4. Modelling the impact of new technologies

New technologies inevitably generate impacts on the transport system and I will focus here on two of the most important ones: the role of remote access and the impact of new mobility technologies.

Teleworking

Remote access, in the form of teleworking and e-commerce is certainly changing trip patterns. Working from home or the nearest coworking space or coffeeshop is eliminating some obvious trips to the office. Hybrid arrangements seem to be the most widely accepted for of remote work as it combines the opportunities to interact with colleagues with the convenience and flexibility of teleworking.

However, there is some evidence that remote work while saving some trips generates some new ones. These new trips are often shorter and avoid peak periods and ideally should be undertaken by active modes; apparently this is not always the case.

For the transport modeller there are crude and better ways to handle this issue. Let’s assume that the classic Trip Generation model has been estimated on a population that do not telework (or telework very little); this needs to be updated post-pandemic. A crude way of handling teleworking is to assert that a certain proportion of trips generated in the peak will not materialise; this percentage can be estimated from the literature or small survey and will have to be calibrated in the full model. This is like the treatment of underreporting of trips but in reverse. The modeller then must speculate how this proportion of avoided trips will change in the future.

A less crude approach can be adopted when the Household Travel Survey is post-covid and asked about the extent of remote work during the survey week. This requires two enhancements to a classic model. First, to it is necessary to identify the type of employment that permits teleworking and the type of person that can take advantage of this possibility. This requires reviewing the segmentation of both travellers and job opportunities and assigning proportions to these segments that will actually adopt remote working. These segments will have their own Trip or Tour generation/attraction models. The modeller still will have to speculate how teleworking will evolve in the future as this is not settled as yet.

There is some anecdotal evidence that remote work is more prevalent on Mondays and Fridays and this may require a new decision: the modelled day. In the past assumptions were made that the model represented an “average normal day” without specifying very clearly what this was: an average Tuesday or Thursday? An average of all five working days? Why an average and not a particularly loaded day? Perhaps it is time to revise this assumption.

Jumping ahead, the use of a synthetic population facilitates this type of analysis as it creates a full population simulated from the 1% or 2% sample from the Household Travel Survey. It will still require care and intelligence in expanding this sample to cover the whole population and this task inevitably will introduce some errors.

e-commerce

Internet procurement has increased significantly since around 2010. IN the form of internet shopping it may eliminate some shopping and leisure trips but create some additional delivery trips previously performed by the shopper itself. An old Household Travel Survey would have ignored this but a more recent one would reflect this changing pattern on the personal movement side; a special type of data collection would be needed to follow the delivery vehicle movements (these are routinely ignored in most strategic models). From a pre-covid survey this requires assumptions with limited evidence in support.

It is also true that the movement of trucks is often treated in a very simplistic way on most strategic models and that of delivery vans even more so. Both are, in essence, Demand Responsive Transport of goods and packages and their simplified treatment follows the lines of other demand responsive trips: taxis, emergency and security vehicles, and alpha males cruising in their red convertibles; trip matrices are not the best descriptors of their movements. Treating them as “traffic noise” is certainly an approximation only.

New mobility technologies

The last ten years have seen an abundance of new mobility technologies; some have been tested and adopted, for example electric bikes and scooters whereas other are in the pipeline, to be commercially available in the “near” future: Connected and Automated Vehicles (CAVs), Hyperloop, flying taxis and road trains.

We need to explore and understand which of these new technologies and services are likely to be viable and, perhaps more importantly, likely to benefit society as a whole reducing emissions and delays and supporting an active and human-friendly urban realm.

There are two key distinctions to bear in mind when planning and modelling for new mobility technologies: Ownership and Demand Response.

Some of the new vehicles will be owned, others will be used as a service. Connected and Automated Vehicles (CAVs, level 4 and one day 5) are likely to become available one day and some will be owned as a better car whereas other will be used as Shared Mobility. Each of these modes of use must be treated differently in our models.

The second distinction is between scheduled and Demand Responsive Transport (DRT) services. There will be always scheduled services as today with fixed routes and frequencies, even if run with automated buses and trains (and less likely hyperloops). The user must adapt its travel plans to the timetable and modelling can follow traditional lines. In contrast, many other services will be demand responsive, and the supply of these service will be adapted, in real time, to individual requests from a specific location and a particular time. In fact, demand responsive systems have existed for some time in the guise of taxis, moto-taxis, rickshaws and more recently public bikes. DRT services pose the most significant modelling challenges as route and timetable are no longer fixed and there is a new relationship between demand and supply to model. How best to model them will be discussed in the next post.