January 16, 2014

The Paradox of IFC

I had an interesting discussion with mr. Velez, the head developer IFC from Autodesk. Which I thought would be nice to share with you all...

I reached the conclusion that there's an interesting loophole in place when it comes to the acceptance of IFC in the AEC industry. I could even say in the Revit-using part of it, but that would probably sell the non-Autodesk disbelievers short. So let's keep it general for now. Don't know how Angel feels about my philosofical insights yet, since he's sleeping now (hopefully). But here it goes:

Standard conversation.

Let's lay out the standard IFC-conversation I have with a lot of people not wanting to use IFC:

Them: We need partners to work with Revit
Me: Why?
Them: because we can't communicate if they don't, which kinds of defeats the purpose of BIM
Me: Why won't you use IFC then?
(from this point onward, the discussion is the same for people using other software, regardless on whether they use IFC or not)
Them: because there's always a loss of information
Me: No there isn't
Them: Yes there is.
Me: Maybe when you do it, I can get any piece of information I want
Them: No you can't
(btw, I never said these conversations are very intellectually challenging)
Me: Have you tried?
Them: Yes, we exported an IFC once. It sucked. All kinds of information missing.
Me: What settings did you apply?
Them: Settings...?
And so on...
Point is: IFC has difficulty in getting accepted because of the supposed loss of information when exchanging models. And that's partly right, however for the wrong reasons. Insufficient implementation in (ALL!) software is what's causing this, not IFC itself. The format is capable of carrying through each and any piece of information you would ever put in your model, and much more. Losless. It's just not implemented that way.
Getting this done is actually the easy part of IFC implementation. It's just a matter of mapping. Which native parameter goes into it's IFC counterpart. And which ones don't have an IFC counterpart? Nice to know, since IFC does allow you to create custom properties for any and all objects.
It's supposed to be relatively simple...

So why aren't we fixing this?

Basically because everybody keeps focussing on getting all the geometry right.
And off course this is important. You don't want to be losing all kinds of objects. But seriously: how often does this happen? 
Let me rephrase: how often does it happen and NOT because of the guy modelling it?

Cause I haven't seen it yet. I haven't seen ANY project where the loss of geometry was NOT caused by the way stuff was modelled, exported, or mapped. Period. And I've seen a lot, trust me. So maybe I haven't seen those few projects where there actually is an inexplainable loss of geometry. But it can't be more then a handful.
Yet there seems to be a huge imbalance between geometry and information in terms of focus and budget. As in: the focus is almost solely on geometry, leaving the information part with improvements that require no more then an hour of programming time max.

Is geometry that important?

No it isn't. Not in my opinion anyway. Now don't start with that "but you're just a consultant, we have real projects" bull. My clients aren't toying around all day and I get paid to deliver results, just the same as anyone else. No results = byebye consultant (even more easily then with employees I might add). It's not about projects, it's about the process.

The majority of firms uses IFC throughout the project as a means of communicating with different project partners. So you get an update every other week or even weekly. Those are used for clash detection, design sessions and so on.

By definition there is stuff missing, incomplete or not there. You're still in the design phase for crying out loud. It's supposed to be that way! Seriously, who cares whether stuff is missing because of the fact it's yet to be designed, or because something went wrong in the export. In fact, there's only one point in time when the IFC needs to be perfect, and that's when it's time to create the deliverable. And I'd rather experience all those errors upfront instead of upon delivery. That way they pop up during design sessions. And I can fix them...

So no, I'm not saying we should except the fact that stuff might go wrong. I am however saying that the impact of those errors is grocely overrated. The first thing I say when someone starts telling me horror-stories about IFC's not being correct is "wow, what did that cost you in penalties from your clients?". I usually get a blank stare... Followed by "uhm, they never knew. We found out during our first design meeting and fixed it".

Fact of the matter is: these are tools. Tools fail sometimes. Revit fails some times. Project files get corrupted all the time. Stuff goes missing, gets omitted from schedules due to wrong filters or modelling methods. Gets accidentally deleted because it's on the wrong Level or a Linked element that is remodelled.
How can someone using Revit tell me that they intend to focus solely on geometry when it comes to IFC until it's 100% reliable? Like Revit is? But when it comes to our primary tool, we build safeguards in our process. So why regard IFC differently?

Now for the Paradox.

On the one hand it's usage in the (Revit-minded) AEC industry is being held back by the general consensus that you lose information when exchanging IFC's. On the other hand, the people that DO use IFC with Revit pressure Autodesk into allocating the vast majority of their resources into focussing on geometry exchange instead of fixing the information drain. 

At the same time, fixing the information loss can be overcome rather easily. It's a User Interface thing. You need to be able to map information from Revit to IFC and back. That's it. The second part, the geometry is way, way more complicated. Needs far more resources.

Now since Autodesk is a commercial company they, to some extend, listen to their customers and drive development on their needs. And budgets are defined based on the size of the (expected) user base and commercial gain. Nothing wrong with that, that's how commerce works. The guy paying the bills gets to make the decisions. 

What we have here is the situation where existing IFC users force Autodesk away from implementing relatively cheap solutions that would actually make IFC more attractive for non-users (eliminating loss of information), which generates a bigger user base, which in turn generates larger funds to tackle the more complicated geometry-related problems.

The Paradox of IFC Implementation is that that the people actually using it with Revit prevent Autodesk from taking big steps in the implementation of it's strongest feature (and major advantage towards geometry based formats such as 3D dwg): the ability to exchange object information.

How to solve this?

I don't know really. Somehow I don't think that after reading this everyone will say "he's right, what was I thinking? Silly me...".
But I am glad I focus on the "easy" part of IFC. That might just make it possible to have another go at the whole programming thing and start doing it  myself.
More on that later...

No comments:

Post a Comment