Background
One of the goals of OBJECTIVE’s next major version was to offer the community a means of generating a CSA-F280 compliant heatloss modeller as that is now required by NBC 2020. This was the easier task because OBJECTIVE pretty much does this already, as the ‘Static’ or ‘One-Zone’ method which CSA supports in the standard. But we also wanted to test OBJECTIVE to ASHRAE 140’s bench tests so that we can make that claim also, for their Class II or one-zone, static model compliance, and yes, ASHRAe 140 has two paths, dynamic hourly and static methods. But NECB for Part 3 Buildings has set ASHRAE 140 Hourly modelling as the bar for energy modellers. That makes our work a lot harder. To run an hourly modeller you pretty much need a Building to model. OBJECTIVE in essence is a spreadsheet of values that represent the building. So what to do? Give up? Or build a building modeller? Because if you’re in for a penny, you’re in for a pound here. If you make an app that can import an entire building, you’re 90% of the way towards building something you can model a building with. So we’re crazy enough to do both it seems.
If you could Clone Revit or ArchiCad, would you?
There are a few things the ArchiCad and Revit using communities have in common, one is complaining about the high cost of the software, but also usability and feature problems. As a BIM user for over 30 years now, I have seen buglists and feature requests grow and die over decades, but one thing remains always, some of the best ideas that come from Alpha and Beta testing forums and user group meetings die at the side of the road.
Legacy software companies either didn’t have the resources to implement feature X or Y, or they considered the users to be philistines, or worse yet, they heard user concerns but then delivered something that is nearly the opposite of what was requested or some feature that was not even half-baked. As a long-time manager of BIM in large and small offices, I’ve seen how BIM software ships the B and M parts (Building and Modelling) while leaving the I part (Information) almost entirely up to the user. This means decades (I’m not exaggerating) are needed to produce a solid template that can be used to generate real building collateral, full drawing sets, performance modelling options, wbLCA modelling, cost estimates and the like. So what if you could start from scratch, focus on the Information parts first, and then build a robust app around that? This is an approach we can call ‘Building Intelligence Modelling’. To answer the question, many AI-powered startups are attempting ‘clean-room’ clones of Revit and ArchiCad, I’ve tried them, and they are so far, easy to crash and rife with bugs. We feel that’s a missed opportunity- what if we can do better?
What would better even look like?
In short: there is power in simplicity. We can consider this simplicity as a form of leverage, and any lever can multiplies the effort applied. For anyone that can remember hand drafting with pencils and pens, the choice of pen is a primary decision. Pen weights confer to architectural functions, heavy structure gets a heavy pen, light fills or hatches get a hairline pen. Stuff needs to be legible when printed, and nobody had 255 pens on their desk (the pen count of most CAD/BIM apps, multiplied by as many times as one has pen tables). Without getting nostalgic, a return to the simplicity of a restrained group of pen weights and colours would help new users obtain better graphic results because, as per Occam’s razor, there are just fewer things to get wrong. So we started with a limited palette of 30 pens, with the standard 1-10 ramp that CAD users are familiar with. Then assign these weights and colours to materials that are defined in the app. When a material is defined, its full graphic and visual properties should form part of that definition, as well as its physical properties and even cost, and embodied carbon. So that when a building is designed, its cost, wbLCA, and thermal performance can be robustly evaluated at any stage. We have done exactly this: materials drive all display and reporting. Simple, powerful.
Drawings are Foundational.
I was so impressed with the work of the folks developing BlenderBIM aka. Bonsai that my architecture firm even co-funded the development of composite or layered wall, roof and floor assemblies in Bonsai, but after some attempts to learn Blender and Bonsai, it became apparent to me that the development team’s approach to the model was taking resources away from creating quality drawings and document management, and as construction is still heavily dependent on drawings as the ‘work instruments’ of an engineer or architect’s contract sets, I felt this was a wrong approach or a missed opportunity, and so set out to see if we could do better with a native app. Our results speak for themselves.
Geometry as the Principal Innovation.
This may be the hardest bit to explain, as it undercuts hundreds, even thousands of years of inherited bias around how geometry is conceived of and performed in a computer. R. Buckminster Fuller is known for the geodesic dome and other inventions, but few realize his primary contribution to human knowledge was the synthesis of a new way of looking at geometry, with the Tetrahedron (the ‘three sided pyramid’) as its foundation rather than the cube. It is easy to mistake this as a trivial re-framing of the relationships of shapes in space, just a re-labelling of the same Euclidean space, but it actually has profound implications for mathematics and computational geometry, and we take full advantage of it. The core idea is that if one takes the tetrahedron as the fundamental frame to assign coordinates to space, we have 4-axes, ABCD, instead of XYZ. Coordinates are written as [0,0,0,0] instead of (1, 1, 1). This is a much broader topic that is fully described in the link (for more information see our Synergetics Cookbook paper on ResearchGate). In short, with ABCD coordinates and Norman J. Wildberger’s Rational Trigonometry, we can do computational geometry with whole integers, completely bypassing irrational numbers (like √3 and √2, the cube diagonal and face diagonal of a square as two examples), resulting in a computational efficiency, speed AND exactness that conventional trig and XYZ cannot. We use these methods directly in energy modelling to generate our ISO-compliant Isothermal contour modelling within our new app. The way we model energy flow is described in this research paper: Huygens-Fuller-Thomson ‘spherelets’.
Public Alpha
If this interests you, email andy[at]openbuilding.ca and I can send you the link to the download package – we’re going to ship a public alpha version for users to test. It’s far from finished, we don’t have the CSA-F280 or ASHRAE 140 bits baked yet, but we do have the full ISO-compliant thermal solvers for Assemblies, Openings and whole Building Sections shipped now. The versions will run on Apple OSX (M1 Silicon machines or newer), and a Linux build will ship in December, 2026. We’ll record a Zoom intro session with our internal group and will post the recording to youtube – and here – this week.
Leave a Reply