AN AUSTRALIAN CLIMATIC DATA BANK FOR USE IN THE ESTIMATION OF BUILDING ENERGY USE. P J WALSH*, M C MUNRO', and J W SPENCER

Transcription

1 AN AUSTRALIAN CLIMATIC DATA BANK FOR USE IN THE ESTIMATION OF BUILDING ENERGY USE by P J WALSH*, M C MUNRO', and J W SPENCER COMMONWEALTH SCIENTIFIC AND INDUSTRIAL RESEARCH ORGANIZATION DIVISION OF BUILDING RESEARCH

2 AN AUSTRALIAN CLIMATIC DATA BANK FOR USE IN THE ESTIMATION OF BUILDING ENERGY USE SUMMARY The structure of a nationally acceptable climatic data base comprising 18 Australian stations in a form suitable for use in heat transfer calculations in buildings is outlined. Solar radiation and surface meteorological parameters are merged in such a way as to make the time of their measurement as coincident as possible. A great deal of data must be estimated in order to produce hourly values, and such estimation has been done in a manner believed to be consistent and reasonable. 1. INTRODUCTION In 1978 the National Energy Research Development and Demonstration Council (NERDDC) sponsored a project entitled 'Estimation of Building Energy Use' to be undertaken jointly by the Divisions of Building Research and Mechanical Engineering. (The latter Division was abolished in September 1981 and the new Division of Energy Technology was created in its place.) The aim of this project was the development of nationally acceptable computer-based design procedures for the estimation of energy consumption in both residential and commercial buildings. An important part of this project involved the compilation of a climatic data bank which can be used with these procedures. The two Divisions had been involved in such a compilation for a number of years. The resultant data had been used for the Divisions' own research purposes but had also been made available to other groups on a commercial basis. A decision was made to compile a data bank which had national acceptability as its basis, and which was generated with the potential user in mind. This compilation has been done, using the earlier work as its basis. The work reported here bears a close similarity to ongoing projects overseas, since the need for comprehensive climatic data for analysis purposes is internationally recognized. Thus the US Department of Commerce (1978) has developed the SOLMET climatic data format, in which available insulation and meteorological data are merged into a single source. The present work cannot match the scope of the SOLMET project. Rather it has been framed to suit some of the specific needs of Australian users after consultations had determined these needs. The computer-based design procedures mentioned above entail, in the first instance, hour-by-hour modelling of the various relevant heat transfer phenomena over at least one year using real climatic data for the locality of interest. The climatic data requirements are for hourly coincident values of dry bulb temperature, wet bulb temperature (alternatively, dew point temperature or absolute moisture content), wind velocity, cloud cover and solar radiation. Such data should be continuous over a period of several years to enable local climatic characteristics to be determined with reasonable accuracy. The data requirements for solar radiation are necessarily more complex and hence more difficult to satisfy than for the other parameters. However, the minimum requirement is for hourly totals of global and diffuse radiation on the horizontal plane. The Australian Bureau of Meteorology maintains separate observation networks for the measurement of the standard surface meteorological parameters (about 900 stations) as well as solar radiation (about 20 stations) across the continent. Observations at threehourly intervals of the standard meteorological parameters from about 60 stations are made available on magnetic tape. Observations at halfhourly intervals of global radiation (20 stations)

3 and diffuse radiation (11 stations) are also available on magnetic tape. The task was to take the data available from the Bureau for those stations with measurements of the standard meteorological and solar radiation parameters, and to make these data coincident at one hourly intervals. In so doing we have departed significantly from the Bureau's standard formats in order to provide data in a concise form. All missing data have been replaced with values considered reasonable. Details of the solar position at each hour, calculated using well known formulae, have also been supplied. Solar position is specified in terms of altitude and azimuth. 2. LOCALITIES INCLUDED IN THE CLIMATIC DATA BANK At the time of writing, there were 18 stations on the Australian mainland or in Tasmania where the Bureau of Meteorology had several years of solar radiation records. These data, together with surface meteorological data for the same 18 stations, were used to compile the Australian Climatic Data Bank. The localities of the 18 stations are listed in Table 1 together with the period of time for which data are available. For many building energy-use estimations the practitioner requires a 'typical year' of data. Unfortunately, a year that is typical for one type of building may not be typical for another. type of building. Nevertheless, to give the user of the tapes some guidance in this matter typical years are listed in Table 1 for some of the localities. These were derived (Walsh, Gurr and Ballantyne 1982) by calculating heating and cooling loads for several years for a small residence and choosing as the 'typical year' the one that gave loads closest to the mean load for both heating and cooling. 3. DATA STRUCTURE AND FORMAT In this section the layout of the data for any location is described briefly. Fixed-length records corresponding to data for one day are used throughout. Each hour of data occupies 60 characters and thus each record contains 1440 characters. Characters are blank, and will allow the user to insert any information of his own. The term 'hour' here refers to local standard time, which will trail local clock time by one hour when daylight saving is in operation. In generals the standard meteorological parameters are measured every three hours and thus two out of every three of these values have been estimated. In contrast, solar radiation measurements are available on a half-hourly basis. For both the standard meteorological parameters and solar radiation, measured data will on occasions be missing and have again beer, estimated. Table 2 provides the format of data for each hour. Note that flags are provided to indicate, whether the various values have been measured or estimated. Table 3 indicates the possible values for flags relating to the standard meteorological parameters while Table 4 indicates the possible values for the flag relating to solar radiation. Table 5 gives a listing of a sample day of data, viz 1 December 1978, for Melbourne. The remainder of this report discusses various aspects of the compilation of these data. 4. SOLAR RADIATION Solar radiation is generally measured as irradiation or radiant exposure (the surface density of the radiant energy received, with units of Wh/m 2 or J/m 2 ) on a given plane, most often the horizontal. Such measurements are integrated totals. Of course the standard meteorological parameters of temperature, pressure etc. are essentially instantaneous values. Note that irradiation measurements are normally taken at local mean solar time (MST), whereas the standard parameters are measured at local standard time (LST). As mentioned previously, the Bureau of Meteorology supplies digitized climatic data of both these types, though separately. The solar radiation data consist of global and diffuse irradiation on the horizontal plane taken on the hour and half-hour MST. The other data are supplied as three-hourly readings

4 on the hour LST. Therefore it is impossible to obtain truly coincident values by combining these data. We can merely attempt to approximate coincidence. This we do by associating hourly meteorological readings with a time as near as possible to the mid-point of the solar hour pertaining to a given irradiation value. The same convention is adopted in the production of SOLMET data. For Australian data it is possible to do this with an error of no more than 15 minutes, since half-hourly irradiation values are available. In effect, the adoption of this convention is equivalent to an assumption that the 'time' of measurement of a given irradiation value is the mid-point of the hour over which the integrated total applies. The irradiation value in Wh/m 2 is numerically equivalent to the mean irradiance in W/m 2 for the hour centred on this measurement time. (By irradiance we mean the radiant flux per unit area incident on a surface, with units of W/m 2. This, of course, is an instantaneous value.) Since energy calculations most often make use of irradiance rather than irradiation, the data are tabulated in these units. These measurement times can be converted to LST by subtracting the longitude correction for the given location from the MST for the irradiance. The longitude correction is defined in minutes as four times the difference between the longitude and standard longitude values. Thus, for Melbourne, with a longitude of 1450 and a standard longitude of 1500, 1130 MST is equivalent to LST. Since the Bureau of Meteorology supplies half-hourly values of irradiation, it is possible to take hourly values ending either on the hour or on the half-hour MST such that when the mid-point of the hour is converted to local standard time it will be within 15 minutes of the hour LST and hence irradiance values will be approximately coincident with temperatures etc. Thus, for Melbourne, the values of irradiation for half hours ending 1130 MST and 1200 MST when added give an irradiation for the hour ending 1200 MST. The mid-point of this hour, viz MST, converts to LST and we can do no better than this. However, for Hobart, with a longitude correction of - 11 minutes, the mid-point of the hour ending 1200 MST converts to 1141 LST. We can do better than this by taking irradiation for the hour ending 1230 MST. The midpoint of this hour converts to LST. Table 6 indicates what this means in practice for a variety of stations. Thus, for the solar irradiance values appearing on the tape as coincident with temperatures etc. at 1200 LST, column A indicates the LST at which this value actually applies (the midpoint of the hour), while column B indicates the MST for the end of the hour over which the integrated total applies. In building heat transfer calculations it is important to be able to estimate the direct solar irradiance falling on a surface of arbitrary orientation and inclination. The direct irradiance on any surface at a given time can be calculated readily if we know both the solar position (altitude and azimuth) and the direct irradiance on a plane normal to the solar beam at the given time. Thus, at every hour, in addition to tabulating global and diffuse irradiance (W/m 2 ) on the horizontal plane, we have calculated and tabulated direct irradiance (W/m 2 ) on a plane normal to the solar beam as well as the solar altitude a and azimuth -i. These latter two quantities are given in degrees. The direct irradiance on a surface normal to the solar beam Gb is given by Gb = G bh /sin α where G bh is the direct irradiance on the horizontal plane. G bh is numerically equivalent to the difference between the global and diffuse irradiation values for the given hour, in effect the average irradiance over the solar hour, whilst α and γ are calculated at the mid-point of the solar hour using the methods of Spencer (1965, 1971). Note that if the hour includes sunrise or sunset (i.e. the first and last non-zero readings of irradiation for the day) then the solar position is calculated at the mid-point of the interval between sunrise (or sunset) and the end of the hour.

5 The above relationship for G b may prove erroneous if the solar altitude is very small, i.e. around sunrise and sunset, giving abnormally high values for the direct normal beam irradiance. If sin a < 0.2, a calculation is made of the theoretical clear sky direct irradiance (refer, e.g., Spencer 1974) normal to the solar beam at the given time. If the value calculated from the measured data exceeds the clear sky value, then it is set equal to the clear sky value. 5. ABSOLUTE MOISTURE CONTENT It was decided that absolute water or moisture content rather than wet bulb or dew point temperature should be included in the list of standard surface meteorological parameters. This is a quantity that may be calculated from a knowledge of wet bulb (or dew point) temperature, dry bulb temperature and atmospheric pressure. The steps involved in its calculation depend on whether wet bulb or dew point temperature is available from the Bureau of Meteorology source data for the particular location. The steps are as follows: a) If the wet bulb temperature Twb is available, calculate the saturation vapour pressure P S at the wet bulb temperature in mb from P S = exp ( Twb/( Twb)) and then calculate vapour pressure (Pv) in mb from P V = P S (Tdb -Twb)(' + Twb/610)Pa where Tdb is dry bulb temperature ( C), and Pa is atmospheric pressure in mb. b) If the dew point temperature Tdp is available, but the wet bulb temperature is not, calculate the vapour pressure in mb from P V = exp ( Tdp/( Tdp)). c) Using the vapour pressure calculated from either a) or b), calculate the absolute moisture content w in g/kg from w = Pv/(Pa - Pv). These relationships have been supplied by the Bureau of Meteorology and relate specifically to the Bureau's standard psychrometric measurements. Note that wet bulb temperatures are measured with naturally ventilated (i.e. nonaspirated) thermometers. Also, all vapour pressures are calculated with respect to water, not ice. 6. MISSING VALUES We want to supply data at one-hourly intervals, and yet the normal surface meteorological parameters are at best supplied at three-hourly intervals. Furthermore, three hourly measurements may on occasions be missing, while whole days of solar radiant exposure, either diffuse or global or both, may also be absent. It is necessary to replace all of these missing values in a manner that is reasonable both climatologically and in terms of the software required to accomplish the replacement. 6.1 Converting three-hourly data to onehourly data Linear interpolation has been used to produce one-hourly data from three-hourly data for all the normal surface meteorological parameters. The original three-hourly data are applicable to the hours 0 (midnight), 3, 6,9, 12 (noon), 15, 18, 21 local standard time during non-daylight-saving periods, and to the hours 2, 5, 8, 11, 14, 17, 20, 23 during daylight-saving periods. The distinction between the original three-hourly data and the interpolated data is quite clearly made by the flags in columns for each 1 hour (60 character) data unit. These flags also serve to indicate whether or not daylight saving is in operation. Thus, if the flag in column 28 takes the value 2 at hour 3 for a given day, then

6 daylight saving is in operation, otherwise it is not. Non-linear interpolation is certainly not warranted for any parameters other than dry bulb temperature. In this case, linear interpolation might cause some undue truncation of diurnal temperature swings. Any non-linear method of interpolation should, if possible, incorporate some information about the daily maximum and minimum temperatures, which are recorded by the Bureau of Meteorology on a separate max./min. thermometer. Unfortunately, the times of occurrence of the daily maximum and minimum temperatures are unknown. Various non-linear methods of interpolation have been tried but none appears to offer significantly better results than does linear interpolation. The US SOLMET tapes also incorporate linearly interpolated data between three-hourly measurements. 6.2 Missing measured values - standard meteorological data Dry bulb temperature If only one measured (three-hourly) value is missing, the value is estimated using linear interpolation. If two successive (three-hourly) values are missing, linear interpolation is clearly unworkable. In this case we use a model of the following form: T t = T + a t (T max - T min ) where T t is the temperature at hour t, T max and T min are daily maximum and minimum temperatures, and at is a function of t. From a brief examination of some three-hourly data, it seems that reasonable approximations to T t can be formed under the assumption that at is constant for a given month and hour. Thus, the above relationship is used to replace the second of the two successive missing values, using a precalculated value of at averaged for the given month and hour. The first of the missing values is then estimated using linear interpolation. Such a procedure can be used to cope with any number of missing values Absolute moisture content Again linear interpolation is used if only one (three hourly) value is missing. If two successive values are missing, the second is replaced by an average value for the particular hour and month and the first then estimated using linear interpolation Other meteorological data The following considerations apply to atmospheric pressure, wind velocity and total cloud cover. If only one (three-hourly) value is missing, linear interpolation is used. If two successive values are missing, they are both replaced by the last measured value Calculation of average data - allowing for daylight saving Average data for a given hour and month are sometimes needed to replace missing three-hourly values as outlined above. These data are also of some interest in their own right. When averaging data over a period beginning, say, in 1970 and ending in 1976, accurate averages for the months October through to March may be difficult to obtain since in daylight-saving periods beginning late 1972, three-hourly observations were taken at different local standard times (refer Section 6.1). A pragmatic approach has been adopted in dealing with this problem. The difference is ignored for the months of October and March, for which very few days ever involve daylight saving. This is because the daylight saving period has commenced on the last Sunday in October and ended on the first Sunday in March. For the months November through to February, in calculating averages only, we assume that an 0200 observation when daylight saving is in operation is equivalent to an 0300 observation when daylight saving is not in operation, 0500 is equivalent to 0600, etc. 6.3 Solar radiation Missing measured values

7 Two possibilities need to be distinguished. The first is when both global and diffuse irradiation on the horizontal plane are missing for a given day. Davies and Hay (1978) present a comprehensive summary of theoretical and empirical methods for estimation of solar radiation under cloudy conditions. Theoretical methods have not been used to estimate hourly values of irradiations; the use of empirical methods in such estimations is the subject of considerable debate. No detailed evaluation of such methods has yet been undertaken for Australian locations. One of the more common forms of empirical methods utilizes sunshine hours data, and takes the form H/H C = a + bn/n where H is the day total of global radiation on the horizontal plane, H C, is the clear sky equivalent of H, n is the number of sunshine hours for the day and N is the clear sky equivalent of n. Where sunshine hours are available, such a relationship is used to estimate H for days when global measurements are missing. The coefficients a and b are determined for each month and location by linear regression. The hourly values of global irradiation for the day are then the average hourly values for the month factored by the ratio of H for that day and the average value of H for the month. This model is crude but at the present stage we cannot justify doing anything more complex. If no sunshine hours data are available for the location, then both global and diffuse irradiation are replaced by average hourly values for the given month. The second is when only global measurements are available. In this situation we need to be able to estimate diffuse irradiance on the horizontal plane. A perusal of the recent solar energy literature reveals a number of different methods available for splitting hourly global irradiance measurements on the horizontal plane into direct and diffuse components. A survey recently carried out by Spencer (1982) using global and diffuse measurements from the Bureau of Meteorology has suggested that the method of Orgill and Hollands 0977) provides the most acceptable error distribution for hourly direct irradiance values. The method is as follows: For G/E < 0.3 D = ag 0.3 <, G/E < 0.75 D = (b + cg/e)g 0.75 < G/E D = d G where G is the hourly measured value of irradiance on the horizontal plane; E is the extraterrestrial radiation for the particular time and location; D is the hourly estimated value of diffuse irradiance on the horizontal plane; a, b, c and d are predetermined coefficients characteristic of the location. As recommended, these coefficients are determined for each location. Where no diffuse data are available for a given location, the coefficients are calculated from measurements for a location with similar patterns of solar radiation. Whichever method is used to calculate direct and diffuse components from a measurement of global irradiance, it must be appreciated that the calculation is inherently statistical in nature and therefore large errors will at times be inevitable Doubtful measurements A close check has been made to ensure that no diffuse measurement exceeds the corresponding global measurement for the given location. When this does occur the diffuse component is generally set equal to the global measurement (i.e. direct component 0). If, however, the 'negative' direct component has an absolute value exceeding 50 W/m 2, then the diffuse measurements for that entire day are abandoned and replaced by values estimated in the manner outlined above. 7. ACKNOWLEDGEMENTS The continued assistance of the Director of the Bureau of Meteorology and his officers, particularly Mr Dan Lee and Mr Peter Shaw, in making available sources of climatic data and in

8 providing advice on their analysis is gratefully acknowledged. The financial support of the National Energy Research Development and Demonstration Council has played a vital role in enabling this work to proceed. 8. REFERENCES Davies, J.A., and Hay, J.E. (1978). Calculation of solar radiation data. Proceedings Canadian solar radiation workshop. Orgill, J.F., and Hollands, K.G.T. (1977). Correlation equation for hourly diffuse radiation on a horizontal surface. Solar Energy 19 (4), Spencer, J.W. (1965). Calculation of solar position for building purposes. CSIRO Aust. Div. Build. Res. Tech. Pap. No. 14. Spencer, J.W. (1971). Fourier series representation of the position of the sun. Search 2, 172. Spencer, J.W. (1974). Melbourne solar tables, SI Units. CSIRO Div. Build. Res. Tech. Pap. (Second Ser.) No. 7. Spencer, J.W. (1982). A comparison of methods for estimating hourly diffuse solar radiation from global solar radiation. Solar Energy 29(1), U.S. Department of Commerce, National Climatic Center (1978). SOLMET Volume 1 - User's Manual TD-9724: Hourly solar- radiation - surface meteorological observations, Department of Energy Contract No. E(49-26)- 1041, Asheville, NC, 44 pp. Walsh, P.J., Gurr, T.A., and Ballantyne, E.R. (1982). A comparison of the thermal performance of heavyweight and lightweight construction in Australian dwellings. CSIRO Aust. Div. Build. Res. Tech. Pap. (Second Ser.) No. 44.

9 Table 1: Australian Climatic Data Bank State No. of or Locality Ident. Latitude Longitude Start Finish No. of Records Typical Territory (S) (E) date date months (days) year New South Wales Wagga + WA 35 10' 147'28' Williamtown* WI 32 49' 151'50' Northern Territory Alice Springs AL 23 49' 133'53' Darwin* DA 12 6' ' Queensland Longreach* + LO 23 26' ' Rockhampton + RO 23 23' ' South Australia' Mt Gambler MG 37 49' ' Oodnadatta* OO 27 33' ' Tasmania Hobart HO 42 53' ' Victoria Laverton LA 37 53' ' Melbourne + ME 37 49' ' Mildura + MI 34 14' ' Western Australia Albany + AB 34 57' ' Forrest* + FO 30 50' ' Geraldton + GE 28 48' ' Hall's Creek* HA 18 14' ' Perth PE 31 57' ' Port Hedland + HE 20 23' ' *All diffuse radiation values estimated for these localities. Solar radiation data for Perth was taken at Pearce until and at Guildford th,,reafter. No sunshine hours data available for these localities. Ident is first two characters in every record.

10 Table 2: Data format for one hour Characters Item 1-2 location identification (e.g. ME represents Melbourne) 3-4 year (e.g. 67) 5-6 month (ie. 1-12) 7-8 day (ie. 1-31) 9-10 hour standard (ie. 0-23, 0= midnight) dry bulb temperature (10-1 C) absolute moisture content (10-1 g/kg) atmospheric pressure (10-1 kpa) wind speed (10-1 m/s) wind direction (0~16; 0 = CALM, 1 = NNE,...,16 N) 27 total cloud cover (oktas, 0-8) 28 flag relating to dry bulb temperature 29 flag relating to absolute moisture content 30 flag relating to atmospheric pressure 31 flag relating to wind speed and direction 32 flag relating to total cloud cover 33 blank global solar irradiance on a horizontal plane (W/m 2 ) diffuse solar irradiance on a horizontal plane (W/m 2 ) direct solar irradiance on a plane normal to the beam (W/m) solar altitude (degrees, 0-90) solar azimuth (degrees, 0-360) 50 flag relating to global and diffuse solar irradiance blank Table 3: Values for flags relating to standard surface meteorological data (columns 28-32) 0 means that the value is a three-hourly measurement 1 means that the value is estimated to replace a missing measurement 2 means that the value is an interpolation between three-hourly measurements Table 4 Values for flag relating to solar radiation data (column50) 0 means that both global and diffuse irradiance values are based on measurements 1 means that both global and diffuse irradiance values are estimated to replace a missing measurement 2 means that the global irradiance value is based on measurement but the diffuse irradiance value is estimated to replace a missing measurement.

11 Table 5: Compiled climatic data for 1 December 1978 for Melbourne Table 6: Longitude correction for the eighteen locations. For the radiation data given as coincident with the normal meteorological parameters at 12 noon, column A gives the LST for the mid-point of the solar hour over which these radiation data apply, while column B gives the MST for the end of the solar hour, i.e. the time at which the original solar measurements were made Station Longitude A B name correction (min) (LST) (MST) Albany Alice Springs Darwin Forrest Geraldton Hall's Creek Hobart Laverton Longreach Melbourne Mildura Mt Gambier Oodnadatta Perth Port Hedland Rockhampton Wagga

12 Williamtown