Object

Local Plan 2040 Draft Plan - Strategy options and draft policies consultation

Representation ID: 6588

Received: 14/09/2021

Respondent: Mr Aiden Farmer

Representation Summary:

• The Transport Modelling undertaken by AECOM in support of the strategic options in the Draft Local Plan 2040 is fundamentally flawed for the following reasons.
• No validation or calibration of the traffic model has been undertaken which is not in keeping with Department for Transport WebTAG guidance.
• It is assumed that the Bedford Borough Transport Model was undertaken using Saturn, which was the model of choice for the Bedford Town Centre modelling project in 2015. In this case, Saturn cannot directly Model Public Transport therefore it is assumed that the engineers have made some significant assumptions with respect to public transport and trips being used, which have not been validated or calibrated.
• TAG unit M1.2 introduces the National Trip End Model (NTEM). It includes forecasts of population, households, workforce and jobs over 30 years which are used in a series of models that forecast population, employment, car ownership, trip ends and traffic growth by Middle Layer Super Output Area (MSOA). The NTEM data set can be viewed using the TEMPro (Trip End Model Presentation Program) software. TEMPro estimates of trip ends at any level below aggregate regions (e.g. MSOA, district, or county level) are subject to uncertainty and should not be used as constraints in matrix development process without verification and possible adjustments. No uncertainty log was prepared which is a recommendation of WebTAG modelling guidance.
• For direct use in matrix development, trip rate information estimated from household survey data should be considered instead to underpin trip end estimates at zone level. There is a risk that model may not be realistic or sensible due to the error around the model parameters used, or limitations in the extent to which the model can represent human behaviour. Therefore, before using any mathematical model, it is essential to check that it produces credible outputs consistent with observed behaviour. This is usually done by running the model for the base year (either the current year or a recent year), and:
• comparing its outputs with independent data (validation);
• checking that its response to changes in inputs is realistic, based on results from independent evidence (realism testing); and
• checking that the model responds appropriately to all its main inputs (sensitivity testing).
• Five types of data can be collected and used to inform most models:
• data on the transport network, including the physical layout, number of lanes, signal timings, public transport frequencies and capacities;
• counts of vehicles or persons on transport services, links or at junctions;
• journey times;
• queue lengths at busy junctions;
• interview surveys, in which transport users are asked to describe trips either through household travel diaries or intercept surveys (e.g. roadside interviews, public transport onboard interview surveys.
• These types of checks have not been undertaken to validate / calibrate the model.
• AECOM have derived trip ends using CTripEnds for a number of journey purposes. Expanding synthetic trip ends produced by CTripEnd to the local zoning system is considered to be subject to significant discrepancies from observed especially if validation and the calibration exercise has not been undertaken.
• It is also important to note that strategic models are not designed for use in a scheme specific assessment. For such an assessment it is recommended a revised forecast model would be produced from a recalibrated base year model using additional and more recent data and targeted to reflect a more specific geographical focus of resources and modelling effort.

1.24 100 word summary

The transport model used by AECOM is fundamentally flawed because the model has not been validated / calibrated (particularly with respect to public transport), the National Trip End Model (NTEM) should not be used below district / county level, no uncertainty log has been prepared, it doesn’t include household survey data, and the data has not been adequately tested. Strategic models are not designed for use in a scheme specific assessment. The latter requires a revised forecast model produced from a recalibrated base year model using additional and more recent data with a more specific geographical focus of resources and modelling.