2. What am I covering?
• Internal processes for risk
and governance
• Case Study : west coast
intercity rail franchise
• Emerging UK
considerations : Aqua
(Quality analysis for
government)
6. Mott MacDonald Credentials
• We are at the forefront of developing guidance and
methods.
• We have developed industry standard software, for
example for demand modelling (DIADEM) and for business
case appraisal (TUBA, INCA, WITA and PAR).
• We contribute to industry guidance, for example we have
written the latest Highways England guidance on Appraisal
of Technology Schemes.
7. Mott MacDonald Credentials
• We have a wide range of application experience,
from large scale strategic and multi-modal
applications to detailed microsimulation models
of traffic management schemes or pedestrian
movements
• Highway modelling - Norwich Ring Road, UK
• PT modelling – Abu Dhabi Surface Transport masterplan, UAE
• Demand modelling
• Microsimulation – Calgary LRT, Adelaide
• Pedestrian modelling – North West Rail link, Melbourne Metro
• Junction modelling – Sydney LRT
• Economic Appraisal – Transparent Economic Assessment
• Freight modelling – Transnet, South Africa
• Land use modelling
• Aviation modelling – London airports
• Rail Modelling – High Speed 2
• Research
• Transport modelling for private financed projects
8. Wider
Thoughts
• It’s not about
software!
• Decision
making
without
traditional
transport
models –
other ways of
quantifying
policy or
project
impacts
9. Who is making the decisions?
Do they really know what they want?
When do they need an answer?
How robust does the answer need to be?
Do they understand the data available?
Have they any confidence in model
outputs?
Have they any technical knowledge?
Are they politically motivated?
Have they pre-conceived answers?
10. Application of Transport Modelling
• As a front-end service to help
determine the policy or
infrastructure solutions
• Provide an evidence base to
support the preparation of a
scheme business case
• Provide an evidence base for
the scheme environmental
appraisal
• To support transport
infrastructure management
business, for example for
Managing Agent Contractor
contracts
14. Best practice papers – Why do we need
them?
• Simple errors in models have stopped schemes
progressing
• Clients have lost confidence and the industry has
recognised change is needed
– HA best practice networks
– TRAMPNET best practice network set up by PTRC
• There are gaps in published guidance
• Modelling is a risky business!
15. Best Practice Papers
1. Peer assist and checking requirements
2. Addressing risks and liabilities in receiving and delivering transport models and data
3. Model design and data requirements
4. Record keeping and version control
5. Spreadsheet management
6. Network building
7. Checking transport models
8. Matrix building
9. Matrix estimation
10. Calibration and validation
11. Forecasting
12. Microsimulation
16. Best Practice – Key general messages
• We need to identify and mitigate risks - our processes are
designed to help us do this
• ALL transport modelling projects must be subject to peer
assist
• Apply standard QES procedures
• Models should be checked, and the checks recorded so
this can be demonstrated
• We must all use best practice procedures (or have a good
reason not to)
17. Best Practice – Peer Assist
• The rationale behind the application of Peer Assist in
transport modelling is to identify problems early, to have an
explicit approach to the identification and mitigation of risk,
and to make better use of the expertise available
• PA at scoping stage to identify and cost a checking plan
• Peer Reviews at key stages, at least at Tender and 1/3 and
2/3 through delivery
18. Best Practice – Spreadsheet management
• Name to describe calculation
• Keep old versions in superseded directory
• Use Spreadsheet with QA front sheet for all calculations
• Describe the calculations: the description should be of
sufficient detail such that any other technical colleague
can open up the spreadsheet and be able to
understand how the calculation works, and how they
might update it if any of the key inputs have changed.
19. Best Practice – Checking networks
• Input data checks
• • Source material
• • New coding against
scheme drawings or other
data
• Model run checks
• Version of software
• File names
• Assignment warning
messages
• Assignment convergence
• Demand model warning
messages
• Demand model convergence
20. Best Practice – Checking networks 2
• Output results checks
– Network statistics
– Changes in trip matrices
– Journey times
– Simulation junction performance
– Routings
– Assignments
– Fixed matrix
– TUBA
21. Best Practice – Matrix Building
• Lots of existing data sources
• Cleaning and verifying survey data
• Building station matrices
• Combining station matrices to get complete observed matrices
• Synthetic matrix methods and infilling
• Validation of matrices
• Sense checking
22. Best Practice - Calibration and Validation
• Check guidelines at beginning – including requirements for
reporting
• Lots of published guidance in DMRB and WebTAG and
from TfL
• Calibration – adjustments to model, but a large part is
checking
• Validation – comparison with independent data
23. Summary
• Transport modelling is a risky
business!
• Protection by using best practice
procedures
• Keep up to date
– Best practice papers
– Transport modelling community
• Plan checking as part of Peer Assist
and scoping
• Record checks
24. West Coast Intercity Rail Franchise
https://www.gov.uk/government/publication
s/report-of-the-laidlaw-inquiry
http://www.nao.org.uk/report/lessons-from-
cancelling-the-intercity-west-coast-
franchise-competition/
• A very public failure
• Technical errors with modelling
• Risk and governance issues
• Comprehensive documentation
of lessons learnt in public
domain
Annual revenue >$1.5bn pa (2009-10)
25. Background
• Nearly all UK intercity rail services are franchised
• Initially 25 Train Operating Companies (TOC’s)
• Net cost operation
• TOC separate from operating Group
• Initial franchises awarded solely on financial bids
• GFC (and overbidding), franchises failed
26. Background
• Macroeconomic risk?
• Cap and collar – surpluses capped, losses collared
• Subordinated Loan Facility (SLF) – Insurance
• GDP Mechanism
• The DfT Model developed for another purpose was
used to calculate the SLF
• Bids submitted in excel model form
27. Findings
• DfT model did not account for inflation (assumed it did)
• Technical confusion regarding elasticity factors to be used
• Bids not properly quantify risk and understated SLF
• DfT model not provided to bidders (lack of transparency)
• Adjustments made by Govt. tender assessment team. Inconsistent, not
clearly documented
28. In laymans terms
• There was a lack of transparency. The Department did not give bidders
enough information on which to base their bids.
• The Department did not follow its own published guidance.
• The amount of capital that the two final bidders were asked to put into their
bids was understated and inconsistently determined.
• The Department’s planning and preparation was inadequate.
• Roles and responsibilities for the project were unclear and resources were
stretched.
• The Department’s governance lacked efficacy.
• Quality assurance was inadequate – limited scope and late (audit
afterwards!)
29. Lessons
• Clarity of objectives helps decision makers to form appropriate judgements
by being a touchstone to refer back to throughout the decision-making
process.
• Strong project and programme management brings together and
coordinates the different streams of work, identifies interdependencies and
the sequence of events – the critical path – a programme needs to follow.
• Senior oversight acts as a sense check.
• Effective engagement with stakeholders, such as suppliers, helps by
contributing their knowledge, signalling problems and brings them into the
process.
• Internal and external assurance provides a sense check and can identify
any areas of concern to management
30. Recommendations
• Clarity of objectives
– Apply project and programme management disciplines to forming policy. Set timetables,
identify key tasks and their dependencies, identify a critical path for making policy changes
and allocate clear roles and responsibilities to deliver individual elements and the policy as a
whole.
– Identify the technical tools and models it requires to implement policy before delivery
commences. Develop, quality assure and test these processes before it moving on
– Provide training to staff on any new tools or policies. Before projects enter operational
stages, staff need training so that they understand objectives and how to apply processes
and tools.
• Project and programme management
– Timetables for major projects and programmes so they are realistic. It should consider the
‘usual’ timescales for typical projects and programmes, identify novel factors that might
impact on these and be cautious in shortening existing timetables.
– Key decision points build in sufficient time to properly consider decisions, include
contingency in case extra work is required, and consider other options if it cannot decide to
proceed.
31. AQUA
• The Aqua Book:
• Guidance on producing quality analysis for government
• https://www.gov.uk/government/publications/the-aqua-book-
guidance-on-producing-quality-analysis-for-government
32. AQUA
effective quality
assurance is achieved
by creating an
environment that is
conducive to quality
assurance and by
embedding
appropriate and
proportionate
processes
33. AQUA
• environment: creating the conditions in which quality
assurance processes can operate effectively, facilitated by
a culture that values quality assurance and welcomes
effective challenge, a well understood chain of
responsibility and sufficient time for quality assurance; and
• process: establishing a clear process for every stage of the
analytical life-cycle. This includes ensuring there is a
shared understanding about the purpose and any
limitations of the analysis
37. Summary : Analysis with RIGOUR
• Repeatable
• Independent
• Grounded in reality
• Objective
• Have understood and managed uncertainty
• Results should address the initial question robustly
• Establish how much we can rely upon the analysis for a
given problem.
Understanding the future demand for road travel is essential to help shape the policies we implement and the investments we make and to ensure that the outcomes for people's lives and livelihoods are fully understood. These are issues that have important outcomes for people's lives and livelihoods and involve billions of pounds of taxpayers' mone
Forecasts are not a target to be met nor do they define the level of road capacity required, but to develop the right strategy it is vital that we are able to understand how road traffic might change over time. This requires a robust forecasting approach that is based on the best available evidence of the underlying drivers of traffic demand, their relationship with changes in traffic and an approach that can model this appropriately.
Often in our analytical work, we expect that quality will be assured by suitable processes alone, more and more arduous processes,
an acknowledgement that an appropriate modelling/analytical environment, an ethos even, is also vital.
The environment involves a culture where leaders value and recognise good quality assurance but also recognise the need for adequate capacity, specialist skills and sufficient time; and finally establish a clear chain of responsibility plus a route for challenge.
The process side is more focused on those carrying out the analyses, and is about a systematic approach to make quality assurance logical, easy and comprehensive but also accessible, based on clear guidance and clear documentation.
Often in our analytical work, we expect that quality will be assured by suitable processes alone, more and more arduous processes,
an acknowledgement that an appropriate modelling/analytical environment, an ethos even, is also vital.
The environment involves a culture where leaders value and recognise good quality assurance but also recognise the need for adequate capacity, specialist skills and sufficient time; and finally establish a clear chain of responsibility plus a route for challenge.
The process side is more focused on those carrying out the analyses, and is about a systematic approach to make quality assurance logical, easy and comprehensive but also accessible, based on clear guidance and clear documentation.
Often in our analytical work, we expect that quality will be assured by suitable processes alone, more and more arduous processes,
an acknowledgement that an appropriate modelling/analytical environment, an ethos even, is also vital.
The environment involves a culture where leaders value and recognise good quality assurance but also recognise the need for adequate capacity, specialist skills and sufficient time; and finally establish a clear chain of responsibility plus a route for challenge.
The process side is more focused on those carrying out the analyses, and is about a systematic approach to make quality assurance logical, easy and comprehensive but also accessible, based on clear guidance and clear documentation.
In particular, it is important to accept that uncertainty is inherent within the inputs and outputs of any piece of analysis. It is important to establish how much we can rely upon the analysis for a given problem.