o0o

TRANSPORTATION PLANNING

Traffic Assignment
Presented by
MSc. Ho Phuoc Duc

Ho Chi Minh City, April 2020

1
 Contents
 Introduction

 Planning Concepts and Four-Step Process Overview

 Trip Generation

 Trip Distribution

 Mode Choice

 Traffic Assignment

 Travel Demand Model Applications

2
 Traffic Assignment
▪ Traffic assignment, the final step in the four-step travel demand forecasting process,
estimates the specific routes on the network travelers take to get from one zone to
another.
▪ This step uses the total trips in the origin – destination (O/D) matrix from the trip
distribution step, which may have been divided among modes, and assigns them to
physical routes.
▪ In a travel demand model application, the input to the traffic assignment step is a
trip table (essentially the O/D matrix), while the result of the assignment is a path
table.
▪ The path table represents each zone-to-zone trip pair through a chained series of
link or segment numbers. Each interzonal pair could have as little as one path
assigned, or could have a number of alternative paths, each carrying a fraction of
the total trips.

https://www.fhwa.dot.gov/planning/processes/metropolitan/planning_practices/
3
 Traffic Assignment
1. Network, Paths, and Skimming
▪ The previous three steps in the four-step process primarily used the data for
different TAZs in the transportation system to be analyzed. These zones are inter-
connected by links, which represent the physical characteristics of the road
segments and facilities of the transportation network.
▪ The traffic assignment step uses the characteristics of all links when determining
which paths are most desirable for travelers, and therefore are expected to carry the
highest traffic loads.
▪ A path is then represented through a series of links, each with its own set of
attributes and characteristics. These attributes can include length, speed limit, travel
time, capacity, functional classification, and a host (whole) of other variables used to
describe the characteristics and expected performance of the link.

4
 Traffic Assignment
1. Network, Paths, and Skimming
▪ The link attributes are translated to the path through a process called skimming.
Conceptually, all the links along a path between two zones are skimmed for their
specific characteristics, which are then aggregated to result in path characteristics.
▪ Take, for example, the simplified 10-zone network in Figure 6.13. If evaluating a trip
with origin in zone 1 and destination in zone 4, the path is represented by link 1 – 4
that connects the two zones. All attributes of link 1 – 4 are translated to the path for
analysis.

Figure 6.13 Simple network for assignment illustration.

5
 Traffic Assignment
1. Network, Paths, and Skimming
▪ But while link 1 – 4 has a clear direct path, other zonal pairs have multiple paths.
For example, take the O/D pair for links 1 and 7. In this case, one potential path is
“1 – 4 → 4 – 7.” But several alternate paths exist in this case, including “1 – 5 → 5 –
7,” “1 – 2 → 2 – 5 → 5 – 7,” or “1 – 4 → 4 – 6 → 6 – 7.” Therefore, the decision of
which path drivers will take becomes a function of the attributes of each path, which,
in turn, are estimated from the aggregated attributes of each link sequence.

Figure 6.13 Simple network for assignment illustration.

6
 Traffic Assignment
1. Network, Paths, and Skimming
▪ In fact, going back to the earlier example of the path from zone 1 to 4, there
technically are other potential paths, including “1 – 3 → 3 – 4,” “1 – 5 → 5 – 4,” or “1
– 2 → 2 – 4,” or even “1 – 2 → 2 – 5 → 5 – 4.” These alternate paths may seem
undesirable compared to the direct path “1 – 4,” but not knowing any characteristics
of the network in Figure 6.13, we can’t disregard that route. What if link 1 – 4 is a
one-way street that doesn’t allow travel in the desired direction? What if it is a
neighborhood street with traffic calming? What if it is a drawbridge that opens for 15
minutes every hour resulting in a long queue, and so on?

Figure 6.13 Simple network for assignment illustration.

7
 Traffic Assignment
1. Network, Paths, and Skimming
▪ In short, we need additional attributes for each path, which are obtained from the
attributes of each link along the path through skimming. Figure 6.14 amends (small
changes) the previous example by adding link travel times (in minutes) for each
zone-to-zone connection.
▪ Applying our previous example, it appears that link 1 – 4 is indeed the shortest
travel time between zones 1 and 4 at 2 min. The alternate path travel times are 3
min for “1 – 3 → 3 – 4”; 4 min for “1 – 5 → 5 – 4”; 6 min for “1 – 2 → 2 – 4”; and 7
min for “1 – 2 → 2 – 5 → 5 – 4”. So while several paths exist, path “1 – 4” clearly
has the lowest travel time.

Figure 6.14 Simple assignment network with link traveltimes.

8
 Traffic Assignment
1. Network, Paths, and Skimming
▪ For the second zonal pair example (1 – 7), the path travel times are 5 min for “1 – 4 →
4 – 7”; 4 min for “1 – 5 → 5 – 7”; 7 min for “1 – 2 → 2 – 5 → 5 – 7”; and 6 min for “1 – 4
→ 4 – 6 → 6 – 7”. So in this case, path “1 – 5 → 5 – 7” produces the lowest travel time
at 4 min.
▪ In this simplified example, the skimming process strictly focused on link travel times.
But a host of other variables may be of interest in a traffic assignment step, including
cost of travel for each link (through tolls for example) or functional classification with,
for example, freeway travel being more desirable and reliable than arterial traffic, even
at the same base travel time. These variables can equally be obtained by skimming the
path in question (being considered = under consideration).

Figure 6.14 Simple assignment network with link traveltimes.

9
 Traffic Assignment
2. Shortest Path Assignment
▪ The simplest assignment algorithm is one based on the shortest path between
zones. Shortest path, in this case, can refer to shortest physical distance, but also
other variables, such as shortest travel time or lowest cost.
▪ Often, only a single variable is used to predict “the best” path or route between two
zones based on this method.
▪ More importantly even, the assignment at its most basic form is done as an all-or-
nothing assignment.
▪ In other words, all trips between a zonal pair will be assigned to that shortest path.
That assumption is further evaluated and refined in advanced assignment
algorithms.

10
 Traffic Assignment
2. Shortest Path Assignment
▪ Referring back to the simple network in Figure 6.14, the provided link travel times
can be used to determine the shortest path assignment.
▪ The assignment process evaluates the cumulative link travel time for each path
between zones and eliminates those route options with higher travel time.
▪ In demand modeling language, each path is skimmed for the total travel time and
compared to other paths for that zone-to-zone connection. Of course, this
skimming process can involve other variables or multiple factors.

11
 Traffic Assignment
2. Shortest Path Assignment
▪ Figure 6.15 shows the shortest path selection process for the simple example network.
The link travel times have been replaced with the cumulative travel times to get to a
particular zone. The example focuses on trips from zone 1 to all other zones, but would
eventually be repeated for all other origin zones as well.

Figure 6.15 Simple assignment network with link traveltimes.

12
 Traffic Assignment
2. Shortest Path Assignment
▪ Starting with link 1 – 4, the cumulative travel time from zone 1 to zone 4 is a simple 2 min.
For an alternate path from zone 1 to 4, the analysis sums the travel time from zone 1 to 3
(1 min) and the travel time from zone 3 to 4 (2 min) for a cumulative travel time of 3 min.
Because that cumulative travel time of 3 min is greater than the link 1 – 4 connection, link
3 – 4 is eliminated from the shortest path analysis for zone 1. This process is repeated for
all subsequent paths to additional zones, but is simplified because links that were
previously eliminated no longer need to be considered.

Figure 6.15 Simple assignment network with link traveltimes.

13
 Traffic Assignment
2. Shortest Path Assignment

▪ The process is repeated until all zones in the network have been connected through
a shortest path from zone 1. Figure 6.16 shows the resulting shortest path diagram
from zone 1 to all other zones.

Figure 6.16 Minimum path tree for zone 1. Figure 6.15 Simple assignment network withlink
travel times.

14
 Traffic Assignment
2. Shortest Path Assignment
▪ From the minimum path tree in Figure 6.16 the total load on each link can be
estimated by assigning the forecasted trips between zones from the trip distribution
step, which is illustrated in Example 6.9. Note that we have only covered trips
originating from zone 1 at this point. In a full travel demand modeling application, this
series of steps would have to be repeated for the remaining nine origin zones.
▪ Even for this simple network, it is evident that this process can be quite cumbersome
and challenging to complete!

Figure 6.16 Minimum path tree for zone 1. Figure 6.15 Simple assignment network withlink
travel times.

15
 Traffic Assignment
6.2 Shortest Path Assignment
EXAMPLE 2.9
The traffic assignment for the sample network in Figure 6.14 resulted in the minimum path
tree for trips from zone 1 as shown in Figure 6.16. The following table lists the number of
trips (from the trip distribution step) for all destination zones, as well as a listing of
impacted links.

➔ Using this shortest path and all-or-nothing assignment, estimate the total number of
zone 1 origin trips on each of the links in the network?

16
 Traffic Assignment
6.2 Shortest Path Assignment
SOLUTION

Figure 6.14 Simple assignment network withlink

travel times.

Figure 6.15 Simple assignment network with link Figure 6.16 Minimum path tree for zone 1.
travel times.
➔ Using this shortest path and all-or-nothing assignment, estimate the total
number of zone 1 origin trips on each of the links in the network?
17
 Traffic Assignment
6.2 Shortest Path Assignment
SOLUTION
➔ This problem is solved by summing up the trips for each link using all zonal
combinations from zone 1. Note that links 2 – 4, 2 – 5, 3 – 4, 3 – 9, 4 – 5, 4 – 6, 4 – 7,
5 – 6, 6 – 7, 6 – 8, 6 – 9, 7 – 9, 7 – 10, and 9 – 10 were eliminated from the shortest
path assignment due to high cumulative travel times. These are not shown in the
table, as the resulting number of trips (originating in zone 1) is zero. The result is
shown in the following table.

Figure 6.16 Minimum path tree for zone 1.

18
 Traffic Assignment
3. Capacity-Constrained Assignment
▪ There are several problems with the shortest path and all-or-nothing assignment
presented in Figure 6.15 and Figure 6.16. For example, the assignment doesn’t take into
account that the total traffic demand assigned to a particular link may result in increases
in travel time on that link, thereby increasing link travel time and decreasing its desirability
in the assignment.
▪ In extreme cases, of course, the demand may actually exceed the capacity of a link,
making it impossible to process the full demand from all zones. A capacity-constrained
assignment takes these travel time increases and link capacity constraints into
consideration.

Figure 6.16 Minimum path tree for zone 1.

Figure 6.15 Simple assignment network withlink
travel times.
19
 Traffic Assignment
3. Capacity-Constrained Assignment
▪ A capacity constrained assignment applies a relationship between the volume-to-
capacity ratio (v/c) on a link and the resulting travel time to each link. The higher the
v/c ratio estimated, the higher is the resulting effect on link travel time.
▪ The most commonly used relationship was developed by the U.S. Bureau of Public
Roads and is referred to as the BPR function. The BPR formula for estimating link
travel time under capacity constraint is as follows:

20
 Traffic Assignment
3. Capacity-Constrained Assignment
▪ The form of the equation results in an increase in the estimated link travel time as a
function of link capacity and v/c ratio. The link travel time is equal to 1.15 times the free-
flow travel time at capacity (v/c = 1) and increases exponentially past the capacity.
▪ A sample application for a link with free-flow travel time of 5 min and capacity of 2000
veh/h is shown in Figure 6.17.

Figure 6.17 Sample application of BPR function for capacity-

constrained assignment.
21
 Traffic Assignment
3. Capacity-Constrained Assignment

▪ Note that there are ways to calibrate the BPR function. For example, changing the
1.15 coefficient to a higher value increases the initial slope of the travel time curve
for low v/c ratios. Similarly, increasing the v/c exponent of 4 to a higher value will
drastically increase the slope beyond the capacity value.
▪ It is important to point out that this assignment method will continue to assign flows
to a link even if the demand exceeds capacity.
▪ In a travel demand application this is important to assure that the model continues
to be able to perform calculations and doesn’t “lock up”.
▪ In other words, the BPR function creates additional link impedance at high v/c
ratios, but doesn’t actually cap (limit) the link volume to the capacity.

22
 Traffic Assignment
3. Capacity-Constrained Assignment
▪ The application of the capacity-constrained assignment is iterative. After an initial
(shortest path) assignment, the link travel times are updated based on the BPR
function or another impedance curve.
▪ The assignment is then repeated with the updated link travel times until an
equilibrium is reached.
▪ Equilibrium in this case means that the assigned flows and link sequences no
longer change, and travel times on all links are thus stable between successive
assignments.

https://www.fhwa.dot.gov/planning/processes/metropolitan/planning_practices/

23
 Traffic Assignment
4. Dynamic Traffic Assignment and Other Advanced Algorithms
▪ The previous assignment methods are useful to illustrate the basic concepts of
traffic assignment, and form the basis for more advanced assignment algorithms.
▪ But in practice, an all-or-nothing shortest path assignment, or even a capacity-
restrained assignment, are often not representative of the assignment algorithms
used in travel demand models.
▪ These algorithms generally use more advanced functions that are applied in an
automated fashion to search for potential assignment solutions in a software
implementation.
▪ A common challenge, for example, with the simpler algorithms is that they may not
result in an equilibrium solution. For example, in a large network with several
thousand zones, an all-or-nothing assignment may result in over-loading of key links
to a point, where even repeated iterations cannot find sufficient alternate paths.

24
 Traffic Assignment
4. Dynamic Traffic Assignment and Other Advanced Algorithms
▪ Travel demand model – based assignments therefore sometimes begin with a
reduced trip distribution table (e.g., 10% of trips), run an initial assignment, and then
gradually add traffic in successive assignment iterations until the full O/D table is
assigned and equilibrium is reached.
▪ A special category of assignment algorithms is referred to as dynamic traffic
assignment (DTA) algorithms (Patriksson, 2015; Meyer, 2015; Weiner, 2012). These are
commonly found in simulation models that are applied to a transportation network,
and may interact with the travel demand model.
▪ Simulation tools are further classified as microscopic or mesoscopic simulation.
These simulation tools differ in the level of complexity of the algorithms used to
model individual vehicles, and both types can contain DTA logic.

25
 Traffic Assignment
4. Dynamic Traffic Assignment and Other Advanced Algorithms
▪ A dynamic assignment uses the actual measured travel times from these vehicles
to inform successive assignment iterations. So, rather than approximating travel
times based on BPR functions and the skimming process, DTA algorithms use the
actual vehicle travel times.
▪ Dynamic Traffic Assignment (DTA) models are therefore often believed to be more
accurate, and are increasingly used to estimate network-wide diversion impacts as
a result of major work zones, bridge closures, or emergency evacuation events.

https://www.fhwa.dot.gov/planning/processes/metropolitan/planning_practices/

26
 Traffic Assignment
6.5 Sample Network Application for Traffic Assignment

Figure 6.18 Sample network: traffic assignmentstep.

27
 Traffic Assignment
5. Sample Network Application for Traffic Assignment
▪ Turning back to the sample network introduced in the trip generation and trip
distribution steps, the traffic assignment now assigns the origin – destination matrix
to the physical paths taken by the trips between zones.
▪ Figure 6.18 shows a sample network with 08 internal TAZs (part of the study area)
and 04 external TAZs (not explicitly being studied, but contribute to trips in and out
of study area).
▪ The figure further shows a roadway network with classification into principal
arterials/freeway, principal arterial/surface street, and minor arterials. Collector
streets and local roads are omitted from the network in a planning context in this
example, but can be included if they carry significant amounts of traffic. As before,
we will focus on the trips originating from zone 1.

28
 Traffic Assignment
5. Sample Network Application for Traffic Assignment
▪ The figure shows the estimated flow between zone 1 and the 13 remaining zones,
with the line thickness. Notice that in this example several zone-to-zone trips use
the same links, such as for trips to zones 6, 7, and 13.
▪ Trips originating from other zones are not shown, but those would ultimately (finally)
all add to the estimated load of trips on each link in the network by superimposing
(combine) them on the zone 1 trips.
▪ If this example illustrates the results of an all-or-nothing assignment, the analyst
may now introduce a capacity constraint, which may further distribute those trips.
Other advanced assignment algorithms can also help to distribute the trips for each
zonal pair across multiple links.

https://www.fhwa.dot.gov/planning/processes/metropolitan/planning_practices/

29
 Traffic Assignment
6.5 Sample Network Application for Traffic Assignment

Figure 6.18 Sample network: traffic assignmentstep.

30
https://www.fhwa.dot.gov/planning/process
es/metropolitan/planning_practices/

31