in the Australian Building Code [37] for most climate  zones。

The lower solar absorptivity value for the external surfaces of walls and roof was chosen to be 0。3 corresponding to a very light colour, for example white or cream。 For walls the upper value was selected to be 0。75 corresponding to a strong colour such as red or green since very dark colour walls are uncommon。 However, for the roof an upper value of 0。9 was selected since a number of houses in Australia are constructed with black  roofs。

Common wall construction types for residential buildings in Australia are lightweight (weatherboard, timber and fibro), brick veneer and heavyweight (double brick, stone and concrete block)。 The overall proportions of these types were estimated to be 36%, 35% and 29% using data from 1986 [6] with the proportions vary- ing by state, building type and year of construction。 In this study we have used these three wall construction types directly to represent the typical variation in wall  construction。

Window area values were taken as 10% and 20% of the total floor area of the room。 These values are slightly below typical ranges for residential buildings [6] based on the fact that here we do not consider shading (either internal or external) and taking into account the calculated maximum allowed window  areas in the Australian Building Code [37] as a function of orientation and window type。 Here two different window types were considered; a single glazed clear window with overall SHGC of 0。85 and    U-value

of 5。68 W/m2/K typical of the majority of windows in residential

buildings in Australia, and a higher performance double glazed window with SHGC of 0。7 and U-value of 3。0  W/m2/K。

For each climate zone or system parameter considered, simu- lations were then run, both with and without conditioning, with 1000 different building models composed with a random selection of these parameter values。 For comparing the mean influence of inpidual parameters on comfort the scaled mean difference in the performance metrics were calculated using the method of Cohen [38]。

This difference was evaluated with the mean of the right most parameter value in Table 1 subtracted from the mean of the left- most parameter value。 That is, a positive scaled mean difference in dd for thermal capacitance parameter would indicate that higher thermal capacitance leads to better performance (i。e。 less discom- fort) on average, while a negative scaled mean difference would indicate that the higher thermal capacitance leads to worse per- formance on average。 For the parameters with more than two values, the scaled mean difference was calculated as the differ- ence between the maximum and minimum parameter means with a positive value again indicating better performance for the param- eter value listed to the right in Table 1。 In this way, a quantitative indication of the relative average influence of the inpidual param- eter variations can be obtained。

For evaluating the correlations between parameters, an ANOVA study  was  conducted  with  second  order  parameter interactions

13

included resulting in 91 (。n) different source terms from the 13

n=1

parameters。 The significant interactions with the 5-largest effect

sizes were then investigated for several climate zones and the results discussed。

Fig。 6。 Performance of an uninsulated brick veneer zone with North facing aspect over a summer day in Brisbane for a system with 2 × 300W modules and 1kWh battery storage。

5。 Results

5。1。 Typical day

Before comparing annual results between systems it is instruc- tive to look at the behaviour of one system on a typical summer day in the sub-tropical climate of Brisbane。 Fig。 6 shows the ambient and indoor temperature variations along with the power output of the PV array, the power consumption of the air-conditioner and the battery state of charge over a 24-h period for a system with 2 collec- tor modules and a 1kWh battery located in Brisbane。 The building model for this case consisted of a thermally inefficient uninsulated brick veneer construction with a natural infiltration rate of 2 air- changes per hour, 1080 kJ/K interior thermal capacitance, north facing 2。5 m2 unshaded single glazed window and internal loads as given in Fig。 3。