Skip to contentUnited States Department of Transportation - Federal Highway Administration FHWA Home
Research Home
Report
This report is an archived publication and may contain dated technical, contact, and link information
Publication Number: FHWA-RD-98-166
Date: July 1999

Guidebook on Methods to Estimate Non-Motorized Travel: Supporting Documentation

2.9 Bicycle Travel Models: QUOVADIS-BICYCLE

Demand Estimation

Descriptive Criteria: What is It?

Categories:

Box with an x insideBicycle Empty BoxPedestrian Box with an x insideFacility-Level Box with an x insideArea-Level

Authors and Development Dates:

DHV Environment and Infrastructure (no date)

Purpose:

QUOVADIS-BICYCLE is a bicycle network model developed by DHV Environment and Infrastructure of the Netherlands. The model can be used to simulate the effects of changes in the bicycle network on the distribution of flows over the network. It can also be used to simulate the effects of changes in socioeconomic characteristics on the generation and distribution of bicycle travel, and can be used to calculate various measures of accessibility and safety. The model is primarily a distribution model and is not intended to forecast changes in total bicycle travel as a result of network improvements.

Structure:

QUOVADIS-BICYCLE is based on QUOVADIS-CAR, an automobile network model, and bears many similarities to standard four-step transportation planning models. The primary exception is that the mode choice step is excluded. The basic steps are as follows:

1. Create a schematic representation of the bicycle network.

2. Calculate trip generation for each zone.

3. Distribute trips between zones.

4. Assign trips to the network.

5. Calibrate the model based on actual counts.

6. Run the model based on future socioeconomic and/or network conditions.

Steps 1 through 3 are repeated for four trip purposes (home-work, home-school, home-shop, and other) for three time periods (7 a.m. - 9 a.m., 9 a.m. - 4 p.m., and 4 p.m. - 7 p.m.). This produces 12 origin-destination trip tables, which are then aggregated into an overall trip table that is assigned to the network.

The details of the steps are as follows:

1. Create a schematic representation of the bicycle network. QUOVADIS-BICYCLE can import network data from car network models including Quovadis-CAR, Tranplan, TRIPS, and Proset, as well as ASCII data. The network can then be modified for bicycle purposes. The network consists of three elements: roads and streets (links), intersections, and connections of zones to the road network. Links can be assigned any of nine road categories (bicycle lane, bicycle track, mixed traffic, etc.) and different bicycle travel speeds can be assigned to each category (15 km/h is used as the default). Intersection delay (waiting time) is calculated based on car traffic, as read from the imported car traffic network. (The user can enter exceptional delays manually.) Speeds and distances for each zone's connection to the network can also be entered.

2. Calculate trip generation for each zone. Trip generation is based on bicycle trip rates derived from analysis by the Central Bureau of Statistics of the Dutch National Household Survey and local surveys. Default rates by time period and trip purpose have been developed based on the number of residents, pupil places at schools, shop employees, and other employees. Alternative data on trip generation rates can also be entered by the user, if such data are available.

3. Distribute trips between zones. Trips are distributed between zones using a resistance function based on the time required to travel by bicycle from one zone to the other. The resistance functions differ by the four trip purposes and are based on the actual distributions of bicycle trip distances as (presumably) determined from the national travel survey. Distances are converted to travel times using an assumed bicycling speed of 15 km/h, based on the shortest distance paths between zones. External trips (as determined from cordon counts) must be distributed manually. The option of intrazonal trips can also be specified, based on assumed intrazonal distances for each zone. These trips are not loaded onto the network.

4. Assign trips to the network. Assignment can be performed using either an all-or-nothing approach based on least-cost (travel time) paths or by stochastic assignment based on travel times. Interzonal travel times are based on link travel speeds plus intersection delays. There is also an option for selective assignment of traffic to certain links, paths, or screen lines.

5. Calibrate the model based on actual counts. Actual counts of bicycle movements can be used to adjust the O-D trip table. The calibration process is iterative. Counts used for calibration can be weighted based on the assumed reliability of the counts. The result of the calibration process is a correction matrix by which O-D tables from the distribution phases of future year plans should be multiplied.

6. Run the model based on future socioeconomic and/or network conditions. Once the model has been calibrated, it can be run with future socioeconomic data and/or an assumed future network to predict the distribution of traffic over the network under future conditions.

Calibration/Validation Approach:

See step 5 above.

Inputs/Data Needs:

The following data are required:

1. Socioeconomic data (number of residents, pupil places or students, shop employees, and other employees) on a zonal basis.

2. Bicycle trip rates by trip purpose and time of day.

3. Bicycle counts: internal counts for testing and calibration of the model, and cordon counts if external traffic is to be considered.

4. Network data. A basic bicycle network can be constructed from the existing road network used for modeling purposes (including car traffic volumes). Modification of this network requires some additional data collection on link characteristics relevant to bicyclists.

Potential Data Sources:

Some of the data sources and formats required for the model might not be readily available in the United States. In particular, data on pupil places is not a standard travel model input in most areas. Also, bicycle trip rates by trip purpose and time of day were obtained from a national travel survey in the Netherlands, and comparable data might not be available in the United States.

Computational Requirements:

Quovadis-BICYCLE can be run on a microcomputer on the MS-DOS platform and is currently being updated to run in a Windows 95 environment.

User Skill/Knowledge:

Some knowledge of travel demand modeling techniques is required.

Assumptions:

Bicycle trip generation rates are essentially assumed to be constant across zones, controlling for population and employment levels, and are based on current bicycle trip generation rates as determined from a national travel survey.

Facility Design Factors:

Since the model assigns trips based on the shortest travel time route, the model is capable of evaluating the impacts on the zone-to-zone distribution of trips and on the assignment of these trips to the network, as a result of new facilities or improvements to existing facilities

which decrease travel time. Travel time on existing facilities can change if the type of facility changes, as long as different speeds have been specified according to facility type. Travel times can also change if measures are implemented to reduce bicyclist delay at intersections (delay reductions must be entered manually).

Conceivably, other factors which influence route choice could be included by assessing penalties to link speed or intersection delay, although this would require assumptions about the bicyclist's tradeoff between speed and other link attributes.

Output Types:

The output of the model includes forecast bicycle flows by facility/link. The number of existing and future bicyclists on each link can also be identified by time period and trip purpose.

The accessibility and safety modules of the model can further be used to:

  • Calculate the extra distance (detour distance) bicyclists have to travel as compared to straight-line distance;
  • Track and graph bicycle accidents;
  • Assess whether bicycle paths are desirable from a safety point of view;
  • Trace unsafe crossings (long-waiting times combined with number of bicyclists and number of accidents);
  • Determine the effects on traffic safety and detour distance of adding a bicycle path.

Real-World Examples:

In addition to applications in the Netherlands, QUOVADIS-BICYCLE has been applied by Allott Transportation to Ipswich, UK. Modeled trip patterns were compared to actual trip patterns and to use of a conventional desktop and field study for identifying a bicycle network. While the model overpredicted bicycle flows, the particular routes identified as part of the bicycle network remained broadly the same. (Department of Transport, 1995).

Contacts/Source:

QUOVADIS-BICYCLE was developed in the Netherlands by DHV Environment and Infrastructure for the Dutch Ministry of Transport. Contact Dick Rooks, DHV Environment and Infrastructure, Laan 1914 No. 35, P.O. Box 1076, 3800 BB, AmersFoort, Netherlands.

Publications:

DHV Environment and Infrastructure. QUOVADIS-BICYCLE User's Manual. (no date)

Department of Transport. Traffic Advisory Leaflet 8/95: Traffic Models for Bicycling. London, UK, 1995.

A bicycle box at a traffic signal in Groningen, Netherlands.
Figure 2.9 A bicycle box at a traffic signal in Groningen, Netherlands.
The box allows bicyclists to wait in front of motor vehicles.

Evaluative Criteria: How Does It Work?

Performance:

See "Real-World Examples."

Use of Existing Resources:

This approach builds on existing travel modeling efforts, and primarily utilizes existing data on travel behavior that must be collected for these efforts. Collection of additional bicycle count data is required, as is modification of the road network to include bicycle facilities and characteristics. The resources required for network modification and conducting bicycle counts increase in proportion to the accuracy desired for modeling purposes.

Travel Demand Model Integration:

Quovadis-BICYCLE is structured like a traditional travel demand model. It can import network and/or socioeconomic data from car models including Quovadis-CAR, Tranplan, TRIPS, and Proset. However, it is not fully integrated with network modeling for other modes; in particular, a mode choice component is lacking.

Applicability to Diverse Conditions:

Trip data and the network model must be developed specifically for the area being modeled. Once the model system is developed, proposed modifications to the local network can be tested.

Usage in Decision-Making:

No information available.

Ability to Incorporate Changes:

Trip data and the network model must be developed specifically for the area being modeled. Once the model system is developed, proposed modifications to the local network can be tested.

Ease-of-Use:

Not evaluated. The software can be run in an MS-DOS environment.

Comments:

This model bears many similarities to the TRIPS bicycle model (described in Section 2.10) in terms of its purpose and structure. Primary differences include the methods used to estimate the base bicycle trip table; the segmentation of trips by purpose and time of day in QUOVADIS; ability to incorporate link-specific travel speeds in TRIPS; the treatment of intersection delay in QUOVADIS; and the ability of QUOVADIS to track accident statistics and calculate accessibility and safety measures. However, TRIPS has advanced recently by integrating mode choice into the modeling process.

 

FHWA-RD-98-166

Previous | Table of Contents | Next

ResearchFHWA
FHWA
United States Department of Transportation - Federal Highway Administration