You are on page 1of 40

High Speed 2 Support

HS2 Model Framework Validation Report:


A Report for HS2

February 2010

Notice
This report was produced by Atkins limited for High Speed Two Limited for the specific purpose of High
Speed Two Modelling Framework Development.

This report may not be used by any person other than High Speed Two Limited without High Speed Two
Limited’s express permission. In any event, Atkins accepts no liability for any costs, liabilities or losses
arising as a result of the use of or reliance upon the contents of this report by any person other than High
Speed Two Limited.
Document History

JOB NUMBER: 5082342 DOCUMENT REF: 5082342 Model Framework


Validation Report (26-02-10).doc

3 Final Steve Jonathan Michael Michael 26/02/10


Miller Foster- Hayes Hayes
Clark

2 Draft 2 Alex Steve Miller Michael Michael 21/02/10


Fortune Hayes Hayes

1 Draft Matt Steve Miller Michael 01/02/10


Carlson Hayes

Revision Purpose Description Originated Checked Reviewed Authorised Date

5082342/5082342 Model Framework Validation Report (26-02-


10).doc
HS2 Model Framework Validation Report

Contents
Section Page
1. Introduction 4
1.1 Overview 4
1.2 Document Structure 4
2. The HS2 Model Framework 5
2.1 Background 5
2.2 The Validation Process 12
3. PLD Validation 15
3.1 Overview 15
3.2 Rail Demand Matrices - Validation Checks 15
3.3 Rail Network and Services 17
3.4 Rail Assignment 18
3.5 Air Model 28
3.6 Highway Model 30
4. PLANET South Model Validation 32
4.1 Overview 32
4.2 Flow Validation 32
5. PLANET Midlands Model Validation 35
5.1 Introduction 35
5.2 Flow Validation 35
6. Conclusions 37
6.1 PLD Model 37
6.2 PLANET South Model 37
6.3 PLANET Midlands Model 37
6.4 HS2 Model Framework 37

List of Tables
Table 2.1 - PLD Network Components 8
Table 2.2 - PLD Matrix Components 8
Table 2.3 - PLD Model Components 8
Table 3.1 - Demand Matrix Validation Checks 16
Table 3.2 - Data Sources for PLD Rail Validation 18
Table 3.3 - London Termini Screenline 21
Table 3.4 - South of Midlands Upper Screenline 22
Table 3.5 - South of Midlands Lower Screenline 23
Table 3.6 - North of Midlands Upper Screenline 24
Table 3.7 - North of Midlands Lower Screenline 24
Table 3.8 - Doncaster Upper Screenline 25
Table 3.9 - Doncaster Lower Screenline 25
Table 3.10 - Newcastle Upper Screenline 26
Table 3.11 - Newcastle Lower Screenline 26
Table 3.12 - Air Demand Check 28
5082342/5082342 Model Framework Validation Report (26-02- 2
10).doc
HS2 Model Framework Validation Report

Table 3.13 - Comparison of CAA Annual Passenger Data & Modelled Flows at Heathrow Airport 29
Table 3.14 - Highway Journey Time Validation (Minutes) 31
Table 4.1 - PLANET South Validation Flows (0700-1000 arrivals in Central London, 2007) 33
Table 5.1 - PM Validation Flows (To Birmingham New Street) 35

List of Figures
Figure 2.1 - PLANET Long Distance Model Framework 6
Figure 2.2 - HS2 Modelling Framework 6
Figure 2.3 - PLD Rail, Highway and Air Networks 7
Figure 2.4 - PLD Matrix Territories 9
Figure 2.5 - PS Model Network 10
Figure 2.6 - PM Model Network 11
Figure 3.1 - London Termini Screenline Location 20
Figure 3.2 - South of Midlands Screenlines 22
Figure 3.3 - North of Midlands Screenlines 23
Figure 3.4 - Doncaster 25
Figure 3.5 - Newcastle 26

Appendices
Appendix A – Rail Service Checks 38

5082342/5082342 Model Framework Validation Report (26-02- 3


10).doc
HS2 Model Framework Validation Report

1. Introduction
1.1 Overview
This report is concerned with the 2007/08 Base Model Validation of the HS2 Model Framework,
used by High Speed Two (HS2) to test options for High Speed Rail between London and
Birmingham.
To form this framework, a new long distance model was developed, using elements of the
previous PLANET Strategic Model (PSM) to form a new PLANET Long Distance (PLD) Model.
This model was then combined with PLANET South (PS) and PLANET Midlands (PM) and
integrated with a separate spreadsheet model for Heathrow passenger access (LHR).
This report describes the validation process for the modelling framework. The purpose of the
validation is to ensure that the base model adequately reproduces trip-making behaviour in terms
of both independently derived count data and journey times. Base year validation is vital to give
assurance that future year forecasts are derived from a representative base year, and is essential
before any future years can be constructed and schemes can be tested with confidence.
Validation is focused on the main corridors affected by high speed rail proposals and associated
released capacity on the existing rail network, principally the West Coast Main Line between
London Euston and Birmingham New Street stations.

1.2 Document Structure


The rest of this report is structured as follows:
 Section 2 describes briefly the Model Framework and describes the version of the model
under consideration, whilst section 2.2 discusses the validation process.
 Sections 3, particularly 3.3 and 3.4 consider PLD rail validation. This is in terms of matrices,
services and assignment validation respectively. Section 3.5 considers PLD air validation,
and section 3.6 considers PLD highway validation.
 Sections 4 and 5 describe the validation of the PS and PM models respectively. Finally,
section 6 summarises the conclusions.

5082342/5082342 Model Framework Validation Report (26-02- 4


10).doc
HS2 Model Framework Validation Report

2. The HS2 Model Framework


2.1 Background
In April 2009, Atkins was appointed by High Speed Two (HS2) to produce a forecasting framework
to support development, testing and appraisal of options for high speed rail between London and
Birmingham, as well as a wider high speed rail network across the UK. Atkins was supported by
Arup and Sinclair Knight Merz in development of the framework and its constituent components.
The forecasting framework was developed in agreement with HS2, the Department for Transport
(DfT) and an external challenge panel of demand forecasting experts over a period of three
months during Summer 2009, with further refinements during Autumn 2009.
Modelling high speed rail has several challenges:
 There is significant interaction between long-distance and short-distance journeys using long-
distance rail services, particularly on the Coventry to Birmingham New Street and Milton
Keynes to London Euston sections of the West Coast Main Line (WCML). This requires both
local detailed and strategic level modelling is required in the same areas;
 The same issues arise on the highway network, where the strategic highway network has
interactions between local trips using motorways for only short distances but affecting long-
distance trips also using the same roads;
 One of the major policy questions surrounds access to Heathrow Airport facilitated by high
speed rail. Passenger behaviour for these trips tends to differ significantly from normal
domestic, non-airport travel; and
 Long-distance trips do not typically match traditional patterns of “peak travel”. The length of
journeys means they tend to straddle both peak and off-peak periods. Time of travel is
heavily affected by demand management systems, using fare incentives to even out demand
between peak and off-peak periods.
For these reasons, a forecasting framework was developed which used linkages between a main
strategic, long-distance forecasting model and local rail models covering London & South East
and the West Midlands and a separate Heathrow Airport access model – the key areas of interest
for HS2 scheme development.

2.1.1 Framework Description


The forecasting framework consists of three PLANET models, adapted to work together, and a
Heathrow spreadsheet-based model combined with overall framework operation batch files which
transfer data between the models.
Details of the development of the framework are given in the separate framework development
report 1 .
The framework is illustrated in Figure 2.1 below.

1
“Model Development Report: A Report for HS2”, Atkins February 2009
5082342/5082342 Model Framework Validation Report (26-02- 5
10).doc
HS2 Model Framework Validation Report

Figure 2.1 - PLANET Long Distance Model Framework

Each of the three models represents separate markets:


The framework developed involved the framework passing pre-loads and matrices between PLD,
PS and PM, as shown in Figure 2.2 .

PLD
Surface  Long  Long 
access  distance  distance 
costs matrix pre‐loads

LHR PS PM

Surface  South  West 


access  East Pre‐ Midlands
trips loads pre‐loads

PLD
.
Figure 2.2 - HS2 Modelling Framework

2.1.2 PLANET Long Distance (PLD)


Developed from the existing PLANET Strategic Model, this model represents longer-distance
(over 50 miles) domestic travel by road, rail, air and high speed rail. For this framework, the model
excludes movements wholly within the West Midlands, wholly within London, South East, South
West and to/from Heathrow Airport. These excluded travel markets generally exhibit different
behaviours than most long-distance trips and are covered by the other three models.
However, as most services between London and the West Midlands also cater for the local West
Midlands and South-East to London commuting market, the effect of local passengers on long-
5082342/5082342 Model Framework Validation Report (26-02- 6
10).doc
HS2 Model Framework Validation Report

distance services are fed into the model from PS and PM by means of “pre-loaded” demand
volumes, i.e. assumed passengers on particular long-distance services which affect the crowding
experienced by long-distance passengers but are not subject to any re-routeing or demand
response effects themselves.
Similarly, demand associated with access to Heathrow Airport is also imported into PLD from a
separate Airport Access spreadsheet model, although this demand is input on a matrix origin-
destination basis and allowed to re-route to take advantage of quicker routes or reduced
crowding.
While the existing PSM model included a high level station choice model for London and
Birmingham (and three other cities), the station choice model did not work at a sufficient level of
detail to capture accessibility of different station sites within the Greater London and West
Midlands areas. As a result, the station choice procedures were updated to take into account
forecast station access times from RAILPLAN (for Greater London) and PRISM (for the
Birmingham urban area) models. For the purposes of this modelling work, the inputs on station
accessibility are static, i.e. they do not include any feedback of knock-on effects of increased
London Underground or highway congestion resulting from accessing alternative station locations.
PLD was created in 2009 by updating the 2001/02 PLANET Strategic Model (PSM) to 2007/8
matrices and networks, and combining it with PS, PM and the LHR model. The PLD Model is
discussed in detail in the PLD Model Development Report, but in basic terms is a multi-modal
national model of strategic movements in mainland Britain. The network coverage is as shown in
Figure 2.3.
Rail Highway Air

Figure 2.3 - PLD Rail, Highway and Air Networks

The version of PLD validated in this report is based on work undertaken between May and July
2009, but includes other amendments identified in the development of initial HS2 forecasts during
Autumn 2009. The source of each of the components of the model is clarified in this section in
Tables below.

5082342/5082342 Model Framework Validation Report (26-02- 7


10).doc
HS2 Model Framework Validation Report

Table 2.1 - PLD Network Components

Element Date Status


Highway Network May 2009 Includes the updates to bring the network up
to date for 2008, by adding the new highway
schemes between 2001 and 2008.
Highway Pre-loads May 2009 Includes regional estimates of traffic growth,
complemented by the TRADS (Highways
Agency count) data in the London to
Birmingham corridor. Pre-loads also allow
for intra SE and WM local traffic.
Rail Network May 2009 Includes the local modifications, such as
Stratford, Kettering, Allerton areas.
Rail Services May 2009 Taken from the 2007/8 CIF (timetable) data.
Air Network June 2009 Updated to 2007/8 services.

Table 2.2 - PLD Matrix Components

Element Date Status


Highway Demand May 2009 Updated to represent 2008 flows.
Matrix
Rail Demand Matrix June 2009 As extracted from LENNON (ticket sales
database) for 2007/8
Rail Fares June 2009 Matrix of fares extracted from EDGE (future
year rail demand forecasting tool) for 2007/8
Fares are removed from the network to
allow a more robust fare matrix to feed into
the mode-choice model.
Air Matrix June 2009 Extracted from CAA data for 2007/8.

Table 2.3 - PLD Model Components

Element Date Status


Assignment July 2009 MSA process enhanced,

Station Choice July 2009 Old 5 city station choice replaced by new
Greater London and West Midlands station
choice.

Mode Choice Not Implemented This is not applicable to a base validation


Rail Pre-loads July 2009 Implemented in PLD
Rail Matrix Cordons / July 2009 Implemented in PLD
Holes / Islands

5082342/5082342 Model Framework Validation Report (26-02- 8


10).doc
HS2 Model Framework Validation Report

As regards the rail passenger demand in particular, but also to some extent the associated
highway (car) demand, the three models differentiate between demand territories. Trips wholly
within the PM area, or wholly within the PS area, are not represented in PLD for either rail or
highway demand. The areas concerned are outlined in Figure 2.4

PLD

PM

PS

Figure 2.4 - PLD Matrix Territories

2.1.3 PLANET South (PS)


The existing PS model has been in use for several years for modelling forecast crowding on the
London & South East rail network and associated impacts on London Underground lines. The
model represents morning peak period (0700-1000) rail movements within an area which covers
the former Network South East territory, with less detailed representation beyond those areas.
PS was adapted for this framework to include only trips within London, South East and South
West areas, to eliminate any overlap with PSM. The model includes both local and long distance
services into London. However, long-distance demand – i.e. rail demand between inside and
outside London, South East and South West areas – is included in the model by means of
“wormholes”.
These “wormholes” are effectively dummy zones that feed origin-destination demand information
from PLD into PS into nodes at the edge of the PS network area, having been scaled from all-day

5082342/5082342 Model Framework Validation Report (26-02- 9


10).doc
HS2 Model Framework Validation Report

to peak-only levels. This demand is then assigned onto long-distance services, with the PS
assignment routines allowing the demand to reach their final destinations within the PS area using
any other rail or underground services. The model works in a similar way for trips from the London
& South East area to the rest of the country.
It is important to note that while long-distance movements are included in PS matrices and are
assigned in a way that allows them to find quickest and least crowded routes to their final
destination, they are not subject to any demand response. By contrast, local trips within the
London & South East area are subject to any changes in crowding as a result of changes to long-
distance services and demand – expected to be a reasonably significant effect in future years.
The matrices in PS are cut down to those origin-destination pairs wholly within the South East, as
shown by the thicker blue line, running east-west from the Wash to South Wales, in Figure 2.5.

PLD

PS

Figure 2.5 - PS Model Network

2.1.4 PLANET Midlands (PM)


The PM model was developed during early 2009 for DfT. The model is generally similar to PS but
covers the West Midlands and East Midlands rail networks and demand. It also has slightly
different functionality from PS in the way that it deals with station choice: while PS uses default
EMME/2 assignment processes to assign demand from zones to stations, PM uses detailed
accessibility information for each station from each zone.
PM was adapted for the framework by cutting back its area of scope to the immediate Greater
Birmingham area. Both local and long-distance services are included in the model. However, in
contrast to PS, long-distance demand is fed from PLD by a pre-load approach, identifying the level
of long-distance passengers on long-distance service groups, and adjusted to reflect the morning
peak demand levels.
This altered approach reflects the much higher levels of car access for longer-distance trips
to/from Birmingham, and the relatively lower importance of knock-on dispersal effects of crowding
5082342/5082342 Model Framework Validation Report (26-02- 10
10).doc
HS2 Model Framework Validation Report

on local rail services feeding long-distance rail passengers from the outskirts of Birmingham into
central Birmingham to catch long-distance services.
PM is an AM Peak rail model for the Midlands. It was constructed by Atkins and Arup, and
validated by Jacobs Consultancy in 2009. It was validated against counts in the following
locations:
 Birmingham;
 Nottingham
 Leicester;
 Lincoln
It was considered fit for use to be integrated with PLD in the Model Framework, in order to
improve the representation of local rail passenger demand in the West Midlands, as far as service
crowding is concerned.
The matrices in PM are cut down to origin-destination pairs wholly within the West Midlands, as
shown by the red line making a cordon in Figure 2.6.

PLD

PM

Figure 2.6 - PM Model Network

2.1.5 Airport Access Model (LHR)


As explained in the previous section, airport access travellers have different demand response
characteristics to most passengers, with relatively little generation, higher values of time and often
with access to relatively high quality coach services. Much research has been undertaken by BAA,
DfT and others on airport access travel models.
The distinct level of detail required for airport trips was met by using a spreadsheet approach
developed using experienced gained from the London Airport Surface Access Model (LASAM) by
sub-consultants SKM. Demand for trips to and from Heathrow Airport, divided by business and
leisure journey purposes is calculated within the spreadsheet using journey time skim information
from the PLD model. As described above, resulting demand to Heathrow Airport by mode was
then fed back into the PLD model to represent the impact of Heathrow Airport passengers on
crowding on long-distance services for long-distance passengers.

5082342/5082342 Model Framework Validation Report (26-02- 11


10).doc
HS2 Model Framework Validation Report

It should be emphasised that airport demand is included in PLD if it is domestic-to-domestic, and


in the Airport Access Model if it is interlining / international. For example, someone flying from
Glasgow to Heathrow to access central London is included in PLD. However, someone on the
same flight to connect at Heathrow for a flight to Sydney is included in the Airport Access Model.
The LHR Model is a spreadsheet-based simplification of the well-established London Airports
Surface Access Model (LASAM). It imports many parameters from LASAM, and takes generalised
costs from PLD. The primary function is to produce Heathrow access demand shares based on
generalised costs derived from PLD.
It was validated in 2009 by SKM, against output elasticities of rail demand to rail fare and rail in-
vehicle time; and car demand to car time.

2.1.6 Use of Framework


Each of the constituent models within the framework performs a different role in developing
forecasts and contribution to the journey time benefits that are the basis for the projected HS2
business case and scheme development:
 PLD is used to provide the main impacts, in terms of passenger loads, abstraction, modal
shift and generation for long-distance passengers on long-distance and high speed rail
services. These changes give rise to over 90% of the scheme benefits in economic terms;
 PS is used to identify the impacts on local commuting passengers in South East England
including those using long-distance services during peak periods. It also provides an outline
view of the issues associated with passenger dispersal in a crowded transport environment
on LUL services. These changes give rise to typically 5% of the scheme benefits in economic
terms.
 PM is used to reflect the impacts on local commuting passengers in the West Midlands,
including on long-distance services. These changes, in a much smaller area of congestion
than in South East England, give rise to typically around 3% of the scheme benefits in
economic terms.
 LHR is used solely to handle passenger access to Heathrow Airport (other than for greater
London) and has a relatively minor role in the overall forecasts.
As such, the validation of the models needs to focus on the core roles of each of the models within
the overall forecasting framework.

2.2 The Validation Process


2.2.1 Overview
Transport models are a simplification of individual behaviour, but remain very complex
combinations of data, assumptions and the application of computer software. The availability of
resources (time, data and budget) can impact on the sophistication and reliability of any model. In
addition, a model being developed at a preliminary stage of a proposed scheme’s development is
likely to be less detailed, and hence less likely to be validated at a detailed level, than one being
used to support the promotion of a detailed scheme design through public scrutiny.
The intention of model validation is to demonstrate how well the model represents the observed
behaviour for the “base year” (usually the most recent year for which consistent data can be
assembled).
The hypothesis is that, if it can be demonstrated that the model faithfully represents the existing
situation, then it can be expected that it will reliably represent some forecast situation. Conversely,

5082342/5082342 Model Framework Validation Report (26-02- 12


10).doc
HS2 Model Framework Validation Report

if the model does not reproduce a particular element of travel behaviour at the validation stage, it
is reasonable to ascribe less confidence in the model’s forecasts regarding that element.
It must be stressed that a good model validation does not, of itself, guarantee a good model
forecast, or vice versa.
It follows that the process of model development and validation requires skilled interpretation
rather than application of standard criteria and procedures. However, conformance with published
guidance on model validation is highly desirable, particularly for a scheme which is likely to be
seeking support from Government sources, either political or financial.

2.2.2 Published Guidance on Model Validation


The Department for Transport has developed a website for guidance on the conduct of transport
studies, or Transport Analysis Guidance, hence known as WebTAG 2 . This is complemented by
advice, including guidance on model validation, related to major road schemes and highway
models, published by the Highways Agency as part of their Design Manual for Roads and Bridges
(DMRB) 3 . There is no equivalent detailed guidance relating to model validation for Public
Transport Models.

2.2.2.1 Public Transport Model Validation


Advice on validation for Public Transport Models takes the form of the three areas to check, as
noted in WebTAG Unit 3.11.2, reproduced in the following box.

10.1.1 The validation of a public transport passenger assignment model


should involve three kinds of check:

 validation of the trip matrix;


 network and service validation; and
 assignment validation.

All of these were undertaken for PLD, and are reported in sections 3.2, 3.3 and 3.4/3.5
respectively. This approach was also adopted in the validation of PS and PM, discussed in
sections 4 and 5 respectively.
Guidance on acceptable validation criteria for public transport models is defined in WebTAG Unit
3.11.2, and is shown below:

10.1.6 Across modelled screenlines, modelled flows should, in total, be


within 15% of the observed values. On individual links in the network,
modelled flows should be within 25% of the counts, except where observed
flows are particularly low (less than 150).

The rail and air flows in the models are compared on this basis in sections 3.4 and 3.5
respectively.

2.2.2.2 Highway Model Validation


The available guidance is skewed heavily towards detailed local traffic models with one-hour
modelled flow comparisons, where a high degree of validation effort is expected.

2
WebTAG: See http://www.dft.gov.uk/webtag/
3
DMRB: See http://www.standardsforhighways.co.uk/dmrb/
5082342/5082342 Model Framework Validation Report (26-02- 13
10).doc
HS2 Model Framework Validation Report

Guidance on acceptable validation criteria and measures for highway models are provided in
DMRB Volume 12. This is shown in the box below:

Criteria and Measures


Assigned Hourly flows * compared with observed flows
Acceptability Guideline
1. Individual flows within 15% for flows 700 - 2,700 vph ) (>85% of cases)
2. Individual flows within 100 vph for flows c 700 vph (>85% of cases)
3. Individual flows within 400 vph for flows > 2,700 vph (>85% of cases)
4. Total screenline flows (normally > 5 links) to be within 5% All (or nearly
all) screenlines
5. GEH statistic:
i) individual flows : GEH < 5 (>85% of cases)
ii) screenline (+) totals: GEH < 4 All (or nearly all) screenlines

Modelled Journey Times compared with Observed Times


6. Within ±15% (>85% of cases)

The highway model in PLD is compared only on the journey time basis in section 3.6, as the main
focus of the highway validation is to ensure realistic highway times are produced to feed back into
the mode-choice model. Importantly, no consistent independent count data exists for strategic
traffic flows. PLD (and PSM) are only concerned with strategic highway flows by car.

2.2.3 Focus of the Validation


The focus of the validation exercise is PLD in its framework form, linked up to PS and PM. Within
PLD, the emphasis of the validation is upon the component parts and assignment of the rail
model, since this is where the tests will be run. The highway and air models are intended as both
comparators for the rail model (in terms of changes in relative journey times) and a source of
competing trips (which are able to switch to - or from - the rail mode.
A secondary focus of the validation is to confirm that the changes introduced in PS and PM, in
integrating them with PLD, have not disrupted unreasonably the recent validation exercises
undertaken on those models.
As discussed above, PLD has the lead role in providing passenger forecasts and economic
benefits for the HS2 scheme, with PS playing a lesser role, and PM having a quite nominal role. It
is therefore all the more important to focus the validation process on the quality of the PLD
component in the Model Framework.
LHR has not been subjected to revalidation, as the only change was to integrate the input and
output mechanisms to the PLD model. As such, no change was expected in the responses that
LHR gave. This was checked during the development stage, by introducing dummy data and
checking the results obtained; these checks confirmed the model responses were unchanged.

5082342/5082342 Model Framework Validation Report (26-02- 14


10).doc
HS2 Model Framework Validation Report

3. PLD Validation
3.1 Overview
This section covers the validation and checking of the core elements comprising the PLD model,
which is the principal forecasting tool in the Model Framework as regards the HS2 forecasts and
Business Case, as discussed above. Accordingly, this section is the major part of the report.
Again, for a major rail based scheme, the rail mode is the most important element of PLD, and this
is covered in sections 3.2 to 3.4. The validation of the more limited air and highway models is
discussed in the remaining sections.

3.2 Rail Demand Matrices - Validation Checks


The most appropriate comparison for PLD is against National Rail Trends (NRT) data on region-
to-region rail movements, published by the Office of Rail Regulation (ORR). This entailed
aggregating demand data for PLD zones (mainly districts) into the parent regions and reversing
the de-annualisation process which was used to produce the PLD data from LENNON ticket sales
data.
Table 3.1 below shows findings for key HS2 flows. Note that the data are not separated by trip
production/attraction, or by direction, and are repeated. For example, 1.3m journeys are recorded
between Scotland and North-west England, and between the North West and Scotland; this
represents a daily flow of around 650,000 passengers in each direction of travel.
Small differences between the two data sources are likely to arise from alternative assumptions
applied to (for example) staff travel, and refunds, as well as how station-to-station ticket data is
allocated to geographic zone areas.
It can be seen that all the region-to-region flows shown in Table 3.1 have PLD versus NRT
journey volumes within 4% of each other. The maximum discrepancy in absolute terms is West
Midlands to London with 239k per annum (or 3.1%).
It can be concluded that, for interregional flows key to HS2, the PLD demand data for 2007/8
closely matches corresponding data from NRT. This confirms that the processing undertaken on
the data at this stage is consistent and gives confidence in the PLD demand matrix data, ahead of
de-annualisation and network assignment checks.

5082342/5082342 Model Framework Validation Report (26-02- 15


10).doc
HS2 Model Framework Validation Report

Table 3.1 - Demand Matrix Validation Checks

Regional Flow NRT Journeys HS2/PLD Difference Difference


journeys % data 0708 (k) %
07/08 (k) (k)
To/from/within Scotland 75,636 100.0
To/from North west 1,313 1.7 1,318 5 0.35%
To/from West Midlands 196 0.3 199 3 1.68%

To/from London 1,362 1.8 1,385 23 1.69%

To/from/within North 71,663 100.0


West
To/from Scotland 1,313 1.8 1,318 5 0.35%
To/from West Midlands 2,,921 4.1 2,952 31 1.07%

To/from London 6,452 9.0 6,267 -185 -2.87%

To/from/within West 42,290 100.0


Midlands
To/from Scotland 196 0.5 199 3 1.68%

To/from North west 2,921 6.9 2,952 31 1.07%

To/from London 7,727 18.3 7,488 -239 -3.09%

To/from/within London 698,782 100.0

To/from Scotland 1362 0.2 1,385 23 1.69%


To/from North west 6,452 0.9 6,267 -185 -2.87%
To/from West Midlands 7,727 1.1 7,488 -239 -3.09%

To/from/within South 254,274 100.0


East
To/from Scotland 189 0.1 193 4 1.90%
To/from North west 1,111 0.4 1,076 -35 -3.17%
To/from West Midlands 2,109 0.8 2,072 -37 -1.78%

5082342/5082342 Model Framework Validation Report (26-02- 16


10).doc
HS2 Model Framework Validation Report

3.3 Rail Network and Services


3.3.1 Overview
The rail network was checked for the accuracy and appropriateness of its coverage. It is important
to note that the PLD rail model is not intended to cover every route and station in the model, but to
cover strategic rail corridors in mainland Britain, for trips of greater than 100km.
The network (nodes and links) was predominately transferred from the previous PSM and thus did
not require an in-depth validation. The new pattern of services, derived from timetable data,
required checking as outlined below and in Appendix A.
Once the network was checked, the transit lines were constructed, as detailed in the Model
Development Report.

3.3.2 Service Validation


The validation of the services is summarised in Appendix A, which represents “best practice” in
developing service coding from the available electronic timetable “CIF” data. The data processing
was undertaken until all the errors identified by these checks had been resolved.
Further checks were then undertaken to ensure that any mismatches between zone and network
definitions did not compromise validation – this can occur where significant amounts of local
demand is loaded onto a network which principally compromises of strategic stations and
services: “network bottlenecks”.
In general terms, this is not a problem for assignment in the base year, but impacts upon
generalised costs when running base and test options in the future years. It has therefore been
discussed in the Model Development Report, where various changes to zone connectors were
made to ensure that local trips were not inappropriately using strategic services.
This process was undertaken on WCML and Trans-Pennine routes on the basis of crowding levels
in future years, focussed on areas of known large local flows which are unlikely to use strategic
services. It should be emphasised that the design of the HS2 framework using specific local
models in the London & South East and West Midlands areas meant that this issue was
addressed directly for the key areas of interest for testing HS2 options.

5082342/5082342 Model Framework Validation Report (26-02- 17


10).doc
HS2 Model Framework Validation Report

3.4 Rail Assignment


3.4.1 Overview
This section outlines the data sources, and the validation of the PLD assigned rail flows against
the available data sources.

3.4.2 Data Sources


Time and budget constraints precluded against a major data collection exercise to provide a truly
independent set of comprehensive observed passenger flows. In addition, the technology to
collect such data in a reliable and consistent way is poorly developed. Accordingly, it was
necessary to rely on the best available data; this is common practice with rail passenger data.
It should also be borne in mind that the model was developed to represent a typical weekday for
2007/08, so a major data collection exercise in 2009 would have not been completely appropriate.
The available count data is drawn from various sources, as shown in Table 3.2 below:
Table 3.2 - Data Sources for PLD Rail Validation

Data Source Location Comments


Terminus counts from 2006 and 2007 by King’s Cross Data is only available for
TOC St Pancras long-distance TOCs.
Euston
MOIRA Midlands flow data for 2007/08 Midlands Not fully independent data
source
MOIRA Northern flow data for 2008/09 North Not fully independent data
source
Guard Counts Data AM Peak East Midlands Inappropriate as PLD is all-
Inter City East Coast day model, although used
for PS validation.
Inter City West Coast

The suitability of the data sources are explained in turn in the following section. The main category
of data to be unsuitable for use is the guard count data from the AM peak. This was available for:
 Sheffield and Leicester stations (East Midlands Trains) for the AM and PM peaks on a single
day in April 2008. This data was collected in terms of boarding and alighting passengers,
together with departing train total loads.
 Various points on National Express East Coast (Leeds, Newcastle, Edinburgh and Glasgow):
counts were undertaken over 5 days in November 2007, these collected the loadings on
trains at various points on the network en route to or from King’s Cross in the AM and PM
peaks.
 Various locations on Virgin West Coast services (Coventry, Wolverhampton, Crewe and
Stoke). This data was taken between September and November 2007, and was broken down
into average weekdays, first and standard class.
This data is not directly usable to validate an all-day model, as further factoring is required to uplift
to daily level. This estimation of locally specific factors from peak to all-day counts violates the
principle of independently observed data, and defeats the object of validation. It could be
considered for use in the validation of a peak period model, such as PS or PM, although it is of
limited coverage - spatially and only represents a single day.
5082342/5082342 Model Framework Validation Report (26-02- 18
10).doc
HS2 Model Framework Validation Report

3.4.2.1 London Terminal Station Counts


This data was available for intercity TOCs at the following London terminal stations:
 King’s Cross (Inter City East Coast). Daily counts were made available by DfT, taken
between September and December 2006 at King’s Cross, broken down by first and standard
class seats, for departing and arriving services. This was based on a full three months data
set for the GNER TOC.
 St Pancras (Midland Main Line). A single set of counts was made available for Autumn 2007
at St Pancras station, broken down by first and standard class seats, for departing and
arriving services. The data was averaged into two groups: Monday to Thursdays, and
Fridays.
 Euston (West Coast Trains). A set of counts was made available, taken between September
and November 2007. The counts were averaged for each weekday, first and standard class.

3.4.2.2 MOIRA Flow Data


The MOIRA data was taken from the following sources:
 MOIRA West Midlands base loading data for 2007/08. This means the same timetable and
demand data as used in PLD was available
 MOIRA Northern base loading data for May 2009. This is closest to the timetable and
demand data in PLD.
Both of these are not strictly 'independent' data, as they were also created from LENNON data, as
was the demand data assigned in PLD. They are nevertheless a valuable benchmark against
which to compare in the absence of any other independent count data.
One issue is that both the MOIRA loadings and the PLD demand are different representations of a
typical day, derived by different approaches to converting annual data to daily data. MOIRA is
described as representing a “typical Wednesday”, implying a usual day between Monday and
Thursday. PLD Rail Demand represents an average day, based on Monday – Friday data.

3.4.2.3 Data Summary


Some of the data is time period specific, while other data is TOC specific. Ideally, all-day data
covering all TOCs is required for validation on any given link, though this rarely available at the all-
day level.
For the purposes of PLD validation, a combination of London termini counts and the MOIRA loads
would appear to be the best source of such data.
The validation was carried out using all suitable available data. This comprised the following
elements:
 London Termini Counts by TOC;
 London to West Midlands MOIRA screenline;
 North of Midlands MOIRA screenline; and
 MOIRA data at other locations.

5082342/5082342 Model Framework Validation Report (26-02- 19


10).doc
HS2 Model Framework Validation Report

3.4.3 Assignment Validation


3.4.3.1 London Terminus Counts
This data is a useful check against the assigned flows on each TOC, but is incomplete in terms of
other TOCs operating from the same stations. This potentially implies that if data for a particular
TOC does not match, it will not be clear whether the overall loading on the link is incorrect, or
whether the balance between TOCs in that corridor is incorrect.
In conclusion, this data is used to validate the respective TOCs at their London termini. This is
shown in Figure 3.1. The counts represent the long-distance TOCs only.
!(

!(

i
Te rmin
don
Lon
Kings Cross (London)
!(
!(

St Pancras (London)
!(
!(

!(

Euston (London)
!(
!(

Figure 3.1 - London Termini Screenline Location


The network diagram is shown to give context to the comparison locations. The results are shown
in Table 3.3:

5082342/5082342 Model Framework Validation Report (26-02- 20


10).doc
HS2 Model Framework Validation Report

Table 3.3 - London Termini Screenline

Route / Station Direction

Guard Count

% Difference
Strategic

(Observed)

PLD Model
(Modelled)

Difference

Pass / Fail
TOC

West Coast Euston Outbound 19124 22343 3219 17% Pass


Main Line
Inbound 18717 22066 3349 18% Pass

Midland St Pancras Outbound 12975 9906 -3069 -24% Pass


Main Line
Inbound 12307 9857 -2450 -20% Pass

East Coast King’s Cross Outbound 15106 15949 843 6% Pass


Main Line
Inbound 14025 16453 2428 17% Pass

Total Outbound 47205 48198 993 2% Pass


Inbound 45049 48376 3327 7% Pass

At an individual count location level, all sites validate within the WebTAG guidance of 25%, and
the London Terminus screenline validates well within the 15% criteria.
On the key route of interest for this study – the WCML route – the overestimate of PLD in
comparison to guards counts could be due to over-allocation of passengers to the strategic TOC
on the route within PLD in comparison to the local operator. However no all-day counts for any of
the three principal north-south lines are available, so it is difficult to confirm this theory.
In addition, guards counts generally tend to be higher than modelled data from ticket sales for
several reasons including:
 Guards counts measure passengers carried on each service rather than “tickets carried”. A
significant number of advance purchase tickets are unused on services. While this results in
a slight bias towards over-estimating modelled train usage from using ticket sales data,
passengers carried data can similarly result in an underestimate of associated ticket revenue;
and
 As responsibility for undertaking counts has generally been a lower priority for on-board train
staff than core duties of operations, train safety or passenger care, guards counts tend to be
missed on busier services when train staff are more concerned with core duties.

3.4.3.2 London to West Midlands


Two adjacent screenlines was created to ensure that the number of rail trips crossing between
London and the West Midlands is broadly correct. The screenlines cover the plausible routes from
London to Birmingham:
 Via Oxford (Cross Country);
 Via Bicester (Chiltern); and
 Via Milton Keynes (West Coast).
This is shown in Figure 3.2.

5082342/5082342 Model Framework Validation Report (26-02- 21


10).doc
HS2 Model Framework Validation Report

The comparison data was MOIRA West Midlands base loading data for 2007/08. This means the
same timetable and demand data as used in PLD was available; however, it is not truly
independent validation data, as MOIRA flows are based on LENNON data.
!(
(!

r
ppe
s: U
!(

nd
idla
of M
!(

!(

th
Sou
!( !(

Milton Keynes Central !(

!( !(

er
!(
!(
!(

ow
!(

:L
!(
!(
!(

ds
!(

an
M idl !(

f
ho
ut
So
Bicester North
!(
!(
!(

!( !(
!(

!(
!(

!(

!(
!(
!( !(

!(

!(

!( !(

!(

!(

!(

!(

Oxford !(
!(

!(
!(
!(
!(

!( !(

!(
!(

!(

!(
!( !(
!(

Figure 3.2 - South of Midlands Screenlines

The results are shown in Table 3.4 and Table 3.5 below.
Table 3.4 - South of Midlands Upper Screenline

Route Station Direction MOIRA PLD Model Difference % Pass /


Model (O) (M) (M-O) Difference Fail

West Milton Northbound 24445 26913 2468 10% Pass


Coast Keynes
Southbound 24397 26515 2118 9% Pass

Chiltern Bicester Northbound 2730 2766 36 1% Pass


North
Southbound 2651 2942 291 11% Pass

Cross Oxford Northbound 3882 3580 -302 -8% Pass


Country
Southbound 3957 3768 -189 -5% Pass
Northbound 31057 33259 2202 7% Pass
Total
Southbound 31005 33225 2220 7% Pass

5082342/5082342 Model Framework Validation Report (26-02- 22


10).doc
HS2 Model Framework Validation Report

Table 3.5 - South of Midlands Lower Screenline

Route Station Direction MOIRA PLD Model Difference % Pass /


Model (O) (M) (M-O) Difference Fail

West Milton Northbound 29483 27080 -2403 -8% Pass


Coast Keynes
Southbound 29433 27053 -2380 -8% Pass
Chiltern Bicester Northbound 3647 3530 -117 -3% Pass
North
Southbound 3668 3554 -114 -3% Pass

Cross Oxford Northbound 3535 3084 -451 -13% Pass


Country
Southbound 3328 3225 -103 -3% Pass
Northbound 36665 33694 -2971 -8% Pass
Total
Southbound 36429 33832 -2597 -7% Pass

As screenlines, the differences are well within the WebTAG validation guidance. Individually, all
movements meet the WebTAG validation guidance of being within 25% of observed on the
modelled link flows, and within 15% of the screenline as a whole.

3.4.3.3 North of Midlands


A smaller pair of screenlines was created on the main branches of the West Coast Main Line at
Crewe and Stoke. These screenlines are shown in Figure 3.3 below.
!(

!(

!(

!(

ppe r
Mid lands: U
North of
Crewe !(

!(
!(
!(

!(

r
we
!(

: Lo
Stoke-on-Trent nds
!(

a
dl
!( !(

f Mi
o !(

orth
!(
N !(

!(

!(

!(

!(
!(

!(

!(

Figure 3.3 - North of Midlands Screenlines

These smaller screenlines covers West Coast and Cross Country services. The results are shown
in Table 3.6 and Table 3.7.
5082342/5082342 Model Framework Validation Report (26-02- 23
10).doc
HS2 Model Framework Validation Report

Table 3.6 - North of Midlands Upper Screenline 

Station Direction MOIRA Model PLD Model Difference % Pass /


(O) (M) (M-O) Differenc Fail
e

Crewe Northbound 8270 8103 -167 -2% Pass


Southbound 8103 8146 43 1% Pass
Stoke Northbound 7229 7695 466 6% Pass
Southbound 7466 8356 890 12% Pass
Total Northbound 15499 15798 299 2% Pass
Southbound 15569 16502 933 6% Pass

Table 3.7 - North of Midlands Lower Screenline

Station Direction MOIRA Model PLD Model Difference % Pass /


(O) (M) (M-O) Differenc Fail
e

Crewe Northbound 8727 10242 1515 17% Pass


Southbound 8544 9579 1035 12% Pass
Stoke Northbound 7772 8388 616 8% Pass

Southbound 7953 8879 926 12% Pass


Total Northbound 16499 18630 2131 13% Pass
Southbound 16497 18458 1961 12% Pass

Individually, all movements meet the WebTAG validation guidance of within 25% of observed on
the modelled link flows, and within 15% as a screenline.

3.4.3.4 Doncaster
East Coast Main Line trips were extracted from MOIRA Northern to give count data either side of
Doncaster. This is close to the timetable and demand data in PLD.
The locations are shown in Figure 3.4 below:

5082342/5082342 Model Framework Validation Report (26-02- 24


10).doc
HS2 Model Framework Validation Report

!(

Doncaster: Upper

Doncaster
!(
!(

Doncaster: Lower

Figure 3.4 - Doncaster

The counts are shown in Table 3.8 and Table 3.9 below.
Table 3.8 - Doncaster Upper Screenline

Route Station Direction MOIRA PLD Model % Pass / Fail


Model (O) (M) Difference

East Doncaster Northbound 13364 14836 11% Pass


Coast
Southbound 13318 15368 15% Pass

Table 3.9 - Doncaster Lower Screenline

Route Station Direction MOIRA PLD Model % Pass / Fail


Model (O) (M) Difference

East Doncaster Northbound 13442 15900 18% Pass


Coast
Southbound 13151 15647 19% Pass

The counts show that both sides of the Doncaster, train loads are well within the 25% criteria.
There was no suitable data to create a wider screenline in the area.

5082342/5082342 Model Framework Validation Report (26-02- 25


10).doc
HS2 Model Framework Validation Report

3.4.3.5 Newcastle
Similarly, East Coast Main Line trips were extracted from MOIRA Northern to give comparison
data either side of Newcastle. The locations are shown in Figure 3.5 below.

er
pp
: U er
tle ow
as L
ewc le
:
N st
ca
ew
N
!(

Newcastle
!(
!(

!(

!(

!(

Figure 3.5 - Newcastle


The comparisons are shown in Table 3.10 and Table 3.11 below.
Table 3.10 - Newcastle Upper Screenline

Route Station Direction MOIRA PLD % Pass / Fail


Model (O) Model Difference
(M)

Northbound 3978 4525 14% Pass


East Coast Newcastle
Southbound 3845 4740 23% Pass

Table 3.11 - Newcastle Lower Screenline

Route Station Direction MOIRA PLD % Pass / Fail


Model (O) Model Difference
(M)

Northbound 6177 7132 15% Pass


East Coast Newcastle
Southbound 5900 7517 27% Fail

The comparisons are all within 25%, except for loadings departing southbound from Newcastle.
This overestimation is likely to be due to the balance between long distance and more local
services: in many cases the balance between long distance and local services is affected by the
forced used of strategic stations, as non-strategic stations are effectively locked out of use.
5082342/5082342 Model Framework Validation Report (26-02- 26
10).doc
HS2 Model Framework Validation Report

The absence of a local model for the area also means that the local demand is concentrated on
the limited strategic corridor, leading to over assignment.
It follows that such over assignment of demand does not reflect a significant error in the model. In
addition, this corridor is far removed from the core corridor (London to the West Midlands) and any
marginal discrepancies in the level of demand and consequential crowding are unlikely to be of
any significance to the HS2 forecasts and Business Case.

3.4.4 Conclusions
Providing comprehensive independent validation of PLD’s rail model will, in time, require further
data collection and analysis. However, on three core indicators the model performs well:
 Comparison of matrix levels with published region-to-region movements shows an excellent
correspondence at an annual level, providing reassurance that the overall volumes of trips in
the model and associated annualisation factors are appropriate;
 The model passes WebTag requirements for validation against independent guards’ counts
data for the main north-south routes affected by high speed rail; and
 Although no independent validation data exists, cross-comparison with other models also
shows good results on the main HS2 corridor and is generally within WebTag requirements
on all other main corridors.
The fact that the PLD rail model validates less well in areas where there are significant numbers of
local trips outside the main HS2 corridor emphasises the value of developing the framework:
validation is far improved in the London & South East and West Midlands areas due to the link
with the local models. If HS2 proposals are developed in more detail in the Leeds and Manchester
area then further linkages with local models in those areas would be likely to improve validation
significantly.
Overall, validation of the PLD rail model on strategic corridors meets the requirements for
development of the HS2 business case.

5082342/5082342 Model Framework Validation Report (26-02- 27


10).doc
HS2 Model Framework Validation Report

3.5 Air Model


3.5.1 Overview
This section outlines the data sources, checks undertaken and validation performed on the PLD
air model.
The air model is included in PLD, principally to provide services and trips to compete with the rail
services. The air model seeks only to include services directly comparable with rail trips, so
services and trips to the Channel Islands, the Isle of Man, or any Scottish islands are not
considered. This is discussed further in the Model Development Report.

3.5.2 Data Sources - Air


SKM provided two-way, end-to-end domestic passenger movements at Heathrow airport. These
were used to check that the air model within PLD was validating appropriately.
Further data was not available, but in broad terms, the important axis is that between London and
the major cities, as this is where the competition with rail (and high speed rail) is to be found.
PSM allowed airport trips to choose between available airports using the strategic highway
network, and this functionality is retained.

3.5.3 Logic Checks


Associated sense-checks were:

3.5.3.1 'Fare Paid' Checks


Checking that the “fare paid” is non-zero for airport available trips, using PLD skims (this ensures
that no trips from, say Berkshire to Buckinghamshire are contained within the air matrix); and

3.5.3.2 Catchment Area Checks


Catchment areas by airport for trips to London and South East England were prepared. Analysis
indicated that more than 70% of the air demand to South East England is generated from
Scotland. Manchester and the North East of England generate the other reasonably significantly
demand. This is summarised in the following table.
Table 3.12 - Air Demand Check

% of Demand to SE England
G. Manchester (& Cheshire) 12%
North East (& Cumbria) 9%
Scotland 72%

A sense check with the CAA “end to end” data at Heathrow airport suggests this data to be in the
right order with more than half of all end to end trips to Heathrow being from Scotland.
Manchester and Newcastle are the other largest generators of demand to Heathrow.

3.5.3.3 Journey Time Data Checks


There was no requirement to validate journey time data for the air model, as these were extracted
from published timetables, and are not subject to modification by congestion in the way they are
for rail or highway. Furthermore, air is a special case where the flight time is very much dominated
by the rest of the air journey, namely access, check-in and egress times.

5082342/5082342 Model Framework Validation Report (26-02- 28


10).doc
HS2 Model Framework Validation Report

3.5.4 Validation
The following validation checks were undertaken on the updated air demand matrices:
 Loading figures on all flights, to ensure reasonable levels of passengers per plane;
 Assignment validation against the SKM/CAA data.

3.5.4.1 Loading Realism Checks


As there is no capacity constraint on air passenger flows within PLD, the loading figures for each
of the transit lines were checked to ensure no flights were carrying unrealistic numbers of
passengers. Only 10% of flights had greater than 100 passengers, and the average is around 57
passengers. This is a reasonable set of loadings.

3.5.4.2 Assignment Validation


The validation comparison is shown in Table 3.13. The model flows have been uplifted (by
application of a multiplicative factor of 365) to be comparable with the available CAA data.
These figures represent domestic travel by air, passengers “inter lining”, or connecting with an
international flight, are not included.
Table 3.13 - Comparison of CAA Annual Passenger Data & Modelled Flows at Heathrow Airport

Airport CAA Observed Modelled Percentage Pass / Fail


Passenger Data Passenger Flows Difference
Aberdeen 303,536 323,755 7% Pass
Edinburgh 695,952 719,780 3% Pass

Glasgow 491,626 516,110 5% Pass


Inverness 25,511 22,630 -11% Pass
Leeds / Bradford 84,813 92,345 9% Pass
Manchester 704,639 644,955 -8% Pass
Newcastle 280,532 269,735 -4% Pass
Teesside 51,533 54,385 6% Pass

Total 2,638,142 2,643,695 0% Pass

3.5.5 Conclusions
The PLD air model validates very well in terms of the flows into and out of London, and has
realistic numbers of passengers on all air routes modelled.

5082342/5082342 Model Framework Validation Report (26-02- 29


10).doc
HS2 Model Framework Validation Report

3.6 Highway Model


3.6.1 Overview
This section outlines the data sources available and checks undertaken for the PLD highway
model.

3.6.2 Limitations of the PLD Highway Model


The highway model is used in a relatively limited way in PLD:
 Access and egress to rail 4 and air 5 services; and
 Strategic highway trips by car.
That is to say, PLD does not model either local highway trips, or the local highway network. This is
due partly to the lack of nationally comprehensive information on local trips, and partly the
strategic nature of the model, which is not designed to model local congestion beyond that on the
main strategic highway network.
The missing local trips are represented by pre-loads on each link. The main purpose of the pre-
loads is to load the network to realistic levels, to allow the speed-flow curves to function correctly.
The pre-loads are calculated as the 'gap' between the assigned flows from the matrices and the
available count data.
The main use of the highway model is to provide comparative generalised costs for highway trips
to compare against rail or air trips. It is the capability to provide a change in journey time, in
response to a change in congestion (resulting from a change in strategic long distance car
demand), which is of most importance.

3.6.3 Data Sources


3.6.3.1 Matrices
No national matrix of strategic highway movements is currently available, so it is not possible to
validate at the matrix level.

3.6.3.2 Link Flows


The pre-loads were calculated by use of Highways Agency count (TRADS) data. Since this data
was used to create the pre-loads, it makes no sense to validate against them. There is no other
equivalent source of observed national traffic flows.

3.6.3.3 Journey Times


A selection of journeys between key cities was used to validate the highway model speeds and
journey time. For each journey, three times were obtained from journey planning websites. The
average of these was compared to the PLD highway time.
Journey Times was extracted from three trip-planning websites:
 AA;
 Green Flag; and
 RAC.

4
Car access at a maximum of one end of the rail trip only (the other ends at a fixed speed as a proxy for
local public transport access).
5
Car access at both ends of an air trip.
5082342/5082342 Model Framework Validation Report (26-02- 30
10).doc
HS2 Model Framework Validation Report

For the reasons outlined, the only appropriate validation to undertake is that of the highway
journey times. The results are shown in Section 3.6.4.

3.6.4 Journey Time Validation


Table 3.14 compares these reported journey times for key city to city movements with the
equivalent journeys forecast in the PLD highway model
Table 3.14 - Highway Journey Time Validation (Minutes)

From To Reported Travel Times

PLD Model Time

(PLD-Average)

% Difference
Difference

Pass / Fail
Green Flag

Average
RAC
AA

Birmingham Leeds 131 127 128 129 118 -11 -9% Pass
Birmingham Manchester 109 101 101 104 110 6 5% Pass
Heathrow Birmingham 120 116 124 120 107 -13 -12% Pass

Heathrow Leeds 202 193 211 202 194 -8 -4% Pass


Heathrow Manchester 215 203 220 213 210 -3 -1% Pass
London Birmingham 148 129 137 138 121 -17 -14% Pass

London Bristol 142 125 128 132 140 8 6% Pass


London Leeds 218 202 221 214 191 -23 -12% Pass
London Manchester 234 216 232 227 207 -20 -10% Pass

Manchester Glasgow 184 212 226 207 208 1 0% Pass

It can be seen that all journeys are within ±15% of the average times.
Generally, reported travel times in PLD tend to be faster than the trip planning data. This is due to
the fact that PLD does not include a detailed representation of the final “local leg” from the
strategic highway network to each zone (“zone connectors”). Since the mode-choice model is
incremental in structure, the urban sections at the extreme ends of the journey do not need to be
modelled in great detail in the end-to-end journey time by highway: it is the 'modellable' change in
journey time on the strategic part of the route, rather than the 'un-modellable' local congestion,
which is unlikely to be impacted by a model scenario test, and is not the focus of the model.

3.6.5 Conclusions
This validation check shows that the model validates well against the only appropriate measure for
the highway mode. The PLD highway model is sufficiently accurate in representing journey times
between major urban centres, albeit slightly faster for the reasons outlined in section 3.6.4 above.
This effect is not of major importance, as the role of the highway model is to represent the change
in journey time associated with a change in congestion (resulting from a change in car demand).

5082342/5082342 Model Framework Validation Report (26-02- 31


10).doc
HS2 Model Framework Validation Report

4. PLANET South Model Validation

4.1 Overview
PS was refined and revalidated in early 2009, including a validation exercise based on TOC
Green Book / PIXC 6 count data. The validation exercise was reported separately in DfT reports 7 .
PS is generally accepted for use on London & SE rail forecasting exercises, subject to validation
exercises on specific corridors where options are being tested.
Only relatively small changes to PS have been made as part of the development of the
framework: the key change has been the separate treatment of peak period long-distance
demand in the model. Whereas the original PS includes matrix representation of long-distance
demand, the framework cuts back the main scope of the model to places closer to London.
Longer-distance demand, calculated in PLD, is fed into PS at the new boundary through new
zones created for that purpose, and appropriately scaled back to represent morning peak period
demand. Certain other minor amendments were also made to PS.
Accordingly, this section shows how the flow validation of the integrated version of PS differs from
the previous full validation, and assesses its appropriateness for use for assessing possible uses
of released capacity on the WCML route as part of HS2 options.

4.2 Flow Validation


4.2.1 Comparison with validation data
There are two data sources available:
 Observed TOC data through the annual Green Book count process, which measures the
highest loading of peak period trains into and out of London;
 Independent modelled loading data from DfT’s Network Modelling Framework (NMF) for the
High Level Output Specification (HLOS) in 2007, which provides an alternative view based on
ticket sales and other count data;
As explained previously in this report, caution has to be applied in using and interpreting count
data for two reasons: first, the data reflects a single day’s count, which can vary by up to 20%
throughout the week and throughout the year. The model reflects annual ticket sales with
assumed conversion to an average weekday: as such, there will also be uncertainty around
whether disparity between modelled figures and counts reflects genuine differences in overall
volumes or the inherent difference between average and typical figures.
Second, Green Book counts are taken at the peak loading point on services. For many commuter
services in Greater London, the point of peak loading is often not at the London terminal station
but at inner London NR / LUL / DLR interchange stations such as at Finsbury Park and West
Hampstead. In addition, the point of peak loading can vary between service groups and within the
peak. While data can be extracted from PS for different loading points, the strategic level of detail
of PS – especially frequency based assignment – means that direct comparisons are not robust.
The independent HLOS / NMF figures provide an alternative view of processing ticket sales data

6
PIXC = Passengers in Excess of Capacity, annual monitoring data of London and South East TOC train
loadings during morning and evening peak periods.
7
“PLANET South AM model development and updates (PSAM v4)”, Atkins October 2009
5082342/5082342 Model Framework Validation Report (26-02- 32
10).doc
HS2 Model Framework Validation Report

and count data; however it should be stressed that this data is not independent count data, and
there is no inherent reason why it should be more reliable than the PS data.
Table 4.1 sets out the independent validation data alongside the framework and non-framework
versions of PS for comparison.
Table 4.1 - PLANET South Validation Flows (0700-1000 arrivals in Central London, 2007)
Route / Green Book PS Non-
HLOS Data PS Framework
Count Point (PIXC) Counts Framework
Great Western
Main Line 22,973 24,100 20,210 21,314
(Paddington) 8
Chiltern
Main Line 10,222 9,100 8,535 7,766
(Marylebone)
West Coast
Main Line 17,256 23,800 18,958 16,434
(Euston) 9
Midland
Main Line 23,543 25,900 20,213 23,556
(St Pancras)
East Coast
Main Line 32,752 31,300 32,737 32,250
(Finsbury Pk)
Total 106,746 114,200 100,653 101,320

Overall validation of the framework against Green Book counts is maintained, with individual
improvements in validation on the West Coast, Great Western and Midland Main Line routes and
a slight worsening on the Chiltern route. PS validation on the Chiltern route is known to be
relatively poor because of the assignment allocation mechanisms between LUL Metropolitan Line
and NR Chiltern services on the Harrow-on-the-Hill corridor.
It should be noted that PS does not include specific representation of demand related to Heathrow
Express in the model, which is not included within the core “travelcard” type data set used to
develop the model. This reflects the relative difficulty in representing premium travel to Heathrow
Airport in a model which ostensibly does not have fares representation – simple inclusion of
demand could make validation worse rather than better and would not represent premium airport
passenger behaviour in any meaningful way.
This means that there is likely to be an underestimate of demand to Heathrow Airport from central
London in PS both in base year and forecasting years. The effect of this on the HS2 business
case was judged to be minor, as crowding on Heathrow Express services is expected to be
extremely limited. For subsequent stages of HS2 scheme development, where more detail is
needed on patterns of access from Greater London to Heathrow Airport it will be appropriate to
extend this representation further.

8
Heathrow Express figures not available or included in observed or modelled figures. It is unclear whether
this demand is included in HLOS figures.
9
London Overground figures not available or included in observed or modelled figures. It is unclear whether
this demand is included in HLOS figures.
5082342/5082342 Model Framework Validation Report (26-02- 33
10).doc
HS2 Model Framework Validation Report

4.2.2 Conclusions
On the critical WCML corridor, the framework slightly under-predicts morning peak demand on the
route, however validation is within 5% of observed figures, generally regarded as very good
validation for public transport modelling and well within daily variation of passenger counts.
No other routes are affected by HS2 option tests; however validation of the rest of the model is
generally unchanged between the framework and non-framework versions of the model.

5082342/5082342 Model Framework Validation Report (26-02- 34


10).doc
HS2 Model Framework Validation Report

5. PLANET Midlands Model Validation

5.1 Introduction
PM was developed as a new demand forecasting model in early 2009, covering morning peak rail
movements in the East and West Midlands areas. A validation exercise was undertaken for the
model in the West Midlands area, focused on local commuter movements into central
Birmingham.
Similar to the process used for the PS model, PM was cut back in scope to provide greater detail
on the effects of capacity release by high speed rail on corridors into central Birmingham. Long-
distance demand is fed into the model through pre-loads on transit-lines using information from
PLD.
The focus of the validation is on corridors into central Birmingham which could be affected by
capacity release through reductions or removal of long-distance passenger demand between
London and Birmingham. This principally affects three corridors:
 West Coast Main Line between Coventry and Birmingham New Street, and between
Wolverhampton and Birmingham New Street; and
 Chiltern Main Line corridor between Leamington Spa and Birmingham Moor Street;
Accordingly, this section shows how the flow validation of the integrated version of PM differs from
the previous full validation, as reported by Jacobs Consultancy in “PM Validation Summary
Report” dated November 2009, on these corridors.

5.2 Flow Validation


5.2.1 Comparison with validation data
Independent validation data for the West Midlands was used from the PM validation report. Table
5.1 below compares flows for the three relevant corridors for the framework and non-framework
versions of PM, together with the observed count data.
Table 5.1 - PM Validation Flows (To Birmingham New Street)

Corridor From Observed Flow PM Non-Framework PM Framework


West Coast Main Line
(Coventry Corridor) 4,228 4,363 (+3%) 3,416 (-19%)
Solihull Corridor to New St
(Long-distance TOCs) 311 235 (-24%) 148 (-53%)
West Coast Main Line
(Wolverhampton Corridor) 4,647 6,074 (+31%) 4,805 (+3%)
All Corridors 9,186 10,672 (+16%) 8,369 (-9%)

The figures show the validation worsens on the routes from the south and east, but improves on
the Wolverhampton corridor. The reduction in validation and passenger volumes on the Coventry
and Chiltern corridors could be due to a number of possible causes:
 Usage levels of long-distance services during peak periods is higher than estimated within
the model development process, resulting in modelled volumes of long-distance passengers
being lower than forecast fed through from PLD to PM;

5082342/5082342 Model Framework Validation Report (26-02- 35


10).doc
HS2 Model Framework Validation Report

 Underlying poor allocation between long-distance and local demand within the underlying PM
model demand data, resulting in a lack of local demand left within the framework; and
 Unlike PS, onward passenger demand from Birmingham New Street and International
stations to local stations in the West Midlands area is not included in the local model. For
modelling purposes, these trips are assumed to start and end at principal stations served by
long-distance services. This may also contribute towards an underestimate on local services
in PM.
It is worth noting that, even with the reduced validation, the models still meet WebTag validation
requirements on the main WCML corridor through Coventry and Wolverhampton. On the Chiltern
corridor – where only a handful of services operate directly to Birmingham New Street during the
morning peak period – validation appears poor but is based on an exceptionally small flow
relatively to the numbers of people using New Street in the morning peak period.
Overall, the reduced validation may result in a slight underestimate of demand, and hence of
crowding relief provided by high speed services on the Coventry corridor in the West Midlands
area. Further investigation of the balance between local and long-distance demand in PM in the
West Midlands area is being undertaken which will enable improvement in validation levels on key
corridors – however the lack of comprehensive data has meant that this investigation could not be
completed within study timescales.

5.2.2 Conclusions
On the main WCML corridor affected by capacity release, the framework under-predicts morning
peak demand on the route compared to both observed data and the non-framework model. This is
likely to result in an underestimate of the impacts of crowding relief on the route.
However this impact is small in terms of the overall benefits from high speed rail, and is
conservative, i.e. improvements to model validation will tend to strengthen the business case for
high speed rail.
Further data collection on peak usage of long-distance services into the West Midlands will
improve the validation on this corridor during any potential further development phases of high
speed rail options.

5082342/5082342 Model Framework Validation Report (26-02- 36


10).doc
HS2 Model Framework Validation Report

6. Conclusions
6.1 PLD Model
The PLD models validate well where information is available at the correct level:
 All day, all TOC counts on rail links;
 Airport trip totals; and
 Journey times by car.
The model can be used with confidence for strategic rail service option testing, particularly in the
corridor between London and the West Midlands.
The PLD models are more likely to over-estimate than under-estimate demand, and hence
crowding levels, in conurbations beyond the main HS2 corridor where integration with local
models has not been undertaken as part of this work, however this is expected to have only a very
small impact on the overall HS2 business case or development of options.

6.2 PLANET South Model


The validation checks undertaken have shown that the core validation of the PS model has been
either maintained or improved by linking into the wider HS2 forecasting framework, especially on
the core HS2 route.
Going forward, detailed HS2 scheme development around Heathrow will require development of a
more sophisticated approach to rail and public transport access to Heathrow Airport from Greater
London which takes into account the premium fares on Heathrow Express and any potential
changes to this policy with the introduction of Crossrail in future years.

6.3 PLANET Midlands Model


Integration of the PM model into forecasting framework has maintained model validation on the
main corridors affected by HS2 proposals and released capacity. However, it should be noted that
the framework version of PM underestimates peak period passenger demand on the main
Coventry corridor into Birmingham. This will tend to result in an underestimate of the benefits of
released capacity from HS2 proposals on the corridor.

6.4 HS2 Model Framework


The constituents of the model framework have been developed and validated and checked
against available data sources in accordance with normal industry “best practice”.
It has been shown to validate for the base year 2007/08 in the key areas for the HS2 project
according to DfT WebTag guidelines. In the few areas where validation is not achieved, these are
not significant as far as the use of the framework in developing forecasts for HS2 on the key
corridors of interest.
When the project proceeds to a detailed design and public scrutiny stage, it is expected that
additional independent validation data would be collected to allow further and more detailed
scrutiny of the model and inform detailed design decisions.

5082342/5082342 Model Framework Validation Report (26-02- 37


10).doc
HS2 Model Framework Validation Report

Appendix A – Rail Service Checks


This was undertaken in the following way, for all models (PLD, PS and PM). This section explains
the checks undertaken when converting the CIF files to PLANET transit line coding. It also
explains the data issues encountered with the CIF.
The checks undertaken are broken down into: Issue, Check and Result. The result shown is
generally the end result rather than the interim results. This is because the majority of checks
were carried out many times over, until the number of errors diminished to zero.
The following checks are carried out each time a modification is made to the specification of the
Perl 10 script. The following checks are made until the errors fall to zero.
All TIPLOCs resolve to a node
Issue: To confirm that each TIPLOC 11 mapping to a ‘relevant’ node was given a node
number.
Check: ‘Find in Multiple Files’ in TextPad was used with a regular expression with all output
files to find any transit segment starting with two spaces followed by a letter:
^ [A-Z]
Result: All segments converted to a node number.

Correct Routing
Issue: To confirm that the correct series of nodes are visited by the transit lines
Check: Sorting transit lines by number of segments in Emme 3 to identify lines with
suspiciously high numbers of segments. These often signify a transit line where an
incorrect node is specified, causing a large detour to be required.
Result: Transit line routings are sensible.

All lines are named


Issue: All lines are named according to their origin and/or destination. Any line failing to be
treated by the process was left with the original CIF identifier of a letter and 5
numbers.
Check: ‘Find in Multiple Files’ in TextPad was used with a regular expression with all output
files to find any transit line header starting with a letter followed by a number:
^a ‘[A-Z][0-9]
Result: All transit lines have a name in the correct format of AA111A.

No import errors
Issue: All transit lines should import without error.

10
Perl: Practical Extraction and Reporting Language. See http://www.perl.org
11
TIPLOC: Timing Point Location
5082342/5082342 Model Framework Validation Report (26-02- 38
10).doc
HS2 Model Framework Validation Report

Check: The error count in the import report files.


Result: No errors.

Trains Per Hour on Links


Issue: The trains per hour by each TOC on each link needed to match a published source
(Railplanner software, paper timetable, PIXC counts at terminus stations)
Check: The number of trains per hour on a link was compared with the existing reference
case. These were largely consistent, and areas where matches were less
convincing were investigated in more detail.
Result: The transit frequencies matched the published data in spot checks.

Train Aggregation
Issue: Pre- and Post-Aggregation trains per hour should be consistent.
Check: A check was made to confirm that the number of trains was equal.
Result: This was found to be correct, subject to minor rounding differences (such as where
7 trains individual trains becomes a train every 25.71 minutes).

Trains Per Hour stopping at Nodes


Issue: The number of trains per hour stopping at nodes should be correct.
Check: The number of trains stopping was checked in Emme 3 by displaying proportional
circles fed by the following configurable attribute:
ca_nlinesstop_i == isIStop*60/hdw
In particular, locations were checked where ca_nlinesstop_i==0, i.e. no trains
stopped there.
Result: This was useful to visually identify areas where a station node had an incorrect
number of stopping trains.

Trains Per Hour stopping at London Termini


Issue: The number of trains per hour stopping at London Termini should be correct.
Check: This was compared with Railplanner 12 software.
Result: This was found to match to a satisfactory level.

Redundant Services
Issue: Services running ‘perpendicular’ to the modelled axis had to be removed.
Check: The ratio of Route Length / Crowfly Length was used to find those services with very
high ratios. The ratios were used to find routes which were deemed not suitable to
be included, due to the peripheral nature of the routes.

12
Railplanner Rail Timetable Tool. See http://www.travelinfosystems.com
5082342/5082342 Model Framework Validation Report (26-02- 39
10).doc
HS2 Model Framework Validation Report

These services are removed by naming lines with such origins and destinations as
‘XX____’ and automatically deleting them after import.
Result: These are removed.

This process ensured that the rail services in the model were optimised to be as error-free as
possible within the time-scale.

5082342/5082342 Model Framework Validation Report (26-02- 40


10).doc

You might also like