15 February 2013

Can BIM alone be used for Construction?

BIM is promoted as the future, the 'new way of doing things'. There is an assumption by BIM evangelists that the BIM model will be the receptacle of all information about a building, making drawings redundant. But is this really true? Is it even possible?


For a contractor to build a building they need to know what they are building. The various sub-contractors and individual trades need enough information to first price their work, then to do their work. But these people do not just follow a set of instructions, they bring their own expertise. The vast majority of building trades require years of training, they are experts in their own right.
So what information do they need? They need to know the Design Intent of what they are doing. They need just enough information to provide a framework, working out the rest themselves. Provide any more information and it is likely the results will be sub-optimal and inefficient.

Current practice is to provide this information as drawings, schedules and specifications. Only relevant information is provided in these documents. If the location of something is not critical it is not dimensioned. If something has to be provided that complies with a description it is not necessarily included in drawings or schedules. If something is included in a schedule, it may not be included in drawings. Therefore, although these documents have never been a full description of the building, they are adequate to build it.


With a BIM model the building has to be completely modelled. You can't leave out a facade just because it is part of a design construct contract, you have to model something. And if something is in a schedule it has to be in the model. Unless every screw, bolt, flashing, seal etc. is modelled in 3D it is not possible to explain how things go together like a drawn detail does.
But most importantly, how do you explain Design Intent when all you can do is create a virtual object? How do you explain which attributes of the virtual object are critical, important, for guidance, or don't matter at all? For example every object has a precise location in the model, yet the precise location may not be important, and may actually be unpredictable because it depend on decisions that can only be made on site.

LOD (Level of Development) tables are an attempt to overcome this shortcoming. But they are not really up to the task. On the one hand an LOD table is not precise enough (how do you use an LOD table to explain that the height of (some) power outlets are critical, but the location along walls is not?). On the other hand, even with this cut down amount of information, LOD tables are becoming far too detailed and complex to be of practical use. The number of element types in a building run into the thousands, let alone the number of actual objects. To assign a LOD to every one, and then track that it is being followed in the model, for each stage of a project, is just not practical in the real world.


None of this should be a surprise to those that work in the AEC industry. Certainly the authors of BIM software are aware of it.
Take Revit as an example. The model is displayed to us via 'views'. We then add notes, dimensions and other annotation to explain our intent. These views are then placed on sheets that can be printed to paper of electronic format (DWF, PDF etc), and exported as 2D CAD files. Revit utilizes traditional practice to solve the problem of communicating design intent.

Bentley are going a step further. They have an initiative called 'Hypermodel' (as in Hypertext) that shows references to other drawings and documents in the 3D model. It is not so much a method of providing design intent information in a BIM model, more like a hybrid solution, providing access to traditional means of communication directly from within a 3D model (as opposed to just from views, as Revit does).

To go back to the original question - is it possible that BIM models will replace drawings - the answer is yes and no.  Yes BIM models will one day be able to communicate design intent, but no, they won't be BIM models as we know them now. There will have to be some sort of added technology. And it is not just a technological problem. If every BIM software vendor has a different approach we will be no better off. It is not realistic to expect contractors and trades people to be familiar with a whole range of software products just to get the information they need to build a building. There has to be some kind of commonality - like drawings, schedules and specifications.
There are probably technologies out there that could do the job now, but with nothing definitive in sight I would say for the foreseeable future it will still be drawings, schedules and specifications. So what does that mean for the BIM models we do now?


Another way to look at it is to define a BIM model as something that is not intended, by itself, to communicate design intent or how to construct the building.
Rob Snyder, who is involved in Bentley's Hypermodel initiative, has succinctly defined the problem. To him the BIM model is the 'environment'. But this environment is not meaningful unless there are 'statements' about it. For example a drawing with notes, dimensions, details is a 'statement'. It only shows the portion of the 'environment' that is relevant, with added 'statements' to point out things that need to be communicated.

So a BIM model is not another way of documenting a building project. It can not communicate in it's own right. It's purpose is as a resource to create deliverables. To create construction documentation, to run structural, thermal and other analysis, to participate in FM, and a whole range other other uses.


By treating a BIM model as a thing in itself confusion is avoided about where information that can be used to construct the building resides. This also clarifies where your contract documentation deliverables can be found.

That said, a BIM model may well be a client deliverable, but make it clear it is only provided as a resource for others to produce their deliverables. So by all means provide your BIM model to others outside the design team, but strip out all traces of documentation - views, sheets, schedules.
You can do this very quickly in Revit:
  1. Create a 3D view only showing what you want to export.
  2. Place that view of a sheet.
  3. Find the sheet in the Project Browser, right click on it, select "Save as file...".
Keep the view and sheet and next time you only have to do step 3.

And make sure project BIM plans are only be about the BIM model. List documentation (production of drawings & schedules) as a BIM Use, but don't include anything about drafting standards or drawing production. These can be referenced in the BIM Plan, but should be completely separate documents.

So don't treat your BIM model as part of your documentation deliverable, and be clear to everyone, from your boss & colleagues to clients, contractors and trades, of this fact.

01 February 2013

Real Collaboration - Working with Engineers

From comments I have garnered it seems the consensus is that engineers will have to model accurately for BIM to work. There was a lot of talk about the many issues that will need to be resolved, work practices that must change, and fees that need realignment, but no alternative strategy to create true BIM materialised.

In my previous post, Should engineers model accurately, Method 4 - Engineers Model Accurately was the only one that received any support, or even serious consideration.

The most common objection (from engineers) to modelling accurately was that "their fee didn't cover it", or "there isn't enough time", which is the same thing. This may well be true, but upon what basis do you ask for increased fees? One of the benefits of BIM, and in particular accurate BIM, is clash detection and avoidance. But can you really go to your client and say "pay me more and I'll reduce clashes between my designed elements and the rest of the building". What do you think they will say? They believe they are already paying you to ensure there are no clashes. As one commenter pointed out, the A/E industry has been getting away with not providing what they promise - fully coordinated design - for years.

Another common objection, was that using BIM software like Revit is more work than traditional CAD. This is just not true.
Whilst you do things in BIM software you don't do in CAD software, there are more things you don't have to do in BIM software that you do have to do in CAD software. As an example, at the request of the contractor we changed the names of some levels. In Revit this took less than a minute, just a simple text edit in one place. When we told the structural drafter, who is using CAD, he groaned as he now has to find, open, and edit every CAD file where these levels are referenced. I'm dreading telling him we are changing door frame sizes to doors in concrete walls. One parameter change for us, hours of stretching lines and redoing dimensions for him. Properly used (i.e. in the way it is designed to be used) Revit will decrease your workload. You won't get far arguing you should be paid more because your BIM software is harder or more time consuming to use (although you might get paid more if you are one of the few who can use it well).

And I can't avoid mentioning the recurring complaint about architects continually changing the design, which apparently is the bane of every engineer's working life and profit margin. Of course the cause of this really has nothing to do with BIM, or does it?
The name Revit is a portmanteau of 'Revise' and 'it'. It's original creators recognised that designing a building is a process of making changes, to optimize the building's design. And that a product that made making changes more efficient had a market. It was only later, after AutoDesk bought it, that the BIM idea took over. So Revit is software designed for making changes easy. And it does. Change one parameter and you can change the size of say, a door, across the BIM model and all drawings and schedules those doors appear in. Even dimensions on drawings change.
So architects using Revit may indeed be making changes to their model that are more vast, and doing it more often, than when they used CAD. Because they can, and because the job of architects is to optimize the design.
Revit Structure and MEP are Revit Architecture with some extra bits added. It is fundamentally the same software. So if engineers used Revit the same way architects do, they wouldn't have such a problem with design changes. They too could use their software to make vast and frequent changes with little extra work.
And who knows, they might also be able to do a better job, by optimizing the building's structural or services design. With the added bonus of annoying the architect with frequent changes!

What does Model Accurately mean?

Before continuing I just want to clarify some things that some commenters were confused about.
By accuracy I mean elements being represented as geometrically accurate 3D representations at their actual location, rather than symbolic representations. They don't need to be realistic (it is best if they aren't), but they do have to represent their spatial requirements.
What I don't mean is sloppy work, where the intention is to place it accurately but that hasn't happened. All AEC consultants can be guilty of this.
Elements need to be spatially accurate during the design process, not just at the end when construction commences. How else can the architect develop the design so services fit? It amazes me when engineers think they can come in at the end of a design they have had no meaningful input in to and expect their services to fit. Or expect the owner to finance an oversized building just so their services will easily fit and make their job easier.
Another clarification is that no-one expects (except BIM evangelists) every single element to be modelled accurately. As architects and engineers we are doing design intent, not a 100% virtual representation of the finished facility. Elements that are large and have specific spatial or location requirements, like ducts, plumbing pipes to falls, light fixtures, large cable trays, do need to be modelled accurately. But elements that are small and can run anywhere, like supply pipes and cabling, don't require as much accuracy. Some elements fall in between. For example power outlets might have a required height, but the exact location along the wall is not critical. So it is not necessary to model all wall studs just so power outlets can be located to avoid them.

In a practical sense, what can be done?

But it is not my intention to beat up on engineers. I feel their pain. Whilst engineers may not be able to argue for more fees, the reality is their fees probably are too low. In a free market fees paid are based on actual work performed, not what is promised or even should, in theory, be done. The other reality is this is not going to go away. Not all engineers have been forced into BIM projects yet, but eventually it will happen.

Whilst no-one can claim extra fees for better coordination, offering to provide evidence of coordination could be construed as an additional service. By evidence I mean actual clash reports, rather than just the promise that there will be no clashes. Another potential extra is providing IFC models. Both of these would involve additional work (and possibly software) to just to deliver, but the potential is there to value add. There may be other opportunities, properly used BIM software can leverage a lot of down-stream processes.

Utilize your BIM software to improve your work practices and work flows. Unlike CAD, which is just a generic drawing package, BIM software is designed with your workflows in mind. For example Revit MEP and Revit Structure are designed to link into analysis software. Revit has a workflow for monitoring changes in linked models (like the Architect's model) called Copy/Monitor. Some commenters reported a lot of success with Copy/Monitor. Some complained it was 'too hard', but I suspect those people weren't committed to making it work. Don't fight your BIM software, use it to its full advantage.

There is an expectation that BIM will involve collaboration. If a client mandates BIM, you can bet they assume collaboration will occur. But what does that mean on the ground? As I have said elsewhere, my definition of collaboration is mutually beneficial cooperation. You scratch my back, and I'll scratch yours.
If the architects say they want you to accurately place power outlets, ask them to do something in return for you. I hear a lot of bitching from engineers about being dudded by architects, but when I ask them what I can do to help (as the architect) I don't get a definitive request to do anything specific, just more bitching. So don't be afraid of asking, but make sure you are explicit about what you want. Just asking the architect to not keep changing the design is not going to do you any good.

Some suggestions (from a non-engineer):
  • Get some commitment that the architects model will be to a certain quality:
    - existing elements will be edited rather then deleted and remade.
    - objects hosted to level or floor rather than wall, floors or ceilings.
    - each ceiling has its own level.
    and make sure to let them know if they don't keep these commitments.
  • Ask the architect to set up views in their model that you can use. If they won't do it ask if you can come into their office and set them up in their model. Might save you from having to recreate views every time you get an updated model. 
  • Ask the architect to use your families. Or at least start of with your families, it probably doesn't matter if they change a family's appearance as long as all the connectors and your parameters are still there.
I'm sure there are many others engineers have dear to their hearts. Don't be afraid to ask. If you face resistance remind them of their duty to collaborate. Offer to do things that help them. Remember it is all about cooperation and negotiation.

So for all you struggling engineers out there, there is hope. Admittedly there is the hurdle of learning your BIM software, but once through that (and it does end), there are practical things you can do to lessen your load to compensate for having to model accurately.