- Introduction
- Methodology of Developing a Transport Demand Model
- Estimation of Transport Demand for the Target Year
- References on Transport Demand Forecast Modelling

The modelling technique provides a quantitative and scientific approach to improving the mobility of people. The approach enables an integrated planning of urban land use and transport systems. The detailed methodology for transport demand modelling is described in a number of references. Several major references are shown at the end of this Annex.

It is recommended, however, that the use of the modelling technique should be limited to project evaluation purposes, rather than plan generation. In the past, many studies made a common mistake in that modelling approaches were used to generate extensive infrastructure projects, such as highways or flyovers, without considering a comprehensive range of options, including the use of public transport systems, NMTs, or various traffic management and/or demand management measures.

It is also proposed that a simplified model should be developed for the purpose of CMP development. As widely recognized, modelling requires substantial cost, workforce and time. Though it is of course true that a detailed model is expected to provide more accurate results, considering the CMP objective to develop a long-term vision and goals for entire city development, the simplified model can provide adequate results with a shorter period and lower cost. Therefore, the following description shows the development methodology for a simplified model as part of CMP preparation.

Page topThe overall methodology for a transport demand model comprises two major steps: 1) Developing a Base Year Transport Demand Model, and 2) Developing a Future Transport Demand Model. The following sections show the details of each step.

Figure 1 Work Flow to Develop Transport Demand Model

As the initial step in the demand modeling, a base-year model may be developed. The methodology for model calibration is given in Figure 2.

Figure 2 Base-Year Transport Demand Model Calibration

Preparation of Base-Year Origin-Destination (O-D) Matrices: The trip generation/ attraction, distribution and modal split can be estimated simultaneously in the base-year model. Person-trip O-D matrices by vehicle type are directly estimated from the O-D survey mentioned above, with the number of person-trip generations and attractions in each zone divided by the sample rate in each zone.

The number of zones is typically around 20 to 40 for the simplified model and a single zone is expected to have around 50,000 – 100,000 population. To estimate modal spilt, the major modes must be classified, but the workload of this step will increase significantly as the number of modes increases. For Indian cities, it is recommended that the trips by the following modes be estimated separately (other modes such as motorcycle or taxi can be included depending on the situation in each city):

- passenger vehicle;
- auto and cycle rickshaws;
- bus;
- truck; and
- bicycle.

The main data required and outputs are as follows:

- Required Input Data
- The number of trips between each zonal pair by vehicle mode, from the results of O-D surveys
- Sample rate in each zone
- Output
- Person-trip O-D matrices by vehicle mode (i.e., private vehicle including passenger car, motorcycle, and bicycle, and public transport)

Preparation of Base-Year Transport Network Data: The transport network data should include the road and public transport networks. For the simplified model, the network should include only trunk routes, such as national and state highways, and roads with high traffic capacity or volume. The network data should consist of a number of link data, including the following:

- coordination;
- traffic capacity;
- free flow speed; and
- road regulations (e.g., one-way, heavy truck ban)

Traffic Assignment: With O-D matrices and road network data, traffic volumes in each road section will be estimated. Two types of O-D matrix will be used: person-trip O-D and vehicle-trip O-D matrices in passenger car units (PCUs), obtained from person-trip O-D matrices divided by both average number of passengers in each vehicle type and the PCU factor for each vehicle type. While assignment of person-trip O-D will show the person travel pattern and major corridor demand in person-trip units, assignment of vehicle-trip O-Ds will show the Volume-Capacity Ratio (V/C ratio) in each road link, indicating the level of congestion and identifying road network bottlenecks. Road network data can be obtained from the road network inventory described above. Required input data and outputs follow:

- Required Input Data
- Person-trip O-D matrices by vehicle mode
- Average number of passengers in each vehicle type
- PCU factor for each vehicle type
- Road network data
- Output
- Traffic volume on each road section
- Vehicle/Capacity ratio on each road section
- Average travel time on each road section

Calibration of the Model: Calibration of the traffic model is essential to ensure accuracy and reliability. The results of the assignment step in the base year should be compared with actual observed traffic count data, such as screen line and cordon line data. The parameters should be adjusted to provide a better match. These parameters will be used for the model in the target year.

Page topDevelopment of a transport demand model is an important task in the CMP. The base-year model will be developed based on the surveys conducted above and it should be calibrated using the existing traffic flow. The model for the target year will then be developed, with the future transport network and land use scenarios. After the development of the model, alternative scenarios and projects will be evaluated with the transport demand model.

A flowchart for developing the model is shown in Figure 3.

Figure 3 Work Flow to Develop Transport Demand Model

The transport demand model includes the following four sub-components:

(i) Trip generation/attraction: This step estimates the volume of person-trips generated/ attracted to/from each zone.

(ii) Trip distribution: The trip distribution step calculates the number of person-trips travelling between zones and develops an O-D matrix.

(iii) Modal split: The modal split step estimates the number of trips by each mode and provides vehicle O-D matrices for each mode.

(iv) Assignment: The assignment step distributes O-D trips onto the road network and estimates the traffic volume on each road.

Page topStep 1: Trip Generation/Attraction: A function that estimates future person-trip generation and attraction should be prepared through a regression analysis from the number of base-year person-trips and explanatory variables, such as population, number of employees and number of non-employees. The function should be estimated by trip purpose.

Next, future trip generation/attraction in each zone are estimated through the function described above, using socio-economic data such as forecast population and employment, which are to be clarified for the future land use scenario described in the previous section.

Required data and outputs are as follows:

- Required Data
- Socio-economic data for each zone
- Trip generation/attraction function
- Output
- Number of trip generations/attractions in each zone

Step 2: Trip Distribution: The number of trips between zonal pairs is estimated from the volume of trip generation/attraction in each zone. Among several methods of estimating trip distribution, the gravity method is most widely used. Required data and output are as shown below.

- Required Data
- Number of trip generations/attractions in each zone
- Trip distribution function
- Distance matrices between each two zones
- Output
- Person-trip O-D matrices for target year

Step 3: Modal Split: Using the modal split scenario and the average number of passengers carried by each vehicle type, the O-D matrix for each vehicle type is estimated from the person-trip O-D.

Required data and output are as follows:

- Required Data
- Modal split
- Trip distribution function
- Distance matrices between each two zones
- Output
- Person-trip O-D matrices for target year

Box 1 summarizes the modal split models.

Box 1 Modal Split Models

There are two major types of model to calculate the modal split rate: aggregate and disaggregate:

- Aggregate Model: An aggregate model uses aggregated data, such as modal choice rate by trip distance based on the O-D survey. An aggregate model can be developed relatively easily, but is not suitable for evaluating specific public transport projects, such as MRT projects, since it does not forecast modal shift from private vehicles to public transport. The future modal share can be provided from policy targets, although there is no certainty that it will be achieved.
- Disaggregate Model: The disaggregate model is based on individual preferences in modal choice. The disaggregate model requires additional data and specialized knowledge to develop and use the model. In particular, it is relatively complicated to develop a model for more than two modes. The advantage of a disaggregate model is that it can assess both modal shift from private vehicles to public transport and the benefits of a public transport project, such as an MRT.
An aggregate model is recommended for the CMP. However, a disaggregate model for a limited corridor should be developed for the Detailed Project Report of a major public transport project, such as an MRT project.

Step 4 Assignment: Using the same methodology as applied in the base-year model, the traffic volume on each road section should be calculated. Future vehicle-trips, person-trip ODs, and the road network will be required as input data.

Page top- Fundamentals of Transportation and Traffic Operations, Carlos F. Daganzo
- Highway Traffic Analysis and Design, R.J.Salter
- Traffic Assignment Techniques, Royal Thomas
- Traffic Engineering, Mc Shane
- Traffic Flow Fundamentals, Adolf. D.May
- Traffic System Analysis for Engineers and Planners, Martin Wohl
- Transportation Engineering and Planning (III Edition), C.S. Papacostos and P.D.Prevedours
- Transportation Engineering an Introduction, C.Jotin Khisty
- Urban Transportation Networks, Yusaf Sheffi

©2008 Asian Development Bank. All Rights Reserved. Legal Site Version 1.0