Sustainable Flexibilization of Carsharing Services: Modeling and Optimization Masterarbeit zur Erlangung des akademischen Grades Master of Science (M.Sc.) am Institut für Wirtschaftsinformatik im Studiengang Wirtschaftswissenschaft der Wirtschaftswissenschaftlichen Fakultät der Leibniz Universität Hannover vorgelegt von Carolina Bormann, B.Sc. geboren am 22.09.1989 in Wolfenbüttel Prüfer: Prof. Dr. Michael Hans Breitner Hannover, den 30.09.2014
Contents List of Figures List of Tables List of Abbreviations IV V VI 1 Introduction and Motivation 1 2 Background of Carsharing Services 4 2.1 Related Work................................... 4 2.2 Sustainability of Carsharing........................... 7 2.3 Existing Carsharing Approaches......................... 9 2.3.1 Station-Based Carsharing........................ 9 2.3.2 Free-Floating Carsharing......................... 11 2.3.3 Peer-to-Peer Carsharing......................... 13 3 Research Design 14 4 Considerations and Factors for Combined Carsharing Systems 16 4.1 One-Way Trips and Further Flexibilization................... 16 4.2 Critical Success Factors.............................. 17 4.3 Design of a Combined and Flexible Carsharing System............ 23 5 Strategic Optimization Model for Flexible Carsharing Systems 27 5.1 Basic Model for Station-Based Carsharing................... 27 5.2 Assumptions and Model Formulation of the Adapted Model......... 30 5.3 Benchmarks and Evaluation of Influencing Parameters............ 36 5.3.1 Hanover as an Exemplary Instance................... 36 5.3.2 Influence of the Maximum Distance Customers are Willing to Walk. 38 5.3.3 Influence of a Varying Demand and the Customer Parameter..... 40 5.3.4 Influence of a Free-Floating Rate.................... 42 5.3.5 Influence of Varying Cost Parameters.................. 44 5.3.6 Influence of the Given Parking Facilities................ 47 5.4 Generalization of the Results........................... 49 5.5 Limitations of the Strategic Optimization Model................ 54 6 Operative Optimization Model for Flexible Carsharing Systems 56 6.1 Development of a Pricing System........................ 56 6.2 Assumptions and Model Formulation...................... 61 6.3 Possible Model Refinements........................... 67 II
6.3.1 Flexibilization of Parking Lots...................... 67 6.3.2 Application of Macro- and Micro Periods................ 68 6.3.3 Simplified Price Function......................... 69 6.4 Limitations of the Operative Optimization Model............... 73 7 Discussion and Recommendations 75 8 Conclusion and Outlook 78 References 81 I Appendix VII A Interview Transcript B GAMS Code of the Strategic Model C GAMS Code of the Operative Model VII XI XVII III
1 Introduction and Motivation New urban mobility concepts are in an upheaval around the world. Thanks to social and technological progress, new mobility markets are developing faster than ever before. Next to the vision of autonomous driving and the further processing of electric vehicles, carsharing is one of the current major trends. Carsharing provides individuals with mobility, without the fixed costs and responsibilities that come with a private car. The basic principal is the sharing of one or more cars between a group of people in order to save money and resources. At this time, carsharing exists predominantly in cities. Carsharing motivates those who own a car to relinquish it and share vehicles with others instead. The individuals who actively use carsharing and have abandoned their own car, profit from the cessation of fixed costs such as insurance, tax, garage costs, as well as the rent for a parking spot. Carsharing not only brings many advantages for individuals but also for the community. Due to a decreasing amount of cars that are necessary to satisfy the total need for mobility the CO 2 emissions are reduced. This again contributes to a better quality of life. City Carshare, the largest non-profit carsharing organization in North America, released in their environmental report an annual decrease of CO 2 emissions. In the San Francisco area 85 million pounds of CO 2 and 4.3 million gallons of gasoline are saved only due to carsharing (City Carshare, 2013). Furthermore, the community profits from a more relaxed traffic situation since carsharing encourages the use of public transportation. Especially young people have adapted the trend of using instead of owning. It has become more and more popular to pay for a service and to place less emphasis on the actual possession of objects. This behavior is reflected in almost all things of daily use: DVDs, software, electrical appliances, tools, clothing, rooms, bicycles and not least cars. Since the maintenance costs of owned cars are particularly high, the sharing of vehicles promises lots of potential. With a steadily increasing proportion of the population living in cities, more and more people are faced with the challenges caused by urbanization. Circumstances such as a lack of space, pollution, rising energy costs and the finiteness of natural resources are reasons to rethink personal vehicle ownership (Rickenberg et al., 2013, p. 1). Seen proportionally to the population, Switzerland is the world leader in the use of carsharing, followed by Germany. In Germany 1.13% of the population over 18 years participated in carsharing concepts in the beginning of 2014 (Loose, 2014, p. 2). In figure 1 the growth of carsharing vehicles and number of participants in Germany, two of the key figures in carsharing, can be seen. Since 2012 the numbers are distinguished by station-based and free-floating operators and it is illustrated that the growth of free-floating is exponential whereas the station-based numbers grow rather continuously. In station-based carsharing customers pick up the car at a fixed station and usually return it to the same station after 1
Figure 1 Carsharing growth in Germany 1997 to 2014 (Source: Loose, 2014, p. 2) use. No major carsharing operator in North America or Germany offers the additional option for one-way trips, however they have successfully been tested in some pilots projects. This one-way option has been successful in Germany for the route Brunswick-Hanover and in Bremen. In contrast, increasing since 2012, some providers offer the so-called free-floating carsharing, which means that vehicles can be picked up at random public parking places in the city and after the use can be returned to any parking spot within the free-floating area. In the beginning of 2014 the number of customers of free-floating services has surpassed the participants of classical providers. This surpass indicates that the free-floating carsharing is more successful, or at least more popular. In contrast to this assumption stands the results of a recent study by the German consultant agency Civity. The study is based on 115 million data points and 18 million occurrences of worldwide car hire. One of their key finding is that the cars of free-floating providers are nearly as inefficient and land-intensive as a private car. Furthermore, they do not have a significant impact on traffic problems (Weigele et al., 2014). These contradictions lead to the conclusion that current carsharing approaches have to be rethought. Customers seem to prefer the inefficient free-floating approach, still the station-based carsharing is more sustainable and customers tend to give up their own car more often (AIM, European Business School, 2013). On the one hand this implies the necessity to find ways to make carsharing more attractive to customers and on the other, sustainable impacts have to be enhanced. Following research questions can be drawn from this: 2
1. How can current carsharing approaches be extended and flexiblelized? 2. How can the optimal locations and the size of carsharing stations be determined in consideration of free-floating options and one-way trips? 3. How can the relocation problem be encountered on a sustainable level? Is a pricing-system sufficient to balance the vehicles in between stations and a free-floating area? How should it be designed? The structure of this thesis is based on the publication scheme of Gregor and Hevner (2013). After the introduction, the background and theoretical foundations of carsharing are mentioned in chapter 2, where a literature review is given. This review is followed by a brief overview of the connection of carsharing and sustainability. Section 2.3 explains the three different approaches of station-based, free-floating and peer-to-peer carsharing in detail. In chapter 3 the research design is presented. The factors that influence a combined carsharing system are addressed and a solution for such a system is proposed in chapter 4. Thus the qualitative component is covered. The following two chapters introduce two mathematical optimizing models. The first, which is discussed in chapter 5, is optimizing on a strategical/- tactical level. In chapter 6 follows a tactical/operative model, which uses the results of the strategic one as a basis. The strategic optimization model for flexible carsharing systems in chapter 5 is based on an existing model of Olivotti et al. (2014). This basic model is introduced in section 5.1 and reasons for a necessary enlargement are given. In section 5.2 this model is modified and enhanced by the option of one-way trips and free-floating trips inside a predefined area. This is followed by benchmarks that are based on the city of Hanover in section 5.3. Those benchmarks are subsequently generalized and evaluated. To conclude, limitations are stated in 5.5. In chapter 6 a new model for operative optimization is developed. With the objective to act sustainable, a user-based approach is chosen which enables the carsharing operator to renounce extra employees who reposition vehicles when needed. To facilitate this a pricing system is developed in section 6.1. The optimization model is then formulated in 6.2. In section 6.3 possible model refinements are given where next to a division in macro- and micro periods another, more limited price function is established. The limitations of the operative model are stated in section 6.4. The thesis concludes with an overall discussion of both models in chapter 7 and in chapter 8 concluding remarks and recommendations for the future are given. Keywords: Flexible Carsharing, Combined Carsharing Approach, One-Way Carsharing, Operations Research 3
8 Conclusion and Outlook This thesis studied for the first time the interesting and practical problem of finding the optimal locations, number of vehicles and parking spots, and the balancing of a flexible carsharing system that combines station-based offers with one-way trips and free-floating. A two-phase modeling was conducted. First a strategical optimization model was established and optimized. Then the results were used to create an operative model. This concluding chapter determines whether the research questions that were raised in chapter 1 could be answered. Additional, an outlook for further research is given. The first research question was concerned with an enhancement of current carsharing approaches. Due to the increased popularity of free-floating services, contrasted with their low impact on a more environmental city, the need to create a new approach arose. To analyze the current situation the advantages and disadvantages of both commercial carsharing approaches, which are currently practiced, were compared. Due to its non-commercial nature, peer-to-peer carsharing was not taken into further consideration, however, a growth of this approach is predicted (Shaheen and Cohen, 2012, p. 26f). Further research should try to take this approach into consideration simultaneously. A solution to how private carsharing can be simplified should be striven for as well. This way the approaches can be connected and new opportunities can be created. Often private car owners also own a parking space, and these could be used to counter the parking lot challenge. After the approaches were enclosed, a new combined design was established. This way both advantages were combined: the popularity of being more flexible with the free-floating, and the ecological impact of the station-based services. The process of using carsharing did not change very much for the customer; the registration and way to book are still the same. The new feature is that a new variety of trips is possible. Customers can choose whether they want to start at a station or in the free-floating area, and if they want to arrive at one or the other. This allows a maximum level of flexibilization and freedom of choice. Current operators do not offer these options. Currently it is only possible to use both approaches, station-based and free-floating, apart from each other via the same platform. The second research question was answered in chapter 5. The chapter resolved how the optimal locations, and the size of carsharing stations can be determined while considering the established design of the flexible carsharing system. This design took both approaches, free-floating options and station-based carsharing with one-way trips into consideration. To answer the question an algebraical model that suits the circumstances was established. This model is based on an already existing model for station-based carsharing. It was revised with adjusted and additional indices, parameters, and decision variables. This resulted in a new model formulation that is adapted to the new challenges. It is assumed that no modeling error was made since benchmarks were conducted and their results display the right values. 78
The benchmarks established different scenarios for when some parameters are changing and it was documented how the decision variables reacted. The changes to the total price can also be seen. The results were then generalized and a constant behavior of the regarded variables could be observed. This can be seen as an example of the model s quality. Thus, the second research question could be answered. The third research question was concerned with the relocation problem and how it can be encountered on a sustainable level. It asked in which way a pricing-system should be designed in order to balance the vehicles between stations and a free-floating area. To encounter the sustainable part of the question, a user-based approach was chosen. This circumvents the operator having to send employees to reposition vehicles, which would be a waste of resources. Both fuel and money can be saved by the chosen approach. In chapter 6.1 a price system is established. It is designed to prevent imbalances that are about to occur. Two different price systems are proposed, their limitations and challenges were described. The first price function is not applicable in the modeling system GAMS. Also no fixed prices were set and customers have to pay a different price for each trip. A second, simplified pricing system, which uses only three different price categories, was created afterwards. This system was tested on a small instance, nevertheless more tests and benchmarks have to be conducted. Thus, the question how a price system should be designed, was answered. However, the thesis did not find out, if a user-based approach is sufficient to balance the vehicles. To answer this question, the proposed model should be implemented in a simulation software and benchmarks have to be performed. The simulation software Arena is suggested for this task. For future research it is recommended to work with this second price function. Further research should be done about the same topic. It is important to find out how the customers feel about a flexible model. Surveys should reveal the peoples readiness to react to different pricing models and how much they are willing to pay for the service. Another interesting aspect would be to implement trip joining and trip splitting in the operative model. It has to be investigated if the offer will be well received by the customers, and if they are willing to share the same vehicle. The openness to make small detours to drop off other customer, has to be researched as well. It can be investigated how many additional costs would be caused by allowing customers to book with the certainty that the vehicle will be available on time. Therefore emergency workers who drive vehicles to stations that are empty can be employed. It has to be examined whether it is better to employ workers, or if it is possible to pay a defined amount per relocation to the customers themselves. In this case, all registered customers would be able take over the task of relocating. Customers may want to take the trip anyway and could then be rewarded with compensation for the urgent relocation or some people might be able and willing to relocate in their free time to earn money. 79
Carsharing will be an urban mobility concept for the future. The strong growth and increasing popularity will play an important role in ensuring that the growth continues. However, it is also necessary to adapt existing concepts to the customers changing needs and demands. In this way, carsharing will stay modern and suit the actual requirements of society while contributing to a decrease in environmental pollution. 80