How to design a city?

Unfortunately you have clearly been given modeling data in a incorrect light. The point of models frequently is not in general to get out a number.

Hm, I'm not quite sure what in my post gave you the impression that I think that, perhaps the 85th percentile remark. That only referred to a traffic engineering requirement to have traffic environments functioning at that sort of level of saturation. The model itself of course contains the actual parameters of the (proposed) scheme.

The criticism I meant to articulate is aimed largely at the requirement to have the model tested with at least existing motor traffic volumes, rather than assuming a reduction in motor traffic volume, which often vitiates any kind of innovative solution, and you get the usual nonsense that's designed to manage road danger rather than reduce it at source--guard railings, pig pen dog leg crossings, dual carriageways, central hatching, etc.

As far as traffic modeling goes you would have to have several models for different situations, much like every other physical situations.

Of course, for different scheme options, for instance.

I expect most government models are using very old methods, but employing some decent mathematicians / engineers to do the job is prob not high on the department for transports list. Many government departments just have some program that they bought and no qualified people to a actually model.

There are some very old programmes around, largely looked after and updated by people like TRL.

As far as modeling cyclists and pedestrians into a traffic flow I see little point how do you mean? If you mean their individual flow and how street planning allows the flow of ped / bikes, in which case that should be easy. Or do you mean adding ped / bike flow in with vehicular flow? If so there is little point as the same effect can be achieved in other ways which I ma willing ot discuss at a pub with pen and paper.

Walking and cycling are not mobility modes, which means that they are less modellable than motors. For instance, one cyclist may ride in the gutter in an intimidated and victimised way, whereas another may take the primary position and not be overtakeable. Pedestrians may step out into the street randomly and cause vehicular flow to stop. It's not programming this so that it occurs in a model that's the problem, but actually getting a handle on how to quantify it/what is likely to happen. It can seriously affect the accuracy of a model. It's one reason why cycling is often a mere afterthought in the design process and why we're generally not even getting cycle audits after schemes that didn't consider cycling had been designed.