During the day the building temperature was maintained below the upper threshold temperature for comfort。 However, in the early morning the temperature rose above the threshold slightly。 For this particular day the battery retained no (usable) charge from the pre- vious day and also carried no charge over to the proceeding day。 Despite this, the battery allowed the air-conditioner to turn on early in the morning and sufficient charge was collected during the day during the off-cycle to allow some cooling in the late afternoon and evening。

For this particular day, the maximum ambient temperature on the previous day was above 30 ◦C and it is apparent that over- night the building remained quite warm and above the ambient

temperature。 This highlights one characteristic of heavy-weighted construction with limited ventilation; that is, when unconditioned they can retain heat for extended periods and remain more uncom- fortable than the  ambient。

For this case, the annual degree discomfort percentage improve- ment relative to the unconditioned building was 85% with a relatively high number of hours remaining outside the comfort band even in the conditioned building (647 h) as compared to the unconditioned building (4239 h)。 For the conditioned building, the time of day and ambient temperature when these uncomfortable conditions occurred is shown in Fig。 7。 More than one third of the

uncomfortable hours occurred during the early hours of the morn- ing when the ambient temperature was <26 ◦C。

182 M。J。 Goldsworthy / Energy and Buildings 135 (2017)  176–186

Fig。 7。 Histogram showing annual total number of uncomfortable hours by time of day for one particular system in Brisbane。 Colours show corresponding ambient temperatures。

Fig。 8。 Box plots showing pddimpr (expressed as a fraction) as a function of battery size for a wide range of buildings relative to unconditioned versions of the same buildings for several key locations for the case of 3 collector modules。

5。2。 PV and battery sizing for maximum PV utilisation

The influence of battery storage capacity and number of PV modules on the degree of discomfort improvement pddimpr was investigated for the same 1000 buildings and for climate zones around Australia。 Simulations with either two or three PV modules corresponding to rated power outputs of 1 x and 1。5 x times the rated a/c power consumption and with battery storage capacities of 0kWh (i。e。 no battery), 0。2, 0。5 and 1kWh were  considered。

Box-whisker type plots of pddimpr (expressed as a fraction) as a function of battery storage size are shown in Fig。 8 for several climate zones for the case with 3 PV modules。 Note here that for readability the results for the different climates zones are not plot- ted exactly at the corresponding horizontal axis location for the given battery size。 Clearly even a very small capacity battery is ben- eficial since the PV panels can only supply enough power to run the air-conditioner when the irradiance is high。 It is also appar- ent that there is a wide variation in performance between different buildings and, that there is a diminishing return from increasing

Fig。 9。 Contours of mean fractional reduction in degree of discomfort across dif- ferent buildings relative to unconditioned versions of the same buildings around Australia for systems with 2 collectors and 0。5kWh battery。 For example, 0。8 con- tour corresponds to an average 80% decrease in the number of discomfort degree hours。

Fig。 10。 Contours of mean PV utilisation fraction across different buildings around Australia for systems with 2 collectors and 0。5kWh battery。 For example the contour

0。4 corresponds to an average PV electricity utilisation of   40%。