Tendering for exhibits
Many didn't have their own workshops but instead bought in experiments in a similar way to how you would buy in building work with a specification and a tendering process.
The impression I have got, which would fit well with my predudices is that this is a very bad idea. Interactive experiments are not like a house, you are often building something which noone else has done before, and even if they have you are probably presenting it in a slightly different way. Very subtle things can entirely change how it works as an educational experience. Exactly how the intelectual feedback works, how hard a knob is to turn, setting up the experiment next to a wall so the fans don't work etc. etc. etc. There are so many unknowns I don't think anyone could get it right first time.
The solution, don't try to get it right first time, ask your suppliers to make you a prototype which they think is what you want, that is quick and dirty, but probably good enough to go on a floor (but will probably break in a couple of days). This will be a lot easier to build than something that is bomb proof. You put it on the floor and watch it, see if works, and whether the visitors use it how you expected, and decide whether that is a good thing or no. Then commission Mk1 which might work.
I am pretty sure the developers would rather work like this, so you will probably get the final version cheaper than you would have done otherwise, and it will work.
Plus of course don't get exhibit developers to do shop fitting modularise...
Also I don't see why you couldn't ask independent exhibit designers to come to you with good ideas, if they haven't got a track record, rent the exhibit off them for a bit and see if it would work. If you like it, buy it or get them to make a better one... this may involve some trust or a lot of contracts, but hey.