appendix b3 – saturn model development report
TRANSCRIPT
Appendix B3 – Saturn Model Development Report
Sheffield & Rotherham District SATURN Model 2008
Model Development Report
Report for Sheffield City Council
September 2009
Document Control
Project Title: Sheffield and Rotherham District SATURN Model 2008
MVA Project Number: C37272
Document Type: Model Development Report
Directory & File Name: M:\tp\C37272 FC 36842 Sheffield Model
Update\Word\LMVR\Report\Final\Main Report v8.0.doc
Document Approval
Primary Author: Andrew Ford
Reviewer: John Allan
Formatted by: Andrew Ford
Distribution
Issue Date Distribution Comments
1 03/07/2009 SCC, MVA Draft
2 05/08/2009 SCC Draft Version 2
3 08/09/2009 SCC, Workspace Final
4
Contents
1 Introduction 1.1 1.1 Background 1.1 1.2 Model History 1.1 1.3 Structure of the Report 1.2
2 Model Overview And Dimensions 2.1 2.1 Introduction 2.1 2.2 The Study Area and Zone System 2.1 2.3 Base Year 2.3 2.4 Time Periods 2.3 2.5 User Classes 2.3 2.6 Network and Junction Characteristics 2.2
3 Data Collection and Collation 3.1 3.1 Introduction 3.1 3.2 Roadside Interview Surveys 3.1 3.3 Count Data 3.3 3.4 Independent Count Set 3.4 3.5 Journey Time Survey Data 3.4
4 Road Network 4.1 4.1 Introduction 4.1 4.2 Network Coverage 4.1 4.3 Data Required for SATURN 4.1 4.4 Range and Logic Checks 4.3
5 Road Traffic Demand Matrices 5.1 5.1 Introduction 5.1 5.2 Summary of the Matrix Building Process 5.1 5.3 Expand RIS 5.2 5.4 Reverse Synthesis 5.3 5.5 Calibrate gravity model and estimate unobserved car trips 5.9 5.6 Estimate LGV and OGV trips within study area sectors 5.12 5.7 Estimate external to external trips 5.12 5.8 Apply Growth to 2006 Matrices 5.13 5.9 Assignment 5.13
6 Model Assignment and Calibration 6.1 6.1 Introduction 6.1 6.2 Assignment Procedure and Convergence 6.1 6.3 User Classes 6.2
Contents
Sheffield and Rotherham District SATURN Model 2008 2
6.4 Generalised Cost Formulation 6.2 6.5 Calibration Procedure 6.4 6.6 Development of the 6 User Class Models 6.4
7 Validation 7.6 7.1 Introduction 7.6 7.2 Calibration count comparisons 7.6 7.3 Cordon Validation 7.14 7.4 Important Count Sets 7.16 7.5 Independent Count Validation 7.18 7.6 Journey Time Comparison 7.21 7.7 Inspection of Typical O-D Routes 7.24 7.8 Matrix Characteristics 7.25 7.9 Trip Length Distribution 7.26 7.10 Trip Ends 7.31 7.11 Origin Destination pairs 7.36
8 Summary and Conclusions 8.1 8.1 Summary 8.1 8.2 Conclusions 8.2
Tables
Table 3.1 Description of Journey Time survey routes 3.4 Table 5.1 Transpose 12 hour to Modelled Hour Percentages – by Purpose 5.4 Table 5.2 Trip-rates used in the Gravity Model 5.10 Table 5.3 Estimated Trip-end Totals and Fully Observed Demand 5.12 Table 5.4 ATC Traffic Growth Factors 5.13 Table 6.1 Convergence Statistics 6.2 Table 6.2 Generalised Cost Parameters – Six User Class 6.3 Table 7.1 Validation against all Calibration Counts prior to Matrix Estimation 7.9 Table 7.2 Validation against all Calibration Counts after Matrix Estimation 7.10 Table 7.3 Comparison of modelled flows against observed counts (All vehicles combined)7.11 Table 7.4 Screenline Flows across Sheffield and Rotherham Cordons – Before Matrix
Estimation 7.15 Table 7.5 Screenline Flows across Sheffield and Rotherham Cordons – After Matrix
Estimation 7.16 Table 7.6 Validation against Important Count Sets after Matrix Estimation 7.17 Table 7.7 Independent Count Validation Statistics 7.21 Table 7.8 Journey Times within 15% (or 1 minute if greater) 7.22 Table 7.9 Percentage of Routes passing DfT criteria 7.23 Table 7.10 Summary of Hourly 6 user class Trip Matrices (pcus) 7.25 Table 7.11 Average Car Trip Lengths Before and After Matrix Estimation 7.28 Table 7.12 Average Car Trip Lengths Before Matrix Estimation – All User Classes, to, from
and within Study Area 7.30 Table 7.13 Correlation between Trip end totals before and after matrix estimation 7.31 Table 7.14 Changes in Trip Ends Before and After Matrix Estimation 7.35
Contents
Sheffield and Rotherham District SATURN Model 2008 3
Table 7.15 Correlation between Origin-Destination pairs before and after matrix
estimation 7.36
Contents
Sheffield and Rotherham District SATURN Model 2008 4
Table of Figures
Figure 2.1 Sheffield and Rotherham SATURN Model Zone system 2.2
Figure 2.2 Sheffield and Rotherham SATURN Model zone system between Sheffield and Rotherham
urban centres 2.2
Figure 2.3 Sheffield and Rotherham SATURN Model zone system - Penistone Rd 2.3
Figure 2.4 Sheffield and Rotherham SATURN Model zone system - Sheffield City Centre 2.3
Figure 3.1 Location of Roadside Interview Surveys 3.2
Figure 3.2 Location of Classified Counts 3.3
Figure 3.3 Location of Independent Counts 3.4
Figure 3.4 Journey Time Survey Routes 3.6
Figure 4.1 Extent of Buffer and Simulation Networks 4.2
Figure 5.1 Matrix Building Flowchart 5.2
Figure 5.2 Schematic showing ‘observed’, ‘unobserved’ and ‘partially observed’ movements
5.7
Figure 5.3 Sector system used for matrix building 5.7
Figure 5.4 Various Observations of a particular O-D trip 5.8
Figure 7.1 Link Flow Validation Plot - Morning Peak 7.13
Figure 7.2 Link Flow Validation Plot - Inter-peak 7.14
Figure 7.3 Link Flow Validation Plot - Evening Peak 7.14
Figure 7.4 Location of Key Counts 7.15
Figure 7.5 Independent Count Validation Plot – Morning Peak 7.26
Figure 7.6 Independent Count Validation Plot – Inter-peak 7.26
Figure 7.7 Independent Count Validation Plot – Evening peak 7.27
Figure 7.8 Trip Length Distribution – Morning peak 7.28
Figure 7.9 Trip Length Distribution – Inter-peak 7.28
Figure 7.10 Trip Length Distribution – Evening-peak 7.29
Figure 7.11 Morning Peak Origin Trip Ends 7.33
Figure 7.12 Morning Peak Destination Trip Ends 7.33
Figure 7.13 Inter Peak Origin Trip Ends 7.34
Figure 7.14 Inter Peak Destination Trip Ends 7.34
Figure 7.15 Evening Peak Origin Trip Ends 7.35
Figure 7.16 Evening Peak Destination Trip Ends 7.35
Appendices
A Roadside Interview Programmes
B Roadside Interview Survey Variables and Value Labels
C Model Bandwidth Plots
D Counts Comparisons
E Journey Time Comparisons
F Route Checking
G Trip Matrix Summaries
Contents
Sheffield and Rotherham District SATURN Model 2008 5
H Trip Length Distributions
I Index of Important Files
J Method for Controlling Matrices to TEMPRO
Sheffield and Rotherham District SATURN Model 2008 i
Summary
Background
In autumn 2008, Sheffield City Council appointed MVA Consultancy to update their multi-
modal model for Sheffield and Rotherham. The new model will be used to support Major
Scheme Business Case submissions and inform future transport policy and strategy
development, post Local Transport Plan 2 (LTP2), within the Sheffield and Rotherham
districts.
The updated multi-modal model SRTM3 has a base year of 2008. It was developed from its
predecessor SRTM2, which had a base year of 2007. It includes three sub-models– a
demand model (SRDM3), a public transport model (SRPTM3) and a highway model (SRHM3).
All three sub-models have been updated – their predecessors were SRDM2, SRPTM2, and
SRHM2.
This report concerns the update of the highway model to form the 2008 highway model of
Sheffield and Rotherham - SRHM3. It describes the data sources and processes used to
calibrate and validate the model. It also presents the results of the validation and seeks to
demonstrate that the model is fit for the purpose of appraising transport schemes and
developments within the area of detailed modelling, an area that encompasses the entire
districts of Sheffield and Rotherham.
The update of the multi-modal model to 2008 had less effect on the highway model than the
other models. Unlike the other two, the highway model has remained on the same software
platform (SATURN) as its predecessor. It was built from the same set of origin-destination
surveys and it retains the same geographical coverage. The key changes is that it has been
calibrated to a set of counts that now include data in the city centre gathered after the
completion of the Northern Inner Relief Road major scheme in the city centre. It has
changed in other ways, with finer zoning on Penistone Road to the north of the city and
refinements made to the network in the City Centre. These changes were made in
preparation for forthcoming studies centred on those areas.
SRHM3 has been built from the following data sources:
2008 Vehicle Occupancy Counts;
2008 Manual Classified Counts – undertaken by SCC and RMBC;
2008 Roadside Interview Survey Data – undertaken in the Waverley area;
2008 Manual Classified Counts – undertaken by Sky High and Nationwide Data
Collection;
2008 to 2006 journey time data;
2007 to 2005 Roadside Interview Survey Data;
2007 to 2003 ATC Data;
2007 to 2003 Manual Classified Counts – undertaken by SCC, RMBC and other
sponsors; and
2001 Census data for all zones within the model updated to 2007 mid-year population
Summary
Sheffield and Rotherham District SATURN Model 2008 ii
Fitness for Purpose
The fitness for purpose of the model is demonstrated through:
the comparison of modelled and observed flows;
− the correlation between modelled and observed flows is good for all time periods
and user classes, with the number of links with a GEH value less than 5 falling
just short of the stringent 85% level. The R squared value, showing the match
between modelled and observed counts, is greater than 0.97 in all time periods.
the comparison of modelled and observed journey times;
− the journey time validation is satisfactory in the morning and evening peaks,
although the inter-peak is less satisfactory due to limitations with the data. This
is further described below;
the inspection of routes between key areas in the model;
− analysis of origin-destination routes between key centres shows that the routing
taken by traffic through the network are plausible;
the description of the processes used to build the model;
− a thorough description of the network and matrix building process is provided,
highlighting the tasks that have been undertaken to ensure the underlying data
behind the model is accurate and robust.
DMRB sets out different criteria for comparison of modelled and observed flows depending on
the magnitude of the flows. Its main guideline is that 85% or more flows pass the criteria.
Against the set of counts used in calibration, our model performs well, with flows in all time
periods passing the guidelines. For the morning peak, inter-peak and evening peaks
respectively 88%, 92% and 89% of flows pass the DMRB guidelines.
For journey times the DMRB requires that 85% of modelled times be within 15% of observed
times (or 1 minute if greater). Our model performs acceptably, almost attaining the
stringent DMRB standard in the morning and evening peak periods. In both the morning and
evening peak 81% of routes pass the DMRB criteria. The inter-peak validation looks poor,
with many routes running too fast. However, we consider that the inter-peak journey time
data is not a valid representation of true inter-peak journey times. This is because many of
the journeys on which it was based span the peak and inter-peak periods. Thus the
observed inter-peak journeys are slower than the average inter-peak would be. Compared
to these slow observations – the inter-peak model appears to run too quickly. However, we
believe the inter-peak model as it stands replicates routings successfully during the inter-
peak period
The validation is good along all corridors of the schemes being brought forward to MSBC
submission, namely Penistone Road Smart Route and the Sheffield to Rotherham BRT routes
(via Meadowhall and Waverley). Both the count and journey time validation along these key
routes is good, enabling us to confidently use the model to appraise future transport
interventions. Further details will be provided in bespoke model development reports that
will accompany each MSBC submission.
Summary
Sheffield and Rotherham District SATURN Model 2008 iii
Model Features
This Model Development Report describes the development and validation of the base year
(2008) Sheffield and Rotherham highway models (SRHM3) for the following three weekday
time periods:
weekday morning peak hour : 0800-0900;
weekday average inter-peak hour : 1000-1200 and 1400-1600; and
weekday evening peak hour : 1700-1800.
The six user class assignment segments the travel demand by income, a requirement of
models used for bids to the Transport Innovation Fund. The six user classes are as follows:
cars – employer’s business;
cars – commute and other – low income;
cars – commute and other – medium income;
cars – commute and other – high income;
light goods vehicles (LGVs); and
other goods vehicles (OGVs).
There are also fixed flows on the network which represent buses and trams.
Summary of Main Data Sources
The main data sources used to build the SATURN highway model were:
Roadside Interview Surveys (RIS) undertaken in 2005, 2006, 2007 and 2008
specifically for this model;
road traffic counts from 2005, 2006, 2007 and 2008;
journey time surveys 2008 (2 routes), 2007 (10 routes) and 2006 (4 routes);
traffic signal timings;
bus timetables; and
tram timetables.
Sheffield and Rotherham District SATURN Model 2008 1.1
1 Introduction
1.1 Background
1.1.1 In autumn 2008, Sheffield City Council appointed MVA Consultancy to recalibrate the existing
Sheffield and Rotherham Multi-Modal Model (referred to as SRTM2) to create a new 2008
base year model (referred to as SRTM3). SRHM3 will be used for several highway studies as
well as forming an integral part of the Sheffield and Rotherham District Multi-Modal Model
(called SRTM3) and the new South Yorkshire / Sheffield City Region Strategic Transport
Model (SYSTM+).
1.1.2 This report covers the highway element of SRTM3, called SRHM3. Separate validation reports
area available for the Demand model (SRDM3) and the Public Transport Model (SRPTM3).
1.2 Model History
1.2.1 The parent model for SRHM3 is SRHM1. This was the first version of the highway model to
form part of a multi-modal model. It extended the geographical coverage of the earlier
highway-only models to cover the whole of the two districts. It was built in 2006, from a
combination of new data and data gathered over the previous 5 years. It was calibrated
concentrating on Sheffield and Rotherham town centres, and also focussing on the Lower
Don Valley corridor. It was developed to support the development of the Sheffield &
Rotherham Bus Rapid Transit (Northern & Southern Routes) and Penistone Road Smart
Route Outline Business Case submissions to the Regional Transport Board and to undertake
initial option testing for the River Don District Masterplan.
1.2.2 By 2007, some of the roadside interview survey (RIS) data on the Sheffield city centre
cordon was 5 years old, too old to be used according to the DfT guidance. New RIS data
were gathered in 2007 for the update to SRHM2.
1.2.3 In 2007, traffic patterns had not yet fully settled down after opening of the Northern Inner
Relief Route in Sheffield City Centre. The update of the other components of the multi-
modal model to 2008 offered the chance to update the highway model again to reflect the
latest traffic patterns.
1.2.4 The development of the 2008 model (SRHM3) drew on a number of existing data sources:
the validated 2006 base year Sheffield and Rotherham SATURN model (SRTM1);
the validated 2007 base year Sheffield and Rotherham SATURN model (SRTM2);
existing RIS data;
manual classified counts and automatic traffic counts;
journey time survey data;
bus routes and frequencies; and
traffic signal timings.
1.2.5 The model required new data collection in the form of:
1 Introduction
Sheffield and Rotherham District SATURN Model 2008 1.2
3 roadside interview survey sites in Rotherham;
manual classified counts; and
new journey time data.
1.2.6 This version of the model, SRHM3, has a base year of 2008 and has the same coverage as
SRHM1 and SRHM2 i.e. the entire area covered by the districts of Sheffield and Rotherham.
It contains trip origin and destination data obtained from a programme of Roadside Interview
Surveys (RIS);
3 sites surveyed in Rotherham in 2008;
50 sites surveyed in Sheffield in autumn 2007;
42 sites surveyed in Sheffield during 2006; and
11 sites surveyed in 2005 in Rotherham.
1.3 Structure of the Report
1.3.1 Following this introductory chapter, the remainder of the Model Development Report is set
out as follows:
Chapter 2 gives an overview of the model covering the study area, zone system, time
periods and vehicle types;
Chapter 3 provides details on the collection of new data and the collation of new and
existing data;
Chapter 4 describes the modelled representation of the Sheffield and Rotherham
highway network;
Chapter 5 outlines the construction of the trip matrices;
Chapter 6 presents details of the model assignment and calibration statistics;
Chapter 7 presents the model validation statistics;
Chapter 8 summarises the main elements of the report and the conclusions drawn.
1.3.2 Supplementary and more detailed information is provided in a series of appendices:
Appendix A details the dates and locations of the Roadside Interview Surveys;
Appendix B details the variable names of the data collected in the Roadside Interview
Surveys;
Appendix C shows SATURN bandwidth plots of vehicle flows in the model;
Appendix D details the counts that make up the validation count cordons and how the
modelled flows match them;
Appendix E shows, both graphically and numerically, the comparison between
observed and modelled journey times;
Appendices F1 to F3 shows the routes assigned by the model for important O-D
movements;
1 Introduction
Sheffield and Rotherham District SATURN Model 2008 1.3
Appendix G summarises the trip matrices before and after matrix estimation at sector
level;
Appendix H compares the distribution of trip lengths before and after matrix
estimation;
Appendix I is an index of the key files used in the model; and
Appendix J is a Technical Note documenting the process used to control matrices to
TEMPRO.
Sheffield and Rotherham District SATURN Model 2008 2.1
2 Model Overview And Dimensions
2.1 Introduction
2.1.1 This chapter provides an overview of the SATURN highway model (SRHM3), presenting the
study area and zone system, the highway users that are represented in the model, and the
time periods modelled.
2.2 The Study Area and Zone System
2.2.1 The SRHM1 zone system is the basis for the SRH3 zone system, and was originally devised
with 500 zones. Ten zones were added in the Manor Top area for use in appraising a new
development in that area, taking the number of zones to 510.
2.2.2 For SRHM2 a further twenty new zones were added enabling us to forecast demand using the
Park and Ride sub-model of the multi-modal model. Of these zones, five were coded into the
network at locations where we would be required to test future year park and ride
interventions, and the remaining zones coded as dummy zones. This took the number of
zones to 530.
2.2.3 The SRTM2 park and ride sub-model has now been merged into the SRDM3 Demand Model,
removing the need for Park and Ride zones in the network. These zones (20 in total) were
removed from the network.
2.2.4 For SRHM3 the zoning system has been revised along the Penistone Rd corridor, in order to
enhance the detail to support a proposed Major Scheme Business Case submission. This
process involved splitting several large zones into smaller zones, resulting in a net addition
of 15 zones.
2.2.5 Following these changes, all components of SRTM3 now consists of 525 zones. The zone
boundaries make use of natural or man-made barriers such as rivers, railways and roads.
The boundaries do not, however, match political boundaries.
2.2.6 The current SRTM3 zone system is shown overleaf;
Figure 2.1 shows the zone system for the whole study area;
Figure 2.2 shows the zone system focussed on Sheffield and Rotherham Urban
Centres;
Figure 2.3 shows the zone system focussed on Penistone Road;
Figure 2.4 shows the zone system in detail for Sheffield City Centre.
2.2.7 The level of detail is sufficient to provide an accurate representation of vehicle flows on the
network and model run times are in the region of 60 minutes using the latest processors.
Sheffield and Rotherham District SATURN Model 2008 2.1
Figure 2.1 Sheffield and Rotherham SATURN Model Zone system
Figure 2.2 Sheffield and Rotherham SATURN Model zone system between Sheffield
and Rotherham urban centres
2 Model Overview And Dimensions
Sheffield and Rotherham District SATURN Model 2008 2.2
Figure 2.3 Sheffield and Rotherham SATURN Model zone system - Penistone Rd
Figure 2.4 Sheffield and Rotherham SATURN Model zone system – Sheffield City
Centre
2 Model Overview And Dimensions
Sheffield and Rotherham District SATURN Model 2008 2.3
2.3 Base Year
2.3.1 SRHM3 has been calibrated to a base year of 2008.
2.4 Time Periods
2.4.1 SRHM3 has retained the three time periods from all previous versions of the model:
weekday morning peak hour : 0800-0900;
weekday average inter-peak hour : 1000-1200 and 1400-1600; and
weekday evening peak hour : 1700-1800.
2.5 User Classes
2.5.1 The model is calibrated at a 3 user class level, as manual classified counts are only available
for cars, LGV and OGV. The final validated model is then segmented into six user class
versions, as required by WebTAG.
2.5.2 The rationale for segmenting the demand in this fashion is that the segments have quite
different values of time and/or vehicle operating costs. The values affect their choice of
routes in the highway model, their response to changes in costs in the demand model, and
also the economic evaluation of time savings in the cost benefits analysis.
Six User Class
2.5.3 The model employs the six segments of highway demand as recommended in the current
DfT guidance on highway modelling (webTAG):
Car – employer’s business;
cars – commute and other - low income;
cars – commute and other - medium income;
cars – commute and other - high income;
light goods vehicles (LGVs); and
other goods vehicles (OGVs).
2.5.4 The full range of validation checks were run on the six user class assignments, to ensure that
the splitting of the car demand into the 3 separate car user classes did not adversely affect
the validation of the model.
2.5.5 SATURN models route choice for cars and goods vehicles. It models buses and trams on
fixed routes, but does not model the number of passengers.
2.5.6 The model uses the following passenger car unit (PCU) factors:
cars and light goods vehicles 1.0;
other goods vehicles 1.7;
standard buses 2.0; and
2 Model Overview And Dimensions
Sheffield and Rotherham District SATURN Model 2008 2.2
trams 6.0
2.5.7 The pcu factor for other goods vehicles was calculated using a sample of five classified
counts at different locations within the Sheffield and Rotherham Districts. The counts had
other goods vehicles separated into two classes:
rigid vehicles over 3.5 tonnes with 2 or 3 axles, with an accepted PCU factor of 1.5;
and
rigid vehicles with 4 or more axles and all articulated vehicles, with an accepted PCU
factor of 2.3.
2.5.8 We took a weighted average of the two PCU factors using the counts for each type over all
movements for all five classified counts. This resulted in an average OGV factor of 1.7.
2.6 Network and Junction Characteristics
2.6.1 The model contains:
2982 simulation nodes;
4682 simulation links;
− 3272 two-way links
− 1410 one-way links
213 buffer links;
571 signalised nodes;
76 roundabouts;
2263 priority junctions; and
661 external nodes.
Sheffield and Rotherham District SATURN Model 2008 3.1
3 Data Collection and Collation
3.1 Introduction
3.1.1 The following data had been collected for previous versions of the model and were included
in the model building and validation:
traffic signal data from the Urban Traffic Control (UTC) teams of both councils;
bus service patterns;
journey time survey data;
roadside interview surveys; and
manual classified vehicle counts.
3.1.2 In addition to previous data, the following data were collected in Sheffield and used
specifically for this version of the model, SRHM3:
manual classified counts along Penistone Rd;
manual classified counts along Ecclesall Rd;
manual classified counts along the Northern Inner Relief Road;
manual classified counts in other key locations around the city centre;
3 roadside interview surveys undertaken in Rotherham;
2 updated journey time survey routes; and
vehicle occupancy survey count data.
3.2 Roadside Interview Surveys
3.2.1 We used RIS data undertaken for the following projects to create nine cordons in Sheffield
and Rotherham Districts:
2008 RIS data from surveys undertaken around the Waverley area;
2007 RIS data from surveys undertaken around Sheffield City Centre, Rotherham
Town Centre and motorway approach roads; and
2006 and 2005 RIS data from interview sites around Sheffield and Rotherham
Districts.
3.2.2 Surveys undertaken prior to 2007 were expanded to 2008 counts to give a consistent
baseline for 2008. The 106 sites constitute an extensive programme of roadside interview
surveys for a medium sized conurbation. The cordons are shown in Figure 3.1 along with
their constituent interview sites. Sector 8 is the rest of the United Kingdom outside of the
cordons pictured overleaf.
3 Data Collection and Collation
Sheffield and Rotherham District SATURN Model 2008 3.2
Figure 3.1 Location of Roadside Interview Surveys
3.2.3 The survey programmes and example coding forms for the surveys are presented in
Appendix A. Each of these surveys involved asking drivers to identify the characteristics of
their current journey including:
time of interview (or receipt of postcard);
type of vehicle;
number of occupants;
origin address;
origin purpose;
destination address; and
destination purpose;
parking location; and
parking tariff (if applicable).
3.2.4 The roadside interviews included manual classified counts in both directions at each site on
the day of the survey and automatic traffic counts in both directions for the week during
which the survey was performed. The manual classified counts were factored in line with the
variation in the traffic volumes over the week of the survey. The ATC’s were carried out at
every location for 1 week, and were used to factor up the MCC’s from the survey day to an
average weekday. We did this as it is expected that the survey reduces the capacity of the
road at the survey site and thus lower flows are often observed on the day of the survey.
3 Data Collection and Collation
Sheffield and Rotherham District SATURN Model 2008 3.3
3.3 Count Data
3.3.1 We had a collection of all counts undertaken in Rotherham and Sheffield for the 2007 version
of the model. In addition to those counts, Sheffield City Council had also conducted more up
to date counts in 2007 and 2008. All counts are from 2005 to 2008, with the vast majority
having been collected between 2006 and 2008. The locations of these counts are shown in
Figure 3.2.
Figure 3.2 Location of Classified Counts
3.3.2 The counts were gathered into the following specific count sets in order to calibrate the
model:
all model counts within both Sheffield and Rotherham Districts;
a cordon around Sheffield City Centre; and
a cordon around Rotherham Town Centre.
3.3.3 Along with the counts sets above, detailed analysis of the count calibration was undertaken
on the following subsets of counts along key corridors within Sheffield:
all counts along Penistone Road (proposed Smart Route);
counts on routes of the proposed Sheffield-Rotherham Bus Rapid Transit schemes; and
counts along Ecclesall Rd, Sheffield.
3 Data Collection and Collation
Sheffield and Rotherham District SATURN Model 2008 3.4
3.4 Independent Count Set
3.4.1 In addition to the calibration counts, an independent count set was retained and not included
in matrix estimation. Figure 3.3 shows the location of the independent counts.
Figure 3.3 Location of Independent Counts
3.5 Journey Time Survey Data
3.5.1 Journey time information was obtained for 17 two-way routes, which are presented in Table
3.1 and Figure 3.4. Information was collected several times for each route in order to be
confident that the times recorded were representative of average traffic conditions.
Table 3.1 Description of Journey Time survey routes
Route Description Year
1 Abbeydale Road 2007
2 Ecclesall Road 2006
3 Sandygate 2007
4 Nether Green 2007
5 Meadowhead 2007
3 Data Collection and Collation
Sheffield and Rotherham District SATURN Model 2008 3.5
6 Sheffield to Rotherham via
Junction 34 north
2006
7 Sheffield to Rotherham via
Junction 34 south
2006
8 Penistone Road / Neepsend
Lane
2007
9 Middlewood Road /
Penistone Rd
2007
10 Chapeltown 2007
11 Crystal Peaks 2006
12 Inner Ring Road 2008
13 Outer Ring Road 2007
14 Sheffield Parkway 2007
15 Mosborough Parkway 2007
16 Crookes 2007
17 Penistone Rd - extended 2008
The routes that are highlighted are key routes that lay on or near major schemes that the
model will be used to test and appraise.
3 Data Collection and Collation
Sheffield and Rotherham District SATURN Model 2008 3.6
Figure 3.4 Journey Time Survey Routes
Sheffield and Rotherham District SATURN Model 2008 4.1
4 Road Network
4.1 Introduction
4.1.1 This chapter summarises the development of the highway definition for the SATURN road
traffic assignment model.
4.2 Network Coverage
4.2.1 Road traffic assignment models require a computerised representation of the highway
network with the following structure:
nodes – the points where roads intersect;
links – the sections of highway between the nodes; and
bus and tram flows on fixed routes.
4.2.2 SATURN offers two levels of network detail:
simulation network, in which capacity restraint is based on gap acceptance (which
represents the extent to which vehicles on minor arms of priority junctions give way to
vehicles on the major arms) applied to the interaction between movements at
junctions; and
buffer network, in which capacity restraint is based on flow delay curves, where
increased flows on a particular link result in increased travel times along that link.
4.2.3 The basis of the SRTM2 network was, broadly speaking, Motorways, A roads, B roads and C
roads plus all bus routes within Sheffield and Rotherham.
4.2.4 This basic structure was amended to add detail in the following areas;
Amendments to incorporate the recently opened Northern Inner Relief Road (NIRR);
Changes to zone connectors within Sheffield City Centre, to ensure a closer match with
the City Centre ‘AIMSUN’ model;
Network detail was added along Penistone Rd to accompany the re-zoning which
occurred in this area; and
Improvements were made to several roundabouts in the vicinity of Rotherham Town
centre.
4.2.5 A skeletal buffer network was added outside of the study area (Sheffield and Rotherham
District) to provide realistic routes for longer distance trips entering the study area. The
extent of the buffer and simulation network is shown in Figure 4.1.
4.3 Data Required for SATURN
4.3.1 The information required for simulation coding encompasses the following attributes:
4 Road Network
Sheffield and Rotherham District SATURN Model 2008 4.2
link length;
link speed;
permitted movements;
saturation flows for each movement;
priorities for each movement;
lane usage and sharing;
flare lengths and stacking capacity;
signal staging; and
signal timings.
4.3.2 The information required for the buffer network coding requires the following attributes:
link length;
speed at capacity;
speed at free flow;
flow at capacity; and
a measure of the steepness of the flow delay curve.
Figure 4.1 Extent of Buffer and Simulation Networks
4 Road Network
Sheffield and Rotherham District SATURN Model 2008 4.3
4.4 Range and Logic Checks
SATNET Errors and Warnings
4.4.1 The SATURN network-building program SATNET produces error and warning messages which
were dealt with individually. All network errors that resulted in fatal errors and semi-fatal
errors were corrected and all the warning messages were inspected and the network was
corrected where necessary.
Bandwidth Plots
4.4.2 Bandwidth plots were used to check that the largest modelled flows occur on the expected
links and network changes were made as necessary. Appendix C presents the bandwidth
plots for total traffic in units of pcu’s.
Queues and Delays
4.4.3 Queues and delays predicted by the model were checked for plausibility in terms of size and
location, and network changes were made where necessary.
Crow-fly Distances
4.4.4 The crow-fly distance between the nodes specifying each link were compared to the coded
link length and any discrepancies checked using the GIS system and corrected where
necessary.
Sheffield and Rotherham District SATURN Model 2008 5.1
5 Road Traffic Demand Matrices
5.1 Introduction
5.1.1 This chapter describes the production of the 2008 Sheffield and Rotherham base year
demand matrices which represent the origins and destinations of the trips in the model.
5.1.2 The production of the base year demand matrices was one of the major tasks in the study.
They were built using data from the 2005, 2006, 2007 and 2008 Roadside Interview Survey
Programmes, which collected travel patterns at 106 locations across Sheffield and
Rotherham.
5.1.3 The study area in this context means Sectors 1 to 7 and Sector 9, which lie within Sheffield
and Rotherham District. Sector 8 is the external area, consisting of some zones on the
periphery of Sheffield and Rotherham Districts and all the zones that cover the remainder of
the UK. Figure 3.1 in Chapter 3 shows the study area.
5.2 Summary of the Matrix Building Process
5.2.1 This section describes the process of building the highway demand matrices. In brief, these
tasks are:
Task 1 - the RIS data is processed to create car, lgv and ogv matrices for fully
observed movements:
Task 2 – the non fully observed cells are in-filled for car using a gravity model:
Task 3 - LGV and OGV non-fully observed cells within the study area are in-filled using
data from SRHM1:
Task 4 - All trips passing through the study area are taken from SRHM1 for all three
user classes.
Task 5 - Finally, these separate components are factored up to the base year of 2008
(if required) and combined, forming the prior matrices.
5.2.2 Figure 5.1 presents the matrix building process in the form of a flow-chart. It uses
parallelograms to represent data and boxes to represent tasks and processes, which are
numbered in the figure and the text. The tasks and sub-tasks are set out as follows:
Task 1 – create matrices from RIS data;
− Task 1a - synthesise missing data;
− Task 1b - expand to counts;
− Task 1c - select fully observed data;
Task 2 – create infill intra sector (study area sectors only) car matrices using gravity
model;
− Task 2a – create trip ends for zones within study area;
− Task 2b – adjust trip ends to match required purpose splits;
− Task 2c – calibrate gravity model;
5 Road Traffic Demand Matrices
Sheffield and Rotherham District SATURN Model 2008 5.2
− Task 2d – create 2008 car infill matrices;
Task 3 – create infill intra sector (study area sectors only) lgv and ogv matrices;
Task 4 – create infill external to external matrices for car, lgv and ogv;
Task 5 – factor up matrices created in Task 3 and 4;
Task 6 – combine matrices;
Task 7 – assign and calibrate.
Task 1 Task 3Task 2
Task 6
Task 5
Task 7
4Create Infill 2008Car, LGV and OGV
Matrices – ExternalTrips
1aSynthesise
Missing Data
5Add Appropriate
Growth
1cSelect Fully
Observed Data
1bExpand
to Counts
6Combine Matrices
2dCreate 2008
Car Infill Matrices
2cCalibrate Gravity Model
2aCreate Trip Ends for
Study Area
2bAdjust trip ends to
match required purpose splits
3Create Infill 2008
LGV and OGVMatrices – Sectors 1
to 7 and Sector 9
7Assign andCalibrate
Task 4
Figure 5.1 Matrix Building Flowchart
5.3 Expand RIS
5.3.1 We expanded the RIS data using 14 different journey purposes – twelve for cars and one
each for LGVs and OGVs. These journey purposes were then aggregated to the six user-
classes used in the model. The fourteen journey purposes for car trips were:
5 Road Traffic Demand Matrices
Sheffield and Rotherham District SATURN Model 2008 5.3
1 home to work;
2 work to home;
3 home to shopping;
4 shopping to home;
5 home to education;
6 education to home;
7 home to employer’s business;
8 employer’s business to home;
9 home to other;
10 other to home;
11 non-home based employer’s business;
12 non-home based other;
13 LGV; and
14 OGV
Synthesise Missing Data
5.3.2 There are three main reasons why we have to synthesise missing data in the roadside
interview surveys:
the surveys are only conducted in one direction at each site because they are very
expensive and at some sites it would be logistically impossible to survey in both
directions;
small roads with limited traffic flow are not surveyed because the extra expense
cannot be justified in terms of the extra data obtained; and
the Police occasionally close sites for a period of time because of safety or to avoid
excessive queues forming on strategic routes.
5.4 Reverse Synthesis
5.4.1 Reverse synthesis is a process allowing RIS data in the survey direction to be used to
estimate the journeys in the reverse direction. The process can be summarised as:
transpose origin and destination zone;
reverse journey purpose;
factor to observed purpose splits by hour; and
factor to observed counts.
5.4.2 The basis for the method is that:
Over the 12 hour survey period, traffic passing the site in the surveyed direction will
be a mirror image of the traffic in the reverse direction (ie it is assumed that issues
5 Road Traffic Demand Matrices
Sheffield and Rotherham District SATURN Model 2008 5.4
such as route variation by direction, single direction journeys, and travel outside the
12 hour period are relatively unimportant);
Information from the forward direction at the site provides the distribution and trip
purpose proportions for the transposed data for the 12 hour period taken as a whole;
and
Information from the RIS observed forward purpose profiles is used to proportion
transposed survey records by hourly period.
Transposition of Origin and Destination and Reversal of Journey Purpose
5.4.3 The first step is simply swapping the origin and destination zones and then reversing the
journey purpose. For the car home-based purposes the journey purpose is reversed (ie
“from home” becomes “to home”), while the non home based and lgv and ogv purposes are
maintained with the same purpose.
5.4.4 We cannot simply transpose morning peak forward direction trips to form evening peak
reverse direction trips because the mix of journey purposes is very different in the two. For
example the inter peak contains lots of shopping trips and the evening peak contains very
few. Therefore we match the journey purpose split in each modelled hour to a target split.
Factoring to Observed Purpose Splits by Hour
5.4.5 The proportion of each purpose’s trips within each hour is calculated from the forward data.
For example 38% of all work to home trips may occur in the hour 1700-1800. These profiles
are built for all purposes and for all hours.
5.4.6 The 12 hour transposed matrix is then factored to the modelled hour by reference to the
profile for that purpose. Thus if there are a total of 100 trips from work to home in a 12
hour period, assuming the profile mentioned above would give 38 trips in the pm peak hour.
This is applied by hour and purpose at all sites. The percentages used are shown in Table
5.2.
Table 5.1 Transpose 12 hour to Modelled Hour Percentages – by Purpose
Purpose AM IP PM
Home to Work 58% 6% 3%
Work to Home 2% 8% 38%
Home to Educ 3% 1% 1%
Educ to Home 0% 1% 1%
Home to Shop 3% 9% 5%
Shop to Home 0% 7% 4%
Home to EB 7% 2% 1%
5 Road Traffic Demand Matrices
Sheffield and Rotherham District SATURN Model 2008 5.5
Purpose AM IP PM
EB to Home 0% 3% 5%
Home to Other 9% 13% 13%
Other to Home 2% 10% 11%
NHBEB 7% 18% 5%
NHBO 9% 21% 13%
Total 100% 100% 100%
5.4.7 This process also has the effect of smoothing the reverse matrix as all 12 hour survey
records for each purpose are used for each period.
Factoring to Observed Reverse Counts
5.4.8 The transposed and factored records were then summed by site, hour, and vehicle type and
factors calculated to get from these totals to the observed RIS non-survey direction counts
for each hour. These factors were created by site, vehicle type, and hour, and then applied
to the reverse matrix.
5.4.9 This gives the reverse matrix with:
Origin zone – original destination zone;
Destination zone – original origin zone;
Purpose – reverse of original purpose;
Period – from period profiles; and
Expansion factor – from 12 hour forward totals factored to modelled hour, factored to
observed counts by site and vehicle type.
5.4.10 Once the transposed matrix is created it is treated the same as the directly observed RIS
matrix and the two are simply added together.
Small Unsurveyed Roads
5.4.11 We have not surveyed every road, as some roads are lightly trafficked and it was not
deemed cost effective to survey these locations. It is important, however, to accurately
estimate the overall level of trip making.
5.4.12 Often the counts for smaller roads are simply added to the count of a nearby road which has
been surveyed. Whilst the counts of vehicles crossing a screenline will be correct, it is
unlikely that the origins and destinations will be truly representative. Vehicles on smaller
roads generally are making shorter trips.
5.4.13 We believe this can be improved upon by making sure that we don’t factor up trips that
would be very unlikely to use the unsurveyed road. We therefore took a sensible view on the
5 Road Traffic Demand Matrices
Sheffield and Rotherham District SATURN Model 2008 5.6
likely range of vehicles using the unsurveyed roads and limited our synthesised trips to those
areas.
Temporary Site Closures
5.4.14 We expanded all the records in each time period to match the modelled hour count so there
was no need to synthesise data for missing individual half hour periods. There was only one
instance where a site was closed for a prolonged period of time leading to a sample size
which was unusable as a set of survey records. Site 616 on Sheffield Parkway was closed
outbound between 5pm and 7pm. We synthesised the records between 4pm and 5pm to
cover the remainder of the time period, and factored up to match the required traffic counts.
Similarly, Site 410 on Penistone Rd was closed between 5pm and 7pm and was synthesised
accordingly.
Expand to Counts
5.4.15 We expanded all the RIS data in a period to match the modelled hour counts for cars, LGVs
and OGVs. The classified counts were performed on the day of the survey and then factored
up using a one week ATC at the survey location. We did this as it is expected that the
survey reduces the capacity of the road at the survey site and thus lower flows are often
observed on the day of the survey.
Select Fully Observed Data
5.4.16 Figure 5.2 presents a schematic sector system to show examples of trips made within the
modelled area. The black lines represent the cordons that are made up of the surveys. Each
time a trip crosses a cordon it is said to be observed. Trips were classified as being:
observed once (shown in green);
unobserved or partially observed (in red); and
observed multiple times (in blue).
5.4.17 We excluded any data which was only partially observed and retained and fully observed
data.
5.4.18 Figure 5.3 presents the sector system applied to the zones in the model.
5 Road Traffic Demand Matrices
Sheffield and Rotherham District SATURN Model 2008 5.7
Figure 5.2 Schematic showing ‘observed’, ‘unobserved’ and ‘partially observed’
movements
Figure 5.3 Sector system used for matrix building
5 Road Traffic Demand Matrices
Sheffield and Rotherham District SATURN Model 2008 5.8
Multiple Observations
5.4.19 For trips that were observed multiple times, we needed to factor them so that they were only
included once in the trip matrix. Figure 5.4 shows that this is not straightforward as a trip
between sectors 2 and 5 (both within Sheffield District), for example, could be observed two
or three times between its origin and destination. Only using a process called barrier
factoring could we estimate the proportion of trips that use each of the three routes,
however this process is not appropriate where there are motorway based cordons. Since we
have several cordons, we decided not to use this method; we decided to use a more robust
method of dealing with multiple observations, as detailed below.
5.4.20 We know that all three routes are common in the fact that they all cross the cordons
surrounding sectors 2 and 5 (the three different route variants are shown in red, blue and
green respectively). Therefore, for such trips, we only included observations crossing both
the origin and destination sector cordons. As every trip between these sectors is observed
twice by these cordons, we simply halved all these observations.
5.4.21 Different routes taken for trips within sector 7 have also been shown in Figure 5.4 below,
giving an example of multiple observed trips for both Sheffield and Rotherham.
Figure 5.4 Various Observations of a particular O-D trip
5 Road Traffic Demand Matrices
Sheffield and Rotherham District SATURN Model 2008 5.9
5.5 Calibrate gravity model and estimate unobserved car trips
5.5.1 Roadside Interview Surveys cannot feasibly capture all the movements made by travellers in
an area the size of Sheffield and Rotherham. Some movements remain wholly or partially
unobserved. These movements were estimated using a gravity model. A gravity model is
one of two methods suggested in DMRB for in-filling trip matrices the alternative being to in-
fill using data from another model.
5.5.2 We ran a separate gravity model for each journey purpose and time period combination. The
first task was to calibrate the gravity model to reproduce the demand in the fully observed
cells of the trip matrices. The calibration produced the parameters that determine the shape
of the deterrence function applied in the model – which helps to determine the trip length
distribution in the output. The parameters were calibrated using functions built into the
gravity model software. The trip-length distributions in the observed data were checked for
plausibility and the trip-length distributions estimated by the gravity model were checked for
their fit to the observed data.
5.5.3 Once the model parameters have been calibrated, gravity models can be applied in two
different modes. In trip-end mode, the user estimates the demand generated in each zone
using independent data on population, trip-rates and land-use. In partial matrix mode, the
gravity model attempts to infer the trip-ends from the same data used to calibrate the model
parameters.
5.5.4 We tried both modes and opted for the trip-end mode because it produced trip matrices that
provided a much better fit to the observed flows of traffic.
5.5.5 Trip-length distributions for the observed and estimated data were checked carefully. They
were checked for the matrices as a whole and for sub-areas – fully observed cells, non-
observed cells.
5.5.6 We estimated trip-ends for the gravity model from zonal population data and trip-rates. The
zonal population data came from the 2001 Census, having been adjusted to match the latest
mid-year population estimates (2007). Table 5.2 overleaf shows the trip-rate values used,
which were derived from TEMPRO by dividing the trip-ends by the population. The process
for undertaking the calculation of the trip-ends is described in Appendix J.
5 Road Traffic Demand Matrices
Sheffield and Rotherham District SATURN Model 2008 5.10
Table 5.2 Trip-rates used in the Gravity Model
From-home To-home Purpose Trip Rate
Denominator AM IP PM AM IP PM
HBW Workers 0.139 0.008 0.007 0.003 0.017 0.124
HBEB Workers 0.012 0.003 0.001 0.000 0.004 0.010
HBED Population *
households with
car / Total
households 0.015 0.003 0.003 0.002 0.005 0.005
HBShop Population *
households with
car / Total
households 0.003 0.011 0.006 0.000 0.011 0.008
HBOther Population *
households with
car / Total
households 0.012 0.016 0.015 0.003 0.017 0.124
NHBEB Jobs 0.004 0.008 0.003 N/A N/A N/A
NHBO Population 0.005 0.012 0.011 N/A N/A N/A
5.5.7 Following the first pass of the gravity model, we improved the fit of the output matrix to the
fully observed data using K-Factors. Trips between the 9 sectors, identified Figure 5.3, were
fully observed in the surveys. The function of the gravity model is to estimate the trips
within each sector. By controlling the gravity model to reproduce the observed totals
between sectors we can ensure that the estimate of the trips within each sector accounts
exactly for the shortfall between the fully observed trips and the total trips. The K-factors
are the method by which we control the gravity model to match the fully observed demand.
K-factors are calculated at sector-level, one is calculated for origin-sector to destination-
sector pairing. The same K-factor is applied to every zonal origin-destination pair that
correspond to the sector level origin-destination pair.
5.5.8 The fully observed cells were eventually overwritten with the demand from the roadside
interview surveys. Thus the estimated demand from the gravity model was used only to in-
fill intra-sector movements.
5.5.9 The gravity model was used to infill only the cells within the study area. Trips with one end
outside the area and one end inside the area were fully observed in the surveys. Trips with
both ends outside the study area were taken from SRHM1.
5 Road Traffic Demand Matrices
Sheffield and Rotherham District SATURN Model 2008 5.11
5.5.10 For the most part the calibration worked well, but for employer’s business there was a
mismatch between the observed demand and the demand estimated from the trip-ends. The
observed demand exceeded the trip-ends totals. For employers business, we retained the
fully observed demand in the inter-sector cells and used the gravity model’s initial estimate
of demand for the intra-sector cells.
5.5.11 Table 5.3 below shows the split between fully observed and estimated trip ends, over a 12
hour period. There are overall about the same number of fully observed trips as there are
estimated trips. Within the individual journey purposes, the percentage of trips that are fully
observed ranges from 45% (Work to Home) to 63% (Non Home Based Employers Business),
with the following exceptions:
Education trips have the shortest trip length distribution of all the trips, as many will
be very short journeys between home and school. Hence it is unsurprising that many
of these trips are intra-sector trips, and fully observed trips make up only 30% of the
total trips; and
Only 30% of OGV trips are fully observed. This is because a large proportion of the
trips in the matrix are through trips along the M1 / M18, which are estimated from the
previous SRHM1 highway model.
5 Road Traffic Demand Matrices
Sheffield and Rotherham District SATURN Model 2008 5.12
Table 5.3 Estimated Trip-end Totals and Fully Observed Demand
Journey Purpose Estimated
Trip-ends
Fully
Observed
% Fully
Observed
Home to Work 86,624 80,004 48%
Work to Home 89,335 73,076 45%
Home to Employers Business 9,845 11,289 53%
Employers Business to Home 14,319 15,551 52%
Home to Education 31,513 14,208 31%
Education to Home 18,466 8,440 31%
Home to Shopping 19,801 30,127 60%
Shopping to Home 20,260 30,290 60%
Home to Other 49,951 56,265 53%
Other to Home 50,878 50,318 50%
Non Home Based Employers Business 32,624 55,205 63%
Non Home Based Other 53,169 77,657 59%
LGV 74,794 73,298 49%
OGV 102,115 44,718 30%
Total 653,693 620,445 49%
5.6 Estimate LGV and OGV trips within study area sectors
5.6.1 The LGV and OGV intra sector trips for sectors 1 to 7 and sector 9 were taken from SRHM1.
5.7 Estimate external to external trips
5.7.1 The Car, LGV and OGV external to external trips were taken from SRHM1.
5 Road Traffic Demand Matrices
Sheffield and Rotherham District SATURN Model 2008 5.13
5.8 Apply Growth to 2006 Matrices
5.8.1 Our starting point was the matrices from the 2006 calibrated version of the Sheffield and
Rotherham model, SRHM1. These matrices were already segmented into the same user
classes as SRHM3 and had the same zone system at the matrix building stage. Therefore we
simply needed to factor the matrices to represent the expected growth in traffic between
2006 and 2008.
5.8.2 We used growth factors from ATC data provided by Sheffield City Council. The data covered
over 50 routes between 2003 and 2007. We only analysed routes which contained data for
all of the observed years, and worked out an average growth rate based upon this data for
each time period. For 2008 we did not have a full set of comparable data to undertake this
comparison. We therefore extrapolated the 2003 to 2007 data to obtain a projected trend for
growth between 2007 and 2008.
5.8.3 Table 5.4 presents these growth rates.
Table 5.4 ATC Traffic Growth Factors
2004 -
2008
2005 -
2008
2006 -
2008
2007 -
2008 2008
AM 0.986 1.000 1.003 1.003 1.000
IP 1.005 1.009 1.001 0.995 1.000
PM 0.989 0.997 1.005 0.996 1.000
12-hour 1.004 1.007 1.005 0.999 1.000
5.8.4 The following elements were factored from 2006 to 2008;
Unobserved car trips for trips within Sector 8 – from existing Car 2006 matrices.
Unobserved LGV and OGV trips within all sectors – from existing LGV and OGV 2006
matrices,.
5.8.5 Having factored the above matrices up to a 2008 base year, they were combined with the
following components to create 2008 prior matrices to be assigned by SATURN:
Unobserved car trips – from gravity model from Sectors 1 through to 9 (excluding
Sector 8); and
Observed car, LGV and OGV trips – from expanded roadside interview surveys.
5.9 Assignment
5.9.1 The first task in calibrating the model was to eliminate network errors. We highlighted these
using several techniques. We looked for excessive delays at junctions and peculiar routes
between origins and destinations. We compared modelled and observed flows on individual
links and turns and the modelled and observed journey times on key routes.
5 Road Traffic Demand Matrices
Sheffield and Rotherham District SATURN Model 2008 5.14
5.9.2 Once we thought we had eliminated all the network errors we moved on to matrix
estimation. This is a powerful but potentially dangerous technique because it can disguise
weaknesses in the model. It makes changes to the trip matrix to try to get the modelled
flows to match the counts, so it implicitly assumes that all the model error lies in the matrix.
It will therefore introduce compensating errors where there are still errors in the network.
5.9.3 The matrix estimation process was iterative with network deficiencies gradually eliminated
during the calibration. Each time a network error was identified we corrected it, assigned the
original matrix, and then re-ran the matrix estimation process. This ensured that any
amendments which were made to the trip matrix as a result of network errors were not
retained once the network error had been corrected.
5.9.4 Several rounds of matrix estimation were run end on end, with the following key points
included in the approach:
fully observed data was frozen in the first iteration, in order to only estimate upon the
non-fully observed (infilled) cells in the first instance;
another iteration was undertaken where only the Sheffield and Rotherham cordon
counts were used, individually and as a cordon set. From our experience in other
projects, this method works well in helping to validate a cordon; and
some iterations were performed only on cars after we observed LGVs and OGVs had
already reached satisfactory levels of calibration.
5.9.5 The SATURN Matrix Estimation program SATME2 is much better behaved when presented
with a good quality prior matrix. Use of the program with matrices that contain partial
observations in some cells produces poor results and is not recommended. We were
confident that our prior matrix was of good quality since all the cells were either fully
observed in the surveys or derived from the recommended gravity model approach.
5.9.6 We set parameters within SATME2 to limit the changes that the program could make to
matrices. The trip ends were capped at a value 50% greater than in the prior matrix, in
order to prevent distortion of the matrix and the creation of many short distance trips, which
SATURN has a tendency to do in order to match counts.
5.9.7 The following checks were undertaken before and after matrix estimation, to ensure that the
matrices had not been overly distorted, and are reported in Chapter 7;
Trip length distribution plots;
Average trip lengths;
Changes in sector to sector movements;
Correlation between trip ends before and after matrix estimation; and
Correlation at an O-D level before and after matrix estimation.
Sheffield and Rotherham District SATURN Model 2008 6.1
6 Model Assignment and Calibration
6.1 Introduction
6.1.1 The objective of this chapter is to present a summary of the steps undertaken to calibrate
and validate the SATURN model. It covers convergence levels, user classes, generalised cost
formulation, calibration procedures, comparison with counts, comparison with journey time
information and characteristics of the final matrices.
6.1.2 Calibration and validation is primarily undertaken using the 3 user class model (Car, LGV and
OGV), as traffic counts cannot be split into different car user classes.
6.1.3 Following calibration and validation using the 3 user class model, the out-turn car matrices
are segmented into 6 user classes as required by the DfT. This process is described in
Section 6.6.6.
6.2 Assignment Procedure and Convergence
6.2.1 There are three iterative loops in SATURN: one within the assignment process, one within
the simulation process and finally an outer loop of assignment followed by simulation. Each
loop runs until it reaches the stopping criteria. The stopping criteria are defined as either the
maximum number of iterations or as a measure of convergence.
6.2.2 The standard Wardrop Equilibrium, using the Frank-Wolfe algorithm, has been used as the
assignment procedure for this model. We set the maximum number of assignment loops
(NITA) to 30, the maximum number of simulation loops (NITS) to 99 and the maximum
number of outer loops (MASL) to 199.
6.2.3 The outer loop convergence criteria have been set to stop the procedure when 99% of the
links (ISTOP) change their flow or delay by 1% (PCNEAR). We set the number of
consecutive iterations for which these criteria had to be met (NISTOP) to four. This is a very
exacting level of convergence compared to the default values in SATURN of 95% of links with
a change of less than 5%. The level of convergence is very important in economic appraisal
as model noise can swamp the scheme benefits if the convergence is not tight.
6.2.4 SATURN produces a number of convergence statistics for the iterative procedures (the
results for the model are shown in Table 6.1):
Delta is reported from the assignment iterations. It is the difference between the
times along the actual routes and the minimum cost routes, summed across the whole
network and expressed as a percentage of the minimum cost times. DELTA is
expected to be below 1%. Table 6.1 shows that the model convergence is well within
the target range;
Epsilon is reported from the simulation iterations. It is a measure of the degree to
which the area under the speed/flow curves is minimised and is expected to be below
2%. Table 6.1 shows that the model convergence is well within the target range; and
P is the proportion of links on which the flows or delays change by less than PCNEAR
% (which has been set to 1% for this model) between outer loop iterations. The final
6 Model Assignment and Calibration
Sheffield and Rotherham District SATURN Model 2008 6.2
loop has been quoted, although we ensure that the 99% stopping criteria is achieved
on 4 consecutive loops to ensure the network is converged. Note that SATURN rounds
to the nearest integer and that 98.5 is rounded up to match the 99% stopping
criterion.
Table 6.1 Convergence Statistics
Modelled Hour Model Assignment
Delta (%)
Simulation
Epsilon (%)
Outer Loop (P)
(PCNEAR=1%)
Morning Peak 6 user class 0.018 0.024 98.9 (45 loops)
Inter-peak 6 user class 0.020 0.012 99.0 (34 loops)
Evening Peak 6 user class 0.055 0.032 98.8 (47 loops)
6.3 User Classes
6.3.1 The final versions of the model employs six segments of demand, but the matrix estimation
stage was undertaken with only a single segment for cars because counts are not available
separately for the three classes of cars.
6.3.2 The six user class version consists of:
cars – employer’s business;
cars – commute and other - low income;
cars – commute and other - medium income;
cars – commute and other - high income;
light goods vehicles (LGVs); and
other goods vehicles (OGVs).
6.3.3 Flows can be output separately by user class for each version of the model, to be used for
economic and environmental appraisal purposes.
6.3.4 Bus and tram flows, operated on fixed routes, are pre-loaded onto the road network prior to
the assignment of the trip matrices.
6.4 Generalised Cost Formulation
6.4.1 The SATURN assignment procedure builds paths through the network based on a behavioural
generalised cost formulation. This is a linear combination of time and distance with the
following form:
Cost (in pence) = PPM * time (in minutes) + PPK * distance (in km)
6.4.2 The values for PPM, PPK, time and distance are presented in Table 6.2
6 Model Assignment and Calibration
Sheffield and Rotherham District SATURN Model 2008 6.3
6.4.3 The actual values used in the model are in the columns titled PPM and PPK, which is an
equivalent, and possibly more familiar, method of formulating the generalised cost in units of
“in-vehicle” time rather than pence. Units of pence have been chosen because tolls and
parking charges can be directly input in pence, with the result that the numbers in the
SATURN output print files are in the correct units.
6.4.4 The values have been calculated for 2008 using the data and formulae in WebTAG (as
published in June 2004).
6.4.5 The advantage of calculating PPM and PPK using published figures is that they are consistent
with the published values of time for considering tolls or charging regimes as well as the
other models in the South Yorkshire LTP area. The values of time are also consistent with
those in the other modules of SRTM 3 (the PT and demand models).
6.4.6 The following points can be made about the six user class values:
The PPM values vary by time period, as the percentage of Commute / Other trips
within the High / Medium / Low income bands varies from one time period to the next;
The inter-peak values of time are the highest (out of the 3 separate income bands),
followed by the Evening Peak and Morning Peak. This is because, as mentioned earlier,
‘Other’ has a higher PPM value than ‘Commute’. Therefore the time period where the
ratio between ‘other’ and ‘commute’ is at its highest (Inter-peak) will have the highest
PPM values for the different income bands. Accordingly, this ratio is lowest in the
Morning Peak, which has the lowest PPM values.
The higher income users have a higher value of time, hence a higher value for PPM.
Table 6.2 Generalised Cost Parameters – Six User Class
Morning Peak Inter-peak Evening Peak
User Class PPM PPK PPM PPK PPM PPK
Cars Employer’s
Business
48.40 12.20 48.40 12.20 48.40 12.20
Car Other – Low Income 7.88 6.28 9.91 6.28 8.85 6.28
Car Other – Medium
Income
10.73 6.28 12.66 6.28
11.70 6.28
Car Other – High Income 14.25 6.28 15.45 6.28 14.78 6.28
LGV 19.53 13.62 19.53 13.62 19.53 13.62
OGV 16.28 40.17 16.28 40.17 16.28 40.17
6 Model Assignment and Calibration
Sheffield and Rotherham District SATURN Model 2008 6.4
Calibration Procedure
6.4.7 Calibration is the process of adjusting the model to improve the fit to the observed data in
two main areas:
adjustments to the coded links and junctions in the network; and
adjustments to the elements of the matrix that were synthesised.
6.4.8 The calibration process was guided by comparison of link flows with observed flows at
individual survey sites, and through comparing modelled journey times with those observed
in surveys.
6.4.9 Adjustments to network specifications were made in order to improve the fit between
modelled estimates and observed link flows and journey times. Care was taken to ensure
that any changes made improved the situation by making the model a better representation
of conditions on the ground. Unrealistic changes to improve the fit were avoided.
6.4.10 After changes were made to the network, the full cycle of matrix estimation (commencing
with the original matrix) was repeated. This avoided retaining matrix changes that occurred
as a result of errors in the network that were subsequently corrected. Each stage of matrix
estimation concentrated on a specific user class.
6.5 Development of the 6 User Class Models
6.5.1 Matrix estimation used matrices of cars, LGVs and OGVs, the three journey purposes for cars
having been summed to match the level at which counts are available. After calibration we
split the car matrix back into the 14 journey purpose level at which the fully observed
matrices were built and at which the gravity model was run.
6.5.2 This process was run as follows:
Prior car matrices were created for all 12 car journey purpose segments, along with
LGV and OGV, from the roadside interview survey data and the synthetic data.
The journey purpose segments were then combined to create the prior car matrices,
which were run through matrix estimation to produce the final car matrices.
Using a 100 sector system, the difference between the prior and final car matrices was
calculated. These factors were then applied to the original 12 journey purpose matrices
prior to matrix estimation, creating 12 final car matrices, segmented by the
appropriate journey purposes.
6.5.3 This process ensures that we have the same split after matrix estimation as before, and that
this is consistent with the DfT guidelines.
6.5.4 The benefit of using a 100 sector system is that whilst splitting at a zonal level may result in
some zero cell values creating errors in the process, and splitting at the 9 sector level is too
coarse, the 100 sector level provides a good compromise. This sector system is the same
system that was used to smooth the original fully observed highway demand matrices.
6.5.5 These 12 out-turn were required for input into Demand Model (SRDM3).
6 Model Assignment and Calibration
Sheffield and Rotherham District SATURN Model 2008 6.5
6.5.6 The 12 car matrices after matrix estimation, as detailed in 6.6.5, were then put into the
relevant income band – low, medium and high. It should be noted that the “Car - home-
based employer’s business” and “Car - non home-based employer’s business” segments were
not segmented by income, as these are simply combined to produce the final ‘Car –
employer’s business’ matrix.
6.5.7 A procedure was developed to calculate income segmentation for consumer purposes by re-
weighting a sample of National Travel Survey (NTS) data to make it representative of
Sheffield and Rotherham zones
6.5.8 The process works as follows:
For each zone in Sheffield and Rotherham, the NTS sample was re-weighted to the
particular population and economic characteristics of the zone;
For each zone, each trip in the NTS sample was allocated to a demand segment
(purpose, car availability, income group) and distance band, enabling a table to be
produced of the form ‘Zone, Segment, Distance Band, Trips’;
In this way, the shares of trips for the three income categories (low, medium and high
income) were calculated for each zone, purpose, car availability and distance segment.
6.5.9 The final step was to combine the income segmented journey purpose matrices to create low
income, medium income and high income matrices.
Sheffield and Rotherham District SATURN Model 2008 7.6
7 Validation
7.1 Introduction
7.1.1 This chapter presents the results of the 6 user class model validation, with the results
presented under the following headings:
calibration count comparisons;
cordon validation;
important count sets;
independent count validation;
journey time comparison;
inspection of typical O-D routes;
matrix characteristics;
trip length distributions;
trip ends; and
origin – destination pairs.
7.2 Calibration count comparisons
7.2.1 A key tool in presenting how well a model performs is to compare the modelled flows against
observed vehicle flows from traffic counts.
7.2.2 As almost all counts were used in calibration of the model, this section is correctly entitled
calibration count comparisons. The comparison of counts to the model flows was undertaken
at a detailed site level using three measures of model performance against observed data,
which are described in turn within this section:
R-squared (R2);
GEH statistic; and
DMRB guideline;
7.2.3 This section presents the comparisons of modelled flows against observed flows. The
comparisons are grouped into a number of categories to highlight different areas of the
model’s performance:
all calibration counts – a headline figure that provides a summary of the goodness of
fit of the modelled flows to observed counts;
cordon totals – these examine whether the model exhibits any systematic bias; and
independent counts – these are a subset of counts that were excluded from matrix
estimation.
7 Validation
Sheffield and Rotherham District SATURN Model 2008 7.7
R-Squared Statistic
7.2.4 The R2 coefficient is a measure of the goodness-of-fit of a linear regression model of the
form Y = A + BX, relating the modelled flow Y to the observed flow X. The regression line
has intercept A and slope B. The R2 coefficient takes a value between zero and one and it
represents the proportion of the variability in the data that is explained by the linear
regression model.
7.2.5 In the case of a traffic assignment model, we are aiming for a situation where the modelled
flows match the counts directly. The R2 coefficient is therefore only valid as a measure of
the goodness-of-fit of the traffic model if the intercept of the regression line is not
significantly different from zero and the slope of the regression line is not significantly
different from one. Therefore an adjusted version, based on a model of the form Y = X, is
presented, which is always lower than or equal to the standard version of the statistic.
GEH Statistic
7.2.6 The analysis of the modelled and observed flows also makes use of the standard GEH
statistic, which is defined as:
)flow modelledflow observed(5.0)flow modelledflow observed( 2
+×−
=GEH
7.2.7 The reason that the GEH statistic is used is the inability of either the absolute or relative
difference measure to cope over a wide range of flows. The GEH statistic is a measure that
looks at both the difference between count and modelled flows, and at the size of each
observation. Thus, where flows are high a low value of GEH can only be achieved where the
percentage difference between observed and modelled flows are small. However, where
flows are very low even quite sizeable percentage discrepancies are considered acceptable.
7.2.8 Note that all GEH values have been calculated based on vehicle flows, by dividing the OGV
count and flow by 2 to move from pcu’ which the assignment uses, to vehicles.
DMRB Guidelines
7.2.9 The DfT guidelines for the validation of highway models are based on those laid out in the
Design Manual for Roads and Bridges (DMRB) Volume 12, Section 2, Part 1, Chapter 4. In
respect of the count comparisons presented in this section there are two separate sets of
criteria against which the counts and modelled flow comparison should be measured. In
both cases the criteria are expected to be met in 85% of cases. The two sets of criteria are:
GEH Statistic:
− links should have a GEH value of less than 5;
DMRB Vehicle Flow Comparison. A link passes the DMRB guidelines if;
− where observed flow is less than 700 vehicles per hour, the modelled flow
should be within 100 vehicles of the observed flow;
− where observed flow is between 700 and 2700 vehicles per hour, the modelled
flow should be within 15% of observed flow; and
7 Validation
Sheffield and Rotherham District SATURN Model 2008 7.8
− where observed flow is greater than 2700 vehicles per hour, the modelled flow
should be within 400 vehicles of the observed flow.
7.2.10 In Tables 7.1 and 7.2 overleaf, we have reported on:
The R squared value – a measure of the goodness of fit between modelled flows and
observed counts;
GEH statistics – the % of all links with a GEH value less than 5;
DMRB guideline – the percentage of links that pass the ‘DMRB Vehicle Flow
Comparison’, as detailed in section 7.7.9; and
Modelled / Observed – the ratio of total modelled flows to total observed counts across
the whole modelled area.
7 Validation
Sheffield and Rotherham District SATURN Model 2008 7.9
All Calibration Counts
7.2.11 The calibration counts are the ones that were used in matrix estimation. We used a total of
2,200 counts, a mixture of turn counts and link counts.
Table 7.1 Validation against all Calibration Counts prior to Matrix Estimation
(all figures expressed as percentages)
R2 GEH <5 DMRB
Guideline countedmodelled
Morning Peak
Car 0.88 41 57 100
LGV 0.88 87 99 93
OGV 0.89 91 98 123
Total 0.90 41 56 101
Inter-peak
Car 0.86 42 63 91
LGV 0.81 82 96 108
OGV 0.93 87 98 138
Total 0.90 43 61 97
Evening Peak
Car 0.88 41 56 97
LGV 0.89 91 100 93
OGV 0.91 96 98 177
Total 0.91 42 55 98
7.2.12 The car user classes for the 2008 model have been derived solely from the observed data,
using a gravity model to infill the non-fully observed movements. The GEH statistic of
between 41% and 44% links with a GEH less than 5 for all time periods is broadly consistent
with other models of large urban areas that MVA have developed and subsequently used
successfully.
7 Validation
Sheffield and Rotherham District SATURN Model 2008 7.10
7.2.13 It is difficult to obtain accurate gravity model estimates of non-fully observed LGV and OGV
trip patterns, and for these user classes we took the 2006 final matrix (factored to 2008) and
used this to infill the 2008 matrix.
Table 7.2 Validation against all Calibration Counts after Matrix Estimation
(all figures expressed as percentages)
R2 GEH <5 DMRB
countedmodelled
Morning Peak
Car 0.98 82 90 98
LGV 0.95 95 100 96
OGV 0.90 95 99 102
Total 0.98 79 88 98
Inter-peak
Car 0.98 86 94 99
LGV 0.93 95 99 97
OGV 0.92 95 100 104
Total 0.98 82 91 99
Evening Peak
Car 0.98 81 91 98
LGV 0.95 96 100 95
OGV 0.92 99 99 101
Total 0.98 79 89 97
7.2.14 The DMRB guidelines are met for all three time periods – the actual figures are 88%, 91%
and 89% for Morning Peak, Inter-peak and Evening Peak respectively. Whilst the three time
periods have very similar networks, the inter-peak has the lowest demand, followed by the
morning peak and then the evening peak, which has the highest demand.
7 Validation
Sheffield and Rotherham District SATURN Model 2008 7.11
7.2.15 The GEH guidelines are almost met for the three time periods – the actual figures for all
vehicles are 79%, 82% and 79%. The figures look a little better when we consider cars on
their own: 82%, 85% and 81%. LGV and OGV are excellent ranging between 96% and
100%.
Table 7.3 Comparison of modelled flows against observed counts (All vehicles
combined)
Y = ax
(Slope)
DMRB
Pass / Fail
R squared DMRB
Pass /
Fail
Total 0.98 Pass 0.98 Pass
Car 0.99 Pass 0.98 Pass
LGV 0.95 Pass 0.95 Pass
Morning
Peak
OGV 0.90 Pass 0.90 Fail
Total 0.98 Pass 0.98 Pass
Car 0.99 Pass 0.99 Pass
LGV 0.95 Pass 0.95 Pass
Inter-peak
OGV 0.91 Pass 0.91 Fail
Total 0.98 Pass 0.98 Pass
Car 0.98 Pass 0.98 Pass
LGV 0.94 Pass 0.95 Pass
Evening
Peak
OGV 0.80 Pass 0.92 Fail
7.2.16 Table 7.3 above details the count correlation against DMRB guidelines. Looking at total
vehicle flows, all time periods exceed the DMRB criteria for both the slope and R squared
value, apart from OGV.
7.2.17 Given that all of the time periods exceed the DMRB guidelines stated in Table 7.2 and Table
7.3, we are pleased with the model performance. Every attempt is made to perform counts
when network conditions are consistent, such as avoiding school holidays and periods where
roadworks are being undertaken. However it is inevitable that there are inconsistencies
between counts, arising from:
day to day variability in traffic;
7 Validation
Sheffield and Rotherham District SATURN Model 2008 7.12
short term changes in traffic capacity due to weather conditions, accidents and vehicle
breakdowns;
land use changes in the area such as a new factory or a housing estate being built;
and
traffic in the vicinity of older counts not growing at the average rate we have used in
factoring up the count.
7.2.18 Because some counts are inconsistent, calibrating the model to satisfy one set of counts can
lead to comparisons with other counts deteriorating. Therefore, as models become larger
and use more counts, they become harder to calibrate to meet the DMRB guidelines. It is a
paradox that as we use more counts in calibration in an attempt to improve the model, it
becomes harder to match all the counts used, and therefore the validation statistics can look
poorer (despite our belief that the model is in fact a better representation of reality).
Link Count Plots
7.2.19 Figure 7.1 to Figure 7.3 show the validation of all the link counts with respect to the DMRB
guidelines. Green flows indicate acceptable flows, with blue representing a degree of under-
assignment and red representing some over-assignment.
Figure 7.1 Link Flow Validation Plot - Morning Peak
7 Validation
Sheffield and Rotherham District SATURN Model 2008 7.13
Figure 7.2 Link Flow Validation Plot - Inter-peak
Figure 7.3 Link Flow Validation Plot - Evening Peak
7 Validation
Sheffield and Rotherham District SATURN Model 2008 7.14
7.2.20 We can conclude from the figures above that the vast majority of links are coloured green,
meaning that they pass the DMRB guidelines. For those links that fail, there is a balance
between red (flow too high) and blue (flow too low) in all networks, which suggests there is
no systematic bias whereby the model over-estimates or under-estimates flows.
7.2.21 Furthermore, those links that fail are spread across the conurbation, signifying that there is
no specific area where the model validates poorly.
7.2.22 Given the large number of counts that we have in the model, and the limiting criteria that
were applied to each zone during matrix estimation to stop unrealistic growth in trips, we are
satisfied that the model validates acceptably.
7.3 Cordon Validation
7.3.1 The DMRB refers to the GEH statistics, requiring it to be less than 4 for the total flows across
screenlines or cordons for “all (or nearly all) screenlines”.
7.3.2 We have two cordons used for the calibration and validation of the model;
Sheffield City Centre Cordon – 18 sites, in general just outside of the Inner Ring Road;
and
Rotherham Town Centre Cordon – 11 sites, just outside of Rotherham Town Centre.
7.3.3 Figure 7.4 shows the locations of these cordons, together with the other key count sets.
7.3.4 Several counts, particularly on the BRT corridors, are in multiple key count sets.
Figure 7.4 Location of Key Counts
7 Validation
Sheffield and Rotherham District SATURN Model 2008 7.15
7.3.5 Tables 7.4 and 7.5 below show the total volume of vehicles crossing each cordon, by time
period and by direction, before and after matrix estimation. The GEH statistic compares the
total modelled flow with the total observed flow.
Table 7.4 Screenline Flows across Sheffield and Rotherham Cordons – Before
Matrix Estimation
Morning Peak Inter-peak Evening Peak
Screenline Mod Obs Links GEH Mod Obs Links GEH Mod Obs Links GEH
Sheffield City -
Inbound
14,064
14,940 6% 7.3
8,648
9,310 8% 7.0
9,625
10,893 13% 12.5
Sheffield City -
Outbound
8,922
8,955 0% 0.3
8,965
8,963 0% 0.0
13,109
13,482 3% 3.2
Rotherham Town
- Inbound
7,179
6,493 -10% 8.3
5,188
4,784 -8% 5.7
5,659
5,082 -10% 7.9
Rotherham Town
- Outbound
4,931
4,512 -9% 6.1
5,465
4,732 -13% 10.3
7,871
7,068 -10% 9.3
7.3.6 The flows across each screenline are generally satisfactory before matrix estimation, ranging
from 13% under assignment to 13% over-assignment.
7.3.7 The percentage of links across each screenline with individual GEH values less than 5 are, for
the morning peak, inter-peak and evening peak respectively:
Sheffield City – Inbound – 50%, 35% and 35%;
Sheffield City – Outbound – 39%, 33%, 33%;
Rotherham Town Centre – Inbound – 73%, 82%, 73%; and
Rotherham Town Centre – Outbound – 55%, 73%, 36%.
7 Validation
Sheffield and Rotherham District SATURN Model 2008 7.16
Table 7.5 Screenline Flows across Sheffield and Rotherham Cordons – After Matrix
Estimation
Morning Peak Inter-peak Evening Peak
Screenline Mod Obs Links GEH Mod Obs Links GEH Mod Obs Links GEH
Sheffield City -
Inbound
14,064
13,820 -2% 2.1
8,648
8,590 -1% 0.6
9,625
9,826 2% 2.0
Sheffield City -
Outbound
8,922
9,075 2% 1.6
8,965
8,914 -1% 0.5
13,109
13,125 0% 0.1
Rotherham Town
- Inbound
7,179
7,330 2% 1.8
5,188
5,483 6% 4.0
5,659
5,908 4% 3.3
Rotherham Town
- Outbound
4,931
5,072 3% 2.0
5,465
5,486 0% 0.3
7,871
7,658 -3% 2.4
7.3.8 The GEH value for the screenline crossings is less than or equal to 4 for all directions and
time periods.
7.3.9 The percentage of links across each screenline with individual GEH values less than 5 are;
Sheffield City – Inbound – 75%, 80%, 90%;
Sheffield City – Outbound – 72%, 94%, 78%;
Rotherham Town Centre – Inbound – 91%, 91%, 82%;
Rotherham Town Centre – Outbound – 91%, 91%, 91%.
7.3.10 Analysis of the links crossing each cordon where the GEH value is greater than 5 shows that
most of these links are minor roads where the observed flow is low, typically less than 700
vehicles per hour.
7.4 Important Count Sets
7.4.1 Whilst we calibrated the model for all counts in Sheffield and Rotherham, we made special
effort to ensure the model would be fit for forthcoming applications of the model. With this
in mind, attention was paid to the count validation along the following corridors;
Sheffield to Rotherham via Waverley – BRT Southern Route;
Sheffield to Rotherham via Meadowhall – BRT Northern Route;
Penistone Rd;
Ecclesall Rd;
Sheffield City Centre Cordon; and
Rotherham Town Centre cordon.
7 Validation
Sheffield and Rotherham District SATURN Model 2008 7.17
7.4.2 The results of the validation for these count sets are shown in Table 7.6.
Table 7.6 Validation against Important Count Sets after Matrix Estimation
(all figures expressed as percentages)
Morning
Peak
Inter-peak Evening
Peak
DMRB GEH
< 5
DMRB GEH
< 5
DMRB GEH
< 5
Sheffield to Rotherham –
BRT Southern 84% 89% 95% 100% 84% 100%
Rotherham to Sheffield –
BRT Southern 94% 94% 94% 94% 100% 100%
Sheffield to Rotherham –
BRT Northern 100% 100% 91% 91% 82% 82%
Rotherham to Sheffield –
BRT Northern 92% 92% 83% 75% 100% 100%
Penistone Rd - Inbound 94% 94% 88% 88% 75% 81%
Penistone Rd - Outbound 79% 86% 93% 86% 86% 86%
Eccdelsall Rd - Inbound 80% 90% 100% 100% 80% 100%
Eccdelsall Rd - Outbound 89% 89% 100% 100% 100% 100%
Rotherham Town Centre
Cordon - Inbound 91% 91% 91% 91% 82% 82%
Rotherham Town Centre
Cordon - Outbound 91% 91% 91% 100% 91% 100%
Sheffield City Centre
Cordon - Inbound 75% 75% 80% 75% 90% 90%
Sheffield City Centre
Cordon - Outbound 72% 94% 94% 94% 78% 83%
7 Validation
Sheffield and Rotherham District SATURN Model 2008 7.18
7.4.3 Table 7.6 shows that all the routes that pass through areas for which major schemes are
planned validate well;
Rotherham to Sheffield (via Waverley) – BRT Southern Route;
− This route validates well across all time periods and directions;
Rotherham to Sheffield (via Lower Don Valley) – BRT Northern Route;
− This route validates well across all time periods and directions;
Penistone Rd;
− Penistone Rd generally validates well in all time periods, with the AM (Outbound)
and PM (Inbound) narrowly missing the GEH criteria;
Sheffield City Centre;
− As mentioned previously, the links with GEH values greater than 5 are generally
minor roads with low traffic flows. The major arterial routes from / to Sheffield
City Centre generally validate well.
Rotherham Town Centre;
− The Rotherham Town centre cordon validates well, meeting the GEH criteria in
every time period and direction apart from the Evening Peak (Inbound).
7.4.4 Appendix D shows the validation statistics for all individual links that comprise these key
routes and cordons.
7.5 Independent Count Validation
7.5.1 We kept behind an independent count set from the calibration process, in order to
independently verify our matrices.
7.5.2 Figure 3.3 showed the locations of the independent counts set.
7.5.3 Figures 7.5 to 7.7 show the validation against this independent count set. Unsurprisingly,
this validation is not as good as the main count set.
7 Validation
Sheffield and Rotherham District SATURN Model 2008 7.19
Figure 7.5 Independent Count Validation Plot – Morning Peak
Figure 7.6 Independent Count Validation Plot – Inter-peak
7 Validation
Sheffield and Rotherham District SATURN Model 2008 7.20
Figure 7.7 Independent Count Validation Plot – Evening peak
7.5.4 Table 7.7 shows the percentage of links in the independent count set with the following
characteristics, for each time period;
Modelled Flow;
Observed Flow;
Percentage Difference;
R squared value;
R squared pass / or fail (criteria – 0.90)
Percentage of Links with GEH less than 5 ( criteria – greater than 85%); and
Percentage of Links passing DMRB criteria ( criteria – greater than 85%);.
7.5.5 The independent count validation is not as good as the calibration count set across all time
periods. The percentage difference between the modelled and observed flow is acceptable
and there is also a good correlation between the modelled and observed across the time
periods. Some of the counts are in areas towards the periphery of the model, where the
network is sparse and there are fewer zones. The plots on Figures 7.5 to 7.7 above show
that the counts closer to the centre of Sheffield, where we have a more detailed network
representation, validate more satisfactorily than those counts located on the periphery of the
model.
7 Validation
Sheffield and Rotherham District SATURN Model 2008 7.21
Table 7.7 Independent Count Validation Statistics
Screenline Mod Obs Abs
Diff
% Diff % GEH
<5
DMRB Slope R
squared
Pass /
Fail R2
Morning Peak Car
34,325 36,719 2,394 7% 42%
51
1.11 0.87 Fail
LGV
5,531 4,733 (798) -14% 80%
94
0.91 0.87 Fail
OGV
2,575 2,234 (341) -13% 88%
100
0.89 0.95 Pass
Total
42,431 43,686 1,255 3% 40%
51
1.07 0.9 Pass
Inter-peak Car
22,521 26,713 4,192 19% 43%
69
1.18 0.89 Fail
LGV
5,147 5,065 (82) -2% 78%
94
1.11 0.87 Fail
OGV
2,914 2,769 (145) -5% 91%
100
0.94 0.96 Pass
Total
30,582 34,547 3,965 13% 49%
68
1.15 0.93 Pass
Evening Peak Car
39,600 40,544 944 2% 39%
48
1.01 0.90 Pass
LGV
3,918 3,933 15 0% 96%
100
1.02 0.95 Pass
OGV
1,006 1,090 84 8% 97%
100
1.11 0.95 Pass
Total
44,524 45,567 1,043 2% 40%
46
1.02 0.92 Pass
7.6 Journey Time Comparison
7.6.1 Sheffield City Council undertook journey time surveys along 17 two-way routes as presented
in Figure 3.4. Several independent surveys were performed for each route and direction to
provide an average of the journey times. All the journey time routes were fully contained
within the simulation network.
7.6.2 The journey times were also tested against confidence intervals of ±15% (or 1 minute if
greater), which is the standard DMRB guideline.
7.6.3 Table 7.8 presents the results for all journey time routes and time periods, showing the
percentage difference between modelled and observed times and whether they are ‘Fast’,
‘Slow’ or ‘OK’ in relation to the DMRB guidelines.
7.6.4 Table 7.9 summarises the route data, showing overall what percentage of routes within each
time period are ‘Fast’, ‘Slow’ or ‘OK’ in relation to the DMRB guidelines.
7 Validation
Sheffield and Rotherham District SATURN Model 2008 7.22
Table 7.8 Journey Times within 15% (or 1 minute if greater)
Morning Peak Inter-peak Evening Peak
Route Mod Obs %Diff DMRB Mod Obs %Diff DMRB Mod Obs %Diff DMRB
1 In 13:51 15:44 -12% OK 13:01 17:22 -25% FAST 13:18 11:46 13% OK
1 Out 11:33 10:21 12% OK 11:00 15:41 -30% FAST 12:19 13:49 -11% OK
2 In 9:30 11:21 -16% FAST 9:08 10:59 -17% FAST 9:03 9:51 -8% OK
2 Out 6:26 6:09 5% OK 9:15 13:36 -32% FAST 9:18 19:49 -53% FAST
3 In 11:31 10:21 11% OK 8:49 10:23 -15% FAST 10:13 9:26 8% OK
3 Out 7:44 6:49 14% OK 7:22 8:17 -11% OK 9:13 9:57 -7% OK
4 In 11:05 10:03 10% OK 9:00 11:24 -21% FAST 11:10 12:41 -12% OK
4 Out 8:01 7:02 14% OK 7:57 9:23 -15% FAST 10:16 10:27 -2% OK
5 In 16:45 20:00 -16% FAST 15:08 21:33 -30% FAST 16:46 14:01 20% SLOW
5 Out 13:24 12:16 9% OK 12:17 21:07 -42% FAST 17:16 20:02 -14% OK
6 In 17:46 16:56 5% OK 16:24 14:58 10% OK 15:19 14:26 6% OK
6 Out 17:29 15:19 14% OK 14:32 14:30 0% OK 14:32 17:12 -16% FAST
7 In 26:45 24:34 9% OK 26:27 27:41 -4% OK 26:30 23:40 12% OK
7 Out 26:50 23:19 15% SLOW 26:44 25:07 6% OK 27:44 26:45 4% OK
9 In 13:17 14:13 -7% OK 12:37 10:47 17% SLOW 13:44 12:53 7% OK
9 Out 11:14 8:20 35% SLOW 10:47 12:03 -10% OK 13:53 19:49 -30% FAST
10 In 21:56 22:12 -1% OK 20:01 30:36 -35% FAST 21:48 21:27 2% OK
10 Out 17:43 16:08 10% OK 17:37 19:03 -8% OK 21:57 22:00 0% OK
11 In 17:45 16:22 8% OK 15:19 15:55 -4% OK 16:23 14:18 15% OK
11 Out 13:53 12:17 13% OK 13:39 13:22 2% OK 15:40 15:15 3% OK
12 In 17:22 16:19 6% OK 14:28 12:09 19% SLOW 19:49 27:30 -28% FAST
12 Out 16:25 16:51 -3% OK 13:03 14:12 -8% OK 16:10 16:19 -1% OK
13 In 29:38 27:53 6% OK 28:23 34:02 -17% FAST 38:00 43:20 -12% OK
13 Out 30:17 32:23 -6% OK 29:39 31:45 -7% OK 33:33 32:31 3% OK
14 In 11:14 15:07 -26% FAST 8:36 11:32 -25% FAST 9:02 10:37 -15% OK
14 Out 10:10 9:22 9% OK 9:21 9:05 3% OK 10:43 10:49 -1% OK
15 In 13:58 16:41 -16% FAST 10:48 15:14 -29% FAST 12:07 13:42 -11% OK
15 Out 12:38 11:24 11% OK 11:38 16:32 -30% FAST 15:36 13:50 13% OK
16 In 11:03 10:25 6% OK 8:26 9:22 -10% OK 11:53 13:37 -13% OK
16 Out 8:49 6:43 31% SLOW 6:09 7:39 -19% FAST 8:52 10:01 -11% OK
17 In 20:01 21:04 -5% OK 14:31 18:07 -20% FAST 19:55 21:04 -5% OK
17 Out 14:19 14:03 2% OK 12:11 12:54 -6% OK 17:30 19:03 -8% OK
7 Validation
Sheffield and Rotherham District SATURN Model 2008 7.23
Table 7.9 Percentage of Routes passing DfT criteria
Screenline Slow OK Fast
Morning Peak 9 % (3) 78 % (25) 13 % (4)
Inter-peak 6 % (2) 44 % (14) 44 % (14)
Evening Peak 3 % (1) 84 % (27) 13 % (4)
Overall 6 % (6) 69 % (66) 23 % (22)
7.6.5 When judging by DMRB guidelines, 78% of all routes are considered to be acceptable within
the Morning Peak and 84% within the Evening Peak. This is slightly below the DMRB
guidelines which state that 85% of routes should have a modelled journey time within 15%
of the observed journey time.
7.6.6 In the Inter-peak it was discovered that a lot of the journey times crossed into the peak
periods, whereby a route may start at 3.40pm but and finish at 4.30pm, and be classified as
‘Inter-peak’ whereas in reality the delays and journey times were more representative of
peak period traffic conditions. The journey times were altered to reflect the perceived delays
and journey times in the true inter-peak, which was also required because the inter-peak
network speeds would be used within the PT model.
7.6.7 In summary, the problem routes in each time period are as follows;
Morning Peak
− Middlewood (Outbound) – surveyed in 2007, this route is too slow. Shares part
of route with Penistone Rd (Outbound), surveyed in 2008, which validates
satisfactorily.
− Crookes (Outbound) – delays at Glossop Rd / Fulwood Rd not replicated in the
model.
Inter –peak – covered in Section 7.6.6 above.
Evening Peak
− Ecclesall Rd Outbound is too fast. There is a slow moving queue of traffic from
Moore Street roundabout to Hunters Bar Roundabout, which takes 10 minutes to
cover approximately 1 mile. The delay is caused by side roads and ‘friction’
effects on the link, which the more strategic SATURN network does not replicate
effectively. If we did replicate this, traffic would simply re-route around the area
giving un-realistic flows on parallel routes.
− Inner Ring Road Clockwise is too fast. This is again due to slow moving traffic
along a long link creating large delays which are hard for SATURN to model. If
this were modelled correctly, traffic would simply re-route and create problems
elsewhere in the network.
7 Validation
Sheffield and Rotherham District SATURN Model 2008 7.24
− Middlewood (Outbound) – surveyed in 2007, this route is too fast. Shares part of
route with Penistone Rd (Outbound), surveyed in 2008, which validates
satisfactorily.
7.6.8 Given the size of this model and the difficulties that large models experience in obtaining the
GEH statistics given by the DfT (that were designed with smaller models in mind) we can
conclude that the journey time validation is acceptable.
7.6.9 Appendix E presents the journey time summary table in more detail. For each route and time
period combination, the following information is presented for the whole route and by route
section:
DMRB Criteria – OK (pass), Fast (fail) or Slow (fail);
modelled journey time;
observed journey time;
absolute difference between modelled and observed; and
percentage difference between modelled and observed.
7.7 Inspection of Typical O-D Routes
7.7.1 Just as important as the count and journey time validation is the routing within the model.
It is vital that key strategic routes are well represented within the model. Appendices F 1 to
F3 show the results of the routing between given origin and destination zones. The following
routes have been chosen to demonstrate that routings are plausible. All routes are on or
near to major highways and / or land developments for which the model will be used to
appraise.
7.7.2 The routes in each Appendix (F1 – Morning Peak, F2 – Inter peak and F3 – Evening Peak) are
as follows:
Figs 1 and 2 - Sheffield City Centre Zone 17 to Ecclesall Rd (and vice versa);
Figs 3 and 4 - Sheffield City Centre Zone 7 to Ecclesall Rd (and vv);
Figs 5 and 6 - Sheffield City Centre to Hillsborough (and vv);
Figs 7 and 8 – North access to Lower Don Valley (and vv);
Figs 9 and 10 – South access to Lower Don Valley (and vv);
Figs 11 and 12 - Sheffield City Centre to Rotherham (and vv);
Figs 13 and 14 - Through trips across Sheffield City Centre – North to South (and vv);
Figs 15 and 16 - Through trips across Sheffield City Centre – East to West (and vv); &
Figs 17 and 18 - Sheffield City Centre to Waverley.
7.7.3 The Green bandwidths alongside the red lines represent the proportion of traffic assigned to
that particular route; the thicker the green line the larger the proportion of traffic on that
route.
7.7.4 The plots show that the model has assigned traffic to sensible routes for the given origins
and destinations.
7 Validation
Sheffield and Rotherham District SATURN Model 2008 7.25
7.8 Matrix Characteristics
7.8.1 Matrices before and after matrix estimation have been aggregated into 8 sectors to allow
comparison at this aggregate level. The trips matrix summaries are presented in Appendix
G. The larger changes are concentrated around trips within the Sheffield urban area, which
is to be expected as 50% of the data has been derived from synthetic data sources.
Considering this, we believe that the changes in sector-to-sector matrix movements are
reasonable.
7.8.2 Tables 7.10 summarises the hourly 6 user class trip matrices after matrix estimation.
Table 7.10 Summary of Hourly 6 user class Trip Matrices (pcus)
Before Matrix
Estimation
After Matrix
Estimation
Percentage
Difference
Morning Peak
Car – Employers Business 11,634 12,943 11%
Car – Commute and Other - Low
income 18,367 19,722 7%
Car – Commute and Other -
Medium income 31,078 32,481 5%
Car – Commute and Other - High
Income 33,870 33,986 0%
Car Total 94,949 99,132 4%
LGV 12,349 13,078 6%
OGV 11,721 9,981 -15%
Total 119,019 122,196 3%
Inter-peak
Car – Employers Business 13,606 15,319 13%
Car – Commute and Other - Low
income 17,071 19,735 16%
Car – Commute and Other -
Medium income 18,623 21,415 15%
Car – Commute and Other - High
Income 19,967 22,063 10%
Car Total 69,267 78,532 13%
7 Validation
Sheffield and Rotherham District SATURN Model 2008 7.26
Before Matrix
Estimation
After Matrix
Estimation
Percentage
Difference
LGV 13,296 13,035 -2%
OGV 13,801 11,579 -16%
Total 96,368 103,151 7%
Evening Peak
Car – Employers Business 9,182 10,104 10%
Car – Commute and Other - Low
income 24,532 27,015 10%
Car – Commute and Other -
Medium income 34,910 37,500 7%
Car – Commute and Other - High
Income 37,522 38,469 2%
Car Total 106,146 113,028 6%
LGV 10,628 10,964 3%
OGV 7,365 4,672 -37%
Total 124,143 128,730 4%
7.8.3 As previously mentioned, approximately 50% of trips within the matrix were derived from
synthetic data. Given this, and the substantial number of traffic counts used for matrix
estimation which can produce large numbers of short distance trips, the changes in matrix
totals for cars after matrix estimation are reasonable.
7.8.4 The OGV matrices have decreased, particularly in the Evening Peak. This decrease is due to
an excessive number of trips along the M1 passing through the study area, which matrix
estimation has factored down. The changes in OGV flows for trips to / from and within the
study area after matrix estimation is below 10%.
7.9 Trip Length Distribution
7.9.1 The distributions of trip lengths within the model are shown in Figures 7.8 to 7.10 for all
vehicles combined, by time period. The variation in trip length distribution, for Car, LGV and
OGV, are provided in Appendix H.
7.9.2 Table 7.11 below also shows the variation in average trip length distribution for car trips
between the prior and post ME matrices. This includes all trips within the network, including
trips that travel through the study area.
7 Validation
Sheffield and Rotherham District SATURN Model 2008 7.27
Morning Peak Trip Length Distribution - All Vehicles
0
5
10
15
20
25
0 to 2 2 to 4 4 to 6 6 to 8 8 to 10 10 to 12 12 to 14 14 to 16 16 to 18 18 to 20 20 to 22 22 to 24 24 to 26 26 to 28 28 to 30 Above30
Trip Length (km)
Perc
enta
ge o
f Tot
al T
rips
(%)
Before Matrix Estimation After Matrix Estimation
Figure 7.8 Trip Length Distribution – Morning peak
Inter Peak Trip Length Distribution - Total Vehicles
0
5
10
15
20
25
0 to 2 2 to 4 4 to 6 6 to 8 8 to 10 10 to 12 12 to 14 14 to 16 16 to 18 18 to 20 20 to 22 22 to 24 24 to 26 26 to 28 28 to 30 Above30
Trip Length (km)
Perc
enta
ge o
f Tot
al T
rips
(%)
Before Matrix Estimation After Matrix Estimation
Figure 7.9 Trip Length Distribution – Inter-peak
7 Validation
Sheffield and Rotherham District SATURN Model 2008 7.28
Evening Peak Trip Length Distribution - Total Vehicles
0
2
4
6
8
10
12
14
16
18
20
0 to 2 2 to 4 4 to 6 6 to 8 8 to 10 10 to 12 12 to 14 14 to 16 16 to 18 18 to 20 20 to 22 22 to 24 24 to 26 26 to 28 28 to 30 Above30
Trip Length (km)
Perc
enta
ge o
f Tot
al T
rips
(%)
Before Matrix Estimation After Matrix Estimation
Figure 7.10 Trip Length Distribution – Evening-peak
7.9.3 Matrix estimation has a tendency to create many short distance trips in order to match the
traffic counts it is presented with. Given that we applied matrix estimation first to the
synthetic cells within the matrix, which tend to have a shorter trip length than the longer
distance trips that cross the cordons and will hence be fully observed, we would expect the
average trip length to be reduced.
Table 7.11 Average Car Trip Lengths Before and After Matrix Estimation
Average Trip Length
(kilometres)
Before ME After ME
Morning Peak 26.8 25.6
Inter-peak 26.5 22.9
Evening Peak 23.4 23.2
7.9.4 Overall, the average trip length has been reduced in all time periods. This is in part to be
expected, given the reasons listed above. The reduction in trip length, however, is relatively
small and we are confident that matrix estimation has not un-realistically altered the trip
lengths within the model.
7 Validation
Sheffield and Rotherham District SATURN Model 2008 7.29
7.9.5 Table 7.12 below shows the average trip lengths in the prior matrices, segmented into the 14
journey purposes from which the matrices were built and what is required for the demand
model. This data in Table 7.12 is only for trips to, from and within the study area, and hence
the average trip lengths presented in Table 7.11 above will be greater than an aggregate of
the data in Table 7.12
7 Validation
Sheffield and Rotherham District SATURN Model 2008 7.30
Table 7.12 Average Car Trip Lengths Before Matrix Estimation – All User Classes,
to, from and within Study Area
Morning Peak Inter-peak Evening Peak
No of
Trips
Average
Trip
Length
(km)
No of
Trips
Average
Trip
Length
(km)
No of
Trips
Average
Trip
Length
(km)
Home to Work
53,556 14 3,113 14 2,738
13
Work to Home
1,495 15 6,189 13
47,197
14
Home to EB
5,552 33 1,326 33
448
29
EB to Home
229 22 1,638 30 4,298
34
Home to
Education 5,873 9 1,012 9 1,253
7
Education to
Home 937 8 1,833 8 2,106
10
Home to
Shopping 1,368 10 5,589 10 3,019
10
Shopping to
Home 172 8 5,738 10 4,128
11
Home to Other
5,877 14 7,769 14 7,759
13
Other to Home
1,933 14 9,014 13
10,549
13
Non Home
Based EB 3,900 25 7,198 26 2,910
28
Non Home
Based Other 5,271 12 11,562 12
11,484
13
LGV
5,875 25 5,869 22 4,297
25
OGV 3,716 43 4,673 43 1,353 42
7 Validation
Sheffield and Rotherham District SATURN Model 2008 7.31
Total
(excluding
LGV and OGV) 86,163 15 61,980 14 97,888 12
7.10 Trip Ends
7.10.1 The trip ends before and after matrix estimation are presented in this section below. Table
7.13 shows the correlation between prior and post matrix estimation trip ends at Origin and
Destination levels.
Table 7.13 Correlation between Trip end totals before and after matrix estimation
Origin Trip Ends Destination Trip Ends
R Squared Y = ax R Squared Y = ax
Morning Peak 0.907 y = 0.94x 0.898 y =0.96x
Inter-peak 0.921 y = 1.00x 0.913 y =0.98x
Evening Peak 0.887 y = 0.97x 0.891 y = 0.98x
7.10.2 Matrix estimation will changes trip ends in order to match link counts, factoring up (and
down) relevant origin-destination movements. We have looked at trip ends to ensure that
these changes do not radically alter origin or destination totals and, importantly, do not
radically alter the distribution of trips.
7.10.3 Figures 7.11 to 7.16 plot this data for all three time periods, focussing on all user classes
combined. Appendix H contains individual plots showing the changes in trip ends before and
after matrix estimation for all time periods and user classes.
7 Validation
Sheffield and Rotherham District SATURN Model 2008 7.32
Morning Peak Total - Differences between Prior and Post ME Origin Vehicle Trip Ends
y = 0.9426xR2 = 0.9072
0
200
400
600
800
1000
1200
1400
0 200 400 600 800 1000 1200 1400
Pre ME
Post
ME
Figure 7.11 Morning Peak Origin Trip Ends
Morning Peak Total - Differences between Prior and Post ME Destination Vehicle Trip Ends
y = 0.9552xR2 = 0.898
0
200
400
600
800
1000
1200
1400
0 200 400 600 800 1000 1200 1400
Pre ME
Post
ME
Figure 7.12 Morning Peak Destination Trip Ends
7 Validation
Sheffield and Rotherham District SATURN Model 2008 7.33
Inter peak Total - Differences between Prior and Post ME Origin Vehicle Trip Ends
y = 1.0014xR2 = 0.921
0
200
400
600
800
1000
1200
1400
0 200 400 600 800 1000 1200 1400
Pre ME
Post
ME
Figure 7.13 Inter Peak Origin Trip Ends
Inter peak Total - Differences between Prior and Post ME Destination Vehicle Trip Ends
y = 0.9843xR2 = 0.9126
0
200
400
600
800
1000
1200
1400
0 200 400 600 800 1000 1200 1400
Pre ME
Post
ME
Figure 7.14 Inter Peak Destination Trip Ends
7 Validation
Sheffield and Rotherham District SATURN Model 2008 7.34
Evening Peak Total - Differences between Prior and Post ME Origin Vehicle Trip Ends
y = 0.9772xR2 = 0.8872
0
200
400
600
800
1000
1200
1400
0 200 400 600 800 1000 1200 1400
Pre ME
Post
ME
Figure 7.15 Evening Peak Origin Trip Ends
Evening Peak Total - Differences between Prior and Post ME Destination Vehicle Trip Ends
y = 0.9757xR2 = 0.891
0
200
400
600
800
1000
1200
1400
0 200 400 600 800 1000 1200 1400
Pre ME
Post
ME
Figure 7.16 Evening Peak Destination Trip Ends
7 Validation
Sheffield and Rotherham District SATURN Model 2008 7.35
7.10.4 The correlation is very good between the trip ends before and after matrix estimation. For all
time periods, the R2 value, and indication of the goodness of fit between two datasets, is
greater than 0.89. This suggests that whilst matrix estimation does change certain origin-
destination movements, it does not introduce large changes to the trip distribution and the
overall distribution of trips in the prior matrix is accurate.
7.10.5 Constraints where applied at a zonal trip end level to the prior matrix in order to prevent
unrealistic changes at a trip end level between the prior and post ME matrices. Whilst ME
attempts to stick to these constraints, sometimes it can’t.
7.10.6 Table 7.14 below shows, for Origins and Destinations for all time periods and user classes,
the percentage of all zones where the increase in the number of trips before and after matrix
estimation is greater than 50%, the percentage of zones where the decrease in the number
of trips after matrix estimation is greater than 50%, and the percentage of trip ends where
the change has a GEH value less than 5.
Table 7.14 Changes in Trip Ends Before and After Matrix Estimation
% of trip
ends > 50%
growth
% of trip
ends > 50%
decrease
% of trip
ends
changed by
less than
50%
% of trip
ends with
GEH less
than 5
Morning Peak Orig 26% 7% 67% 74%
Dest 16% 12% 72% 71%
Inter-peak Orig 12% 6% 82% 82%
Dest 15% 4% 81% 83%
Evening Peak Orig 16% 12% 82% 70%
Dest 19% 10% 71% 74%
7.10.7 Many of the zones where trips have changed by more than 50% are zones where the
existing demand is very low, and hence a large percentage change does not always equate
to a large absolute change in the number of trips.
7.10.8 The GEH value takes into account instances where there could be a large percentage change
in trips ends, greater than 50%, but a low absolute increase in flow. Overall, nearly three-
quarters of trips ends in the morning and evening peak are modified by matrix estimation
such that the GEH value is less than 5. This ratio increase to 80% for the inter-peak.
7 Validation
Sheffield and Rotherham District SATURN Model 2008 7.36
7.10.9 From the data we can conclude that the trip end constraints have worked well, and that the
number of zones that have experienced large charges in trips between the prior and post ME
matrices have been minimised.
7.11 Origin Destination pairs
7.11.1 The analysis of the changes in trip ends shows a very good correlation between the trip ends
before and after matrix estimation. Taking this analysis one level further, we have looked at
changes between the prior and post ME matrices at an O-D level.
7.11.2 Table 7.15 shows these changes for Car, LGV and OGV.
Table 7.15 Correlation between Origin-Destination pairs before and after matrix
estimation
O-D Trips
R Squared Y = ax
Morning Peak Car 0.6513 0.893
LGV 0.8317 1.009
OGV 0.8191 0.511
Inter-peak Car 0.7635 0.941
LGV 0.7654 1.001
OGV 0.8649 0.670
Evening Peak Car 0.7059 1.058
LGV 0.8309 0.845
OGV 0.8673 0.640
7.11.3 As expected, this correlation (based upon the R squared value) is not as good as for the trip
ends, as in a lot of instances we are dealing with many small numbers in the prior matrix.
The table does show that there is a decent correlation between prior and post ME matrices at
the very detailed O-D level, showing that we are not radically altering the distribution and
magnitude of trips in the prior matrix in order to reach the DMRB criteria.
Sheffield and Rotherham District SATURN Model 2008 8.1
8 Summary and Conclusions
8.1 Summary
8.1.1 This report has detailed the steps undertaken to update the SATURN Model of Sheffield and
Rotherham by rebasing it to 2008 and recalibrating it across the whole study area. The new
SRHM3 model represents the morning peak, inter-peak and evening peak modelled hours.
8.1.2 The overall geographic coverage has been retained from the 2007 versions of the model
(SRHM2), the network representing all motorways, A, B and C and bus routes in the districts
of Sheffield and Rotherham. The model has been slightly refined in some areas to aid
calibration.
8.1.3 Fully observed matrices were created from a series of 106 roadside interview surveys
undertaken between 2005 and 2008:
3 new sites surveyed in March 2008 in the Waverley area;
50 new sites conducted in Autumn 2007 to improve the capture of trips entering
Sheffield City Centre and Rotherham Town Centre, the Meadowhall screenline and the
motorway screenline (SWYMM’s update);
42 sites conducted in Spring 2006 to improve the capture of trips in the Sheffield
District; and
11 sites conducted in 2005 to improve the capture of trips in the Rotherham District.
8.1.4 Non-fully observed movements were obtained from a gravity model for car, and from a
previous version of the model for LGV and OGV.
8.1.5 Traffic signal timings were updated where necessary using information supplied by the Urban
Traffic Control (UTC) teams from Sheffield City Council and Rotherham Borough Council.
8.1.6 The model employs tight convergence criteria for the assignment, specifically ISTOP set to
99 and NISTOP set to 4. This is considerably tighter than what is required by the DfT for
economic appraisal.
8.1.7 The model has met or is very close to the guidelines in DMRB in the following areas:
count comparisons by vehicle type;
count comparisons for important count sets;
journey time comparisons;
link lengths compared to crow fly distances;
junction coding compared against aerial photographs and site visits; and
modelled routes checked by inspection for plausibility.
8.1.8 The model has been developed with 6 user classes including income segmentation for car
users. This is so that the model is sufficiently flexible that it can be used to support an
application to the Transport Innovation Fund (TIF), which requires demand to be income
segmented.
8 Summary and Conclusions
Sheffield and Rotherham District SATURN Model 2008 8.2
8.1.9 A five user class model has also been created should Sheffield City Council require this
functionality. The model validates to the same level as the 6 user class model summarised in
this report, and will primarily be used to provide demand data for current or proposed
AIMSUN models within Sheffield District
8.1.10 The five user classes are;
Car – Employer’s Business;
Car – Commute;
Car – Other;
LGV; and
OGV.
8.2 Conclusions
8.2.1 It is considered that the Sheffield and Rotherham District SATURN model, SRHM3, has been
successfully developed and the models for all time periods are considered fit for purpose,
namely to appraise transport schemes and the effects of new developments within the
districts.
Appendices
Appendix A – Roadside Interview Programme
Table A.1 2005 Rotherham Inner/Outer Cordon
ID Location Date Type Sample
Size (%)
422 B6086 Grange Lane Wed 2 Nov 05 SG/PC 688 (71)
424 Meadowhall Road, Rotherham Tue 1 Nov 05 SG/PC 812 (55)
501 A629 New Wortley Road Tue 18 Oct 05 RIS 1235 (13)
502 Coronation Bridge, Rotherham Wed 19 Oct 05 RIS 1180 (24)
503 A6178 Sheffield Road Wed 19 Oct 05 RIS 1017 (17)
505 A631 Bawtry Road Thu 3 Nov 05 RIS 1121 (9)
506 A630 Doncaster Road Tue 18 Oct 05 RIS 1053 (10)
507 Barber's Avenue Thu 20 Oct 05 SG/PC 1253 (29)
508 Rawmarsh Hill Thu 20 Oct 05 RIS/PC 1047 (14)
511 B6086 Brook Hill Wed 2 Nov 05 RIS 774 (40)
512 Badsley Moor Lane Tue 1 Nov 05 SG/PC 1240 (68)
Table A.2 2006 Sheffield Outer Cordon & Screenlines
ID Location Date Type Sample
Size (%)
401 A6102 Manchester Road Wed 22 Mar 06 SG/PC 1303 (31)
403 A61 Penistone Road, Burncross Thu 23 Mar 06 PC 1507 (20)
404 Bracken Hill, Burncross Thu 23 Mar 06 SG/PC 1324 (47)
405 A6135 Chapeltown Rd, Chapeltown Mon 27 Mar 06 SG/PC 1049 (17)
408 Wincobank Avenue, High Wincobank Mon 27 Mar 06 SG/PC 1358 (46)
409 Shirecliffe Road Thu 30 Mar 06 SG 1212 (24)
410 A61 Penistone Road Thu 30 Mar 06 PC 1559 (13)
411 B6079 Middlewood Rd, Hillsborough Tue 18 Apr 06 RIS 935 (37)
412 B6077 Loxley Road, Malin Bridge Tue 18 Apr 06 SG/PC 1399 (33)
420 A6135 White Lane Tue 28 Mar 06 SG/PC 1292 (32)
421 Jumble Lane, Thorpe Common Tue 28 Mar 06 SG/PC 417 (73)
423 New Droppingwell Road Wed 29 Mar 06 SG/PC 1130 (54)
425 Blackburn Road Wed 29 Mar 06 RIS 1079 (21)
428 B0667 Worksop Road, Netherthorpe Thu 27 Apr 06 SG/PC 1277 (39)
429 A618 Mansfield Rd, Wales Common Fri 28 Apr 06 RIS 1314 (22)
430 B6058 Sheffield Road, Nether Green Tue 2 May 06 RIS 1290 (33)
431 B6053 Rotherham Rd, Windmill Hill Tue 2 May 06 RIS 1334 (20)
432 A6135 Sheffield Road Wed 3 May 06 SG/PC 1218 (34)
433 Ford Lane Wed 3 May 06 SG/PC 1078 (83)
435 B6057 Sheffield Road, Dronfield Mon 8 May 06 RIS 1330 (24)
436 A61 Dronfield By Pass Tue 9 May 06 RIS 1207 (12)
437 B6054 Greenhill Parkway Tue 9 May 06 RIS 1257 (28)
438 A621 Abbeydale Road Wed 10 May 06 RIS 1375 (19)
439 A625 Ecclesall Road, Whirlow Wed 10 May 06 SG/PC 1349 (32)
440 Ringinglow Road, Ringinglow Thu 11 May 06 SG/PC 1155 (58)
441 A57 Manchester Road Thu 11 May 06 SG/PC 1534 (60)
442 Carr Road, Deepcar Wed 22 Mar 06 SG/PC 902 (78)
443 B6088 Manchester Road Tue 21 Mar 06 SG/PC 1130 (35)
444 A616 Valley View Tue 21 Mar 06 RIS 1031 (17)
446 A629 Halifax Road Mon 20 Mar 06 RIS 1000 (23)
448 Tankersley Lane Mon 20 Mar 06 SG/PC 946 (60)
461 Europa Link Wed 19 Apr 06 RIS 1150 (30)
462 Handsworth Road Wed 19 Apr 06 RIS 1177 (16)
463 A6102 Prince of Wales Road Thu 20 Apr 06 RIS 1173 (11)
465 Harborough Avenue, Manor Park Thu 20 Apr 06 SG/PC 1282 (60)
466 A6135 City Road Mon 24 Apr 06 RIS 1274 (22)
467 East Bank Road Mon 24 Apr 06 SG/PC 1464 (45)
468 B6388 Gleadless Road Tue 25 Apr 06 SG/PC 1517 (37)
469 Blackstock Road Tue 25 Apr 06 SG/PC 1484 (48)
470 Hemsworth Road Wed 26 Apr 06 SG/PC 1571 (42)
471 A61 Meadowhead, Greenhill Mon 8 May 06 RIS 1274 (15)
472 Bocking Lane, Greenhill Wed 26 Apr 06 PC 2084 (25)
Table A.3 2007 Sheffield and Rotherham Motorway Corridor
ID Location Date Type Sample
Size (%)
609 A6109 Meadowhall Road Tue 27 Mar 07 PC 1122 (9)
616 A630 Sheffield Parkway Tue 27 Mar 07 PC 1043 (7)
612 A6178 Sheffield Road Wed 28 Mar 07 PC 892 (13)
614 A6102 Shepcote Lane Wed 28 Mar 07 PC 1127 (5)
Table A.4 2007 Surveys – Rotherham Town Centre
ID Location Date Type Sample
Size (%)
621 B6089 Greasbrough Road Tue 6 Nov 07 RIS 1155 (11)
622 A633 Rotherham Rd / Rawmarsh Rd Tue 6 Nov 07 PC 847 (12)
623 A630 Fitzwilliam Road Wed 7 Nov 07 RIS 1198 (19)
624 A630 Doncaster Road Wed 7 Nov 07 PC 1116 (25)
625 A6021 Broom Road Thu 8 Nov 07 RIS 1167 (18)
626 A618 Moorgate Road Thu 8 Nov 07 PC 1058 (24)
627 A630 Centenary Way Mon 12 Nov 07 RIS 1364 (12)
Table A.5 2007 Surveys – Sheffield/Meadowhall Screenline
ID Location Date Type Sample
Size (%)
628 B6082 Ecclesfield Road Mon 12 Nov 07 PC 795 (16)
629 A6102 Upwell Street Wed 21 Nov 07 PC 880 (11)
630 A6135 Barnsley Road Wed 21 Nov 07 PC 869 (16)
Table A.6 2007 Surveys – Sheffield City Centre
ID Location Date Type Sample
Size (%)
631 B6074 Mowbray Street Mon 10 Dec 07 PC 189 (5)
632 C426 Pitsmoor Road Mon 10 Dec 07 PC 160 (6)
633 A6135 Spital Hill Tue 11 Dec 07 PC 111 (4)
634 A6109 Savile Street Tue 11 Dec 07 PC 496 (7)
635 B6073 Furnival Road Wed 12 Dec 07 PC 295 (9)
636 A57 Sheffield Parkway Wed 12 Dec 07 PC 1091 (5)
637 B6072 Broad Street Tue 27 Nov 07 PC 334 (8)
639 B6071 Shrewsbury Road Tue 27 Nov 07 PC 337 (12)
640 B6070 Granville Road Wed 28 Nov 07 PC 433 (10)
642 A61 Queens Road Wed 28 Nov 07 PC 285 (4)
643 UNC Shoreham Street Tue 13 Nov 07 PC 528 (16)
644 A621 Bramall Lane Thu 13 Nov 08 PC 909 (12)
645 B6388 London Road Mon 3 Dec 07 PC 375 (4)
647 A625 Ecclesall Road Mon 3 Dec 07 PC 1449 (12)
648 B6069 Glossop Road Tue 4 Dec 07 PC 332 (4)
649 A57 Western Bank Tue 4 Dec 07 PC 568 (8)
650 C431 Bolsover Street Thu 22 Nov 07 PC 592 (13)
651 UNC Meadow Street Thu 29 Nov 07 PC 219 (13)
652A A61 Penistone Road (Main Road) Thu 13 Dec 07 PC 359 (2)
652B A61 Penistone Road (Subsidiary Rd) Thu 13 Dec 07 PC 72 (11)
653 A61 Netherthorpe Road Thu 29 Nov 07 PC 751 (6)
654 A57 Broad Lane Wed 5 Dec 07 PC 952 (10)
656 UNC Leopold Street Mon 26 Nov 07 PC 159 (13)
657 A621 Arundel Gate Thu 22 Nov 07 PC 307 (4)
658 A61 Sheaf Street Wed 5 Dec 07 PC 339 (3)
659 UNC Upper Allen Street Mon 26 Nov 07 PC 187 (13)
Table A.7 2007 Surveys – Motorway Surveys
ID Location Date Type Sample
Size (%)
713 A618 Pleasley Road Tue 20 Nov 07 PC 1350 (25)
714 A630 Rotherway Mon 19 Nov 07 PC 1449 (12)
715 A631 Bawtry Road Mon 19 Nov 07 PC 696 (11)
716 A6178 Sheffield Road Thu 15 Nov 07 PC 562 (9)
717 A6109 Meadow Bank Road Thu 15 Nov 07 PC 688 (10)
718 A629 Cowley Hill Wed 14 Nov 07 PC 1058 (17)
719 A629 Upper Wortley Road Wed 14 Nov 07 RIS 1372 (21)
720 A616 Roundabout leading to M1 J35A Thu 6 Dec 07 PC 590 (7)
721 A61 Westwood New Road Thu 6 Dec 07 PC 879 (9)
733 A57 Aston Way Tue 20 Nov 07 RIS 774 (10)
Table A.8 2008 RSI Surveys – Waverley Area
ID Location Date Type Sample
Size (%)
801 Treeton Road Tuesday 8th July 2008 PC 1341 (38)
802 Retford Road Thu 6 Dec 07 PC 1452 (23)
803 Chesterfield Road Tue 20 Nov 07 PC 1747 (22)
1.1 Roadside interview survey forms are shown overleaf. The 2007 form is similar to the 2005
and 2006 forms, except there is an extra set of questions regarding (if applicable) asking
respondents which car parks they used and, if a public car park, what charge they paid.
Figure A1 Example Roadside Interview Survey form from MVA Surveys in 2002, 2005 and 2006
Figure A2 Example Roadside Interview Survey form from MVA Surveys in 2007
Appendix B – Roadside Interview Survey Variables
MVA Roadside Interview Surveys – Origin / Destination Purposes
1 Home
2 Usual Workplace
3 Employer’s Business
4 Education
5 Shopping
6 Personal Business
7 Visit Friends
8 Recreation/Leisure
9 Other
Common Vehicle Types
1 Car
2 LGV
3 OGV
MVA Surveys Vehicle Types
1 Car
2 Taxi
3 LGV
4 OGV 1
5 OGV 2
6 Bus/Coach
7 Minibus
8 Motorcycle/Moped
9 Pedal Cycle
10 Other
Notes
OGV 1 is defined as a rigid goods vehicle greater than 3.5 tonnes with 2 or 3 axles.
OGV 2 is defined as a rigid or articulated vehicles with 4 or more axles.
Table B1 Value Labels for All Surveys
Description MVA 2006 MVA 2007
Original Site Number Site Site
New Site Number
Date Date Date
Half hour time period Time Time
Serial_No Serial_No
Interview_No Interview_No
Original Vehicle Type Type Type
VehOther VehOther
Occupancy Occupancy Occupancy
Origin
Origin Postcode Sector
Origin Postcode Opostcode Opostcode
Origin Purpose Opurpose Opurpose
Alternative Origin Purpose Oother Oother
Origin X Coordinate
Origin Y Coordinate
Destination
Destination Postcode Sector
Destination Postcode Dpostcode Dpostcode
Destination Purpose Dpurpose Dpurpose
Alternative Destination Purpose Dother Dother
Destination X Coordinate
Destination Y Coordinate
Parking Location
Origin Check Results_O Results_O
Destination Check Results_D Results_D
Illogical Flag Illogical Illogical
Appendix C – Bandwidth Plots
Figure C1 Morning Peak Model Bandwidth Plot
Figure C2 Inter-Peak Model Bandwidth Plot
Figure C3 Evening Peak Model Bandwidth Plot
Appendix D - Count Comparisons
1 Key
1.1 Table D1 shows the count comparison (total vehicles) for the individual links along each of the key routes, namely BRT North, BRT South,
Penistone Rd, and Eccdesall Rd, and across both the Sheffield City Cordon and the Rotherham Town centre Cordon
1.2 The data is presented for Morning Peak, Inter-peak and Evening Peak
1.3 Figure D1 to D6 show scatter-plots of the observed counts against modelled flows, again focussed on total vehicles.
1.4 They are presented for all three times periods, with one plot for each time period showing the correlation with respect to the DMRB
guidelines, and another plot showing the correlation with respect to the GEH guidelines.
Table D 1 Count Comparison for Key Count Sets
Morning Peak Inter-peak Evening Peak Obs Mod GEH DMRB Obs Mod GEH DMRB Obs Mod GEH DMRB
Penistone Rd - Outbound 853
571
10.6 LOW
889
817
2.5 OK
1,365
1,169
5.5 OK
Penistone Rd - Outbound 569
523
2.0 OK
657
674
0.6 OK
1,129
1,093
1.1 OK
Penistone Rd - Outbound 1,449
1,246
5.5 OK
1,512
1,313
5.3 OK
2,022
1,773
5.7 OK
Penistone Rd - Outbound 901
944
1.4 OK
1,039
1,098
1.8 OK
1,372
1,545
4.5 OK
Penistone Rd - Outbound 996
1,128
4.1 OK
1,182
1,212
0.9 OK
1,570
1,764
4.8 OK
Penistone Rd - Outbound 1,131
1,159
0.8 OK
1,097
1,115
0.5 OK
1,334
1,515
4.8 OK
Penistone Rd - Outbound 1,208
1,389
5.0 HIGH
1,279
1,348
1.9 OK
1,526
1,753
5.6 OK
Penistone Rd - Outbound 1,433
1,452
0.5 OK
1,432
1,449
0.4 OK
1,894
1,853
0.9 OK
Penistone Rd - Outbound 1,707
1,686
0.5 OK
1,518
1,528
0.3 OK
1,996
1,864
3.0 OK
Penistone Rd - Outbound 1,683
1,735
1.3 OK
1,571
1,600
0.7 OK
2,033
1,883
3.4 OK
Penistone Rd - Outbound 1,601
1,636
0.9 OK
1,809
1,827
0.4 OK
2,162
2,039
2.7 OK
Penistone Rd - Outbound 1,431
1,462
0.8 OK
1,466
1,559
2.4 OK
1,929
1,983
1.2 OK
Penistone Rd - Outbound 1,535
1,547
0.3 OK
1,665
1,678
0.3 OK
2,213
2,172
0.9 OK
Penistone Rd - Outbound 1,726
1,780
1.3 OK
1,887
1,896
0.2 OK
2,329
2,325
0.1 OK
Penistone Rd - Outbound 1,762
1,714
1.1 OK
1,749
1,790
1.0 OK
2,194
2,149
1.0 OK
Penistone Rd - Outbound 1,554
1,529
0.6 OK
1,349
1,422
2.0 OK
1,544
1,527
0.4 OK
Penistone Rd - Outbound 1,527
1,529
0.1 OK
1,469
1,422
1.2 OK
1,664
1,527
3.4 OK
Penistone Rd - Outbound 963
937
0.8 OK
973
974
0.0 OK
842
814
1.0 OK
Penistone Rd - Outbound 1,556
1,556
0.0 OK
1,325
1,332
0.2 OK
1,455
1,469
0.4 OK
Penistone Rd - Inbound 1,587
1,602
0.4 OK
1,239
1,278
1.1 OK
1,602
1,650
1.2 OK
Penistone Rd - Inbound 868
938
2.3 OK
892
877
0.5 OK
975
978
0.1 OK
Penistone Rd - Inbound 1,458
1,619
4.1 OK
1,347
1,266
2.2 OK
1,558
1,530
0.7 OK
Penistone Rd - Inbound 1,665
1,619
1.1 OK
1,283
1,266
0.5 OK
1,503
1,530
0.7 OK
Penistone Rd - Inbound 2,290
2,249
0.8 OK
1,659
1,642
0.4 OK
1,772
1,746
0.6 OK
Penistone Rd - Inbound 2,252
2,348
2.0 OK
1,596
1,670
1.8 OK
1,820
1,892
1.7 OK
Penistone Rd - Inbound 1,724
1,848
2.9 OK
1,087
1,143
1.7 OK
1,163
1,173
0.3 OK
Penistone Rd - Inbound 1,846
1,898
1.2 OK
1,230
1,214
0.5 OK
1,367
1,274
2.6 OK
Penistone Rd - Inbound 2,215
2,387
3.6 OK
1,650
1,585
1.6 OK
1,492
1,517
0.7 OK
Penistone Rd - Inbound 2,011
2,112
2.2 OK
1,485
1,413
1.9 OK
1,359
1,387
0.7 OK
Penistone Rd - Inbound 2,037
1,998
0.9 OK
1,362
1,351
0.3 OK
1,551
1,401
3.9 OK
Penistone Rd - Inbound 1,948
2,158
4.6 OK
1,306
1,420
3.1 OK
1,482
1,502
0.5 OK
Penistone Rd - Inbound 1,911
2,104
4.3 OK
1,358
1,383
0.7 OK
1,498
1,592
2.4 OK
Penistone Rd - Inbound 1,628
1,675
1.2 OK
1,131
1,123
0.2 OK
1,257
1,319
1.7 OK
Penistone Rd - Inbound 1,411
1,489
2.0 OK
988
1,001
0.4 OK
1,066
1,109
1.3 OK
Penistone Rd - Inbound 2,123
2,096
0.6 OK
1,281
1,310
0.8 OK
1,438
1,482
1.2 OK
Penistone Rd - Inbound 2,307
1,960
7.5 LOW
1,458
1,236
6.0 LOW
1,628
1,542
2.1 OK
Penistone Rd - Inbound 1,383
1,486
2.7 OK
990
1,024
1.1 OK
1,192
1,288
2.7 OK
Ecclesall Rd - Outbound 990
979
0.3 OK
891
889
0.1 OK
1,044
1,043
0.0 OK
Ecclesall Rd - Outbound 871
979
3.6 OK
786
889
3.6 OK
883
1,043
5.2 HIGH
Ecclesall Rd - Outbound 638
587
2.0 OK
589
510
3.4 OK
646
646
0.0 OK
Ecclesall Rd - Outbound 580
673
3.7 OK
580
552
1.2 OK
722
748
0.9 OK
Ecclesall Rd - Outbound 477
456
1.0 OK
560
429
5.9 LOW
615
622
0.3 OK
Ecclesall Rd - Outbound 810
891
2.8 OK
887
914
0.9 OK
1,180
1,202
0.6 OK
Ecclesall Rd - Outbound 676
751
2.8 OK
727
716
0.4 OK
946
1,003
1.8 OK
Ecclesall Rd - Outbound 733
679
2.0 OK
666
636
1.2 OK
832
876
1.5 OK
Ecclesall Rd - Outbound 996
948
1.5 OK
1,095
1,059
1.1 OK
1,654
1,283
9.7 LOW
Ecclesall Rd - Outbound OK OK OK
685 586 3.9 671 591 3.2 1,015 943 2.3
Ecclesall Rd - Outbound 563
530
1.4 OK
565
553
0.5 OK
890
894
0.1 OK
Ecclesall Rd - Inbound 589
562
1.1 OK
564
529
1.5 OK
628
624
0.1 OK
Ecclesall Rd - Inbound 730
698
1.2 OK
695
654
1.6 OK
734
734
0.0 OK
Ecclesall Rd - Inbound 1,010
899
3.6 OK
1,043
1,008
1.1 OK
951
978
0.9 OK
Ecclesall Rd - Inbound 452
413
1.9 OK
667
656
0.4 OK
783
732
1.8 OK
Ecclesall Rd - Inbound 610
545
2.7 OK
742
747
0.2 OK
745
813
2.4 OK
Ecclesall Rd - Inbound 761
962
6.8 HIGH
850
1,056
6.7 HIGH
824
894
2.4 OK
Ecclesall Rd - Inbound 633
563
2.8 OK
482
370
5.4 LOW
351
411
3.1 OK
Ecclesall Rd - Inbound 811
851
1.4 OK
669
668
0.1 OK
582
622
1.6 OK
Ecclesall Rd - Inbound 661
708
1.8 OK
599
590
0.4 OK
515
542
1.2 OK
Ecclesall Rd - Inbound 1,046
1,171
3.8 OK
874
1,006
4.3 HIGH
795
897
3.5 OK
Ecclesall Rd - Inbound 1,197
1,172
0.7 OK
1,016
1,006
0.3 OK
916
898
0.6 OK
Ecclesall Rd - Inbound 3,127
3,048
1.4 OK
2,432
2,425
0.1 OK
2,281
2,131
3.2 OK
BRT North - Outbound 824
770
1.9 OK
559
531
1.2 OK
657
282
17.3 LOW
BRT North - Outbound 1,176
1,159
0.5 OK
910
884
0.9 OK
1,234
866
11.3 LOW
BRT North - Outbound 690
637
2.0 OK
505
475
1.4 OK
478
440
1.8 OK
BRT North - Outbound 703
760
2.1 OK
528
533
0.2 OK
561
620
2.4 OK
BRT North - Outbound 319
402
4.4 OK
272
300
1.7 OK
220
336
7.0 HIGH
BRT North - Outbound 294
315
1.2 OK
358
353
0.2 OK
392
353
2.0 OK
BRT North - Outbound 514
549
1.5 OK
527
496
1.4 OK
619
613
0.2 OK
BRT North - Outbound 778
749
1.0 OK
682
674
0.3 OK
823
808
0.5 OK
BRT North - Outbound 931
937
0.2 OK
777
838
2.1 OK
928
934
0.2 OK
BRT North - Outbound 488
700
8.7 HIGH
570
725
6.1 HIGH
624
660
1.4 OK
BRT North - Outbound 1,223
1,077
4.3 OK
1,379
973
11.9 LOW
2,003
1,722
6.5 OK
BRT North - Outbound 780
730
1.8 OK
1,250
1,209
1.2 OK
1,978
1,881
2.2 OK
BRT North - Outbound 615
623
0.4 OK
490
543
2.3 OK
618
678
2.4 OK
BRT North - Outbound 540
568
1.2 OK
565
610
1.8 OK
1,196
1,181
0.4 OK
BRT North - Outbound 799
793
0.2 OK
528
516
0.5 OK
822
838
0.5 OK
BRT North - Outbound 181
138
3.4 OK
176
155
1.7 OK
218
153
4.8 OK
BRT North - Inbound 417
163
14.9 LOW
381
172
12.6 LOW
471
150
18.3 LOW
BRT North - Inbound 555
543
0.5 OK
471
472
0.0 OK
568
543
1.0 OK
BRT North - Inbound 1,097
1,105
0.2 OK
486
518
1.4 OK
469
470
0.0 OK
BRT North - Inbound 607
648
1.6 OK
431
478
2.2 OK
560
496
2.8 OK
BRT North - Inbound 1,924
1,928
0.1 OK
1,279
1,246
0.9 OK
1,323
1,271
1.5 OK
BRT North - Inbound 1,853
1,582
6.5 OK
1,336
1,202
3.7 OK
1,492
1,375
3.1 OK
BRT North - Inbound 778
757
0.7 OK
622
667
1.8 OK
693
701
0.3 OK
BRT North - Inbound 993
969
0.8 OK
834
886
1.8 OK
1,074
1,110
1.1 OK
BRT North - Inbound 775
768
0.3 OK
696
805
4.0 HIGH
791
822
1.1 OK
BRT North - Inbound 427
473
2.2 OK
348
385
2.0 OK
223
341
7.0 HIGH
BRT North - Inbound 478
507
1.3 OK
367
390
1.2 OK
242
281
2.4 OK
BRT North - Inbound 385
413
1.4 OK
319
337
1.0 OK
384
403
1.0 OK
BRT North - Inbound 628
679
2.0 OK
483
502
0.8 OK
596
611
0.6 OK
BRT North - Inbound 707
514
7.8 LOW
583
627
1.8 OK
722
727
0.2 OK
BRT South - Outbound 1,773
1,804
0.7 OK
1,448
1,503
1.4 OK
2,491
2,496
0.1 OK
BRT South - Outbound 1,519
1,489
0.8 OK
1,156
1,091
1.9 OK
2,109
1,945
3.6 OK
BRT South - Outbound 1,539
1,469
1.8 OK
1,172
1,123
1.4 OK
2,242
1,997
5.3 OK
BRT South - Outbound 1,108
1,442
9.3 HIGH
1,076
1,090
0.4 OK
1,778
1,546
5.7 OK
BRT South - Outbound 2,213
2,512
6.2 OK
1,899
2,078
4.0 OK
3,195
3,085
2.0 OK
BRT South - Outbound OK OK OK
1,193 1,162 0.9 865 832 1.2 1,101 1,199 2.9
BRT South - Outbound 985
965
0.6 OK
893
860
1.1 OK
1,372
1,357
0.4 OK
BRT South - Outbound 873
877
0.2 OK
428
448
1.0 OK
441
442
0.0 OK
BRT South - Outbound 470
484
0.6 OK
353
376
1.2 OK
425
372
2.6 OK
BRT South - Outbound 446
455
0.4 OK
429
450
1.0 OK
477
440
1.7 OK
BRT South - Inbound 532
544
0.5 OK
424
438
0.7 OK
709
715
0.3 OK
BRT South - Inbound 514
523
0.4 OK
385
393
0.4 OK
596
600
0.2 OK
BRT South - Inbound 484
488
0.2 OK
492
496
0.2 OK
834
859
0.9 OK
BRT South - Inbound 1,203
1,189
0.4 OK
862
875
0.4 OK
996
1,018
0.7 OK
BRT South - Inbound 1,174
1,083
2.7 OK
934
960
0.8 OK
1,087
1,094
0.2 OK
BRT South - Inbound 3,047
2,920
2.3 OK
1,634
1,693
1.4 OK
1,955
1,985
0.7 OK
BRT South - Inbound 1,706
1,240
12.1 LOW
1,086
937
4.7 OK
1,020
952
2.2 OK
BRT South - Inbound 2,120
1,915
4.6 OK
1,523
1,462
1.6 OK
1,962
1,810
3.5 OK
BRT South - Inbound 2,323
2,231
1.9 OK
1,857
1,875
0.4 OK
2,379
2,361
0.4 OK
Rotherham City Cordon - Inbound 615
623
0.4 OK
490
543
2.3 OK
618
678
2.4 OK
Rotherham City Cordon - Inbound 687
702
0.6 OK
528
532
0.2 OK
604
573
1.3 OK
Rotherham City Cordon - Inbound 1,047
1,045
0.1 OK
777
788
0.4 OK
864
838
0.9 OK
Rotherham City Cordon - Inbound 1,144
1,093
1.5 OK
740
706
1.3 OK
907
830
2.6 OK
Rotherham City Cordon - Inbound 574
781
8.0 HIGH
607
773
6.3 HIGH
584
812
8.6 HIGH
Rotherham City Cordon - Inbound 774
722
1.9 OK
547
572
1.0 OK
538
528
0.4 OK
Rotherham City Cordon - Inbound 560
570
0.4 OK
320
314
0.4 OK
295
307
0.7 OK
Rotherham City Cordon - Inbound 261
258
0.2 OK
196
186
0.7 OK
244
226
1.1 OK
Rotherham City Cordon - Inbound 983
975
0.2 OK
479
498
0.9 OK
447
460
0.6 OK
Rotherham City Cordon - Inbound 446
455
0.4 OK
429
450
1.0 OK
477
440
1.7 OK
Rotherham City Cordon - Inbound 89
103
1.5 OK
74
123
4.9 OK
82
214
10.9 HIGH
Rotherham City Cordon - Outbound 607
648
1.6 OK
431
478
2.2 OK
560
496
2.8 OK
Rotherham City Cordon - Outbound 506
523
0.8 OK
527
560
1.4 OK
772
786
0.5 OK
Rotherham City Cordon - Outbound 725
714
0.4 OK
847
824
0.8 OK
1,128
1,163
1.0 OK
Rotherham City Cordon - Outbound 707
722
0.6 OK
823
829
0.2 OK
1,165
1,065
3.0 OK
Rotherham City Cordon - Outbound 461
473
0.5 OK
622
608
0.6 OK
691
685
0.2 OK
Rotherham City Cordon - Outbound 482
586
4.5 HIGH
609
635
1.1 OK
790
712
2.9 OK
Rotherham City Cordon - Outbound 267
293
1.6 OK
412
423
0.6 OK
665
655
0.4 OK
Rotherham City Cordon - Outbound 133
134
0.1 OK
189
174
1.1 OK
270
307
2.2 OK
Rotherham City Cordon - Outbound 414
405
0.4 OK
484
502
0.8 OK
989
1,015
0.8 OK
Rotherham City Cordon - Outbound 532
544
0.5 OK
424
438
0.7 OK
709
715
0.3 OK
Rotherham City Cordon - Outbound 96
29
8.5 OK
98
15
11.0 OK
133
58
7.6 OK
Sheffield City Cordon - Inbound 3,047
2,920
2.3 OK
1,634
1,693
1.4 OK
1,955
1,985
0.7 OK
Sheffield City Cordon - Inbound 370
376
0.3 OK
253
234
1.3 OK
366
342
1.2 OK
Sheffield City Cordon - Inbound 215
264
3.2 OK
208
284
4.8 OK
384
477
4.5 OK
Sheffield City Cordon - Inbound 722
677
1.7 OK
229
236
0.4 OK
217
231
0.9 OK
Sheffield City Cordon - Inbound 175
284
7.2 HIGH
73
230
12.8 HIGH
67
281
16.2 HIGH
Sheffield City Cordon - Inbound 594
575
0.8 OK
152
144
0.7 OK
133
110
2.1 OK
Sheffield City Cordon - Inbound 1,036
1,151
3.5 OK
691
668
0.9 OK
685
690
0.2 OK
Sheffield City Cordon - Inbound 713
651
2.4 OK
543
487
2.5 OK
621
555
2.7 OK
Sheffield City Cordon - Inbound 1,046
1,171
3.8 OK
874
1,006
4.3 HIGH
795
897
3.5 OK
Sheffield City Cordon - Inbound 352
314
2.1 OK
269
319
2.9 OK
241
270
1.8 OK
Sheffield City Cordon - Inbound 972
954
0.6 OK
909
752
5.5 LOW
808
696
4.1 OK
Sheffield City Cordon - Inbound 697
673
0.9 OK
242
249
0.5 OK
254
254
0.0 OK
Sheffield City Cordon - Inbound 307
418
5.9 HIGH
125
106
1.7 OK
145
206
4.6 OK
Sheffield City Cordon - Inbound LOW LOW OK
2,307 1,960 7.5 1,458 1,236 6.0 1,628 1,542 2.1
Sheffield City Cordon - Inbound 764
821
2.0 OK
512
553
1.8 OK
746
637
4.1 OK
Sheffield City Cordon - Inbound 433
431
0.1 OK
196
188
0.6 OK
271
326
3.2 OK
Sheffield City Cordon - Inbound 317
180
8.7 LOW
278
205
4.7 OK
312
325
0.7 OK
Sheffield City Cordon - Outbound 2,213
2,512
6.2 OK
1,899
2,078
4.0 OK
3,195
3,085
2.0 OK
Sheffield City Cordon - Outbound 125
127
0.1 OK
138
152
1.2 OK
160
142
1.5 OK
Sheffield City Cordon - Outbound 275
330
3.2 OK
201
231
2.0 OK
224
378
8.9 HIGH
Sheffield City Cordon - Outbound 173
184
0.9 OK
203
221
1.3 OK
239
300
3.7 OK
Sheffield City Cordon - Outbound 221
144
5.7 OK
257
249
0.5 OK
550
555
0.2 OK
Sheffield City Cordon - Outbound 48
117
7.6 OK
33
51
2.9 OK
26
32
1.2 OK
Sheffield City Cordon - Outbound 119
128
0.8 OK
194
225
2.1 OK
301
354
2.9 OK
Sheffield City Cordon - Outbound 499
515
0.7 OK
792
852
2.1 OK
1,115
1,046
2.1 OK
Sheffield City Cordon - Outbound 430
445
0.7 OK
460
387
3.6 OK
736
693
1.6 OK
Sheffield City Cordon - Outbound 871
979
3.6 OK
786
889
3.6 OK
883
1,043
5.2 HIGH
Sheffield City Cordon - Outbound 353
207
8.7 LOW
341
344
0.2 OK
320
343
1.2 OK
Sheffield City Cordon - Outbound 1,003
1,037
1.1 OK
942
797
4.9 LOW
1,258
1,258
0.0 OK
Sheffield City Cordon - Outbound 305
264
2.4 OK
365
326
2.1 OK
963
884
2.6 OK
Sheffield City Cordon - Outbound 141
141
0.0 OK
195
224
2.0 OK
346
465
5.9 HIGH
Sheffield City Cordon - Outbound 1,449
1,246
5.5 OK
1,512
1,313
5.3 OK
2,022
1,773
5.7 OK
Sheffield City Cordon - Outbound 268
313
2.7 OK
204
186
1.3 OK
222
197
1.7 OK
Sheffield City Cordon - Outbound 213
180
2.4 OK
207
192
1.0 OK
373
350
1.2 OK
Sheffield City Cordon - Outbound 217
206
0.7 OK
238
197
2.7 OK
178
228
3.5 OK
Figure D.1 Morning Peak Count Comparison – Total Vehicles
y = 0.9804xR2 = 0.9814
0
1000
2000
3000
4000
5000
6000
7000
0 1000 2000 3000 4000 5000 6000 7000
Observed
Mod
el F
low
Model Flow DMRB Criteria High Limit DMRB Criteria Low Limit Actual Flow = Count Linear (Model Flow) Linear (Model Flow)
Figure D.2 Inter-peak GEH Count Comparison – Total Vehicles
0
1000
2000
3000
4000
5000
6000
7000
0 1000 2000 3000 4000 5000 6000 7000
Observed
Mod
el F
low
Model Flow DMRB Criteria High Limit DMRB Criteria Low Limit Actual Flow = Count Linear (Model Flow)
Figure D.3 Evening Peak GEH Count Comparison – Total Vehicles
y = 0.9759xR2 = 0.9817
0
1000
2000
3000
4000
5000
6000
7000
0 1000 2000 3000 4000 5000 6000 7000
Observed
Mod
el F
low
Model Flow DMRB Criteria High Limit DMRB Criteria Low Limit Actual Flow = Count Linear (Model Flow)
Figure D.4 Morning Peak GEH Count Comparison – Total Vehicles
0
1000
2000
3000
4000
5000
6000
7000
0 1000 2000 3000 4000 5000 6000 7000
Observed
Mod
el F
low
Model Flow GEH Value = 5 GEH Value = - 5 GEH Value = 0
Figure D.5 Inter Peak GEH Count Comparison – Total Vehicles
y = 0.9829xR2 = 0.9855
0
1000
2000
3000
4000
5000
6000
7000
0 1000 2000 3000 4000 5000 6000 7000
Observed
Mod
el F
low
Model Flow GEH Value = 5 GEH Value = - 5 GEH Value = 0 Linear (Model Flow)
Figure D.6 Evening Peak GEH Count Comparison – Total Vehicles
0
1000
2000
3000
4000
5000
6000
7000
0 1000 2000 3000 4000 5000 6000 7000
Observed
Mod
el F
low
Model Flow GEH Value = 5 GEH Value = - 5 GEH Value = 0
Appendix E – Journey Time Comparisons
Table E 1 Morning Peak Journey Time – By Route Segment
By Route Section
Route Mod Obs Diff %Diff DMRB %
Slow
% Ok % Fast
Abbeydale Rd - Inbound 832 944 -112 -12% OK 27% 18% 55%
Abbeydale Rd - Outbound 694 621 73 12% OK 36% 27% 36%
Ecclesall Rd - Inbound 570 681 -111 -16% FAST 29% 14% 57%
Ecclesall Rd - Outbound 386 369 17 5% OK 43% 43% 14%
Crookes Rd / A57 Western Bank -
Inbound 686 621 65 10% OK 33% 22% 44%
Crookes Rd / A57 Western Bank -
Outbound 463 409 54 13% OK 38% 13% 50%
A57 Fulwood Rd / Glossop Rd -
Inbound 662 604 58 10% OK 44% 11% 44%
A57 Fulwood Rd / Glossop Rd -
Outbound 482 423 59 14% OK 56% 0% 44%
Chesterfield Rd / London Rd -
Inbound 1005 1200 -195 -16% FAST 7% 43% 50%
Chesterfield Rd / London Rd -
Outbound 804 736 68 9% OK 36% 43% 21%
Rotherham to Sheffield via
Meadowhall 1046 1016 30 3% OK 36% 5% 59%
Sheffield to Rotherham via
Meadowhall 1043 919 124 13% OK 35% 25% 40%
Rotherham to Sheffield via Lower
Don Valley 1609 1475 134 9% OK 46% 12% 42%
Sheffield to Rotherham via Lower
Don Valley 1610 1400 210 15% OK 38% 24% 38%
Middlewood Rd / Penistone Rd -
Inbound 795 853 -58 -7% OK 22% 17% 61%
Middlewood Rd / Penistone Rd -
Outbound 674 500 174 35% SLOW 44% 19% 38%
Chapeltown to Sheffield 1316 1333 -17 -1% OK 31% 23% 46%
Sheffield to Chapeltown 1064 969 95 10% OK 47% 7% 47%
Crystal Peaks / Manor Top to
Sheffield 1064 982 82 8% OK 36% 14% 50%
Sheffield to Manor Top / Crystal
Peaks 833 737 96 13% OK 33% 8% 58%
Inner Ring Rd - Clockwise 1042 980 62 6% OK 44% 16% 40%
Inner Ring Rd - Anticlockwise 983 1011 -28 -3% OK 43% 14% 43%
Outer Ring Rd - A6102 - Clockwise 1784 1673 111 7% OK 38% 25% 38%
Outer Ring Rd - A6102 - Anti-
Clockwise 1822 1944 -122 -6% OK 38% 25% 38%
Sheffield Parkway - Inbound 677 907 -230 -25% FAST 50% 25% 25%
Sheffield Parkway - Outbound 612 562 50 9% OK 13% 38% 50%
A57 Mosborough Parkway /
Sheffield Parkway - Inbound 836 1001 -165 -16% FAST 44% 22% 33%
A57 Mosborough Parkway / 759 684 75 11% OK 11% 22% 67%
Sheffield Parkway - Outbound
Walkley to Sheffield 668 625 43 7% OK 80% 0% 20%
Sheffield to Walkley 529 403 126 31% SLOW 50% 0% 50%
Penistone Rd - Inbound 1198 1265 -67 -5% OK 46% 8% 46%
Penistone Rd - Outbound 859 844 15 2% OK 41% 23% 36%
Overall 29407 28691 716 30% 25% 45%
Table E 2 Inter Peak Journey Time – By Route Segment
By Route Section
Route Mod Obs Diff %Diff DMRB %
Slow
% Ok %
Fast
Abbeydale Rd - Inbound 781 1042 -261 -25% FAST 27% 36% 36%
Abbeydale Rd - Outbound 661 941 -280 -30% FAST 55% 27% 18%
Ecclesall Rd - Inbound 549 659 -110 -17% FAST 29% 43% 29%
Ecclesall Rd - Outbound 556 816 -260 -32% FAST 57% 0% 43%
Crookes Rd / A57 Western
Bank - Inbound 529 623 -94 -15% FAST 56% 11% 33%
Crookes Rd / A57 Western
Bank - Outbound 442 497 -55 -11% OK 63% 13% 25%
A57 Fulwood Rd / Glossop Rd
- Inbound 539 684 -145 -21% FAST 56% 0% 44%
A57 Fulwood Rd / Glossop Rd
- Outbound 477 563 -86 -15% FAST 44% 22% 33%
Chesterfield Rd / London Rd -
Inbound 909 1293 -384 -30% FAST 57% 14% 29%
Chesterfield Rd / London Rd -
Outbound 737 1267 -530 -42% FAST 43% 36% 21%
Rotherham to Sheffield via
Meadowhall 981 898 83 9% OK 32% 18% 50%
Sheffield to Rotherham via
Meadowhall 862 870 -8 -1% OK 25% 15% 60%
Rotherham to Sheffield via
Lower Don Valley 1582 1661 -79 -5% OK 38% 19% 42%
Sheffield to Rotherham via
Lower Don Valley 1604 1508 96 6% OK 38% 21% 41%
Middlewood Rd / Penistone Rd
- Inbound 757 647 110 17% SLOW 17% 28% 56%
Middlewood Rd / Penistone Rd
- Outbound 648 723 -75 -10% OK 56% 13% 31%
Chapeltown to Sheffield 1201 1836 -635 -35% FAST 31% 38% 31%
Sheffield to Chapeltown 1057 1143 -86 -8% OK 33% 40% 27%
Crystal Peaks / Manor Top to
Sheffield 920 955 -35 -4% OK 50% 7% 43%
Sheffield to Manor Top /
Crystal Peaks 820 802 18 2% OK 33% 33% 33%
Inner Ring Rd - Clockwise 868 730 138 19% SLOW 24% 24% 52%
Inner Ring Rd - Anticlockwise 783 852 -69 -8% OK 38% 33% 29%
Outer Ring Rd - A6102 -
Clockwise 1692 2043 -351 -17% FAST 42% 17% 42%
Outer Ring Rd - A6102 - Anti-
Clockwise 1752 1905 -153 -8% OK 33% 33% 33%
Sheffield Parkway - Inbound 512 692 -180 -26% FAST 63% 25% 13%
Sheffield Parkway - Outbound 561 546 15 3% OK 25% 38% 38%
A57 Mosborough Parkway /
Sheffield Parkway - Inbound 649 914 -265 -29% FAST 56% 44% 0%
A57 Mosborough Parkway /
Sheffield Parkway - Outbound 698 992 -294 -30% FAST 33% 33% 33%
Walkley to Sheffield 506 562 -56 -10% OK 60% 20% 20%
Sheffield to Walkley 370 459 -89 -19% FAST 75% 25% 0%
Penistone Rd - Inbound 872 1087 -215 -20% FAST 54% 25% 21%
Penistone Rd - Outbound 732 774 -42 -5% OK 36% 27% 36%
Overall 26607 30984 -4377 39% 24% 37%
Table E 3 Evening Peak Journey Time – By Route Segment
By Route
Section
Route Mod Obs Diff %Diff DMRB % Slow % Ok % Fast
Abbeydale Rd - Inbound 798 706 92 13% OK 27% 18% 55%
Abbeydale Rd - Outbound 739 829 -90 -11% OK 36% 27% 36%
Ecclesall Rd - Inbound 543 591 -48 -8% OK 29% 14% 57%
Ecclesall Rd - Outbound 557 1190 -633 -53% FAST 43% 43% 14%
Crookes Rd / A57 Western
Bank - Inbound 613 566 47 8% OK 33% 22% 44%
Crookes Rd / A57 Western
Bank - Outbound 555 597 -42 -7% OK 38% 13% 50%
A57 Fulwood Rd / Glossop Rd
- Inbound 669 761 -92 -12% OK 44% 11% 44%
A57 Fulwood Rd / Glossop Rd
- Outbound 617 627 -10 -2% OK 56% 0% 44%
Chesterfield Rd / London Rd -
Inbound 1006 842 164 19% SLOW 7% 43% 50%
Chesterfield Rd / London Rd -
Outbound 1038 1202 -164 -14% OK 36% 43% 21%
Rotherham to Sheffield via
Meadowhall 917 866 51 6% OK 36% 5% 59%
Sheffield to Rotherham via
Meadowhall 865 1033 -168 -16% FAST 35% 25% 40%
Rotherham to Sheffield via
Lower Don Valley 1966 1420 546 38% SLOW 46% 12% 42%
Sheffield to Rotherham via
Lower Don Valley 1665 1606 59 4% OK 38% 24% 38%
Middlewood Rd / Penistone Rd
- Inbound 826 773 53 7% OK 22% 17% 61%
Middlewood Rd / Penistone Rd
- Outbound 833 1189 -356 -30% FAST 44% 19% 38%
Chapeltown to Sheffield 1308 1288 20 2% OK 31% 23% 46%
Sheffield to Chapeltown 1317 1320 -3 0% OK 47% 7% 47%
Crystal Peaks / Manor Top to
Sheffield 984 858 126 15% OK 36% 14% 50%
Sheffield to Manor Top /
Crystal Peaks 938 915 23 3% OK 33% 8% 58%
Inner Ring Rd - Clockwise 1189 1651 -462 -28% FAST 44% 16% 40%
Inner Ring Rd - Anticlockwise 969 979 -10 -1% OK 43% 14% 43%
Outer Ring Rd - A6102 -
Clockwise 2279 2600 -321 -12% OK 38% 25% 38%
Outer Ring Rd - A6102 - Anti-
Clockwise 2002 1952 50 3% OK 38% 25% 38%
Sheffield Parkway - Inbound 542 637 -95 -15% OK 50% 25% 25%
Sheffield Parkway - Outbound 644 649 -5 -1% OK 13% 38% 50%
A57 Mosborough Parkway /
Sheffield Parkway - Inbound 725 822 -97 -12% OK 44% 22% 33%
A57 Mosborough Parkway /
Sheffield Parkway - Outbound 936 830 106 13% OK 11% 22% 67%
Walkley to Sheffield 713 817 -104 -13% OK 80% 0% 20%
Sheffield to Walkley 532 601 -69 -11% OK 50% 0% 50%
Penistone Rd - Inbound 1196 1265 -69 -5% OK 46% 8% 46%
Penistone Rd - Outbound 1051 1143 -92 -8% OK 41% 23% 36%
Overall 31532 33125 -1593 37% 19% 44%
Appendix F1 – Route Checking – Morning Peak
Fig F1 - 1 Ecclesall Rd to City Centre Zone 17
Fig F1 - 2 City Centre Zone 17 to Ecclesall Rd
Fig F1 - 3 Ecclesall Rd to City Centre Zone 7
Fig F1 - 4 City Centre Zone 7 to Ecclesall Rd
Fig F1 - 5 Hillsborough to City Centre
Fig F1 - 6 City Centre to Hillsborough
Fig F1 - 7 North to Lower Don Valley
Fig F1 - 8 Lower Don Valley to North
Fig F1 - 9 South to Lower Don Valley
Fig F1 - 10 Lower Don Valley to South
Fig F1 - 11 Rotherham to Sheffield
Fig F1 - 12 Sheffield to Rotherham
Fig F1 - 13 Through Trip – North to South
Fig F1 - 14 Through Trip – South to North
Fig F1 - 15 Through Trip – East to West
Fig F1 - 16 Through Trip – West to East
Fig F1 - 17 Sheffield to Waverley
Fig F1 - 18 Waverley to Sheffield
Appendix F – Route Checking – Inter-peak
Fig F2 - 1 City Centre Zone 17 – Ecclesall Rd
Fig F2 - 2 Ecclesall Rd – City Centre Zone 17
Fig F2 - 3 City Centre Zone 7 – Ecclesall Rd
Fig F2 - 4 Ecclesall Rd – City Centre Zone 7
Fig F2 - 5 Hillsborough to City Centre Zone 17
Fig F2 - 6 City centre Zone 17 to Hillsborough
Fig F2 - 7 Lower Don Valley - North
Fig F2 - 8 North to Lower Don Valley
Fig F2 - 9 South to Lower Don Valley
Fig F2 - 10 Lower Don Valley to South
Fig F2 - 11 Sheffield to Meadowhall
Fig F2 - 12 Meadowhall to Sheffield
Fig F2 - 13 Rotherham to Sheffield
Fig F2 - 14 Sheffield to Rotherham
Fig F2 - 15 Through Trip – North to South
Fig F2 - 16 Through Trip – South to North
Fig F2 - 17 Through Trip – East to West
Fig F2 - 18 Through Trip – West to East
Fig F2 - 19 Sheffield to Waverley
Fig F2 - 20 Waverley to Sheffield
Appendix F3 – Route Checking – Evening Peak
Fig F3 - 1 Ecclesall Rd to City Centre Zone 17
Fig F3 - 2 City Centre Zone 17 to Ecclesall Rd
Fig F3 - 3 City centre Zone 17 to Ecclesall Rd
Fig F3 - 4 Ecclesall Rd – City Centre Zone 7
Fig F3 - 5 Hillsborough to City Centre
Fig F3 - 6 City Centre to Hillsborough
Fig F3 - 7 North to Lower Don Valley
Fig F3 - 8 Lower Don Valley to North
Fig F3 - 9 South to Lower Don Valley
Fig F3 - 10 Lower Don Valley to South
Fig F3 - 11 Meadowhall to Rotherham
Fig F3 - 12 Rotherham to Meadowhall
Fig F3 - 13 Rotherham to Sheffield
Fig F3 - 14 Sheffield to Rotherham
Fig F3 - 15 Through Trip – North to South
Fig F3 - 16 Through Trip – South to North
Fig F3 - 17 Through Trip – South to North
Fig F3 - 18 Through Trip North to South
Fig F3 - 19 Sheffield to Waverley
Fig F3 - 20 Waverley to Sheffield
Appendix G – Trip Matrix Summaries
1 Key
1.1 For each of the three time periods, the prior matrix and post matrix estimation trip totals ( in
pcu’s) are shown, together with the absolute difference, percentage difference and GEH
value (a composite measure taking both absolute and percentage difference into account).
1.2 Any cells where the percentage change is greater than 50% is also highlighted in red.
1.3 Similarly, for the GEH tables, anything with a GEH value greater than 5 is highlighted in red.
Table G1 Initial Morning Peak Matrix
1 2 3 4 5 6 7 8 9 Total
1 829 31 111 825 211 27 41 346 24 2,445
2 158 734 460 479 100 36 79 661 7 2,714
3 602 450 5,146 2,130 387 113 190 1,070 21 10,108
4 2,246 244 1,417 15,728 1,813 208 573 3,880 79 26,189
5 1,135 114 523 3,779 7,697 232 387 2,913 189 16,969
6 25 6 32 100 47 668 503 523 15 1,919
7 256 40 219 976 285 1,046 5,392 2,591 82 10,886
8 2,201 768 1,331 6,513 2,597 1,887 3,590 26,711 305 45,903
9 145 7 37 261 300 72 156 432 418 1,830
Total 7,597 2,393 9,276 30,792 13,439 4,289 10,911 39,126 1,140 118,964
Table G2 Final Morning Peak Matrix
1 2 3 4 5 6 7 8 9 Total
1 1,117 63 158 1,480 601 47 115 576 83 4,241
2 197 381 743 200 60 36 67 518 3 2,206
3 858 424 5,177 1,978 278 164 281 1,399 19 10,579
4 2,474 208 1,289 15,795 2,020 181 639 3,941 100 26,646
5 1,163 52 493 4,217 8,334 142 291 3,062 221 17,976
6 17 12 23 95 35 881 477 511 18 2,068
7 212 85 254 1,134 254 1,068 5,767 2,569 160 11,503
8 2,044 571 1,685 5,833 2,703 2,358 3,180 26,280 308 44,962
9 140 1 26 320 328 67 217 483 434 2,016
Total 8,222 1,797 9,848 31,052 14,613 4,945 11,035 39,338 1,346 122,196
Table G3 Absolute Difference – Morning Peak
1 2 3 4 5 6 7 8 9 Total
1 288 33 47 655 390 20 74 230 59 1,796
2 40 -353 283 -280 -40 0 -12 -142 -4 -508
3 256 -26 31 -152 -109 52 91 330 -2 470
4 227 -36 -128 67 206 -27 66 61 20 457
5 28 -61 -30 438 637 -91 -95 149 33 1,007
6 -8 6 -9 -5 -12 213 -27 -12 3 149
7 -44 45 35 158 -31 22 376 -21 79 617
8 -157 -196 355 -680 106 471 -410 -432 3 -942
9 -5 -6 -11 59 27 -5 61 50 16 186
Total 625 -596 572 260 1,174 656 124 212 206 3,232
Table G4 Percentage Difference – Morning Peak
1 2 3 4 5 6 7 8 9 Total
1 35% 108% 42% 79% 185% 75% 178% 66% 245% 73%
2 25% -48% 62% -58% -40% 0% -15% -22% -55% -19%
3 43% -6% 1% -7% -28% 46% 48% 31% -8% 5%
4 10% -15% -9% 0% 11% -13% 12% 2% 26% 2%
5 2% -54% -6% 12% 8% -39% -25% 5% 17% 6%
6 -32% 99% -28% -5% -26% 32% -5% -2% 22% 8%
7 -17% 111% 16% 16% -11% 2% 7% -1% 96% 6%
8 -7% -26% 27% -10% 4% 25% -11% -2% 1% -2%
9 -3% -84% -29% 22% 9% -7% 39% 12% 4% 10%
Total 8% -25% 6% 1% 9% 15% 1% 1% 18% 3%
Table G5 GEH Difference – Morning Peak
1 2 3 4 5 6 7 8 9 Total
1 9.2 4.8 4.0 19.3 19.4 3.3 8.3 10.7 8.1 31.1
2 3.0 15.0 11.5 15.2 4.4 0.0 1.3 5.9 1.7 10.2
3 9.5 1.3 0.4 3.4 6.0 4.4 5.9 9.4 0.4 4.6
4 4.7 2.4 3.5 0.5 4.7 1.9 2.7 1.0 2.1 2.8
5 0.8 6.7 1.3 6.9 7.1 6.6 5.2 2.7 2.3 7.6
6 1.7 2.0 1.7 0.5 1.9 7.7 1.2 0.5 0.8 3.3
7 2.9 5.6 2.3 4.9 1.9 0.7 5.0 0.4 7.1 5.8
8 3.4 7.6 9.1 8.7 2.1 10.2 7.1 2.7 0.1 4.4
9 0.4 3.0 1.9 3.4 1.5 0.6 4.4 2.4 0.8 4.2
Total 7.0 13.0 5.9 1.5 9.9 9.6 1.2 1.1 5.8 9.3
Table G6 Initial Inter-peak Matrix
1 2 3 4 5 6 7 8 9 Total
1 1,275 53 270 1,745 489 30 85 814 47 4,808
2 48 376 490 254 106 17 67 552 4 1,914
3 231 428 3,315 1,845 352 48 142 828 16 7,205
4 1,744 262 1,858 13,339 2,460 128 796 4,194 117 24,898
5 411 91 353 2,390 5,726 88 285 2,243 306 11,894
6 37 16 54 144 95 451 1,083 1,014 36 2,930
7 78 69 148 783 288 1,227 3,573 2,862 98 9,127
8 669 496 847 4,135 2,197 962 2,676 20,195 284 32,462
9 26 2 18 113 312 36 103 278 245 1,132
Total 4,519 1,793 7,353 24,748 12,024 2,986 8,811 32,978 1,156 96,368
Table G7 Final Inter-peak Matrix
1 2 3 4 5 6 7 8 9 Total
1 1,776 46 263 1,915 748 28 115 986 73 5,951
2 43 165 710 177 50 9 43 478 1 1,676
3 319 660 3,757 2,021 143 51 182 942 7 8,082
4 2,048 165 1,993 15,161 2,641 121 833 4,307 112 27,382
5 387 54 218 2,432 6,379 48 253 2,151 429 12,351
6 25 27 62 124 63 868 1,144 1,112 55 3,480
7 82 90 156 786 235 1,218 4,575 3,023 176 10,342
8 807 364 829 4,929 2,329 1,011 2,618 19,434 288 32,608
9 32 1 11 110 393 34 128 310 260 1,279
Total 5,520 1,572 7,999 27,656 12,980 3,386 9,891 32,744 1,402 103,151
Table G8 Absolute Difference – Inter-peak
1 2 3 4 5 6 7 8 9 Total
1 501 -8 -7 170 259 -2 30 173 26 1,143
2 -5 -211 220 -77 -56 -8 -25 -74 -3 -238
3 88 233 442 177 -209 2 40 114 -9 877
4 304 -97 136 1,822 180 -7 37 114 -5 2,485
5 -24 -37 -135 41 653 -40 -32 -92 123 458
6 -12 11 8 -19 -32 416 61 99 18 550
7 4 21 8 4 -53 -9 1,002 161 78 1,215
8 138 -132 -19 794 133 49 -59 -762 4 147
9 6 -1 -8 -2 81 -2 25 32 15 148
Total 1,001 -221 646 2,909 956 400 1,080 -234 246 6,783
Table G9 Percentage Difference – Inter-peak
1 2 3 4 5 6 7 8 9 Total
1 39% -15% -2% 10% 53% -7% 36% 21% 54% 24%
2 -10% -56% 45% -30% -53% -47% -37% -13% -72% -12%
3 38% 54% 13% 10% -59% 5% 28% 14% -57% 12%
4 17% -37% 7% 14% 7% -5% 5% 3% -4% 10%
5 -6% -41% -38% 2% 11% -46% -11% -4% 40% 4%
6 -32% 70% 15% -13% -34% 92% 6% 10% 51% 19%
7 5% 30% 5% 0% -18% -1% 28% 6% 79% 13%
8 21% -27% -2% 19% 6% 5% -2% -4% 1% 0%
9 25% -48% -42% -2% 26% -5% 25% 12% 6% 13%
Total 22% -12% 9% 12% 8% 13% 12% -1% 21% 7%
Table G10 GEH Difference – Inter-peak
1 2 3 4 5 6 7 8 9 Total
1 12.8 1.1 0.4 4.0 10.4 0.4 3.0 5.8 3.3 15.6
2 0.7 12.8 9.0 5.3 6.3 2.2 3.3 3.2 1.8 5.6
3 5.3 10.0 7.4 4.0 13.3 0.3 3.1 3.8 2.7 10.0
4 7.0 6.6 3.1 15.3 3.6 0.6 1.3 1.7 0.5 15.4
5 1.2 4.4 8.0 0.8 8.4 4.9 1.9 2.0 6.4 4.2
6 2.2 2.4 1.1 1.7 3.6 16.2 1.8 3.0 2.7 9.7
7 0.4 2.3 0.6 0.1 3.3 0.3 15.7 3.0 6.6 12.3
8 5.1 6.4 0.6 11.8 2.8 1.6 1.1 5.4 0.2 0.8
9 1.2 0.7 2.0 0.2 4.3 0.3 2.4 1.9 0.9 4.3
Total 14.1 5.4 7.4 18.0 8.6 7.1 11.2 1.3 6.9 21.5
Table G11 Initial Evening peak Matrix
1 2 3 4 5 6 7 8 9 Total
1 732 108 487 2,214 898 24 212 1,820 116 6,612
2 44 855 641 343 118 10 83 775 4 2,873
3 243 521 5,507 1,963 514 22 202 1,174 35 10,181
4 1,142 446 2,450 15,951 3,434 108 1,018 6,247 241 31,036
5 261 99 430 2,411 8,849 46 309 2,658 349 15,412
6 13 34 94 200 205 533 1,208 1,565 68 3,919
7 39 93 220 928 413 652 5,385 3,722 171 11,623
8 451 711 1,119 4,611 3,156 605 2,923 27,182 473 41,231
9 26 8 17 103 220 16 94 342 430 1,257
Total 2,951 2,875 10,965 28,723 17,807 2,017 11,435 45,484 1,887 124,143
Table G12 Final Evening Peak Matrix
1 2 3 4 5 6 7 8 9 Total
1 1,109 100 569 2,325 1,061 28 243 1,678 171 7,285
2 65 434 731 255 47 5 39 738 0 2,314
3 369 1,013 5,608 2,129 264 38 210 1,554 13 11,199
4 1,441 248 2,378 16,147 3,385 106 1,243 6,797 282 32,029
5 283 35 352 2,469 9,200 35 296 2,414 475 15,559
6 8 28 96 136 186 824 1,224 1,851 78 4,431
7 29 56 286 870 387 704 6,152 3,623 291 12,397
8 481 603 1,298 4,499 3,588 707 2,901 27,399 540 42,016
9 26 3 18 131 300 37 178 338 469 1,500
Total 3,811 2,520 11,335 28,962 18,417 2,484 12,486 46,392 2,320 128,730
Table G13 Absolute Difference – Evening Peak
1 2 3 4 5 6 7 8 9 Total
1 376 -8 82 111 164 4 31 -142 56 673
2 21 -421 90 -88 -71 -5 -44 -37 -3 -559
3 126 492 100 165 -250 16 9 380 -21 1,018
4 299 -198 -72 197 -48 -1 225 551 41 993
5 22 -63 -78 58 350 -11 -13 -244 126 148
6 -5 -5 2 -63 -19 291 15 286 10 512
7 -9 -37 65 -58 -27 51 767 -99 120 774
8 30 -108 179 -112 432 102 -22 217 67 785
9 0 -5 1 28 80 22 83 -4 38 243
Total 861 -355 370 239 611 468 1,051 908 434 4,587
Table G14 Percentage Difference – Evening Peak
1 2 3 4 5 6 7 8 9 Total
1 51% -8% 17% 5% 18% 15% 15% -8% 48% 10%
2 48% -49% 14% -26% -60% -54% -53% -5% -88% -19%
3 52% 95% 2% 8% -49% 72% 4% 32% -62% 10%
4 26% -44% -3% 1% -1% -1% 22% 9% 17% 3%
5 9% -64% -18% 2% 4% -23% -4% -9% 36% 1%
6 -36% -16% 2% -32% -9% 55% 1% 18% 14% 13%
7 -24% -40% 30% -6% -6% 8% 14% -3% 70% 7%
8 7% -15% 16% -2% 14% 17% -1% 1% 14% 2%
9 1% -67% 5% 27% 36% 138% 88% -1% 9% 19%
Total 29% -12% 3% 1% 3% 23% 9% 2% 23% 4%
Table G15 GEH Difference – Evening Peak
1 2 3 4 5 6 7 8 9 Total
1 12.4 0.8 3.6 2.3 5.2 0.7 2.0 3.4 4.6 8.1
2 2.8 16.6 3.4 5.1 7.8 2.0 5.7 1.3 2.3 11.0
3 7.2 17.8 1.3 3.7 12.7 2.9 0.6 10.3 4.4 9.8
4 8.3 10.6 1.5 1.6 0.8 0.1 6.7 6.8 2.6 5.6
5 1.4 7.7 3.9 1.2 3.7 1.7 0.7 4.9 6.2 1.2
6 1.4 1.0 0.2 4.9 1.4 11.2 0.4 6.9 1.1 7.9
7 1.6 4.3 4.1 1.9 1.3 2.0 10.1 1.6 7.9 7.1
8 1.4 4.2 5.2 1.7 7.4 4.0 0.4 1.3 3.0 3.8
9 0.0 2.4 0.2 2.6 5.0 4.2 7.1 0.2 1.8 6.5
Total 14.8 6.8 3.5 1.4 4.5 9.9 9.6 4.2 9.5 12.9
Appendix H – Trip Length Distributions
1 Key
1.1 The trips length distribution plots show variations in trip length distribution for Car, LGV and
OGV between the prior matrices and post matrix estimation matrices.
1.2 This analysis is presented looking at the 3 user class matrices that were run through matrix
estimation.
Morning Peak Car Trip Length Distribution
0
2
4
6
8
10
12
14
16
18
20
0 to 2 2 to 4 4 to 6 6 to 8 8 to 10 10 to 12 12 to 14 14 to 16 16 to 18 18 to 20 20 to 22 22 to 24 24 to 26 26 to 28 28 to 30 Above30
Trip Length (km)
Perc
enta
ge o
f Tot
al T
rips
(%)
Before Matrix Estimation After Matrix Estimation
Figure H1 Morning Peak Trip Length Distribution for Cars
Morning Peak LGV Trip Length Distribution
0
5
10
15
20
25
0 to 2 2 to 4 4 to 6 6 to 8 8 to 10 10 to 12 12 to 14 14 to 16 16 to 18 18 to 20 20 to 22 22 to 24 24 to 26 26 to 28 28 to 30 Above30
Trip Length (km)
Perc
enta
ge o
f Tot
al T
rips
(%)
Before Matrix Estimation After Matrix Estimation
Figure H2 Morning Peak Trip Length Distribution for Light Goods Vehicles
Morning Peak OGV Trip Length Distribution
0
10
20
30
40
50
60
0 to 2 2 to 4 4 to 6 6 to 8 8 to 10 10 to 12 12 to 14 14 to 16 16 to 18 18 to 20 20 to 22 22 to 24 24 to 26 26 to 28 28 to 30 Above30
Trip Length (km)
Perc
enta
ge o
f Tot
al T
rips
(%)
Before Matrix Estimation After Matrix Estimation
Figure H3 Morning Peak Trip Length Distribution for Other Goods Vehicles
Inter Peak Car Trip Length Distribution
0
5
10
15
20
25
0 to 2 2 to 4 4 to 6 6 to 8 8 to 10 10 to 12 12 to 14 14 to 16 16 to 18 18 to 20 20 to 22 22 to 24 24 to 26 26 to 28 28 to 30 Above30
Trip Length (km)
Perc
enta
ge o
f Tot
al T
rips
(%)
Before Matrix Estimation After Matrix Estimation
Figure H4 Inter-peak Trip Length Distribution for Cars
Inter Peak LGV Trip Length Distribution
0
5
10
15
20
25
0 to 2 2 to 4 4 to 6 6 to 8 8 to 10 10 to 12 12 to 14 14 to 16 16 to 18 18 to 20 20 to 22 22 to 24 24 to 26 26 to 28 28 to 30 Above30
Trip Length (km)
Perc
enta
ge o
f Tot
al T
rips
(%)
Before Matrix Estimation After Matrix Estimation
Figure H5 Inter-peak Trip Length Distribution for Light Goods Vehicles
Inter Peak OGV Trip Length Distribution
0
5
10
15
20
25
30
35
40
45
50
0 to 2 2 to 4 4 to 6 6 to 8 8 to 10 10 to 12 12 to 14 14 to 16 16 to 18 18 to 20 20 to 22 22 to 24 24 to 26 26 to 28 28 to 30 Above30
Trip Length (km)
Perc
enta
ge o
f Tot
al T
rips
(%)
Before Matrix Estimation After Matrix Estimation
Figure H6 Inter-peak Trip Length Distribution for Other Goods Vehicles
Evening Peak Car Trip Length Distribution
0
2
4
6
8
10
12
14
16
18
0 to 2 2 to 4 4 to 6 6 to 8 8 to 10 10 to 12 12 to 14 14 to 16 16 to 18 18 to 20 20 to 22 22 to 24 24 to 26 26 to 28 28 to 30 Above30
Trip Length (km)
Perc
enta
ge o
f Tot
al T
rips
(%)
Before Matrix Estimation After Matrix Estimation
Figure H7 Evening Peak Trip Length Distribution for Cars
Evening Peak LGV Trip Length Distribution
0
5
10
15
20
25
30
0 to 2 2 to 4 4 to 6 6 to 8 8 to 10 10 to 12 12 to 14 14 to 16 16 to 18 18 to 20 20 to 22 22 to 24 24 to 26 26 to 28 28 to 30 Above30
Trip Length (km)
Perc
enta
ge o
f Tot
al T
rips
(%)
Before Matrix Estimation After Matrix Estimation
Figure H8 Evening Peak Trip Length Distribution for Light Goods Vehicles
Evening Peak OGV Trip Length Distribution
0
10
20
30
40
50
60
70
80
0 to 2 2 to 4 4 to 6 6 to 8 8 to 10 10 to 12 12 to 14 14 to 16 16 to 18 18 to 20 20 to 22 22 to 24 24 to 26 26 to 28 28 to 30 Above30
Trip Length (km)
Perc
enta
ge o
f Tot
al T
rips
(%)
Before Matrix Estimation After Matrix Estimation
Figure H9 Evening Peak Trip Length Distribution for Other Goods Vehicles
Appendix I – Index of Important Files
1 Networks
Morning Peak sr07a447_6uc.dat
Inter Peak sr07i447_6uc.dat
Evening Peak sr07p447_6uc.dat
2 Matrices
Morning Peak sr07a447_6uc.ufm
Inter Peak sr07i447_6uc.ufm
Evening Peak sr07p447_6uc.ufm
3 Assigned Networks
Morning Peak sr07a447_6uc.ufs
Inter Peak sr07i447_6uc.ufs
Evening Peak sr07p447_6uc.ufs
4 Calibration Counts Files
All Counts Counts{A,I,P}027All.dat
Sheffield and Rotherham District Counts
Counts{A,I,P}027SheffandRoth.dat
5 Pelican Data
sr08pel08.dat
6 Coordinates File
Sr08a128.xy
7 Bus Route Files
Morning Peak sr08abus001.dat
Inter-peak sr08ibus001.dat
Evening Peak sr08pbus001.dat
8 Journey Time Files
Morning Peak 2007 routes - sr06aJTSh101a.dat
2006 routes - sr06aJTSh101b.dat
Inter-peak 2007 routes - sr06iJTSh101a.dat
2006 routes - sr06iJTSh101b.dat
Evening Peak 2007 routes - sr06pJTSh101a.dat
2006 routes - sr06pJTSh101b.dat
9 KR Files for Matrix Estimation
All Counts
Fully Observed Trips Frozen sr06{a,i,p}kr{c,l,o}001.dat
Sheffield and Rotherham Counts
Fully Observed Trips Frozen sr06{a,i,p}kr{c,l,o}002.dat
Sheffield and Rotherham Counts sr06{a,i,p}kr{c,l,o}003.dat
All Counts sr06{a,i,p}kr{c,l,o}004.dat
Appendix J – Method for Controlling Matrices to Tempro
Technical Note
Project Title: SRHM3
MVA Project Number: C37272
Subject: Method for Controlling Matrices to TEMPRO
Note Number: 02 Version: 1
Author(s): John Allan
Reviewer(s): Alice Woolley, Peter Kidd, James Blythe, Nick Benbow
Date: 10 March 2009
Introduction
To build the SATURN model matrices for the SRHM3 highway model, we carried out matrix
estimation on the SRHM2 matrices using new counts gathered since the Northern Inner Relief
Road was built. The matrix estimation was carried out with all car based journey purposes
gathered together into a single user class – because the traffic counts do distinguish between
journey purposes.
The SRDM3 model requires matrices at a very detailed level of journey purpose segmentations. To
build matrices for SRDM3 we had to transfer the impact of matrix estimation from the all-car
matrices to the individual journey purpose matrices built for SRHM2.
In building matrices for SRDM3, we have discovered a mismatch between the journey purpose
proportions in the model and those in TEMPRO. The model matrices have more employer’s
business trips and more non-home based trips than TEMPRO. However, we cannot match the
trip-ends exactly because for one journey purpose the fully observed trip ends exceed the
TEMPRO total by a large margin. We have investigated the processing of the roadside interview
data and concluded that the mismatch is carried through from the raw data.
The mismatch between the RIS data and TEMPRO probably reflects the differing biases inherent in
the two different survey approaches. In RIS, we may expect long distance trips to be over-
represented and short-distance trips to be under represented, this means that RIS may contain
a high proportion of employer’s business trips and a low proportion of eductation trips. OIn the
other hdand, household surveys, such as the one on which TEMPRO is based, tend to under
represent trips in the middle of trip chains so they will under represent non-home based trips
including non-home based employers business The true proportions are probably somewhere
between those estimated in the two different surveys.
We considered three possible approaches for controlling the SATURN trip-ends to match TEMPRO.
We have decided to use an approach that uses trip rates calculated from TEMPRO with
population data at zonal level.
Outline of the Approach
The broad approach is as follows:
Task 1 - Calculate 12-hour tour trip-rates from TEMPRO using the trip ends and
population data for the combined Sheffield and Rotherham districts, for the purposes HW,
HEB, HED ..
Task 2 - Calculate 12-hour to modelled-hour factors by purpose separately for from-home
and to-home matrices
Task 3 - Multiply the 12 hour trip rates by the 12-hour to modelled hour factors to
produce hourly trip rates
Task 4 - Apply the modelled hour trip rates to the zonal population data to produce hourly
trip ends - note that the zonal population will need to be controlled to the latest mid-year
population estimates because the population in Census seems very different from the
population in TEMPRO.
Task 5 - Use the hourly trip ends in MVGRAM to produce synthetic matrices (with intra-
zonals estimated using an estimated cost of half the cost to the nearest zone)
Task 6 - Calculate K-factors at sector level to ensure that the gravity model matches the
observed data and that the in-fill data sums to the target trip-end less the observed data
Task 7 - Re-run MVGRAM to produce revised matrices
Task 8 - Over write the fully observed cells with the fully observed data
Problem with NHBEB and an Adjustment
This approach will work for all the journey purposes except NHBEB for which the fully
observed demand exceeds the total forecast by TEMPRO. For NHBEB, we will build synthetic
matrices using trip ends calculated from the TEMPRO trip rates and we will overwrite the fully
observed cells with fully observed demand. We will not adjust the total to match TEMPRO.
The approach we have adopted for NHBEB will yield matrices that do not match the journey
purpose proportions in TEMPRO for either EB of NHB trips. In fact the matrices will end up
with EB and NHB proportions that lie somewhere in between the proportions in TEMPRO and
those in the original roadside interview surveys.
From what we know about the types of survey from which the two alternatives were
estimated, it seems reasonable to suppose that the true proportions lie somewhere between
the two. We expect that the roadside interview surveys have a bias towards boosting the
proportion of observed trips on longer journey purposes (such as employer’s business ) and
reducing the proportion of trips on shorter journey purposes (such as education trips). We
also expect that a household interview survey (such as NTS on which the TEMPRO trip ends
were based) is likely to yield underestimates of non-compulsory trips, employer’s business
trips and non-home based trips.
Task 1 Daily Trip Rates from TEMPRO
Task 1
Task 1 Calculate daily trip rates from TEMPRO
Objective To generate trip rates for each TEMPRO journey purpose – for car an public
transport
Inputs TEMPRO PA trips
TEMPRO variables
Census Variables
Processes Find a set of variables that are common to Census and TEMPRO so we can
calculate trip rates
Find variables that can take some account of
Outputs 14 trip rates – one 12-hour trip rate for car trips for each of the 7 journey
purposes and one rate for PT trips
We plan to use trip rates and population data to estimate production trip ends for the zones
in the study area. For the attractions, we plan to use attraction weights – which MVGRAM
will factor so that the trip productions are preserved.
We need to calculate rates in TEMPRO that can be applied at zonal level so we need
segmentation variables are available in both TEMPRO and our zonal data.
Trip rates could be calculated using the following TEMPRO variables
Households
Workers
Jobs
Population
Households by car ownership 0, 1 , 2, 3+
Total number of cars
Zone level Census data is available at for the trip end calculations for the following variables:
Residential population
Daytime population
Households
Households with a car
Households without a car
Jobs
Jobs reached by car
Jobs reached by PT
Jobs reached by other mode
Workers
Resident workers who normally drive
Resident workers who normally use PT
Resident workers who normally use another mode
Number of non-car owning households
We could try to take some account of car ownership levels by building car ownership data
into the trip rates. To achieve this, we need to use variables that are common to the two
data sets. The only possibility is to use the ratio of car owning households to total
households.
We assume that workers who want to buy a car will buy one so the car ownership level
within a zone will already be reflected in the numbers of workers within the zone. We
therefore have not applied the car ownership adjustment to commute trips or employers
business trips.
To produce trip rates for each purpose we have divided numbers of productions-attractions in
TEMPRO by the denominators in Table 1.
Trip Rate Denominators
Purpose Trip Rate Denominator
HBW Workers
HBEB Workers
HBED Population * households with car / Total
households
HBShop Population * households with car / Total
households
HBOther Population * households with car / Total
households
NHBEB Jobs
NHBO Population
Note that the productions-attractions in TEMPRO are outbound and return legs – to convert
home-based productions-attractions to the number of journeys you must multiply by 2. The
non-home based productions-attractions journeys in TEMPRO do not need to be multiplied by
two – they are already origin-destination journeys.
Having chosen the TEMPRO variables for the denominators for trip rates, we must identify
the corresponding variables in the Census data. These variables are set out in Table 2. Note
that the some of the fields in this table are not directly available from the census data rather
they have been calculated from other census variables. Examples include the number of
workers who travel to work by each mode, which we have summed over the modes provided
in the census data.
The calculation of trip demand for home based purposes will be constrained by the
production totals so we don’t need to produce attraction trip-rates. Instead we can use
attraction weights. The benefits of using attraction weights rather than trip rates is that the
variables used for the weights need not be common to both data sets, they can exist just in
the zonal data. That gives us a little more information that we can use to distribute the
commute trips according the car commute usage within the attraction zone.
Production Variables and Attraction Weights
Purpose Production Variable Attraction Weight
HBW Workers Daytime TTW by Car
HBEB Workers r Daytime TTW by Car plus PT
HBED resident population * Population
* households with car / Total
households
Resident population
HBShop resident population * Population
* households with car / Total
households
Daytime population
HBOther resident population * Population
* households with car / Total
households
Daytime population
NHBEB Jobs Daytime TTW by Car plus PT
NHBO Daytime population Daytime population
The output of this step will be a set of 14 trip rates: 7 journey
12-Hour to Modelled-hour Factors
Task 2
Task 2 Calculate hourly trip rates
Objective Generate factors to convert 12-hour trip rates into modelled hour trip rates
Inputs Observed RIS journey purpose proportions by modelled hour for the 12
hour day
Processes Separately for from-home and to-home trips, calculate the proportion of the
12-hour total that occur in the three modelled hours in the highway model –
0800 to 0900, average of 1000 to 1600 , and 1700 to 1800.
Multiply the 12-hour trip rates by the hour factors
Outputs 84 trip rates – one trip rate for for each combination of the 2 directions, 3
periods, 7 journey purposes, and 2 modes (car and PT)
Hourly Trip Ends
Task 3
Task 3 Calculate hourly trip-ends
Objective Create production trip-ends and attraction weights for the gravity model
Inputs Hourly trip-rates
Census data by zone – including attraction end data in the external zones
Observed productions from the external area to the internal area only (ie
exclude external to external trips)
Processes For internal area production trip-ends, multiplying the hourly trip rates by
the census data
For external area productions, use the observed trips to attractions within
the study area
Generate attraction weights by extracting the relevant census data by zone,
then factor the attraction weights to match the total productions – last step
may not be needed as MVGRAM may take care of it. We need attraction
data in the external area because some trips will be attracted from the study
area. Although the gravity model will only deal with internal area
productions we can use the full attraction data because the gravity model
will be set up so that it does not produce any external-to-external trips.
Outputs 84 sets of production trip ends and attraction weights – one for each
combination of the 2 directions, 3 periods, 7 journey purposes, and 2
modes (car and PT)
Task 4 - Initial MVGRAM
Task 4
Task 4 Calculate hourly trip rates
Objective Generate initial synthetic matrices for each purpose and time period
combination
Inputs Trip-ends from the previous process
X1 and X2 factors for the gravity models calibrated from the fully observed
RIS data
Distance skim matrices
Processes Zeroise the external-to-external cells in the skim matrices
Transpose the skim matrices for use with transposed the to-home
productions and attractions to make productions into rows in the matrices
Run gravity models separately for each journey purpose – separately for
home to work and work to home.
Run the models so productions are rows and attractions are columns – use
the normal skim for from-home and the transposed skim for to-home.
Outputs 84 trip matrices– one for each combination of the 2 directions, 3 periods, 7
journey purposes, and 2 modes (car and PT)
Task 5 Calculate K-factors
Task 5
Task 5 Generate K-Factors
Objective Generate a set of K-factors that can be used in a second run of the gravity
model to ensure it matches all the fully observed sector-to-sector and the
sector productions, so that we can be sure that the intra sector movements
have the correct totals
Inputs Observed RIS data
Mask to identify fully observed data and not fully observed data
Synthetic matrices from MVGRAM
Processes Identify the fully observed RIS
SQEX both fully observed RIS and synthetic matrices to the RIS sector
system
Divide SQEXed synthetic by RIS – to produce K-factors at sector level – note
the leading diagonal should be set to 1.0 as an experiment. We may need
to return to this step to set the leading diagonal factor so that the
productions in each sector match the synthetic totals. Check whether this
second step is necessary using one or two examples
Unsqex the factors to zone level.
Outputs 84 matrices of k-factors – one for each combination of the 2 directions, 3
periods, 7 journey purposes, and 2 modes (car and PT)
Task 6 Re-run MVGRAM
Task 6
Task 6 Generate K-Factors
Objective Rerun MVGRAM to produce synthetic matrices that both match both the
trip-end targets from TEMPRO and the fully observed movements at sector-
to-sector level
Inputs Trip-ends
X1 and X3 from the calibration
K-factors
Processes Run MVGRAM
Transpose the to-home matrices
Sum the from-home and to-home matrices
Outputs 84 synthetic matrices of hourly trips in PA format – one for each
combination of the 2 directions, 3 periods, 7 journey purposes, and 2
modes (car and PT)
Task 7 - Over write the fully observed cells with the fully observed data
Task 7
Task 7 Overwrite the fully observed cells
Objective Re-insert the fully observed RIS data into the trip matrices
Inputs Synthetic trips
Observed data from RIS
Mask matrices
Processes Mask the observed RIS to retain just the fully observed data
Mask the synthetic data to retain everything except the fully observed cells
Sum the masked matrices
Outputs 84 synthetic matrices of hourly trips in OD format – one for each
combination of the 2 directions, 3 periods, 7 journey purposes, and 2
modes (car and PT)
Task 8 - Over write the through trips
Task 8
Task 8 Overwrite the through trips
Objective Re-insert the through trips
Inputs Assignment matrices from the previous version of the model
Observed journey purpose proportions from RIS
Mask matrices
Matrices of trips from Task 8 (combined observed and synthetic for all the
trips with at least on trip-end inside the study area)
Processes Mask the old assignment matrices to produce just the external to external
trips
Apply JP splitting factors from RIS to get 5 user class assignment matrices
to our 14 user-class matrices
Add the masked matrices to the matrices of Task 8
Outputs 84 synthetic matrices of hourly trips in PA format – one for each
combination of the 2 directions, 3 periods, 7 journey purposes, and 2
modes (car and PT)
John Allan
Managing Consultant
Sheffield and Rotherham District SATURN Model 2008 4
MVA Consultancy provides advice on transport and other policy areas, to central, regional and local government, agencies, developers, operators and financiers. A diverse group of results-oriented people, we are part of a 350-strong team worldwide. Through client business planning, customer research and strategy development we create solutions that work for real people in the real world. For more information visit www.mvaconsultancy.com
Head Office
MVA House, Victoria Way
Woking, Surrey GU21 6DD United Kingdom
T: +44 (0)1483 728051 F: +44 (0)1483 755207
Birmingham
Second Floor, 37a Waterloo Street,
Birmingham, B2 5TJ, United Kingdom
T: +44 (0)121 233 7680 F: +44 (0)121 233 7681
Dubai
PO Box 123166, 803-805 Arbift Tower, Baniyas Road,
Deira, Dubai, UAE
T: +971 (0)4 223 0144 F: +971 (0)4 223 1088
Dublin
1st Floor, 12/13 Exchange Place, IFSC, Dublin 1, Ireland
T: +353 (0)1 542 6000 F: +353 (0)1 542 6001
Edinburgh
Stewart House, Thistle Street, North West Lane
Edinburgh EH2 1BY United Kingdom
T: +44 (0)131 220 6966 F: +44 (0)131 220 6087
Glasgow
Seventh Floor, 78 St Vincent Street
Glasgow G2 5UB United Kingdom
T: +44 (0)141 225 4400 F: +44 (0)141 225 4401
London
One Berners Street
London W1T 3LA United Kingdom
T: +44 (0)20 7612 3700 F: +44 (0)20 7436 9293
Lyon
11 rue de la Republique, 69001 Lyon, France
T: +33 (4) 72 10 29 29 F: +33 (4) 72 10 29 28
Manchester
25th Floor, City Tower, Piccadilly Plaza
Manchester M1 4BT United Kingdom
T: +44 (0)161 236 0282 F: +44 (0)161 236 0095
Marseille
13, rue Roux de Brignoles, 13006 Marseille, France
T: +33 (4) 91 37 35 15 F: +33 (4) 91 54 18 92
Paris
12-14, rue Jules Cesar, 75012 Paris, France
T: +33 (1) 53 17 36 00 F: +33 (1) 53 17 36 01
Email: [email protected]
Offices also in
Bangkok, Beijing, Hong Kong, Shenzhen and Singapore