battery storage size。 The influence of different building parameters is investigated in more detail  below。

Contour plots showing the variation of mean pddimpr (expressed as a fraction) and the mean PV utilisation fraction around Australia for the cases with 2 collector modules and 0。5kWh storage are also shown in Figs。 9 and 10。 That is, in Fig。 9 the contour pddimpr =

0。8 corresponds to an average 80% decrease in the annual number

of degree hours above the comfort bound (25 ◦C) across all build- ing variations; while in Fig。 10 the contour 0。4 corresponds to an average of 40% of the annual PV electricity collected being used for cooling, again across all building  variations。

The highest annual improvement in mean degree of comfort occurs, not in the locations with the highest total irradiance, but in those locations with high mean irradiance during times when the ambient temperature is also high。 Indeed a regression analysis has 94% of the variation in pddimpr  across climate zones  explained

by the annual mean values of the total horizontal radiation and

temperature both evaluated only when the temperature is above 24 ◦C。 The  former  has  a  positive  correlation,  the  latter negative。

M。J。 Goldsworthy / Energy and Buildings 135 (2017) 176–186 183

Fig。 11。 Comparison of the relative mean influence of various building parameters on the degree of discomfort for PV-AC conditioned buildings for four selected repre- sentative climates。 Shaded regions indicate small (±0。2), medium (±0。6) and large (> ± 0。6) effect sizes。

That is, the best locations for mean comfort improvement averaged across house types are those combining high coincidence of irradi- ance with temperatures just hot enough to lead to a conditioning requirement。

On the other hand, the mean PV utilisation is highly correlated with the number of cooling degree days alone with 77% of the vari- ation across climate zones explained by this number。 The addition of the mean incident solar radiation when the temperature is above

24 ◦C improves the correlation to 83%。 Thus, the best locations for

high PV utilisation are, unsurprisingly, those with a year round cooling requirement。

5。3。 Influence of inpidual building parameters on comfort

The influence of different building parameters on the total degree of discomfort dd was investigated for both the PV-AC condi- tioned buildings and unconditioned versions of the same buildings。 Cohen’s scaled mean difference (d) was used to assess the relative influence of each parameter variation。 The results are summarised in Fig。 11 (PV-conditioned) and Fig。 12 (un-conditioned) for four selected representative climate  zones。

In general, the buildings most appropriate for off-grid PV-AC air-conditioning are those with minimal window area with west- erly aspect, low external thermal mass construction in the tropics and subtropics, and higher thermal mass construction in Victoria and far southern Australia, higher infiltration, low internal loads, low solar absorptivity on external surfaces to reduce radiative heat gains and high interior thermal capacitance to smooth fluctua- tions in internal temperature。 Wall insulation and double glazed windows (particularly if westerly facing) provide some benefit by limiting unwanted heat gain from outside。 However ceiling insu- lation actually slightly increases the mean degree of discomfort in all regions with the exception of one climate zone centred on the Kimberly in Western Australia (Although solar radiation and hence heat gain through the roof is very high in this region, it is unclear whether this is an anomaly due to incorrect weather data since the results at nearby climate locations did not show such a large varia- tion)。 The reason that ceiling insulation generally slightly increases