Computational Construction

Refugee at a Trump Rally

In the construction world, we don’t like computers too much. Big yellow hydraulic powered machines- love them. But technology, not so keen.

In the past, I have been about as welcome as a refugee at a Trump rally when I turn up at a construction site with my models, databases & graphics.

The good news for me is that is is changing, albeit slowly.

Constructing it is more difficult than designing it

I’ll upset my friends by saying this- but as an architect who has spent most of his career working in construction, I do at least have a view from each side of the fence:

Constructing something is much more difficult than designing it.

By this, I mean that for a contractor, it is very easy to loose your shirt with a couple of quite minor errors. I have described elsewhere in this blog that construction is a high risk/low margin business.

For a designer, short of the building/structure collapsing and your professional indemnity insurance company inevitably wrangling their way out, a single project gone bad will not sink your company. An ugly duckling of a building or one with dodgy plumbing might end up in court, or with fewer commissions in the future, but you’ll probably survive.

For a contractor- the financial  losses from one project could wipe out the profits from the last 100 projects, or the entire company. To construct something successfully involves taking an infinte number of permutations of method, program & sequence and finding the few that work.

Winning projects often comes down to who will accept the thinnest margins or biggest risk, but not nessasarily having the best delivery method.

Some aspects of construction are counter-intuitive. For example you can compress (accelerate) a program only so much by putting more resources onto a project. After that, putting on more crews or subcontractors means that they hold each other up and overall progress slows.

In construction- the usual argument for doing something a particular way is usually “that’s the way we have always done it” or there is a single tried & tested accepted method. Generally, the construction industry is not forward thinking or innovative

What does BIM have to do with this ?

BIM in a traditional sense, very little. However, using technology and computing power for a better outcome: a lot

Computational design

Computational design has been around for quite a while.

The basic idea is that you define the logic, constraints, limits or rules and let the computer generate options or come up with an optimum solution.

So in a design context, it might be about juggling the infinite number of permutations of column spacing, beam span/depth/concrete strength against functional or commercial aspects such as lettable area or number of carparks.

You might use the brilliant ‘Grasshopper’ if you are concerned with geometry, or the newer kid on the block ‘Dynamo‘ in a parametric model environment.

So in other words, you design the algorithm and not the solution. The algorithm tells you what the solution is.

Furthermore, there are tools for specific purposes such as ClashMEP  which will automatically route building services such as pipes & ducts, work out alternative routes and so on.

In a construction context, my belief is that similar technology will play a future part in informed decision making.

You are getting ahead of yourself

As with some of the other posts on my blog– I can’t demonstrate exactly what I’m talking about, because the technology does not yet exist. But I’ll try to explain how it might work with a couple of examples:

precast panels for a tunnel, building or other structure:

  • Big or slender panels might be efficient in crane lift or material terms- but are at a greater risk of damage. Perhaps it is quicker and easier to place a greater number of smaller but less fragile panels.
  • Make panels too early and you need acres of storage space, but ‘just in time’ production introduces a risk of precast production issues affecting other work.
  • A remote production facility might be cheaper due to existing plant, but the transport cost and risk of delays in transport outweighs the savings.

pipe spools, cable tray & end devices on a plant project

  • certain spools might be critical in terms of the commissioning regime, but installing early is against the natural construction sequence, or could interfere with other work.
  • A linear installation approach might make sense, but prioritising specific end devices, then going back and filling in less critical ones might be more effort, but accelerate commissioning & testing & therefore overall work
  • Smaller spools might be easier to fabricate and transport, but the increased number of joints increases installation time, potential for leaks and complicates other work such as insulation installation or future operations/replacement.

and so on….

To my mind- you would be able to define these constraints or rules and use technology to come to an informed decision, or generate options.

Computational Construction

Of course, you can’t distill everything down to an algorithm or formula. Sometimes it really does just come down to experience and intuition.

But surely, if you accept my characterisation above of the construction industry, technology would have to play a part in analysis and identification of optimum solutions, to improve profitability & certainty or reduce risk.

So in the case of my examples above, there has to be something more sophisticated than a spreadsheet or the ‘that’s the way we have always done it’ argument.

Even the ability to generate a number of possible permutations or options would be a step forward. I’d see this as a good upstream decision-making counterpart to data analytics (downstream in analysis and reporting on decisions) for design & construction.

I’m going to call this ‘computational construction’ (although I admit this does sound a bit geeky). I’m looking forward to a future where technology will further improve construction delivery.