diffuse irradiance is not so readily obtainable. For such sites that measure global irradiation alone a simple but reasonably accurate method is required to estimate diffuse irradiance from its global counterpart. This study investigates the applicability of various standard models correlating hourly diffuse fraction k and hourly clearness index k_t in the South Korea. Orgill and Hollands¹⁴ were the first to develop an equation to determine the hourly ratio of diffuse-to-total irradiation received in a horizontal surface. Erbs et al.¹⁵ derived the correlation between the hourly diffuse fraction and the hourly clearness day from four US locations. Reindl et al.¹⁶ estimated the diffuse fraction using two different models developer with measurements of global and diffuse irradiation on a horizontal surface from five European and North American locations. In the absence of a site-specific diffuse estimation algorithm, CIBSE Guide J¹⁷ proposed equations to estimate the horizontal hourly diffuse irradiation from the observed hourly global horizontal irradiation. Lee et al.¹⁸ estimated the hourly solar irradiation on inclined surfaces in South Korea.

To quantify the performance of the models and ascertain whether there was any underlying trend in the different location, the mean bias error (MBE) and root mean square error (RMSE) were determined for the five cities. MBE provides information on the long-term performances of the modeled regression equations. A positive MBE value indicates the amount of overestimation in the predicted hourly solar irradiation and vice versa. RMSE provides information on the short-term performances, and indicates the scattering of data around the linear regression. According to Lee et al., comparisons between measured I_d,m and calculated I_d,c diffuse irradiation along with the values of RMSE and MBE, are summarized in Table 2. The entries in Table 2 indicate that the CIBSE Guide J method provided lower MBE and RMSE values and so this method was used.

Table 2.

Statistical result of diffuse-to-global solar radiation correlation for the different models with Daejeon measured data (2010).

	MBE (%)	RMSE (%)
CIBSE Guide J (2002)	3.09	37.50
Reindl (1990)	3.50	41.23
Watanabe et al. (1983)	−12.50	58.47
Erbs et al. (1982)	12.67	42.49
Orgill & Holland (1977)	12.27	41.29

MBE: mean bias error; RMSE: root mean squared error.

Weather data selection using the Finkelstein–Schafer statistic

TWD consist of individual months of meteorological data sets selected from different years over the available data period, which is called a long-term measured data series. These are synthetic years based on the most average months taken from a 20-year period. The most average months are commonly selected using the Finkelstein–Schafer statistic (FS) by Finkelstein and Schafer.¹⁹ FS statistics are the most common methodology for generating typical weather data. This method is an empirical methodology for selecting individual months from different years over the available period. According to FS statistics, if a number of observations, n, of a variable x are available and have been sorted into an increasing order x₁, x₂, … , x_n. The FS comparison statistics (between the long-term CDF for each month and the CDF for each individual year of the month) are calculated according to

FS (y, m) = (1 / n) \sum_{i = 1}^{n} | CD F_{m} (x_{i}) - CD F_{y, m} (x_{i}) |

(1)

where CDF_m is the long-term and CDF_y,m is the short-term (for the year y) cumulative distribution function of the daily variable x for the month m, x_i is an ordered sample value in a set of n observations sorted in ascending order, and n is the number of daily readings for the month. In CIBSE TRY and TMY, the representative year for each month of the data set was determined by observing that the representative year has the smallest FS value, i.e.

TRY = min (FS)

(2)

In CIBSE TRY method, the best month is the lowest total FS by the standard deviation. For TMY method, the best month is the lowest total FS by the weights. In ISO TRY method, for each calendar month and each year, add the separate ranks for the three climate parameters. For each calendar month, for the three months with the lowest total ranking, calculate the deviation of the monthly mean wind speed from the corresponding multi-year calendar-month mean. The month with the lower deviation in wind speed is selected as the “best” month to be included in the reference year.

Generic office building

A base case office building was developed to serve as a baseline reference for the comparative energy studies. The base case was a 25 m × 25 m, 10-story building with curtain wall design, 3.4 m floor-to-floor height and 60% window for all four walls (Figure 2). The walls had a U-value of 0.35 W/m²K and the windows a U-value of 2.4 W/m²K¹. The air-conditioned and non-air-conditioned floor areas were 5440 m² and 810 m², respectively. The air-conditioned space had four zones at the perimeter. Table 3 shows a summary of the key building envelope design parameters. The design standard covers all the cities. Since this study is about comparison of the relative energy performance using different weather years. ApacheSIM for the model system calculates hourly heating and cooling energy requirement over a period of 1 year. The heating setpoint and the cooling setpoint are 19℃ and 23℃, respectively. The humidity control is from 0% to 70%.

Figure 2.

Generic office building.

Table 3.

Summary of building envelope parameters. All buildings were considered for Seoul, Daejeon, Daegu, Gwangju, and Busan.

Building	U-value (W/ m² K)
Base case	Wall-to-window ratio: 60%, window overhang: 0 m, window transmittance: 0.9, window U-value: 2.4 W/m²K, wall U-value: 0.35 W/m²K, Roof U-value: 0.25 W/m²K
1	Wall-to-window ratio: 50%
2	Wall-to-window ratio: 40%
3	Window overhang: 0.5 m
4	Window overhang: 1.0 m
5	Window transmittance: 0.7
6	Window transmittance: 0.5
7	Window U-value: 1.8 W/m²K
8	Window U-value: 1.2 W/m²K
9	Wall U-value: 0.26 W/m²K
10	Wall U-value: 0.17 W/m²K
11	Wall-to-window ratio: 40%, window overhang: 1.0 m, window transmittance: 0.5, window U-value: 1.2 W/m²K, wall U-value: 0.1.7 W/m²K

Typical weather data

A TWD consists of 8760 values of various selected meteorological parameters such as ambient temperature, solar radiation, relative humidity, and wind velocity that are originally derived from long-term data. Lee et al.²⁰ generated the typical weather data using the ISO TRY method for six major cities of South Korea. Levermore and Parkinson²¹ had suggested the selection process of the selected year and the algorithm for new Test Reference Years and Design Summer Years for the United Kingdom. Argiriou et al.²² compared the methodologies for TMY generation using 20-years data for Athens Greece. Kalamees et al.²³ suggested the weighting factors for climate variables for selecting the TWD. Wilcox and Marion²⁴ had compared with Sandia Method and TMY weighting values. The ISO TRY, CIBSE TRY, and TMY used in this paper differ in some details (Table 4), but are all based on constructing a year from 12 individual months of real weather data. Using the method described above, the TWDs are generated for major cities in South Korea from meteorological data for 20 years (1986 to 2005). Table 5 shows final selected months for the major cities.

Table 4.

Weather parameters of typical weather data.

		Method
Weather parameter	Statistic	TMY	Weight	ISO TRY	Weight	CIBSE TRY	Weight
Dry bulb temperature	Mean	O	2/24	O	1/3	O	1/3
Dry bulb temperature	Maximum	O	1/24	×		×
Dry bulb temperature	Minimum	O	1/24	×		×
Dewpoint temperature	Mean	O	2/24	O	1/3	×
Dewpoint temperature	Maximum	O	1/24	×		×
Dewpoint temperature	Minimum	O	1/24	×		×
Wind velocity	Mean	O	2/24	O		O	1/3
Wind velocity	Maximum	O	2/24	×		×
Global solar irradiance	Total	O	12/24	O	1/3	O	1/3

TRY: test reference year; TMY: test meteorological year.

Table 5.

Final selected months for the three major cities of South Korea.

Month	Seoul			Daejeon			Busan
	ISO TRY	CIBSE TRY	TMY	ISOTRY	CIBSE TRY	TMY	ISO TRY	CIBSE TRY	TMY
January	1998	1993	1998	1987	1998	1994	1987	1998	1994
February	1994	1999	1994	2001	1994	1994	1997	1994	1992
March	1997	2003	1999	1996	2003	2003	1995	2003	1989
April	1997	1997	1997	1986	1997	1997	1991	1999	1999
May	1993	1993	1999	1986	1991	1994	1993	1993	1993
June	1994	1994	1995	1986	2005	2005	1988	2005	2000
July	2005	2005	2005	1995	2005	1993	1997	1998	1995
August	1992	1992	1987	1992	1993	1992	1996	1999	1996
September	1992	2002	1992	1990	2002	1991	1991	2002	2003
October	1990	1990	1990	2005	1993	1993	1999	1990	1995
November	1996	2000	1994	2004	1996	1991	1998	2000	1991
December	1987	1994	1993	1994	1996	1994	1992	1999	1993

TRY: test reference year; TMY: test meteorological year.

Results and discussion

Comparison of the ISO, CIBSE TRYs, and TMY in simulation

The impact of modeling using the full 20 years of meteorological data base set (BS) was compared to the use of three TWD for five location in the South Korea (Seoul, Daejeon, Daegu, Gwangju, and Busan) by simulating 12 office buildings. The buildings were modeled using a fully dynamic thermal simulation software (a commercial software package), and the range of building parameters is shown in Table 3. The heating and cooling energy used by the air-conditioned 12 office buildings is shown in Figure 3, from which we see that the three TWD slightly underestimate the average total energy use of the heating and cooling plant. Table 6 demonstrates the TMY underestimate cooling load more than other, both ISO TRY and TMY underestimate heating load comparing with CIBSE TRY in Seoul. Except in the CIBSE TRY for Buildings 7, 8 and 9, the three TWD are slightly underestimate the average total energy use of the heating and cooling plant. The maximum difference between the simulated total heating and cooling loads is about 6% in Seoul.

Figure 3.

The heating and cooling load used by the office buildings in Seoul.

Table 6.

Result of base set and three TWD simulation for the heating and cooling energy use (Diff = difference in percentage). Data shown is for Seoul.

Building	Mean annual heating load				Mean annual cooling load				Mean annual heating and cooling load
Building	Base Set (MWh)	CIBSE TRY (Diff %)	ISO TRY (Diff %)	TMY (Diff %)	Base Set (MWh)	CIBSE TRY (Diff %)	ISO TRY (Diff %)	TMY (Diff %)	Base Set (MWh)	CIBSE TRY (Diff %)	ISO TRY (Diff %)	TMY (Diff %)
Base case	195.74	−0.71	−3.65	−3.27	388.77	0.10	−0.05	−3.56	584.52	−0.17	−1.23	−3.47
1	191.44	−0.88	−3.80	−3.58	327.86	0.00	−0.35	−3.89	519.30	−0.32	−1.59	−3.78
2	188.16	−1.07	−3.94	−3.87	267.14	−0.15	−0.81	−4.34	455.30	−0.53	−2.08	−4.14
3	209.48	−1.04	−3.78	−3.62	323.80	0.02	−0.31	−4.02	533.28	−0.40	−1.64	−3.86
4	222.94	−1.39	−3.93	−3.94	271.49	−0.02	−0.58	−4.47	494.44	−0.64	−2.07	−4.23
5	228.86	−2.28	−4.56	−4.72	291.65	−0.04	−0.70	−4.37	520.51	−1.01	−2.36	−4.53
6	263.96	−2.25	−3.77	−4.32	206.81	−0.64	−1.95	−5.78	470.77	−1.53	−2.96	−4.96
7	152.83	−0.12	−3.75	−2.95	416.25	0.33	0.44	−2.98	569.08	0.21	−0.66	−2.98
8	108.56	0.71	−4.22	−2.61	454.63	0.05	0.34	−2.94	563.19	0.17	−0.51	−2.87
9	190.98	−0.21	−3.16	−2.75	390.27	0.10	−0.02	−3.52	581.25	0.00	−1.03	−3.27
10	187.40	−0.64	−3.62	−3.19	391.45	0.10	0.01	−3.48	578.85	−0.14	−1.14	−3.39
11	182.98	−2.44	−4.34	−5.15	106.13	−2.82	−3.89	−8.34	289.11	−2.58	−4.18	−6.30

TRY: test reference year; TMY: test meteorological year.

The simulation of the heating and cooling load of office buildings was divided five components (wall-to-window ratio, window overhang, window transmittance, window U-value, and wall U-value) for energy saving. These values are found in Figure 4. Comparing the five components, window and wall U-values were very close to the base case. However, the wall-to-window ratio had the largest differences, followed by window transmittance and window overhang. When the wall-to-window ratio decreases by 20%, the total load for the office building decreases from 585 MWh to 455 MWh (22.1%). The heating and cooling load is in the case almost linear with the flow as any building type exclude window U-value.

Figure 4.

The heating and cooling load of office buildings by five components (wall-to-window ratio, window overhang, window transmittance, window, and wall U-value) in Seoul.

Building heating and cooling loads

The computed monthly heating loads from each of the 23 simulation runs were analyzed. To ascertain whether the computed heating loads from the TWD follow the long-term mean, the monthly load profile from the TWD was compared with those from the 20 individual years. Figure 5 shows four profiles—the monthly mean loads for long-term mean (i.e. 20-year average) and the three TWD (ISO TRY, CIBSE TRY, and TMY).

Figure 5.

Comparison of monthly heating and cooling load profiles for Seoul, Daejeon, and Busan.

It can be seen that the TWD predictions fall within the minimum and maximum ranges and tend to follow the long-term means quite closely in Figure 5. Daejeon appears to have the largest deviations during the winter months from December to February. The peak monthly heating load tends to occur in January. Not surprisingly, Seoul has the largest long-term mean peak monthly heating load of 59.32 MWh, more than twice that for Busan. The difference between the maximum and minimum profiles can be significant. For instance, in Busan, the maximum and minimum heating loads in January of 41.83 MWh and 28.22 MWh, respectively, differ by more than 67%. To put the errors in perspective, MBE and RMSE were normalized by the respective monthly mean values, and the findings are shown in Table 7. The deviation of 20 years is from 5.95% to 12.02% for heating loads in five cities. In Seoul, MBE ranges from 0.70% over estimation in the CIBSE TRY to 3.52% over estimation in the ISO TRY. RMSE varies from 7.70% in CIBSE TRY to 8.44% in ISO TRY. Busan appears to have a large RMSE and deviation for the three TWD, ranging from 17.54% in TMY to 18.81% in ISO TRY. This is probably due to the small heating requirements in warm winters—a small variation could result in a large percentage difference.

Table 7.

Summary of MBE and RMSE for heating loads.

		Seoul	Daejeon	Daegu	Gwangju	Busan
20 years	Mean (MWh)	195.74	164.36	131.80	134.42	101.01
	Median (MWh)	197.38	162.21	132.94	135.12	98.07
	Deviation (%)	6.63	7.59	7.32	5.95	12.02
MBE	ISO TRY (%)	3.52	7.56	0.47	4.55	7.73
	CIBSE TRY (%)	0.70	2.85	27.62	25.83	−5.07
	TMY (%)	3.17	5.80	1.28	3.09	3.11
RMSE	ISO TRY (%)	8.44	12.18	10.19	8.59	18.91
	CIBSE TRY (%)	7.70	9.96	29.44	26.84	17.99
	TMY (%)	8.30	11.17	10.26	7.92	17.54

MBE: mean bias error; RMSE: root mean squared error; TRY: test reference year; TMY: test meteorological year.

Similar analysis was conducted for the building cooling loads, and the results are shown in Figure 5. Again, it can be seen that the ISO TRY tends to follow the long-term means very closely. The difference among Seoul (long-term max peak load of 98.04 MWh), Daejeon (long-term max peak load of 124.60 MWh), and Busan (long-term mean peak load of 122.65 MWh) is shown in Figure 5. Table 8 shows that the deviation of 20 years is from 6.97% to 9.53% and the median is lower than the mean for cooling loads in five cities. The entries in Table 8 indicate that the ISO TRY provided lower MBE and RMSE values for four of five cities. For humid summer in South Korea, the weighted factor of 0.33 for relative humidity in ISO TRY is better than other TWD. In Seoul, MBE ranges from 0.05% over estimation in ISO TRY to 3.44% over estimation in TMY, and RMSE ranges from 8.96% (ISO TRY) to 9.60% (TMY). All five cities have the MBE ranges from 0.05% in Seoul to 8.50% in Gwangju and the RMSE ranges from 8.51% in Daejeon to 13.09% in Busan. Compared with heating loads, there are smaller variations in cooling load predictions from different weather years.

Table 8.

Summary of MBE and RMSE for cooling loads.

		Seoul	Daejeon	Daegu	Gwangju	Busan
20 years	Mean (MWh)	387.19	472.09	489.35	491.91	465.01
	Median (MWh)	384.76	464.01	474.11	483.34	463.52
	Deviation (%)	7.08	6.99	7.47	6.97	9.53
MBE	ISO TRY (%)	0.05	0.38	2.20	2.30	1.28
	CIBSE TRY (%)	−0.10	3.62	6.10	8.50	4.06
	TMY (%)	3.44	2.06	4.29	1.49	−5.56
RMSE	ISO TRY (%)	8.96	8.51	9.60	9.40	11.92
	CIBSE TRY (%)	8.96	9.24	11.16	12.46	12.52
	TMY (%)	9.60	8.75	10.28	9.23	13.09

MBE: mean bias error; RMSE: root mean squared error; TRY: test reference year; TMY: test meteorological year.

The long-term annual mean total load varies from 633 MWh in Daejeon to 564 MWh in Busan. Predictions from the TWD are very close to the long-term means, ranging 7 MWh underestimation in Daejeon to 30 MWh overestimation in Busan, representing only 1.1–5.32% of the respective annual total load. In general, the TWD predictions tend to follow the long-term means quite closely, and both are well within the maximum and minimum ranges of the 20 individual year predictions. There appears to be some noticeable differences between the TWD predictions and the long-term means in June and August for Seoul. These percentage errors tend to be at the lower end of the 20-year ranges and, once again, indicate that there is good agreement between predictions from the TWD and the corresponding long-term means.

Sensitivity of weather parameters

Global warming in Asia means greater climatic extremes with drier dry seasons and wetter wet seasons. The summer climate of South Korea is characterized by a climate of high temperature and humidity. Table 9 shows that the cooling load of the base case office building changed as the weather parameters (temperature and solar irradiation of summer season) changed over the years for Seoul. 1994 produced the largest cooling load. According to Table 2, Figure 6 shows the cooling load for office buildings based on the components of building envelop by a greatest climate extremes year (1994) for Seoul. The maximum difference among the simulated cooling loads was 191 MWh, or 42.4%, the window transmittance had the best positive effects. Taking all the cases into consideration, the positive effects far outweigh the negative effects exclude the window U-value for a climate of humid and hot.

Figure 6.

The cooling load of the office building by a climate of high temperature and humidity for Seoul in 1994.

Table 9.

The cooling loads of the base case model by changing the weather parameter.

	Temperature		Solar irradiation		Cooling load
	℃	Rank	W/m²	Rank	MWh	Rank
1986	21.39	20	176.72	3	321.30	3
1987	21.50	18	182.08	2	311.04	8
1988	22.47	10	185.91	1	344.32	2
1989	21.99	13	168.00	6	311.39	7
1990	21.90	15	123.98	20	255.45	20
1991	22.23	12	156.60	13	295.35	13
1992	21.58	17	159.83	12	284.12	17
1993	21.48	19	156.56	14	276.24	18
1994	23.61	1	170.43	4	359.95	1
1995	21.90	16	144.15	18	272.82	19
1996	22.64	8	155.20	15	298.18	12
1997	22.76	6	160.84	10	318.98	4
1998	22.78	4	142.64	19	285.11	16
1999	23.05	3	160.74	11	317.82	5
2000	22.76	5	146.53	17	307.93	9
2001	23.25	2	151.78	16	317.03	6
2002	22.28	11	168.90	5	302.76	10
2003	21.98	14	163.97	9	290.27	15
2004	22.69	7	166.97	7	300.74	11
2005	22.52	9	164.55	8	294.98	14
Average	22.34		160.32		303.29
Diff.	5.70%		15.96%		18.68%

Figure 7 shows the relationship between the measured summer and winter mean temperature of each location over the 20 years and the heating and cooling load. The cooling load is considerably increased according to the summer mean temperature, but the heating load is slightly decreased. Figure 8 shows the relationship between measured summer and winter mean solar irradiation and the simulated heating and cooling loads. For the three cities, there are more increased the cooling load than decreased the heating load as the mean solar irradiation increases. It is interesting that the high coefficient of determinations for the solar irradiation parameter is strong so that about 60% of the data is explained by the linear relationships. Figure 9 shows that the relationship with higher cooling loads tend to have lower heating loads for 20 years of the three cities. A reason for scatter in the relationship being Busan with higher summer mean solar irradiation.

Figure 7.

The relationship between measured summer and winter mean temperature and simulated heating and cooling load.

Figure 8.

The relationship between measured summer and winter mean solar irradiation and simulated heating and cooling load.

Figure 9.

The relationship between simulated annual heating load and cooling load during 20 years.

Conclusions

Computer simulation of buildings and solar energy systems is being used increasingly in energy assessments and design. In this paper, TWDs have been generated, using three different methodologies, from hourly meteorological data measured in Seoul Daejeon, Daegu, Gwangju, and Busan. A total of 368 simulation runs were performed and building heating and cooling loads were analyzed. The aim of this study was to compare the simulation results from three methodologies (ISO TRY, CIBSE TRY, and TMY) with those from individual years and their long-term means. It also assessed the importance of five important components of the building envelope (wall-to-window ratio, window overhang, window transmissivity, window U-value, and wall U-value) for the influence on the heating and cooling load and to study the influence of the different weather parameters (temperature and solar radiation).

It was found that monthly load and energy use profiles from the TWD did follow the long-term means quite well. From the yearly average values of MBE and RMSE errors it can be seen that MBE values range between 1.21% and 4.04% and that RMSE values range between 6.18% and 17.68%. The cooling loads indicate that the ISO TRY provided lower MBE and RMSE values for four of the five cities. For the humid summer in South Korea, the weighted factor of 0.33 for relative humidity in the ISO TRY is better than the other TWDs. Comparing the five components, window and wall U-values were very close to the base case. However, the wall-to-window ratio had the largest differences, followed by window transmittance and window overhang. When the wall-to-window ratio decreases by 20%, the total load for the office building decreases from 585 MWh to 455 MWh (22.1%). The cooling load is considerably increased according to the summer mean temperature, but the heating load is slightly decreased. The three cities are considerable increased annual cooling load according to mean solar irradiation. It is interesting that the high coefficient of determinations for the solar irradiation parameter is strong so that about 60% of the data is explained by the linear relationships. The reduction in window transmittance had the best reduction in cooling load, outweighing the changes in the window U-value. However, this is for the South Korean climate of high temperature and high humidity.

Footnotes

Funding

This research was supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Minister of Education (NRF-2013R1A1A2058276).

Conflict of interest

None declared.

References

BS EN ISO 15927-4: 2005. Hydrothermal performance of building. Calculation and presentation of climate data. Part 4: Hourly data for assessing the annual energy use for heating and cooling, London: British Standards Institution, 2005.

Lam

. Climatic influences on the energy performance of air-conditioned buildings. Energy Convers Manage 1999; 40: 39–49.

Gugliermetti

Passerini

Bisegna

. Climate models for the assessment of office buildings energy performance. Build Environ 2004; 39: 39–50.

Westphal

Lamberts

. The use of simplified weather data to estimate thermal loads of non-residential buildings. Energy Build 2004; 36: 847–854.

Enshen

. Are the efficient rates (RVRs) approximate in different climate locations for the same buildings with the same reform? Build Environ 2005; 40: 557–562.

Kolokotroni

Zhang

Watkins

. The London Heat Island and building cooling design. Solar Energy 2007; 81: 102–110.

Mui

Wong

. Cooling load calculations in subtropical climate. Build Environ 2007; 42: 2498–2504.

Karlsson

Rohdin

Persson

M-L

. Measured and predicted energy demand of a low energy building: Important aspects when using Building Energy Simulation. Build Serv Eng Res Technol 2007; 28: 223–235.

Yang

Lam

Liu

. Building energy simulation using multi-years and typical meteorological years in different climates. Energy Convers Manage 2008; 49: 113–124.

10.

Jentsch

Bahaj

James

PAB

. Climate change future proofing of buildings - Generation and assessment of building simulation weather files. Energy Build 2008; 40: 2148–2168.

11.

Radhi

. A comparison of the accuracy of building energy analysis in Bahrain using data from different weather period. Renew Energy 2009; 86(3): 869–875.

12.

Kershaw

Eames

Coley

. Comparison of multi-year and reference year building simulation. Build Serv Eng Res Technol 2010; 31: 357–369.

13.

Chow

DHC

Levermore

. The effects of future climate change on heating and cooling demands in office buildings in the UK. Build Serv Eng Res Technol 2010; 31: 307–323.

14.

Orgill

Hollands

KGT

. Correlation equation for hourly diffuse radiation on a horizontal surface. Solar Energy 1977; 19(4): 337–359.

15.

Erbs

Klein

Duffie

. Estimation of the diffuse radiation fraction for hourly, daily and monthly-average global radiation. Solar Energy 1982; 28(4): 293–302.

16.

Reindl

Beckman

Duffie

. Diffuse fraction correlations. Solar Energy 1990; 45(1): 1–7.

17.

CIBSE. Guide J-Weather, solar and illuminance data, London: The Chartered Institution of Building Services Engineers, 2002.

18.

Lee

Yoo

Levermore

. Quality control and estimation hourly solar irradiation on inclined surfaces in South Korea. Renew Energy 2013; 57: 190–199.

19.

Finkelstein

Schafer

. Improved goodness of fit tests. Biometrika 1971; 58: 641–645.

20.

Lee

Yoo

Levermore

. Generation of typical weather data using the ISO Test Reference Year (TRY) method for major cities of South Korea. Build Environ 2010; 45: 956–963.

21.

Levermore

Parkinson

. Analyses and algorithms for new test reference years and design summer years for the UK. Build Serv Eng Res Technol 2006; 27: 311–325.

22.

Argiriou

Lykouis

Kontoyiannidis

. Comparison of methodologies for TMY generation using 20 years data for Athens, Greece. Solar Energy 1999; 66: 33–45.

23.

Kalamees

Jylhä

Tietäväinen

. Development of weighting factors for climate variables for selecting the energy reference year according to the EN ISO 15927-4 standard. Energy Build 2012; 47: 53–60.

24.

Wilcox S and Marion W. User's manual for TMY3 data sets (revised). Technical Report NREL/TP-581-43156. National Renewable Energy Laboratory (NREL). www.nrel.gov/docs/fy08osti/43156.pdf 2008.