Set parking requirements from data

One number for the whole city

Many German cities require developers to build parking when they build housing. The rule sits in a local parking bylaw, and its purpose is to keep residents from relying on public parking or parking on the street. For a long time, the requirement was often a single number for the whole city, such as one space per home. It did not matter whether the building stood next to a tram stop or far from any public transport.

A single number does not match real demand. Where a city asks for more parking than people need, construction costs go up, and land goes to garages instead of homes or green space. So many cities have made their rules more flexible. We compared the residential parking requirements of 12 German cities. The base value ranges from 0.5 spaces per home in Potsdam to 1 in Munich, Dresden or Stuttgart. Eleven of the 12 adjust it by location, often by dividing the city into zones, and eleven adjust it for extra mobility services.

What is usually missing is where the numbers come from. How a city sets its base value, and how much it changes from zone to zone, is rarely explained. It is often unclear whether the values come from experience, from a political choice or from an analytical method. That is a weak basis for a rule that developers and the city council have to accept.

Other countries already link parking to public transport. In Switzerland, municipalities use public transport quality classes to set how many parking spaces a building may have, and Austria links its parking rules to public transport access in a similar way.

How FLEKS works

FLEKS stands for Flexibilisierung Kommunaler Stellplatzsatzungen, making municipal parking bylaws more flexible. The method builds on an approach from TU Hamburg that was developed with data from Hamburg and Osnabrück. Together with stattbau münchen, we developed it further for the City of Munich, to set parking requirement zones for housing there.

It works in two steps. First, the model learns from today. The city is divided into small squares, for example the 100-metre census grid. For each square we know how many cars people own and what is around them: how good public transport is, how far the nearest shops are, how dense the area is, and some social characteristics. Comparing all the squares shows how much each of these factors pushes car ownership up or down.

Then FLEKS applies what it learned. For any place, it estimates how many parking spaces per home the people living there would need. This also works for places that do not exist yet, such as a new housing area. Areas with similar demand are grouped into zones, each with a minimum requirement the city can use in its parking bylaw.

Developers start from the minimum for their zone and adjust it for the mix of home sizes and types in their project. If the project offers mobility services such as car sharing, the requirement drops further. Our glossary explains what a mobility concept has to show for that.

FLEKS is a model of car ownership, not a finished bylaw. It gives the city a starting point it can explain and defend. The final values are still for the city council to decide.

Model parking demand in GOAT

In GOAT, FLEKS is built as a workflow. Each step of the calculation is a node on the workflow canvas, for example Join Features, which links two layers by distance, or Aggregate Points, which aggregates points onto areas. Values such as the search distance are set as Variables. The workflow is saved in the project, so you can run the calculation again with new data or for a different scenario. The GOAT documentation explains how workflows are built.

Public transport quality is one of the main inputs. FLEKS uses public transport quality classes, which GOAT calculates from timetable data with a tool of its own. When the service changes, for example with a new tram line, you put the new class into the model and see how parking demand changes there, before anyone decides.

The result is a layer on the map. For each zone it shows the parking demand, the range from minimum to maximum, and the number of residents. New transit lines and new shops change what people need, and the zones can then be recalculated for the new situation.

GOAT supports parking work in other projects too. The consultants at stattbau münchen have used GOAT since 2022 for accessibility analyses that support mobility concepts for housing and reduce parking demand. In Munich’s Harthof district, as part of the EU project ASCEND, we calculated high-resolution parking demand from accessibility, sociodemographics and land use.

You can find these projects directly below.