Parking Drives Sites

In the study below, 200 homes on a 200,000 SF site at 1.5 parking stalls a home put down almost twice as much parking as building: 99,000 SF of stalls and aisles beside a 50,000 SF housing footprint. That's why parking can organize a site plan. Stalls, aisles and access routes take substantial area, and where they go sets the building footprint, the pedestrian routes, loading and the open space left on the parcel. Nobody tours an apartment for the drive aisle, but the drive aisle may have decided how many apartments there are to tour.

Stacking the stalls changes the land they occupy

Illustrative 200,000 SF site. Stalls = homes × stalls per home, rounded up to whole stalls; the ratio is an editable assumption, not any jurisdiction’s requirement. Housing footprint = homes × 850 net SF / 85% efficiency / four housing floors. Parking is a separate structure, 330 gross SF/stall including circulation divided by selected parking levels. No podium overlap or underground zero-footprint assumption is used. Cost/stall is independently editable, not inferred from level count. A negative remainder means the two footprints exceed the site before setbacks or open-space rules.

Stalls are homes times a ratio

The arithmetic is short. Required stalls are homes × the parking ratio, rounded up to whole stalls. Each stall also needs its share of aisle and drive; the study folds that into one assumption, 330 gross SF per stall including circulation. Stack the stalls on more than one level and the footprint divides:

parking footprint SF = homes × stalls per home × 330 ÷ parking levels

The ratio is local. Your jurisdiction's requirement and how people are expected to reach the project set it, and I won't pretend there's one national number. The 1.5 here is the study's assumption, not a recommendation, and the 330 is an allowance, not a stall layout.

A worked example: one 200,000 SF site

These are the interactive's starting assumptions, and they're illustrative: a 200,000 SF site (about 4.6 acres), 200 homes at 1.5 stalls a home, and housing at 850 net SF a home, 85% efficient, on four floors.

• 200 × 1.5 = 300 stalls; 300 × 330 = 99,000 SF of parking on one level.
• 200 × 850 ÷ 0.85 ÷ 4 = 50,000 SF of housing footprint.
• 99,000 + 50,000 = 149,000 SF, which leaves 51,000 SF of the site.

Per home, that's 250 SF of building footprint and 495 SF of parking. Every home you add costs land twice, and the parking bill is the bigger one.

Now put the same 300 stalls on two levels. The parking footprint halves to 49,500 SF, the housing footprint doesn't move, and the land left nearly doubles, from 51,000 to 100,500 SF. Nothing about the housing changed. The parking decision handed back a quarter of the site.

It hands back homes, too. Push the count up until the remainder hits zero. With one level of parking, this site tops out between 260 and 270 homes: 260 leaves 6,300 SF and 270 runs 1,150 SF past the site line. With two levels, 400 homes fit with 1,000 SF to spare. Same site, same ratio, same buildings; the parking levels moved the ceiling from about 260 homes to 400. The ratio moves it as well. Stay on one level and drop to 1.0 stall a home and the ceiling rises to 340 homes; raise it to 2.0 and it falls to 210 (counting, as the interactive does, in steps of ten homes). That's the argument Garden Apartments makes about surface-parked plans, where every stall lands on grade: the parking field, not the buildings, usually decides how many units the site holds.

Those ceilings are not yields. A zero remainder leaves nothing for setbacks, the drive to the street, open space or a stormwater basin, so the count on a real site is lower. The point is which term sets it.

What stacking costs

Stacking trades land for structure. The study keeps cost per stall as its own assumption rather than inferring it from the level count, and its examples use $8k a stall on one level, $30k on two and $45k on four. Those are placeholders you can drag, not quotes. On them, the 300 stalls cost $2.4M on one level and $9.0M on two: $6.6M more to recover 49,500 SF, about $133 per square foot of site. Four levels at $45k cost $13.5M and recover only 24,750 SF more than two, about $182 a square foot.

The geometry explains the slide. The footprint is 99,000 ÷ levels, so the second level gives back 49,500 SF, the third 16,500 and the fourth 8,250. The first level you stack does most of the work; after that you're paying more for less ground.

Surface parking uses land at grade; a garage or podium shifts parking into a structure and changes building cost, circulation and massing. Neither approach can be judged from stall count alone. Whether $133 a square foot of site is a bargain depends on what the land costs and what you'd put on it, and that's a local answer.

What the interactive changes, and what it holds fixed

Four assumptions move. Drag the end of the cyan housing bar to change the number of homes, the end of the orange parking bar to change stalls per home, and the bottom of the parking stack to add or remove levels; slide the cost per stall below it. Held fixed: the 200,000 SF site (the dashed red line), the housing footprint formula and 330 SF per stall. Parking is drawn as its own structure beside the housing. The model doesn't put housing over a podium or bury parking with no footprint, and a negative remainder means the two footprints overrun the site before setbacks, open space or stormwater take anything.

Watch the orange bar against the red line. At 400 homes and 2.5 stalls a home on one level, parking alone is 330,000 SF and the site is 230,000 SF short. Four levels pull the parking back to 82,500 SF, with 17,500 SF to spare, at an assumed $45M of parking. The land problem became a money problem.

The judgment

A useful site study compares parking layouts alongside the building and its other site obligations. Required stalls, accessible routes, stormwater, setbacks and service access all compete for space. Solving parking first can obscure those relationships; bringing them into the same drawing makes the trade-offs visible.

So here's my judgment. Settle the parking strategy at concept, the ratio you can actually get and the number of levels you're willing to pay for, because it sets the unit count before any unit plan does. On this site the choice between one level and two is the choice between roughly 260 homes and 400, and no floor plan recovers that later. Then draw it with the building: a site plan makes the choices easier to see when parking is drawn with the building, not treated as a separate calculation. The line this page started as, “the site plan is a parking decision,” overstates it. A site plan is also a stormwater, access and code decision. But in this example, parking is the decision that eats the most ground.

It also shrinks the pond. #23, Site and Stormwater runs these two schemes as hardscape under its illustrative 3-inch, 90%, 3-foot assumptions: 99,000 SF of parking plus the 50,000 SF footprint needs about 11,175 SF of basin; 49,500 plus 50,000 needs about 7,463.

That is the end of the plans reading path. It started with a sheet that has to fit a building and ends with a site that has to fit the cars and the water before it fits the building, and at every stop the useful thing to store was a relationship, not a drawing.