Justifying trip rates from a national survey database

A highway officer reviewing your trip rates is not checking the arithmetic. They are checking the site selection, because the selection decides the answer and everything after it is multiplication. Two consultants working from the same national database, on the same site, can produce peak hour rates that differ by a third, and neither has done anything dishonest. The difference lies in which surveyed sites they kept.

So the defensible approach is to make the selection visible and reasoned. State the filters, state why each one was applied, and hand over the output so somebody else can reproduce it.

Why selection decides the outcome

The trip rate databases used in the United Kingdom, principally the TRICS database, hold surveys of real developments with counts of vehicle and person movements. You choose a land use category, apply filters, and the system returns a set of surveyed sites and an average rate per unit or per 100 m².

Every filter changes the sample. Narrow the region and you may drop from 40 sites to six, with the average swinging accordingly. Include town centre sites in a sample meant to represent an edge of settlement scheme and the rate falls, because those sites have lower car use. Include car dominated sites in a sample for a highly accessible location and it rises.

None of this is a flaw in the database. It reflects the fact that trip generation depends on location and context as much as on land use, which is exactly why the selection has to be argued rather than assumed.

The filters that matter, in order

Work through them in a fixed sequence and record each decision.

1. Land use category and sub-category. Get this right first. Houses privately owned and flats privately owned behave differently, and a mixed scheme may need both applied to their respective unit counts.
2. Region and area type. Most practice excludes Greater London from samples for sites outside it, and treats Ireland separately. Include or exclude on a stated basis.
3. Location type and population within one mile. This is the filter that aligns the sample with the accessibility of your site. An edge of town site should be compared with edge of town surveys.
4. Development size range. Set a band around your scheme rather than accepting the full range, because very large and very small sites behave differently per unit.
5. Parking provision. Where the database records spaces, a sample of sites with generous parking will overstate rates for a scheme with restricted parking, and understate them the other way round.
6. Survey date range. Ten years is the usual window. Surveys carried out during the pandemic period are normally excluded, and any decision to include them should be agreed with the highway authority rather than taken quietly.
7. Days of week and survey period. Weekday surveys for a weekday assessment, with the peak periods matched to the network peaks rather than to the database defaults.

Record every filter in a table in the appendix, with the number of sites remaining after each step. That table answers most of the questions a consultee would otherwise ask in writing.

How many sites are enough

There is no statutory minimum and no published national figure. What officers look for is a sample large enough that one unusual site cannot dominate the average, and a spread of sites rather than several surveys of the same development.

Practical guidance that holds up in most reviews: if a filter takes the sample into low single figures, relax the least important filter rather than proceeding on two sites. Check the individual site rates for outliers and comment on them rather than silently deleting them. And if the sample is thin because the land use is unusual, say so and support the rate with a first principles estimate or a survey of a comparable operating site.

Deleting individual surveys without a stated reason is the single fastest way to lose credibility with a highway authority. If a site is excluded because it is a retirement development in a sample of general housing, write that down.

Multi-modal rates and mode share

Vehicle rates answer the junction capacity question. They do not answer the sustainable transport question, and increasingly authorities want both.

Multi-modal surveys record person trips by mode, which lets you present total person movements and the split between car, public transport, walking and cycling. For a scheme in an accessible location, that evidence supports a lower car mode share and a smaller highway impact than a vehicle rate alone would suggest.

Where multi-modal data for a comparable location is thin, the alternative is to apply a mode share derived from census travel to work data for the local area, adjusted for the characteristics of the scheme. Be careful with the adjustment. A claim that a development will achieve a car mode share well below the area average needs a mechanism behind it, usually parking restraint plus a funded Travel Plan with monitoring, not an assertion about the type of resident expected.

Netting off an existing use

Where a site has a lawful existing use, the assessment compares the proposal against that use rather than against nothing. Derive the existing trips the same way you derive the proposed trips, from the database or from a survey of the site while it is still operating, and present both.

Survey the existing use before it closes if you can. A count taken while the unit is trading is far stronger evidence than a rate derived from national averages, and it costs a fraction of the argument it prevents. Where the site is already vacant, the database route is the only one available, and the assessment should acknowledge that the fallback is theoretical.

Sensitivity testing and presentation

Run the assessment on the central rate, then test it. A sensitivity case using a higher rate, or the rate from a slightly different selection, shows that the conclusions are not balanced on a single choice. Where the higher case still produces acceptable junction performance, say so plainly, because that sentence removes the objection before it is made.

The presentation that survives review contains the selection criteria in full, the list of surveyed sites with their identifiers and dates, the resulting rates with the sample size, the calculation applying the rate to the development quantum, and the sensitivity case. Put the database output in the appendix unedited. A Transport Assessment that shows its working is read once, and a selective one is read three times by three different consultees.

Illustrative example: a 90 dwelling scheme on the edge of a market town. The initial selection returns 34 sites nationally, narrowed to 11 after excluding Greater London, restricting to edge of town locations and applying a size band of 50 to 200 dwellings. Two sites with rates well above the rest are retained and discussed rather than removed. The resulting evening peak rate is applied, a sensitivity case at the upper quartile is run, and the highway authority accepts the selection without comment. Illustrative only.

Where the scheme also needs parking evidence, keep the two exercises separate. Parking accumulation from a Parking Survey answers a different question from trip generation, and mixing them in one chapter invites confusion about what the figures represent. The threshold question of which transport document is needed at all is covered in transport assessment thresholds and trip generation.

This is general guidance rather than legal advice. Check the position with your local planning authority or a qualified professional before relying on it.

PlanScope provides AI-assisted draft reports for planning applications. Reports should be reviewed before submission. We do not guarantee planning approval. See PlanScope pricing if you want a structured first draft to build the transport evidence around.