-453-
System Dynamics in Urban Transportation
Planning and Policy Analysis
Indra K. Khanna, N. Singh and Prem Vrat
Indian Institute of Technology,Delhi.
ABSTRACT
A simulation model of the passenger transportation system is presented. The
model has been built in order to carry out a series of simulation experiments.
The purpose of these experiments is to compare the effects of same transpor-
tation policies on road congestion, modal-split, air pollution and transporta~
tion fuel consumption. System dynamics principles have been used for simulating
the model. The statistics of Delhi urban area have been used to calibrate the
model.
INTRODUCTION
The population of Delhi urban area has been growing at a fast rate. In 1971,
Delhi had an urban population of 36.47 lakhs and in 1981 it had grown to 57.68
lakhs, and is expected to reach a staggering figure of 122 lakhs by 2001. Ex-
pansion in Government employment and the industrial base of the Delhi area have
been the major stimuli for rural to urban migration (Gupta 1982, pp.21-28). In
addition, an estimated million people come to Delhi every day from the satellite
towns. Out of this million strong floating population, about two lakh persons
settle down every year, thereby adding to the already strained urban passenger
transportation system of Delhi.
The mass transportation facilities in Delhi are provided by buses and to a
lesser extent, by ring rail. The dependence on public transport in Delhi is
much less as campared to Bambay, Calcutta, and Madras. Other modes of pass-
enger transportation in Delhi, include motor-cars, taxis, scooters, auto-~
rickshaws and bicycle rickshaws, horse-driven tongas and bicycles. The
heterogeneous nature of traffic accentuates the transportation problem by
increasing road-congestion, which also increases transportation fuel con~
sumption and air pollution.
PURPOSE
The purpose of this study is to construct a Siilation model of the trans-
portation system and use it as a policy analysis tool. The study has been
carried out in close cooperation with the experts connected with transpor-
tation planning. The model has been so developed that policy alternatives
as suggested by the planners, can be tested.
~454-
The initial data inputs to model are of two kinds. Firstly, largely quali-
tative information and camments obtained from experiencedtransport planners
Secondly, quantitative published data obtained from various sources. Values
of same parameters have been supplied exogeneously.
SELECTION OF THE MODELLING STRATEGY
Traffic and transportation planners must deal with a very complicated system
with human, economic, physical and technical factors involved. It is possible .
to set up a wide range of different models depending upon which sub-systems are
considered important to the problem in question(Holst 1977, p. 7). The models
can be divided into two main groups, simulation models and optimization models.
In each group several different modelling and solution techniques can be applied
e.g. static or dynamic models, different algorithms, etc.
System dynamics offers us a methodology for understanding certain types of
complex problems (Richardson 1981, p. 1) as are encountered in today's
transportation systems. It focusses on the structure and behaviour of systems
composed of interacting feedback loops. The structure defines how the
variables interact (pidd 1984, p. 187). Casual diagrams allow the analyst
to quickly communicate the structural assumptions underlying his model
(Goodman 1974, p.5). They help to identify the feed back structure of
systems as a prelude to analysing the behavioural characteristics of the
feedback systems . In general a system dynamics study has a two-fold
objective (Coyle 1977, p. 19).
1. Explaining the systems behaviour in terms of structures and policies.
2. Suggesting changes to structure, policies, or both, which will lead
to an improvement in the behaviour.
It is because of the above mentioned characteristics that system
dynamics methodology has been chosen to simulate and analyse the trans-
portation system, in this study.
THE MODEL
For the purpose of analysis, the model is divided into four main sectors:
Socio-econamic , passenger transportation, environment and energy (Tran 1979}
In the socio-econamic sector the atributes of urban area, such as population,
industries, houses, employment and land-use are modelled. In the transportation
sector, buses, personalized modes of transport and rail transit are modelled.
In the environment sector, air pollution, and in the energy sector , fuel con-
sumption are modelled. These four sectors interact with each other and the
sectoral interactions form the main feedback loops. The feedback loops of the
transportation sub-system are shown by influence diagrams, in the next section.
TRANSPORTATION SUB-SYSTEM
The main attributes of the transport sub-system are the roads, buses, personalized
modes, and ring rail. Figure 1 represents the bus transit activities of the
transportation sub-system.
-455~
wat Mag ip L
FOR
BUSES
TRANSPORTATION
() AVAILABILITY
+
+ _—_ an
BUSES * gusES
PURCHASED BUS
+ : RIDERSHIP
(3) +
_ Bus {2)+
EXPENSES ay
+
£4]
BUS
NSB FARE
FIG- 1 BUS-TRANSIT FEEDBACK LOOPS
The positive loop 1 represents an explosive growth in the number of buses,
with an increase in transit travel demand.
As number of buses increase, bus ridership will go up, generating more
revenues. This will further increase the number of buses as shown by
positive loop 2.
However, the positive loop 1 and 2 are counteracted by negative loop 3.
With the increase in the number of buses, bus expenses will go up, thus
reducing the bus fund.
Negative loop 4 shows the effect of bus fare on bus revenues and bus fund.
Figure 2 represents the feedback loops of the personalized mode usage in
the transportation sub-system.
Negative loop 1 shows the effect of traffic density on the average vehicle speed,
Positive loop 2 shows the relationship between
wna transit’ eri p personalized vehicle trips and
-456-
AVERAGE TRIP
HeNoTn VEHICLE 6 ENERATION™
OCCUPANCY
| TRANSIT
= a OO~N . CAPACITY
TRAFFIC + +
DENSITY PERSONALIZED
VEHICLE TRIPS —.
ou 2)
TRANSIT
ta) TRIPS
<= AVERAGE
VEHICLE x”
SPEED
FIG.2 FEEDBACK LOOPS FOR PERSONALIZED
MODE OF TRANSPORT
Figure 3 represents the feedback loops of the ring rail transit.
RAIL TRANSIT
RAIL TRANSIT RIDERSHIP
a EXPENSES \/ NN
RAIL RAIL— TRANSIT
TRANSIT _DEMANO
FUND (*)
re
RAIL
TRANSIT
<4 RAIL-TRANSIT
— FARE
FIG-3 RING-RAIL TRANSIT FEEDBACK LOOPS
The positive loop 1 represents the growth of rail ridership with decrease of
rail fare.
The negative loop 2 represents the decline of rail transit fund with decrease
in rail fare.
The negative loop 3 shows the effect of declining rail transit fund on the
rail transit services available.
~457-
MODEL CALIBRATION
The base year chosen for initializing the simulation model is 1971. The
date from 1971 to 1981 has been used to calibrate the model. Figure 4 shows
the actual and simulated values of some of the parameters used for calibration.
Actual value Simulated value
Population 57.68 lakhs 56.48 lakhs
Buses 3488 3502
Bus ridership 30.96 lakhs/day 31.66 lakhs/day
Transit modal split 0.60 0.58
FIGURE 4
COMPARISON OF ACTUAL AND SIMULATED VALUES FOR 1981
The model, so calibrated, has been used to project the values upto year 2001.
POLICY ANALYSIS
An improvement of urban transportation system seems unlikely to occur unless
new policies are implemented. In this study three different policies have been
used to simulate the behaviour of the system, beyond 1985.
POLICY 1
Petrol price has been increased by 15% by Central Government in the year 1985.
The first policy analysis relates to this price hike. Due to this price in-
crease the generalized cost of travel by personal vehicle will increase. This
will result in a decreasing auto modal split fraction as shown in figure 5.
The mass transit systems will take the diverted trips, resulting in an increase
of mass transport modal split. This shift from private vehicle trips to mass
transit trips also results in a decrease in highway traffic density.
The effects of this policy on fuel consumption are shown in figures 6 and 7.
This policy is expected to bring about the following changes by 2001,
Petrol consumption 8.74 % decrease
Diesel consumption 0.54 % increase
Pollution 2.26 % decrease
POLICY 2
In the second policy run the car parking fee has been raised to rupees
10/- from the present rupees 2/-, in addition to the increased fuel price.
As shown in figure 5 this policy will result in a further decrease in private
vehicle usage. The diverted trips will be taken up by the mass transit system.
A=AMSFO-30 0.315 0.330 0345 0360 0.375 0.390 0.405 0.420 0.435 0.450
B= BMSFO.50 0.5) 0.62 0-53 0854 0.55 0.56 0.57 0.58 0.59 0.60
R= RMSF.035 04/5 .0480 .0545 .0610 .0675 0740 0805 .0870 .0935 .10
H=HTD 10 ‘ 63.5 81.3 99.1 7 135 153 170 188
1971 T
BASE RUN ———
PoLicy 1 ——~
1976 PoLicy 2
POLICY 3
1981
&
1986) a
t
i991 ‘
SS
»~
1996 R
|
Yo
7 SS]
200i Fic
FIG.5 MODAL SPLIT AND HIGHWAY TRAFFIC DENSITY
7.50 E07 3.00E08 5.25 £08 7-50E08 9.-75E08 1.20E09
1971 T T T T
I976- BASE RUN ——
POLICY 1 ai
POLICY 2 Se
POLICY 3 se
i981
(986 &
S
1
{991
ig96-
2001
FIG-6 PETROL CONSUMPTION (LITRES)
2.00E08
1971
1976
196)
i986
i991
2001
9.60 E08 1.72 E09 2.48E09 3.24E09 4.00E09
T
T T T t
BASE RUN ——
POLICY 1 ar
PoLIcCY 2. —-—
PoLicy 3) —--—
FIG.7 DIESEL CONSUMPTION (LITRES)
-09%-
~461-
However, the modal split changes are not very significant. The change in
traffic density is also very small.
Figure 6 and 7 show the effects of this policy, on fuel consumption. With
reference to the base run values, the following changes are expected by 2001.
Petrol consumption 9.71 % decrease
Diesel consumption 0.58 % increase
Pollution 2.88 % decrease
POLICY 3
In addition to changes in the first two policies, the bus fares have also
been revised upwards, with effect fram 1985, in this policy run. The
present and proposed bus fare structures are as follows:
Distance Present fare Proposed fare
Upto 4 km 0.30 rupee 0.40 rupee
over 4 km, upto 12 km 0.40 rupee 0.50 rupee
Over 12km, upto 16 km 0.40 rupee 0.75 rupee
Over 16km, upto 20 km 0.50 rupee 1.00 rupee
Over 20mk 0.75 rupee 1.00 rupee
Due to the bus fare increase, the generalized cost of bus travel increases.
As the rail fare is not increased, the mass transit trips are expected to
shift towards ring rails, as seen in the figure 5. This will result in an
increased rail modal split fraction, and a sight decrease in bus modal split
fraction.
OTHER POLICIES
Some other policies should also be analysed. Ring rail, at present charges
a flat fare of 1 rupee per trip. The effect of reduction in ring rail fare
and relating it with trip length can also be studied. Another policy which .
can be analysed is-to reduce the number of private vehicles on roads, say
by half. This can be done by allowing only vehicles with even registration
numbers, on roads, on even dates and with odd registration numbers on odd
dates. This will result in reduction of peak hour traffic density. For such
a policy to work an efficient public transport system is a must.
CONCLUSIONS
A continuous system simulation model has been used to investigate the
effectiveness of a number of policy changes to reduce the traffic congestion,
fuel consumption and air pollution generation.
The policy changes examined are by no means a complete set of changes which
might have been studied. Although this has enabled to provide insight regard-
ing the structure and operations of the urban transportation system . Worthy
~462-
suggestions for policy alternatives were difficult to obtain. Some more policies,
in addition to those suggested above should also be analysed, before a specific
recommendation can be made.
REFERENCES
Coyle, R.G. Management System Dynamics. New York: Wiley, 1977.
Goodman, M.R. Study Notes in System Dynamics. Cambridge: Mass. Wright-Allen
Press, 1974.
Gupta, J.D. " Regional Land Use and ‘Transportation Development Optimization
Model for Delhi Region of India", Transportation Research Record, 848,
pp. 21-28.
Holst, O. A Survey of Traffic and Transportation Models, Lingby, IMSOR, 1977.
Pidd, M. Computer Simulation in Management Science Chichester, Wiley, 1984.
Richardson, G.P., and A.L. Pugh III. Introduction to System Dynamic Modelling
with Dynamo. Cambridge, Mass. MIT Press, 1981.
Tran, T.K. An Urban Transportation Policy Planning Methodology: System Dynamics
Approach. Blacksubrg, Virginia, 1979.
