Spaces are counted twice: those drawn and those that work

On the plan the numbers add up: 48 stalls drawn, standard met, application ready. Then the car park opens and the maths changes: four spaces reserved for management, two taken by charging stations, three unusable because the real manoeuvre is not the CAD one, one eaten by trolleys. The spaces drawn are not the spaces that work — and the planning standard, when the inspection or the change of use arrives, is measured on the second category. This calculator lines up the two ledgers: the spaces required by the local standard, the actually usable spaces net of the unavailable share, and their ratio — the provision index δ — which says whether the project passes, struggles, or locks up useless asphalt.

The formula: standard, plan and unavailable share

The model is a double ledger. On one side the rule's demand: R = A·r/100, the reference area times the standard in spaces per 100 m² — a number that changes with use and jurisdiction, and which here is an input, not an assumption. On the other the real supply: U = P·(1−q/100), the drawn spaces cleaned of the share q that does not work for general users. The index δ = U/R is dimensionless: it works with European standards per 100 m² as with North American ones per 1,000 square feet, because it compares two quantities of the same unit. Above 1 the provision covers the demand; below, the deficit S = U − R has a precise number.

The standard scale: the same car park, five verdicts

The standard r is the variable the designer does not control: the rule sets it for the use. The table shows how the same car park (48 spaces, 10% unavailable share, 1,200 m²) changes fate as the use changes:

As an office with standard 2 the car park is 80% oversized; as a supermarket with standard 6 it is twenty-nine spaces short. Same square metres, same asphalt: it is the use that decides the verdict — and the reason every change of use starts again from this count.

A complete numerical example

A 1,200 m² commercial building with a local standard of 4 spaces per 100 m²: the rule asks for R = 1,200 × 4 / 100 = 48 spaces. The plan draws 48 — a perfect tie, on paper. But the unavailable share is 10%: reserved stalls, a couple with chargers, a couple unusable. The spaces that work are U = 48 × 0.90 = 43.2, the index is 43.2 / 48 = 0.90 and 4.8 spaces are missing from the count. The project that "met the standard" is 10% short — not through a calculation error, but because the two ledgers had never been reconciled.

How to read the provision index

The four bands describe four situations. Below 0.8 parking commands: the deficit decides permits, turnover and liveability, and the options are structural. Between 0.8 and 1.0 you are below the standard in the grey zone: the project passes or stalls on the difference between drawn and usable. Between 1.0 and 1.3 the provision is compliant with a physiological margin. Beyond 1.3 the problem flips: asphalt that does not work has a cost — each stall with its aisles takes 25-30 m² that produce nothing, and the surplus is locked capital to put to work (charging, logistics, densification) before it becomes mere maintenance. No band is neutral: the index always says where to look.

Where the standard comes from: uses and jurisdictions

The value of r is not universal: it is the product of local rules and uses. Offices typically sit between 1 and 3 spaces per 100 m²; neighbourhood retail between 2 and 4; big-box retail between 4 and 8; housing is often counted per dwelling rather than per area. North America writes the same numbers as spaces per 1,000 square feet (4 per 1,000 sq ft ≈ 4.3 per 100 m²), and many European cities reason the other way round — maximums instead of minimums — in transit-served areas. The calculator treats r as a declared input precisely for this: the source of the standard belongs in the application, and scenario comparison is done by changing one number, not redoing the count.

The unavailable share: where the plan lies

The share q is the most underestimated line of the count. Into it go the reserved spaces (management, car sharing, contractors), the charging stalls when they serve only a fraction of users, the accessible spaces where the local rule does not count them in the general standard, and above all the stalls unusable in fact: the column that stops a door from opening, the manoeuvre that works in CAD with theoretical radii and fails with a real SUV, the dead corner used as storage. In mature car parks the real share is measured by walking: a one-hour survey often moves the index more than any argument about the standard.

International perspective: parking reform

Parking requirements are at the centre of a global reform. Cities like Minneapolis, Buffalo and San José have abolished parking minimums; London and Zurich have worked for years with maximums; many European capitals cut the standards in high-accessibility areas, and the research — from Donald Shoup onwards — has documented how much urban land the mandatory parking space has consumed. For designers, the practical consequence is that r has become unstable: it changes with plans, zones and years. A parametric calculator serves exactly this — redoing the verdict in ten seconds when the rule changes, and comparing the same provision under different standards.

Limits of the model and correct use

The model verifies provision against the standard: it does not estimate the real parking demand, which depends on schedules, accessibility and overlapping uses — a compliant car park can be empty or saturated, and peak-hour occupancy remains the figure to measure in the field. The area A must be taken in the exact definition of the rule (gross leasable, usable or built), because changing the base changes R in proportion. For this count's twin on the people side there is the maximum occupancy calculator — same logic, standard per area; for yards to build or extend, the excavation cost calculator completes the economic picture.

The unavailable-share scale: the car park that empties while standing still

The share q is the variable nobody votes on and everybody suffers: it grows one reserved space at a time. The table sweeps it across the base case (1,200 m², standard 4, 48 drawn spaces) — on the first row the tie is exact, then the car park empties without a single stall disappearing from the plan:

The first row deserves a note: exactly 1.00 is compliance — the border belongs to the compliant band, but without a single space of margin. From 25% upwards the model adds the warning: one space in four exists only on paper, and before designing new asphalt it pays to return spaces to general use — every 5 points of share recovered, the index climbs back without a metre of construction.

The area scale: same spaces, growing building

Extension is the quietest way to lose compliance: square metres are added, no spaces are removed. The table keeps the 48 spaces fixed (10% share) and grows the building:

At 400 m² the car park is nearly three times what is needed — asphalt that does not work; at 4,000 one hundred and seventeen spaces are missing. The usable-spaces column does not appear because it does not move: 43.2 from start to finish. It is the denominator that changes the fate, and that is why every extension application reopens the provision count.

Measuring real occupancy: the 85 percent rule

The index δ answers the rule; the car park answers the facts. The measurement that reconciles the two worlds is the peak-hour occupancy count: vehicles present are counted at three or four critical moments of the typical week, for at least two weeks, and compared with the usable total. The operating threshold used by planners is 85%: below it, there is real reserve even if the regulatory index is low; above it, the hunt for a space has already begun even if the paper says "compliant". The combination of the two numbers decides the strategy: low index + low occupancy = a purely administrative problem (payment in lieu, agreements); low index + high occupancy = a physical problem, and there you go back to the design.

The cost of a parking space, measured in square metres

Each surface stall takes 12-15 m² net, but with aisles, ramps and landscaping the real consumption climbs to 25-30 m² per space. The arithmetic is quick: a standard of 4 spaces per 100 m² means dedicating 100-120 m² to parking for every 100 m² of building — the car park can take more land than the activity it serves. That is why the surplus is never free: twenty spaces beyond the need are five to six hundred square metres of sealed soil, drainage, lighting and maintenance. And why underground parking, which costs a multiple of surface parking per space, becomes rational where land is valuable: the choice between the two forms is a problem of the denominator's value, not of preference.

Typical mistakes in verifying provision

The first mistake is verifying on drawn spaces: the unavailable share turns the tie into a deficit — and it grows over time, one "reserved" sign at a time. The second is using the wrong area: if the rule speaks of usable area and you compute on the gross one, the requirement comes out 15-25% inflated. The third is applying one use's standard to another: switching from office to medical or from shop to restaurant can double r on the same metres. The fourth is adding uses without looking at schedules: offices and homes peak in complementary ways, and adding their standards at 100% produces car parks empty for half the day. The fifth is forgetting that standards evolve: the application approved under yesterday's minimums can face tomorrow's maximums at renewal.

Shared parking: when one space works twice

Uses peak at different times: the office fills at 10 on Tuesday, the restaurant at 9 on Saturday night, housing overnight. Shared parking exploits this complementarity: the same stall serves the accountant by day and the resident by night, and the provision needed by the whole is lower than the sum of the single uses' provisions. Many rules recognise it with concurrency coefficients or with the option of contracting spaces between neighbouring properties; where the rule is silent, the hour-by-hour count per use is the technical argument to bring to the table. In the calculator, the shared scenario is explored by reducing the standard r to the combined peak value — and it is often the difference between a stalled project and one approved without an extra metre of asphalt.

Frequently asked questions

How do you calculate the required parking provision?

With R = A·r/100: the reference area times the local standard in spaces per 100 m². With 1,200 m² and a standard of 4, 48 spaces are required. The comparison must be made with the usable spaces — those drawn net of the unavailable share — not with the plan total.

What is the provision index?

It is the ratio δ = U/R between the actually usable spaces and those required by the standard. It is dimensionless: 0.90 means the spaces that work cover 90% of the regulatory demand. Below 0.8 the deficit is critical; above 1.3 the surplus locks up asphalt and capital.

Why are the drawn spaces not enough to verify the standard?

Because a part does not work for general users: reserved stalls, charging bays, accessible spaces where not countable, stalls unusable due to real geometry. With a 10% unavailable share, 48 drawn spaces become 43.2 usable — and the tie on paper becomes a real deficit.

How many parking spaces are needed per 100 m²?

It depends on use and jurisdiction: offices typically 1-3 spaces per 100 m², neighbourhood retail 2-4, big-box retail 4-8, housing often counted per dwelling. The exact value is set by the local rule — which is why the calculator treats it as a declared input, to be cited with its source.

How do you convert American standards per 1,000 square feet?

Multiply by roughly 1.076: 4 spaces per 1,000 sq ft equal about 4.3 spaces per 100 m². The index δ does not change with the convention, because it is a ratio between homogeneous quantities: just use area and standard in the same unit.

What counts in the unavailable share?

Reserved spaces (management, car sharing, contractors), charging stalls when they serve a fraction of users, accessible spaces if the rule does not count them in the general standard, and stalls unusable in fact: impossible manoeuvres, columns, corners used as storage. It is measured by walking the car park, not from CAD.

Does a parking deficit really block a project?

Often yes: the standard is a condition of the building permit and of changes of use. The typical ways out are cutting the unavailable share, recovering surface for parking, paying in lieu where the rules allow it, contracting spaces in nearby structures — or downsizing the use.

Is a very high index a good sign?

Not necessarily: above 1.3 the surplus is asphalt that does not work. Each stall with its aisles takes 25-30 m²: a surplus of twenty spaces is five hundred square metres locked up. The options are putting them to work — charging, urban logistics, densification — or not building them at all.

Why does the same car park go from compliant to insufficient when the use changes?

Because R depends on the use's standard: 48 usable spaces cover an office with r=2 (index 1.80) and do not cover a supermarket with r=6 (0.60). Every change of use starts again from the provision count — it is often the real constraint of the deal.

What is parking minimum reform?

A global movement: cities like Minneapolis, Buffalo and San José have abolished mandatory minimums, London and Zurich use maximums, many European cities cut standards near public transport. The practical consequence is that the standard r changes over time: the verdict must be redone when the rule changes.

Does the calculator also estimate real parking demand?

No: it verifies provision against the regulatory standard. Real demand depends on schedules, accessibility and sharing between uses — a compliant car park can be saturated or empty. The figure to measure in the field is peak-hour occupancy, before any decision on new asphalt.

Which floor area should be entered?

The one in the exact definition of the local rule: gross leasable (GLA), usable or built. Changing the base changes the required spaces in proportion: if the standard speaks of usable area and you enter the gross one, the requirement comes out inflated — the most frequent base error in verifications.

Updated July 2026