VISUM

VISUM is a comprehensive, flexible software system for transportation planning, travel demand modeling and network data management. VISUM is used on all continents for metropolitan, regional, statewide and national planning applications.

Designed for multimodal analysis, VISUM integrates all relevant modes of transportation (i.e., car, car passenger, truck, bus, train, pedestrians and bicyclists) into one consistent network model. VISUM provides a variety of assignment procedures and 4-stage modelling components which include trip-end based as well as activity based approaches.

PTV Vision VISUM 11 – What’s new

Improvements in VISUM 11
Detailed release information
Keep being updated: the PTV Vision newsletter

VISUM 11 is:

More user-friendly and efficient (new junction editor, procedures dialog)
More realistic (powerful new line blocking operation)
Faster (Multi-Threading assignments)
Consolidated (more compact data model (matrices, attributes))
Easier to extend (new Add-In-Architecture)
Groundbreaking in terms of a new Assignment method, called LUCE

Learn more about the new features of the must have VISUM 11...

HBEFA Emission Calculation in VISUM

Improved Environmental Assessments: Emission Calculation and Traffic Count Data Management

In a world of dwindling resources and changing climate transportation planning can no longer be a simple exercise of matching supply to demand. Intelligent solutions to today’s traffic problems need to demonstrate that they are environmentally sound in addition to providing the desired mobility. Recent legislation and the findings from environmental research define a framework against which the sustainability of a given planning scenario can be measured as quantitatively as the capacity from the traffic engineering perspective.

Among different measures of effectiveness, the traffic-related emissions and immissions of various greenhouse gases and other pollutants remain the focus of environmental assessments. In practical terms planners take the traffic volumes predicted by a demand model and compute first emissions per network link, then aggregate in space or feed the emissions into a second model for immissions.

The Approach

The basic approach for emission calculation follows a very straightforward equation:

Emission = Traffic Volume * Emission factor

In this equation “Emission” stands for the total mass of a pollutant species such as CO₂ or NO_x emitted by the vehicles on a network link during a given time interval. The traffic volume is the number of vehicles which traverse the link within that time interval. Demand models, like those developed in VISUM, yield traffic volumes, either as totals or disaggregated by vehicle type (e.g. cars, light trucks, heavy trucks). These volumes are then multiplied by emission factors, the “unit costs” in emission modelling. Obviously, emission factors are not constants, but functions of several factors. Speed, link type (e.g. motorway vs. city street), engine type, gradient, even temperature all have a significant impact. The functional relationship is determined through a vast amount of empirical work, measuring actual emissions in the field and in the lab taking a wide variety of driving patterns into consideration.

Why HBEFA 3.1?

The effort for such a research project is so great that it favours a supranational approach. More than ten years ago the environmental agencies of Germany, Switzerland and Austria pooled their resources to compile a comprehensive database of emission factors. The result was published as the Handbuch Emissionsfaktoren (HBEFA). After several revisions the HBEFA faces the next major revision. Not only will the emission factors be updated to take into account new engine concepts and emission standards, but the project partners will also include the findings from the EU-funded ARTEMIS research project on Transport Emission Models and Inventory Systems. ARTEMIS proposes a new set of systematic descriptors for traffic situations which simplifies mapping from transportation models to emission models. The HBEFA consortium also took on board several other European countries: among others Sweden, the UK and France are joining, so that the revised HBEFA is on its way to become a truly European standard for emission factors. With those recent developments in mind the VISUM development faced an easy choice when picking an emission model for integration with VISUM: the next release of HBEFA fits the bill perfectly.

Mapping the Fleet

How does the VISUM - HBEFA 3.1 link simplify life for the transportation planner? The first step is mapping the vehicle types used in the demand model to those used in HBEFA. Demand models typically distinguish only a very small number of different vehicle types (aka transportation systems), often only cars and trucks. This is much too aggregate for emission calculations. In fact, each transportation system represents a mix of many vehicle types which behave similarly in terms of demand modelling, but have very different emission factors. Fortunately, the providers of HBEFA take this aspect into consideration and offer ready-made mixed fleets for different countries and years as part of the database. For a national model of a region in Austria, the planner might then map the VISUM transportation system “Car” to the HBEFA standard fleet mix “Car Austria 2015”, actually composed of dozens of specific vehicle types. For applications in countries not covered by HBEFA or for specific technology scenarios users can take a standard fleet mix as a starting point and then modify the shares of the vehicle types.

Mapping Traffic Situations

Emission factors depend not only on the vehicle type, but also on the traffic situation in which the vehicle is operating. In terms of emissions coasting at 100 km/h on a motorway is dramatically different from stop and go in front of a traffic signal on an urban arterial. HBEFA captures these differences by the traffic situation descriptors recommended by ARTEMIS. Each traffic situation is defined by four descriptors, three of which relate to facility type and location (urban / rural, functional road class, free-flow speed). These three descriptors have an immediate counterpart in a VISUM model, as they are attributes of links (or link types). The fourth descriptor (LOS) is a qualitative four-step scale describing the traffic flow state from free to traffic jam. Finding a proxy for it in a demand model is less straightforward, but a classification of volume-capacity (v/c) ratio or of actual to free-flow speed ratio are natural choices. Guidance on threshold values for the four LOS classes is the subject of ongoing research, and users will be able to customise the relationship.

With the two mappings in place, emission calculation becomes a very easy post-processing of an assignment result from the demand model: for each network link VISUM disaggregates the volume to the HBEFA fleet mix, looks up the traffic situation and its emission factor, corrects for gradient, and multiplies the result.

Cold Start Emissions

All of the above relates to warm emissions and ignores one particularly important part of total emissions: those on the first few kilometres. Cold start emissions are often neglected or explained away, because ordinary traffic counts can be disaggregated by vehicle type, but obviously not by time since engine start. The picture changes completely if traffic volumes are taken from a demand model. Modern software packages such as VISUM store the full trajectory for each assigned trip, so by inverting this information it is indeed possible to tell which fraction of the volume on a given network link corresponds to vehicles within the first few hundred metres of their trip. After adjusting for cordoning effects near the boundary of the model, this information can be used to calculate from the model, instead of guesstimate, cold-start emissions – which may alter the total emission distribution significantly.

Link to Immissions

Emissions are often not the end of the processing pipeline. In order to compare the environmental impact with thresholds imposed by legislation, they have to be converted to immissions first. This conversion, taking into account dispersal around detailed 3D topography and air chemistry, is the realm of specialised software outside the transportation planning suites. The good news is that due to open GIS standards interoperability is not a problem, and in packages like VISUM it is a snap to export emissions per link such as ESRI Shapefiles to any GIS-enabled immission model.

Release Plan

The VISUM development team is now preparing the software for the integration of HBEFA and PTV is proud to announce that the link will be available very shortly after the official release of HBEFA 3.1.

: Data flow between VISUM and HBEFA

: Example of changes of the fleet composition in Sweden. Each colour represents a HBEFA vehicle type. The HBEFA database includes shares of the entire fleet for each year between 1990 and 2020.

LUCE: Fast algorithm for user equilibrium traffic assignment

In cooperation with Prof. Guido Gentile, one of the leading scientists in the field of traffic planning at the University of Rome, a new algorithm called LUCE (Linear User Cost Equilibrium) has been implemented in PTV Vision VISUM: LUCE represents a step change from the earlier path-based equilibrium assignment by improving runtime, convergence and path proportionality:

It runs much faster,
reaches higher levels of convergence
and loads a richer set of paths per OD pair.

New era of research

For decades the state of the art in equilibrium assignment was dominated by methods like Frank & Wolfe and by the path-based assignment in VISUM. But some years ago Hillel Bar-Gera’s work on Origin-Based Assignment started a new era of research into alternative methods. Based on these concepts VISUM now fully integrates an assignment algorithm called LUCE.

How does it work?

LUCE is an origin-based gradient method. Like all origin-based methods LUCE equilibrates volumes on all paths from one origin to all destinations simultaneously. The LUCE assignment procedure in VISUM uses optimized data structures called bushes for the storage of all efficient paths for one origin. The way bushes encode path volumes ensures that the paths for one origin satisfy the fair proportionality condition proposed by Bar-Gera and Boyce.

LUCE partitions the problem by origins. The main idea is to seek at each node a user equilibrium for the local route choice of drivers coming from the same origin among the arcs of its backward star. The travel alternatives that make up the local choice sets are the arcs that belong to the current bush. The cost functions associated with these alternatives express the average impedance from the origin to each intermediate node, linearized at the current flow pattern. The unique solutions to such local linear equilibria in terms of origin flows, recursively applied to each node of the bush in topological order, provide a descent direction with respect to the classical sum-integral objective function. The network loading is then performed through turn proportions, avoiding explicit path enumeration. Like all gradient methods, LUCE does not directly equilibrate travel cost along alternative routes, but instead uses the gradient of the cost function with respect to volume as an approximation of travel cost around the current solution. Capitalizing on the fast calculation of the derivatives, LUCE achieves a very high convergence speed.

How does it perform?

Discover the significant reduction of calculation times of path based assignments: The effects on run-time of the improved algorithm are striking. The chart (left) compares run-times for the Chicago regional model used in many benchmarking exercises. This model contains ~1800 zones. The horizontal axis is run-time, the vertical axis is relative gap. Compared to equilibrium assignment in VISUM 10, the same path-based algorithm performs already ~3 times faster in VISUM 11, but LUCE is another improvement by a factor 3-5. To reach a gap of close to 10-5, VISUM 10 took over 1.5 hours. In VISUM 11 the path-based assignment takes only ~30 minutes to reach the same gap, but LUCE reaches it in ~8 minutes. Convergence continues down to around 10-8 at much the same speed. The same qualitative results were obtained for many other networks.

With LUCE you are able to load a richer set of OD pairs. Note that while link volumes are uniquely determined in equilibrium, path volumes are not. In fact, a vast set of path volumes will fit a given link volume pattern. VISUM’s path-based assignment is known to be parsimonious, i.e. it covers the equilibrium link volumes with a very small number of paths, often just 1 path per OD pair. As long as path volumes are only a by-product on the way to good link volumes this can be neglected. But if the path volumes are analyzed in their own right, proportionality becomes important. The picture below illustrates how on a congested network VISUM’s path-based assignment will still only load a single path for a given OD pair, whereas LUCE will load no less than 48 – all plausible.

Faster post-assignment analysis Many of the post-assignment analysis procedures have been tuned to work on LUCE bushes if available, resulting in vast performance improvements. Computing skim matrices, evaluating flow bundles, blocking back etc. have never been faster! Obviously, there are parts of the software which cannot easily benefit from the power of LUCE bushes. In some uses cases such as path lists, sub-network calculation, COM access to paths and AMN export, traditional paths must be generated from the LUCE bushes. Editing of path flows and geometry is not possible on LUCE bushes. Furthermore, conversion functions between paths/path sets and assignment paths are not available for LUCE assignments.

REFERENCES Gentile G. (2009), Linear User Cost Equilibrium: a new algorithm for traffic assignment, submitted to Transportation Research B.

Perspectives

Planned after Release VISUM release 11.5

Warm start: Like other assignment methods in VISUM, LUCE will be able to accept an existing assignment result as an initial solution. Because the warm start functionality requires a set of bushes, the prior assignment result must be of type LUCE

Assignment with ICA: Though not initially, LUCE will be available as an alternative assignment method to standard equilibrium assignment for the assignment with ICA planned for VISUM 11.5

Proportionallity across origins: It is planned to extend the procedure in a way to achieve fair assignment of volumes to paths across all origins and destination instead of all destinations per origin only.

: Comparison of path proportionality

: Comparison of performance of LUCE with existing equilibrium assignment methods