Always a topic for animated debate, uncertainty was a main theme for this year’s Modelling World, held in Birmingham in early November.
Apart from some confusion around the exact location of the event - with one attendee taking a taxi from New Street Station to the old Edgbaston venue, rather than walking to the much more accessible new city centre location - discussions inside and outside of the technical sessions explored uncertainty and what direction to take in modelling and analytics.
These were some of the most prominent issues:
Increasingly a conference where both modellers and their end users meet, the vibe was more amicable and more collaborative, optimistic even, than you might have expected: last year’s ‘Computer Says Road’ report by Create Streets, giving the profession a bit of a kicking, was not mentioned once.
In recognition of the uncertain and unstable times in which we live, [TV1] The Department for Transport released its Uncertainty Toolkit in May 2021, now some two and a half years ago. At the event, Liz Jacobs, DfT’s Deputy Director Transport Appraisal and Strategic Modelling, explained progress in the application of Common Analytical Scenarios (CAS).
Not only has the narrative style of the scenarios supported communication about uncertainty, it has also led to a culture change in thinking about the issue , Liz explained, and the use of scenarios has increased the Department’s resilience to uncertainty.
Having acknowledged that, it remains hard to move the discussion away from the Dft’s core scenarios, whereas the need is great for effective visualisation and presentation of multiple scenarios. In this regard , Daniel Fisher, Appraisal and Analysis Manager at TfGM, usefully described how the authority uses ‘Policy On’ and Policy Off’ scenarios to test the need for, and value of interventions in futures in which their overall transport strategy is assumed to have succeeded or failed.
Explaining that this approach supports a more strategy- focused application of their analytical/appraisal toolkit as per the Treasury Green Book review- which by the way resonated with at least some decision-makers- Daniel voiced what others must surely also think: “is using the Common Analytical Scenarios true scenario planning, or really deploying a set of elaborately framed sensitivity tests?”.
Continuing the theme, Stephen Cragg, Head of Appraisal and Model Development at Transport Scotland, reflected on designing for a single optimised future, illustrated by the lovable panda, versus looking for a resilient future, illustrated by the trash panda- American slang for a raccoon.
Like Daniel Fisher, Stephen reflected that in practice , and up till now , scenario development testing has tended to focus on futures with or without strategy success, rather than deep, steep uncertainty. Stephen’s recommendation for this : ideally, design for a vision-led future, using the self-fulfilling prophecy to create a virtuous circle.
Transport for London has perhaps the longest pedigree in developing and using scenarios in practice. Stefan Trinder and Laura Putt, both at TfL , illustrated how their quite different scenarios from the DfT’s CAS are nowadays actively used in major project design and benefit estimates, but also in the development of TfL’s corporate strategy. Embracing the process of developing scenarios collaboratively as one of the benefits, their advice was to ensure that every scenario is a combination of positive and negative elements; and the TfL pair [TV2] identified their challenge as finding a balance between qualitative and quantitative insights in developing, presenting and applying the scenarios.
There certainly was no consensus at Modelling World on whether the future of transport modelling will continue to be delivered via commercial off-the-shelf packages such as VISUM, Aimsun, Cube, EMME and SATURN ( other software packages are also available), or whether the lure of Open Source, and confidence in its use, is now so great that it’s only a matter of time before the commercial model of modelling is changed completely.
Robin Lovelace, Associate Professor of Transport Data Science at Leeds Institute for Transport Studies has been a long-time proponent of open source software, and certainly of open access to model results. His 2020 paper “Open access transport models: A leverage point in sustainable transport planning”, co-authored with John Parkin and Tom Cohen, is definitely worth a read. At Modelling World Robin argued for many of the advantages of an open source approach, not least having access to the increasingly large ecosystem available for development, testing and use – with MATSIM a prime example in the transport modelling field.
These views were echoed by Crispin Cooper and Pedro Camargo who asked us : “can you afford not to invest in open source?”. Crispin, who is a Lecturer at the Department of Computer Science in Cardiff University, created a stir at last year’s Local Transport Summit in Cardiff with his open access TrafRed tool, to create and assess the effectiveness of interventions aimed at reducing car mileage. Try it out for yourself here: https://users.cs.cf.ac.uk/CooperCH/TrafRed/. Pedro, from Australian-based Outer Loop Consulting, is the main developer and promotor of the open source AequilibraE tool.
As the name suggests, AequilibraE is an equilibrium assignment model, written in open source Python, with a growing set of supplementary tools and interfacing with open source QGIS. Making use of widely available open data sources, its use by start-ups, research institutions and increasingly consultancies ( though currently non-UK) illustrates that it is possible to create that ecosystem and critical mass that Robin talked about.
And that’s exactly what Ben Taylor and Isaac Scott from Transport for the North told us they are doing: developing a Common Analytical Framework of models and tools to be shared and used by all subnational transport bodies and other open-source partners. What they saw as the benefits are it being small and modular, easy to access, optimised and properly tested.
Disappointingly for some, maybe, there was no cage fight between open source fans and commercial package promotors. The reality is that many, if not all, commercial packages already allow for and actively encourage open source add-ins – the most popular being QGIS and Python. Both develop, sometimes in parallel and independently, sometimes hand-in-hand.
But as so often, delegates were reminded of the practical issues that hinder the wider uptake of open source solutions. One public sector audience member remarked how difficult it is institutionally to get software from non-commercial sources installed, with TEMPro apparently having taken months to be approved and become usable.
A MW session on alternative model forms to inform decision-making, led by Helen Bowkett, Senior Technical Director at Arcadis, inevitably discussed activity based models - now sometimes known as AcBMs or activity- based demand models (ABDMs), to stop the confusion with agent-based models: (AgBMs or ABMs).
Laurence Chittock from PTV and Lou Mason-Walsh from Cambridgeshire County Council described the activity- based model in development for Cambridge, as an add-on or alternative, rather than replacing their four-stage model. Making best use of existing open source tools such as the PopulationSim population synthesizer, and free-to-access data including the National Travel Survey and the Census, the team has managed to build a model within 6 months.
Much is made in activity- based modelling of its much greater spatial detail and population segmentation – but Luis (Pilo) Willumsen of Nommom Solutions drew a round of applause from sceptics in the audience when suggesting that this can also be achieved within existing four-stage models – an activity-based modelling paradigm was not essential to that make a granularity difference.
He also pointed out the paucity in supporting (household) survey information necessary to estimate activity scheduling behaviour - so that that many practical AcBMs, despite their name, do not allow for actual rescheduling of activities.
Is all this creating an even blacker box? Tim Price from TfL took a pragmatic position, illustrating how Transport for London is building on its existing suite of strategic models in their development of activity-based capabilities. Being to some extent ahead of the game, Tim offered advice: Modularity of the demand model is useful, allowing the plugging in and out of model enhancements only where and when this delivers value.
TfL is of course fortunate, compared to other authorities, to have its London Travel Demand Survey (LTDS), interviewing around 8,000 households annually, as an excellent basis for its population synthesizer – an AcBM model building block that is more generally subscribed to, not just for modelling purposes, but also for post-analyses at a much more diverse population representation level.
These sentiments were echoed by Nick Bec and Theodore Chatziioannou from Arup’s City Modelling Lab: Viewing AcBMs as an ecosystem of models that add value in isolation and as the basis for further analysis, their advice was to “focus on things that only an activity- based model can do to make them complementary to traditional models”.
These practical experiences with activity based models start to permeate the profession, building on open source tools already available, reducing some of the hype and introducing good advice for those still in the earlier phases of that journey: focus on the problem you’re trying to solve, rather than starting from an AcBM as the solution, build modular, test rigorously, foster sharing and collaboration, and be open to scrutiny.
MW took a timely look at active travel modelling , something of particular interest to me as I completed my Master’s dissertation in 1985 on cyclist route choice modelling. I asked in my conference opening address the important question: has cycling modelling improved since then?
Pleasingly, I can report, it has! Until very recently the Cinderella of transport modelling, increasing local authority interest in and funding for active travel interventions has challenged the modelling of walking and cycling, both in terms of route choice and in the estimation of such demand. All proposed methods in the Modelling Active Travel session built extensively on open source data and tools – is it their ubiquitous availability that is at last making (certainly) cycling modelling mainstream?
Networks are easily built from Open Street Map and OS Highway Mastermaps, with demand data synthesized from NTS and Census data, supplemented by mobile phone data, STRAVA and cycle hire information. Terrain data on, for example, elevation levels, adds relevant route choice detail. Integration with an existing variable demand model seems to remain a challenge – instead of estimating future demand, SYSTRA’s Ian Burden and Neil Raha presented their Cycling Propensity Tool (with a similar objective to the existing Propensity to Cycle tool) investigating where is cycling currently common and where does cycling have the greatest potential to grow?
As so often, Professor James Woodcock who leads the Public Health Modelling group at the University of Cambridge, provided plenty of food for thought in this field, together with his PhD student Corin Staves, challenged by the title given to them - Bringing Walk & Cycle Modelling into the 21st Century.
For cycling, it’s all about the detail. To be able to assess health impacts of physical activity, traffic injuries and air pollution effects, their new cycling demand model, adapted from MITO, the Microscopic Transport Orchestrator developed at the Technical University of Munich, the open source demand model is based on the traditional four-stage approach, but agent-based, with very high-resolution in the zoning (50 x 50m), and modelling a full week of travel.
Further work that the Cambridge University researchers are looking into is modelling how the built environment influences active travel – representing not just time and distance,, but also physical characteristics such as infrastructure width, surface quality, separation from motorised traffic, and for junctions, signal type and crossing speed / volume.
It is in this space of active mode modelling that most progress seems to be being made, enabled by open source data and tools, but also by a willingness of developers to share their insights openly, be scrutinized and criticized, with a healthy collaboration between academics and practitioners. Given all these developments, and recognizing the intended role of active modes in decarbonizing the transport sector, I would personally like to see a TAG unit emerge on this.
The MW session that originated from an idea to discuss how well the modelling of Levelling Up was progressing became even more poignant after the Government’s announcement of the HS2 northern legs cancellation, just a month earlier. An intended discussion on modelling transformational schemes (and by the way, the recent DfT – commissioned report on transformative impacts of transport investment , prepared by CEPA( Cambridge Economic Policy Associates) is well worth a read), instead transformed into a discussion between the subnational transport bodies on their initiatives to level up modelling within their regions.
The shared set of tools, data, expertise and economies of scale now on offer for local authorities were described by Richard Bradley of Midlands Connect and Chris Storey of Transport for the North. Sarah Valentine, Head of Analysis and Appraisal at Transport for the South East, illustrated how an STB with fewer resources is starting to benefit from employing these.
It is a case of not just Levelling Up modelling within the STBs, but also between them. Tools are increasingly extended to provide insights into outcomes outside of just traditional transport effects, and also include levelling up the balance between forecasting, monitoring and evaluation. And as the tools and data are open source and software-agnostic, they are generally also available to consultants working on public sector projects too.
There’s never a dull moment in the modelling world, and at Modelling World, and ,as so often, I expect that delegates left with more questions buzzing in their heads than when they arrived. But I feel heartened by the progress that others seem to have made, and are willing to share . The celebration of successes had as a highlight John Bates being awarded the Lifetime Achievement Award. We’re already working on next year’s event, probably in June and again in Birmingham. I’ve no doubt at all that there will again be plenty to talk about!
Tom van Vuren is a Strategic Business Partner at Amey, Visiting Professor at the University of Leeds and Policy Director at the Transport Planning Society. He chairs the Modelling World organising team
TransportXtra is part of Landor LINKS
© 2024 TransportXtra | Landor LINKS Ltd | All Rights Reserved
Subscriptions, Magazines & Online Access Enquires
[Frequently Asked Questions]
Email: firstname.lastname@example.org | Tel: +44 (0) 20 7091 7959
Shop & Accounts Enquires
Email: email@example.com | Tel: +44 (0) 20 7091 7855
Advertising Sales & Recruitment Enquires
Email: firstname.lastname@example.org | Tel: +44 (0) 20 7091 7861
Events & Conference Enquires
Email: email@example.com | Tel: +44 (0) 20 7091 7865
Press Releases & Editorial Enquires
Email: firstname.lastname@example.org | Tel: +44 (0) 20 7091 7875
Web design london by Brainiac Media 2020