CVEN 2401/2402 Workshop: Week 4

Trip Distribution

This week we will review the second step, trip distribution of the 4-step planning process.

Trip Distribution
Trip distribution is the 2nd step of the 4-step planning process; the focus of this step is to convert the
attraction and production levels at each zone (node) and convert them into origin-destination (OD)
trips. The main idea behind the models for trip distribution is that we can develop methods which will
split the number of trips produced (or attracted) among the available destinations (origins).
The model we will focus on today is the Gravity Model. This model is based on Newtons law of
universal gravitation:
1 2
=
2
To adapt this concept to trip distribution, we make the assumption that rather than using the masses of
the objects (1 , 2), we will use the trip attraction and distribution factors; instead of using the
distance ( 2 ), we will use some function that represents the difficulty of getting from origin to
destination, i.e., some function of the origin-destination travel cost ( ). Finally, instead of using the
gravitational constant, we will use a constant (or 2) which will help us calibrate the model so that the
"right" number of trips are assigned (we will discuss what "right" means).

Question 1

Show why the correction factor, =
, preserves the number of trips coming out of
[ ]
( )

each origin.

CVEN2401/2402 Workshop Week 4: Trip Distribution/ Mode Choice 1

Solution
To prove this, we must show that

= =
( )

since the term and are constant for all , we can move it out of the summation, leaving:


( ) ( )
= = =
( )
[ ] [ ]
( ) ( )

1
To simplify the formulation, we can let the term ( )
be equal to which is a friction factor
between each origin and destination pair. Therefore, substituting in the friction factor, the gravity
model for trip distribution can be expressed as;
1
= =
( )
[ ]
( )

Gravity Model prediction

The production constrained model is used in situations where we are much more confident on the
number of trips produced than the number of trips attracted (which is common). While the number of
people in a household is normally fixed, the number of people in a workplace may change depending
on clients and other visitors, and the size of the workplace in terms of employees may change much
more rapidly (and randomly) than that of a household.
In the production constrained model we first assume that the number of trips produced at each origin
will be exactly the number of trips that leave each origin; however, we do not place this same
restriction on trips attracted/number of trips that arrive at each destination. Having made this
assumption, the model then predicts that the number of trips from each destination to each origin will
be given by the following equation:
1
=
( )
[
( )]
1
The term
which we will refer to as , is a factor associated with each origin, which is equal
[ ]
( )

to the inverse of the sum of the expression ( )
over all destinations. By using this particular factor,
we ensure that the total number of trips coming out of each origin is preserved, and we split the trips

coming out of a specific origin in a manner that is proportional to the value (
)

The production constrained trip distribution model only ensures that trip productions match initial
forecasts. Thus, in order to preserve the production and attraction forecasts, row and column factoring
is conducted until a given threshold is reached for RF and CF:

= and =

CVEN2401/2402 Workshop Week 4: Trip Distribution/ Mode Choice 2

Question 2
Using the gravity model
Following the traffic generation assessment of the newly proposed suburb Freeflow, The Traffix
Municipality has hired you to forecast the trip distribution of the entire Municipality using the
following data where the region has been separated into 3 zones (shown in Figure 1).

Figure 1: Zone Division Traffix

The expected productions and attractions for the zones are given in the following table:
Zone 1 2 3 Total
Productions 250 450 300 1000
Attractions 395 180 425 1000

The expected travel times between the zones are described using the following matrix:
Zone 1 2 3
1 6 4 2
2 2 8 3
3 1 3 5

The relationship with the travel time and the friction factor has been empirically derived as shown
below;
Travel
1 2 3 4 5 6 7 8 9 10
Time
Friction
Factor 82 52 50 41 39 26 20 13 9 5
value

Determine the trip distribution throughout Traffix using row and column factoring approach

Solution
The intention is to determine the zone-to-zone trips and fill out the following trip distribution matrix.
Remember that

CVEN2401/2402 Workshop Week 4: Trip Distribution/ Mode Choice 3

 Destinations Productions
1 2 3 ( )

1 250

Origins 2 450

3 300
Attractions
395 180 425
( )

=

Friction Factor Table;
Zone 1 2 3
1 26 41 52
2 52 13 50
3 82 50 39

250 395 26
11 = = 64.59
(395 26 +
180 41 +
425 52)

Iteration 1:

1 2 3
1 250 395 26 250 180 41 250 425 52
11 = 12 = 13 =
(395 26 + (395 26 + (395 26 +
180 41 + 180 41 + 180 41 +
425 52) 425 52) 425 52)
= 64.59 = 46.42 = 138.99
2 450 395 52 450 180 13 450 425 50
21 = 22 = 22 =
(395 52 + (395 52 + (395 52 +
180 13 + 180 13 + 180 13 +
425 50) 425 50) 425 50)
= 209.45 = 23.86 = 216.69
3 300 395 82 300 180 50 300 425 39
31 = 32 = 33 =
(395 82 + (395 82 + (395 82 +
180 50 + 180 50 + 180 50 +
425 39) 425 39) 425 39)
= 167.64 = 46.58 = 85.78

CVEN2401/2402 Workshop Week 4: Trip Distribution/ Mode Choice 4

 Destinations Observed Estimated
1 2 3 Productions Production

1 65 46 139 250 250

Origins 2 209 24 217 450 450

3 168 46 86 300 300

Estimated
442 116 442 1000 1000
Attractions
Observed
395 180 425
Attractions
(395/442)= (180/116) (425/442)

0.89 =1.55 =0.96

After the first iteration, the production values match but the attraction values differ greatly from the
observed attraction. Therefore, in the table above, the column factors ( have been computed to
complete the second iteration.

Iteration 2:

1 2 3
1 11 = 65 0.89 12 = 46 1.55 13 = 139 0.96
= 58.088 = 71.38 = 133.65
2 21 = 209 0.89 22 = 24 1.55 22 = 217 0.96
= 186.78 = 37.24 = 208.65
3 31 = 168 0.89 = 150.14 32 = 46 1.55 33 = 86 0.96
= 72.93 = 82.69

Destinations Observed Estimated

1 2 3 Productions production

1 58 71 134 250 263 0.95057

Origins 2 187 37 209 450 433 1.039261
3 150 72 83 300 305 0.983607
Calculated
395 180 425 1000
Attractions
Estimated
395 180 425
Attraction

After the second iteration, the attraction values match, but the production values differ greatly from
the observed production. Therefore, in the table above, the row factors ( have been computed to
complete the third iteration.

CVEN2401/2402 Workshop Week 4: Trip Distribution/ Mode Choice 5

Iteration 3:
1 2 3
11 = 58 0.95 13 = 134 0.95
1 12 = 71 0.95 = 67.49
= 55.13 = 127.38
21 = 187 1.04 22 = 37 1.04 22 = 209 1.04
2
= 194.34 = 38.45 = 217.21
32 = 72 0.98 33 = 83 0.98
3 31 = 150 0.98 = 147.54
= 70.82 = 81.64

Destinations Observed Estimated

1 2 3 Productions production

1 55 67 127 250 250

Origins 2 194 38 217 450 450

3 148 71 82 300 300

Calculated
395 180 425 1000 1000
Attractions
Estimated
395 180 425
Attraction

CVEN2401/2402 Workshop Week 4: Trip Distribution/ Mode Choice 6