research

Using RP data from a bespoke travel-activity survey conducted in the Austrian Alps, I developed discrete choice models of transport mode choice of tourists during their vacation stays at the resorts.

When tourists decide about which transport mode to choose for trips at the destination, we observe significant effects of travel time, travel cost, travel party composition, trip purpose, respondent’s level of fitness, knowledge about long-distance travel to the destination and mobility options at the destination, and selected weather elements. Tourists are found to be very inelastic to changes in travel cost, whilst more responsive to changes in travel time. This research delivers unique evidence that can advance transport policy design and thus contribute to more sustainable travel at tourist destinations.

With the popularization of private vehicles and low-budget airlines, leisure travel has become the most significant category of trips made and miles traveled in western societies. However, its impact the natural environment can be substantial, in particular in national parks, conservation areas, and highly attractive tourist hotspots. Authorities and destination managers are struggling to encourage visitors to arrive by train rather than private cars, but mostly to no avail.

My research indicates that the long-distance travel to the destination and the local travel at the destination should not be treated separately since they affect each other substantially. For instance, people choose to travel by car on vacation out of fear of insufficient transportation options at the destination. Through discrete choice models, I found large effect of travel time, which eventually results in high preference for choosing car compared to other modes. Winter travelers are less likely to choose rail than summer visitors, which is caused by bulky and heavy sports equipment (skis, ski boots), difficult to carry on a train. Furthermore, the effect of company size is substantial and should not be underestimated. As long as an adult couple can easily decide to go on a train to the Alps for a hiking hut-to-hut trip, a family with two kids under six years would be more inclined to choose car. This has to do with the uncertainty of conditions during the trip to the destination (crowdedness in trains, too little space, comfort, etc.) and uncertainty regarding the activities and mobility on-site. Parents traveling to tourist resorts in the Alps are usually good-earners and thus have a high Value of Leisure. That is, they likely assign high importance to the flexibility of having car at hand all the time and being able to adapt easily should the activity schedule change. Traveling by train and using transit at the destination appears to be too much hassle for them, even though they might overpay for using car.

I have been working on developing solutions to the above mentioned problems, which include: (1) seamless luggage trasport services from home to destination, (2) innovative mobility services at the destination (e.g., carsharing), and (3) a one-stop-shop for booking all mobility services in a single smartphone app.

First, it turns out that international luggage service is extremely difficult to implement due to burreucracy of rail operators, competing interests, and incompatible IT interfaces. Second, launching a carsharing service in non-urban tourist areas is of no interest for commercial providers due to economic factors - low return rate, seasonal fluctuations in demand. Third, econometric models based on Stated Preference experiments among tourists prove that the effects of such measures are dramatically low when compared to the effect of travel time and travel cost. It is unlikely that visitors will change their behavior by virtue of these new services alone. Without the 'stick' component (penalizing, pricing, ...), the change to sustainable modes will not happen (in time).

Design of a survey is a responsible task which requires time and personal resources as well as experience and knowledge in the field. Scope, type and form of the survey has implications for all following stages of the research. Wrong or inappropriate method can lead to low quality results or a complete failure in answering the research questions. Therefore, I developed a rating system, which, based on a set of performance indicators and weighting parameters, returns weighted scores for existing traditional (survey-based) and tracking (GPS, phone traces, social media, ...) data collection methods.

In a longitudinal study, conducted before, during and after the 2018 UCI Cycling World Championship, I analyzed whether, besides promoting the host region as a cycling destination, a large cycling event can also leverage a positive change in residents' travel behavior towards active mobility.

It turns out that while residents expect the event to change their travel behavior and are actually willing to cycle more, this willingness does not translate into real behavior after the event. Also, the investments in cycling infrastructure, made because of the UCI, fall short of their expectations. However, spectating the event live alleviates the disappointment.

While the level of physical activity at home does influence daily mobility patterns, that is, those physically active are significantly more likely to cycle and walk for transportation, this does not hold on vacation. Also, mobility patterns at home are unrelated to the ones adopted during a vacation – active cyclists/walkers at home do not cycle/walk significantly more on vacation, nor are the daily drivers driving more at their vacation resorts.

Apparently, there are other, more influential factors shaping mobility decisions during a vacation, such as flexibility, luggage transportation and on-site mobility. People do not want to give up comfort and flexibility when traveling on vacation, even though their overall awareness level about sustainability, environment and climate is high.

Standard techniques in congestion analyses exploit the cross-sectional traffic data aggregated over all drive lanes. These are however not reliable for the motorway sections, where the traffic volume, vehicle composition, and velocities differ significantly between the offside and nearside drive lanes. Since these conditions are becoming more frequent with the increasing congestion on motorways in alpine regions, their effect on traffic flow is not negligible.

The proposed approach consists in employing the adaptive smoothing method (ASM) to the disaggregated data for each of the traffic lanes separately, which enables the lateral interactions between vehicle flows and the influence of congestion incidents on adjacent lanes to be taken into account. The outcomes show a dissimilar overall number of jams detected for the separate drive lanes than in the analysis of the whole cross-section. The difference is also in types of congested states. Some traffic patterns (mega jams and wide jams) appear more often on the nearside than on the offside lane.

I worked on the usability of probe vehicles data (aka: Floating Car Data) of low quality for congestion analysis on motorways. Based on a dataset characterized by large GPS time interval, low number of trajectories, and overrepresentation of trucks, I analyzed the influence of these quality parameters on congestion detection and identification of its type. After a validation with stationary detector data (SDD), I fond that low-quality probe vehicles data does not allow to identify short-term congestion patterns (short speed drop, stop-and-go wave) reliably. Especially the low number of trajectories proves problematic, whereas the influence of other parameters was less distinct. Despite these flaws, for large-scale congestion incidents (wide jam, mega jam), floating car data provide outcomes similar to those derived from stationary detectors.

Two-lane motorways with heavy truck traffic are very prone to large-scale congestion caused by minor incidents. Construction work that requires the closure of one lane makes the situation even more sensitive, and large-scale congestion on the motorway puts at risk the proper functioning of adjacent residential areas where the alternative road network is sparsely developed (as it is in mountainous regions).

One solution is to limit the inflow of trucks on the highway on selected days and hours. The study explores how the unregulated inflow of trucks is accelerating the collapse and to what extent a regulated inflow can shift the capacity limit. Initial results prove the effectiveness of limiting the truck inflow for alleviating large congestion incidents. However, allowing vehicles in at even and large time intervals does not prevent platoons from forming, as drivers tend to drive one after the other and after a few kilometers small time gaps are observed again.