Please use this identifier to cite or link to this item:
Title: Investigating Car Driver’s On-Street Parking Decisions
Authors: KHALIQ, Annum 
Advisors: JANSSENS, Davy
Issue Date: 2018
Abstract: Car parking is an interesting domain in transportation. In the past, a limited number of studies investigated parking issues but the attention paid to car parking research is increasing with time. This is because car ownership rates are increasing. Large number of cars need to be parked somewhere at the end of the journey, which is difficult to manage. Parking management has become an interesting domain to explore. Investigating short term (shopping, leisure etc.) car trips to city center is necessary because the traffic searching for cheap on-street parking spaces creates nuisance in city centers. It increases travel time, fuel consumption and drivers’ frustration. The increasing number of cars cruising on downtown streets are creating alarming situation with respect to traffic congestion. Studies suggest that the share of traffic cruising for parking is 30% and the average cruising time is about 8 minutes (Shoup, 2005). Although there is a debate regarding myths of parking (Syden & Scavo, 2006) but the importance of on-street parking cannot be ignored. The literature suggests that local governments should manage on-street parking by controlling price and allocation of parking spaces during different times of the day. In order to find out which other factors can be controlled to manage on-street parking efficiently a research related to car drivers’ on-street parking choices needs to be executed. Motivated by the limitations of previous parking choice models, one of the major contributions of this research is to model and simulate car drivers parking decisions keeping in view the road conditions. This is necessary to investigate the effect of changes in parking conditions on car drivers’ parking choices i.e. where do people park (either on-street, off-street or continue search) if certain conditions of road are changed. The study of car drivers’ parking choice behavior investigates effects of changes in parking policy on traffic flow. Another major contribution of this research is to evaluate which parking and road conditions urge car drivers to search for parking. Road condition depict the circumstances prevailing on the road such as space availability, allowed speed limit or surrounding land uses. During the initial phase of my PhD research project, it was observed that no research has been carried out to identify combination of parking and road conditions that induce search traffic. The design of a conceptual framework to simulate the parking choice process of car drivers’ is also presented in this research. This framework depicts when car drivers enter a street and start looking for parking, they have to make a choice. If they find a suitable parking place they park on-street; if they do not find an optimal parking space, they drive to the nearest parking garage; and if they do not find a suitable place in the current street and they do not want to park in a garage, they keep on searching and drive to a next street looking for parking. This framework is setup to analyze the realistic behavior of car drivers’ parking search. To collect data for evaluating car drivers’ parking decisions with respect to road conditions a comprehensive online questionnaire was prepared. A stated choice experiment was designed to present the road conditions (prevailing parking situation) to a car driver. It was very difficult to design such a choice task that can detail a particular parking situation on a road. For designing such a comprehensive choice task a large set of attributes was required. A standard stated choice experiment cannot handle a large number of attributes. Therefore, the hierarchical information integration (HII) approach was applied (other approaches were avoided to keep the level of complexity as minimal as possible). A simplistic yet controlled method for dealing with large number of attributes and hypothetical situations was required. There is evidence in the literature that hierarchical information integration (HII) has provided useful results when combined with stated choice experiments (Louviere, 1984) (Bos et al, 2003) (Richter and Keuchel, 2012). For this research, the integrated hierarchical information integration (HII-I) approach has been used. According to this approach, attributes and attribute levels are grouped together in different categories known as ‘decisions constructs’. These decision constructs depict descriptive summary labels for coherent sets of attributes that define alternatives in the study. Different experimental designs were developed for each construct such that each design included a detailed description of one decision construct while the other decision constructs are added to the design as a ‘global attribute’. This process makes each choice task less detailed, consisting of eleven attributes in total; seven general parking related attributes, three attributes of detailed construct, and one global attribute (high-order description of decision construct treated as additional attribute, replacing three detailed attributes). The designed choice tasks were presented in an online questionnaire. The respondents were asked to assume that they are driving from their home to a (generic) destination located in the center of a city. While driving they enter a certain road segment (with specific parking and road conditions) to search for a suitable parking space. The full questionnaire consisted of three parts and was constructed using an online system developed by the Eindhoven University of Technology. Two screening questions were conducted at the beginning of the questionnaire, to select respondents who own one or more cars and have some experience with on-street parking in urban areas. The first section of the questionnaire was related to car drivers’ experiences with on-street parking. This section included questions regarding on-street parking issues faced by the respondents, search time, parking purpose, parking frequency, walking time between parking place and final destination, parking information sources, preference for on-street and off-street parking, and which type of road (commercial, residential) they find most difficult to park at. The second section consisted of the hypothetical choice conditions. At the beginning of each section a detailed explanation was provided to prepare the respondents regarding the forthcoming questions. The context of parking decision was mentioned in the explanation of section 2. After designing the choice task and the questionnaire, the next task was to identify a well-balanced sample and disseminate the questionnaire to the respondents. For this purpose, the online company ‘PanelClix’ was hired to get required number of responses. Belgian members of the panel were invited to fill out the online questionnaire. In total, 548 respondents completed the questionnaire. The sample characteristics show that the group of respondents is well representative of the referenced population (car drivers’ who visit inner city areas by car and have parking experiences). After data collection the next task was to perform data cleaning. The collected data was checked thoroughly for inconsistencies in questionnaire completion behavior (i.e. by length of time taken to fill out the survey). After carefully performing data cleaning, the data was coded (dummy coded) and analyzed using NLOGIT version 5 (Econometric Software Inc., 2012). The data was first analyzed using standard multinomial logit model and later on, the data was analyzed using mixed multinomial logit model. The data fitted well with standard multinomial logit model (MNL) and mixed multinomial logit model (MMNL). With standard multinomial logit model (MNL) only a few attributes are significant at the conventional level (95 percent) such as ‘parking tariff’, ‘expected parking duration’, ‘speed limit’, ‘surrounding activities’, ‘off-street parking tariff‘ and ‘maximum parking duration’. The significant parameters are in anticipated direction. This result might indicate that car drivers do not consider ‘security’ or ‘payment options’ while looking for a parking space on-street. Mixed multinomial logit model also appeared as a good fit for the data because more number of the means and standard deviations of attributes such as ‘off-street parking tariff’, ‘on-street parking cost’, ‘level of parking convenience’, ‘surrounding activities’, ‘maximum parking duration’, ‘speed limit’ , ‘distance between parking location and destination’ and ‘number of streets visited’ are significant. This means that there is random taste variation across the respondents regarding these attribute levels (since the MNL model assumes that all the individuals are same while MMNL considers that every individual is different therefore the results of model estimation are different). The model estimation results show that if speed limit is 20km/h, the expected parking duration is less than 60 minutes, and the off-street parking tariff is high (2.50 euro/hour) then there is a higher probability that a car driver parks the car on-street. Moreover, free parking, providing parking closer to destination, roads with reduced speed limits (preferably in residential areas), and without any security feature are the major conditions that induce parking search. Similar conditions that induce search traffic can be identified using the model results. After analyzing the data, the estimates of mixed multinomial logit are used to predict the probability of car drivers’ parking choices. The model predictions are based on the principles of the Monte-Carlo simulation, these simulations are generated using excel worksheet. The results of the simulation provide a range of possible outcomes to the decision-maker and the probabilities related to parking decisions of car drivers. This simulation setup is able to illustrate the working of the model when investigating the effects of parking measures. With the simulation, the effects of (a) surrounding activities (b) speed limit (c) parking duration (d) the introduction of paid parking or an increase of the on-street parking cost from 1.00 euro to 2.00 euro are evaluated. The simulation shows clearly the changes in probabilities of car drivers parking decisions i.e. park on-street, park off-street, and continue to search due to the change in parking cost. Similarly, other inferences can be drawn for other significant attributes such as ‘distance between parking location and destination’ and ‘number of streets visited’. The model predicts that the activities located on a road also have a strong impact on the car drivers’ parking decisions. The model identifies that cruising would be high in case of a roads having recreational and commercial activities such as sports stadium, playground, shopping malls, etc. Residential streets usually have several parking spaces available, which is realistic. This model is also capable to predict the number of cars that will be cruising for parking with respect to certain road conditions. If we change a road condition the change in car drivers parking decisions can be predicted and thus the most preferable road conditions for on-street parking can be identified. Certain other policy related inferences (policy implications) can be drawn from this research. Keeping in view, the existing literature attributes such as walking distance, price, search time, vehicle type, access time, etc. only provide limited knowledge related to car drivers parking decisions, in order to raise the policy implications a detailed understanding of road related attributes is required, which are highlighted in the current research. The adopted approach has highlighted relative importance of the parking and road related attributes resulting in demonstration of how through adjustment of these road related parking factors (such as speed reduction and changing parking tariff), the number of cars cruising for parking can be reduced. In a nutshell, policy makers can use the results of this study to find out which parking and road conditions contribute to search traffic. Therefore, certain road situations can be avoided that induce search traffic by devising suitable policies. Also this type of models can be useful for decision makers to promote certain parking management strategies that complement built environment and urban planning strategies. Moreover, this research can further be used as a part of multi-agent simulation systems to express in more detail car drivers parking decisions for predicting effect of change in parking measures on traffic conditions of a city.
Document URI:
Category: T1
Type: Theses and Dissertations
Appears in Collections:PhD theses
Research publications

Files in This Item:
File Description SizeFormat 
  Until 2023-10-22
Peer-reviewed author version4.14 MBAdobe PDFView/Open    Request a copy
Show full item record

Page view(s)

checked on Jul 4, 2022


checked on Jul 4, 2022

Google ScholarTM


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.