Should Engineers model accurately?

Here in Australia (I assume it is the similar elsewhere) engineers (M, E, P, and Structural) are responsible for specifying what things are and how many there are, but the architect is responsible for location. So traditionally engineers only need to show their objects schematically, whilst the architect has to show those same objects (or at least the ones that are visible) at their actual size and exact location so they can be dimensioned.
Examples include structural elements (like columns) on setout plans, A/C register and lights on ceiling plans, electric and data outlet on wall elevations, plumbing fixture locations on plans and elevations.

The theory behind BIM tells us that we shouldn't have duplicates of elements in the model. Besides waiting for the nirvana of Integrated Project Delivery (IPD) and the single BIM model how do we deal with this problem in a practical way?


METHOD 1:
Only objects authored by the architect are official BIM information.
Elements may be duplicated but only the architect's version is used for BIM. Effectively the architect becomes a drafting service for other consultants on the project.
Problems:

• architects would have to duplicate all engineer designed objects, even ones not relevant to design intent (like ducts above ceilings, plant, switchboards etc).
• all engineering data accompanying these elements would also have to be duplicated within those elements.
• the benefits of engineers using BIM software for analysis diminishes if there is no need for them to author BIM data.
• adds significant extra work and legal liability for architects, a liability possibly not covered by PI insurance. 

Summary:
Probably the reality on many current projects where engineers haven't taken up BIM, but not desirable into the future.
In theory possible on large IPD projects where BIM authoring for everyone is by a third party (the project BIM author?), but not a practical solution on the vast majority of building projects.


METHOD 2:
The architect can move objects authored by engineers.
The engineers accurately model their elements but the architect has the ability (and authority) to move those elements.
Problems:

• not possible with current BIM authoring softwares.
• moving engineer's elements may affect the engineering design. For example moving a column may increase a beam span.
• moving engineer's elements may break the setup for analysis. Moving a ceiling register may disconnect it from the thermal analysis system.

Summary:
Maybe technically possible if software written to do it. Solves the architects problem, but adds to the engineers problems. The architect being able to make changes to engineers elements would be a legal minefield.
A hybrid could be a workflow that allows the architect to move objects in a copy of the engineer's BIM, that is then used by the engineers, after their review, to move their actual objects.


METHOD 3:
Elements are duplicated by different authors,
each version can only be used for specific purposes.
Each engineer authored element is flagged as only usable for scheduling, and each architect authored element flagged as only usable for location.
Problems:

• still duplicate elements in a BIM model.
• not practical, or even possible, with all current BIM authoring and reviewing softwares.
• introduces risk of wrong information being used. An LOD table buried in a BIM Execution plan is not going to stop an electrical apprentice using the wrong information.

Summary:
The reality on projects where Revit is used with no BIM requirements.
Technically possible if BIM software written to accommodate it. May also be possible within some BIM federating software with an appropriate setup/workflow.


METHOD 4:
Engineers model accurately.
Engineer authored elements are both geometrically accurate and placed accurately.
Problems:

• Additional modelling (drafting) work beyond current practice for engineering offices.
• greater co-ordination required as architects still have to instruct engineers where to place these  elements.
• potential program delays if engineers don't keep up with architect's design (particularly for distributed BIM like Revit).
• allied workflow issues - do engineers also provide BIM for the architect's design options and design studies or do architects duplicate engineer's elements for these?

Summary:
If engineers say they are doing BIM, or are required to by their agreements, this is (according to BIM theory) what they should be doing.
On the face of it engineers would be required to provide more than they have in the past - accurate 3D model rather than 2D schematic. But is it actually that much more work with BIM software like Revit? And is this extra compensated by the benefits of having an accurate BIM for other engineering uses like scheduling, analysis and clash detection?
However more effort is definitely required in the workflow of ensuring engineer's elements are placed accurately where the architect wants them. Is this effect of what the BIM evangelists mean when they talk about "collaboration"?


METHOD 5:
Sub-contractors do accurate BIM.
Sub-contractors provide shop drawings as accurately located and sized BIM. Engineers just provide schematic information.
Problems:

• during design phase effectively the same as METHOD 1. The architect will have to model all engineer's elements to create a useful BIM.
• BIM can't be used for analysis or clash detection before construction starts.
• a lot of design work won't be able to be finalised until construction is under way, leading to possible delays in completion.
• in current practice not all engineer's elements are shop drawn. Power & data outlets, ceiling fixtures are some examples that are not.

Summary:
This method often occurs by default in Design & Construct projects. It is workable, but defeats the purpose of building a complete virtual building before construction, one of the benefits of BIM.
It is also a method advocated by IPD; engage contractor and sub-contractors during design phase so they can contribute. Besides the issues that early selection and engagement of this group entails, it relegates engineers to the role, at best, of advisor, at worst, no involvement at all.

CURRENT SITUATION

This problem can not be unknown to the software developers. Revit's solution is a 'copy/monitor' functionality. Objects between different discipline Revit models can be copied, and then monitored for changes. So this is a type of METHOD 3; Elements are duplicated by different authors but each version can only be used for specific purposes. But it is only a partial solution. There is no direct way to allocate duplicate objects to different purposes in an aggregated or federated BIM model. The other problem with it is the sheer amount of data. For example it is not practical in Revit to monitor thousands of electrical and data outlets or ceiling fixtures found in even moderately sized projects.
Another issue is that a lot of engineers don't see the benefit of BIM to them. I believe one of the reasons for this is due to the slow take up of BIM for analysis by engineers. One of the drivers behind developing BIM for Revit structure and MEP was to use the same model to communicate a design and to perform the various analyses required to inform that design. You would think the advantage of not having to create a new model for every different analysis would drive BIM within engineering offices. But the reverse seems to have happened. Most see Revit as another model they have to do in addition to their analysis models, and treat that model as a drafting tool for doing drawings - schematic drawings.  

I've been involved with projects where all methods but METHOD 2 have been used. It generally depends on how BIM proficient the engineers are and what their client service agreements stipulate. Because, at least in Australia, the client engages engineers rather than the architect, the architect has no power to dictate what they do so we just have to go with the flow.
So no method is ideal, but which one should we strive for?

WHICH METHOD IS LEAST WORST?

Firstly I'd like to emphasise that the list above is not exhaustive. I'm sure others have alternatives they have thought of and possibly used. The other point is that there may not be one method for all situations and projects. But it would be nice if there was.

A word about METHOD 5. It is the one advocated by BIM evangelists. But it fundamentally changes the type of service a consulting  engineering firm provides. Perhaps the work lost to sub-contractors could be picked up by working for those sub-contractors, but it still diminishes the role of engineering consultants. Read my view on the effects of IPD on architects in my post Integrated Project Delivery: Bad News for Architects?

As an architect I'm torn between METHODs 3 & 4. METHOD 3 allows me to hurtle forward doing my work without having rely on other offices, but METHOD 4 ensures BIM data is properly coordinated. METHOD 3 means I have to ensure my duplication work matches the engineers, METHOD 4 means I have to ensure the engineers work arrives in a timely manner.
On balance I think I prefer METHOD 4. It is the proper way to do BIM. Unless I can get that hybrid of METHOD 2 working.

What do you think?