Now where do I put it
The question of when to go into high level modelling software is a tricky one. But it needs to at least be asked and marginally answered simply to manage costs within your organisation.
I have spoken with MEP Engineers who have modeled their entire work before they have even been awarded the contract, and occassionally, as you might expect, they don't win the contract. That would upset me and it upsets them, but with the combination of human nature and the software to hand it is tempting to try and 'get it right' first time. Of course, when it comes time to make amendments, it can be more work to undo what has already been done to a higher level of detail than if it was done at a lower level of detail in the first place.
The problem has to be the detail. It's easy to imagine that a low level of detail is less work than a high level of detail. So for example, an illuminated bollard would be less work if it was just a blob called 'illuminated bollard' than if said bollard were finished to a high level of detail including fantastic geometry, textures, scheduling and manufacturer's information. However, that isn't even the problem because designers and engineers may have only one copy of a model for a given object, so they just sling 'it' in. The implication is that this is bad LoD practice, because you need to clearly communicate to what level of resolution a given design has been taken. That obviously means more work initially creating more models of the same thing at different levels of physical detail (never mind the information). However, what that would produce is a much clearer design that a project manager could examine and understand where more work needs to be done or perhaps conversely, less work needs to be done, because the project isn't at the degree of resolution that the designer has supposed.
Just to turn that on its head. When bidding for work visualisations need to look as good as possible. This invariably means in the minds of designers that the model should be in as high a level of detail as possible. So what do you get? You use the same models that you've previously prepared and drop them in, with all the high level of detail joys that they bring (or don't, depending on the model object). This is sowing a minefield for future design of that project. Leaving in legacy, highly detailed objects at conception phases could lead to all sorts of problems founded on the assumption that the model is already at a high level of completion, or perhaps the object is just 'lost' and inadvertantly gets priced. There really are so many things that can go wrong that I won't go into it in any great depth, mainly because we'll need to start handing round the brown paper bags to breathe into when we collectively hyperventilate. Needless to say, we've intentionally introduced a false assumption into the model and that can lead to problems.
