It is a model, not a simulation, and it is traditional to argue the flaws in the methodology or assumptions of the model if you seek to invalidate. What I heard you to say was: “what if you have to knock down N affordable units to build M market-rate units?”

And the model exactly models this, showing that as long is M > N, everyone will will get more housing. If you run the simulation you can see how it does this, market rate housing captures people who can afford it, they move out of mid-tier housing, which captures the folks who want mid-tier, which frees up affordable units that were being occupied by people who could move up.