Abstract
On fully enclosed freeways, the service area is the only key node for certain crowd activities such as resting, toileting, and dining. In these activities, using the toilet is the most important purpose for people to enter the service area. A reasonable toilet location and size in the service area is the basis for its efficient operation. This study conducts field surveys on several service areas, including Meicun, Tai’an, and Qianxian, through video recording and crowd tracking questionnaires. We summarize and analyze the characteristics of population flow in Chinese freeway service areas. The key data obtained through field surveys are reported, such as the priority and length of stay of customers in the facilities during peak hours at weekends, to analyze the characteristics of pedestrian traffic in the service areas. Finally, based on a large number of pedestrian simulation methods used in studies on the layout of subways and airport facilities, the AnyLogic simulation software is applied to establish a crowd movement model for these areas. The study not only obtained the pedestrian characteristics of the service area but also showed that only changing the location of toilets has no effect on the efficiency of toilet use without changing the level of service, whereas changing the number of toilets has a significant impact on the efficiency. Most importantly, this paper establishes a relationship model between the toilet size of the service area and the service area pedestrians flow; when the toilet seats 266/158/148 people, the maximum flow of people in the service area is about 3,115/2200/2000 p/h, which can provide help for the scale design of the service area.
1. Introduction
The service area is a key node connecting freeways and surrounding scenic spots. A reasonable geometric layout of the internal facilities and traffic organization is the basis for ensuring an efficient operation of service areas. Accurately evaluating the internal road layout and pedestrian traffic organization mode in the design stage of service areas is of great significance for predicting the traffic state of such areas during the operation phase, adjusting the design plan in advance, and improving the traffic organization means. Moreover, ensuring a smooth traffic flow within the service area is a prerequisite. The design specification for highway alignment (2017) requires freeways in China to be fully enclosed; the service areas here are quite different from those in Europe and the United States (Green Book, 2001 [1]). On fully enclosed freeways, vehicles and personnel cannot freely enter and exit the freeways. Passengers and vehicles can only enter the service area for resting [2–4], refueling, and dining. This also makes the service area of the fully closed freeway full of pedestrians. However, there are no complete data on pedestrian characteristics in service areas, so the interaction between crowd gatherings and internal facilities is not considered in the design of the internal facilities of service areas, resulting in a large number of unreasonable internal facility design schemes, which is a major issue that should be resolved.
The main contributions of this study include the following.
This study aims to resolve the bottlenecks in the current service area design and provide a reference for future internal design of service areas. In this paper, we first summarize and analyze the characteristics of population flow in Chinese freeway service areas. Subsequently, key data obtained through field surveys are reported. Finally, based on a large number of pedestrian simulation methods used in studies on the layout of subways and airport facilities, we apply the simulation software AnyLogic to establish a pedestrian flow model within the freeway service area.
2. Literature Review
To design the internal facilities of freeway service areas based on crowd gatherings, it is first necessary to identify the several aspects of a service area, such as the role, function, and scale. From the 1950s to the early 21st century, freeway service areas have expanded considerably in terms of their role, function, and definition [5, 6]. As a facility built on both sides of a high-grade freeway, a freeway service area is a hub for drivers and passengers to rest, eat, shop, and refuel/charge without having to leave the road on which they are driving. Related research during this period mainly focused on location selection and operation management of service areas. Gordon [7] studied a roadside service area to solve problems of vehicle driving failure or driver fatigue in the initial stage of freeway construction. Subsequently, Fowler et al. [8] comprehensively studied the theory of service area infrastructure construction, including spacing, site requirements, facility building structure, materials, mechanical systems, and operation and maintenance, and suggested measures for the construction of power sources, water systems, and sewage treatment systems in the service area. Since the turn of the 21st century, based on in-depth discussion on existing theories, various countries have started paying attention to the economic benefits of service areas and have tried applying new technologies to improve their operational efficiency. Many scholars [9–12] are now using various types of simulation software and a variety of monitoring equipment to model and analyze the scale and location of freeway service areas. Currently, research on rest areas in American freeways mainly focused on reducing the impact of rest area construction on environment and the application of constructed wetlands. Kao et al. [9] proposed the application of constructed wetland systems in freeway rest areas. In UK, the focus of research on freeway service areas has shifted toward the following aspects: mitigation of driver fatigue through environmental construction with attempts to reduce road accidents [10], using retail and rental markets to increase freeway service area turnover [11], construction of service area landscape and characteristic architectures [12], and sewage treatment in freeway service areas. However, the above studies did not consider the arrangement of facilities from the perspective of pedestrians and have not applied the mature pedestrian traffic characteristics theory. This makes the design of freeway service areas lack pedestrian traffic characteristics, yielding unreasonable designs. As a result, some areas in a service area are crowded with people, whereas some other areas are left unused.
Second, to study the design of the internal facilities in freeway service areas based on crowd gatherings, it is necessary to identify the characteristics of pedestrian traffic flow within service areas. Since the 1950s, scholars [13] have explored the pedestrian traffic flow theory. Images and video films, as the main research methods, have been applied to analyze the behavior of pedestrian groups, in order to establish service levels of road facilities, design reasonable pedestrian service facilities, and formulate transportation planning guidelines. Hankin and Wright [13] laid the foundation for research on pedestrian traffic flow; however, they did not apply computer technology or supporting software, and their results cannot be applied to the research of transportation hub facilities such as subways and airports. With the advent of computers and supporting pedestrian flow simulation software, the pedestrian traffic flow theory has advanced rapidly [14–17]. Various countries have been studying pedestrian facilities based on the pedestrian traffic flow theory. Fruin’s research results were collated and published in 1971. The book covers most types of pedestrian transportation facilities and corresponding research contents. The freeway Capacity Manual (HCM) prepared by the United States Transportation Research Board (TRB) [18] is highly regarded in the field of traffic safety. The latest version, HCM 2000, has two chapters introducing the concept of pedestrian traffic, analysis methods for pedestrian traffic characteristics, and design process of pedestrian facilities. In addition, TRB prepared the Transit Capacity and Quality of Service Manual (TCQSM) [19], namely, the “Transit Capacity and Quality of Service Manual.” It was designed to provide guidance for engineers in planning and designing hub facilities and systems to obtain the latest research results. The manual covers bus stations, ferry terminals, and rail transportation hubs, and other stops, hubs, and terminals, involving a large number of pedestrian transportation facilities. In 2001, the China freeway society published the “Traffic Engineering Handbook” [20], summarizing domestic research results and learnings from foreign research; Chapter 12 presents a systematic summary of pedestrian and bicycle traffic. Scholars [21] have reported pedestrian traffic flow characteristics within service areas of hub facilities including subways and airports. However, these characteristics are different from those observed in freeway service areas. First, there are no queuing facilities inside service areas, such as gates, security channels, unmanned ticket machines, and manual ticketing windows. Second, most of the service areas are single-story buildings; i.e., there are no floor facilities, such as elevators, stairs, or escalators. However, the ratio of the commercial area within the service area to the total area far exceeds the ratio of the commercial area of the subway stations and terminal buildings to the total area. Based on the analysis of pedestrian characteristics in an airport terminal by Gu et al. [21], the pedestrian pace in the commercial areas of terminal buildings is found to be always higher than that in queuing areas (the area near the check-in counters), which is much lower than those in circulation areas (the main location for pedestrian circulation, which is the section connecting the hall entrance with each functional area) and at hall entrances.
Therefore, the pedestrian traffic characteristics of freeway service areas are different from those of subways and terminal buildings. However, none of the above studies have considered service areas, and there is no relevant study on the characteristics of population flow within a service area. Nevertheless, the above studies provide a theoretical basis for this paper.
3. Materials and Methods
In this study, the priority and probability of using the facilities, the time of using the facilities, and the characteristics of pedestrians in the service area were obtained through video recording, questionnaires, and pedestrian tracking. Subsequently, based on these data, the crowd flow in the high-speed service area was modeled using the AnyLogic simulation software. In the simulation and by changing the location of the facility, the change in the crowd aggregation degree was determined.
3.1. Field Test
After getting the construction form and some relevant data of service areas in various countries [22, 23] service areas with different architectural forms were selected for field testing. The main methods used in field tests include video recording, questionnaire survey, and pedestrian tracking. The main purpose is to obtain pedestrian characteristics at the peak time of the service area (11:00–13:00).
3.1.1. Video Records
Test instrument: Video camera. Test purpose: To obtain the time for pedestrians stay in the facilities inside the service area and service area. Test process: Setting up cameras at the entrance and exit ramp of the expressway service area in advance, then starting work after reaching the time specified in the test, and recording the time of vehicles and pedestrians to enter the service area or the internal time of departure, so as to get the stay time of pedestrians in the service area.
3.1.2. Questionnaire Survey
Test questionnaire: The whole questionnaire is provided in Figures 1 and 2. Test purpose: To obtain the travel characteristics and test process of pedestrians in the service area. Test process: Selecting pedestrians randomly in the service area and distributing relevant questionnaire information.
3.1.3. Pedestrian Tracking
Although we can determine the residence time of pedestrians in the service area and each facility through video, the activity track of each pedestrian within the service area cannot be accurately tracked through video, so as to determine the frequency and importance of each facility in the service area. Therefore, the pedestrian tracking method needs to be adopted to obtain data on the use frequency and importance of various facilities in the service area. Test purpose: To obtain the order and time of various facilities for pedestrians. Test process: Pedestrians of all ages and gender are randomly selected to investigate and sign the investigation agreement to track their activity path in the service area after their consent, so as to obtain the corresponding data. Scope selection: Meicun service area (comprehensive service area), Qianxian service area (independent-type service area), and Tai`an service area (square service area). Ethical considerations: This research was reviewed and approved by Chang`an University.
3.2. Simulation
Based on preliminary field test data, we refer to the pedestrian traffic characteristics of terminal buildings and subways and divide the interior of the service area into a circulation area (the main location for pedestrian circulation and the connection between various facilities) and various facility areas (toilets, shops, and restaurants). Because of the large number of small commercial stores in a freeway service area, the small stores are merged into commercial areas for the analysis, in order to facilitate calculation and simulation.
The software used in the simulation, namely, AnyLogic, is widely used in research and engineering [24]. This software is based on the system dynamics theory and integrates the complex system theory. Moreover, for the first time, it uses a hybrid state machine to depict the evacuation of personnel in the event of an emergency.
4. Results
4.1. Survey Results
The survey was conducted in the following three service areas for each service area as follows:
In the Meicun service area (Figure 3), which represents a comprehensive service area, all the facilities are set up in a comprehensive building. Pedestrians gather at the center (the toilet is at the center). Figure 3 shows the layout of the internal facilities of the comprehensive service building. The brown structures in Figure 4 and all the service area structure diagrams and population flow diagrams in this paper are the building walls, while the green dotted line represents the scale of each facility. Supermarkets, restaurants, and toilets are located around the building center in Figure 3. The interior area of the comprehensive building is approximately 3500 m2, and the scale in the picture is 1 m = 2.5 pixels.
Tai’an service area (Figure 4) can be considered a representative of a square service area. There is no comprehensive service building with integrated functions such as toilets, shops, and restaurants in the east of the Tai’an service area. Instead, small buildings are arranged in sequence, and each building has a function. The boundaries of each facility area of the service area are connected to the outside, which is convenient for pedestrians to quickly enter each facility. However, when the passenger flow is high, the crowd scattered around is difficult to manage. The area of its main facility is approximately 1800 m2.
The Qianxian service area is an independent-type service area. Although the Qianxian service area has a comprehensive building that integrates structural facilities, such as toilets, supermarkets, and restaurants, the entrance to each facility is scattered at different locations, and the entrances are directly constructed outdoors. Figure 5 shows its layout. The main facility area is approximately 2400 m2.
Table 1 lists the scale of the facilities in the three service areas (number of services), obtained from the survey.
4.1.1. Analysis of Pedestrian Speeds in Service Areas
Pedestrian speeds in the service area were counted based on video records. Table 2 lists the pedestrian sample amounts based on the video and field measurement.
After investigation, it is found that there is no significant difference in pedestrian walking speed among all age groups except those over 60 years old with too few samples; the maximum differences are 0.07 m/s (Meicun), 0.07 m/s (Qianxian), and 0.02 m/s (Tai’an); similarly, the difference of pedestrian walking speed between men and women in service areas is also small, with the values of 0.01 m/s, 0m/s, and 0.01 m/s. Figure 6 shows the overall average speed and the average speed of each age group and gender in the three service areas. The reason for this phenomenon may be that there are many open shops in the service area, and pedestrians of all ages and genders are also browsing shops and watching goods due to curiosity and their own needs when walking, thus reducing the walking speed, which leads to the fact that the impact of gender and age on walking speed cannot be clearly reflected.
(a)
(b)
4.1.2. Analysis of Pedestrian Aggregation in the Service Area
After statistics of vehicles entering the service area according to the video, the data are shown in Table 3. According to the video, small and medium-sized buses and trucks entering the service area account for more than 80 percent of the total traffic. Due to the limited carrying capacity of small vehicles, the vast majority of them are 1–3 passengers. According to the questionnaire and follow-up survey, the vast majority of the respondents had no more than three peers. Furthermore, field surveys found that even large bus passengers were mostly clustered in 1–3 units and few in larger groups.
4.1.3. Analysis of the Pedestrian Residence Time in the Service Area
Because of technical constraints, we cannot obtain enough data on the time spent by pedestrians in service areas at present. Therefore, in this study, in the field investigation of the three locations, the vehicle residence time was obtained through the license plate recognition technology. Table 4 lists the length of stay of vehicles in the service area. The data show that a total of 63.846% of the vehicle residence time was within the interval of (5,35) min. It is inferred that passengers stay for a short time inside the service area.
From the survey (Table 5), we found that, due to the relatively monotonous facilities in the service areas, the passengers and drivers tend to use only 1 to 3 facilities. Among these facilities, the usage times of toilets and supermarkets are generally 2–3 and 3–5 min, respectively, whereas the residence time in restaurants is more scattered, generally more than 20 min.
4.1.4. Analysis of the Facilities Utilization Rate in the Service Area
Field surveys are conducted in the service areas of Meicun, Tai’an, and Qianxian by means of video recording and flow tracing. We recorded the population flow during peak hours [25] on weekends (i.e., 11:00–13:00). Furthermore, multiple observation records are taken to obtain the average value of the population flow. The results show that the population flows in the Meicun service area, Tai'an service area, and Qianxian service area at peak hours during the weekend are 2615, 1200, and 932 person times per hour (p/h), respectively. The usage probability of each facility is obtained from random population flow tracking records and face recognition technology. Table 6 lists the specific values.
4.2. Simulation Validation
4.2.1. Parameter Settings
Based on the above pedestrian traffic characteristics analysis and investigation data, the pedestrian simulation parameters in the service area can be set. Since the different gender, age, personality, etc. can cause significant differences in the pedestrian pace and the use time of each facility, setting a fixed value does not accurately reflect the real situation. In this paper, only the intervals with the highest frequency of pedestrian data were obtained through the survey and are not fully determined for the specific distribution, so the pedestrian characteristics data in this simulation is set to be uniformly distributed within the intervals with the highest occurrence frequency. Table 7 lists the parameter settings of each facility turnover rate.
4.2.2. Simulation Calibration and Validation
Because the field survey data may not reflect the real situation completely and in detail, the parameter settings may be inaccurate. Therefore, in the calibration and validation part of this study, the pedestrian flow of each internal facility in the field experiment is compared with the corresponding pedestrian flow of each facility obtained by simulation. The hypothesis test method is used to test it. If the error between the simulation data and the real data exceeds the allowable value, the probability of using each facility is adjusted to keep the error of the simulation results within the allowable range.
Table 8 lists the hourly population flow in the three service areas during weekend peak hours and in each facility through video analysis and face recognition technology.
The corresponding model of each service area is run to obtain the population flow of each service area to each facility, as listed in Table 9.
Suppose that whether an individual enters a particular facility is subject to a binomial distribution with probability p, B∼(1,p); then the probability distribution of the population n is B∼(n, p). Its mean value and variance are
According to the central limit theorem, when the sample n is big enough, approximately follows a normal distribution. Therefore, we have
When n is sufficiently large, approximation equals Sn, so the confidence interval of with a confidence of 1- is approximately
Based on the interval estimation of the simulation data, the 95% confidence interval of the number of people entering each facility is obtained, as listed in Table 10.
By comparing the corresponding data in Tables 9 and 10, it is found that the simulation results are all within the confidence interval of 95%, which proves that there is no significant difference between the simulation results and the actual situation at the significance level α = 0.05, so the model accuracy meets the requirements of the study.
4.3. Simulation Results
At the flow rate set above, we assume that each facility in the service area can meet the relevant demands of all people, i.e., no queuing. The facilities in the three service areas are simulated; the simulation parameters have been given in the previous section.
4.3.1. Impact of Toilet Location in the Comprehensive Service Area on Crowd Gathering
(1) Impact of location change of a single toilet. Modeling simulation is conducted based on the data and assumptions made, as listed in the tables provided in this paper. Because of the large number of shops in the service area, in order to simplify the model, various small shops are merged, and simulation modeling is performed taking the form of commercial and catering areas. Figure 7(a) shows the simplified model diagram. Figure 7(b) shows the resulting structure diagram where the position of the toilet is at the center on the same side of the door.
(a)
(b)
As the toilet in the model is a service area, pedestrians can enter from all directions, which is contrary to the actual situation; therefore, as shown in Figure 7(b), a wall encloses the toilet. Because of the distance between the user and the toilet, the location of the entrance and exit of the toilet is set at the center. Moreover, considering the possible interference due to the change in the facility area, the area of the toilet is equal to that of a commercial area at the same position in a double-toilet scheme, so it is approximately equal to the commercial area in practice.
Simulation experiments are conducted to obtain the crowd density map (Figure 8). In the case of only changing the facility location, the number of users in each facility remains unchanged. The route with the highest crowd is the route from the entrance to the toilet, and the most crowded areas on the opposite side of the toilet are the left and right sides in front of the toilet. For the key paths from both sides of the entrance to the toilet, the al and ar sections, shown in Figures 8(a) and 8(b), are selected, respectively. Figure 9 shows the population flow with time, where the left figure shows the relationship from toilet, which is at the central location on the opposite side of the entrance and exit, and right one is of toilet that is in the central location on the same side of the entrance and exit.
(a)
(b)
(a)
(b)
Figure 9 shows the relationship between the population flow and time, where the x-axis is the start time of the peak period, and y-axis is the person-time per hour. By comparing Figures 9(a) and 10(a), we find that the two toilet layout methods have similar changes in the path of the highest population flow to the toilet. However, with the toilet location changed, the average population flow drops from 995 p/h (left) and 1332 p/h (right) to 933 p/h (left) and 1173 p/h (right), respectively. Through the simulation data, it can be found that the rate of changing the position of the toilet is 6.23% for the pedestrian flow on the left path and 11.94% for the flow on the right path. In addition, the change in the population flow shows a certain tidal pattern. After 20 min, the first group of diners begin to enter the toilet after they finish their meal. Therefore, a second wave of population flow peak is formed. Subsequently, there is a wave of population flow peak every 20 min, which coincides with the dining time. Through the comparison of the two figures on the left and right, it can be found that after the toilet location is changed, the change trend is the same; however, the data size shows a certain decreasing trend compared with the original. This proves that changing only the toilet location cannot change the population flow characteristics, but it can have an impact on crowd gathering. Figure 10 shows the changes in the population flow of the most congested areas in the original structure (sections bl and br in Figure 8(a)) over time.
In the density graph, the most congested area and the most unfavorable path after changing the toilet location coincide. By comparing the average population flow in the most unfavorable area after changing the location, we find a significant decrease in the average population flows on the left and right paths. The average population flow on the left path drops from 1364 to 933 p/h and from 1967 to 1332 p/h on the right path. The rates of change were 31.60% and 32.28%, respectively. This proves that optimizing the toilet location in the comprehensive service area can effectively reduce crowd gathering.
(2) Impact of a double-toilet scheme on crowd gathering. Another toilet optimization scheme is evaluated below, which is to change the number of toilets and set up double toilets in the service area (Figure 11). After the change, double toilets are set on both sides. To facilitate the population flow to the restaurant after going to the toilets, the toilets are set up with two exits, and the total service capacity of the toilet remains unchanged at 266 persons (i.e., the maximum service capacity of each toilet is 133 persons). The number of people serving is still the highest (i.e., there is no queuing phenomenon). Figure 12 shows the density of the population flow of the scheme. After running the model, we noted queuing in front of the toilets on both sides. The maximum number of users was 129 persons in the right toilet and 101 persons in the left toilet. After implementing the double-toilet scheme, we observed a significant drop in the population flow density. There are no longer any areas with a population flow density higher than 1.5 persons/m2 in the entire graph. Comparing the number of toilet users in the single-toilet and double-toilets schemes, we found that the average number of simultaneous users in the single-toilet scheme is 132 persons, and the maximum number is 172 persons. In the double-toilet scheme, the average numbers of simultaneous users in the left and right toilets are 63 and 100 persons, respectively. The maximum number of simultaneous users in the left toilet is 80 persons, whereas the maximum number of simultaneous users in the right toilet is 128 persons.
Through simulation, it is found that implementing double toilets on the left and right sides of the service area can help significantly reduce crowd gathering. With this implementation, there is no longer any area where the crowd density exceeds 1.5 people/m2 in the crowd density graph. From the analysis of the number of users and the use efficiency, the use efficiency of a single toilet is found to be less than 50%, i.e., only 49.718%, which is consistent with the actual situation. This shows that the current toilet design in the service area is unreasonable, with approximately 50% of the toilet resources being wasted during peak hours on weekends. The average usage efficiency of the double-toilet scheme is 61.549%, which is greater than that of the single-toilet scheme, and the maximum usage rate is 78.195%. This confirms that setting up double toilets can help make better use of toilet resources than setting up a single toilet. Comparing the user numbers for the two toilets, we find that the change trend in the user number for the right-side toilet is similar to that for the central single toilet. The change trend in the user number for the left-side toilet is different from those for the other two toilets. There are two peaks of population flow when the model runs for 40 and 80 min, whereas the other two toilets have a population flow peak approximately every 20 min. Based on the analysis of the population flow density graph, the left-side toilet is closer to the commercial area, so a large number of pedestrians enter it after shopping, whereas the right-side toilet is closer to the catering area, so the change in the population flow is consistent with the meal time.
4.3.2. Impacts of Toilet Location in Square-Type Service Area on Crowd Gathering
Based on survey data and previous assumptions, AnyLogic is used for the modeling and simulation. Because the design of the parking location in the square-type service area is more scattered, it is simplified into two entrances, namely, a small entrance above and a main entrance below, and the proportion of the number of people entering is 1:2. Figure 13(a) shows the simplified model diagram. Figure 13(a) shows the structural diagram after changing the toilet position. The scale in Figure 13 is 1 m = 1 pixel.
(a)
(b)
Figure 14 shows the population flow density in the Tai’an service area after the experiment. As shown in Figure 14(a), the most congested path is the path from the toilet to the store. As shown in Figure 14(b), the most congested path is the path from the toilet to the main parking area.
(a)
(b)
The section a where is the most crowed area in Figures 13(a) and 13(b) is selected, respectively, and the relationship between the population flow and time is calculated, as shown in Figure 15. Through simulation experiments, it is found that setting the toilet at the center can effectively reduce the population flow density in the most congested areas of the service area. After changing the location, the population flow in the most congested area drops from 1344 to 1234 p/h, a decrease of 8.18%. The population flow trend is similar to that in the comprehensive service area; i.e., the peak of the second wave of population flow is reached in approximately 40 min, and the peak of the next wave of population flow is reached every 20 min. However, the center location of the toilet causes the restaurant staff to reach the toilet closer. Hence, the population flow trend is slightly earlier than before. A comparison of the number of simultaneous toilet users shows that the two layout methods have almost no impact on the number of simultaneous toilet users. The number of people before the change was 44 persons, and the number of people after centering the toilet is 45 persons. The maximum number of people does not change either. The maximum value is 65 persons for the toilet on one side and 61 persons for the toilet centering scheme. This also indicates that simply changing the toilet location has no effect on the number of users and the use efficiency.
4.3.3. Impact of Toilet Location in Independent-Type Service Area on Crowd Gathering
The independent-type service area is different from the square-type service area. First, all the facilities have a single entrance and exit. Second, the parking area is directly in front of the gate of the main building in the service area; i.e., after passengers get off the bus, they will be facing all the facilities in the main building of the service area. Due to the extremely low use frequency of accommodation facilities in the service area, this experiment does not involve accommodation services. The drinking water station is set outdoors and can be assessed without having to enter the main building of the service area, so the drinking water station is also not considered. Figure 16(a) shows the simplified model diagram of the traditional service area. Figure 16(b) shows the structure diagram after changing the toilet position and setting it at the center. The scale in Figure 15 is 1 m = 1 pixel.
(a)
(b)
Figure 17 shows the simulated population flow density. As shown, the most congested area in the two graphs is the area in front of the toilet, that is, section a in Figures 16(a) and 16(b).
(a)
(b)
For section a in Figures 15(a) and 15(b), respectively, Figure 18 shows the relationship between the population flow and the time in the most congested path.
After setting the toilet at the center, we find a significant decrease in the average population flow in the most congested area of the service area. Moreover, the trend in the population flow changes due to variations in location. After the toilet is located at the center, the population flow advances a certain time earlier than before. The average population flow drops from 1307 to 922 p/h, decreased by 29.46%. The change trend in the population flow is still similar to that in the first two service areas. However, due to the low population flow, the peak value at the peak of population flow does not increase significantly. Similar to the previous results, the average numbers of simultaneous users of the toilet in the two schemes are 46 and 51, respectively. There is no significant difference in the average value between the two schemes, and the maximum values are 68 and 70, respectively. Therefore, changing the toilet location has no effect on the number of toilet users and usage efficiency in the independent-type service area.
4.3.4. Relationship between the Facility Scale and the Flow Based on the Crowd Gathering
The demands and number of passengers in the service area are different at different times, leading to a change in the number of people entering the facilities in the service areas with time. Previous studies [9–12] focused on evacuation as well as emergencies in the case of subways and airports. They did not conduct any in-depth study on the changes in the population flow and the characteristics of the scale of internal facilities. Therefore, we change the total population flow to the service area to simulate the changes in the passenger flow considering the varying demand during weekdays and holidays. In addition, through the application of the AnyLogic software, the number of users of each facility in the service area is simulated under different flows, so as to evaluate the interior of the service area. Finally, the relationship between the facility scale and the flow based on the crowd gathering is obtained.
The incoming flow to each service area is increased by −500, −1000, 500, 1000, and 1500 (p/h), respectively, and the running time of the model is extended to 180 min. After the increase, the average population flows of each facility in each service area are listed as in Table 11.
After the experiment, when the total incoming flow reaches 3115 p/h, there is a more evident queuing phenomenon in the restaurant after the 66th min from the beginning of the peak period. A large number of people gather at the restaurant entrance, making the population flow density exceed 1.5 m2/person in certain areas of the restaurant entrance. When the total incoming flow reaches 3615 p/h, after the 39th min, i.e., at 11:39, there is a line of more than a hundred people in front of the toilet. At this time, the data show that the number of people who need to use the toilet is 363, and the scale of facility is only 262 persons. Therefore, more than a hundred people gather at the entrance of the bathroom, causing safety hazards. At this time, the queue length is 101 people. After the 54th min (11:54), the density graph shows that the crowd extends to the right entrance. When the population flow increases to 4115 p/h, the toilet is congested at the 28th min (11:28). At this time, the queue length is 177 people, and the service capacity of the service area cannot meet the requirements. Therefore, when the toilet seats 266 people, the maximum flow of people in the service area is about 3,115 p/h.
Table 12 lists the average population flows of each facility in the square style service area (the toilet is on one side). When the population flow in the service area increases to 2200 p/h, there is congestion in front of the toilet at the 68th min (12:08), and a large number of people enter and exit the toilet. At this time, the number of toilet services is close to its upper limit, reaching 147 persons. By the 100th min (12:40), a large number of people gather in front of the toilet, and the service capacity of the toilet reaches the upper limit. At the 180th minute, the entrance and exit of the toilet are crowded, and the service function of the toilet fails. However, the actual peak time exceeds 180 min; i.e., the result will not appear in the weekend peak hours of the service area. After the population flow in the service area increases to 2700 p/h, the toilet is heavily congested at the 35th minute with 217 persons in line. The total queue length is 375 persons, and there is severe congestion in front of the toilet. At this time, the service capacity of the service area can no longer meet the requirements. Therefore, when the toilet capacity is 158, the service area must not exceed 2200 p/h. The service area should take flow limiting measures before reaching this value and control vehicle entry.
Table 13 lists the population flows of each facility in the independent-type service area. When the initial population flow increases from 500 to 1432 p/h, the number of people in the restaurant increases monotonously with time. At 13:00, the number of people in the restaurant is close to 400 persons. At this time, the number of people significantly exceeds the carrying capacity. However, peak hours generally do not exceed 2 h, and hence, the traditional service area can still serve the population flow at this time. After the population flow increases to 1932 p/h, there is no queuing in front of the toilet, and the average number of people is 93 persons. However, there are too many people in line at the restaurant. When the population flow reaches 2432 p/h, and the model runs to the 35th minute, a crowd gathers in front of the toilet. By the 130th min, there is a large queue to the toilet. At this time, the population flow in the most congested area is 1012 p/h, and the service area cannot meet the usage requirements. Therefore, when the toilet capacity is 158, the traffic flow should be controlled at about 2000 p/h.
5. Discussion
To study the design of the internal facilities in the service areas of fully enclosed freeways, we obtained data, such as the pedestrian traffic characteristics, from freeway service areas through literature analysis and field investigation. With this, we established a crowd flow model, with crowd gathering as the evaluation criterion. Simulations were performed on service areas with three different building types. The following conclusions can be drawn from the study:(1)Research on service areas with different types of pedestrians was conducted through field experiments to determine the number of facilities required and the probability of using them, average residence time of the pedestrians in the service area, and pedestrian walking speed. Through a statistical analysis, we summarize the service area four characteristics of flow of pedestrian, and the results provide reference for further research and simulation experiments.(2)During weekend peak hours, the toilet usage efficiency in the different types of service areas under the current scale was less than 50%, which led to significant wastage of resources. The reason may be that, in the three types of service areas, the time interval of using the toilet is [16, 26] min, which makes the toilet turnover rate very high, thereby decreasing the number of people who wish to use the toilet at the same point in time. Therefore, in the design of toilets in service areas, the flow of population should be accurately estimated to avoid any shortage or wastage of resources due to too small or too large toilets. Moreover, simulation experiments showed that changing only the toilet location has no effect on the usage efficiency of the toilet, whereas changing the number of toilets can affect the usage efficiency. Setting up double toilets on both sides closer to the commercial and catering facilities made it convenient for personnel, with the usage efficiency of the double-toilet scheme being 10% higher than that of the single-toilet scheme.(3)The simulation showed that changing the location and number of toilets in the service areas can reduce the congestion in the path to the toilet and staff gathering in front of the toilet. In the interior design of the service area, if the path to the toilet is separated from the path to the supermarket, congestion due to overlapping or similar paths can be avoided.(4)The carrying capacity of different types of service areas may be related to the service scale of the toilets. After a significant increase in the flow of people, the toilets are the first facilities in the service areas where crowds queue. Through the simulation, it was found that when the passenger flow of the comprehensive service area exceeds 3115 person times/h, the toilet scale should not be less than 266 persons; the capacities of square-type and independent-type service areas should not be less than 158 and 148 persons when the hourly passenger flows reach 2200 and 1932 person-time/h, respectively.
This study also considered the outburst of travel during holidays, and the results can provide a reference for the design of internal facilities in freeway service areas. Moreover, we investigated and analyzed the traffic characteristics of service areas and used them to simulate the population flow. Compared with previous studies, this research is more in line with the actual conditions of service areas, and the results have better reference value. In future research, the differences in the use of various facilities by women, children, and disabled people will be considered in the simulations. Moreover, the use of more facilities will be considered, such as drinking water areas and garden rest areas. The data in this paper are applicable only to a few domestic service areas. The sample data are insufficient, and the results may need to be further verified. Although we selected data that can reflect the general situation, the focus of the service area in different regions may be different. Hence, the data may not accurately reflect the true situation of the facility layout scheme.
6. Conclusions
Through field investigation, data pertaining to pedestrians in service areas were obtained and analyzed. The behavior characteristics of the pedestrians in the service areas were obtained and summarized. The results provide reference for future research and simulation experiments.
Through simulation, this study determined the use efficiency of three types of toilets in service areas and analyzed the factors influencing the toilet use efficiency. Changing the number of toilets was found to have a greater influence than changing the location of the toilets without changing their total service scale. On the premise of determining the scale of the toilet service, the pedestrian carrying capacities of three types of service areas were simulated. Simulation determines the ultimate carrying capacity of the toilet of the three service areas. This model can test the design scale of the toilet in the service area to ensure that the service level meets the standard and has no resource waste. Our results can provide reference for the scale design of internal facilities in service areas. In future, we intend to generate a large-scale customized dataset and use big data analytics tools to get more insights and simulate the problem in a better way (Figures 18).
Data Availability
The data used to support the findings of this study are included within the article.
Conflicts of Interest
The authors do not have any conflicts of interest with other entities or researchers.
Authors’ Contributions
Conceptualization was done by Xufeng Li and Jinliang Xu; data curation was done by Xufeng Li and Shuo Han; formal analysis was done by Xufeng Li; funding acquisition was done by Jinliang Xu; investigation was done by Xufeng Li and Shuo Han; methodology was done by Xufeng Li and Jinliang Xu; project administration was done by Jinliang Xu; supervision was done by Jinliang Xu and Yaping Dong; validation was done by Jinliang Xu and Yaping Dong; visualization was done by Xufeng Li; writing–original draft was done by Xufeng Li; writing–review and editing were done by Xufeng Li.
Acknowledgments
The authors are grateful to the Shaanxi Provincial Department of Transportation for generously providing relevant information and the freeway design documents of the test road sections. They also thank the drivers for their cooperation during the field experiment. This research was funded in part by the Transport Technology Project of Shaanxi Province (Grant no. 18-23R).