31 August 2012

ACCURACY vs PRECISION


Accuracy and precision are often used interchangeably, but they are not quite the same. A point relevant to documenting design intent in Revit (and other software).

Accuracy 

The root of accuracy is “to prepare with care”. Exactness is only one possible quality of being careful.  Appropriateness is another possible quality.
When designing and documenting buildings we need to be mindful they exist in the real world, are built by real people, using equipment that interacts with the physical world (as opposed to the virtual).
This starts with the surveyor. Although their CAD file might report to 16 decimal places they haven’t consciously measured to 16 decimal places. You might have noticed their heights and other figures are in metres to 2 decimal places. That is because their degree of accuracy is 10mm; they measure to the nearest 10mm.
On site contractors would like a tolerance of 25mm, but realistically work to the nearest 5 or 10mm. The Australian Building Commission’s Guide to Standards and Tolerances 2007 has a tolerance of 5mm for most things (up to 20mm for a concrete floor).
As an example walls are (usually) made up of a number of elements. Each element has a nominal thickness, and a manufacturing tolerance. In construction the supporting structure is rarely perfectly aligned, and where scrimmed joints occur there is always an extra 2-3mm of thickness. So in reality a wall is never its theoretical thickness (joiners always assume rooms will be narrower when measured corner to corner).
The traditional solution is to dimension ‘zones’ rather than dimensions based on theoretical sizes. A 90 stud wall with 2 layers 13mm plasterboard is theoretically 116mm wide, but it’s zone should be 120mm (note we are only talking about an extra 2mm per side). This ensures room dimensions are measurable by a human, a room with a theoretical width of 4004 becomes 4000 in your documents.
So to document a wall thickness as its theoretical thickness is not being accurate. It may be exact, but it is not “prepared with care”.

Precision 

Which brings me to Precision. Software doesn’t operate in the real world, it operates in the theoretical world of mathematics. It require exactness to calculate, and in Revit, to maintain relationships (constraints).
Whereas in the real world the exact thickness of a wall is unknowable to within 5mm, in software it has to be known to many decimal places.
If you don’t draw precisely in software it will break. We have already experienced Revit crashing because walls were created 0.00000012 degrees off grid lines. Recently I have seen multiple error messages saying joined elements and other relationships have been broken because a level was changed by a mere 0.125mm.

The root of precision is to “curtail” or “shorten”. To get software to work in a predictable and consistent manner avoid accidental dimensional relationships and very small differences; “curtail” trailing decimal places and unintentionality.

This means thinking about where your building sits on the site, how fits together, how different parts relate.
 It means not blindly tracing data from elsewhere, whether a survey plan, CAD file or rhino import.
A simple rule; leave complexity where it matters (which includes the design), simplify everything else.


So don’t get accuracy and precision mixed up.
Dimensioning walls to their theoretical thickness may be precise, but it is not accurate. You may consider setting your building out from an existing wall or boundary to be accurate, but if that wall or boundary is 0.3 degrees off perpendicular, it is not precise.
When setting out a Revit project, think both about accuracy; how useful the information being created is to others, and precision; how robust is my Revit model going to be.

24 August 2012

Revit: Why can’t I Dimension in a Section?


If any objects are not EXACTLY perpendicular to the view direction of a section Revit can’t place horizontal dimensions to those objects.
This includes grids. Which affects elevations as well. If a grid isn't perpendicular to an elevation you can't dimension to that grid.

All that actually makes perfect sense. If an object is not perpendicular to a view the distance measured will depend on where the section is cut. Move the section and the dimension would change.

I suspect this behaviour is not a bug but was designed in to Revit. It ensures consistency between what is modelled and what is documented.
But Revit, being software and not human, is very precise about this. ANY angle off perfect perpendicularity will cause this behaviour. To a computer 90.000002355205 degrees is not the same as 90.000000000000. 

Here is an example where different walls are not parallel:


Here is an example where walls are not perpendicular to grids:


Here is an example where the grids are not perpendicular to each other:

WHY IS THIS A PROBLEM?

You can't place dimensions in certain views. Period. And you can't fake dimensions in Revit.
Worse, Revit may crash when some-one tries to place a dimension in these views. I've seen it happen in Revit 2011, although admittedly not yet in Revit 2012 or 2013.

ARE THESE PROBLEMS ARE FIXABLE?

Yes and No. Yes they are, but is it practical - sometimes no.
It was not practical on the project these images came from because it was 3 weeks away from documentation being complete. Not only was there little time, there was a real danger of loosing a whole lot of work already done. For example deleting objects and redrawing them (like the grids) causes Revit to automatically delete all dimensions to those objects. Good-bye concrete set-out drawings!

If you don't fix the underlying problem, the work around is to draw Symbolic lines over the top of what you want to dimension in the affected views, and then dimension between them. Not ideal for accuracy, and a lot of work if there are a lot of dimensions.

So how do you stop this happening in the first place?

TIPS FOR DRAFTERS

DON’T trace CAD files: CAD lines are often at random angles, particularly if they are based on imports from Rhino.
DON’T use lines from Rhino: Rhino changes geometry when converting to meshes or lines for importing.
in fact DON’T trace ANYTHING: Sure its quicker – for you – for now. But some-one will pay later on. Set things out properly.
DON’T follow title boundaries: Title boundaries are sometimes slightly off being orthogonal. Fit your building inside the tile boundary, don’t follow it.
Set grids out rationally FROM THE START: Someone has to actually build what you are drawing. Think about how you will explain the setout to them.
Why do this?
DON’T treat Revit objects as if they are merely lines: Revit walls, floors etc. are 3D objects that appear in lots of views, not just the one you are working in. What you have just done to get your plan looking right may have completely stuffed a whole lot of sections and elevations.

TIPS FOR PROJECT LEADERS

DON'T ask for CAD files to be traced: Ask them to be used as a guide, be explicit they are NOT to be traced.
DON'T insist Title boundaries be exactly followed: Instead make the decision yourself how the building will fit within the boundary. Don't leave it up to your drafting staff.
Make sure Grids are RATIONAL: This can be a challenge when dealing with existing buildings, but remember grids are a documentation tool used to rationalise dimensions. Grids are there to locate things, they don't HAVE to pass through column centres.
Ask for PROOF: Ask you team to give you print-outs (or show you on the screen) with dimensions to 12 decimal places to PROVE things are modelled accurately. And do it early and progressively. The longer you leave it the more objects effected and the more work required to fix it.

17 August 2012

Integrated Project Delivery: Bad News for Architects?

This month the Australian Institute of Architects (RAIA) and Consult Australia are presenting their first suite of practice documents related to BIM in Australia. (more info)

Although there are a number of other Australian BIM documents around (NatSpec, CRC), these will be the first by groups with some skin in the market.

Undoubtedly these documents will be heavily influenced by (if not actually based on, like NatSpec) existing American documents, particularly the work done by the American Institute of Architects (AIA).

Fair enough, but I have an issue with the AIA approach to Integrated Project Delivery (IPD).

It is inherently anti-architect. It explicitly reduces the traditional influence of architects at early stages of a project, and therefore the main driver of design excellence.

Background – The AIA IPD Document

The AIA’s Integrated Project Delivery (IPD) is an excellent example of the direction Integrated Practice is heading. It makes sense; it ties together all the concepts that have been discussed over the last 10 years. But on reading it I see some implications for architects:
  • They see integrated practice as collaboration between owner, contractor and architect (in that order!).
  • Places the owner at the head of an integrated team, being responsible for managing the BIM model, as well as providing “final arbitration”.
  • Their whole scenario relies on early involvement of contractor and at least some trades / suppliers.
  • They clearly state that traditional “design-bid-built” type contracts are not capable of integrated practice, yet offer only one alternative model.
  • AIA is focusing on contractual arrangements, 3 out of the 4 IPD sub-groups focus on legal & legislative issues.
  • Suggest that members of an integrated team be “pre-qualified”. What does this mean and who is arbitrator of this pre-qualification?
  • Foresees BIM model ownership transferring to contractor at construction phase.
  • Assume architects will have less involvement during construction because “all issues are resolved during earlier stages”.
  • The AIA call it “Integrated Project Delivery”, which accurately describes their model. But is this model really necessary for “Integrated Practice”?

I'm at a loss as to why an architectural practice would agree to this arrangement, or why the AIA is pushing it.

Is it money?

IPD arrangements don’t suggest architects fees will benefit from IPD. One of the problems with IPD is that architects require higher fees at earlier stages to even provide BIM, let alone profit from it.

Is it influence?

Current IPD models don’t increase the influence of the architect. Collaborative arrangements are more likely to decrease the architect’s influence, particularly if the contractor and owner are involved from the start. At least under standard practice the architect has the ear of the owner until the building has been completely designed.

Is it control?

There is a perception that architects will be able to control a BIM model because they are the first to initiate it. Yet the AIA IPD arrangements place control of the model to a 3rd party specialist, engaged by the owner or contractor. Architects will not be able to argue they should maintain control of the model unless they also provide model management services, something that, as essentially an IT skill, is not a core skill of architectural practices. (remember what happened with Project Management)

Is it market share?

It is hard to see how IPD will increase the need for architects in the industry. With its focus on project delivery, architect specialities like design excellence and design co-ordination are mere bit players, unlike current practice where the beginning of a project is dominated by these factors.

Is it Quality?

In theory BIM and IPD will provide improved quality of outcomes. But that improvement doesn’t necessarily include better architectural outcomes. It does include reduced time, reduced co-ordination mistakes, the ability to model alternative scenarios. But those scenarios are not necessarily ones involving improving architectural design. As only one member of a collaborative team, it is unlikely the team will appreciate the advantage of letting the architects work through design alternatives. Contrast that with current practice where the architect spends most of the early stages of a project doing just that.

Is it industry leadership?

Current IPD models rely on the owner and contractor being active participants in IPD. Yet, at least in Australia, extremely few owners or builders are interested enough to actively engage in IPD. The truth is architects have little influence, and even less ability to encourage, organisations many times larger than any architectural practice, (some larger than all practices combined). I’m not sure why architects are the ones trying to convince them.

Is it benevolence?

Widespread IPD will create a more efficient building industry. It will cost less to build and operate the same amount of built space. And architects like being benevolent. But the reality is the ultimate monetary beneficiary will be the owner, as they will pay less construction costs. Some contractors may initially be able to increase profit, but only if they are competing against other contractors who don’t use IPD. I think the AIA needs to ask itself; are they doing this to solely benefit others?

Is it fear?

IPD will happen. The efficiencies for owners and contractors are too attractive. The reality is if architects aren’t prepared to involve themselves in IPD arrangements, or can’t at least provide a BIM service, these major players will go elsewhere. They will either employ their own architects, engage large drafting services, or engage large multi-discipline practices (invariably owned and controlled by engineers).

The AIA IPD approach is WRONG

I’ve come to the conclusion that architects as a profession have a lot to fear from IPD. Current proposed IPD models marginalise architects. The AIA IPD model is suited to large scale, large owner, large contractor type projects. They push the architect out of their role at the beginning of projects, when traditionally architects have had the most influence.

That is not to say IPD as such is bad and should be resisted. I believe architects should treat it as inevitable. But I do think architects, through their representative associations, should be developing IPD model(s) that protect the control and influence of architects. And I don’t believe an approach like the AIA IPD model does this.

Lets not repeat this mistake in Australia.

Introduction

I've created this blog to share my thoughts and experiences in making BIM practical.

There is a lot of theory flying around about BIM, which is great. But untested theory is no better than science fiction writing. And just as useless in the real world (would you try making a warp drive engine just because you read about it?).

I'm a Revit user so some posts will be about using Revit. I make no apology for this, Revit is the most common BIM software (for now). So practical REVIT is effectively the same as practical BIM.

For those who don't know, BIM is  Building  Information  Modelling, best described as the creation of a virtual computer model of a building before actual construction starts.