Date post: | 29-Jul-2018 |
Category: |
Documents |
Upload: | truongquynh |
View: | 223 times |
Download: | 0 times |
Los AlamosNATIONAL LABORATORY
LA-UR-Approved for public release;distribution is unlimited.
Title:
Author(s):
Submitted to:
Form 836 (10/96)
Los Alamos National Laboratory, an affirmative action/equal opportunity employer, is operated by the University of California for the U.S.Department of Energy under contract W-7405-ENG-36. By acceptance of this article, the publisher recognizes that the U.S. Governmentretains a nonexclusive, royalty-free license to publish or reproduce the published form of this contribution, or to allow others to do so, for U.S.Government purposes. Los Alamos National Laboratory requests that the publisher identify this article as work performed under theauspices of the U.S. Department of Energy. Los Alamos National Laboratory strongly supports academic freedom and a researcher’s right topublish; as an institution, however, the Laboratory does not endorse the viewpoint of a publication or guarantee its technical correctness.
99-7
The TRANSIMS Simulation Framework
B. W. Bush
WWW
TRANSIMS http://transims.tsasa.lanl.gov Page 1 of 66
The TRANSIMS Simulation Framework
B. W. Bush and the TRANSIMS teamLos Alamos National Laboratory
22 February 2001
TRANSIMS Page 2 of 66
Abstract
TRANSIMS (Transportation Analysis and Simulation System) isan integrated system of travel forecasting models designed togive transportation planners accurate, complete information ontraffic impacts, congestion, and pollution. The underlyingTRANSIMS philosophy is that individual behaviors and theirinteractions, as constrained by the transportation system,generate the transportation system’s performance. To effectthat performance in a simulation, individual behavior must bemodeled. This presentation outlines the framework of softwaremodules that constitute TRANSIMS, providing details on theirpurpose, input and output data, and algorithms; it also explainshow the TRANSIMS Selector holds the framework together.
Los Alamos National Laboratory is leading this effort to develop these newtransportation and air quality forecasting procedures required by the Clean Air Act,the Intermodal Surface Transportation Efficiency Act, and other regulations; it is partof the Travel Model Improvement Program sponsored by the U.S. Department ofTransportation, the Environmental Protection Agency, and the Department of Energy.
TRANSIMS Page 3 of 66
Outline
■ approach■ software modules
• population synthesizer• activity generator• route planner• traffic microsimulator• emissions estimator• output visualizer
■ the framework■ the “selector”■ examples■ future directions■ conclusion
TRANSIMS Page 4 of 66
TRANSIMS Approach
■ virtual metropolitan region created comprising completerepresentation of a region’s . . .• individuals• activities• transportation infrastructure
■ trips planned to satisfy individuals’ activity patterns■ movement of individuals across transportation network
simulated on a second-by-second basis• realistic traffic dynamics produced from interactions of
individual vehicles• vehicle pollutant emissions and fuel consumption estimated
■ models iterated• stabilizes simulation• allows travelers to react to information about the satisfaction of
their preferences
TRANSIMS Page 5 of 66
Key Ideas
■ modeling• identify limiting system constraints• preserve significant correlations• represent necessary behaviors• incorporate appropriate fidelity• construct relevant disaggegrations
■ simulation• information flow controls the
scenario• feedback loops cause system
characteristics to emerge
seconds
minutes
hours
years
meters
streets
block groups
personshouseholds
cohorts
space
time
demographics
TRANSIMS Page 6 of 66
Research Areas
■ computer science• parallel algorithms• large data set compression & distribution• pattern recognition• visualization• computational complexity and algorithms
■ theory of simulation• sequential dynamical systems• dependency graphs• coupled/nested simulations
■ complex systems• emergent behavior
■ feedback studies• uses: convergence, stabilization, modeling• approaches: control theory, game theory, information theory
TRANSIMS Page 7 of 66
Major TRANSIMS Components
Householdsand
Activities
Routesand
Plansµµµµsimulation
EmissionsMODELS3
TRANSIMS Page 8 of 66
Population Synthesizer: Purpose
■ creates a regional population imitation• demographics closely match real population• households are distributed spatially to approximate regional
population distribution• household locations determine some of the travel origins and
destinations■ synthetic population’s demographics form basis for individual
and household activities requiring travel
TRANSIMS Page 9 of 66
Population Synthesizer: Data Flow
PopulationSynthesizer
Synthetic Households� location� census tract / block group
Synthetic Persons� gender� age� schooling� employment (type, location,
hours)� transportation� income
Vehicles� vehicle id� household� initial network location� type of vehicle� emissions type
STF-3A� summary tables of
demographics� available for block groups
PUMS� 5% sample of census records� PUMA consisting of census
tracts, etc.� approximately 5,000 people
TIGER/Line� using MABLE/Geocorr� geographic layout of census
tracts and block groups
Network Data� activity locations
Forecast� marginals by block group
TRANSIMS Page 10 of 66
Population Synthesizer: Algorithm
STF-3A
PUMS
TIGER/Line
Network Data SyntheticHouseholds
SyntheticPersons
choose geographiclevel of detail
select demographicsand assemble
summary tables
construct PUMA-based multiway table
of demographics
estimate multiwaytable for each census
tract
draw householdsfrom multiway tables
in census tracts
Vehicles
Forecast
transform multiwaytables using forecast
TRANSIMS Page 11 of 66
Block Group 31200.1 in Portland, Oregon
block group
activity locations
streets
TRANSIMS Page 12 of 66
1996 Forecast for Block Group 31200.1
Size Households Age of Head Households Income Households1 84 � 24 24 � 4999 02 42 25–34 42 5000–9999 473 0 35–44 35 10,000–14,999 84 6 45–54 28 15,000–24,999 195 0 55–64 0 25,000–34,999 376 0 65–74 2 35,000–49,999 0� 7 0 � 75 1 50,000–74,999 21
Total 132 Total 132 75,000–99,999 0� 100,000 0
132
■ any forecasting methodology may be used■ forecast represented as a marginal distribution over block
groups of . . .• household size• age of head of household• annual household income
TRANSIMS Page 13 of 66
Iterative Proportional Fitting for Block Group 31200.1
■ correlation structureof demographicvariables preserved
■ marginal distributionsof forecast matched
General Correlation Structure
Correlation Structure in 31200.1
1,22,1
2,21,1
pppp
⋅⋅
=ϕ
■ correlation structuremeasured by “oddsratio,” e.g.,
TRANSIMS Page 14 of 66
Example Household in Block Group 31200.1
PersonsID 255552 255553 255554 255555Age 42 42 19 7Relationship Householder Husband/wife Son/daughter Son/daughterSex Male Female Female FemaleWorked in 1989 Yes Yes Yes No (under 18)EducationalAttainment
Some college,but no degree
High schoolgraduate,diploma or
GED
Somecollege, butno degree
1st , 2nd, 3rd,or 4th grade
Industry ElectricalMachinery,Equipment,
and Supplies,N.E.C
Not SpecifiedRetail Trade
Offices andClinics of
Chiropractors
Occupation Managers andAdministrators,
N.E.C
SalesWorkers,
OtherCommodities
Managers,Medicine and
Health
Total Income $45,000 $13,000 $6000Hours Worked 40 40 15Lived Here in 1985 No No No (under 5)Means oftransportation towork
Car, truck, orvan
Car, truck, orvan
Car, truck, orvan
Vehicle occupancy 1 1 1Time of departurefor work
6:50 1:00 14:00
Travel time to work 0:20 0:15 0:10
HousholdID 111733Size 4Vehicles 3Activity Location 23101PUMS Record 44789Anyone under 18 YesWorkers in 1989 3+Total Income $64,000Tenure Owned with
mortgageor loan
Value $90,000 -$99,999
TRANSIMS Page 15 of 66
Activity Generator: Purpose
■ creates . . .• household and individual activities• activity priorities• activity locations• activity times• mode and travel preferences
■ generates travel demand sensitive to demographics of syntheticpopulation
■ activities form basis for determining individuals’ trip plans for theregion
TRANSIMS Page 16 of 66
Activity Generator: Data Flow
ActivityGenerator
Household ActivitySurvey
� representative sample ofpopulation
� including travel and activityparticipation of all householdmembers
� recorded continuously for 24+hours
Network Data� nodes� links� activity locations (includes
land use and employment)
Synthetic Population
Activities� participants� activity type� activity priority� starting time, ending time,
duration (preferences andbounds)
� mode preference� vehicle preference� possible locations
TRANSIMS Page 17 of 66
Activity Generator: Algorithm
HouseholdActivity Survey
Network Data
SyntheticPopulation
Activities
create skeletal activitypatterns by strippinglocations from survey
and organizing via trips
match synthesizedhouseholds with survey
households usingregression keyed on
householddemographics
choose activity timesby randomizing survey
household times
generate trip chainsand activity locations
using continuousgravity model based onsynthesized household
location
handle commercialactivities and itinerant
travelers throughorigin-destinationmatrix methods
TRANSIMS Page 18 of 66
Example Prediction Tree Using Household Demographics
workers = 0all households
workers = 1
workers = 2
workers >= 3
persons = 1
persons = 2
persons >= 3
persons = 1
hhage < 38.5
hhage >= 38.5
hhage < 53
income > 5.5
income < 5.5
hhage >= 53
persons = 2 ages5to17 = 0
ages5to17 = 1
ages5to17 <= 1persons = 3 hhage < 29.5
hhage >= 29.5
ages5to17 >= 2
persons = 4 ages5to17 <= 1
ages5to17 >= 2
persons >= 5 ages5to17 <= 2
ages5to17 >= 3
persons = 2
persons = 3
persons >= 4 hdensity < 1.295
hdensity >= 1.295persons = 3
persons >= 4
TRANSIMS Page 19 of 66
Example Activities in Portland, Oregon
HOME
WORK
SHOP
HOME
WORKLUNCH
WORK
DOCTOR
SHOP
HOME
first person in household second person in household
TRANSIMS Page 20 of 66
Route Planner: Purpose
■ generates regional individual activity-based travel demand byassigning . . .• activities• modes• routesto individuals in the form of trip plans
■ trip plan is a sequence of . . .• modes• routes• planned departure and arrival times at origins, destinations, and
mode changing facilities■ trip plan selection related directly to each individual’s goals■ individual trip plans form basis for traffic simulation that
accounts for interactions among travelers
TRANSIMS Page 21 of 66
Route Planner: Data Flow
RoutePlanner
Transit Data� route paths in network� schedule of stops� driver plans� vehicle properties (e.g. bus
capacity)
Network Data� nodes� links� lane connectivity� activity locations� parking places & transit stops� "process" links
Vehicles Traveler Plans� vehicle start and finish
parking locations� vehicle path through network� expected arrival times along
path� travelers (driver and
passengers) present invehicle
� traveler mode changes
Activities
Link Travel Times
TRANSIMS Page 22 of 66
Route Planner: Algorithm
Transit Data
Network Data
Activities
Traveler Plans
Vehicles
convert activitypreferences for a traveler
into a constraint (anexpression in a formal
language) for the graph
decompose thetransportation networkinto a layered graph
find the path in thelayered graph with
minimum generalizedcost that satisfies thetraveler's constraints
express the optimal pathas a series of legs for the
traveler’s plan
Link TravelTimes
TRANSIMS Page 23 of 66
Example Layered Multi-Modal Network
walk
auto
bus
light railrail stop
bus stop
parking lot
activitylocation
proc
ess
link
TRANSIMS Page 24 of 66
Formal Language for Mode Preferences
■ Symbols represent different modes:• w = “walk,” c = “car,” b = “bus,” l = “light rail,” t = (b|l) = “bus or light rail”
■ A series of symbols expresses a mode preference:• wcw = “walk, then drive a car, then walk”• wctw = “walk, then drive to a transit stop, then take transit, then walk”• blb = “ride bus, then transfer to light rail, then ride bus”
walk network
bus network
car network
proc
ess
link
time delays areincurred duringmode transfer
generalizedcosts areincurred duringmode transfer
generalizedcosts areincurred duringmode transfer
proc
ess
link
proc
ess l
ink
parkinglocation
transitstop
activitylocation
activitylocation
activitylocation
• w = “only walk”■ Each mode
transfer passesthrough a processlink where timeand other costsare incurred.
TRANSIMS Page 25 of 66
Example Route Plans in Portland, Oregon
HOME
WORKLUNCH
WORK
DOCTOR
SHOP
HOME
HOME
WORK
SHOP
second person in householdfirst person in household
TRANSIMS Page 26 of 66
Traffic Microsimulator: Purpose
■ simulates the movement and interactions of travelers throughouta metropolitan region’s transportation system• executes travel plans provided by the Route Planner• computes the overall intra- and inter-modal transportation
system dynamics■ combined traveler interactions produce emergent behaviors
such as traffic congestion■ microsimulation output forms basis for environmental
calculations and for iteration decision-making
TRANSIMS Page 27 of 66
Traffic Microsimulator: Data Flow
TrafficMicro-
Simulator
Transit Data� route paths in network� schedule of stops� driver plans� vehicle properties (e.g.
starting location)
Network Data� nodes� links� lane use and connectivity� intersections (signs and
signals)� activity locations� parking� transit stops
Traveler Plans
Traveler Events� traveler id, trip id, leg id� time, location� inconvenience measures� anomalies� events
Summary Data� link travel times� link/lane densities� turn counts
Vehicles Snapshot Data� vehicles on links� vehicles in intersections� traffic controls� vehicle sub-populations
TRANSIMS Page 28 of 66
Traffic Microsimulator: Algorithm
Transit Data
Network Data
Traveler Plans
Traveler Events
Summary DataVehicles Snapshot Data
partition networkover computational
nodes (CPNs)
queue vehicles onparking places
update trafficsignal states
place travelers insimulation
let vehicles leaveparking places
perform lanechanges
move vehiclesforward on links
let vehicles enterparking places
transfer vehicles toother CPNs
let vehicles enterintersections
let vehicles leaveintersections
collect output
TRANSIMS Page 29 of 66
Cellular Automaton Microsimulation
7.5 meter × 1 lane cellularautomaton grid cells
intersection with multipleturn buffers (not internallydivided into grid cells)
single-cell vehicle
multiple-cell vehicle
TRANSIMS Page 30 of 66
Cellular Automaton Driving Rules
■ movement forward on grid based on . . .• gap to next vehicle• current speed• maximum speed
■ lane changes based on . . .• chosen approach lane to next intersection• current speed• gap to next vehicle in current lane• gaps to previous and next vehicles in new lane
(additional special cases for turn and merge pocket lanes)■ intersection entry based on . . .
• position/speed on link• occupancy of intersection buffer• state of oncoming/interfering traffic
■ total of about twelve adjustable parameters for driving rules
TRANSIMS Page 31 of 66
Traffic Microsimulator: Output Types
The state of eachvehicle on thelink is reported.
Snap-shotData
vehicle id, time, link id, position, velocitiy, lane, status
The stateof thetrafficcontrol isreported.
node id, time, phase, allowed movements
The state ofeach vehicle inthe intersectionis reported.
vehicle id, time, node id, position
Thetraversaltimes forvehiclesthat havetraveledthe lengthof the linkare summ-arized.
Summ-ary
Datalink id, vehicle count, sum of travel times
{The vehicle counts and velocities in"boxes" along the link are summarized.
link id, box position, vehicle count, sum of velocities
The traveler hasjust become lostbecause he/shecannot make theleft turn he/sheplanned onmaking at thisintersection.This event isreported.
TravelerEvents
traveler id, vehicle id, time, location, event
TRANSIMS Page 33 of 66
Emissions Estimator: Purpose
■ translates traveler behavior into consequent . . .• air quality• energy consumption• pollutant emissions
■ produces estimates of tailpipe and evaporative emissions forlight- and heavy-duty vehicles as a function of vehicle . . .• fleet composition• status• dynamics
■ emissions output forms basis for the computation of pollutantconcentrations, atmospheric conditions, local transport anddispersion, and chemical reactions
TRANSIMS Page 34 of 66
Emissions Estimator: Data Flow
EmissionsEstimatorExternal Data Sets
� California Air Resource Board(CARB) stratified trajectories
� “three cities” driving behaviors
Network Data� nodes and links� activity locations, parking
places, and transit stops
Microsimulator Output� traveler events� summary data
Emissions Inventory� CO, NOx, non-methane
hydrocarbons, particulatematter
� CO2, fuel consumption� 30 meter resolution along
road segments� 15 minute resolution in time
MODELS3 Database
Vehicles� make and age� technology� power-to-weight ratio� functioning or malfunctioning
emission control system
TRANSIMS Page 35 of 66
Emissions Estimator: Algorithm
ExternalData Sets
Network Data
Microsimul-ator Output
EmissionsInventory
MODELS3Database
Vehicles
estimate light-dutytailpipe emissions viaRiverside / Michiganmodal emissionsmodel
infer smooth vehicletrajectories
estimate heavy-dutytailpipe emissions viaWest Virginia model
estimate fuelevaporation viaMobile 5 & Mobile 6algorithms
infer types ofthrottling from gaps infront of vehicles
TRANSIMS Page 36 of 66
Emissions Estimator Details
F l e e t S t a t u s
E m i s s i o n I n v e n t o r y
· N O X · C O · H C · P a r t . 3 0 m b y 1 5 m i n .
· C h e m i s t r y · T r a n s p o r t · D i s p e r s i o n
· O z o n e · P a r t . · N O 2 · C O · H C 3 - D c o n c e n t r a t i o n s h o u r l y h o r i z o n t a l r e s . 1 ” 5 k m v e r t i c a l r e s . 2 0 ” 5 0 0 m .
D e c e l s I n s i g
P l a n n e r T r u c k L o a d s
M i c r o - s i m u l a t i o n
R e s u l t s
V e h i c l e S y n t h e t i c
P o p u l a t i o n
1 1 , 2 , . . . , 1 2 1 3 1 4 , . . . , 2 5
2
4 0
D e c e l s A c c e l s
A c c e l s
1 1 2 3
2
4 0
R e g i o n a l A i r Q u a l i t y
L D V A g g r e g a t e D y n a m i c s
H D V A g g r e g a t e D y n a m i c s
E V A P
L D V T A I L P I P E
H D V T A I L P I P E
P o w e r D i s t . b y T i m e a n d
S e g .
F l e e t D y n a m i c s
P o w e r ( v / A D i s t . )
M O D E L S 3
TRANSIMS Page 38 of 66
Output Visualizer: Purpose
■ allows an analyst to view and animate data generated by anyother TRANSIMS module
■ provides a unified and flexible means for exploring thevoluminous output data potentially available
TRANSIMS Page 39 of 66
Output Visualizer: Data Flow
OutputVisualizer
"Box" Data Files� time� link (with node being
approached)� length and position of box� arbitrary data columns for any
floating-point data to beviewed
Network Data� nodes, links, lanes� traffic controls� activity locations, parking
places, transit stops
Traveler Plans two or threedimensionalpresentation
Emissions Inventory
Microsimulator Output
static oranimated view
individual orsummary display
interactive orbatch mode
TRANSIMS Page 41 of 66
TRANSIMS Network Data
1
1
Node #8606 Node #8524
Node #8521Node #8523
Node #8600
Node #8525
Node #14136
Node #8520
Node #14141
Node # 14142
Node #8522
Node #8610
Node #8608
Node #14340
Node #8603
0
2
3 4 5
6
1
1234
1
2
1
23
3
2
4 5
6
01
3 21
1
2
3
2
3
4
56
1
2 1
1 2 3
543211234
1
2
12
3
1123 2123 21 3
1
2
21
1
2
2
1 1
1
1
23
2
3
4
5
6
1
1
2
2
3
4
1
1
2
1
2
211 32
1
1
2
2
2112
1
1
elev. 1000m.
northing 500m.easting 500m.
elev. 900m.
elev. 1000m.elev. 1000m. elev. 1000m.
elev. 1000m.
elev. 1000m.
elev. 1000m.elev. 1000m.
elev. 1000m.elev. 750m.
elev. 1000m. elev. 1000m. elev. 1000m.
elev. 1000m.
6.7% grade
4.3%
gra
de
16.7% grade
1000m.
1500m.
1000m.
1000m.
500m. 1000m. 1000m. 1000m.
3m.
6m.
6m.
6m.
9m.
6m.
6m.9m.
9m.
18m.
3m.3m.
3m.
3m.
12m.
18m. 6m.
12m.
6m.6m.
3m.
9m.
9m.
12m. 13.5m.
13.5m. 13.5m.
13.5m.
0m.
6m.
6m.
6m.6m.6m.
6m.
6m.
200m.
100m.450m.
100m.
3m.
300m.
200m.Link #11487 Link #11495
Link #28800Link #12384
Link #2750Link #2751
Link #11486
Link #2752Link #2753
Link #2754
Link
#27
55Li
nk #
2756
Link
#97
05Li
nk #
9706
Link #9704
Link #12407
Link #2758Link #2757Link #2759
Link
#28
804
3m.
LIGHTRAIL
LIGHTRAIL
LIG
HTR
AIL/
AUTO
LIGH
TRAIL/AU
TO
AUTOAUTO
AUTOAUTOAUTO
AUTO
AUTO
AUTO
AUTOAUTO
AUTOAUTOAUTOAUTOAUTO
AUTO
AUTO
AUTO
AUTO
AUTO
AUTO
AUTO
AUTO
LIG
HTR
AIL/
AUTO
AUTO
AUTO
LIGH
TRAIL/AU
TOAU
TO
LIGHTRAIL/AUTO
LIGHTRAIL/AUTO
LIGHTRAIL/BUS
AUTO
AUTO/BUSAUTO
LIGHTRAIL/BUSAUTO/BUS
LIGHTRAIL/BUS
LIGHTRAIL/BUSAUTO/HOV3/BUS
AUTO/BUS
AUTO
/BU
S
AUTO
/BUS
AUTO
/BUS
AUTO
/BUS
AUTO
/BU
S
AUTO
/BUS
AUTO
/BUS
AUTO
AUTOAUTO
AUTOAUTO
AUTOAUTOAUTO
AUTO
/BU
SAU
TO/B
US
AUTO
/BU
S
AUTO
AUTO
AUTO
AUTO
AUTO
AUTO
AUTO
AUTO/BUSAUTO/BUS
AUTO/BUSAUTO/BUS
AUTO
AUTO
AUTO
AUTO
AUTO
AUTO
AUTO
AUTO
AUTO
AUTO
AUTO
AUTO
AUTO
AUTO
AUTO
AUTO
AUTO
AUTO
AUTO
AUTO
AUTO
AUTOAUTO
AUTOAUTOAUTO
AUTO
AUTOAUTO
AUTO
speed limit 20 m/s(buses 15 m/s)
speed limit 15 m/s
speed limit 20 m/s
Parking #1002
300m.
Parking #1001
400m.
Parking #1003
200m.
Parking #1004200m.
Barrier #9001
450m.
200m.
Stop #3005
Stop #3002Stop #3003
350m.
650m.
650m.
Stop #3004
600m.
Stop #3001400m.
Detector #5001
Detector #5002
Detector #5005
250m.
300m.
350m.
3m.
3m.
3m.25% grade
speed limit 20 m/s(buses 15 m/s)
AUTO/TAXI
AUTO
Parking #1005
Parking #1006
Stop #3006
■ nodes■ links
• grade• mode• functional class
■ lanes• restrictions• connectivity
■ intersections• setbacks• signs• signals (rings, entries)
■ parking places■ transit stops■ activity locations
• land use• employment
■ “process” links
TRANSIMS Page 43 of 66
TRANSIMS Software Modules
■ general characteristics• can be treated as “black boxes”• simple invocation• well-defined parameter sets• well-defined input/output file specifications
■ several currently available• population synthesizer• activity generator• route planner• traffic microsimulator• emissions estimator• output visualizer
■ alternate modules performing identical functions (but usingdifferent algorithms) can coexist
■ completely new types of modules can be created
TRANSIMS Page 44 of 66
Data Flow for Current TRANSIMS Modules
Inpu
t File
sM
odul
esIn
put &
Out
put F
iles
PopulationSynthesizer
TravelerSurvey
SyntheticPopulation
Census
RoutePlanner
Activity
ActivityGenerator
OutputVisualizer
Traffic Micro-simulator
EmissionsEstimator
Network
TravelerPlans
Transit
Vehicle SimulationOutput
EmissionsInventory
Arbitrary BoxData
MODELS3Database
Air QualitySurveys
■ A TRANSIMS selector and iteration script control when modules arerun and how the data are routed between modules.
TRANSIMS Page 45 of 66
TRANSIMS Framework
■ flexible software system■ for transportation planning studies/experiments■ supports the future growth of TRANSIMS technology■ building blocks
• software modules– standardized command file– standardized input/output interface requirements– several major modules already available– third-parties may replace or add new conforming modules– reusable C++ libraries for building TRANSIMS objects
(network, plan, activity, and simulation output )– high-performance, parallel/distributed computing
• simulation data files– well-documented text formats– interface library callable from C, C++, FORTRAN, etc.
TRANSIMS Page 46 of 66
TRANSIMS Framework (continued)
• data manipulation tools– filtering, sorting, indexing, merging, searching, summarizing,
“noising”– for standard data files
• tools for controlling iteration between modules– “iteration database” with history of iterations– “selector” controlling and supervising iteration process
• iteration “scripts”– define a study or experiment– predefined for typical studies
+ calibration+ sensitivity analysis+ convergence/equilibration of activities, plans, and traffic
■ many possible combinations of above “building blocks” many possible realizations of TRANSIMS
TRANSIMS Page 47 of 66
Building Blocks in the TRANSIMS Framework
ActivityGenerator
filter, sort,merge,noise
reassign
travelers
Link Summary
Output
Traveler Event
Output
Activity
Set
Route Planner
Traffic Micro-
simulator
update
Plan SetIterationDatabase
new activities
replantravelers
merge/update
filter, sort,
merge,
noise
new plans
update
filter, sort,
merge,
noise
new output
Synt
hetic
Popu
latio
n
PopulationSynthesizer
selector
resimulatetravelers
roll back time, or pause
SelectorStatistics
update
update
EmissionsEstimator
EmissionsInventory
filter, sort,m
erge,noise
update
new emissions
recalculateemissions
arch
ive
software modules
iteration data files
data flows
software tools
simulation data files
TRANSIMS Page 48 of 66
One Realization of TRANSIMS
ActivityGenerator
filter, sort,m
erge,noisereassign
travelers
SummaryOutput
TravelerEvents
Activity Set
RoutePlanner
Traffic Micro-simulator
update
Plan SetIteration
Database
new activities
replantravelers
merge/update
filter, sort,m
erge,noisenew plans update
filter, sort,m
erge,noisenew output
SyntheticPopulation
PopulationSynthesizer
selector
resimulatetravelers
roll back time, or pause
SelectorStatistics
update
update
EmissionsEstimator
EmissionsInventory
filter, sort,m
erge,noise updatenew emissionsrecalculate
emissions
arch
ive
TRANSIMS Page 49 of 66
Selector: Purpose
■ controls when modules are run and how the data are routedbetween modules
■ operates in conjunction with an “iteration script” that providesthe top-level control for a series of TRANSIMS iterations
■ no single, “standard” Selector component• different study designs involve different iteration schemes• a variety of Selectors have uses in different studies or other
contexts
TRANSIMS Page 50 of 66
Selector: Data Flow
Iteration Database� record of traveler iterations
within a study� attributes representing
quasi-static information abouttravelers
� expectations encompassingplanned activities, routes, andtimes
� experiences comprisinginformation extracted fromdetailed microsimulationoutput
� analyst may customizecontents for a particularstudy
SelectorSelector Statistics
� basic summary of choicesmade
� how many travelers are beingreassigned activities or plans
� distributions of the differencebetween expected andexperienced travel times forvarious traveler populations
Selection Choices� list of the travelers that will be
reassigned activities,replanned, resimulated, etc.
� embodies the detaileddecisions of the Selector
TRANSIMS Page 51 of 66
Selector: Generic Algorithm
select travelersto resimulate
IterationDatabase
SelectorStatistics
SelectionChoices
extract statistics usedto decide how toproceed with the
iteration
select travelersto reassign
decidewhether to reassign,replan, or resimulate
travelers
select travelersto replan
ActivityGenerator
RoutePlanner
TrafficMicrosim.
select travelersto resimulate
TRANSIMS Page 52 of 66
Example Selection Strategies
■ replan routes for travelers who have simulated travel timesdiffering too much from their planned travel times
■ reassign activities for households if any member is too late forwork
■ average microsimulation output from several runs■ switch to a higher fidelity microsimulation midway through the
iteration process■ reject newly-generated route plans for some travelers based
on their travel preferences■ alter transit schedules based on travel demand■ adjust pricing based on network congestion■ mimic traveler information system by adding different levels
of random noise to feedback data for different travelers■ change selector to be used in next iteration
TRANSIMS Page 53 of 66
Four Example Iteration Schemes
Population
Activities
Plans
Traffic
Emissions
iteration number
1
2
3
4
75
6 8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
Population
Activities
Plans
Traffic
Emissions
1
2
3
4
75
6 8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
iteration number
Population
Activities
Plans
Traffic
Emissions
13 14 15 16 17 18 19 20 21 22 23
iteration number
1
2
3
4
75
6 8
9
10
11
12
Population
Activities
Plans
Traffic
Emissions
5 10 11 12 20
iteration number
1
2 4 6 8 9 13 15 16 18 19 21 22
3 7 14 17 23
TRANSIMS Page 54 of 66
Iteration in TRANSIMS
P r i n t i n g . . .
W e d 1 0 : 0 0 : 0 0 1 9 9 6 9 6 1 2 1 9 / n e x t - p a 0 t h i t e r a t i o n
Z o o m I n Z o o m O u t C e n t e r Z o o m T o R e s e t M a p S e t T i m e S t e p S t e p B a c k A n i m a t e S t o p R e s e t T i m e F i n d O p t i o n s P r i n t Q u i t
P r i n t i n g . . .
W e d 1 0 : 0 0 : 0 0 1 9 9 6 9 6 1 2 1 9 / n e x t - p a / 9 0 0 - 1 s t i t e r a t i o n
Z o o m I n Z o o m O u t C e n t e r Z o o m T o R e s e t M a p S e t T i m e S t e p S t e p B a c k A n i m a t e S t o p R e s e t T i m e F i n d O p t i o n s P r i n t Q u i t
P r i n t i n g . . .
W e d 1 0 : 0 0 : 0 0 1 9 9 6 9 6 1 2 1 9 / n e x t - p a / 9 0 0 - 1 0 t h i t e r a t i o n
Z o o m I n Z o o m O u t C e n t e r Z o o m T o R e s e t M a p S e t T i m e S t e p S t e p B a c k A n i m a t e S t o p R e s e t T i m e F i n d O p t i o n s P r i n t Q u i t
■ feedback is required to stabilize anonlinear system
■ the iteration process letsactivities, route plans, and trafficconverge to quasi-equilibrium
■ some experiments/studies needto control the flow of informationamong TRANSIMS componentsbetween iterations
Iteration 0 Iteration 1
Iteration 10
TRANSIMS Page 55 of 66
Feedback in TRANSIMS
■ The route planner only “sees”link capacities and travel timedelays.
■ The traffic microsimulationaccounts for intersectionimpedances and other vehicleinteractions in addition to linkcapacities.
■ Feedback of link travel timedelays output from the trafficmicrosimulation into the routeplanner is necessary in orderto generate realistic travelerplans.
■ Example: Without microsimulationfeedback, the planner would thinkthat link C is congested and notroute any traffic through link Donto link C.
D
A
B D
B
A
C
C
congestion in route planner
congestion in traffic microsimulation
TRANSIMS Page 56 of 66
Portland, Oregon, Case Study
Clean-up
Stabilize
Constrainto Scenario
Clean-up
Stabilize
State Space
Scenario Constraints
User Equilibrium
Solution
P A R
E
R T
A R
S
T
A R
R T
simulation
TRANSIMS Page 57 of 66
Status
■ main effort “winding down”• some research issues still outstanding• software completed but not fully tested
■ licensed to universities• actively used at Texas A & M, and Southern Florida State
■ commercialization process underway• PriceWaterhouseCooper first licensee
■ already applied in several case studies at LANL■ broader research continues
TRANSIMS Page 66 of 66
Summary
■ flexible software system• loosely coupled building blocks• integrated• customizable• extensible
many possible realizations ofTRANSIMS
■ to meet research community andMPO needs
■ strong theoretical basis
Product “Shell”
Product “Shell”
TRANSIMS-LANL
Studies & R
esearch
DO
T & Legislation
MPO