Analysis of the Factors Affecting the Construction of Subway Stations in Residential Areas

Abstract

To design a more suitable scheme under different conditions so that subway stations can play their role better, this study investigated the construction elements of subway stations in residential areas. Metro stations in residential areas are generally located at the intersection of urban roads. As the “handover space” of the city, the construction principle should be based on people’s experience. As the core basis for traffic planning and future operation management, construction of subways in residential areas should take into consideration factors such as trip volume, distribution, and mode selection of residents, and be determined based on such mathematical models as the unit factor method, gravity model, and disaggregated analysis method. Station site selection is based on the point, line, and surface elements, and the importance of a station in the line network is judged by degree and betweenness centrality; the accessibility of the line network is determined by connectivity. In this study, the influencing factors of residential area subway station construction are divided into construction elements and site selection. Internal construction elements of stations include station entrances and exits, escalators, ticket machines, and transfer routes, and the conclusion of the mathematical model is used to select or give opinions about the internal construction elements of the subway. The point, line, and surface elements and the connection relationship between the subway and buses are used to determine the site selection of the subway. Furthermore, this paper discusses the three elements that affect the construction of the subway, comprehensively considers the functional requirements of the subway, and makes reasonable adjustments to each element. Finally, the requirements for the elements of the subway construction are determined.

1. Introduction

Underground rapid transit rail systems (subway systems) are a mode of public transport and efficiently service a large number of people. Their service quality is closely related to the spatial location of above-ground pedestrians, the degree of saturation, and the structure of different subway lines. Solutions to urban traffic congestion and inadequate transport capacity cannot be limited to constructing more motorized roads; the prioritized development of subways should be achieved through the development of a rational rail transport scheme. As the design of cities has shifted from a single-center layout to a multi-center layout, subway lines have become the best choice for public transportation in terms of capacity and efficiency. A city’s central areas can be connected to many suburban areas via a subway network to achieve effective and coordinated development of each area. In the process of subway station network design selection, the construction of stations depends on whether the spatial location of a subway station in the network is in a central area or a peripheral area, and whether station passenger flow is high or low. Additionally, the construction of subway stations should comply with the requirements of connectivity in line planning.

Studies have shown that residential development and population growth around new subway stations is more pronounced in suburban areas than in city centers and satellite towns, especially in new suburban areas that are burgeoning with subway development. As urbanization progresses, transportation infrastructure becomes more sophisticated and residents’ demand for public transportation services increases, leading to the diversification and increased complexity of residential subway station construction. More specifically, direct benefits to residents include convenience of travel and reduced travel time and cost.

Consequently, cities have begun to develop rail transit construction on a large scale, making it particularly important to analyze problems related to the construction of subway stations in residential areas. Presently, research on the construction of stations in residential areas is focused on the form and layout of the network based on predicted passenger flow; few systematic studies have been conducted on the construction of subway stations. Therefore, based on existing research, this study provides an in-depth analysis of subway station planning and construction. On the basis of passenger flow forecasting using mathematical models such as the unit factor method, gravity model, and the dis-aggregated analysis method, it combines residents’ travel volume, distribution, and mode selection, in addition to station internal construction, to scientifically analyze relevant influencing factors of subway station construction in residential areas. Additionally, it provides construction opinions, highlights considerations, suggests a method and theory of subway station construction for decision makers, and suggests improvements in the spatial integration between subway stations and residential areas. The specific process is shown in Figure 1.

2. Literature Review

In 1962, when the Chicago Area Transportation Study was published, the four-stage prediction method was first proposed [1]. The application of the four-step approach proposed by Church and Clifford, which is the most commonly used method for passenger flow forecasting in rail transit construction, provides an important theoretical basis for addressing passenger travel path selection. Passenger trip forecasting is used to predict total trip generation and attraction. Initially, most cities in the United States used linear regression models; thereafter, cross-classification methods were proposed. Finally, the currently used unit factor method was developed based on the non-home-based trips model proposed by Gordon et al. and the work trip generation forecasting model proposed by Yam et al. [2,3].

The most typical gravity model for predicting the distribution of passenger trips was first proposed by Casey and used to analyze daily trips between settlements within a certain area. Many types of gravity model have emerged through several improvements, but they do not completely depart from the basic form of the gravity model. After a period of practice and application, Furness [4] proposed the growth factor method. In addition, Stouffer [5] constructed the intervention opportunity model on the basis of gravity model and growth factor method, which was later improved by Schncider.

The division of the choices of transportation modes for trips between transport distribution and assignment is mainly performed using the gravity model combined with traffic distribution and the disaggregated model. These models were initially explored and developed to a practical level by scholars such as Warnr and Lemman; however, the research was not very effective. The logit and probit models proposed by Domencich, which use different probability distribution functions to frame their discrete models, are still used [6]. By the end of the 1980s, the BCD and Dogit models proposed by Gaudry and Dagenias, and the generalized logit model proposed by Gerken, made many improvements to the shortcomings of the logit model. Now, the F-W algorithm proposed by Frank and Wolfe [7] is widely used. Wang Shusheng [8] carried out in-depth research on passenger flow forecasting methods of urban rail transit. Based on the transport corridor, he proposed the basic framework of corridor analysis and passenger flow forecasting, and the disaggregated analysis method represented by the logit model. In addition, by introducing the transport corridor, he studied the attraction range of rail transportation and, combined with TransCAD traffic planning software, he proposed the utility matrix-based passenger flow prediction methods, RP and SP.

Additionally, many scholars have focused on the prediction of urban rail passenger flow in the short term. Yochi et al. [9] proposed a real-time passenger flow prediction method for rail transportation; Wei [10] proposed a hybrid empirical mode decomposition and back-propagation neural network forecasting approach; and Ozerova [11] analyzed factors affecting intercity passenger flows and used a regression model to predict them. Based on the discussion at the passenger flow law from the train schedule and other aspects, T. M. hrushevska [12] predicted the daily passenger flow of suburban railways.

The current development of short-term passenger flow forecasting is mainly based on time series forecasting, which uses the past to analyze and speculate on the sequence of things and the law of change to fit and identify. Ahmed et al. [13] first proposed the application of the auto-regressive integrated moving average (ARIMA) model to predict short-term passenger flows, which achieved satisfactory results and is favored by many scholars at home and abroad, especially in the field of transportation. Sangsoo [14] used the ARIMA model to predict the short-term passenger flow and estimated the model parameters through the Maximum Likelihood Estimate. The results show that the ARIMA model has higher accuracy than other time series models. Kumar and Vanajakshi [15] used a trunk line in India to build a seasonal ARIMA (SARIMA) model to predict the morning peak passenger flow using a small amount of data, eliminating the limitations of the ARIMA model, which requires a large amount of data input. Ding et al. [16] decomposed subway passenger flows into two parts: a constant term and an error term, and predicted the constant term of passenger flow using the ARIMA model and a generalized auto-regressive conditional heteroskedasticity (GARCH) model to detect the volatility of passenger flow during peak periods or large events. Zhao Yawei [17] established a multi-dimensional time series short-term traffic flow forecasting model by comprehensively considering the impact of legal holidays, tolls, weather, and other external factors on expressway traffic flow. Meng Pinchao [18] proposed a short-term passenger flow prediction method for rail transit based on the moving average method. The historical passenger flow data in the same time interval every day are processed with the moving average (MA), and the traffic flow forecasting in the same corresponding period is obtained. At the same time, the real-time passenger flow data are collected to correct the prediction results.

As can be found from the above documents, many studies have extended and expanded the four-stage prediction method. According to the development law of the time axis, some previous studies focused more on the gravity model, logit model, and probit model, whereas the prediction of the current model is more inclined to the short-term prediction of the autoregressive model. Despite the above achievements of the four-stage prediction method, the previous studies still have some limitations. This study forms a more complete methodology by integrating passenger flow prediction with subway construction, as shown in

Table 1. Comparison of research gaps.

Limitations of Previous Studies	Improvements of This Study
The form and layout of the traffic network are determined by predicting the traffic passenger flow, and there are few studies on the internal construction of subway stations.	The variables in the mathematical model are used to predict the number of future traffic passenger flow, travel distribution and traffic choice, so as to deeply analyze the construction of subway stations.

.

3. Methodology

At present, the method of predicting the annual traffic passenger flow in the future obtained by the mathematical model is called the forecast passenger flow, which has the characteristics of stage, similarity, and growth. In this study, the classical four-stage prediction method, which is popular in urban rail transit passenger flow prediction at home and abroad, was used to put forward suggestions for the construction of subway stations both inside and outside. In addition, it was verified by an example to explore a more practical and applicable construction of subway stations in residential areas.

This paper mainly analyzes the first three stages of the four-stage prediction method and puts forward a model based on the analysis of the influencing factors of subway station construction, as shown in Figure 2.

3.1. Forecasting Passenger Trips by Unit Coefficient Method

The level of refinement of the content of rail transit passenger flow forecasting varies, as do the requirements for station construction—a situation related to the stage of the rail transit project. Scientific and accurate prediction of travel volume can provide a means to assist in decision making for passenger flow inducement and operation planning of urban rail transit.

The unit factor method has been used to forecast long-range passenger flows for passenger trip generation and attraction, using different trip purposes of passengers as the independent variables and trip generation as the dependent variable, to determine high-, medium-, and low-density passenger trip forecasts based on socio-economic forecasts.

$$Pi={\displaystyle \sum _{i=1}^n{C}_{pi}{X}_{pi}}$$

(1)

where p_i is the trip generation for the i district; X_pi is the total population, number of adults, students, or jobs in the i transportation district; and C_pi is the trip generation factor corresponding to X_pi.

The time series method is used to build a mathematical model using curve fitting and parameter estimation based on time series data observed by a system. The model is a type of dynamic series analysis, and all that is required are historical data of the series itself. The time series method includes the auto-regressive moving average (ARMA) and ARIMA models. The most commonly used is ARMA, whereby the model is built and fitted with historical information to determine the parameters, and then the short-term passenger flow is predicted.

The ARMA model’s extrapolation requires only historical annual data without considering other influencing factors and is not suitable for forecasting when there is significant long-term variation. Therefore, the model is more suitable for routes with short-term passenger forecasts and for which historical annual operating data are available.

3.2. Forecasting Passenger Trip Distribution by Gravity Model

The passenger trip distribution model was developed by decomposing trips in a district into trips between individual districts, which are usually calculated using the gravity model. The model assumes that the amount of travel between districts i and j is inversely proportional to the amount generated by i and the amount attracted by j, and inversely proportional to the impedance parameter (time or distance) between the two districts, as shown in the following equation:

$$T_{ij}=k\frac{T_i^{\alpha }{T}_j^{\beta }}{d^{\gamma }}$$

(2)

where $T_{ij}$ is the trips between transport districts i and j; $T_i$ is the trips occurring in transport district i; $T_j$ is the trips attracted to traffic district j; $d$ is the impedance parameter (time or distance) between transport districts i and j; $k$ is the socio-economic correction coefficient; and $\alpha ,\beta ,\gamma$ are constant terms.

Traffic impedance can be due to time, distance, or cost, but in most cases a factor is usually taken as the impedance parameter for simplicity of calculation, and the supplementary modified gravity model again refers to the urban layout coefficient to make it more consistent with the actual urban situation. It is called a composite gravity model, and expressed as follows:

$$\theta =\frac{P_i{A}_j{F}_{ij}{K}_{ij}}{{\displaystyle \sum _j{A}_j{F}_{ij}{K}_{ij}}}$$

(3)

where $\theta$ is the trip distribution between transport districts i and j; $P_i$ is the trip occurrence in transport district i; $A_j$ is the total attraction in transport district j; $F_{ij}$ is the trip impedance factor between traffic districts i and j; and $K_{ij}$ is the urban layout coefficient adjustment between transport districts i and j.

3.3. Forecasting Travel Mode Options by Dis-Aggregated Analysis

The classification of the choice of travel mode for residential trips is generally performed at different stages of the planning process to derive models for the choice of different travel modes. The gravity model and the probabilistic prediction model are used to model the transport distribution and assignment in the planning process, respectively.

In the gravity model combined with traffic distribution, the dependent variable is determined by the volume of passengers generated and attracted between transport sub-districts i and j on a given route. The greater the impedance coefficient between sub-districts, the smaller the distribution on that route for that traffic mode, as shown in the following equation:

$$T_{ijm}=p_i\frac{A_i\times T_{ijm}}{{\displaystyle \sum _j{\displaystyle \sum _m{F}_{ijm}}}}$$

(4)

where $T_{ijm}$ is the distribution of mode m between transport sub-districts i and j, and $p_i$ and $A_i$ are generated and attracted volumes, respectively. $F_{ijm}$ is the impedance coefficient of mode m between sub-districts i and j.

Where the subway is considered as the transportation mode of choice for trips, this gravity model considers the impact of the subway on the distribution of trips through the impedance factor between two districts.

The disaggregated analysis method divides travel modes into distribution and assignment, which makes the choice of transportation mode specific to the individual, and the basis for the choice is determined based on the differences in the utility of different transportation modes. Choices are in the form of probabilities. The dependent variable is the condition that externally affects the individual when using that mode of transport, as shown in the following equation:

$$P_{mk}=\frac{e^{V_{mk}}}{{\displaystyle \sum _j{e}^{V_{mk}}}}$$

(5)

where $P_{mk}$ is the probability of the K individual using mode m, and ${}^{V_{mk}}$ is the influence factor of the K individual using mode m, typically as a function of travel time t, cost c, and comfort s.

For the utility of using a certain mode, the equation is:

$$V=\alpha \times t_{mk}+\beta \times c_{mk}+\gamma \times s_{mk}$$

(6)

where $\alpha$, $\beta$, and $\gamma$ are constants.

The probability of residents using the subway obtained via the dis-aggregated analysis method has the advantages of a small sample size, strong time and regional transferability, and high precision; however, it is more difficult to determine the influence factor using this method.

4. Results

4.1. Trip Volume and Station Construction

The different urban functions around subway stations can be divided into three main types: work, study, and shopping. The attraction intensity under different conditions places different requirements on station construction. Stations near schools, commercial centers, and office buildings have high attraction intensity for study, shopping, and work trips, respectively. When the attraction intensity of study trips to residential area passenger flow is high, the subway station should be built on the transport line to the school district to avoid transit as far as possible, and the distance between the station and the entrance and exit to the residential area should be short. The location of a residential area and residents’ income level determine a district’s passenger flow for shopping. When constructing a station, it should be located on the line to a mall or shopping center, which can be used as a transit station. In residential areas, the highest passenger flows are for the purpose of work trips, so stations are usually built as interchange stations to facilitate passenger flows to different districts with concentrations of work areas.

A reasonably connected residential subway station can shorten the travel time of residents, improve the attractiveness of the subway, and attract greater passenger flow to the residential area around the station for interchanges. Internal interchanges are divided into three forms: platform, station hall, and passageway interchanges. Due to the small scale of current interchange stations and the lack of existing interchange space, passageway interchange is most often used, but the distance between passageway interchanges is too long.

The reasonable distribution of residential subway station entrances and exits can effectively alleviate pressure during peak passenger flows. Based on the density of residential areas within the coverage area, there are three forms of distribution of residential station entrances and exits, as shown in

Table 2. Distribution and width of entrances and exits.

Subway Station Area Coverage	Distribution Form	Relationship between B and b
Low- and medium-density residential areas	Two unidirectional openings on both sides	$B\ge b$
High-density residential area	Two bidirectional openings on both sides	$B\ge b\times \partial /2$
Mixed residential, commercial, and administrative areas	Four bi-directional openings on both sides	$B\ge b\times \partial /2$

and Figure 3.

In Table 2, b is the passageway width (m); B is the width of entrance/exit; and ∂ is the diversity factor, generally from 1 to 1.25.

The entrance and exit capacities of subway stations in residential areas are a factor affecting passenger travel. A wide entrance/exit passageway will cause technical and construction difficulties, but a narrow entrance/exit passageway will affect flow during peak travel times and pose a safety hazard. To meet the above requirements, the net width of a passageway should satisfy the following equation:

$$b_1\ge \frac{R\times \alpha }{C_1}$$

(7)

where $b_1$ is the designed net width of an entrance/exit passageway (m); $R$ is the in/out passenger flow at the entrance/exit during peak hours (persons/hour); $\alpha$ is the uneven coefficient of entrance/exit passenger flow, generally from 1 to 1.25; and $C_1$ is the mixed capacity of a 1 m-wide passageway (persons/hour).

For passenger flow data in the direction of entrances and exits during peak hours, which are not easily available, an approximate range of passenger flow can be obtained with the support of data from ticket gates. The entrance ticket checkpoint is located between the ticket machine and the station concourse of the subway station, and the exit ticket checkpoint is located between the station concourse and the ground-level entrance/exit, as calculated in the following equation:

$$R=\frac{Nm}{K}$$

(8)

where $N$ is the number of ticket gates in and out of the station; $m$ is the capacity of each ticket machine per minute, which is 20 to 25 tickets/machine/minute; and $K$ is the over-peak hour factor, which is 1 to 1.4.

Beijing’s Tiantongyuan district is the largest housing district in Asia. It is a typical high-density population residential area, surrounded by three subway stations: the Tiantongyuan Station, Tiantongyuan South Station, and Tiantongyuan North Station. Additionally, the under-construction Tiantongyuan East Station, as the third station from the north of Line 17, is an important node for flows between Tiantongyuan and the city center, where residents travel mostly for work. The main travel times are the morning and evening peaks, creating a concentrated commuter flow. The form, type, and interchange routes of the station planned for the Tiantongyuan East Station are shown in

Table 3. Tiantongyuan East Subway Station.

Station	Station Format	Type of Station	Interchange Routes	Number of Subterranean Levels	Platform Type	Number of Entrances and Exits
Tiantongyuan East Station	East T	Interchange station	Interchange between Lines 17 and 13A and branch Line 17	2 levels	Dual island platform s with four lines	4

.

The locations of entrances and exits of subway stations in high-population-density residential areas determine the convenience of travel and the choice of travel modes for residents. The northeast quadrant of the Tiantongyuan East Station is the Likang Hongqiao Market, the southeast quadrant is green space, the southwest quadrant is the Tiantongyuan II residential area, and the northwest quadrant is the Tiantongyuan III residential area [19]. The locations and purposes of the entrances and exits at the Tiantongyuan East Station are shown in

Table 4. Locations and purposes of the entrances/exits of the Tiantongyuan East Station.

Entrance/Exit Location	Purpose
Exit A, in front of the residential area of Tiantongyuan East III, in a north–south direction along the road	Allows for passenger flows of Tiantongyuan III
Exit B, in front of the Likang Hongqiao Market, south-to-north along the road	Allows for passenger flows of the market
Exit C, within green space in the southeast quadrant	Leaves an entrance/exit for unplanned areas to accommodate passenger flows
Exit D, in front of the residential area of Tiantongyuan II, south-to-north along the road	Allows for passenger flows of Tiantongyuan II

and Figure 4.

Tiantongyuan station, Tiantongyuan south station, and Tiantongyuan north station are serial stations on Metro Line 5, relieving the traffic passenger flow of the Tiantongyuan community. The terrain formed by the four subways on Metro Line 5, Line 7, and Line 13A and the Tiantongyuan district they serve is, shown in Figure 5.

4.2. Trip Distribution and Station Construction

The transportation network is a large system, and its core content is network node analysis. Residential subway station site selection can use the point, line, and surface element method. In the course of site selection, it is necessary to pay attention to the connections between the three elements. Points are the basis of the transportation network. Their scale and function depend on the line network structure, and they usually relate to the attraction and occurrence points of large transport. Lines are the main traffic corridors in a city and usually the main routes through which a city’s passenger flows pass. They are the links preceding and following points and surfaces. Surfaces are used to comprehensively research districts and analyze influencing factors of subway construction in planning areas.

Key nodes on a subway network should be selected based on the actual situation of the subway network, combined with the ability to concentrate and disperse passenger flows in the surrounding area. Nodes are graded according to their topological structure, vulnerability, attraction weights, generation weights, and other important indicators. The specific measurement factors are shown in Figure 6 [20].

The index weights of connectivity, inbound passengers, and station exits were investigated in this study. Degree and betweenness centrality are static indicators of topological structure and are defined as metrics of node centrality in transportation network analysis, where the degree and betweenness centrality of a node represent node connectivity and the number of shortest paths through the node, respectively. The larger the result, the more important the node in the transportation network. The specific equation is as follows:

$${DC}_i=\frac{k_i}{N-1}$$

(9)

where ${DC}_i$ is degree centrality; $k_i$ denotes the number of existing edges connected to node i; and ($N-1$) denotes the number of edges where node i is connected to all other nodes.

$${BC}_i={\displaystyle \sum _{s\ne i\ne t}\frac{n_{st}^i}{g_{st}}}$$

(10)

where ${BC}_i$ is between centrality; $n_{st}^i$ is the number of shortest paths that pass through node i; and $g_{st}$ denotes the number of shortest paths connecting s and t.

In urban transportation planning and regional planning, when urban residents travel for work, study, or shopping, they generally converge along traffic arteries and form passenger flows. If a transport line between districts passes through the commercial center, cultural and entertainment centers, urban transportation hubs (such as train and bus stations), and other places with large passenger flows, it will effectively reduce the non-linear coefficient of the line as well as the travel time of residents. Consequently, during the planning phase, the accessibility of the route is obtained from the network connectivity to analyze whether it is consistent with a trunk line between districts. It is the ratio of the number of lines and the number of stations in the network:

$$C=\frac{L/\xi }{HN}=\frac{L/\xi }{\sqrt{AN}}$$

(11)

where C is the line network connectivity; L is the total length of the planned subway line network; H is the average spatial linear distance between subway stations in the planned area (km); N is the number of stations in the planned area; A is the area of the planned area (km²); ξ is the non-linear coefficient of the line network; and the meaning is the ratio of the actual distance between stations of the line network to the linear distance.

In terms of connectivity, the district area and the number of stations are the results of planning. Additionally, different intervals represent different line network structures. In the calculation of different line connectivity between two districts, when C is close to 1, the transport network layout is a radial network, and the interchange stations are mostly for two routes, as in Figure 7. When C is 2, the transport network layout is a raster network, and the interchange stations are mostly for four routes, as in Figure 8 [21,22].

After determining the station location in the residential area, trip distribution is calculated between districts i and j based on the trip distribution model, which provides the trip distribution between districts i and j influenced by the impedance coefficient. The larger the impedance coefficient, the smaller the amount of residential trip distribution. Using time and distance as examples, the elements of site construction for different types of planning scenarios to achieve higher trip volumes are shown in

Table 5. Site construction elements for different planning types.

Type of Planning	Site Construction Elements
Urban transport planning	The subway stations in districts i and j should be direct stations to reduce transit. Access widths should be designed to maximize peak hour pedestrian flow to save travel time.
Regional planning	The entrances and exits of the subway stations should be connected to the entrances and exits of housing communities to avoid excessive distances. Reduce interchanges between the subway and other modes of transportation in the interval from district i to j.

.

4.3. Travel Mode Selection and Station Construction

The factors influencing the choice of the mode of transport for travel are diverse and coupled with each other. They are graded according to the human–vehicle–route factors of influence. The “human” factor refers to the main body of the passenger flow, which is attracted by different social indicators. “Vehicle” refers to the form of transport chosen by passengers to travel, and the influencing factors are comfort, accessibility, and speed. “Road” is the influence of the route conditions on subway trips.

Presently, the factors influencing the choice of transportation are mainly time, economy, and personal experience, and the requirements for the construction of subway stations are different. Taking the influencing factor of “time” as an example, the subway is fast and time-saving compared with other modes of transportation [23]. By reasonably choosing the construction conditions of the internal entrance/exit escalators and the length of interchange passageways, travel time can be reduced, which can increase the proportion of residents who choose the subway as their mode of transportation; see

Table 6. Site condition capabilities and construction options.

Construction Conditions	Capacity Analysis	Site Construction Options
Entrance/exit escalators	The 1 m-wide escalator moves 6720 passengers per hour at a speed of 0.5 m/s, and the same width of stairs moves 4200 passengers per hour (going up).	Escalators are more prominent in terms of time saving or passenger throughput than subway station entrances and exits that use stairs to travel up and down.
Length of interchange passageways	The number of people passing through a 1 m-wide lane is 5000 per hour in one direction and 4000 per hour in mixed traffic in both directions.	The latest regulations on interchange passages stipulate that the interchange time for passenger flow shall not exceed 5 min, and the distance of the passages shall be appropriately shortened under the condition of meeting the peak passenger flow passage.

.

An insufficient number of in-station ticket machines will result in congestion and wastage of passengers’ time, which will cause them to choose other modes of transport. The formula for calculating the number of ticket machines in stations is as follows:

$$N=\frac{Mk}{m}$$

(12)

where $N$ is the number of ticket machines; $M$ is total inbound and outbound passenger traffic (peak hour); $k$ is the over-peak factor, which is 1.2–1.4; and $m$ is the hourly capacity of manual ticket selling facilities, which is 1200 tickets/ (H person) or the ticket selling capacity of the automatic ticket vending machine, which is 4~6 tickets/minute/set.

Buses and subways are the two main modes of public transport. In addition to connecting the two, we should consider how to increase the proportion of subways as the main mode of travel. The factors affecting the choice of subways for passenger trips in the network size formula derived from traffic demand are the rail transit line load intensity and traffic interchange coefficient, as shown in the following equation:

$$L=\frac{Q\alpha \beta \lambda }{\gamma }$$

(13)

where L is the length of the line network (km); Q is the total number of urban trips (million trips); α is the proportion of public transport trips and β is the proportion of subway trips to public transport trips; λ is the transport interchange coefficient; and γ is the rail transit line load intensity [million trips/(km-day)].

The equation shows that to increase the share of public transport trips that are subway trips, the load intensity of rail transit lines can be strengthened or the interchange coefficient can be reduced. The interchange coefficient can be reduced by adjusting the two influencing factors of the number of lines in the network and the network structure. The higher the number of lines in the network, the lower the interchange coefficient; the lower the number of interchanges in the network structure, the lower the interchange coefficient. Alternatively, a network with a small interchange coefficient is not necessarily efficient, and a large interchange coefficient indicates that the proportion of the passenger flow from other lines is greater.

The connection relationship between subway and buses in terms of their spatial location can be divided into centralized layout and decentralized layout according to the layout method.

Table 7. Station locations with a centralized subway and bus layout.

Layout Features	Subway Station Locations	Bus Stop Locations
The interchange between subway and bus is concentrated in one place, making it easy to use and efficient to transfer.	Middle of the road	Set on both sides of the access road and connected to the site via an underpass or pedestrian bridge.
	Side of the road	Set on both sides of the subway station, the bus stop is located on one side with the subway station.

, and Figure 9 and Figure 10, show the layout characteristics and the relationship between the locations of subway stations and the locations of bus stops when a centralized layout is used.

When a decentralized layout is used, the form of bus stops is the same as in the centralized layout, but the entrances and exits of subway stations are influenced by the layout of bus stops. The relationship between decentralized layout characteristics, bus stop forms, and subway station entrance and exit locations are shown in

Table 8. Station locations with a decentralized subway and bus layout.

Layout Features	Bus Stop Format	Subway Station Entrance/Exit Locations
Subways and bus interchanges are independent of each other, but weakly connected to each other and inefficient for interchange.	Island	The entrances to the subway stations are located on either side of the center of the bus stop.
	Side	The subway station entrances are connected to each bus stop via underground passages.

, Figure 11 and Figure 12 [24].

5. Case Study

Located in Jinggangshan road in the center of Huangdao District, Qingdao, the surrounding commercial and residential areas are densely distributed, and the occupancy rate of residential areas is extremely high. Based on the analysis of entrance and exit C of Jinggangshan subway station, Liqun Plaza, AEON shopping center, and other shopping malls are distributed nearby. The main traffic passenger flow comes from the surrounding residential areas such as Dongfang yinzuo, Wuyishan district, No. 18 yard of Fuchunjiang, etc. According to Formula (1), the number of current residents multiply the travel generation coefficient to obtain the value of future travel volume, as shown in

Table 9. Data of surrounding communities of Jinggangshan station.

Residential Area Name	Total Number of Households	Travel Generation Coefficient	Number of Current Residents (Valuation)	Forecast of Future Travel Volume	Proportion of Taking Subway as the Mode of Travel	Predicted Number of Subway Trips (Persons/Hour)
Dongfang yinzuo	532	1.5	1596	2394	40%	958
Wuyishan district	983		2949	4423	42%	1858
No.18 yard of Fuchunjiang	994		2982	4473	45%	2013

.

It can be seen from the above table that the number of people using the subway in the future travel volume is calculated according to the proportion. At the same time, the number of passenger flow entering the station in the peak period is taken as the predicted number of subway passengers; the value is 4829 (persons/hour). According to the data in the above table, in order to explore whether the existing internal construction of Jinggangshan Road station entrance C can meet the demand of predicted passenger flow in the future, we conducted a field investigation on the current situation of Jinggangshan Road station entrance C, as shown in

Table 10. Construction status of Jinggangshan C entrance and exit.

The Width of Entrance/Exit Passageway (m)	Number of Tickets Machines	Number of Ticket Gates	Number of Passengers Leaving the Subway Station at Peak (Persons/Hour)	Number of Entrance/Exit Escalators (Down)	Number of Entrance/Exit Stairs
5.8	8	9	1440	1	1

.

According to Formula (7), The peak passenger flow at the entrance and exit $R$ is the sum of the number of subway trips in the future and number of passengers leaving the subway station at the peak, which is 6269 (persons/hour). $\alpha$ ranges from 1 to 1.25, and $C_1$ is 4000 (persons/ hour). It can be concluded that the demand range of the entrance width is 1.6–2 m, and the width of the current entrance and exit passage meets the requirements.

According to Formula (8), nine ticket gates allow a maximum of 13,500 passengers to pass per hour, meeting the requirements of future passenger flow during rush hours.

According to Formula (12), eight ticket machines and two manual ticket windows in the station allow a maximum of 5018 people to buy per hour, meeting the requirements of future passenger flow during rush hours.

According to the conditions and capacities of the entrance and exit escalator sections in Table 6, the 1 m-wide escalator allows 6720 passengers to pass per hour, and the number of people who go down the stairs with the same width is 4200 per hour. The design of the existing entrance and exit escalator section meets the requirements of the future peak passenger flow.

6. Discussion

Regarding the influencing factor of the volume of residents’ trips on the construction of stations, the three most important aspects are the social indicators that attract residents to travel, the length and width of interchange passageways, and the form and number of entrances and exits. Depending on the functional positioning of the surrounding area, the requirements for the construction of the station can vary, and the design of the residential subway station as an interchange station should meet the predicted results of travel volume. Additionally, the distribution and capacity of entrance/exit passageways are influenced by the density of residential areas in the district.

Regarding the influencing factor of residents’ travel distribution on station construction, the siting analysis of residential subway stations is based on the point, line, and surface element method, which introduces impedance coefficients (time and distance) in the gravity model after considering the station’s importance in the transportation network and its ability to collect and disperse surrounding passengers.

Regarding the influencing factor of residents’ travel mode choice on station construction, the proportion of passenger subway trips can be increased through the construction choice of subway stations, strengthening the load intensity of rail transit lines, or reducing the traffic interchange coefficient. Subway stations and bus stops can be spatially divided into the two layouts of centralized and decentralized connections, and the two significantly differ from each other in terms of interchange and connection methods. These should be flexibly chosen based on their respective advantages and disadvantages.

7. Conclusions

As the most important component of a city, the planning and construction of subway stations around a residential area affects the travel efficiency and occupancy rate of residents and has a bearing on the economic and social development of a city. Given the obvious advantages of rail transit in relieving urban pressure, many subway stations have been built near residential areas, and it is important to study the site selection, station construction, and connectivity of rail transit networks.

This study analyzed the three elements of trip generation, trip distribution, and trip mode selection of residents around subway stations. The unit factor method was used to predict residents’ trip generation, the gravity model was used to predict trip distribution, and a combination of the gravity model and disaggregated analysis was used to predict trip transportation mode selection. The prediction results highlight an important method and key to determine the construction of subway stations in residential areas.