Transportation and Spatial Modelling: Lecture 6
-
Upload
tu-delft-opencourseware -
Category
Sports
-
view
377 -
download
3
description
Transcript of Transportation and Spatial Modelling: Lecture 6
![Page 1: Transportation and Spatial Modelling: Lecture 6](https://reader035.fdocuments.net/reader035/viewer/2022070302/548175d35906b5f56c8b4686/html5/thumbnails/1.jpg)
17/4/13
Challenge the future Delft University of Technology
CIE4801 Transportation and spatial modelling
Rob van Nes, Transport & Planning
(Uncongested) Assignment
![Page 2: Transportation and Spatial Modelling: Lecture 6](https://reader035.fdocuments.net/reader035/viewer/2022070302/548175d35906b5f56c8b4686/html5/thumbnails/2.jpg)
2 CIE4801: (Uncongested) Assignment
Content
• What do we want to know?
• Shortest path Reprise
• All or Nothing assignment
• Multiple routes
• Trucks
![Page 3: Transportation and Spatial Modelling: Lecture 6](https://reader035.fdocuments.net/reader035/viewer/2022070302/548175d35906b5f56c8b4686/html5/thumbnails/3.jpg)
3 CIE4801: (Uncongested) Assignment
1.
What do we want to know
![Page 4: Transportation and Spatial Modelling: Lecture 6](https://reader035.fdocuments.net/reader035/viewer/2022070302/548175d35906b5f56c8b4686/html5/thumbnails/4.jpg)
4 CIE4801: (Uncongested) Assignment
What do we want to know?
• Often the main goal of a modelling study
• Flows on links or lines • Bottlenecks • Travel times
![Page 5: Transportation and Spatial Modelling: Lecture 6](https://reader035.fdocuments.net/reader035/viewer/2022070302/548175d35906b5f56c8b4686/html5/thumbnails/5.jpg)
5 CIE4801: (Uncongested) Assignment
Introduction assignment
Trip production / Trip attraction
Trip distribution
Modal split
Period of day
Assignment
Zonal data
Transport networks
Travel resistances
Trip frequency choice
Destination choice
Mode choice
Time choice
Route choice
Travel times network loads
etc.
![Page 6: Transportation and Spatial Modelling: Lecture 6](https://reader035.fdocuments.net/reader035/viewer/2022070302/548175d35906b5f56c8b4686/html5/thumbnails/6.jpg)
6 CIE4801: (Uncongested) Assignment
Introduction to assignment
assignment model
travel demand
infrastructure supply
flows
![Page 7: Transportation and Spatial Modelling: Lecture 6](https://reader035.fdocuments.net/reader035/viewer/2022070302/548175d35906b5f56c8b4686/html5/thumbnails/7.jpg)
7 CIE4801: (Uncongested) Assignment
Solving the assignment problem
1O
2O
1D
2D
2 1
1
3
1
2
2
1 1
1
2
1 1O
2O
1D 2D
10 5
8 3
OD matrix network
Given the travel demand and the infrastructure supply, determine assignment using the shortest paths.
![Page 8: Transportation and Spatial Modelling: Lecture 6](https://reader035.fdocuments.net/reader035/viewer/2022070302/548175d35906b5f56c8b4686/html5/thumbnails/8.jpg)
8 CIE4801: (Uncongested) Assignment
Solving the assignment problem
1O
2O
1D
2D
1O
2O
1D 2D
10 5
8 3
OD matrix network
10 10
5 5
5 5
For the first origin, find all shortest paths to all destinations, and assign the corresponding trips to the shortest paths.
![Page 9: Transportation and Spatial Modelling: Lecture 6](https://reader035.fdocuments.net/reader035/viewer/2022070302/548175d35906b5f56c8b4686/html5/thumbnails/9.jpg)
9 CIE4801: (Uncongested) Assignment
Solving the assignment problem
1O
2O
1D
2D
1O
2O
1D 2D
10 5
8 3
OD matrix network
11 3 8
8 8
For the second origin, find all shortest paths to all destinations, and assign the corresponding trips to the shortest paths.
![Page 10: Transportation and Spatial Modelling: Lecture 6](https://reader035.fdocuments.net/reader035/viewer/2022070302/548175d35906b5f56c8b4686/html5/thumbnails/10.jpg)
10 CIE4801: (Uncongested) Assignment
Solving the assignment problem
1O
2O
1D
2D
1O
2O
1D 2D
10 5
8 3
OD matrix network 11 8
5 8
8 10 10
5
5 0 0
Combine all assignments between all OD pairs, and compute the total link flows.
0
8 0 0 0
0
0 0
0 0
0 0
![Page 11: Transportation and Spatial Modelling: Lecture 6](https://reader035.fdocuments.net/reader035/viewer/2022070302/548175d35906b5f56c8b4686/html5/thumbnails/11.jpg)
11 CIE4801: (Uncongested) Assignment
2.
Main building block: shortest path Reprise
![Page 12: Transportation and Spatial Modelling: Lecture 6](https://reader035.fdocuments.net/reader035/viewer/2022070302/548175d35906b5f56c8b4686/html5/thumbnails/12.jpg)
12 CIE4801: (Uncongested) Assignment
Main concept for ‘tree algorithms’ (Moore and Dijkstra)
• For all nodes • Set travel time tt(x) to and set the back node bn(x) to 0
• For the origin i set time to 0 and back node to -1 • Node i is the first active node a • Select all links (a,j) and check travel times
• If tt(a)+time(a,j) < tt(j) tt(j)=tt(a)+time(a,j) , bn(j)=a and node j becomes an active node
• Node a is no longer active
• Select a new active node from the stack and repeat previous step until there are no active nodes left
• For Dijkstra: select the link having the lowest travel time and select the active node having the lowest travel time
∞
![Page 13: Transportation and Spatial Modelling: Lecture 6](https://reader035.fdocuments.net/reader035/viewer/2022070302/548175d35906b5f56c8b4686/html5/thumbnails/13.jpg)
13 CIE4801: (Uncongested) Assignment
2 1
3
2
1 1
2
1
1
5
6
4
Shortest paths
![Page 14: Transportation and Spatial Modelling: Lecture 6](https://reader035.fdocuments.net/reader035/viewer/2022070302/548175d35906b5f56c8b4686/html5/thumbnails/14.jpg)
14 CIE4801: (Uncongested) Assignment
2 1
3
2
1
2
1
1
Shortest path tree Contains shortest paths to all nodes from a certain origin
Shortest paths
![Page 15: Transportation and Spatial Modelling: Lecture 6](https://reader035.fdocuments.net/reader035/viewer/2022070302/548175d35906b5f56c8b4686/html5/thumbnails/15.jpg)
15 CIE4801: (Uncongested) Assignment
1 1
4
7
2
5
8
3
6
9
2
3 4 5 6 7
8 9 10 11 12
Shortest route from node 7 to node 5: node 7 à node 4 à node 1 à node 2 à node 5 or link 8 à link 3 à link 1 à link 4
Representation of shortest paths
![Page 16: Transportation and Spatial Modelling: Lecture 6](https://reader035.fdocuments.net/reader035/viewer/2022070302/548175d35906b5f56c8b4686/html5/thumbnails/16.jpg)
16 CIE4801: (Uncongested) Assignment
1 1
4
7
2
5
8
3
6
9
2
3 4 5 6 7
8 9 10 11 12
Very compact, less practical: Back node representation
4 1 6 7 2 9 -1 7 8
node
1
node
2
node
3
node
4
node
5
node
6
node
7
node
8
node
9
route “7 à x”
Number for elements for storing all shortest routes in the network:
9 9 81N N× = ⋅ =
Representation of shortest paths
![Page 17: Transportation and Spatial Modelling: Lecture 6](https://reader035.fdocuments.net/reader035/viewer/2022070302/548175d35906b5f56c8b4686/html5/thumbnails/17.jpg)
17 CIE4801: (Uncongested) Assignment
1 1
4
7
2
5
8
3
6
9
2
3 4 5 6 7
8 9 10 11 12
Very compact, more practical: Back node representation Plus link numbers 4 1 6 7 2 9 -1 7 8
node
1
node
2
node
3
node
4
node
5
node
6
node
7
node
8
node
9
route “7 à x”
Number for elements for storing all shortest routes in the network: 2 2 9 9 162N N× × = ⋅ ⋅ =
Representation of shortest paths
3 1 5 8 4 10 -1 11 12
![Page 18: Transportation and Spatial Modelling: Lecture 6](https://reader035.fdocuments.net/reader035/viewer/2022070302/548175d35906b5f56c8b4686/html5/thumbnails/18.jpg)
18 CIE4801: (Uncongested) Assignment
Algorithm using back nodes
• Set all link flows equal to 0
• For each origin
• Determine shortest path tree
• For each destination
• Backtrack the route and add the OD-flow to the link flow of each link in the route
• Advantage: fast, efficient memory usage
![Page 19: Transportation and Spatial Modelling: Lecture 6](https://reader035.fdocuments.net/reader035/viewer/2022070302/548175d35906b5f56c8b4686/html5/thumbnails/19.jpg)
19 CIE4801: (Uncongested) Assignment
1 1
4
7
2
5
8
3
6
9
2
3 4 5 6 7
8 9 10 11 12
link 1 0 1 0 0 1 0 0 0 link 2 0 0 0 0 0 0 0 0 link 3 1 1 0 0 1 0 0 0 link 4 0 0 0 0 1 0 0 0 link 5 0 0 1 0 0 0 0 0 link 6 0 0 0 0 0 0 0 0 link 7 0 0 0 0 0 0 0 0 link 8 1 1 0 1 1 0 0 0 link 9 0 0 0 0 0 0 0 0 link 10 0 0 1 0 0 1 0 0 link 11 0 0 1 0 0 1 1 1 link 12 0 0 1 0 0 1 0 1
rout
e 7à
1 ro
ute
7à2
rout
e 7à
3 ro
ute
7à4
rout
e 7à
5 ro
ute
7à6
rout
e 7à
8 ro
ute
7à9
Less compact, very practical: Assignment map
Number for elements for storing all shortest routes in the network:
( 1) 12 9 8 864A N N× × − = ⋅ ⋅ =
3 0 2 4 0 0 0 1 0 0 0 0
Alternative representation
![Page 20: Transportation and Spatial Modelling: Lecture 6](https://reader035.fdocuments.net/reader035/viewer/2022070302/548175d35906b5f56c8b4686/html5/thumbnails/20.jpg)
20 CIE4801: (Uncongested) Assignment
How to use an assignment map
Link costs
Assignment map
For every OD a row of 0/1 indicating which links are part of the route
OD
flo
ws
Link flows
Rout
e co
sts
All calculations can be performed using matrix algebra
![Page 21: Transportation and Spatial Modelling: Lecture 6](https://reader035.fdocuments.net/reader035/viewer/2022070302/548175d35906b5f56c8b4686/html5/thumbnails/21.jpg)
21 CIE4801: (Uncongested) Assignment
Example 1 2 3 4 5 6 7 8 9 10 11 12
1 1 2 1 2 5 4 3 6 1 2 1
..
0 0 1 0 0 0 0 1 0 0 0 0
1 0 1 0 0 0 0 1 0 0 0 0
0 0 0 0 0 1 0 0 0 1 1 1
0 0 0 0 0 0 0 1 0 0 0 0
1 0 1 1 0 0 0 1 0 0 0 0
0 0 0 0 0 0 0 0 0 1 1 1
0 0 0 0 0 0 0 0 0 0 1 0
0 0 0 0 0 0 0 0 0 0 1 1
..
.. ..
7 1 5
7 2 5
7 3 5
7 4 2
7 5 2
7 6 2
7 8 3
7 9 3
.. ..
X
5+1+2+1 =9
X 5+5+2+2=14
![Page 22: Transportation and Spatial Modelling: Lecture 6](https://reader035.fdocuments.net/reader035/viewer/2022070302/548175d35906b5f56c8b4686/html5/thumbnails/22.jpg)
22 CIE4801: (Uncongested) Assignment
Selected link analysis
• Assignment map tells which OD-pairs use a specific link
• So you can create an OD-matrix of OD-pairs using that link and assign that to the network
• Note that using a back node representation requires extra administration in the calculations to perform a selected link analysis
![Page 23: Transportation and Spatial Modelling: Lecture 6](https://reader035.fdocuments.net/reader035/viewer/2022070302/548175d35906b5f56c8b4686/html5/thumbnails/23.jpg)
23 CIE4801: (Uncongested) Assignment
This was how you might do it
• Does it capture the relevant phenomena for the assignment?
• Differences in perception between users • Single route per OD or multiple routes • Multiple route concept implies route choice modelling
• Effect of congestion • Included or excluded
![Page 24: Transportation and Spatial Modelling: Lecture 6](https://reader035.fdocuments.net/reader035/viewer/2022070302/548175d35906b5f56c8b4686/html5/thumbnails/24.jpg)
24 CIE4801: (Uncongested) Assignment
Main assignment types
Congestion effect modelled?
No Yes
All-or-nothing assignment
Deterministic user-equilibrium
assignment
Stochastic assignment
Stochastic user-equilibrium
assignment
Multiple routes modelled?
No
Yes
![Page 25: Transportation and Spatial Modelling: Lecture 6](https://reader035.fdocuments.net/reader035/viewer/2022070302/548175d35906b5f56c8b4686/html5/thumbnails/25.jpg)
25 CIE4801: (Uncongested) Assignment
3.
All or Nothing assignment
![Page 26: Transportation and Spatial Modelling: Lecture 6](https://reader035.fdocuments.net/reader035/viewer/2022070302/548175d35906b5f56c8b4686/html5/thumbnails/26.jpg)
26 CIE4801: (Uncongested) Assignment
All-or-nothing (AON) assignment Reprise
Assumptions: • All traffic between an OD pair is assigned to just one route (usually the ‘shortest’ route (in terms of cost (min, €, km))) • There is no congestion
Topics • Mathematical formulation • Some comments
![Page 27: Transportation and Spatial Modelling: Lecture 6](https://reader035.fdocuments.net/reader035/viewer/2022070302/548175d35906b5f56c8b4686/html5/thumbnails/27.jpg)
27 CIE4801: (Uncongested) Assignment
AON: Formulated as an optimisation problem (route based)
ijr
ijr
ij
T
t
T
=
=
=
number of OD trips from i to j taking route r
travel time on route r from i to j
number of OD trips from i to j
minimize the total travel time experienced by all travellers
ijr ijrT T=∑ flow conservation
0ijrT ≥ nonnegativities
minijr
ijr ijrT i j rZ t T=∑∑∑
s.t.
![Page 28: Transportation and Spatial Modelling: Lecture 6](https://reader035.fdocuments.net/reader035/viewer/2022070302/548175d35906b5f56c8b4686/html5/thumbnails/28.jpg)
28 CIE4801: (Uncongested) Assignment
asa
a
ms
q
qtT
=
=
=
=
traffic flow on link a
traffic flow on link a with destination s
travel time on link a
number of OD trips from node m to node s
minimize the total travel time experienced by all travellers
mina
a aq aZ t q=∑
s.t.
;sa a
sq q=∑ s s
ms a aa M a M
T q q m s− +∈ ∈
+ = ∀ ≠∑ ∑ flow conservation
0saq ≥ nonnegativities
AON: Formulated as an optimisation problem (link based)
![Page 29: Transportation and Spatial Modelling: Lecture 6](https://reader035.fdocuments.net/reader035/viewer/2022070302/548175d35906b5f56c8b4686/html5/thumbnails/29.jpg)
29 CIE4801: (Uncongested) Assignment
AON-assignment: some comments
• Very fast method • Easy to interpret
• Misses essential concepts • Very sensitive for small changes
• Acceptable if you have many zones?
• Note that AON of software package A might differ from AON of software package B
![Page 30: Transportation and Spatial Modelling: Lecture 6](https://reader035.fdocuments.net/reader035/viewer/2022070302/548175d35906b5f56c8b4686/html5/thumbnails/30.jpg)
30 CIE4801: (Uncongested) Assignment
4.
Multiple routes: route choice (stochastic assignment)
![Page 31: Transportation and Spatial Modelling: Lecture 6](https://reader035.fdocuments.net/reader035/viewer/2022070302/548175d35906b5f56c8b4686/html5/thumbnails/31.jpg)
31 CIE4801: (Uncongested) Assignment
Stochastic assignment
Shortest path: All-or-nothing assignment: For each OD pair, all travellers take the shortest route Multiple routes: Stochastic assignment For each OD pair, all travellers take their perceived shortest route
1t 2t 3t
1t2t
3t
travel time
prob
abili
ty
![Page 32: Transportation and Spatial Modelling: Lecture 6](https://reader035.fdocuments.net/reader035/viewer/2022070302/548175d35906b5f56c8b4686/html5/thumbnails/32.jpg)
32 CIE4801: (Uncongested) Assignment
Main concept
ijr ijr ijrt ε= +t
ijr aa r
t t∈
=∑
( )ijr a a a a ijr ijra r a r a rt t tε ε ε
∈ ∈ ∈
= + = + = +∑ ∑ ∑t
However is it about perception at route or link level?
Usual assumption: Gumbel or Normal?
Gumbel leads to the Logit model
![Page 33: Transportation and Spatial Modelling: Lecture 6](https://reader035.fdocuments.net/reader035/viewer/2022070302/548175d35906b5f56c8b4686/html5/thumbnails/33.jpg)
33 CIE4801: (Uncongested) Assignment
Scaling or the error term
( )ijr ijr rt L= + Θt z Dispersion parameterSize of route
Random variable e.g. (0,1)rL r
N
Θ =
=
=z
( )a a at L= + Θt z Dispersion parameterSize of link
Random variable e.g. (0,1)aL a
N
Θ =
=
=z
The error term is related to the ‘size' of route or link
Size could be the length or free flow travel time of route or link
![Page 34: Transportation and Spatial Modelling: Lecture 6](https://reader035.fdocuments.net/reader035/viewer/2022070302/548175d35906b5f56c8b4686/html5/thumbnails/34.jpg)
34 CIE4801: (Uncongested) Assignment
Stochastic assignment at route level
Assume that z is Gumbel distributed:
exp( )exp( )
ijrijr
ijpp
tt
µβ
µ
−=
−∑(Logit)
10
5 5
5
But ... logit assignment cannot deal properly with overlapping routes!
All 3 route alternatives: 10 min.
33%
67% 33%
33%
![Page 35: Transportation and Spatial Modelling: Lecture 6](https://reader035.fdocuments.net/reader035/viewer/2022070302/548175d35906b5f56c8b4686/html5/thumbnails/35.jpg)
35 CIE4801: (Uncongested) Assignment
Stochastic assignment at route level
If there are overlapping routes, we can use the nested logit model
10
5 5
5
39%
61% 30%
30%
10
5 4.5
22
2
1 ln (exp( ))
( 1.5)
aa
t tµµ
µ
= − −
=
∑
![Page 36: Transportation and Spatial Modelling: Lecture 6](https://reader035.fdocuments.net/reader035/viewer/2022070302/548175d35906b5f56c8b4686/html5/thumbnails/36.jpg)
36 CIE4801: (Uncongested) Assignment
Stochastic assignment: route or link level?
• Nested logit doesn’t work in complex networks, e.g. Grid network in New York
• Alternatively we can switch from route to link level • Variation in perception at link level • While maintaining consistency at route level
![Page 37: Transportation and Spatial Modelling: Lecture 6](https://reader035.fdocuments.net/reader035/viewer/2022070302/548175d35906b5f56c8b4686/html5/thumbnails/37.jpg)
37 CIE4801: (Uncongested) Assignment
Stochastic assignment at link level
Assume that z is Normally Distributed:
...ijrβ = (Probit)
No explicit analytical expression, but can be computed by simulation. Probit assignment has no problem with overlapping alternatives.
10
5 5
5
62% 31%
31%
38%
For n=1 to N do
1. 2. Compute shortest route 3. If shortest route is alternative r then
0 0: ( ) , (0,1)a a at t t N= + Θ z z :
: 1/ijr ijr Nβ β= +
![Page 38: Transportation and Spatial Modelling: Lecture 6](https://reader035.fdocuments.net/reader035/viewer/2022070302/548175d35906b5f56c8b4686/html5/thumbnails/38.jpg)
38 CIE4801: (Uncongested) Assignment
Stochastic assignment
9.3600 9.5860 10.4766 10.9212 7.9679 9.1340 9.8664 8.5351 9.0683 9.5078 11.0549 11.2326 9.1196 10.3319 10.0110 9.8607 10.5413 9.2257 9.6406 10.8822 11.0726 10.4757 9.9555 8.0073 11.4155 9.5061 9.1937 10.3215 9.8450 10.7971
10 iterations:
route 1 route 2 route 3
40% 30% 30%
Probit assignment:
10
5 5
5
![Page 39: Transportation and Spatial Modelling: Lecture 6](https://reader035.fdocuments.net/reader035/viewer/2022070302/548175d35906b5f56c8b4686/html5/thumbnails/39.jpg)
39 CIE4801: (Uncongested) Assignment
Stochastic assignment and network representation
• Iterative scheme is tailor-made for tree search algorithms (or vice versa)
• Assignment map approach can be used as well • Generate routes first,
e.g. manually or using repetitive shortest path searches while systematically eliminating or penalizing links (k-shortest path algorithm)
• Check whether routes are realistic (e.g. large or very short detours)
![Page 40: Transportation and Spatial Modelling: Lecture 6](https://reader035.fdocuments.net/reader035/viewer/2022070302/548175d35906b5f56c8b4686/html5/thumbnails/40.jpg)
40 CIE4801: (Uncongested) Assignment
Assignment map and stochastic assignment
Link costs
Assignment map
For every OD one or more rows of 0/1 indicating which links are part of the route
OD
flo
ws
Rout
e co
sts
Rout
e Fl
ows
Choice model
AON or Logit (per OD)
Link flows
Probit
![Page 41: Transportation and Spatial Modelling: Lecture 6](https://reader035.fdocuments.net/reader035/viewer/2022070302/548175d35906b5f56c8b4686/html5/thumbnails/41.jpg)
41 CIE4801: (Uncongested) Assignment
Stochastic assignment: comments
• In which cases would it be a realistic assignment? • Networks without congestion
• Bicycle, 24-hour period (?)
• How many iterations? • Old models: maximum of e.g. 6 (Burrell) • Theoretically: iterate until convergence
• Normal distribution? • Only positive travel times => e.g. Gamma distribution
![Page 42: Transportation and Spatial Modelling: Lecture 6](https://reader035.fdocuments.net/reader035/viewer/2022070302/548175d35906b5f56c8b4686/html5/thumbnails/42.jpg)
42 CIE4801: (Uncongested) Assignment
5.
Trucks
![Page 43: Transportation and Spatial Modelling: Lecture 6](https://reader035.fdocuments.net/reader035/viewer/2022070302/548175d35906b5f56c8b4686/html5/thumbnails/43.jpg)
43 CIE4801: (Uncongested) Assignment
What’s the problem (and possible solutions)
• Simply put, there are also trucks on the road……
• Add an OD-matrix of trucks to the OD-matrix of cars • Note that a truck is not a car • Use a PCU-value to correct for the size and impact:
PCU for a truck is 1.5 or 2 • Assign the resulting OD-matrix of vehicles
• Assign OD-matrix of trucks separately (e.g. using different speeds) and add the flows • Use a PCU-value