Having spent much of 2009/2010 designing a small extension, and learning about the building design process, it has become clear to me that there is a terrible dearth of Free Software for building design. And I want to do something about it.
There are many aspects to building design: drawing/CAD, energy-efficiency (U-value calculation, thermal design, thermal bridge analysis, solar gain modelling), dew-point anaysis, cost estimation, structural calculations, heating sizing, window design and so on. There is piles of software to help with these various aspects but almost all of it is proprietary and only available for Windows. Very little of it is integrated such that having described a wall construction for U-value calcs you don't have to describe it again in a different bit of software for dewpoint, or cost-estimation calcs.
Most of this software is simple implmentations of documents such as standardised govt energy calculations (SAP in the UK), or standards (e.g on structural calcs, U-value calcs). And writing some software to do it is not at all hard. This would be very worthwhile, but the real gains would come from integrating the modules such that you would only have to describe the construction once, then get all sorts of relevant information calculated. This is called 'BIM' (Building Information Modelling).
A practical problem with many of the bits of software that are available is that they have limited selections of things such as insulation types, or manufacturers' products. Free Software would make it very easy to add more things to such lists so that you still use your modelling software with some interesting new product.
There are various bits of Free Software that can be used, but they are not necessarily easy to use or at all integrated. Work to re-implement a lot of simple tools and to integrate the fancier tools into some kind of cohesive whole would both be very valuable.
I would very much like to hear from anyone else who thinks this is an interesting area and would like to help out. I know a fair amount about the design process and the simpler maths, but have little idea how best to implement some kind of model that would be reasonably generic. I am also very conscious that trying to design the whole thing first will simply result in getting nothing done due to being overwhelmed by complexity. We need simple modular tools for the various problem-spaces and some way of exchanging data. I note that Numpy might be an appropriate implementation mechanism.
One problematic aspect of this is regulatory controls. For example in the UK you need to present a SAP calculation for a new building to show that the design is energy-efficient enough. Making some software to do those sums is not difficult, but the software used in an application has to be certified by the BRE (Building Research Establishment) otherwise it's output doesn't count. This certification costs money and we would have to find the necessary GBP 1000 from somewhere to make the software useful for official purposes. It is still useful without that of course. There are no doubt similar issues in other areas and other countries. I don't think they are too serious, but we may need to argue with rules that unreasonably favour proprietary software.
There seems to be very limited software already available for this stuff, but I have found a few things:
I've written a few spreadsheets for things which really needs tarting up and turning into applications with a helpful UI, as 'spreadsheet' is an almost unmaintainable language and it's hard to give them a nice interface and sufficient genericness
There are lots of proprietary windows apps, some of which run under wine, which give useful ideas about how to arrange interfaces.