Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
25% decrease in in- vehicle travel times in the Dal- lasâFort Worth area. This analysis was conducted to assess the reasonableness of the predications. The activity- travel patterns were predicted for the entire syn- thetic population of 3,452,751 from 1,754,674 house- holds for the base case and each of the four changes in vehicle travel times. The impact of the changes in in- vehicle travel time on aggregate activity- travel patterns was examined for trip frequency, person miles of travel, vehicle miles of travel (VMT), and person hours of travel (PHT). ⢠The 10% increase in in- vehicle travel times reduced the total number of trips by 1%, whereas a 25% increase in in- vehicle travel times decreased the total number of trips by 2.4%. A 10% decrease in the in- vehicle travel time increased total trips by 1.1% and a 25% decrease resulted in an increase in total trips of 3.1%. An increase in in- vehicle travel times decreases VMT and a decrease in in- vehicle travel times results in an increase in VMT. An increase in in- vehicle travel times increases the PHT for work and decreases the PHT for nonwork purposes, resulting in an overall increase in PHT. A decrease in in- vehicle travel times reduces the PHT for work and increases the PHT for nonwork purposes, resulting in an overall decrease in PHT. MATSIM/PLANOMAT: A MICROSIMULATION SYSTEM OF ACTIVITY DEMAND Kay Axhausen Kay Axhausen described an open- door Java- based toolkit, which provides the user with various instruments to implement activity- based models and scheduling- based models. The model is called Multi- Agent Trans- portation SIMulation Toolbox (MATSIM- T). The following points were covered in his presentation. ⢠First, it is beneficial to examine how current behav- ior is being modeled at the microscopic level. Elements include generalized costs of the route- mode- location alternative. Budgets and long- term commitments are included. Tastes include values, attitudes, and life style by sociodemographics. One of the big attractions of using microsimulation is that there is a national frame- work to account for differences in tastes between per- sons. There is also an increased awareness that the choices that individuals make are driven not only at the individual level and at the household level, but also within the larger social network, which will decide and influence location choices and activity choices. ⢠The generalized cost of a route- mode- destination alternative includes time and reliability, adjusted for both comfort and risk. Reliability is such a large element of the travel experience that individuals who are risk adverse will have a different behavior than those who are risk prone. Monetary expenditures are also included. Numerous activities have a social content, which may focus on doing things with or for others. Agent- based microsimulation might offer the opportunity to address these issues. ⢠Microsimulation models should include a learning approach. On the one hand, they model schedulingâ what an agent does, by which mode and route, and with whom. On the other hand, they model competition for slots on networks and facilities. Initially, iterations between scheduling, the mental map, and the competi- tion will help revise the cost estimates. The parameter estimation is typically not included because of complex- ity, but it should really be included. ⢠A first step in the use of microsimulation models is creating a description of the world. The availability of accurate data is critical in this step. MATSIM- T provides various tools to deal with these and other issues. MATSIM- T implements numerous elements to create the world and to manage the different resolutions. It pro- vides an agent database, which is in memory. It provides various tools to implement the competition for a slot on the network. Various dynamic traffic assignment tools can be selected. There are also various tools to schedule activities. ⢠The focus is on modeling household interaction. This household interaction includes choosing an optional allocation of time over a day and decisions on joint activities, journey destination, and journey mode. A tool searches for the optimum schedule, which takes numerous iterations. These iterations currently take a lot of time to run. ⢠Zurich is being used as a test bed because it has a detailed navigation network, available timetables for public transport, and information on facilities available for each mode. The agent population has been generated using seven dimensions. Estimates of travel demand are available from a national travel survey and from observed counts. The initial analysis indicated that the smarter the agent and the more variability of adjustment by the agent, the faster convergence or a steady state is reached. If the agents in the optimization are allowed a wider search base, they find solutions quicker. This analysis indicates that fewer interactions may be needed to reach a steady- state system. ⢠The software will be available at www.source forge.org for others to use. Efforts are under way to regain the capabilities of the full scheduler. Parameter estimation also needs to be performed. Visualization and analysis tools are also being considered, along with methods to integrate social networks. 30 INNOVATIONS IN TRAVEL DEMAND MODELING, VOLUME 1
