D1.2 Theoretical principles and methodological approach of the PHOEBE framework and selection of tools

Deli e able 1.2
Theo e ical p inciples and
me hodological app oach o he
PHOEBE amewo k and
selec ion o ools
PHOEBE
-
PROJECT.EU
This p ojec has ecei ed unding om
he Eu opean Union’s Ho izon Eu ope
esea ch and inno a ion p og amme
unde g an ag eemen No 101076963
UK pa icipan s in Ho izon Eu ope P ojec
PHOEBE a e suppo ed by UKRI g an
numbe s 10038897 (The In e na ional
Road Assessmen P og amme – iRAP)
and 10056912 (The Floow)
The sole esponsibili y o he con en o his
documen lies wi h he au ho s. I does no
necessa ily e lec he opinion o he Eu opean
Union. Nei he CINEA no he Eu opean
Commission a e esponsible o any use ha may
be made o he in o ma ion con ained he ein.
phoebe-p ojec .eu
Copy igh © by PHOEBE
1
Documen Con ol Page
Deli e able numbe
1.2
Deli e able i le
Theo e ical p inciples and me hodological app oach o he PHOEBE
amewo k and selec ion o ools
Deli e able e sion
3.1
Wo k Package numbe
1
Wo k Package Ti le
PHOEBE F amewo k - Me hodological and echnical app oach
Due da e o deli e y
31/10/2023
Ac ual da e o deli e y
27/10/2023
Dissemina ion le el
Public
Type
Repo
Edi o (s)
Ami Pooyan A gha i, Amna Chaudh y, Ami Hossein Kalan a i, Shanna
Lucchesi, Monica Olyslage s
Con ibu o (s)
Ami Pooyan A gha i, Amna Chaudh y, Ami Hossein Kalan a i, Shanna
Lucchesi, Monica Olyslage s, Ma cel Sala, An onio Pellice , A una a
Pu a unda, Ch is elle Al Haddad, Cons an inos An oniou, Apos olos
Ziakopoulos, Ma ia Oikonomou, Sam Chapman, Eleono a Papadimi iou
Re iewe (s)
EIRA, UPV
P ojec name
P edic i e App oaches o Sa e U ban En i onmen s
P ojec Ac onym
PHOEBE
P ojec s a ing da e
01/11/2022
P ojec du a ion
45 mon hs
Righ s
PHOEBE conso ium
phoebe-p ojec .eu
Copy igh © by PHOEBE
2
PM e o s pe bene icia y ha con ibu ed o he deli e able
#
Pa ne
PM e o in he Deli e able
1
TUD
5.00
2
NTUA
0.50
3
TUM
3.75
4
iRAP
1.00
5
AIM
1.53
6
FLOOW
1.36
7
POLIS
0.01
8
EIRA
0.05
To al
PHOEBE Conso ium
13.20
phoebe-p ojec .eu
Copy igh © by PHOEBE
3
Documen His o y
Ve sion
Da e
Bene icia y
Desc ip ion
0.1
06/07/2023
Ami Pooyan A gha i &
Eleono a Papadimi iou (TUD)
Ou line o he Deli e able s uc u e and
d a o e iew and beha iou al models
0.2
29/08/2023
Amna Chaudh y, Ami Hossein
Kalan a i, Ami Pooyan A gha i
(TUD)
Apos olos Ziakopoulos & Ma ia
Oikonomou (NTUA)
A una a Pu a unda, Ch is elle
Al Haddad, Cons an inos
An oniou (TUM)
Shanna Lucchesi, Monica
Olyslage s & James B ad o d
(iRAP)
An onio Pellice & Ma cel Sala
(AIM)
Sam Chapman (FLOOW)
Inpu o he di e en sub-chap e s,
namely:
Road sa e y assessmen (iRAP),
T a ic mic osimula ion (AIM),
Beha iou al models (TUD),
Mode choice and modal shi (TUM),
Socioeconomic impac s (TUM),
Consolida ion o he amewo k (NTUA),
Selec ion o ools (iRAP)
1.0
26/09/2023
Amna Chaudh y, Ami Hossein
Kalan a i, Ami Pooyan A gha i
(TUD)
Fi s d a o he comple e deli e able
sha ed wi h con ibu ing pa ne s
1.1
02/10/2023
Amna Chaudh y, Ami Hossein
Kalan a i, Ami Pooyan A gha i
(TUD)
Final d a o he comple e deli e able
wi h commen s om all pa ne s
add essed Execu i e summa y and
conclusions o he Deli e able,
p oo eading, consolida ion o inal d a
o in e nal e iew (by EIRA, UPV)
2.0
25/10/2023
Amna Chaudh y, Ami Hossein
Kalan a i, Ami Pooyan A gha i
(TUD)
Upda ed e sion o he deli e able wi h
commen s om in e nal e iews
add essed and sen o all pa ne s o
las check be o e submission
2.1
26/10/2023
A una a Pu a unda (TUM)
Addi ional inpu p o ided and commen s
add essed by all pa ne s in he di e en
sub-chap e s
3.0
27/10/2023
Amna Chaudh y, Ami Hossein
Kalan a i, Ami Pooyan A gha i
(TUD)
Final documen
3.1
30/10/2023
Ami Pooyan A gha i, Amna
Chaudh y (TUD)
P og ess beyond he s a e o he a
added o he inal documen
phoebe-p ojec .eu
Copy igh © by PHOEBE
4
P ojec summa y
The EU- unded ‘P edic i e App oaches o Sa e U ban En i onmen ’ (PHOEBE) p ojec aims o de elop
an in eg a ed, dynamic human-cen ed p edic i e sa e y assessmen amewo k in u ban a eas. This will
be achie ed by b inging oge he he in e disciplina y powe o a ic simula ion, oad sa e y assessmen ,
human beha iou , mode shi and induced demand modelling and new and eme ging mobili y da a.
Focused on ulne able oad use s' sa e y, he 3.5-yea -long PHOEBE p ojec will d aw inspi a ion om
eal-wo ld scena ios in he h ee pilo ci ies o A hens (GR), Valencia (ES) and Wes Midlands (UK).
Tes ing ac i i ies will be pe o med ac oss he use cases o simula e and o ecas he impac o changes
on sa e y in di e en scena ios o dis up ions o ansi ions ac oss u ban anspo ne wo ks.
P edic ing and isualising he sa e y and socioeconomic ou comes o new o ms o anspo , new
echnologies, o egula o y and beha iou al changes om he indi idual (mic o) le el up o he ne wo k-
wide (mac o) le el will also be a signi ican game-change o u ban s akeholde s. The esul s o PHOEBE
can be used as a bluep in by o he Eu opean ci ies o de elop hei knowledge p oduc s, such as
socioeconomic analysis model, u ban oad sa e y assessmen , human beha iou and choice modelling.

phoebe-p ojec .eu
Copy igh © by PHOEBE
5
PHOEBE pilo ci ies
Lis o pa icipa ing ci ies:
• A hens (G eece)
• Valencia (Spain)
• Wes Midlands (Uni ed Kingdom)
Social Links:
h ps:// wi e .com/P ojec _PHOEBE
h ps://www.linkedin.com/company/phoebe-p ojec /
h ps://www.you ube.com/@phoebep ojec
Fo u he in o ma ion please isi WWW.PHOEBE-PROJECT.EU
phoebe-p ojec .eu
Copy igh © by PHOEBE
6
P ojec pa ne s
O ganisa ion
Coun y
Abb e ia ion
EVROPSKI INSTITUT ZA OCENJEVANJE CEST - EURORAP
SI
EIRA
ETHNICON METSOVION POLYTECHNION
EL
NTUA
TECHNISCHE UNIVERSITEIT DELFT
NL
TUD
TECHNISCHE UNIVERSITAET MUENCHEN
DE
TUM
AIMSUN SLU
ES
AIM
POLIS AISBL
BE
POLIS
FACTUAL CONSULTING SL
ES
FC
UNIVERSITAT POLITECNICA DE VALENCIA
ES
UPV
OSEVEN SINGLE MEMBER PRIVATE COMPANY
EL
O7
THE FLOOW LIMITED
UK
FLOOW
INTERNATIONAL ROAD ASSESSMENT PROGRAMME
UK
iRAP
phoebe-p ojec .eu
Copy igh © by PHOEBE
7
Lis o abb e ia ions and ac onyms
Ac onym
Meaning
AADT
A e age annual daily a ic
ADC
A e age di e ence o cos
ADE
A e age di e ence o uni o e ec i eness
API
Applica ion p og amming in e ace
AV
Au oma ed ehicle
BCR
Bene i -cos a io
CBA
Cos -bene i analysis
CF
Calib a ion ac o
CEA
Cos -e ec i eness analysis
CMF
C ash modi ica ion ac o
DR
Decele a ion a e
EVT
Ex eme alue heo y
EAI
Ex e nal agen in e ace
FHWA
Fede al Highway Adminis a ion (US)
FSI
Fa al and se ious inju y
GTFS
Gene al ansi eed speci ica ion
ICER
Inc emen al cos e ec i eness a io
KPI
Key pe o mance indica o
LMV
Ligh mobili y ehicle (mic o-mobili y)
LOC
Loss o con ol (c ash ype)
ML
Machine lea ning
NPV
Ne -p esen alue
OSM
Open S ee Map
PET
Pos -enc oachmen ime
PR
Pe cen age educ ion
SEA
Socio-economic analysis
SEM
S uc u al equa ion model
SSM
Su oga e sa e y measu e
SRS
S a Ra ing sco e
TD
T a el dia ies
TTC
Time o collision
ViDA
iRAP’s compu a ional pla o m
VoI
Value o inju y
VoSL
Value o s a is ical li e
VoT
Value o ime
VRU
Vulne able oad use s
WTP
Willingness- o-pay
phoebe-p ojec .eu
Copy igh © by PHOEBE
8
Table o Con en s
Execu i e summa y ................................................................................................................................... 12
P og ess beyond he s a e o he a ......................................................................................................... 14
S uc u e o he deli e able and links wi h o he wo k packages/deli e ables ...................................... 14
1 Theo e ical p inciples & me hodological app oaches ........................................................................ 16
1.1 Resea ch ques ions and objec i es ......................................................................................... 16
The How Ma ix explained ................................................................................................... 19
2 The PHOEBE F amewo k.................................................................................................................. 21
2.1 O e iew................................................................................................................................... 21
S ep 1 – Baseline calcula ions ............................................................................................. 23
S ep 2 – scena io change calcula ions ................................................................................ 24
S ep 3 – Socioeconomic analysis ........................................................................................ 25
2.2 Demand models ....................................................................................................................... 29
P ocesses ............................................................................................................................ 30
Da a ...................................................................................................................................... 30
Theo e ical basis .................................................................................................................. 35
Model es ing ........................................................................................................................ 37
2.3 T a ic mic osimula ion ............................................................................................................. 37
P ocesses ............................................................................................................................ 39
Models in a ic simula ion................................................................................................... 40
Da a ...................................................................................................................................... 40
Model es ing ........................................................................................................................ 41
2.4 Beha iou al models .................................................................................................................. 42
P ocesses ............................................................................................................................ 42
Da a ...................................................................................................................................... 44
Models .................................................................................................................................. 44
Model es ing ........................................................................................................................ 45
2.5 Road sa e y assessmen .......................................................................................................... 46
Da a ...................................................................................................................................... 46
Models and p ocesses ......................................................................................................... 50
Model es ing ........................................................................................................................ 54
2.6 Sa e y indica o s and dynamic isk ou pu s .............................................................................. 54
Taxonomy o sa e y ( isk) indica o s .................................................................................... 56
Risk indica o s in a ic mic osimula ion .............................................................................. 58
Dynamic sa e y ( isk) indica o s ........................................................................................... 60
phoebe-p ojec .eu
Copy igh © by PHOEBE
15
The ex en o he de ails co e ed in his deli e able a e ela ed o he o e all me hodology o he PHOEBE
amewo k meaning ha he desc ip ion wi hin he deli e able p ima ily co e s he heo e ical de ails on
he me hods/models unde each componen wi hin he amewo k along wi h hei in e ac ion wi h each
o he . The de ails on how di e en me hods/models wi hin each o he abo e componen s will be
p ac ically implemen ed in he ele an ools (e.g. AIMSUN mic osimula ion and iRAP oad sa e y
assessmen ), will be p o ided in PHOEBE Wo k Packages 3 (Model de elopmen and enhancemen )
and 4 (Sa e y use-case implemen a ion).
Finally, i is wo h men ioning ha he PHOEBE me hodological amewo k ( he ‘HOW’ ma ix oge he
wi h he p ocess lowcha ) se es wo pu poses: (i) o be a new esea ch and de elopmen (R&D)
amewo k ha ad ances he s a e-o - he-a and can be used by o he esea che s and schola s o
dynamically pe o ming he oad sa e y assessmen s h ough in eg a ing he exis ing, eme ging, and u u e
changes wi hin he demand models, a ic mic osimula ion, beha iou al models, and oad sa e y
assessmen models; and (ii) o be a “bluep in ” o how ci ies can es ablish and apply he p edic i e sa e y
assessmen amewo k e icien ly and cos -e ec i ely, p o iding a p ac ical guide on how i wo ks and
how o implemen i h ough he knowledge p oduc s such as socioeconomic analysis model, u ban oad
sa e y assessmen , human beha iou and demand modelling.

phoebe-p ojec .eu
Copy igh © by PHOEBE
16
1 Theo e ical p inciples &
me hodological app oaches
1.1 Resea ch ques ions and objec i es
The PHOEBE F amewo k is a me hodological app oach designed o ci ies o be e unde s and he u u e
sa e y implica ions o changes in he anspo sys ems, such as changes in oad use beha iou s, he
edesign o c ea ion o new in as uc u e, o he in oduc ion o a new mode. I aims o be a “bluep in ” o
how ci ies can es ablish and apply he p edic i e sa e y assessmen amewo k e icien ly and cos -
e ec i ely.
To do so, i aims o b ing oge he se e al exis ing elemen s cu en ly used in anspo planning,
assessmen and modelling, including:
• T a el demand and mode shi models, which can p edic consume choices and oad use
low changes;
• T a ic simula ion en i onmen s, which p o ide insigh s in o how hese a ious elemen s
in e ac in i ual a ic scena ios;
• Human beha iou models which can help an icipa e how oad use s espond in gi en
scena ios, and he ex en o which hey may de ia e om ypical beha iou s;
• Road sa e y assessmen , which can p o ide objec i e measu es o oad use sa e y linked o
speed, low and he physical oad ea u es; and
• Socio-economic impac assessmen which can unde s and he heal h, sa e y, economic and
en i onmen al impac s o changes in he anspo sys em.
The PHOEBE F amewo k b ings hese elemen s oge he o enable analysis on he combined e ec o
u u e changes, as well as how hey e ol e o e ime. The de elopmen o he PHOEBE F amewo k
in ol es h ee p incipal aspec s:
1 The heo e ical de elopmen s, which includes he esea ch, whe e necessa y, o key
p inciples and me hodology o he F amewo k. These include, o example, how o ecas
sa e y impac s can be ex apola ed ac oss a ci y-wide ne wo k.
2 The echnical de elopmen s equi ed o ensu e ha each componen o he F amewo k has
he necessa y unc ionali y o p oduce he analysis equi ed. This includes, o example, he
model de elopmen s o allow sa e y a ings o be inco po a ed in o a ic simula ion.
3 The ools and ma e ials equi ed o he use and exploi a ion o he F amewo k o i s
in ended end use s, o ensu e he F amewo k can be u ilised o i s s a ed pu pose and mee
i s desi ed objec i e. The PHOEBE F amewo k hus needs o ul ima ely be a p ac ical guide
on how i wo ks and how o implemen i .
This deli e able is he ou pu o he PHOEBE P ojec Tasks 1.2 (p oposi ion o he dynamic isk
assessmen me hodology), 1.3 (socioeconomic analysis me hodology) & 1.4 ( echnical de elopmen
p epa a ion). I aims o add ess how he ollowing esea ch and echnical aspec s equi ed o he
de elopmen o he PHOEBE F amewo k will be achie ed, namely:
phoebe-p ojec .eu
Copy igh © by PHOEBE
17
• How o model mode shi and induced demand o new modes o anspo and ex apola e i
o ne wo k-le el;
• How o in eg a e us in o a ic mic osimula ion en i onmen s, mo e accu a ely and mo e
comp ehensi ely (in e ms o hei sa e y);
• How o inco po a e he pe inen beha iou al models in o a ic mic osimula ion
en i onmen s;
• How o make use o a ic mic osimula ion and c ea e sui able sa e y indica o s o he
c ea ion o dynamic sa e y ou pu s in oad sa e y assessmen ;
• How o conduc a socio-economic analysis o he sa e y impac s; and
• How o de e mine he echnical me hods, esou ces and ools needed o deli e he echnical
aspec s o he p ojec and achie e he in eg a ion o he abo e aspec s (including echnical
speci ica ions and ela ed in e -dependencies).
I is impo an o no e ha while he esea ch and de elopmen pa o he F amewo k may include a b oad
ange o da a/models o each o he i e componen s in PHOEBE, no all o hem may ul ima ely be
implemen ed in PHOEBE. Only hose ha a e ele an and easible (pa icula ly wi h an eye on he use
case needs) will be implemen ed.
Figu e 1-1 shows how he di e en elemen s o he PHOEBE a e linked wi h one ano he in an in eg a ed
amewo k.
Figu e 1-1 The dynamic dimension o he PHOEBE amewo k
phoebe-p ojec .eu
Copy igh © by PHOEBE
18
As discussed in he s a e-o - he-a and end use needs e iew epo o PHOEBE (Deli e able 1.1), he
PHOEBE F amewo k spans di e en le els o analysis (mic o-, meso- and mac oscopic). Figu e 1-2
shows, concep ually, how each elemen connec s wi h hese le els.
Figu e 1-2 The mul ile el dimension o he PHOEBE amewo k
As he PHOEBE F amewo k wo ks o sys ema ise he in eg a ion o hese elemen s, he p ojec eam has
iden i ied se e al ocus ques ions o guide i s esea ch and echnical de elopmen s. These a e:
• Wha in o ma ion needs o be p o ided by mode sha e and induced demand models o in o m
he dynamic simula ion amewo k? How can changes in demand o new modes be modelled?
• Wha a e he en y poin s o a ic simula ion models o ecei e inpu o he dynamic isk
assessmen – how can induced demand and beha iou al models be in oduced in o a ic
simula ion?
• How can new modes o anspo and o he in e en ion scena ios in u ban a eas be modelled
in a ic simula ion en i onmen s?
• Wha beha iou al models a e bes sui ed o conside ing he mos c i ical beha iou s o oad
use s wi h po en ial sa e y impac s? How can beha iou al changes be linked o sa e y
ou pu s?
• How can oad sa e y assessmen be calcula ed ‘dynamically’, ha is, o di e en a ic
speeds and lows o e he cou se o a day?
• How can he ine spa ial dimensions o a ic simula ion be in eg a ed wi h he mesoscopic
sa e y assessmen om oad sa e y assessmen models?
• Wha sa e y indica o s a e needed o es ima e socio-economic impac s o policies and
changes in he anspo ne wo k? How can he PHOEBE F amewo k p o ide hese
indica o s?
phoebe-p ojec .eu
Copy igh © by PHOEBE
19
To add ess he ques ions, i is necessa y o unde s and he equi ed da a, models and p ocesses ha a e
wi hin each elemen ha make up he PHOEBE F amewo k, as well as he ela ionship be ween he
componen s. This includes: he inpu da a and en y poin s; all models o be applied on he da a coming
h ough hose en y poin s; and he ou pu s o hose models.
To assis wi h his, a ma ix in Figu e 1-3 maps he a ge ed da a, models and p ocesses in o he i e main
componen s o PHOEBE: demand models, a ic mic osimula ion, beha iou al models, oad sa e y
assessmen and socio-economic impac assessmen . The ma ix is ein o ced by he analysis scale in
each componen (mic o, meso and mac o).
Figu e 1-3 The “how” ma ix o da a, models and p ocesses in PHOEBE me hodology
The ma ix shows how each componen is in e ela ed wi h o he componen s (i.e., wha inpu s i ecei es
om o he componen s, wha models i applies on hose inpu s, and wha ou pu s i gi es o o he
componen s) and how he p edic i e sa e y assessmen mo es om mic o o meso o mac o scales o
analysis and he o he way a ound (i.e., scalabili y). Hence, he ma ix is e e ed o as he How Ma ix.
The How Ma ix explained
S a ing om he op le cell in he ma ix, he demand models ake se e al inpu s as explana o y a iables
o compu e a el demand and p o ide his demand in e ms o OD ma ices and mode choice p obabili ies
o a ic mic osimula ion (PDM12). In e u n, a ic mic osimula ion p o ides upda ed inpu s o explana o y
a iables due o he changes o he demand models o e-calcula e he a el demand, including he
induced demand (PDM21).
This back and o h gi ing and ecei ing occu o o he componen s in PHOEBE oo. Howe e , he e may
be no di ec exchange be ween some componen s in he amewo k. Fo example, demand models will
phoebe-p ojec .eu
Copy igh © by PHOEBE
20
no di ec ly p o ide any inpu s o beha iou al models, oad sa e y assessmen and socioeconomic impac
assessmen and hence he co esponding cells in he ma ix (PDM13, PDM14 and PDM15) a e g eyed ou .
T a ic mic osimula ion p o ides su oga e sa e y measu es as p oac i e indica o s o sa e y o beha iou al
models (PDM23) and dynamic low and speed o oad sa e y assessmen (PDM24). The e is no di ec link
be ween a ic mic osimula ion and socioeconomic impac assessmen (PDM25 and PDM52), o be ween
beha iou al models and socioeconomic impac assessmen (PDM35 and PDM53).
While beha iou al models will no gi e any di ec inpu o demand models (PDM31), hey p o ide a ic
mic osimula ion wi h he likelihood o oad use beha iou s (PDM32) and oad sa e y assessmen wi h
beha iou al FSI alues (PDM34). Road sa e y assessmen is he only componen in PHOEBE ha will gi e
inpu o all o he componen s. I will p o ide s a ic sa e y assessmen o oads ( o example, isk sco es)
o demand models (PDM41), a ic mic osimula ion (PDM42), and beha iou al models (PDM43). I also
p o ides he socioeconomic analysis wi h inal FSI es ima ions (PDM345) based on dynamic sa e y
assessmen . In e u n, he socioeconomic analysis will gi e he o e all impac assessmen back o he
demand models (PDM51) o adjus he demand a e an in e en ion has been implemen ed.
In he ollowing sec ions, de ails o he abo e da a (inpu and ou pu ), models and p ocesses in each
componen a e explained, and he in e - ela ionship be ween he cells in he How Ma ix a e hen
discussed. The esul o consolida ing he da a, models, p ocesses and he in e - ela ionship be ween
PHOEBE componen s is hen he in eg a ed scalable amewo k.

phoebe-p ojec .eu
Copy igh © by PHOEBE
21
2 The PHOEBE F amewo k
2.1 O e iew
The PHOEBE F amewo k assembles exis ing ools and models in such a manne ha , wi h he equi ed
amendmen s o hei inpu s and ou pu s, hey can exchange in o ma ion o o ecas he impac s on oad
sa e y mo e p ecisely, wi h emphasis on ulne able oad use s (VRU). In addi ion o he esea ch and
echnological de elopmen s equi ed o connec he PHOEBE F amewo k elemen s, he inco po a ion o
VRU in o some o hese elemen s equi es dedica ed a en ion.
1
Figu e 2-1 shows in de ail how hese in e connec ions wo k. I is impo an o no e i does no del e in o
he pa icula i ies o each componen . These pa icula i ies will be discussed in he ollowing sec ions.
The inpu da a wi hin each componen include necessa y da a om ex e nal sou ces and om o he
componen s wi hin he PHOEBE F amewo k. Al hough depic ed in S ep 1, he ex e nal da a inpu s a e
assumed o be a ailable, collec ed, cleaned and s anda dised be o e he p ocess s a (as in any ypical
oad sa e y e alua ion/calcula ion p ocess).
Speci ically, he da a inpu s o each componen ( om ou side he o he PHOEBE componen s) include
he ollowing:
1 Da a o oad sa e y assessmen includes ha ele an o p edic ion o u u e c ashes such
as his o ic c ash da a, ope a ional a ic low da a (speeds, lows o simila – e.g.
occupancies), geome ical and in as uc u e/ne wo k cha ac e is ics da a such as he numbe
o lanes on he oads, p esence o cycle lanes, pedes ian c ossings, bus s ops, a ic signals
e c.
2 Da a o mode choice and induced demand models includes ha ele an o mode choice
and a el demand modelling, such as socio-demog aphic da a, land-use da a, ne wo k da a,
da a om o igin-des ina ion (OD) su eys, e c.
3 Da a o he beha iou al modelling equi es human ac o s based beha iou al da a, such as
coun s o speci ic beha iou al ins ances (e.g. dis ac ion, speeding illegal c ossings), ela ed
geome ic, ne wo k & in as uc u e da a, ele an c ash da a and pe inen su ey da a.
4 Da a o he a ic mic osimula ion includes ha needed o modelling he s udy a ea including
ne wo k cha ac e is ics (geome ic layou , anspo ea u es (e.g. pedes ian c osswalks,
unc ional a ibu es, con igu a ions, e c.), a ic demand and composi ions, a ic con ol
da a, ehicula kinema ics, e c.
1
VRU-speci ic aspec s a e p esen in all elemen s, such as beha iou al modelling, a ic simula ion and oad sa e y
analysis (especially wi h su oga e c ash measu es). Howe e , he VRU pa icula i ies, in da a and eal condi ions,
will be handled in e nally as hey a e encoun e ed wi hin each o he elemen s and a e no discussed he e.
phoebe-p ojec .eu
Copy igh © by PHOEBE
22
Figu e 2-1 The PHOEBE F amewo k lowcha
phoebe-p ojec .eu
Copy igh © by PHOEBE
23
I should be no ed ha o e lap exis s ac oss he da a needs o each elemen . The simila i ies in da a ypes
can o en se e o alida ion ac oss da a sou ces, and e i ica ion ha he same a eas a e desc ibed wi h
he same signs and magni udes ac oss da abases (e.g. o geome ic da a). In la e s ages o he
F amewo k, he sou ces ha a e deemed as mos accu a e and mos comple e will be selec ed and used
exclusi ely, while he o he s will be e ained o compa ison and e e ence pu poses in da a ames.
The ollowing sub-sec ions u he p o ide a summa y o he indi idual s eps wi hin he amewo k, he
componen s in ol ed, and hei in e ac ion wi h each o he while sec ions 3 and 4 del e in o mo e
heo e ical de ails on he models wi hin each componen as well as he inpu -ou pu low om one o he
o he componen wi hin he amewo k. The de ails o how he models will be implemen ed in he
espec i e ools (i.e. AIMSUN mic osimula ion and iRAP oad sa e y assessmen ) will be co e ed in WP3.
Finally, he PHOEBE p edic i e sa e y assessmen amewo k will measu e he impac s (KPIs) o e he
p esen o a baseline yea and in u u e a e implemen ing he changes, o e some ime ame (e.g. in 5,
10,15 yea s, o longe ime pe iod), o accoun o he changes in he ne wo k h ough in oduc ion o
in e en ions, egula ions, beha iou s, o new echnologies e c., and e en i he e a e no such changes,
o simply accoun o he na u ally occu ing changes in u u e such as h ough he popula ion g ow h o
a el pa e ns.
His o ical ends such as popula ion g ow h, ou ism g ow h, p e-exis ing ends in mode choice and/o
a el pa e ns, land use changes, e c. can be used o baseline o ecas s. Fo p ojec ions in any u u e
baseline scena io, an annual mul iplie ac o s based on his o ical da a o es ima es will be equi ed.
Conside a ion mus be gi en o any skewness in he da a due o any easons e.g. impac o he global
pandemic. The p ecise o ecas ing o planning ho izon will be decided la e in o hcoming wo k packages
(WP3 and WP4).
S ep 1 – Baseline calcula ions
The pu pose o S ep 1 in he lowcha is o accu a ely cap u e he exis ing baseline scena io ( he cu en
si ua ion). The baseline scena io is he e e ence poin agains which he impac o any u u e changes
can be compa ed (Figu e 2-2).
S ep 1 Baseline scena io (wi hou any changes)
Figu e 2-2 The PHOEBE F amewo k lowcha – S ep 1
phoebe-p ojec .eu
Copy igh © by PHOEBE
24
In his s ep, da a passes be ween each o he elemen s as ollows:
• Road sa e y assessmen p o ides s a ic oad sa e y c ash isk and c ash occu ence
es ima es by segmen o he mode choice and beha iou al modelling.
• Mode choice modelling (incl. induced demand) p o ides mode choice p obabili ies and
beha iou al modelling p o ides p obabili ies o speci ic beha iou occu ence o a ic
mic osimula ion.
• A e he a ic mic osimula ion uns, he a ic mic osimula ion p o ides simula ed
ajec o ies o he oad use s which ( h ough pos -p ocessing in a su oga e sa e y
assessmen model) p o ides numbe o con lic s o he beha iou al modelling. I also p o ides
low and speed da a o oad sa e y assessmen .
• The beha iou al modelling p o ides he a al and se e e inju y (FSI) calib a ion pa ame e s
o oad sa e y assessmen o calcula ing he upda ed sa e y a ings (S a Ra ings) o a iable
speed and lows (“dynamic S a Ra ings”).
• Road sa e y assessmen p o ides upda ed S a Ra ings pe segmen o he a ic
mic osimula ion o isualisa ion o dynamic isk.
• The co e ou pu s om S ep 1 o he PHOEBE F amewo k a e he KPIs ecei ed om he
a ic mic osimula ion and oad sa e y assessmen o be used o scena io compa ison.
S ep 2 – scena io change calcula ions
In S ep 2, oad sa e y (o o he anspo -o ien ed) changes a e assumed o ha e been implemen ed in
he s udy a eas. S ep 2 ecei es inpu s compa able o hose o S ep 1 o conduc he necessa y
calcula ions o he KPIs a e aking he speci ic changes (e.g. in as uc u e, egula o y, beha iou al, e c.)
in o accoun (Figu e 2-3).
S ep 2 New scena io (wi h changes)
Figu e 2-3 The PHOEBE F amewo k lowcha – S ep 2
phoebe-p ojec .eu
Copy igh © by PHOEBE
31
When compa ed o he baseline scena io, he mode choices o di e en scena ios p oduce an impo an
indica o called mode shi . The ele an inpu s and ou pu s (da a) o he mode choice modelling p ocess
a e lis ed and desc ibed below (Depa men o T anspo , 2014; O úza & Willumsen, 2014).
2.2.2.1 Inpu da a
The demand o mode choice models a e sophis ica ed disc e e choice models (s a is ical models) ha
can p edic he mode choices and mode shi s. Th ee modelling s eps a e ca ied ou o de elop hese
models: model es ima ion, calib a ion, and alida ion (Depa men o T anspo , 2020a). Speci ic da a a e
used as inpu da a o de elop such disc e e choice models (Depa men o T anspo , 2020b; D. A. ;
Henshe & Bu on, 2008; O úza & Willumsen, 2014). This da a is e med inpu da a and a e ypically o
he ollowing ypes:
(Socio-)demog aphic da a:
Demog aphic o sociodemog aphic da a a e he mos c ucial inpu da ase in demand modelling. This da a
p o ides essen ial in o ma ion ega ding he demog aphy o he egion unde conside a ion, e.g.,
household s uc u es, employmen s a us, ip gene a ion a es, ca /o he modes o anspo owne ship
and a ailabili y ends, socioeconomic s uc u e, and license holding (Depa men o T anspo , 2020a),
as depic ed in he ollowing Figu e 2-6.
Figu e 2-6 Sociodemog aphic da a inpu equi emen s o demand models
Sou ce/s: This da a can be ob ained om he egional o na ional census da a, na ional a el su eys,
o specialised local household ( a el) su eys can be se up, especially o his pu pose.
Land-use da a
Land-use da a ies he popula ion da a o he geog aphical dimension and enables he calcula ion o he
ips spa ially o doubly (wo k, educa ion) and singly (shopping, leisu e) cons ained pu poses o he
conside ed pe son g oups.
Sou ce/s: This da a can be ob ained om he egional o na ional census, wo kplace s a is ics, and local
au ho i y planning da a o esiden ial, employee, e ail, and comme cial loo space by zone.
Ne wo k da a
Ne wo k da a e e s o da ase s encompassing de ails o he ne wo k o he anspo supply side. I
p o ides he u ili ies, such as ime, dis ance, e c., o he demand calcula ions. I also p o ides inpu s
abou an essen ial mode o anspo , e.g., equency and headways o public anspo , as well as oad
sa e y assessmen isk sco es.
Sou ce/s: This da a can be ob ained om Open S ee Map (OSM) and local public anspo se ice
p o ide s. O en, public anspo da a a e a ailable in he o m o GTFS da ase s oo. Road sa e y

phoebe-p ojec .eu
Copy igh © by PHOEBE
32
assessmen isk sco es (including p e-exis ing da a) can be de i ed om he exis ing oad sa e y
assessmen ools o any egion o in e es .
T a el o zonal des ina ions by Pu pose, Mode, and Time o Day (Pe iod)
These da ase s p o ide he ends conce ning he des ina ions and he pu pose o he ips, mode choices,
and he imes and cos s o he ips. These ends a e used o es ima e he mode choice models. Thus,
hese da ase s a e conside ed c i ical o es ima ing he mode choice models.
Sou ce/s: P incipally specialised su eys, including he s a ed p e e ence su eys and e ealed
p e e ence. I can also be collec ed om sou ces, e.g., oadside in e iews, mobile ne wo k da a,
au oma ic numbe pla e ecogni ion, and local/na ional/ egional household a el su eys. T a el dia ies
(including longi udinal a el dia ies) a e also used o ga he his da a.
Explaine no es
S a ed p e e ence su eys (SP)
The SP su ey ( he Con ingen Valua ion) is a echnique o es ablishing alua ions be ween
al e na i es. The subjec is asked o choose be ween policies, p oduc s, o se ices wi h desi able and
undesi able cha ac e is ics. In doing so, hey indi ec ly e eal which pa s a e mos impo an o hem.
SP su eys a e ins umen al in cases wi hou eal-wo ld da a o make conclusions. In he ollowing
s eps, de ailed s a ed p e e ence su eys will be designed and dissemina ed among he subjec s o
pilo ci ies (use cases) o collec he SP da a. The SP da a will acili a e he p ocesses o he sui able
demand model class choice and he es ima ion o he chosen models.
Re ealed p e e ence da a (RP)
The RP da a assumes consume s' pu chasing (demand) habi s e eal hei p e e ences. RP da a will
be ga he ed om se e al sou ces, e.g., he ac ual pu chase da a (e.g., icke pu chases) and
consump ion o demand da a (e.g., ac ual a el demand coun s). The RP da a will acili a e he demand
model choice, es ima ion, and alida ion o he chosen demand models (mode choice models).
(Longi udinal) T a el dia ies (TD)
Longi udinal a el dia ies a e eco ds o s udy and ack indi iduals' a el pa e ns and beha iou s
o e an ex ended pe iod. Unlike egula a el dia ies ha keep eco ds o a speci ic pe iod, i.e., a day
o a week, longi udinal a el dia ies collec in o ma ion o e se e al weeks, mon hs, o yea s.
Longi udinal a el dia ies a e a ype o e ealed p e e ence su ey. Thus, hese da a sou ces will
cap u e he ac ual beha iou and choices o he esponden s o speci ic con ex s.
T ip leng hs and imes
T ip leng hs and ip imes a e specialised da ase s used indi ec ly in he mode choice model de elopmen
p ocess. This da a is no used du ing he model es ima ion bu in he model alida ion p ocess. T ip leng hs
and imes a e o en used wi h hei spa ial, empo al, and pu pose con ex s, i.e., in conjunc ion wi h he
o igin-des ina ion, o igin-des ina ion imes, and pu pose o he ips.
Sou ce/s: The local/ egional/na ional household a el su eys a e he chie sou ces o hese da ase s.
In addi ion o hese su eys, (longi udinal) a el dia ies a e also conside ed ich da a sou ces.
Vehicle occupancies
Vehicle occupancies ac oss all he pe son g oups, pu pose, and mode segmen s a e no ewo hy da ase s
used in demand modelling p ocesses. These da ase s a e u ilised o es ima e he numbe o ehicles
mo ing in he ne wo k. In addi ion o he numbe o ehicles, hese da ase s a e also used o es ima e he
anspo pe o mance (subsequen ly he KPIs, e.g., emissions, consump ions, e c.) o he di e en modes
o anspo .
phoebe-p ojec .eu
Copy igh © by PHOEBE
33
Da a sou ces: The local/ egional/na ional household a el su eys a e he chie sou ces o hese
da ase s. In addi ion o hese su eys, (longi udinal) a el dia ies a e also conside ed.
Vehicle ope a ing cos s (gene alised cos s) and Value o Time (VoT)
The ehicle ope a ing cos s and he gene alised cos s o ehicle ope a ion o di e en pe son g oups and
pu poses o ips and he VoT a e used o es ima e he impedances a ising while making ips be ween
he o igin-des ina ion pai s in he demand modelling p ocess.
Sou ce/s: The ehicle ope a ing, gene alised cos s, and he VoT a e gene ally p o ided by local planning
and s a is ical au ho i ies. These alues can also be es ima ed om he " a el o zonal des ina ions by
Pu pose, Mode, and Time o Day (Pe iod)" da ase s. Table 2-1 lis s all he equi ed da a and hei p obable
sou ces o demand modelling.
Table 2-1 Demand modelling da a equi emen s and hei sou ces
N .
Main da ase
Sub da ase
To be used in
Sou ces
1
Popula ion
Household
s uc u e
Employmen
s a us
T ip gene a ion
a es
T ip gene a ion a e
calcula ions
Socioeconomic and
pe son g oup
ca ego isa ion
Mode spli dis ibu ion
Regional o na ional census
da a
Na ional a el su eys
Specialised local household
( a el) su eys
1.1
1.2
1.3
1.4
Socioeconomic
s uc u e
License holding
Ca s and o he
modes o
anspo
a ailabili y
1.5
1.6
2
Land-use da a
T ip gene a ion
T ip dis ibu ion
T a ic Analysis Zone
dema ca ion
Regional o na ional census
Wo kplace S a is ics
Local au ho i y planning
da a
3
Ne wo k da a
U ili y calcula ions
Inco po a ion o anspo
supply side in calcula ions
Open S ee Maps (OSM)
Local au ho i ies and public
anspo se ice p o ide s
4
T a el o zonal
des ina ions by
Pu pose, Mode,
and Time o
Day
T ip gene a ion
T ip dis ibu ion
Time pe iod choice
Asce aining ends
Specialised a el su eys
including s a ed p e e ence
su eys.
Re ealed p e e ences
T a el dia ies
5
T ip leng hs
and imes
Model calib a ion
Model alida ion
Specialised a el su eys
T a el dia ies
6
Vehicle
occupancies
Es ima ion o he numbe
o ehicles
Specialised a el su eys
phoebe-p ojec .eu
Copy igh © by PHOEBE
34
N .
Main da ase
Sub da ase
To be used in
Sou ces
Es ima ion o anspo
pe o mances
Es ima ion o KPIs, e.g.,
emissions, consump ions
e c.
T a el dia ies
7
Vehicle
ope a ing cos s
(gene alised
cos ) and Value
o ime (VoT)
Es ima ion o u ili ies and
disu ili ies.
Es ima ion o
impedances.
Local planning and
s a is ical au ho i ies
T a el su eys
T a el dia ies
2.2.2.2 Ou pu da a
The demand model gene a es speci ic ou pu da a. The main ou pu s a e desc ibed as ollows.
Es ima ed mode choice p obabili ies
Mode choice p obabili ies, as he ou pu s o a mode choice model, ep esen he likelihood o p obabili y
ha a g oup o indi iduals will choose o a speci ic pu pose, a speci ic anspo a ion mode (such as
walking, bike, ca , o public ansi ) among he a ailable op ions o a gi en ip, based on hei
cha ac e is ics and he a ibu es o he modes. These p obabili ies add up o 1 (Figu e 2-7), indica ing
ha one o he modes will be chosen o he ip.
These p obabili ies a e essen ial in unde s anding and p edic ing mode choices in a el demand
modelling. A highe p obabili y o a pa icula mode sugges s ha he model p edic s a g ea e likelihood
ha indi iduals will choose ha mode o he gi en ip, conside ing he speci ied ac o s and a ibu es.
Figu e 2-7 Mode choice p obabili ies
phoebe-p ojec .eu
Copy igh © by PHOEBE
35
Es ima ed mode spli s
Indi idual mode choice p obabili ies associa ed wi h speci ic modes o anspo can be in e p e ed as
mode spli amongs all he a ailable modes. Mode spli can be u he used o gene a e mode shi .
Mode shi s
A change in he inpu a iables o he mode choice models, e.g., a el ime, cos , e c., will de elop a new
mode spli o mode sha e; hus, he baseline mode sha e can be compa ed wi h he new sha e o calcula e
mode shi .
Es ima ed OD mo emen s ( ips) ma ices including he induced a el demand
The mode sha e o spli can be used o gene a e he OD mo emen s in e ms o he ips ca ego ised by
modes o anspo . These ip ma ices can be de i ed by mul iplying he OD ip ma ices ob ained om
he ip dis ibu ion calcula ions wi h he mode choice p obabili ies. In addi ion o he egula OD ma ices,
he (disc e e choice) model es ima ed o o ecas ing he induced demand will also gene a e he OD
ma ices o he induced a el demand in e ms o he numbe o new ips added due o he changes
in oduced (Kalliga, 2021).
Demand elas ici ies
Fo a gi en scena io (i.e., a di e en se o inpu a iables compa ed o he baseline si ua ion), he ou pu s
o he mode choice models can be used o gene a e demand elas ici ies. These demand elas ici ies help
unde s and he changes o he dependen a iables unde he changing magni ude o he independen o
explana o y a iables.
𝑒 =log(𝑇1)−log(𝑇0)
log(𝐶1)−log(𝐶0)
(2-1)
Whe e,
𝑒 is he elas ici y,
𝑇 is he demand wi h supe sc ip s 0 and 1 indica ing he alues be o e and a e he change o cos ,
𝐶 is he cos wi h supe sc ip s 0 and 1 indica ing he alues be o e and a e cos s.
Theo e ical basis
In PHOEBE, he de elopmen o sui able disc e e choice models is en isioned o la ge-scale o ecas ing
o he mode shi (M. E. Ben-Aki a & Le man, 2006). The es ima ion o mo e complica ed models, such
as hyb id class models (models wi h la en a iables) and la en class choice models a e also en isioned
o o ecas speci ic en i ies such as expe ience and e ec s o ex e nal ac o s like eal- ime in o ma ion
lows h ough he in e ne and apps. These models will be de eloped as pe he h ee s eps (es ima ed,
calib a ed, and alida ed). The ele an pa s o such models will be conside ed in he p ojec 's disc e e
choice model, which will be planned o la ge-scale o ecas ing.
The p ima y aims o he choice model de elopmen a e lis ed below:
1 Fo ecas ing he la ge-scale mode choice and mode shi by di e en pe son g oups (Ben-
Aki a & Le man, 2006; Domencich & McFadden, 1975);
2 Inco po a ing a ious modes o anspo in he model (Domencich & McFadden, 1975);
3 Inco po a ing speci ic in angible p ope ies in he o m o inpu (la en ) a iables in he model
(Mo oaki & Daziano, 2015); and
phoebe-p ojec .eu
Copy igh © by PHOEBE
36
4 Inco po a ing he la en class choice models o cap u e obse ed and unobse ed
he e ogenei y. These models can ela e a se o obse ed disc e e mul i a ia e a iables o a
se o la en a iables ( a iables ha a e no di ec ly obse ed bu a e a he in e ed) (Ga cía-
Mele o e al., 2022; Lee e al., 2003).
Mos o he modelling p ocess depends on he inpu da a (explained in he p eceding sec ions) and he
SP/RP su eys. The inal model ype and modelling p ocesses will be hus inalised a e he comple ion
o he inpu da a collec ion, including he comple ion o he SP/RP su eys. A gene alised model
ep esen a ion is gi en below:
Explaine no es
Gene alised disc e e choice model o o ecas ing mode choice (Ben-Aki a & Le man, 2006)
A disc e e choice model wo ks wi hin he ounda ional assump ion o a ional choices he indi idual
(g oups) makes when con on ed wi h a disc e e se o al e na i es o maximise hei bene i o u ili y.
The disc e e choice e alua es he inpu o explana o y a iables conce ning he cha ac e is ics o he
use g oups and he a ailable al e na i es. The choice is exp essed in e ms o p obabili y. The
gene alised o m o he disc e e choice model is p esen ed below:
𝑃𝑖𝑗 =𝑓(𝑆𝑖,𝑋𝑗)
(2-2)
Whe e,
𝑃𝑖𝑗 selec ion o choice p obabili y o mode (al e na i e) 𝑗 by a pe son g oup 𝑖.
𝑓 Decision unc ion (disc e e choice unc ion)
𝑆𝑖 se o cha ac e is ics o he pe son g oup 𝑖, e.g., wo ke s, s uden s e c.
𝑋𝑗 se o cha ac e is ics o he al e na i es (modes o anspo , e.g., ca , bike) 𝑗, e.g., cos , ime.e c.
The se o cha ac e is ics 𝑺𝒊 and 𝑿𝒋 (Ben-Aki a & Le man, 2006)
The se o cha ac e is ics, namely 𝑆𝑖 and 𝑋𝑗, can be encapsula ed unde a single e minology commonly
known as u ili y (𝑈). I can be de ined as he a ac i eness o each al e na i e ha he decision-make
will enjoy i hey choose i . A u ili y has wo componen s, namely:
● Sys ema ic o measu ed u ili y, also known as de e minis ic u ili y (𝑉).
● S ochas ic/ andom componen (𝜀).
As shown below, he o al u ili y is exp essed by adding hese wo u ili y ypes.
𝑈 =𝑉+𝜀
(2-3)
The de e minis ic u ili y (𝑉) con ains he explana o y a iables ha can be measu ed, e.g., Ride ime,
a e, and numbe o ans e s. The coe icien s (𝛽) o he pa ame e s con ol he sensi i i y o he
explana o y a iables. A gene ic u ili y (de e minis ic) equa ion is gi en below:
𝑉𝑗𝑞 =𝛼𝑗𝑞 +𝛽𝑗1.𝑥𝑞1 +𝛽𝑗2.𝑥𝑞2 +⋯ = 𝛼𝑗𝑞 +∑𝛽𝑗𝑘.𝑥𝑞𝑘𝑘
(2-4)
𝑉𝑆𝑡𝑢𝑑𝑒𝑛𝑡𝑠.𝑃𝑇 =1.0+(−0.1).𝑅𝑖𝑑𝑒𝑇𝑖𝑚𝑒+(−0.75).𝐹𝑎𝑟𝑒+⋯
(2-5)
Whe e,
𝑉𝑗𝑞 de e minis ic u ili y alue
𝛼𝑗𝑞 cons an o al e na i e 𝑞 o g oup 𝑗
𝛽𝑗𝑘 pa ame e de e mining impo ance o a ibu e 𝑘 o g oup 𝑗
𝑥𝑞𝑘 alue o a ibu e 𝑘 o he al e na i e 𝑞
The decision unc ion 𝒇 (Ben-Aki a & Le man, 2006)
The decision unc ion 𝑓 is also known as he model o choice model class. The ype o 𝑓 depends upon
he dis ibu ion o he s ochas ical u ili y 𝜀, e.g., i he 𝜀 is assumed o be independen and iden ically

phoebe-p ojec .eu
Copy igh © by PHOEBE
37
dis ibu ed as an ex eme alue dis ibu ion 𝐸𝑉(0,𝜇) (i.i.d. Ex eme Value) 𝑓 is called as logi model. A
logi model can be gi en by:
𝑃(𝑖|𝐶𝑛)=𝑒𝜇.𝑉𝑖𝑛
∑𝑒𝜇𝑉𝑗𝑛
𝑗𝜖𝐶𝑛
(2-6)
whe e,
𝐶𝑛 a choice se o al e na i es wi h 𝑛 al e na i es
𝜇 scaling ac o
𝑉𝑖𝑛 de e minis ic u ili y o al e na i e 𝑖 om 𝐶𝑛.
The la en a iables (Lee e al., 2003)
Some phenomena canno be measu ed; hus, indi ec measu emen s a e used o in oduce o
inco po a e hem in a choice model. O en, psychome ics is used o sol e hese p oblems. An example
o indi ec measu emen s o la en concep is, ease o a elling wi h a pa icula mode o anspo wi h
child en. Such a iables a e called la en a iables and a e used o explain he u ili y unc ion. Thus, he
in angible pe cep ion can be inco po a ed in o he choice modelling amewo k. These ypes o choice
models a e o en e med hyb id choice models.
Model es ing
The demand models' es ing sequence o he PHOEBE use cases is desc ibed below:
• Da a collec ion: Good da a sou ce is he mos c i ical componen o he model es ing. The
da a come om he SP/RP su eys and o he sou ces, e.g., census da a, land-use da a, a el
dia ies, a ic coun s, VehKM, and T ipKM da a.
• Model choice: A sui able choice model will be chosen based on he da a collec ed and he
equi emen s o he use cases.
• Model es ima ion: The inpu a iables and o he componen s o he chosen model will be
inalised, and subsequen ly, he model will be es ima ed wi h he da a om he su eys and
he a el dia ies.
• Model calib a ion and alida ion: The es ima ed model will hen be alida ed, calib a ed,
and e- alida ed wi h he help o eal-wo ld coun da a. This s ep will p o ide a demand model
o simula ing he cu en o baseline scena io. This model will p oduce a baseline mode
sha e, which will be used as a e e ence o gene a e he mode shi o he planned scena io.
• Fo ecas o he use cases scena ios: Acco ding o he pa icula use case scena io, he
inpu a iables o he baseline demand model will change, p oducing a o ecas ed mode
sha e. This mode sha e will hen be compa ed wi h he baseline mode sha e o gene a e he
mode shi o he succeeding p ocesses.
2.3 T a ic mic osimula ion
T a ic mic osimula ion in PHOEBE is a key enable o combining he inpu om di e en componen s o
he PHOEBE amewo k o simula e and o ecas he sa e y impac s o changes and u u e scena ios
ac oss u ban anspo ne wo ks. Tes ing in a i ual en i onmen speeds up es ing, wi hou he delays o
eal-wo ld implemen a ions and educe cos , bo h social and mone a y, and o allow ep oducibili y and
con ol o e expe imen s ha would no be possible in he eal wo ld.
Mic osimula ion enables he ep oduc ion o he ajec o ies o each single oad use (pedes ians,
bicycles, e-scoo e s, ca s, buses, and o he s) un eiling hei po en ial unsa e in e ac ions. Le e aging
om he powe and capabili ies o mic osimula ion o in eg a e o he di e en models and ools, he
PHOEBE amewo k combines he mic osimula ion componen wi h mode choice models, human ac o
phoebe-p ojec .eu
Copy igh © by PHOEBE
38
based beha iou al models, and oad sa e y assessmen models, o ep oducing mo e ealis ic
in e ac ions be ween di e en oad use s and consequen ly imp o ing he assessmen amewo k o oad
sa e y in a p ospec i e manne h ough gi ing a me hodological amewo k which can be expanded o
he inclusion o eme ging and u u e mobili y op ions.
T a ic mic osimula ion modelling s a s wi h a anspo oad g aph which in ol es de eloping abs ac
ep esen a ion o anspo ne wo ks ha consis s o links and nodes. These objec s ha e hei own
a ibu es ha make each o hem unique. When mul iple links and nodes a e linked oge he , a
ep esen a ion o he mobili y ne wo k is achie ed. The ollowing a e he gene al equi emen s o build a
mic osimula ion om sc a ch. This s a s wi h a ne wo k ep esen a ion o he a eas o simula e.
A ne wo k g aph is ypically impo ed om Open S ee Map (OSM) and co esponds o 3-dimensional
abs ac ep esen a ion o he anspo oad ne wo k. This means ha ing all s a iona y elemen s wi hin
he oad ne wo k, enabling he accu a e simula ion o ehicula mo emen s, such as shown in Figu e 2-8
(a).
The ne wo k model pe ains o he depic ion o a ne wo k employed by mic oscopic simula ion-d i en
models. This equi es mo e in ica e da a, including elemen s such as a ic managemen s a egies,
pedes ian walkways, nodes go e ned by a ic signals, he ype o in e sec ion con ol, capaci ies, a el
demand, and so o h. An example o a ne wo k model is illus a ed in Figu e2-8 (b).
a)
b)
Figu e 2-8 Ne wo k g aph s. ne wo k model a) Ne wo k g aph in OSM b) Ne wo k model in Aimsun Nex wi h a ic
ligh s, amway and bus schedules, p io i ies a in e sec ions, e c.
A e he ne wo k is buil , he p ocess o calib a ing he supply and demand ac o s is ini ia ed. This is o
ensu e ha he simula ion en i onmen ep esen s eal a ic condi ions and can be used o deploy and
phoebe-p ojec .eu
Copy igh © by PHOEBE
39
assess he changes, wi hin he PHOEBE F amewo k. The da a which a e ypically equi ed o pe o m
his ask include:
• Ne wo k geome y – Da a desc ibing he geome ical aspec s o he use case a ea.
• Demand – Da a, exp essed in numbe o ips be ween o igin and des ina ion pe
anspo a ion mode.
• T a ic con ol – Cha ac e isa ion o a ic con ol elemen s like a ic ligh s.
• T ansi ope a ion – Da a cha ac e ising he ope a ion o public anspo , such as ansi
ou es, s ops, and schedule.
• Ne wo k a ic s a e and pe o mance – Da a cha ac e ising how a ic beha es along
oadway elemen s, such as olumes, speeds, a el imes, loca ion o bo lenecks, e c. Da a
should be collec ed o all c i ical ime pe iods being s udied, e.g., AM peak, Midday peak, PM
peak, weekend.
P ocesses
The mic osimula ion componen wi hin he PHOEBE F amewo k links o o he componen s in he
amewo k, as shown in Figu e 2-9 h ough exchange o se e al inpu s and ou pu s. To be mo e speci ic,
he mode choice modelling componen p o ides OD ma ices pe mode as inpu o he mic osimula ion
componen . I also akes he human ac o based beha iou al models inpu om he beha iou al modelling
componen o he amewo k o ep esen he d i ing and walking beha iou s o mo o ised and non-
mo o ised oad use s in he simula ion model. This means ha mo e ealis ic d i ing o walking s yles a e
modelled in o he mic osimula ion en i onmen .
Wi h hose inpu s mic osimula ion can p oduce esul s on a iables o in e es (KPIs) such as a ic lows
pe oad use ype and oad segmen , a el imes, emissions, and many o he s. The ou pu om
mic osimula ion uns also include indi idual ehicle / pedes ian ajec o ies o u he p ocessing h ough
a su oga e sa e y assessmen model o iden i y con lic s, which will be used by he beha iou al modelling
componen o es ima ing he FSI calib a ion pa ame e s o be used in he oad sa e y assessmen
componen . In addi ion, he ele an a iables (such as speeds and lows, e c.) om he mic osimula ion
ou pu will also be u ilised by he oad sa e y assessmen componen o calcula e he upda ed isk sco es
(dynamic) which can hen be in eg a ed wi hin he simula ion model bu only o isualisa ion pu pose.
Figu e 2-9 In e ac ion be ween a ic mic osimula ion and o he componen s
phoebe-p ojec .eu
Copy igh © by PHOEBE
40
Models in a ic simula ion
T a ic mic osimula ion can be o e whelmingly complex and he e o e equi es a so wa e package ha
comes wi h all he necessa y ea u es. In gene al, he models in ol ed a e (i) he ne wo k ep esen a ion,
which by i sel is a model, (ii) he pa h inding and sho es pa h models, (iii) demand loading models, (i )
s ochas ic and andom choice models and ( ) he models o mo e ehicles o pedes ians. The las models
a e he ones o ha e he g ea es impac on sa e y as hey go e n he ac ions o ehicles and pedes ians.
Pedes ians a e ypically modelled by a social o ce model. On he o he hand, ehicula mo emen s a e
go e ned by kinema ics models commonly known as ehicle beha iou models wi hin mic osimula ion
modelling domain. Fo example, ypically he ehicles mo e in he longi udinal axis acco ding o a ca -
ollowing model, and h ough he ans e sal axis acco ding o a lane-changing model. In some occasions,
i is possible o ha e a 2-D model ha combines bo h o ha e non-lane-based a ic dynamics.
Addi ionally, he e a e he gap accep ance models o in e sec ions, which go e n he beha iou o
ehicles on unsignalised in e sec ions (no a ic ligh s) and pedes ian decision making a pedes ian
c ossings. Figu e 2-10 illus a es a gene al modelling low o a ic mic osimula ion along wi h an ex e nal
con ol.
Figu e 2-10 Gene al modelling low o a ic mic osimula ion enabling ex e nal con ol
Wi hin he con ex o PHOEBE p ojec , which in addi ion o combining o he models and ools wi hin he
amewo k also in ol es inco po a ing human ac o based beha iou al models in mic osimula ion models,
some ex e nal applica ion p og amming in e aces (APIs) will be used as pe he needs, o ins ance, o
e lec he d i e s’ cha ac e is ics (such age g oup, gende e c.) and speeding beha iou , based on he
human ac o based beha iou al models (as explained in sec ion 2.4) on a segmen in he model. Simila
app oach will be applied o modelling he unsa e and/o non-complying beha iou s o mo o ised and non-
mo o ised oad use s, p io i ised wi hin he PHOEBE p ojec .
Da a
T a ic mic osimula ion can gene a e as amoun s o da a, as da a can be ga he ed o each single
simula ed objec a each simula ion s ep. Typically o a oid being o e whelmed wi h such as quan i ies
o da a, agg ega ed KPIs a e s o ed and shown o he end use .
phoebe-p ojec .eu
Copy igh © by PHOEBE
47
2 Obse ed lows: Mo o cycles, bicycles and pedes ians along o c ossing obse ed lows a e
pa o his g oup. Obse ed lows a e no di ec ly used in S a Ra ings. They a e used o
alida e and check oad use low da a.
3 Speed limi : Speed limi is collec ed o gene al a ic. The speed limi o mo o cycles and
hea y ehicles may also be collec ed, bu hese a e used o analy ical pu poses only (and
a e no au oma ically used o S a Ra ing calcula ions). Majo di e ences in ope a ing speed
o speed limi a e eco ded, as well as speed managemen and a ic calming ea u es.
4 Mid-block a ibu es: Road ea u e a ibu es ha do no ela e o in e sec ions. Includes
in o ma ion abou he numbe o lanes, lane wid h, cu a u e ( adio and quali y), median ype,
g ade, sigh dis ance, p esence o ligh ing and delinea ion mus be eco ded. Also includes
pa emen quali y a ibu es (skid esis ance and oad condi ion) as well as he p esence o
ehicle pa king, se ice oads, and oad wo ks along he sec ion. These a ibu es can be
co ela ed wi h links in he a ic simula ion ypes o ne wo ks. Upg ade cos s a e also
collec ed o suppo sa e oad in es men plans (i.e., i is no equi ed o S a Ra ings o FSI
Es ima ion models).
5 Roadside a ibu es: Roadside objec s and hei dis ance om he oad. Includes oad
shoulde wid h and i he e a e umble s ipes p esen in he edge lines.
6 In e sec ion a ibu es: Fea u es ela ed o in e sec ions, such in e sec ion ype (including
numbe o legs, signalisa ion, u ning lanes), in e sec ion quali y, connec ing oad lows, and
p esence o channelisa ion and p ope y accesses.
7 VRU ea u es: Fea u es ela ed o pedes ians, bicyclis s and mo o cyclis s, such as
sidewalks, bicycle and mo o cycle acili ies, and oad c ossings and school zones. Includes
a ea ype which in luences isk ac o s and land use which is used o help p edic he p esence
o VRUs.
2.5.1.2 Ope a ional pa ame e s
The ope a ion pa ame e s e e o speed and lows o he di e en oad use s. This includes ehicle
a e age annual daily a ic (AADT) low, pe cen age o mo o cycles, pedes ian and bicycle peak hou
lows, and ope a ing speeds (85 h pe cen ile and mean). Mo e in o ma ion on he ope a ion pa ame e s
can be ound in he iRAP S a Ra ing and In es men Plan Manual.
Speeds
Speed is an essen ial a ibu e wi hin he iRAP p o ocols. The iRAP S a Ra ing models use he 85 h
pe cen ile ope a ing speed and speed limi o de e mining S a Ra ings, while mean ope a ing speed is
used o gene a e FSI Es ima ions.
T adi ionally, S a Ra ings a e based on he la ge o he speed limi and he ope a ing speed (85 h
pe cen ile speed - he speed a o below which 85 pe cen o all ehicles a e obse ed o a el unde
ee- lowing condi ions pas a moni o ed poin ). The model enhancemen as pa o PHOEBE will p io i ise
he ope a ing speeds o allow speed luc ua ion along he day.
Speed isk ac o s also in o m coun e measu e igge s and economic bene i s. The speed isk ac o s
wo k as mul iplie s o e e y c ash ype included in he model, highly impac ing he S a Ra ing Sco e.
Flows
The S a Ra ing models use di e en low pa ame e s. They a e:
Vehicles lows: Vehicle lows a e ep esen ed by he Annual a e age daily a ic (AADT). The AADT is
used in he calcula ion o S a Ra ings and o es ima e he numbe o a ali ies o each 100m segmen o
oad. Highe a ic lows inc ease a oad use ’s exposu e o ce ain c ash ypes, such as a head-on

phoebe-p ojec .eu
Copy igh © by PHOEBE
48
collision. Gene ally, a la ge AADT will esul in a lowe S a Ra ing and a highe es ima ed numbe o
a ali ies o a 100m segmen o oad. Mo o cyclis s and hea y ehicles a e coun ed as he pe cen age o
hose ypes o ehicles in he daily lows. A mo o cyclis S a Ra ing will only be calcula ed i mo o cyclis s
a e p esen . I he mo o cycle % is ze o, hen no mo o cyclis S a Ra ing will be p oduced and he es ima e
o mo o cyclis a ali ies will be ze o.
Bicyclis s low: Conside s all bicycles, including con en ional bicycles and e-bikes, passing h ough he
segmen in he peak hou . Bicycle low is used o es ima e he numbe o bicyclis a ali ies o each 100m
segmen o oad. A bicyclis S a Ra ing will only be calcula ed i bicyclis s a e p esen . I he bicyclis peak
hou low is ze o, hen no bicyclis S a Ra ing will be p oduced, and he es ima e o bicyclis a ali ies will
be ze o. Ideally his da a will be supplied by he oad au ho i y o o he ele an o ganisa ion. In
ci cums ances whe e eliable low da a a e no a ailable es ima es mus be made. The p ocess o
es ima ing bicycle lows should ollow a simila me hodology as o pedes ian lows.
Pedes ian lows: Pedes ian peak hou lows a e eco ded indi idually o lows along he d i e -side,
along he passenge -side and ac oss he oad. Changes in pedes ian lows do no a ec he S a Ra ings,
hough la ge pedes ian lows will esul in a lowe and highe es ima ed numbe o pedes ian a ali ies.
I pedes ian low ac oss and along he oad is ze o, hen no co esponding S a Ra ing will be p oduced
and he es ima e o pedes ian a ali ies will be ze o. Ideally, his da a will be supplied by he oad au ho i y.
In ci cums ances whe e eliable low da a is no a ailable, es ima es mus be made. The inal peak hou
pedes ian lows used in he analysis ul ima ely need o e lec he speci ic oad en i onmen .
Rela ionships be ween lows and oad a ibu es can a y signi ican ly be ween coun ies and egions. I
is he e o e good p ac ice o de elop a me hodology o es ima ing lows ha e lec s he speci ic con ex
o he assessmen .
2.5.1.3 C ash da a
Numbe s o a ali ies by oad use and c ash ype a e used o calib a e he da ase and p oduce he FSI
Es ima ion and In es men Plan epo s in ViDA (The iRAP compu a ional pla o m o s a a ing and FSI
es ima ion models). Calib a ion ac o s a e used o ensu e ha he o al es ima ed numbe o FSI on he
ne wo k is equal o he ac ual numbe o FSI on ha ne wo k.
Ideally, c ash da a o a 3-yea pe iod will be supplied by he oad au ho i y o ano he ele an agency,
such as police. The a ali ies mus be dis ibu ed in o use g oups and c ash ypes and inpu ed as numbe
o a ali ies o pe cen ages o he o al a ali ies. Table 2-2 displays he use g oups and he ypes o
c ashes o which he in o ma ion is needed.
Table 2-2 Use g oups and c ash ypes in ques ion
Use g oups
C ash ypes
Ca occupan s
Mo o cyclis s
Pedes ians
Bicyclis s
Mic o mobili y (e.g. e-scoo e s)
Run-o LOC d i e -side
Run-o LOC passenge -side
Head-on LOC
Head-on o e aking
In e sec ion
P ope y access
Along
C ossing in e sec ed oad
C ossing inspec ed oad
O he
*LOC: Loss o Con ol
phoebe-p ojec .eu
Copy igh © by PHOEBE
49
The CycleRAP model will also be used in selec ed use cases whe e cyclis s and o he Ligh Mobili y
Vehicle (LMV) use s a e he p ima y ocus. The da a collec ed o CycleRAP is di e en o ha desc ibed
abo e since he assessmen s a e cen e ed on acili ies bicyclis s and LMV a e using which may no
necessa ily cap u e om he oad ne wo k as is in he S a Ra ing models.
2
Howe e , he wo may be
used oge he . De ails and speci ica ions o he CycleRAP da a can be ound a he CycleRAP Use
Guide.
The ou pu da a he S a Ra ing and FSI Es ima ion models includes:
• C ash ype sco es pe oad use g oup ( ehicle occupan , pedes ian, bicyclis , mo o cyclis );
• S a Ra ing Sco es (SRS) and S a Ra ings pe oad use g oup;
• FSI Es ima ions pe oad use g oup and o all oad use s; and
• Fo CycleRAP, isk sco es by c ash ype ( ehicle-bicycle/LMV; bicycle/LMV- bicycle/LMV;
bicycle/LMV-pedes ian; single bicycle/LMV) and o al isk sco e (combined).
The ull lis o S a Ra ing and FSI Es ima ion ou pu da a is p o ided in Table 2-3.
Table 2-3 S a Ra ing and FSI Es ima ions model ou pu da a
Use s
Risk Sco es
Fa ali y es ima ions
S a Ra ings
Vehicle occupan s
Vehicle SRS Run-O LOC
D i e -Side
Vehicle SRS Run-O LOC
Passenge -Side
Vehicle SRS Head-On LOC
Vehicle SRS Head-On
O e aking
Vehicle SRS In e sec ion
Vehicle SRS P ope y Access
Vehicle SRS To al
Vehicle SRS To al Smoo hed
Vehicle Occupan FSI Es ima ion Run-O
LOC D i e -side pe km pe yea
Vehicle Occupan FSI Es ima ion Run-O
LOC Passenge -side pe km pe yea
Vehicle Occupan FSI Es ima ion Head-
On LOC pe km pe yea
Vehicle Occupan FSI Es ima ion Head-
On O e aking pe km pe yea
Vehicle Occupan FSI Es ima ion
In e sec ion pe km pe yea
Vehicle Occupan FSI Es ima ion P ope y
Access pe km pe yea
Vehicle Occupan FSI Es ima ion To al pe
km pe yea
Vehicle S a
Ra ing Raw
Vehicle S a
Ra ing
Smoo hed
Mo o cyclis s
Mo o cyclis SRS Run-O
LOC D i e -Side
Mo o cyclis SRS Run-O
Passenge -Side
Mo o cyclis SRS Head-On
LOC
Mo o cyclis FSI Es ima ion Run-O LOC
D i e -side pe km pe yea
Mo o cyclis FSI Es ima ion Run-O LOC
Passenge -side pe km pe yea
Mo o cyclis FSI Es ima ion Head-On LOC
pe km pe yea
Mo o cyclis S a
Ra ing Raw
Mo o cyclis S a
Ra ing
Smoo hed
2
Like S a Ra ing da a, CycleRAP da a includes acili y- ela ed a ibu es and ope a ional a ibu es.
phoebe-p ojec .eu
Copy igh © by PHOEBE
50
Use s
Risk Sco es
Fa ali y es ima ions
S a Ra ings
Mo o cyclis s (con .)
Mo o cyclis SRS Head-On
O e aking
Mo o cyclis SRS In e sec ion
Mo o cyclis SRS P ope y-
Access
Mo o cyclis SRS Along
Mo o cyclis SRS To al
Mo o cyclis SRS To al
Smoo hed
Mo o cyclis FSI Es ima ion Head-On
O e aking pe km pe yea
Mo o cyclis FSI Es ima ion In e sec ion
pe km pe yea
Mo o cyclis FSI Es ima ion P ope y
Access pe km pe yea
Mo o cyclis FSI Es ima ion Along pe km
pe yea
Mo o cyclis FSI Es ima ion To al pe km
pe yea
Pedes ians
Pedes ian SRS Along
Pedes ian SRS C ossing
In e sec ing Road
Pedes ian SRS C ossing
Inspec ed Road
Pedes ian SRS To al
Pedes ian SRS To al
Smoo hed
Pedes ian FSI Es ima ion Along D i e -
side pe km pe yea
Pedes ian FSI Es ima ion Along
Passenge -side pe km pe yea
Pedes ian FSI Es ima ion C ossing Side
Road pe km pe yea
Pedes ian FSI Es ima ion C ossing
Th ough Road pe km pe yea
Pedes ian FSI Es ima ion To al pe km
pe yea
Pedes ian S a
Ra ing Raw
Pedes ian S a
Ra ing
Smoo hed
Bicyclis s
Bicyclis SRS Along
Bicyclis SRS In e sec ion
Bicyclis SRS Run-O
Bicyclis SRS To al
Bicyclis SRS To al Smoo hed
Bicyclis FSI Es ima ion Along pe km pe
yea
Bicyclis FSI Es ima ion In e sec ion pe
km pe yea
Bicyclis FSI Es ima ion Run-O pe km
pe yea
Bicyclis FSI Es ima ion To al pe km pe
yea
Bicyclis S a
Ra ing Raw
Bicyclis S a
Ra ing
Smoo hed
**LOC: Loss o Con ol
SRS: S a Ra ing Sco es
Models and p ocesses
The S a Ra ing and FSI Es ima ion model use an e idence-based, s anda dised app oach. This allows
eliabili y and ans e abili y o he PHOEBE s uc u e o o he ci ies. The i s deli e able in PHOEBE
(D1.1 S a e o he a and end-use s need) desc ibed how he S a Ra ing, CycleRAP and he FSI
Es ima ions models wo k and i s cha ac e is ics ha a e ad an ageous o a oad sa e y amewo k.
As pa o he PHOEBE F amewo k, he S a Ra ings and he CycleRAP models b ing he in as uc u e
isk componen o measu e oad sa e y impac . S a Ra ings ep esen a isk o an indi idual use ype
( ehicle occupan , mo o cyclis , bicyclis , and pedes ian) and is he sum o indi idual c ash ype sco es
phoebe-p ojec .eu
Copy igh © by PHOEBE
51
( un-o - oad, heads-on, in e sec ions and access poin s, a elling along he oad and c ossing he oad).
They a e an objec i e measu e o :
i. How likely i is ha a oad c ash will occu o an indi idual oad use , based on he speed,
olume and physical ea u es o he oad; and
ii. The se e i y o he ou come when a c ash does occu .
The iRAP models conside he con ibu ion o se e al in as uc u e and ope a ional elemen s desc ibed
in he p e ious sec ion and hei con ibu ion o one o bo h p e iously discussed componen s. They a e
combined in a mul iplica i e model o de e mine o al isk, associa ed S a Ra ing, and FSI Es ima ions.
While he S a Ra ing model indica es he isk o an indi idual oad use , he FSI Es ima ion Model
conside s he S a Ra ing esul s o es ima e how many FSI c ashes will occu on he oad ne wo k,
conside ing exposu e ( oad use lows). The e o e, he FSIs will be he main indica o o sa e y o he
PHOEBE amewo k.
Figu e 2-14 Road Sa e y assessmen p ocess low wi hin PHOEBE amewo k
phoebe-p ojec .eu
Copy igh © by PHOEBE
52
The oad sa e y assessmen p ocess is de ailed in Figu e 2-14. F om he heo e ical pe spec i e, oad
sa e y assessmen ( ha is, he S a Ra ings, FSI Es ima ions and whe e ele an , CycleRAP) in e ac wi h
he o he elemen s wi hin he PHOEBE F amewo k in i e di e en s ages o he p ocess.
F om oad sa e y assessmen o o he componen s:
i. S a ic isk sco es a e inpu s o demand and beha iou al models;
ii. Dynamic isks sco es a e an inpu o a ic simula ion, howe e , only o isualisa ion; and
iii. C ash isk sco es and FSIs a e used o gene a e PHOEBE KPIs.
F om o he componen s o oad sa e y assessmen :
i. Dynamic isk sco es which use a iable speed and low da a om a ic simula ion; and
ii. Beha iou al calib a ion ac o s o he FSI Es ima ion model.
The oad sa e y assessmen low can s a in wo di e en ways depending on whe he he assessmen s
al eady exis o i i is an en i ely new assessmen . Bo h esul in ‘s a ic’ sco es, ha is, a isk sco e o a
poin in ime.
Fo new assessmen s, he p ocess s a s wi h da a ga he ing ( oad su ey and/o da a ga he ing om AI
and o he sou ces). A comp ehensi e da ase is compiled which align wi h ha desc ibed in sec ion 2.5.1.
The collec ed da a is hen used o calcula e he p elimina y c ash sco es o each indi idual oad segmen .
Whe e he e is an exis ing assessmen (and he e o e p e-exis ing da a as well as ou pu da a in he o m
o S a Ra ings and FSI Es ima ions), he p ocess s a s wi h con e ing he 100-segmen esul s o be
compa ible wi h he simula ion ne wo k (links and nodes). Exis ing o new da ase s hen se e as he
ounda ion o he baseline assessmen phase.
Explaine no es
Changing he da a s uc u e o e lec nodes and links
The S a Ra ing Sco e (SRS) and he FSI Es ima ion a e calcula ed o each 100m oad segmen o
ehicle occupan s, mo o cyclis s, pedes ians and bicyclis s. This means ha all oad a ibu e da a
comp ises 100m coding segmen s.
The 100m segmen s a e equally dis ibu ed om he assessmen 's s a ing poin de ined by he su ey.
When a new su ey s a s, a new e e ence s a ing poin is es ablished. Each ow o he ViDA co e
iles ep esen s one 100m segmen . The la i udes and longi udes p o ided e e o he beginning o
each segmen .
Howe e , simula ion models s uc u e he da a di e en ly: using nodes and links. Th ee app oaches
will be de eloped and es ed o PHOEBE:
(i) P ocedu e o ne wo ks wi h mos links wi h mo e han 100m
When mos o he links in he ne wo k a e mo e han 100m long, he da a coding can s ill be done pe
100m segmen , and he da a poin will be spa ially joined by he link, wi h he isk sco e a e aged by
he link and he FSI's summed. P ocedu es o compa ibilise he ne wo ks will be es ablished and es ed
du ing WP3.
(ii) P ocedu e o ne wo ks wi h mos links wi h less han 100m
When mos o he links in he ne wo k a e less han 100m long, coding needs o be done pe link. The
impac s o changing he coding s anda disa ion will be es ed and analysed as pa o WP3 using
PHOEBE use cases.
(iii) P ocedu e o ma ch in e sec ion sco es and nodes
In e sec ions, pa icula ly in ci ies and u ban a eas, can ake many o ms. In eali y, in e sec ions may
ha e as many as 6- o 8-legs and be a combina ion o slip lanes, on and o amps o o e passes, and

phoebe-p ojec .eu
Copy igh © by PHOEBE
53
s aged a ic signals. The e o e, ma ching S a Ra ing da a poin s ha p esen an in e sec ion coded
wi h mic o-simula ion nodes is a mo e complex han ma ching mid-block sec ions wi h links. Due o
he complexi y desc ibed abo e, he PHOEBE amewo k will p o ide guidelines on ma ching
in e sec ion sco es and nodes pe in e sec ion ype. The in e sec ions in he use case s udy a eas o e
many examples, including 3 and 4-leg junc ions, oundabou s and me ge lanes in single o double-
ca iageways.
Use guidelines on how o iden i y he bes p ocedu e and how o ope a e he analysis will be de eloped
as pa o WP3.
A e he s a ic esul s a e gene a ed and s uc u ed acco ding o links and nodes, he S a Ra ing models
wai o ou pu da a ( a iable speed and lows) om he a ic mic osimula ion and he FSI calib a ion
pa ame e s ecei ed om he beha iou al modelling componen . These a e hen used o gene a e he
‘dynamic’ isk p o iles. The dynamic c ash sco es a e hen sen back o he a ic simula ion ool o
isualisa ion pu poses and will be used o in o m PHOEBE KPIs. Visual maps a e hen used o illus a e
he spa ial dis ibu ion o isk sco es ac oss he en i e expanse o he oad ne wo k.
S a Ra ings will be ecalcula ed in s ep 2, o ake in o accoun changes in oad condi ions, speeds and
oad use lows.
2.5.2.1 Es ima ing FSI
Es ima ing FSI uses he iRAP FSI Es ima ion model o p edic he numbe o FSI associa ed wi h each
speci ic oad segmen o e a gi en pe iod ( ypically 20 yea s). The p ocess elies on a comp ehensi e
unde s anding o he oad a ibu es in conjunc ion wi h hei in e ac ion wi h he p e ailing a ic
condi ions. The FSI Es ima ion model will gene a e he i s es ima es based on he in as uc u e c ash
isk ( he S a Ra ings), numbe o oad use s and is calib a ed using c ash da a ( o he baseline scena io).
3
Va iable speed and low da a om he a ic mic osimula ion and beha iou al calib a ion ac o s Es ima es
om beha iou models will adjus hese alues. The inal FSIs will in o m PHOEBE KPIs and socio-
economic analysis.
FSI Es ima ions will be ecalcula ed in s ep 2 based on changes in he S a Ra ing, oad use lows and
beha iou calib a ion ac o s. This allows p o ides he basis o unde s and he sa e y impac s o he
changes.
Explaine no es
Va iable speed and lows o p oduce dynamic isk sco es
The S a Ra ings, CycleRAP, and FSI es ima ions models p o ide a s a ic isk e alua ion. This means
he esul s e e o a speci ic ime- ela ed poin ela ed o he momen he images we e cap u ed and
he suppo ing da a collec ed.
T a ic speeds and lows a e highly a iable compa ed o in as uc u e- ela ed elemen s o a oad. Wi h
new echnologies cap u e da a much as e , he isk assessmen model can cap u e hese a ia ions
and p o ide dynamic isk sco es. This helps oad au ho i ies unde s and changes ha may be occu ing
h oughou a ypical day, and add ess imes when isk is highes .
The dynamic alloca ion o speed and lows equi es enhancemen s in he cu en models. The ollowing
enhancemen s a e planned in WP3:
3
This calib a ion akes in o accoun ac o s which a e beyond hose conside ed by he S a Ra ing models, such as
ehicle sa e y s anda ds and ea u es (e.g. sea bel s and ai bags), oad use beha iou s (aside om ope a ing speed
which is di ec ly conside ed) and c ash esponse which may in luence FSI ou comes.
phoebe-p ojec .eu
Copy igh © by PHOEBE
54
Es ablishmen o he da abase s uc u e o accoun o mo e han one isk esul pe loca ion: o allow
isk calcula ion, o ins ance, o di e en hou s o he day, he model inpu ile needs o be s uc u ed
in a way ha would enable compu a ion and isualisa ion o he esul s.
Me hodology o non-daily lows: Cu en ly, he model uses Road AADT o ehicles and mo o cycle
lows and peak hou lows o bicycles and pedes ians. The model needs o be p epa ed and es ed o
ecei e inpu s o di e en pa ame e s o ehicle and VRU lows.
Da a ans e abili y p ocedu es o au oma e isk calcula ion based on a ic simula ion inpu s:
S anda disa ions o inpu s/ou pu s o ma s and he API connec ions need o be de eloped oge he wi h
he a ic simula ion ool o allow smoo h da a sha ing among models.
Calib a ion o FSI o e lec beha iou al changes
The FSI Es ima ion model is calib a ed using c ash da a o he assessed oads. The calib a ion ac o s
(CF) check ha he o al es ima ed numbe o FSI on he ne wo k is equal o he ac ual numbe o FSI
on ha ne wo k. The ollowing equa ions show he CF o mula o he pedes ian c ash c ossing he
inspec ed oad (PED CR-IR) c ash a ali ies as an example:
𝐶𝐹𝑃𝐸𝐷⁡𝐶𝑅−𝐼𝑅 =⁡𝐴𝑐𝑡𝑢𝑎𝑙⁡𝑛𝑜.⁡⁡𝑜𝑓⁡𝑝𝑒𝑑𝑒𝑠𝑡𝑟𝑖𝑎𝑛⁡𝑐𝑟𝑎𝑠ℎ⁡𝑓𝑎𝑡𝑎𝑙𝑖𝑡𝑖𝑒𝑠⁡𝑤ℎ𝑒𝑛⁡𝑐𝑟𝑜𝑠𝑠𝑖𝑛𝑔⁡𝑡ℎ𝑒⁡𝑖𝑛𝑠𝑝𝑒𝑐𝑡𝑒𝑑⁡𝑟𝑜𝑎𝑑⁡𝑜𝑛⁡𝑡ℎ𝑒⁡𝑛𝑒𝑡𝑤𝑜𝑟𝑘
∑(𝑆𝑅𝑆𝑃𝐸𝐷−𝐶𝑅−𝐼𝑅×𝑎(𝐴𝐴𝐷𝑇)𝑏×𝑉𝑁𝑂𝑁−𝑀𝐶×𝐹𝐺)⁡ (2-12)
whe e,
𝐶𝐹𝑃𝐸𝐷⁡𝐶𝑅−𝐼𝑅= Calib a ion ac o o pedes ian c ash a ali ies when c ossing he inspec ed oad
n= he numbe o 100-me e segmen s o oad
SRS PED-CR-IR = S a Ra ing Sco e o pedes ians c ossing he inspec ed oad
a = AADT mul iplie
AADT = annual a e age daily a ic
b = AADT powe
V NON-MC = non-mo o cycle AADT
FG = a ali y g ow h exponen
In his way, he FSI Es ima ion model akes accoun o ac o s ha in luence he numbe o a ali ies on
he oad o he han in as uc u e, speed and low. The e o e, when c ash da a is a ailable, he FSIs
accoun o FSI o igina ing om unsa e beha iou . FSIs will be calib a ed based on he beha iou al
model esul s. The p oposi ion is ha he calib a ion ac o o mula will ha e an ex a componen
ep esen ing he pe cen age o inc ease/dec ease in he numbe o dea hs o he pa icula beha iou
s udied and p o ide mo e g anula i y in he calib a ion. The app oach and es ing will be inalised in he
WP3 ac i i ies.
Model es ing
All planned enhancemen s o he models will be es ed. The iRAP ViDA sys em p o ides access o es ing
da a in mo e han 90 coun ies, ensu ing ha aining and es ing phases encompass a di e se ange o
di e se a ic scena ios. This igo ous analysis o all he me hods no only subs an ia es i s e ec i eness
wi hin he PHOEBE p ojec bu also showcases i s po en ial o enhancing oad sa e y on a b oade scale.
2.6 Sa e y indica o s and dynamic isk ou pu s
The basic de ini ion o isk in sa e y science is he occu ence p obabili y o an e en mul iplied by he
consequence o ha e en . Based on his de ini ion, many indica o s can be de ined o isk and sa e y on
he oad.
T adi ionally, oad c ashes a e used as indica o s o oad sa e y (in which he isk can be calcula ed as
he p obabili y o c ash occu ence mul iplied by he se e i y o c ashes). Howe e , he e a e many issues
phoebe-p ojec .eu
Copy igh © by PHOEBE
55
wi h using c ashes as an indica o . Fi s i is a eac i e app oach, meaning one o mo e people need o
be in ol ed in FSI c ashes be o e sa e y issues a e no iced o add essed; and second, many c ashes
(pa icula ly he p ope y damage only c ashes) a e no epo ed and so hei da a a e no a ailable in he
c ash da abases. This is pa icula ly he case o non- ehicle occupan oad use s.
Howe e , p edic i e models can iden i y zones a high isk o FSI c ashes wi h an accep able le el o
con idence. F om many yea s o esea ch in oad sa e y, i is al eady well-known wha oad componen s
a ec he likelihood and se e i y o c ashes. Many p edic ion o mula ions ha e been es ablished and
used as p esen ed in D1.1.
Recen ly, su oga e sa e y measu es ha e been commonly used o oad sa e y assessmen (as explained
in sec ion 3.1.7.1 o D1.1), o iden i y a ic con lic s by p ocessing he ehicula ajec o ies om he
ou pu o a ic mic osimula ions. T a ic con lic s ha e been used mo e ecen ly as p oac i e and mo e
equen indica o s o sa e y (in which case he isk would be he p obabili y o con lic s mul iplied by hei
se e i y).
Acco ding o he Swedish T a ic Con lic s Technique, he same causes can be a ibu ed o nea -c ashes
o a ic con lic s and can ac as a ool o sa e y assessmen . The a ious possible in e ac ions be ween
oad use s and he associa ed isk can be u he elabo a ed h ough he “sa e y py amid” by Hydén
(Figu e 2-15; Hydén, 1987), p esen ing di e en zones o sa e y c i ical e en s while also e lec ing he
associa ed se e i y.
Figu e 2-15 Classi ica ion o sa e y c i ical e en s acco ding o he Swedish T a ic Con lic s Technique (Hyden, 1987)
The Fede al Highway Adminis a ion (FHWA)’s Su oga e Sa e y Assessmen Model (SSAM) p o ides
a ious pa ame e s o iden i ying he a ic con lic s including:
• TTC = Time o collision;
• PET = Pos -enc oachmen ime;
• DR = Decele a ion a e;
• MaxS = Maximum o he speeds o wo ehicles;
• Del aS = Maximum ela i e speed;
• Classi ica ion as lane-change, ea -end, o pa h-c ossing e en ype;
• Del aV = Vehicle eloci y change had he e en p oceeded o a c ash.
In addi ion o he di ec sa e y measu es, he e a e some ac o s which a e p edic o s o sa e y (as opposed
o di ec indica o s). Fo example, ha sh accele a ion o co ne ing may di ec ly lead o con lic s o c ashes
and so hey a e good p edic o s o sa e y. Fo bo h indica o s and p edic o s o sa e y, some a e ine-
phoebe-p ojec .eu
Copy igh © by PHOEBE
56
g ained (sui able o mic oscopic sa e y analysis), and some a e mo e agg ega ed (sui able o
mac oscopic sa e y); some a e only ele an o d i e s (e.g., ha sh accele a ion), and some a e ele an
o ulne able oad use s oo (e.g., con lic s). As a esul , i is o high impo ance o c ea e a axonomy o
sa e y indica o s o be e in eg a e he da a, models and p ocesses o he componen s in PHOEBE and
es ablish he me hodology.
Taxonomy o sa e y ( isk) indica o s
The axonomy o sa e y indica o s (Table 2-4) is de eloped conside ing bo h indica o s and p edic o s o
sa e y. The indica o s include:
• Di ec measu es such as c ashes (numbe and se e i y);
• Su oga e measu es such as con lic s ( equency and se e i y);
• P edic o a iables such as haza dous ac o s such as ela ed o in as uc u e ( om oad
sa e y assessmen s a a ings); and
• Human ac o s ( isky beha iou s o oad use s), which can lead o po en ial collisions.
Fac o s associa ed wi h c ash causes a e gene ally ca ego ised as d i e - ela ed, ehicle- ela ed o
ela ed o he oadway en i onmen (in as uc u e, pa emen su ace condi ion, sigh obs uc ions, and
o he s).
The p edic o a iables include a a ie y o sa e y c i ical ac o s ela ed o in as uc u e and oad use s’
beha iou s, which can po en ially lead o collisions, such as:
• Inadequacies wi hin he In as uc u e o sa e a ic ope a ions;
• Lack o compliance o oadway ules, dis ac ed d i ing, a igued d i ing, agg essi e d i ing
beha iou s, e c.
Vehicle s uc u al cha ac e is ics wi hin a ious ehicle ypes a e ou o he scope and he eby ha e no
been conside ed.
Table 2-4 Taxonomy o Sa e y/Risk Indica o s
Road
Use s
Va iable
Le el
Rela ionship
wi h Risk
D i e s and ehicle occupan s
Con lic s ( equency and
se e i y)
(calcula ed using TTC, PET,
DR, MaxS, Del aS, ype,
Del aV)
Mic o: Indi idual ins ances
Meso: Numbe o con lic s ac oss a
segmen / a an in e sec ion
Mac o: Numbe o con lic s o e a ne wo k
Indica o
C ashes ( equency and
se e i y - FSI)
Mic o: Indi idual c ashes
Meso: Numbe o c ashes ac oss a
segmen / a an in e sec ion
Mac o: Numbe o c ashes o e a ne wo k
Indica o
S a Ra ing Risk Sco es
(In as uc u e)
Meso: isk sco e o a segmen / in e sec ion
Indica o
phoebe-p ojec .eu
Copy igh © by PHOEBE
63
3.1 P ocedu e
This sec ion elabo a es on wo me hods o SEA ha a e chosen o assessing he socioeconomic impac s
o he deployed changes o be conside ed in PHOEBE. The chosen me hods a e:
• Cos -Bene i Analysis (CBA); and
• Cos -E ec i eness Analysis (CEA).
Cos -Bene i Analysis (CBA)
CBA is a sys ema ic socioeconomic e alua ion echnique deployed o e alua e a p ojec , policy, o
decision's p os and cons. I is ca ied ou by compa ing he o al cos s associa ed wi h implemen ing he
p ojec , policy, o decision o he o al bene i s gene a ed by i .
Since a deployed change is such an in e en ion, he CBA can be used o e alua e he social and
economic impac s o a pa icula change. CBA is a quan i a i e me hod o e alua ion. The p ima y
objec i e o a cos -bene i analysis is o de e mine whe he he bene i s o an ac ion ou weigh i s cos s in
mone a y e ms.
The s eps o ca ying ou a CBA agains a change a e desc ibed below. Figu e 3-1 p o ides a concep ual
low o he p ocess.
a) Iden i ying he bene i s: The bene i s o a deployed change chie ly comp ise wo sepa a e elemen s
(Daniels e al., 2019), namely:
I. The numbe o c ashes p e en ed by he deployed change; and
II. O he possible bene i s ela ing o he mobili y and en i onmen al impac s.
b) Quan i ying o mone ising he bene i s: Speci ic bene i s gene a ed by any change a e
quan i a i e, e.g., p ope y damage a oidance and eme gency cos sa ings. Howe e , pa o he
bene i s a e quali a i e, e.g., he numbe o dea hs and se e e inju ies p e en ed. These KPIs a e
quali a i e because he cos s associa ed wi h hese KPIs a e a iable and depend upon se e al
ac o s. To es ima e he cos s associa ed wi h such KPIs, a echnique like Willingness- o-Pay (WTP)
is deployed.
• Willingness- o-Pay me hod: The ollowing scena io is conside ed o desc ibe he WTP me hod.
Suppose a change o policy is expec ed o educe he numbe o dea hs caused by oad c ashes
𝑛 by o e a gi en ime ame o a popula ion o size 𝑃. Now, i 𝑣 is he a e age amoun each
membe o he popula ion is willing o pay o und he oad sa e y measu e o isk educ ion, hen
he o al amoun o be paid by he popula ion is 𝑣.𝑃. Thus, he willingness o pay pe a ali y is
(𝑣.𝑃)/𝑛 (OECD, 2001). WTP can be u he used o es ima e he Value o S a is ical Li e (VoSL)
o he Value o P e en ing a Fa ali y (VPF). VoSL, o VPF, is a quan i a i e (o en mone a y)
alue used o quan i y he bene i o a oiding a a ali y. Gene ally, empi ical app oaches, e.g.,
Re ealed (OECD, 2001) o S a ed p e e ence me hods (con ingen alua ion) (An oniou, 2014;
Rizzi & O úza , 2006), a e deployed o es ima e WTP and VoSL.
• S a ed p e e ence (SP) me hod: This me hod is dependen on s a ed p e e ence su eys
(An oniou, 2014). The esponden s a e asked o deli e hei esponses h ough a a ing scale.
A sui able disc e e choice model (e.g., Mul inomial o Nes ed logi (in case he IIA does no hold)
is hen selec ed, wi h each po en ial esponse coded as an al e na i e. An example o he SP
me hod wi h o de ed p obi model is p esen ed below. Suppose 𝑌 is he esponse a iable wi h
𝐾 le els (scales); in his case he model can be ep esen ed as (An oniou, 2014):

phoebe-p ojec .eu
Copy igh © by PHOEBE
64
𝑃(𝑥)=𝛷(𝜃𝑗−𝛽′𝑥)
(3-1)
whe e,
𝛷 is he cumula i e no mal unc ion
𝜃0= −∞< 𝜃1<⋯ <𝜃𝑘<∞ a e he b eak poin s
𝑥 is he ec o o explana o y a iables
𝛽 is he ec o o unknown pa ame e s
The dis ibu ion o he choice p obabili y 𝑃 as he unc ion o he u ili y 𝑈 can be p esen ed as
Figu e 3-2.
The oad sa e y bene i s o a pe iod 𝑛, depending on le el o se e i y 𝑠 ha esul s om he
in oduc ion o a measu e, can hen be calcula ed as (Daniels e al., 2019):
𝐵𝑒𝑛𝑒𝑓𝑖𝑡𝑠𝑛=⁡∑𝑇𝑎𝑟𝑔𝑒𝑡𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑠∗𝐸𝑓𝑓𝑒𝑐𝑡𝑖𝑣𝑒𝑛𝑒𝑠𝑠𝑠∗𝐶𝑟𝑎𝑠ℎ𝑐𝑜𝑠𝑡𝑠⁡
𝑛
(3-2)
Whe e, Acco ding o Daniels e al. (2019) and Wijnen & S ipdonk, (2016) he 𝐸𝑓𝑓𝑒𝑐𝑡𝑖𝑣𝑒𝑛𝑒𝑠𝑠 is
ypically exp essed by means o he pe cen age educ ion (PR) o ei he he numbe o c ashes o
he numbe o casual ies as a consequence o he measu e as a consequence o he measu e
(calcula ed, o ins ance, wi h C ash Modi ica ion Fac o s).
The e ec i eness o en a ies acco ding o he le el o se e i y conce ned. The a ge c ashes
(𝑇𝑎𝑟𝑔𝑒𝑡𝑐𝑟𝑎𝑠ℎ𝑒𝑠) a e he numbe o c ashes (o inju ies) o a ious se e i y le els ha possibly can
be a ec ed by he measu e, so ypically in be o e-a e s udies his is he es ima ed numbe o c ashes
in he be o e pe iod co ec ed o eg ession o he mean and o end e ec s.
The bene i s can be exp essed in mone a y alues by mul iplying he numbe o p e en ed c ashes
o inju ies wi h he mone a y alue o he bene i , i.e., he cos pe c ash o inju y (𝐶𝑟𝑎𝑠ℎ𝑐𝑜𝑠𝑡). C ash
cos s ypically consis o se e al componen s o which human cos s, i.e., imma e ial cos s, end o be
he mos impo an . C ash cos s a e s ongly dependen on he se e i y le el o he c ash (Daniels e
al., 2019; Wijnen & S ipdonk, 2016).
Figu e 3-2: Dis ibu ion o he esponses Adap ed om An oniou (2014)
phoebe-p ojec .eu
Copy igh © by PHOEBE
65
c) Iden i ica ion and quan i ica ion o cos : Since cos s a e he second mos c ucial ac o , hey
should be iden i ied and quan i ied comp ehensi ely. Sec ion 3.2.1.2 p esen s di e en kinds o
cos s associa ed wi h he CBA. Howe e , mos cos ypes a e quan i a i e and mus be added
oge he o quan i y he o al cos associa ed wi h he deployed change.
d) Discoun ing and calcula ing he Ne P esen Value (NPV): The quan i ied cos s and bene i s
mus be discoun ed o coun e ac o s like in la ion. The discoun ing should be ca ied ou wi h a
sui able and well- esea ched discoun a e o achie e he co esponding NPV o bo h he cos s (-
) and bene i s (+). The ollowing o mula can be used o discoun he cos s and bene i s.
𝑁𝑃𝑉 = ∑𝑛𝑜𝑚𝑖𝑛𝑎𝑙⁡𝑣𝑎𝑙𝑢𝑒
(1+𝑑𝑖𝑠𝑐𝑜𝑢𝑛𝑡⁡𝑟𝑎𝑡𝑒)𝑛
𝑁
𝑛=1
(3-3)
Ne P esen Value o he deployed can be calcula ed by sub ac ing 𝑁𝑃𝑉𝐶𝑜𝑠𝑡𝑠 om 𝑁𝑃𝑉𝐵𝑒𝑛𝑒𝑓𝑖𝑡𝑠,
i.e., 𝑁𝑃𝑉𝑅𝑆𝑀 =𝑁𝑃𝑉𝐵𝑒𝑛𝑒𝑓𝑖𝑡𝑠 −𝑁𝑃𝑉𝐶𝑜𝑠𝑡𝑠. The alue o he NPV o he cos s and bene i s decides he
alidi y o he CBA, i.e., he CBA only makes sense i : 𝑁𝑃𝑉𝑅𝑆𝑀 >0.
e) Calcula ing BCR: As desc ibed in he p eceding sec ions, BCR is an impo an me ic o CBA. I
is calcula ed by di iding he 𝑁𝑃𝑉𝐶𝑜𝑠𝑡𝑠 by 𝑁𝑃𝑉𝐵𝑒𝑛𝑒𝑓𝑖𝑡𝑠. The CBA makes sense i 𝐵𝐶𝑅⁡>⁡1. Thus,
a deployed change is conside ed e ec i e i 𝑁𝑃𝑉𝑅𝑆𝑀 >0 and 𝐵𝐶𝑅⁡> ⁡1.
Cos -E ec i eness Analysis (CEA)
The Cos -E ec i eness Analysis (CEA) is ano he me hod o analyse he socioeconomic impac o oad
sa e y measu es. Me hodologically, CEA is closely ela ed o CBA. Some imes, CEA is conside ed a
a ian o CBA (Wesemann, 2000). In CEA, he cos s associa ed wi h a change deployed a e quan i ied
along wi h he numbe o a ali ies and inju ies.
One way o ca ying ou CEA o e alua e a pa icula oad sa e y measu e is by using Inc emen al Cos
E ec i eness Ra io (ICER) (Chan i h e al., 2021) ICER is also used o ank oad sa e y measu es in e ms
o hei e ec i eness. Ma hema ically, ICER is desc ibed as:
𝐼𝐶𝐸𝑅 =𝐴𝐷𝐶
𝐴𝐷𝐸
(3-4)
ADC ep esen s he a e age change in he budge in he gi en ime in e al ( he pe iod om no oad
sa e y measu e ill i s engagemen ). In con as , ADE ep esen s he a e age change in he numbe o
dea hs caused by oad c ashes du ing he same ime in e al. The esul an ICER o each measu e is
hen accoun ed o along wi h he ou quad an s o he CEA plane shown in Figu e 3-3 below.
Figu e 3-3 CEA plane ( ou quad an s) (Chan i h e al., 2021)
phoebe-p ojec .eu
Copy igh © by PHOEBE
66
3.2 Da a
Inpu da a
The inpu da a o he componen o socioeconomic analysis (SEA) can be ca ego ised in o wo classes
(OECD, 2001), namely:
• Inpu da a o calcula e he bene i s; and
• Inpu da a o calcula e he cos s.
Bo h hese classes o inpu da a a e desc ibed in he sec ions below.
3.2.1.1 Inpu da a o calcula e he bene i s
The inpu da a o he componen o socioeconomic analysis o calcula ing he bene i s (due o he
in oduc ion o oad sa e y measu es) p ima ily consis s o mul iple Key Pe o mance Indica o s (KPIs
4
).
The KPIs p esen he measu able o m o all he bene i s ha he deployed changes p oduce. A
measu able o objec i e o ma o he KPIs enables he mone isa ion p ocess o he bene i s.
Table 3-1 p esen s he h ee p incipal issues o ca ego ies, and he co esponding o eseeable KPIs
(Chan i h e al., 2021; Daniels e al., 2019; El ik, 2001). The p eceding componen s o p ojec PHOEBE's
me hodological amewo k, namely demand modelling, beha iou al modelling, a ic mic osimula ion, and
oad sa e y assessmen , a e se (in p ope o de and loops) o p oduce hese KPIs. Howe e , i is o be
no ed ha addi ional KPIs may be added, and he cu en ly lis ed KPIs can be edi ed o emo ed om he
lis depending on he de ailed modelling conside a ions.
Table 3-1 P incipal issues, oad sa e y measu es and co esponding KPIs
N
P inciple issues
Co esponding KPIs
1
Heal h and Sa e y
Pe cen age change in FSI
Numbe . o quali y-adjus ed li e yea s due o up ake o ac i e modes
and changes in ai quali y
Pe cen age o a el a 3-s a o be e by oad use ype
Pe cen age pa icipa ion in walking, cycling & o he o ms o ac i e
mobili y
2
En i onmen
Pe cen age change in mo o ised ips
Pe cen age change in non- mo o ised ips
Pe cen age change in ca bon emissions and/o uel consump ion
3
Economy
Heal h sec o sa ings pe km ( om less auma / be e heal h e c.)
Rising cos pe quali y-adjus ed li e yea due o up ake o ac i e
modes and changes in ai quali y
Pe cen age change in economic cos o FSI
Household sa ings esul ing om up ake o ac i e modes
4
The selec ion o he KPIs is a i al pa o he p ojec PHOEBE. In he la e phase o he p ojec , he
p ojec commi ee and he espec i e s akeholde s can join ly selec he KPIs.
phoebe-p ojec .eu
Copy igh © by PHOEBE
67
The KPIs lis ed in Table 3-1 will be calcula ed be o e and a e in oducing he changes. This way, he
bene i s gene a ed by he changes can be delinea ed and quan i ied (mone ised), which will u he ac as
he bene i side inpu da a o he SEA componen .
3.2.1.2 Inpu da a o calcula e cos s
The o he side o he inpu da a o he SEA componen consis s o he da a desc ibing he o e all cos s o
he changes (Daniels e al., 2019). The cos side can be de ined as all he cos s in ol ed in se ing up he
changes, including all he one- ime in es men and ecu ing cos s implica ed in se ing up he change.
The cos pa o he inpu da a is mo e objec i e han he bene i pa since i includes p ima ily he
in es men cos s, which can be easily quan i ied in mone a y e ms. The inpu cos s can be classi ied in o
se e al ca ego ies o ypes, namely:
• Ini ial capi al cos s: These cos s include all he one- ime expenses incu ed a he beginning
o he p ojec implemen ing changes, e.g., cos s o design, cons uc ion, equipmen
pu chases, and ins alla ion.
• Ope a ion and main enance cos s: These cos s include any ongoing cos s and ecu ing
expenses equi ed o keep he change ope a ional o e ime, e.g., inspec ion cos s, epai
cos s, o imp o emen cos s.
• Adminis a i e cos s: These cos s include any one- ime o ecu ing adminis a i e
expenses associa ed wi h managing he deployed change o e ime, e.g., S a sala ies,
aining, and o e head cos s ela ed o he RSM.
• Moni o ing and e alua ion cos s: These cos s in ol e he expenses associa ed wi h
acking he deployed changes in e ms o i s e ec i eness in p o iding oad sa e y o e ime.
These cos s include expenses o da a collec ion, analysis, and epo s on sa e y e alua ion.
Ou pu s and u he analyses
The socioeconomic analysis o deployed changes p o ides aluable ou pu s, insigh s, and in o ma ion o
decision-make s. I also enables u he analyses ha shed ligh on he po en ial impac s, bene i s, and
cos s o implemen ing he changes. The ou pu s and he u he analyses enabled by he socioeconomic
analysis a e lis ed below.
3.2.2.1 Ou pu s
The ou pu s o socioeconomic analysis ypically include:
1 Ne P esen Value (NPV): The NPV is a s anda d ou pu o socioeconomic analysis (cos -
bene i analysis) o any changes (Daniels e al., 2019; D èze & S e n, 1987). The NPV is a
nume ical alue ep esen ing he di e ence be ween he p esen alue o he bene i s
p o ided by he change and i s p esen cos . A posi i e NPV indica es ha he change is
economically iable.
2 Bene i -Cos Ra io (BCR): Like NPV, he BCR is ano he s anda d ou pu o he
socioeconomic analysis (cos -bene i analysis) o any change. The BCR p esen s he alue
o he a io o he bene i s gene a ed by he change deployed o i s cos s. A BCR > 1 signi ies
ha he bene i s gene a ed by he change a e mo e han he cos s in ol ed in se ing up ha
change (Daniels e al., 2019; D èze & S e n, 1987).
3 Inc emen al Cos E ec i eness Ra io (ICER): Like BCR, ICER is he a io o he a e age
change in he budge in he gi en ime in e al, i.e., he pe iod om no deployed change ill
i s engagemen o he a e age change in he numbe o dea hs caused by c ashes du ing he
same ime in e al. The esul an ICER o each change is compa ed o he ou quad an s o
he CEA plane o p o ide i s e ec i eness (Chan i h e al., 2021; Wesemann, 2000). I is an
ou pu o he cos -e ec i eness analysis (CEA). The CEA plane is p esen ed in Figu e 3-3.
phoebe-p ojec .eu
Copy igh © by PHOEBE
68
4 Mone ised bene i s and cos s: The socioeconomic analysis p o ides de ailed and
quan i ied in o ma ion abou he bene i s and he cos s in e ms o mone a y alues (CBA)
associa ed wi h he deployed change.
3.3 Fu he analyses
Fu he analyses suppo ed by SEA ha can be ca ied ou ollowing he SEA p ocess a e:
1 Sensi i i y Analysis: The sensi i i y analysis explo es he e ec s o changing alues o he
c i ical assump ions o a iables on he esul s.
2 O he economic (indica o s) analyses: Economic indica o s such as he In e nal Ra e o
Re u n (IRR) calcula ion can be used o assess he a e o e u n on in es men and he
inancial a ac i eness o he deployed change.
3 Risk assessmen analysis: The isk assessmen analysis can p o ide a quali a i e o
quan i a i e assessmen o he po en ial isks and unce ain ies associa ed wi h he oad
sa e y measu e's ou comes and cos s.
4 Inpu s o Policy Decisions: The analyses o he ou pu s o he SEA p o ide c i ical
in o ma ion o making in o med policy decisions. Decision-make s can use he esul s o
p io i ise and alloca e esou ces e ec i ely. Addi ionally, eedback loops can be a anged o
con ol he po ency o e ec i i y o he change o he p eceding modules o he p ojec
PHOEBE, e.g., mode choice modelling, a ic mic osimula ion, beha iou al modelling, and
oad sa e y assessmen module.
5 Moni o ing and E alua ion Plan: The SEA ou pu s can also p o ide he necessa y
in o ma ion in se ing up ecommenda ions o a moni o ing and e alua ion plan o ack he
ac ual ou comes o he implemen ed changes and compa e hem o he p ojec ed bene i s
and cos s.

phoebe-p ojec .eu
Copy igh © by PHOEBE
69
4 Solu ion equi emen s and
echnical speci ica ions
This sec ion is ocused on he exis ing and ‘ o become a ailable’ echniques and ela ed echnologies
equi ed o he PHOEBE F amewo k, and se s ou he p io i ies and ela ed in o ma ion equi ed o WP3.
I includes:
• Technical shee s which ou line all he ele an de ails abou each componen , such as model
desc ip ions, inpu s and ou pu s, equipmen , and license equi emen s;
• The echnological de elopmen s o he models’ in eg a ion and ha monisa ion o concep s,
inpu s and pa ame e s and hei deli e y ime able; and
• Unce ain ies and issues o be add essed in WP3.
4.1 Technical shee s
The echnical shee s compile echnical da a abou each o he elemen s in he PHOEBE F amewo k. The
pu pose o he echnical shee s is o p o ide a e e ence o he PHOEBE echnical pa ne s o suppo he
de elopmen wo k in WP3 and o acili a e in o ma ion indings in u u e applica ions o he PHOEBE
F amewo k.
The echnical shee con en e lec s he knowledge o he cu en s age o PHOEBE de elopmen .
The e o e, he echnical shee s will be upda ed as he WP3 and he PHOEBE p ojec p og esses. The
echnical shee s a e included in he appendices o his documen .
4.2 WP3 echnical de elopmen plan
To pu PHOEBE’s heo e ical me hodology in o p ac ice, he echnical pa ne s mus pe o m a se ies o
ac i i ies o achie e he le el o echnical de elopmen necessa y o es he p oposed heo e ical
amewo k.
The ac i i ies can be spli in o h ee phases: (i) p epa a ion, (ii) de elopmen and (iii) in eg a ion and
e inemen s. Figu e 4-1 p esen s he main ac i i ies in each de elopmen phase.
phoebe-p ojec .eu
Copy igh © by PHOEBE
70
Figu e 4-1 WP3 de elopmen phases
PHASE I: P epa a ion (6 mon hs)
Phase I, he p epa a ion phase, aims o se he speci ici ies o he app oach. Wi h he de ini ion o he
in o ma ion lows es ablished in he heo e ical amewo k, pa ne s mus ag ee on he echnical aspec s
o success ully ecei e inpu s. Examples o ma e s ha need o be discussed a e:
• Ne wo k compa ibili y be ween a ic simula ion models and oad sa e y assessmen
me hods;
• Modal spli ou look o ma o inpu in o a ic simula ion; o
• Beha iou al pa ame e s ha need o be calib a ed in o a ic simula ion models and how o
calib a e hem.
This phase has some in e sec ions wi h o he wo k packages. The design o new su eys will be
o mula ed as pa o he p epa a ion o model de elopmen s bu will ha e close coope a ion wi h he da a
managemen e o s handled in WP2. The same is ue o he esul s o he assessmen s and he
in o ma ion hey gene a ed ha will be a ailable o o he pu poses in he p ojec unde he umb ella o
WP2. Likewise, he F amewo k needs o calcula e he changes in KPIs, heo e ically es ablished in his
deli e able bu also ag eed wi h he local s akeholde s as pa o he WP4 ac i i ies.
Simila ly impo an is he p epa a ion o he a ic mic osimula ion en i onmen o in eg a ion. The
simula ion ool p o ided by AIMSUN will be he pla o m ha agg ega es he inpu s and eed models wi h
hei ou pu s, ensu ing he dynamic ou pu s expec ed o PHOEBE.
PHASE II: De elopmen (12 mon hs)
The Phase II ac i i y esul s will de e mine he sui abili y and eplicabili y o he PHOEBE p edic i e
app oach. I is one o he p ojec 's mos complex and c i ical momen s because i occu s when he model
de elopmen s and enhancemen s a e elabo a ed, calib a ed (i ha is he case) and es ed. This p ocess
phoebe-p ojec .eu
Copy igh © by PHOEBE
71
is expec ed o be an i e a i e, con inuous and e olu iona y p ocess, whe e in each ound o es ing,
encoun e issues will be wo ked ou o adjus models' s a is ical and echnical i s.
Phase II has a s ong alignmen wi h he ac i i ies on WP2. In case i is necessa y, a con inuous p ocess
o da a delinea ion is planned o suppo new da a collec ion. The same is ue o WP4, since local
s akeholde s need o e alua e he p ojec esul s and p o ide eedback on he isualisa ion ool, which
will also be de eloped o allow end use s o unde s and he impac s o es ing changes. The isualisa ion
ool will be pa o AIMSUN solu ions.
Phase II also includes some esea ch aspec s. The me hodological amewo k ( he ‘HOW’ ma ix oge he
wi h he p ocess lowcha ) se es as a new esea ch and de elopmen (R&D) amewo k oo ha
ad ances he s a e-o - he-a and can be used by o he schola s o in eg a ing he exis ing o eme ging
elemen s o demand models, a ic mic osimula ion, oad sa e y assessmen , and beha iou al models.
Howe e , i is impo an o no e ha while he R&D pa o he amewo k may include a b oad ange o
da a/models o each o he i e componen s in PHOEBE, no all o hem may ul ima ely be implemen ed
in PHOEBE and only hose ha a e ele an and easible (pa icula ly wi h an eye on he use case needs)
will be implemen ed. Fo example, d i ing unde in luence (DUI) is one o he majo illegal beha iou s
among d i e s in u ban en i onmen s and hus may need o be in eg a ed wi hin a ic mic osimula ion
oo.
Collec ing he da a o his beha iou is no a i ial ask and may be cos ly and ime consuming. Ye , he e
is no obs acle in ha ing his beha iou in he R&D pa o he p ojec and as a ype o illegal beha iou s o
mo o ised oad use s (Sec ion 2.4.2 o his deli e able), especially since he analy ical models pe aining
o his beha iou may no be subs an ially di e en han o he ele an beha iou s in PHOEBE.
PHOEBE pa ne s ha e al eady iden i ied esea ch oppo uni ies wi hin he p ojec scope. Depending on
he esea ch indings, he esul s may o may no be in eg a ed wi h he inal con igu a ion o PHOEBE,
bu hese will be documen ed and p esen ed in WP3 deli e ables.
PHASE III: In eg a ion and e inemen s (4 mon hs)
Phase III ocusses on he in eg a ion and possible e inemen s o he F amewo k. The comple e
in eg a ion will happen as pa o WP5, bu WP3 need o de elop and es he means o in eg a ion.
The PHOEBE F amewo k will p o ide a se ies o suppo ing ma e ials o ensu e eplicabili y. The
p epa a ion o he use guides, as well, as he inal model de elopmen documen a ion will be pe o med
du ing Phase III and be pa o deli e able D3.2.
4.3 WP3 imeline
The WP3 imeline conside s he h ee phases desc ibed abo e and he p ojec miles ones and
deli e ables planned. The WP GANTT cha s a e in Appendix B. As highligh ed in he imeline (in Ap il
2024 and Ap il 2025), wo deli e ables need o be p epa ed espec i ely o documen he echnical
de elopmen wo k. They a e:
D3.1 - New and enhanced models/simula ion en i onmen s and use suppo ma e ials (Be a ed.)
D3.1 will p esen he models' p epa a ions o he use case es ing and he upg ades in he simula ion
en i onmen . The Deli e able D3.1 is conside ed a mid- e m deli e able since he wo k would s ill be in
p og ess on he deli e able due da e. The na u e o he models applied on PHOEBE implies di e en
le els o de elopmen a he s age whe e D3.1 need o be p esen ed. In o he wo ds, i is expec ed ha
models ha a e da a dependen , such as he demand models and beha iou al models, can only be
de eloped en i ely a e he p ojec achie es he miles one o wo king da a a ailable o WP3, WP4 and
WP5, which he due da e is he same as ha o D3.1.
phoebe-p ojec .eu
Copy igh © by PHOEBE
72
The D3.1 p esen a ion o ma is s ill unde discussion bu mus combine echnical documen a ion and
isualisa ion o upg ades. The deli e able con en will p esen he ad ances in he simula ion en i onmen
and he oad sa e y assessmen s o demons a e he PHOEBE amewo k in he use cases.
D3.2 - Finalised models/simula ion en i onmen s me hodology ac shee s and use suppo
ma e ials
D3.2 will p esen he inal e sion o he model, including he e inemen s and lessons lea ned as pa o
he use case demons a ion. The use documen a ion will also be pa o D3.2, whe e he p ac ical and
heo e ical in o ma ion o guide he end use in applying he PHOEBE amewo k will be p esen ed.
PHOEBE de elopmen in e dependencies
The complexi y o he PHOEBE F amewo k is s ongly ela ed o he numbe o in e -dependencies among
i s componen s. The connec ions and da a lows be ween he elemen s o he PHOEBE F amewo k (as
depic ed in Figu e 1-1) a e ele an o he p ojec 's echnical de elopmen s. This sec ion desc ibes he
in e dependencies ha need o be esol ed in WP3. The echnical shee s in Appendix A p o ide mo e
de ail on he echnical in e dependencies.
Technical speci ici ies o be esol ed ega ding he in e dependencies:
• De ini ion o dynamic isk p o iles (S a Ra ings, c ash isk sco es and/o FSI Es ima ions); To
c ea e he condi ions o isk sco es o be used in such models, hey will be calcula ed i s as
s a ic sco es based on oad AADT and 85 h pe cen ile ope a ing speed, ep esen ing he
o e all condi ion o he ne wo k o e he day o he models. The esul s can be upda ed u he
in he p ocess a e he speed and low p o iles a e a ailable o he dynamic isk assessmen s
om he simula ions.
• Validi y and ele ance o S a Ra ings as a su oga e measu e o sa e y pe cep ion ( esea ch
and alida ion equi ed): Al hough he iRAP S a Ra ing does no e lec oad use
pe cep ions, he cha ac e is ics o he in as uc u e a ec he demand o speci ic modes
(e.g., mo e cycling ips when cycling in as uc u e is a ailable) o he way oad use s use
he in as uc u e (e.g., d i ing close o cyclis when he e is oad ma k sepa a ing he lows).
The e o e, isk sco es mus be a ailable be o e es ima ing he demand and beha iou al
models.
• Simula ion ime ames and demand da a a ailabili y: The simula ion o he scena ios depends
on he low in o ma ion in he shape o OD ma ix and zoning. The e o e, he demand da a's
a ailabili y will in o m he scena ios o be es ed as pa o PHOEBE (on-peak, o -peak,
hou ly, o o he ). AIMUSN Nex de ines he da a pa ame e s well since his is a s anda d inpu
o a ic simula ion scena ios.
• Es ablishmen o calcula ion p ocedu es o adjus mic osimula ion beha iou al ac o s: The
a ic simula ion depends on he indi idual beha iou al p obabili ies o beha iou al
pa ame e s in he ca - ollowing, lane-changing and gap accep ance models. How he models
ha desc ibe he a ge beha iou s will be inco po a ed in o he AIMSUN nex ool will be
discussed in WP3.
• Da a ans e p ocedu es om gi e s o ecei e s: The PHOEBE in eg a ion will be success ul
i he da a lows a e well es ablished in da a con igu a ion o ma ch ecei e s' equi emen s
and he low imeline (in each momen o he low p ocess, he ou pu s and inpu s should be
sha ed). Each o he in e dependencies's a ows p esen ed in Figu e 1-1 has i s own se o
a angemen s ha need o be made by he pai o pa ne s. The WP3 deli e ables will
documen and desc ibe he da a ans e p ocedu es o suppo u he applica ions o
PHOEBE.
phoebe-p ojec .eu
Copy igh © by PHOEBE
79
Re e ences
An oniou, C. (2014). A s a ed-p e e ence s udy o he willingness- o-pay o educe a ic isk in u ban
s. u al oads. Eu opean T anspo Resea ch Re iew, 6(1), 31–42.
h ps://doi.o g/10.1007/s12544-013-0103-3
Ba , R., Mackay, D., & Olsha sky, R. W. (1982). A Selec i e Re iew o T a el-Mode Choice Models.
Jou nal o Consume Resea ch, 8(4), 370. h ps://doi.o g/10.1086/208877
Ben-Aki a, M. E., & Le man, S. R. (2006). Disc e e choice analysis: Theo y and applica ion o a el
demand (Vol. 9). The MIT P ess.
Be azzon, S., & Elikan, O. (2009). Al e na i e neighbou hood speci ica ions o he spa ial weigh
ma ix; e ec s on spa ial au oco ela ion index and mul i a ia e analysis o heal h da a. In 17 h
ACM SIGSPATIAL In e na ional Con e ence on Ad ances in Geog aphic In o ma ion
Sys ems (ACM GIS 2009): 4-6 No embe 2009; Sea le.
Bie lai e, M., 1998. Disc e e choice models. In Ope a ions esea ch and decision aid me hodologies
in a ic and anspo a ion managemen (pp. 203-227). Be lin, Heidelbe g: Sp inge Be lin
Heidelbe g.
Chan i h, C., Pe mpoonwiwa , C. K., & Fowles, R. (2021). Cos -E ec i eness o Road Sa e y Policy
o P e en ing and Reducing Road T a ic Fa ali ies in Thailand. Thailand and The Wo ld
Economy, 39(2), 1–17. h ps://so05. ci- haijo.o g/index.php/TER/a icle/ iew/251882
Chen F. (2007). Sensi i i y o goodness o i indexes o lack o measu emen in a iance, S uc . Equ.
Modeling. 14, 464–504.
Daniels, S., Ma ensen, H., Schoe e s, A., Van den Be ghe, W., Papadimi iou, E., Kaise , S., Aigne -
B euss, E., So e opoulos, A., Wijnen, W., Weije ma s, W., Ca nis, L., El ik, R., Ma in Pe ez,
O., & Ziakopoulos, A. (2019). A SYSTEMATIC COST-BENEFIT ANALYSIS OF 29 ROAD
SAFETY MEASURES.
Depa men o T anspo . (2014). TAG UNIT M1, P inciples o Modelling and o ecas ing.
Depa men o T anspo . (2020a). TAG UNIT M1.2, Da a Sou ces and Su eys.
Depa men o T anspo . (2020b). TAG UNIT M2.1, Va iable Demand Modelling.
Domencich, T. A., & McFadden, D. (1975). URBAN TRAVEL DEMAND - A BEHAVIORAL ANALYSIS.
D èze, J., & S e n, N. (1987). Chap e 14 The heo y o cos -bene i analysis (Vol. 2, pp. 909–989).
Else ie . h ps://doi.o g/10.1016/S1573-4420(87)80009-5
El ik, R. (2001). Cos -bene i analysis o oad sa e y measu es: applicabili y and con o e sies. In
Acciden Analysis and P e en ion (Vol. 33). www.else ie .com
Ga cía-Mele o, G., Sainz-González, R., Co o-Millán, P., & Valencia-Vásquez, A. (2022). Ridesou cing
mode choice: A la en class choice model o Ube X in Chile. T anspo a ion Resea ch
In e disciplina y Pe spec i es, 16, 100722. h ps://doi.o g/10.1016/j. ip.2022.100722
Henshe , D. A. ;, & Bu on, K. J. ; (2008). Handbook o anspo modelling: Vol. Volume 1 (D. A.
Henshe & K. J. Bu on, Eds.; Second edi ion). Eme ald.
h ps://doi.o g/10.1108/9780857245670
Hydén, C. (1987). The de elopmen o a me hod o a ic sa e y e alua ion: The Swedish T a ic
Con lic s Technique. Bulle in Lund Ins i u e o Technology, Depa men , (70).
Kalliga, V. (2021). Who Does New T ips and Why?-An Analysis Towa ds he Modelling o Induced
Demand [Mas e hesis]. Technical Uni e si y o Munich.
Lee, B., Fujiwa a, A., Zhang, J., Con e ence, Y. S.-10 h I., & 2003, unde ined. (2003). Analysis o
mode choice beha iou s based on la en class models. Academia.EduBJ Lee, A Fujiwa a, J

phoebe-p ojec .eu
Copy igh © by PHOEBE
80
Zhang, Y Sugie10 h In e na ional Con e ence on T a el Beha iou Resea ch, Luce ne,
2003•academia.Edu. h ps://www.academia.edu/download/34170559/lee.pd
Mo an, P. A. (1950). No es on con inuous s ochas ic phenomena. Biome ika, 37(1/2), 17-23.
Mo oaki, Y., & Daziano, R. A. (2015). A hyb id-choice la en -class model o he analysis o he e ec s
o wea he on cycling demand. T anspo a ion Resea ch Pa A: Policy and P ac ice, 75, 217–
230. h ps://doi.o g/10.1016/j. a.2015.03.017
Na ayanan, S., & An oniou, C. (2023). Sha ed mobili y se ices owa ds Mobili y as a Se ice (MaaS):
Wha , who and when? T anspo a ion Resea ch Pa A: Policy and P ac ice, 168, 103581.
h ps://doi.o g/10.1016/j. a.2023.103581
No iko , D., Ko epano , V., & Chkha ish ili, A. (2018). Re lexion in ma hema ical models o decision-
making. In e na ional Jou nal o Pa allel, Eme gen and Dis ibu ed Sys ems, 33(3), 319–335.
OECD. (2001). Economic E alua ion o Road T a ic Sa e y Measu es. OECD.
h ps://doi.o g/10.1787/9789282112861-en
O úza , J. de D., & Willumsen, L. G. (2014). Modelling T anspo . Wiley.
h ps://doi.o g/10.1002/9781119993308
Pes el, E., F ied ich, M., HEIDL, U., PILLAT, J., Schille , C., & SCHIMPF, M. (2016).
Quali ae ssiche ung on Ve keh snach agemodellen. In M. Taille en & R. Deschaux (Eds.),
S aßen e keh s echnik. D oz. h ps:// id. b.o g/ iew/1437391
Rizzi, L. I., & O úza , J. de D. (2006). Es ima ing he willingness- o-pay o oad sa e y imp o emen s.
T anspo Re iews, 26(4), 471–485. h ps://doi.o g/10.1080/01441640600602302
on Schmid , A., López Díaz, M., & Schengen, A. (2021). C ea ing a Baseline Scena io o Simula ing
T a el Demand: A Case S udy o P epa ing he Region Tes Bed Lowe Saxony, Ge many.
In e na ional Con e ence on Ad ances in Sys em Simula ion (SIMUL), 51–57.
h ps://elib.dl .de/144358/
Washing on, S., Ka la is, M., Manne ing, F. and Anas asopoulos, P., 2020. S a is ical and
econome ic me hods o anspo a ion da a analysis. Chapman and Hall/CRC.
Wesemann, P. (2000). Economic e alua ion o oad sa e y measu es.
Wijnen, W., & S ipdonk, H. (2016). Social cos s o oad c ashes: An in e na ional analysis. Acciden
Analysis & P e en ion, 94, 97–106. h ps://doi.o g/10.1016/j.aap.2016.05.005
Winkle , C., & Mocanu, T. (2017, Oc obe ). Me hodology and Applica ion o a Ge man Na ional
Passenge T anspo Model o Fu u e T anspo Scena ios. P oceedings o he 45 h
Eu opean T anspo Con e ence. h ps://elib.dl .de/117999/
Ziakopoulos, A., & Yannis, G. (2020). A e iew o spa ial app oaches in oad sa e y. Acciden Analysis
& P e en ion, 135, 105323.
Ziakopoulos, A., Vlahogianni, E., An oniou, C., & Yannis, G. (2022). Spa ial p edic ions o ha sh
d i ing e en s using s a is ical and machine lea ning me hods. Sa e y science, 150, 105722.
Sha, H., Haoua i, R., Singh, M. K., Tympakianaki, A., Hu, B., Zach, M., Oikonomou, M., Chaudh y,
A., Thomas, P., Quddus, M., Mo is, A. (2022). T ans e abili y o Resul s wi hin he Le i a e
P ojec , T ans e abili y Wo king G oup Wo king Pape o he H2020 p ojec LEVITATE.
Ullman, J.B. and Ben le , P.M., 2012. S uc u al equa ion modeling. Handbook o Psychology, Second
Edi ion, 2.
phoebe-p ojec .eu
Copy igh © by PHOEBE
81
Appendices
Appendix A:
Technical shee s
phoebe-p ojec .eu
Copy igh © by PHOEBE
82
Technical shee
Beha iou al Modelling
Models Desc ip ion
Beha iou al Models:
Two majo modelling echniques o modelling human beha iou s will be used
including:
o Disc e e Choice Models: These models ake he pe spec i e o
disc e e decision-making p ocesses (e.g. disc e e ins ances o going
o e he speed limi )
- Mul inomial logi models
- O de ed p obi models
- Random pa ame e s logi /p obi models
- La en class logi /p obi models
o S uc u al Equa ions Models: These models look a he con inuous
p opo ions o beha iou s o e a ce ain pe iod o ime ( he a io o
agg essi e mobili y o e kilome es a elled).
- P incipal componen analysis
- Con i ma o y ac o analysis
- Gene alised linea models
- La en a iable models
Th ee gene al g oups o beha iou al models will be de eloped co esponding
wi h he h ee human ac o s p io i ised in PHOEBE:
● G oup #1: speeding beha iou o mo o ehicles on u ban a e ials
● G oup #2:
o illegal c ossing o pedes ians a in e sec ions and o zeb a c ossings
o ed-ligh unning o bicycles and e-scoo e s
● G oup #3: ha sh manoeu ing (accele a ion, decele a ion) co ne ing o
d i e s and o cyclis s
Expec ed inpu s
Indi idual d i ing ins ances (o p opo ions) wi h speed o e he speed limi
Indi idual d i ing ins ances o iola ions o oadway egula ions (all oad use
ypes)
Red-ligh unning e en s (all oad use ypes)
J-walking o illegal c ossing e en s
No yielding e en s (e.g., mo o ised ehicles no yielding o VRUs)
E en s o agg essi e d i ing
E en s o dis ac ed d i ing (e.g., use o cell phone)
Roadway and T a ic Cha ac e is ics ela ed da a
Inpu da a sou ces
Came as
Senso s
Floa ing Ca Da a
Telema ics
T a ic Su ey Da a
T a el A i udes Su ey (could be he sou ce o iola ions and dis ac ed
d i ing e c.)
phoebe-p ojec .eu
Copy igh © by PHOEBE
83
Technical shee
Beha iou al Modelling
Expec ed ou pu s
Indi idual p obabili ies o p opo ion o he popula ion conduc ing each o he
modelled beha iou
Con lic s esul s in he numbe o FSI c ashes
Con lic s caused by he a ge ed beha iou s
PHOEBE
F amewo k
Dependencies
(wha is his
dependen on
happening be o e)
Acqui ing ele an da a om a ailable sou ces as well as collec ing necessa y
da a in he case o una ailabili y om he a ailable sou ces (e.g. some use ul
da a o he p oposed beha iou al models could be collec ed h ough he
s a ed/ e ealed p e e ence su eys)
PHOEBE
F amewo k
P edecesso s
(wha does his
in o m)
Mic osimula ion models o calib a ing he beha iou al pa ame e s o be used
in he mic osimula ion models
Road sa e y assessmen ools h ough p o iding calib a ion ac o s o
beha iou s and es ima ion o FSI because o he modelled beha iou s
Equipmen
(So wa e
equi emen s)
Aimsun so wa e
Necessa y S a is ical Analysis So wa e
License
equi emen s
(i any)
Aimsun License
License o any pa icula s a is ical analysis so wa e (i no al eady a ailable)
PHOEBE
de elopmen s
Th ee gene al g oups o beha iou al models will be de eloped co esponding
wi h he h ee human ac o s p io i ised in PHOEBE:
● G oup #1: Speeding
- speeding beha iou o mo o ehicles on u ban a e ials
● G oup #2: Viola ions
- illegal c ossing o pedes ians a in e sec ions and o zeb a c ossings
- ed-ligh unning o bicycles and e-scoo e s
● G oup #3: Ha sh manoeu ing
- accele a ion, decele a ion, co ne ing o d i e s and o cyclis s
Addi ional
in o ma ion
De ailed in o ma ion abou he beha iou al models can be ound in PHOEBE
Deli e able 1.2.
phoebe-p ojec .eu
Copy igh © by PHOEBE
84
Technical shee
T a ic Simula ion
Models Desc ip ion
Aimsun simula ion pla o m
Aimsun p o ides a comp ehensi e simula ion so wa e. I includes all he
ele an aspec s in a ic simula ion: ne wo k ep esen a ion, a ic ligh s
con ol, demand modelling, di e en use beha iou modelling, a ic ou ing,
public anspo a ion and pedes ians among he mos ele an . On op o ha
he so wa e o e s a ious coding capabili ies o expand hei unc ionali y
using ad-hoc solu ions.
Aimsun de aul models
Aimsun so wa e pla o m has a wide a ie y o models o achie e a
comp ehensi e simula ion so wa e. Fo PHOEBE p ojec he mos ele an
a e hose ela ed o a ic dynamics, hese a e he ca ollowing model,
Modelling Vehicle Mo emen - Aimsun Nex Use ’s Manual. Then in mul i lane
s ee s o oads, he lane changing model Modeling Vehicle Mo emen -
Aimsun Nex Use ’s Manual, and in unsignalised in e sec ions he gap
accep ance model Modelling Vehicle Mo emen - Aimsun Nex Use ’s Manual.
Las bu no leas he pedes ian model d i en by a social o ce model ha
enables o simula e indi iduals pedes ians Simula ing Pedes ians - Aimsun
Nex Use ’s Manual and Gap Accep ance Model o Pedes ians a T a ic
Signals - Aimsun Nex Use ’s Manual.
Expec ed inpu s
● Ne wo k gene al geome ic da a: To se up a simula ion model a
ep esen a ion o he ne wo k is equi ed, including he links, nodes
and u nings. This can be impo ed om a ious sou ces o
p eexis ing simula ion models.
● T a ic ligh ope a ional ea u es: a ic ligh s iming and con ol
schemes need o be in oduced in o he simula ion pla o m.
● Demand da a: o igin des ina ion ma ices by anspo a ion mode
a e equi ed.
● Public anspo a ion: i any e ec s o he public anspo a ion
ehicles mo ing h ough he ne wo k a e equi ed, hen he public
anspo a ion da a, including, pa hs, s ops and schedule a e needed.
Open GTFS is ypically enough.
● T a ic low da a: in o de o alida e a ic models, a leas a ic
low coun a some key loca ions is needed, speeds and densi y o
occupancy alues a e app ecia ed.
● Beha iou al da a: o highly de ailed model calib a ion use s’
beha iou al da a like ajec o ies, accele a ion o speed p o iles a e
necessa y.
Inpu da a sou ces
● Open-sou ce da a: ne wo k in o ma ion can be ga he ed om
OSM, public anspo a ion om GTFS, an many imes he e a e
public AADT.
● P ojec pa ne s: a e expec ed o p o ide de ailed da a like O igin
Des ina ion ma ices and complimen a y da a o he open-sou ce
da a.

phoebe-p ojec .eu
Copy igh © by PHOEBE
85
Technical shee
T a ic Simula ion
● Use case s akeholde s: a e expec ed o p o ide de ails on he
in as uc u e changes o model and some addi ional da a o en ich
he da a al eady ob ained om open sou ce and p ojec pa ne s.
Expec ed ou pu s
● Calib a ed simula ion models: Aimsun Nex simula ion models o
assess he wha i scena ios a ound he zones o in e es .
● Dynamic a ic pa ame e s: om he simula ion models he h ee
main a ic a iables will be a ailable as ime se ies ( luc ua ing o e
ime). These a e: low, speed and densi y.
● Vehicles ajec o ies: his will enable an in-de ail analysis o he
con lic s ha a ise a di e en places o he ne wo k as well as he
use o Su oga e Sa e y Measu es wi h so wa e packages like
SSAM.
PHOEBE
F amewo k
Dependencies
(wha is his
dependen on
happening be o e)
● Da a: o de elop he Aimsun models a signi ican amoun o da a is
equi ed, his is al eady explained in he expec ed inpu s, bu he
PHOEBE amewo k dependencies a e:
○ iRAP s a a ing da a
○ Mode choice models
● Beha iou al models: calib a ion o simula ion models is dependen
on he beha iou al models used o pe o m he simula ion; hus,
modelling canno be comple ed un il he beha iou al models o use
a e ully de eloped.
PHOEBE
F amewo k
P edecesso s
(wha does his
in o m)
● iRAP dynamic s a a ing: o de elop dynamic s a a ing,
simula ion is needed o es di e en scena ios.
● Mode choice i e a ions: o asses he op imal mode choice spli he
simula ion pla o m needs o be eady.
Equipmen
(So wa e
equi emen s)
● Aimsun Nex 23
● Py hon 3.10 and he op ional lib a ies needed ( his will depend on
he PHOEBE p ojec de elopmen )
License
equi emen s
(i any)
● Adequa e Aimsun licence. Aimsun is a p op ie a y so wa e ha
needs a licence o be used. The e a e di e en ie s o licences,
depending on he modules needed
PHOEBE
de elopmen s
● Simula ion model wi h in eg a ed PHOEBE modules
○ Simula ion model de elopmen
○ In eg a ion o new beha iou al models
○ In eg a ion o he mode choice model
○ In eg a ion o he iRAP s a a ings in o a simula ion pla o m
○ In eg a ion o he new dynamic s a a ing
phoebe-p ojec .eu
Copy igh © by PHOEBE
86
Technical shee
Mode choice and modal shi models
Models Desc ip ion
Disc e e choice models:
The disc e e choice models will be deployed o o ecas he mode choice o
he indi idual scena ios, including he modal shi be ween wo scena ios.
Disc e e choice models speci y he p obabili y o choosing an al e na i e (ou
o all he al e na i es a ailable) by a g oup o indi iduals (ou o all he g oups
unde conside a ion). A sui able model ype will be chosen, conside ing he
inpu da a.
Expec ed inpu s
● S a ed-P e e ence (SP) su ey da a: The SP su ey da a (s a ed
p e e ence) will be an essen ial inpu o he model. Th ough he s a ed
p e e ence su eys, hypo he ical si ua ions a e p esen ed o he
esponden s, who a e hen asked o choose based on he gi en
a ibu es o each al e na i e wi hou necessa ily expe iencing hem
in eal si ua ions.
● Re ealed P e e ence (RP) da a: The RP da a cap u e a elle s’
cu en o e ealed a el beha iou . Fo ins ance, one is in e es ed in
knowing he mode a a elle uses, a el imes, and des ina ions.
● T a el dia ies: A a el dia y is an indi idual's eco d o his o he
a els. This essen ially is a ype o e ealed da a.
● Telema ics da a: The elema ics da a cap u ing loca ion, speed, uel
consump ion, ehicle aul s, e c., will be used as inpu s o he model.
● Demog aphic da a: Demog aphic da a con aining de ails like age,
gende , loca ion, household in o ma ion, ca a ailabili y, e c., will be
used as inpu o he model.
● Coun da a: Coun da a, e.g., he numbe o ehicle ips, ehicle
miles a elled, he numbe o pe son miles a elled, e c., will be used
o calib a ing he model.
Inpu da a sou ces
● SP da a: The SP da a will be ga he ed om su eys.
● RP da a: The RP da a can be ga he ed om se e al sou ces:
● F om a elle s’ a el dia ies
● Ope a ional da a e ealing he p e e ences o he a elle s om
he oad au ho i ies
● T a el dia ies: F om su eys.
● Telema ics da a: F om p o essional da a p o iding companies, local
au ho i ies.
● Demog aphic da a: F om local au ho i ies, census e c.
● Coun da a: F om p o essional da a p o iding companies, local
au ho i ies.
Expec ed ou pu s
The model is expec ed o deli e he p obabili ies o choosing di e en
al e na i es by di e en g oups o indi iduals. These p obabili ies can hen be
used o gene a e he OD ma ices o all he indi idual g oups using all he
al e na i es.
phoebe-p ojec .eu
Copy igh © by PHOEBE
87
Technical shee
Mode choice and modal shi models
PHOEBE
F amewo k
Dependencies
(wha is his
dependen on
happening be o e)
Be o e model es ima ion: The models ha e a dependency on he quali y and
ype o he inpu da a.
A e model es ima ion: A e model es ima ion, he calib a ion o he model
depends on he quali y and ype o he coun da a.
In e scena io dependencies: The models will ha e dependencies on all he
s eps and models ha can al e he inpu o explana o y a iables o he mode
choice models, e.g., T a el ime, cos , numbe o ans e s, speed e c.
PHOEBE
F amewo k
P edecesso s
(wha does his
in o m)
Su eys and o he inpu da a.
Equipmen
(So wa e
equi emen s)
BIOGEME: Py hon- and Pandas-based open-sou ce so wa e.
Apollo: R-based open-sou ce package.
MS Excel, AIMSUN ( a el demand modelling so wa e)
License
equi emen s
(i any)
AIMSUN
PHOEBE
de elopmen s
As pa o he PHOEBE de elopmen s, he mode choice models a e planned
o:
● Re lec he mul idimensionali y o human beha iou while choosing
modes, e.g., pe sonal p e e ences, socioeconomic cha ac e is ics,
a i udes, cul u al in luences, d i e cha ac e is ics, esponse imes,
adap a ion beha iou s, and psychological ac o s.
● Inco po a e he pe cep ion o isks associa ed wi h modes while
choosing modes.
● Re lec he spa io empo al cha ac e is ics o mode choice.
● Inco po a e he e ec s o he eal- ime in o ma ion low h ough apps
and in e ne in mode choice.
● Inco po a e he VRUs in mode choice amewo k.
phoebe-p ojec .eu
Copy igh © by PHOEBE
88
Technical shee
Road Assessmen Me hodology
Models Desc ip ion
iRAP S a Ra ing Me hodology
Road inspec ion da a o p o ide a simple and objec i e measu e o he le el o
sa e y 'buil -in' o he oads o ehicle occupan s, mo o cyclis s, pedes ians
and bicyclis s.
iRAP Fa al and Se e e inju ies es ima ion me hodology
Fa al and Se ious Inju y (FSI) Es ima es d aw on he oad a ibu e da a used
o S a Ra ings, low da a o each oad use and ne wo k-le el c ash da a o
p o ide an es ima ion o FSIs along each segmen o a oad and suppo
localised p io i isa ion o in es men .
CycleRAP
Road sa e y isk model o bicycle and ligh mobili y use s o e alua e he isk
o inciden s o oad and bicycling in as uc u e.
Expec ed inpu s
• En iched Geome y da a: encoded oad- ela ed a ibu es each
de ailing he immedia e oad en i onmen and isk in luencing a ibu es
o oad segmen s.
The a ibu es o be eco ded can be ound on he iRAP Coding Manual
(le hand d i e and igh hand d i e e sions).
• Ope a ional da a: speed and lows
Guidelines on how o adi ionally collec speed and low da a o s a ic
S a Ra ing analysis can be ound a iRAP Su ey Manual.
• Geo e e enced c ash da a by oad use ca ego y ( ehicle occupan s,
mo o cyclis s, pedes ians and bicyclis s) and c ash ype ( un-o
passenge and d i e side, head-on o loss o con ol o o e aking,
in e sec ion, p ope y access, along, c ossing inspec ed o in e sec ed
oad). C ash da a inpu o ma can be ound a he iRAP S a Ra ing and
In es men Plan Manual
Road use ca ego y and c ash ypes conside ed in he analysis can be
ound a iRAP Me hodology Fac Shee #4 C ash ypes
Inpu da a sou ces
• Road ea u e da a eco ded om ideo su ey o om an AiRAP da a
supplie
• Ope a ional da a on speed and lows can be aken om local au ho i y
da a when p esen , oad ope a ion measu emen o use o da a
supplie s p o iding measu ed speed and low da a ( o example The
Floow o O7 may p o ide measu es oad speed o o he da a o suppo
he model)
• C ash da a om oad au ho i ies
All da a need o be compa ible wi h iRAP speci ica ion a ailable a :
h ps://i ap.o g/speci ica ions/
Mo e de ails on he inpu da a o he PHOEBE p ojec can be ound a
PHOEBE Deli e able 2.1 - Consolida ed da a equi emen s and use case
egion da a a ailabili y epo .
phoebe-p ojec .eu
Copy igh © by PHOEBE
95
GANTT cha o oad sa e y assessmen enhancemen s – Pa 2
GANTT cha o a ic mic osimula ion enhancemen s