When you perform solar and thermal calculations in Ecotect, it does not do these for any specific year, but rather a standard 'average' energy year. This is true of nearly all thermal and energy analysis tools. Even the weather files used for solar and energy analysis are usually averaged to better represent long-term conditions. This article explains the basic assumptions behind the 'average' energy year and why this approach is more appropriate than using any particular year.
Solar calculations in Ecotect, like most energy analysis tools, are not designed to accurately replicate the position of the Sun in the sky and it's radiative output during any specific year. Instead, they are designed to replicate as accurately as possible the conditions within a long-term 'average' year for use in comparative analysis. In thermal and energy analysis, direct comparability is the most important consideration as the degree of relative change is often far more informative than the quest for absolute accuracy, which is virtually impossible to achieve.
However, because Ecotect does not allow you to enter a year, some people have questioned how accurate it's solar calculations can be compared to those used by the US National Oceanic and Atmospheric Administration (NOAA), which do vary slightly from year to year.
The Average Year
The concept of the 'average' year is important and needs some explanation. As a year with 365 days does not evenly divide into 52 weeks, each consecutive year starts on a different weekday. For example, the first day of the year in 2010 was a Friday, in 2009 it was a Thursday and in 2008 it was a Tuesday. This means that different years may contain a different number of weekends and/or holiday periods and these may occur at slightly different times within each season.
Whilst the effect of this on annual energy use is likely to be quite small, it is an arbitrary and unnecessary variation that may mask other effects that the designer is more interested in. It also presents a problem when designing annual operational or holiday schedules as these would need to be either manually edited to match each test year and locale-specific holiday sequence, or the logic required to apply generic schedules to any particular year and locale would be very complex and extensive indeed.
Thus, an 'average' year is used, starting and finishing on a Monday. This is how dedicated energy analysis tools such as DOE-2 and EnergyPlus also work – allowing their users to arrange seven daily schedules into a set of characteristic weekly schedules, and then arrange 52 of these weekly schedules over the year.
This does mean that both the very last day (December 31) and the following day (January 01) are both taken to be Mondays, however New Years day is invariably a holiday so the overall effect of this compared to the utility it offers is negligible.
The solar cycle in this case needs to be symmetrical over the year, meaning that you should get exactly the same annual energy results for any consecutive 365 days regardless of the start date. For example, running an analysis from January to January should give exactly the same annual loads as from August to August.
If year-on-year variation was considered, you would get slightly different results in the range January 2009 to January 2010 than from August 2009 to August 2010. This is because there would potentially be a different number of weekends and holidays in Summer (where, in a hot climate, office air-conditioning is a significant load) and the Sun may be slightly lower/higher. Though small in value, this adds unnecessarily complexity to the two results when the very purpose for which they were calculated in the first place was likely to be direct comparison.
Differences in Solar Position
There are many algorithms for the calculation of solar position and they are used in a multitude of applications for a range of different purposes. The accurate replication of observed data is one important use case, but there are others. Due to the complexity of the Earth-Sun geometric relationship, there is no absolute trigonometric solution that captures all aspects of their behavior. In fact, most algorithms use a series of modification factors for the equation of time that are derived directly from best-fit analysis of many years of observatory data.
The main difference between the 'average' year solar position algorithm in Ecotect and those provided by observatory web sites is that one uses a set of fixed modification factors for some of the linear equations involved in the calculation and the other derives it's factors dynamically from the input year value. Obviously there are also simplifications in the mathematics involved because of the use of these fixed factors, but the overall differences between the fixed equations in Ecotect and the NOAA methodology for the year 2010 are less than 0.8 of a degree in azimuth and 0.6 of a degree in altitude angles across the whole year.
The differences in calculated solar position between Ecotect and the NOAA methodology are relatively small and do not significantly affect the shadowing, solar and thermal analysis in which they are used. The benefits of using an average energy year in terms of simplifying the comparison of result sets, on the other hand, are significant.
If you do need a year in order to work out a specific distribution of holidays and weekends, you can use 2007 or 2001 as they both started and ended on a Monday.