National Academies Press: OpenBook

Intelligent Transportation Systems in Headway-Based Bus Service (2021)

Chapter: Chapter 2 - Literature Review

« Previous: Chapter 1 - Introduction
Page 8
Suggested Citation:"Chapter 2 - Literature Review." National Academies of Sciences, Engineering, and Medicine. 2021. Intelligent Transportation Systems in Headway-Based Bus Service. Washington, DC: The National Academies Press. doi: 10.17226/26163.
×
Page 8
Page 9
Suggested Citation:"Chapter 2 - Literature Review." National Academies of Sciences, Engineering, and Medicine. 2021. Intelligent Transportation Systems in Headway-Based Bus Service. Washington, DC: The National Academies Press. doi: 10.17226/26163.
×
Page 9
Page 10
Suggested Citation:"Chapter 2 - Literature Review." National Academies of Sciences, Engineering, and Medicine. 2021. Intelligent Transportation Systems in Headway-Based Bus Service. Washington, DC: The National Academies Press. doi: 10.17226/26163.
×
Page 10
Page 11
Suggested Citation:"Chapter 2 - Literature Review." National Academies of Sciences, Engineering, and Medicine. 2021. Intelligent Transportation Systems in Headway-Based Bus Service. Washington, DC: The National Academies Press. doi: 10.17226/26163.
×
Page 11
Page 12
Suggested Citation:"Chapter 2 - Literature Review." National Academies of Sciences, Engineering, and Medicine. 2021. Intelligent Transportation Systems in Headway-Based Bus Service. Washington, DC: The National Academies Press. doi: 10.17226/26163.
×
Page 12
Page 13
Suggested Citation:"Chapter 2 - Literature Review." National Academies of Sciences, Engineering, and Medicine. 2021. Intelligent Transportation Systems in Headway-Based Bus Service. Washington, DC: The National Academies Press. doi: 10.17226/26163.
×
Page 13
Page 14
Suggested Citation:"Chapter 2 - Literature Review." National Academies of Sciences, Engineering, and Medicine. 2021. Intelligent Transportation Systems in Headway-Based Bus Service. Washington, DC: The National Academies Press. doi: 10.17226/26163.
×
Page 14
Page 15
Suggested Citation:"Chapter 2 - Literature Review." National Academies of Sciences, Engineering, and Medicine. 2021. Intelligent Transportation Systems in Headway-Based Bus Service. Washington, DC: The National Academies Press. doi: 10.17226/26163.
×
Page 15
Page 16
Suggested Citation:"Chapter 2 - Literature Review." National Academies of Sciences, Engineering, and Medicine. 2021. Intelligent Transportation Systems in Headway-Based Bus Service. Washington, DC: The National Academies Press. doi: 10.17226/26163.
×
Page 16
Page 17
Suggested Citation:"Chapter 2 - Literature Review." National Academies of Sciences, Engineering, and Medicine. 2021. Intelligent Transportation Systems in Headway-Based Bus Service. Washington, DC: The National Academies Press. doi: 10.17226/26163.
×
Page 17
Page 18
Suggested Citation:"Chapter 2 - Literature Review." National Academies of Sciences, Engineering, and Medicine. 2021. Intelligent Transportation Systems in Headway-Based Bus Service. Washington, DC: The National Academies Press. doi: 10.17226/26163.
×
Page 18
Page 19
Suggested Citation:"Chapter 2 - Literature Review." National Academies of Sciences, Engineering, and Medicine. 2021. Intelligent Transportation Systems in Headway-Based Bus Service. Washington, DC: The National Academies Press. doi: 10.17226/26163.
×
Page 19

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

8 Literature Review Headway-Based Service Introduction Headway-based service is a form of bus operation where the main objective is to maintain the headways between buses. This objective can be formulated in different ways. For example, the transit agency could try to keep buses spaced at a specified target headway, or simply try to maintain regular bus spacing on the route, in which case the average headway might vary based on other factors such as real-time passenger demand and traffic conditions. HBS can be contrasted with schedule-based service, where the main objective is for buses to maintain a schedule consisting of arrivals at specific locations at specific times. The transit literature has long made a distinction between infrequent bus service and frequent bus service (Jolliffe and Hutchinson 1975; Turnquist 1978), which differ in the way that passengers consult information and arrive at stops. The dividing line between the two is typically around a 10-minute headway (Fan and Machemehl 2009; Kittelson & Associates et  al. 2013; Ibarra-Rojas et  al. 2015), although some recent studies have suggested that it is affected by service reliability, and headway could be lower for routes with high on-time reliability (Luethi et al. 2007; Ingvardson et al. 2018). In infrequent service, passengers rely on the bus schedule and plan to catch a specific bus. They may need to adjust their desired departure time based on when the bus comes and will typically arrive at the bus stop just before the scheduled time. In this scenario, it is important for buses not to leave early because they might leave passengers behind. A late bus would typically encounter the same number of passengers as an on-time bus but impose additional out-of-vehicle waiting time, which passengers are known to dislike (Thobani 1984; Truong and Hensher 1985), proportional to the lateness. In frequent bus service, passengers may or may not consult the bus schedule or real-time information. For many passengers, the expected wait time is short enough that they will simply arrive at the bus stop at their desired departure time (Abkowitz and Lepofsky 1990; Hickman 2001). Mixed passenger behavior has also been observed in which some passengers consult the schedule and others do not (Luethi et al. 2007; Ingvardson et al. 2018). In this context, the number of passengers that a bus encounters at a stop tends to vary based on the headway between it and the bus in front of it, referred to as the forward headway. The resulting varia- tions in dwell time at stops lead to instability (Welding 1957; Turnquist and Bowman 1980) and problems with bus bunching, which are described in the following section. Headways can be measured anywhere along the route but are typically computed by comparing the arrival times at a particular stop. The now-widespread availability of real-time information has been shown to change the distribution of passenger arrivals at bus stops (Watkins et al. 2011), but C H A P T E R 2

Literature Review 9   the modeling of Fonzone et al. (2015) showed that non-uniform passenger arrivals may even exacerbate the bunching tendency in certain conditions. Headway-based service is most useful on frequent bus routes. In this context, focusing on a headway objective is consistent with passenger behavior. Schedule-based strategies have been found to be ineffective at controlling bunching (Newell and Potts 1964; Berrebi et al. 2018), while HBS is able to reduce bus bunching and the expected waiting time of passengers (Chang et al. 2004; Pulley et al. 2006). Headway-based operation may also allow buses to travel faster than in schedule-based operation by reducing or eliminating slack in the schedule (Chang et al. 2004). Challenges A key challenge in the operation of high-frequency bus service is bus bunching, also referred to as bunching and gapping. Bunching is a phenomenon in which adjacent buses tend to form pairs (Newell and Potts 1964; Chapman and Michel 1978). From the passenger perspective, this pairing effect leads to large gaps in service followed by the arrival of two (or more) buses in short succession. Bunching often starts with a small perturbation, such as a group of boarding passengers, which delays a bus. Figure 2 shows four buses initially spaced at an even headway, h. The second bus then picks up some delay that puts it behind its headway target by δ. Now the second bus has a larger forward headway, h+δ, and the third bus has a smaller forward headway, h-δ. On average, the second bus will now encounter a few extra passengers who might otherwise have boarded the third bus. The extra dwell time associated with these additional passengers will put it further behind. Similarly, the third bus will encounter fewer passengers and tend to run a little ahead. With no intervention, the second and third buses will eventually converge, opening large gaps in front of and behind the bunch. Researchers and transit agencies use various definitions of bunching. One definition considers any headway below a minimum threshold (e.g., 2 minutes) to be a bunching event (Feng and Figliozzi 2015). This definition with a 1-minute threshold is used by the Chicago Transit Agency (CTA) (Chen et al. 2013) and the Los Angeles County Metropolitan Transportation Authority (LA Metro) (Flynn et al. 2011). Another method for detecting bunching is to use the ratio between observed and target headway (Strathman et al. 1999). The literature has mostly focused on defining bunching, but gapping can also be defined and measured. CTA defines a big gap as twice the scheduled headway or 15 minutes, whichever is greater (McKone et al. 2009). Figure 2. Bus bunching example.

10 Intelligent Transportation Systems in Headway-Based Bus Service Bunching affects reliability, which passengers are known to value (Chang et al. 2004), through its impact on wait times and in-vehicle travel times. In a system with irregular headways, passengers’ wait time depends on when they arrive at the stop. Randomly arriving passengers are more likely to arrive during the longer headways. Irregular headways also lead to variations in in-vehicle travel times. As described previously, a bus behind a long headway will encounter more passengers, spend more time at stops, and as a result have a higher passenger load. Passengers riding these buses will experience slower travel times, and others could be denied boarding if the bus capacity is reached (Toledo et al. 2010; Muñoz, et al. 2013). Conversely, passengers on a bus behind a short headway will experience faster travel times. These sources of travel time variability in wait and in-vehicle travel times increase the variability of total travel times. Passengers who need to arrive at a particular time (e.g., for trips to work or school) respond by leaving earlier (Berrebi et al. 2015). Furth and Muller (2006) found that these time-sensitive passengers plan their travel on the 95th percentile arrival times. These add-on effects reduce the appeal of transit by more than might be suggested by average travel times. Performance Indicators Performance indicators and incentives can sometimes be a challenge for providers of HBS. Transit agencies have traditionally used on-time performance as a key measure, but it is not meaningful in HBS (Rietveld 2005), and these routes may not even have a schedule. Many different performance indicators are currently used or have been proposed for HBS. The Transit Cooperative Research Program’s TCRP Report 88: A Guidebook for Developing a Transit Performance Measurement System (Kittelson & Associates et al. 2003) lists four: • Headway adherence, defined as the standard deviation of headway deviations divided by the mean of scheduled headways (Kittelson & Associates et al. 2013). The coefficient of variation of headway, a similar measure, has also been used (Toledo et al. 2010; Cats et al. 2011). • Headway ratio, the observed headway divided by the scheduled headway. Nakanishi (1997) considered headways between 50% and 150% of the target headway to be acceptable. Carris in Lisbon, Portugal, and the Coast Mountain Bus Company in Vancouver use this measure with a range of 80% to 120% considered to be on target (Trompet et al. 2011). This measure is less useful if the scheduled headway is not constant (Trompet et al. 2011). • Headway regularity index, based on the Gini ratio (Henderson et al. 1991). The Gini ratio is a measure of dispersion and ranges from 0 (all values are equal) to 1 (one value is large, and all others are zero). • Service regularity, expressed as the percentage of headways within a certain interval. New York City Transit (NYCT) considers buses within 3 minutes of the scheduled headway to be on target (Cramer et al. 2009). The citations within parentheses are confirming citations. Henderson et al. (1991) is the source of the headway regularity index measure listed in the TCRP report. Other performance indicators for HBS include the standard deviation of headway (Toledo et al. 2010) and an irregularity index designed to penalize long headways (Golshani 1983). Trompet et al. (2011) proposed using the standard deviation of the difference between the target and actual headway for routes where the target headway varies over time. Bellei and Gkoumas (2010) analyzed headway distributions. Another type of performance measure focuses on passenger waiting time. London Buses uses a measure of excess waiting time assuming that passengers arrive at a uniform rate (Transport for London 2012). A similar excess waiting time measure is used by the Land Transport Authority (LTA) in Singapore (LTA 2014). Transit agencies that contract out bus service to third-party operators often use incentives. Punctuality incentives are not well aligned with HBS (Jansson and Pyddoke 2010) and lead to counterproductive actions by bus drivers to make the time window (Cats et al. 2012). Regularity

Literature Review 11   incentives are used by LTA in Singapore (LTA 2014), Transport for London, and Transantiago in Santiago, Chile (Transport for London 2012). Uniman (2009) and Jansson and Pyddoke (2010) discussed various considerations for the development of performance incentives. Operational Strategies A variety of operational strategies have been developed in the academic literature for use in HBS. Unless stated otherwise, the data mentioned in this subsection are based on simulation studies. Those that describe agencies’ current practices or the performance of operational strategies in the field are given special mention. Boarding limits restrict the number of passengers that can get on a bus at a stop. All buses have a capacity constraint and are unable to accept any more passengers once the capacity has been reached. Boarding limits triggered by the capacity constraint have been modeled by Zolfaghari et al. (2004). Imposing boarding limits before capacity has been reached has been proposed as a control strategy (Delgado et al. 2009) since part of the instability in high-frequency routes is tied to variable dwell times. The strategy discussed by Delgado et al. (2009) combines holding and boarding limits. Delgado et al. (2012) compared this combination to a strategy with holding alone and found that the main value of boarding limits is in a scenario of high passenger demand and short headways where the operator is able to reduce expected waiting time, cycle time, and cycle time variability by balancing the load factor across buses. Zhao et al. (2016) developed a strategy that uses only boarding limits to manage headways. Holding is a common control action in which buses wait for a specific amount of time before continuing on their route. Typically, holding time is applied at a stop after the bus has finished boarding and alighting passengers. Stops where holding time is applied are referred to as control points (Newell 1977). The choice of control points varies between transit agencies, ranging from every stop being a potential control point to only major stops and transit centers (Cats et al. 2011). Some stops may not be good control points because a holding bus would block traffic or prevent other buses from reaching a busy stop. Numerous holding strategies exist in the literature. Broadly speaking, they can be divided into two groups: • Optimization approaches, which emphasize passenger objectives and permit some variations in headway. • Control approaches, which emphasize regular headways. Optimization Approaches Early optimization approaches focused only on the average waiting time of passengers at stops (Barnett 1974; Turnquist 1981; Abkowitz and Tozzi 1986). More recent works include both out-of-vehicle waiting time and onboard delay caused by holding (Zhao et al. 2003; Xuan et al. 2011). This approach is computationally intensive and difficult to solve fast enough for real- time use. Heuristics (Zolfaghari et al. 2004; Cortés et al. 2010; Chen et al. 2013; Asgharzadeh and Shafahi 2017; Wood et al. 2018; Gkiotsalitis and Cats 2019), rolling horizon (Sánchez-Martínez et al. 2016; Manasra and Toledo 2019), and simplified assumptions (Delgado et al. 2009) can reduce the computational burden enough to permit real-time use. Other recent research has considered multiple bus lines, either sharing a corridor (Hernández et al. 2015) or interacting through transfers (Manasra and Toledo 2019). Optimization methods can differ from control approaches according to when they intervene. Wood et al. (2018) noted that their strategy tends to not hold near the end of the line because relatively few passengers are waiting downstream. In this scenario, the holding delay imposed on onboard passengers may outweigh the benefits of regular headways.

12 Intelligent Transportation Systems in Headway-Based Bus Service Control Approaches Control approaches may try to hold buses to maintain a target headway, or try to equalize headways in general without a specific objective. Adamski and Turnau (1998) and Rossetti and Turitto (1998) proposed a control approach to headway management. Daganzo (2009) wrote a seminal paper on the topic, which proved mathematically that holding buses based on the forward headway can stabilize a bus route and prevent bunching. One downside to a forward headway approach is that it remains possible for large gaps to open if a bus suffers a large distur- bance and the bus in front of it stays on target. This problem can be solved by considering the backward headway, the separation between a bus and the bus behind it. Bartholdi and Eisenstein (2012) developed a holding strategy using only the backward headway and showed that their approach allows the bus route to stabilize at a natural headway. The natural headway is determined by the cycle time under current condi- tions (inclusive of slack) divided by the number of buses in service. This property is especially valuable during major disruptions like inclement weather or special events because a fixed- target headway (e.g., 10 minutes) will become unachievable if the cycle time grows too long or if buses are taken out of service due to mechanical issues, driver shortages, or other issues. Argote-Cabanero et al. (2015) and Laskaris et al. (2019) developed holding strategies for a scenario with multiple interacting bus lines, which may arise when a route has branches or where multiple routes share a corridor. Liang et al. (2016), Andres and Nair (2017), and Zhang and Lo (2018) developed strategies that calculate holding time based on both the forward and backward headways. Some recent works have incorporated headway prediction (Fu and Yang 2002; Cortés et al. 2010; Xuan et al. 2011; Delgado et al. 2012; Berrebi et al. 2015; He 2015; Moreira-Matias et al. 2016; Andres and Nair 2017). Prediction can be particularly useful for strategies that use back- ward headways because the last measured headway is several stops behind the bus applying the strategy. Research has also been done to compare the performance of different holding strategies in simulation (Cats et al. 2011; Berrebi et al. 2018). Cats et al. (2011) simulated a bus route in Stockholm, Sweden, under schedule-based operation and headway-based operation with two possible holding strategies: enforcing a minimum headway, or attempting to equalize the for- ward and backward headways. Cats et al. found that both headway-based strategies reduced generalized cost for passengers, and that the even headway strategy reduced 90th percentile cycle time by 1.6% and improved punctuality at the driver relief point. Berrebi et al. (2018) simulated a bus route in Portland, Oregon, under schedule-based operation and under headway-based operation with several different holding strategies from the literature. The authors reported that schedule- based operation was unable to stabilize headways, while the holding strategies considered were generally effective. Berrebi et al. (2018) noted that the best strategy depends on context. The method of Daganzo and Pilachowski (2011) was found to perform best with shorter holding times spread across a greater number of control points, while the method of Berrebi et al. (2015) performed best when longer holding times were allowed at fewer control points. Berrebi et al. (2018) also discussed parameterization of the different strategies considered. Holding strategies may be limited by other constraints. Some strategies have a side effect of schedule sliding, where all buses are slowed down to maintain the headway objective (Daganzo 2009; Muñoz et al. 2013). This phenomenon may disrupt driver relief opportunities and vehicle scheduling (Cats 2014), potentially increasing overtime pay if drivers regularly work beyond the end of their scheduled shifts. A maximum holding time may be desirable to limit impacts on passengers and the system as a whole, and van Oort et al. (2010) suggested a parameter value of 60 seconds. Gkiotsalitis (2020) proposed a control approach to holding

Literature Review 13   for an electric bus fleet where the return time to the depot is treated as a hard constraint to maintain the vehicle charging schedule. Real-time holding has also been studied in rail operations (O’Dell and Wilson 1999; Ding and Chien 2001; Eberlein et al. 2001; Puong and Wilson 2008; van Oort et al. 2010). Wood et al. (2018) described holding procedures and performance measures used in practice at NYCT. Short turns are a strategy where some buses turn around before the end of the line and reenter service in the opposite direction. Short turns can be planned in the schedule to provide additional frequency and capacity to a portion of the route (Jordan and Turnquist 1979; Furth 1986), but are also used to recover from service disruptions (Li et al. 1993). Short turns are often restricted to specific locations where there is sufficient space both to transfer continuing passengers to another bus and to turn a bus around. Short turns are likely to be unpopular with passengers on the outer portion of the route, who face reduced service and forced transfers. Several studies have investigated how to generate short turns at a service planning level (Delle Site and Filippi 1998; Cortés et al. 2011; Verbas and Mahmassani 2013; Verbas et al. 2015; Gkiotsalitis et al. 2019). Zhang et al. (2017) proposed a real-time stop skipping and short-turning strategy. Related to short turns is the concept of layover time. Similar to slack in scheduled service, layover time at the end of a route is a buffer that is able to smooth out some disruptions and help buses begin each trip spaced at regular headways (Li et al. 1993; van Oort and van Nes 2009). Layover time can also be used after a short turn to ensure that the bus reenters service with a suitable forward headway (Zhang et al. 2017). Speed guidance is a strategy where bus drivers are asked to adjust their driving based on real- time headways (Chen et al. 2013). Drivers have discretion over several components of a bus trip, including cruising speed, door opening, and door closing, and can make minor adjustments without passengers noticing. This strategy may therefore have some public relations benefits compared to others, like short turns and stop skipping, which are more noticeable to passen- gers (Chandrasekar et al. 2002). Variations exist. In more qualitative versions, drivers may be reminded by dispatchers when their forward headway deviates from the target and be asked to speed up or slow down (Flynn et al. 2011). In more quantitative versions, drivers have an onboard device that gives them a specific cruising speed target. Daganzo and Pilachowski (2011) proposed a theoretical control strategy to determine cruising speed based on real-time forward and backward headways. Argote-Cabanero et al. (2015) and He (2015) combined holding and speed guidance. Sirmatel and Geroliminis (2018) developed a hybrid model predictive controller with a dual objective of headway regularity and speed. Varga et al. (2018) developed a receding horizon model predictive controller. He et al. (2019) developed a strategy for headway control on a corridor with dedicated bus lanes. Manasra and Toledo (2019) combined holding and speed guidance in an optimization system. Stage vehicles (bus insertion) are buses that are held in reserve and can be placed in service when conditions warrant (Cats 2014). Typical use cases are to break up a long headway, provide additional capacity, or replace a bus that has broken down (Morales et al. 2020). Cats (2014) and Petit et al. (2019) recommend positioning stage vehicles close to locations where service tends to deteriorate, such as a set of high-demand stops. Petit et al. (2018) developed an anti-bunching strategy where stage vehicles substitute for buses that are significantly early or late. Petit et al. (2018) noted that the Champaign-Urbana Mass Transit District currently uses a similar ad-hoc strategy: buses that are significantly late (in scheduled service) may be replaced by a stage vehicle at the dispatcher’s discretion. The late bus continues in drop-off-only mode until it is empty, at which point it becomes a stage vehicle and is available for insertion elsewhere. Petit et al. (2019) extended the previous research to a scenario where multiple bus routes share a pool of stage vehicles.

14 Intelligent Transportation Systems in Headway-Based Bus Service Zhang and Lo (2018) simulated the time for a route controlled by holding to stabilize after a bus is inserted or removed. Morales et al. (2020) developed a strategy for determining when to insert a bus into a long headway. The rule they used is threshold based, which makes the choice of parameter value strategic: too low and all stage vehicles are inserted early on; too high and they are never used. Testing the strategy with real data from Santiago, Chile, Morales et al. (2020) found an optimal threshold of nearly twice the average headway and an optimal insertion point slightly later than half the inserting headway because the stage vehicle starts empty. Stop skipping is a strategy that allows buses to make up time by skipping selected stops (Fu et al. 2003; Chen et al. 2013). The downside of this strategy is that it increases the waiting time of passengers waiting at skipped stops, who are forced to wait an additional headway. Variations exist. For instance, some agencies (e.g., LA Metro and RTC Transit of Southern Nevada) will not skip any stop requested by an onboard passenger (Kim et al. 2005; Flynn et al. 2011). Continuing to serve all requested stops helps with passenger satisfaction but may also reduce the effectiveness of the strategy depending on passenger demand patterns (Sun and Hickman 2005). Research has suggested allowing every other bus to skip stops (Fu et al. 2003) or at least maintaining some all-stop service (Zhang et al. 2017) as a way to limit the impact on waiting passengers. Several researchers have developed bus control strategies that combine holding and stop skipping (Cortés et al. 2010). Cortés et al. (2010) and Sáez et al. (2012) used hybrid predic- tive control to select control actions, while Liang et al. (2019) used a rule-based approach. Generally, buses that are ahead of their headway target hold, while those behind target can skip stops. Liang et al. (2019) noted the risk of overcorrection when using stop skipping to make up time. A related strategy, termed deadheading or segment skipping in the literature, is to skip several stops in a row. The term deadheading is typically used when there are no passengers on board. In this case, the purpose is to reposition the bus to a portion of the route with higher passenger demand, similar to short turning or inserting a stage vehicle (Eberlein et al. 1998). The term segment skipping is used when passengers are on board (Nesheli and Ceder 2014). In this case, the challenge is to communicate the control action to passengers in advance so that they may transfer to or continue to wait for an all-stop bus if their stop is to be skipped. Stop skipping strategies are also used in rail transit systems. Wood et al. (2018) described the strategies used in practice at NYCT. Combinations of Strategies Many of the sources listed previously implemented more than one operational strategy. Table 1 summarizes these combinations. Some strategies are complementary because they act in different directions. For example, holding can be used to slow down buses with a short forward headway, while stop skipping can be used to speed up those with a long forward headway. In other cases, one strategy is used to make small adjustments (stop skipping), and a second is used to make larger adjustments (short turns) if the first is unable to correct the problem. Other Strategies Exclusive lanes can support HBS by removing conflicts with traffic, which are an important source of travel time variability (Turnquist and Bowman 1980; Chang et al. 2004; Flynn et al. 2011). As one example, LA Metro’s Orange Line reports only a 20-second difference between peak and non-peak periods in average end-to-end running times (Flynn et al. 2011).

Literature Review 15   Queue jump lanes, typically provided at signalized intersections, allow buses to depart first when a signal turns green and can provide some of the benefits of exclusive lanes at a much lower cost (Chang et al. 2004). Operating procedures can help reduce the potential for bunching. Since small headway deviations can grow quickly, it is important for buses to start each trip at the correct head- way (Flynn et al. 2011). Along the route, any action that reduces travel time variability will also improve the stability of the route. For example, LA Metro emphasizes the importance of a dwell time of 15 to 20 seconds per stop in its guidance for Orange Line bus operators (Flynn et al. 2011). Finally, the discussion of a natural headway in Bartholdi and Eisenstein (2012) has implications for service planning. Because the natural headway is related to the cycle time and the number of buses in service, agencies may need to periodically reevaluate the cycle time and fleet allocation in order to maintain a specific headway. Passing is a strategy that can be used when buses are already bunched (Flynn et al. 2011). The second bus in the bunch, which typically has fewer passengers, is instructed to pass its leader at a stop or other safe location. Passing may modestly shorten the long headway in front of the bunch and reduce crowding on board, but the two buses will likely remain close together with- out additional control actions. Various stop improvements and policies, such as all-door boarding, can reduce dwell times and dwell time variability. All-door boarding reduces dwell times by allowing passengers to board using any door (El-Geneidy et al. 2017). In some cases, this policy is implemented with off-board fare collection, while in other cases rear doors have smartcard validation. Level boarding is a feature found in some bus rapid transit (BRT) systems. Level boarding is particularly helpful for wheelchair users and other mobility-impaired individuals, and can reduce or eliminate the need for ramps, lifts, and assistance from the bus driver (Chang et al. 2004). Raised curbs and low-floor buses offer some of the same benefits. Off-board fare collection means that tickets and smartcards are not purchased or validated on the bus, which speeds up the boarding process and reduces dwell times. This policy is usually paired with all-door boarding, and Diab and El-Geneidy (2012) suggested that the two policies are particularly valuable when articulated buses are used. Research Bo ar di ng L im its H ol di ng Sh or t T ur ns Sp ee d G ui da nc e St ag e V eh ic le s St op S ki pp in g Argote-Cabanero et al. (2015) Cats (2014) Cortés et al. (2010) Delgado et al. (2009) He (2015) Liang et al. (2019) Lizana et al. (2014) Manasra and Toledo (2019) Petit et al. (2018), Petit et al. (2019) Sáez et al. (2012) Wood et al. (2018) Zhang et al. (2017) Table 1. Combinations of operational strategies.

16 Intelligent Transportation Systems in Headway-Based Bus Service Intelligent Transportation Systems Technologies Introduction ITS is a broad category. This subsection introduces those ITS technologies that have existing or proposed applications to HBS. The following subsection, Uses, describes various field experiments and full deployments of these technologies that have been reported in the literature. An APC counts passengers boarding and alighting (Boyle 2008). APCs were initially deployed to automate the collection of ridership data for performance measurement and scheduling (Strathman 2002; Tétreault and El-Geneidy 2010). This continues to be their primary purpose, but some agencies now monitor APC data to track passenger load in real time (Yu et al. 2016). Identifying buses with a high passenger load is useful because they tend to run slower and fall behind their headway target. AVL allows transit agencies to track the position of buses in real time and to archive vehicle trajectories for later analysis (Parker 2008). The real-time position of buses is needed to calculate headways. Current systems typically use GPS to identify bus location (Chang et al. 2004). The update frequency varies by system, such as 1 minute for Metro Transit (in the Minneapolis– Saint Paul, Minnesota, area) (El-Geneidy et al. 2007) and 2 minutes for LA Metro circa 2011 (Flynn et al. 2011), since updated to 10 seconds. Bus location data are typically shared with dispatchers and passengers in real time and archived for later analysis and performance measurement. CAD is a method for real-time monitoring and control of transit vehicles. Capabilities vary between systems but typically include a centralized control room (where dispatchers are located), software for tracking bus locations and events (e.g., headway violations, incidents, and mechanical problems), and a means of communicating with vehicle operators (voice or text). Since tracking bus locations is a core function, CAD is generally implemented along- side AVL, and the two systems are often referred to jointly as CAD/AVL. Some CAD systems provide feeds of raw information to dispatchers, while others have additional functionality such as event-based alerts (e.g., headway deviation exceeding a certain threshold) and decision- support systems that recommend specific actions (e.g., holding and stop skipping). In addition to the features available to dispatchers, CAD systems also include a component in the driver cabin called a mobile data terminal (MDT). The MDT is used by drivers to communicate with dispatchers and displays relevant information to the driver, such as schedule time points, forward and backward headways, speed guidance, and so on. Figure 3 shows examples of CAD displays available to dispatchers and drivers. EFP is a category of fare payment that includes magnetic stripe media, smartcards, digital wallets, and smartphone apps. Most of these payment types can be validated automatically and therefore reduce boarding time relative to cash payment or flash passes (Chang et al. 2004) and reduce fraud (Diab and El-Geneidy 2012). EFP can also be used to track passenger boardings by location, either in real time or for later analysis (Pelletier et al. 2011). PIS include screens at bus stops and transit centers, voice- and text-based services, and online trip planners. These systems are used to communicate real-time or scheduled bus arrivals, messages about service disruptions, and other agency announcements (Chang et al. 2004). SS include text and voice messaging systems as well as cameras on board buses and in larger stations and transit centers (Chang et al. 2004). These technologies are typically paired with CAD so that dispatchers can monitor them remotely. SS is primarily used to deter crime and respond to incidents, but it has been suggested that dispatchers could use the real-time video feeds to monitor passenger loads at stations and on board buses (Adewumi and Allopi 2013).

Literature Review 17   Smartphone apps are increasingly common. Some agencies make their data available to third- party app developers, while others develop and maintain their own apps. The General Transit Feed Specification is one common data format originally developed by Google for transit direc- tions in Google Maps. Agency apps typically include features for EFP (mobile ticketing) and PIS (real-time bus information, bus schedules, service disruptions, and agency announcements). TSP allows buses to save running time by receiving special treatment at traffic signals. Recent research has proposed using TSP to support headway-based operation by speeding up buses that are behind their headway targets (Estrada et al. 2016; Anderson and Daganzo 2020). Estrada et al. (2016) proposed combining speed guidance with modifications to signal offsets and green extensions. Anderson and Daganzo (2020) proposed combining a control approach to holding with early greens and green extensions. Uses Transit agencies use archived ITS data in a variety of ways. CTA has used AVL data to analyze where and why bus bunching occurs (Hammerle et al. 2005) and has used CAD/AVL data to simulate scenarios for dispatcher training (Golani 2007). Metro Transit has used AVL (a) Sample dispatcher view (b) Sample MDT display Figure 3. Sample CAD displays (Source: Trapeze Group).

18 Intelligent Transportation Systems in Headway-Based Bus Service and APC data to analyze running time and headway deviation (El-Geneidy et  al. 2007). El-Geneidy et al. (2007) found that buses on one Metro Transit route typically served only 50% of scheduled stops, and the authors recommended stop consolidation to reduce both running time and running time variability. Kim et al. (2005) conducted an evaluation of the Metropolitan Area Express (MAX) BRT line in Las Vegas, Nevada, soon after service began in 2004. MAX had dedicated lanes in downtown and stations with level boarding, all-door boarding, and off-board fare collection. ITS included CAD/AVL, APC, TSP, and an optical guidance system for station docking. MAX was designed as a rapid overlay, with Route 113 providing local service on the same corridor. MAX was oper- ated by headway, with a target of 12 minutes from 5 a.m. to 7 p.m. and 15 minutes from 7 to 10 p.m. The main operational strategy was stop skipping. Capital costs included $2.1 million (2004 dollars) for ticket vending machines at 22 stations, $298,810 for CAD/AVL and APC systems, and $216,171 for TSP at 11 intersections. The TSP cost is not comprehensive because some traffic signals had been upgraded to TSP-compatible equipment prior to the MAX project. Performance data from 2004 showed that MAX had high headway reliability (the exact percentage and threshold were not specified), and experienced lower average dwell times and dwell time variability than Route 113 and the RTC system average. Kim et al. (2005) credited the dedicated lanes and station amenities for MAX’s high reliability. Total ridership on the corridor (MAX and Route 113) increased by 25% in the 5 months after MAX service began. McKone et al. (2009) reported that CTA was conducting four pilot projects: • Eliminating mid-route time points. • Installing countdown clocks at terminals to ensure buses depart at even headways. • Having control centers monitor headways and advise on-street supervisors. • Providing headway-based speed guidance to drivers using audio/visual cues. Cats (2014) described a field experiment in Stockholm, Sweden, where bus line 1 was converted from schedule to headway-based operation. An indicator based on the forward and backward headways was displayed on a screen in the driver cabin. Drivers were encouraged to adjust their speed as much as possible and hold at any stop if needed. The trial showed a change in headway distribution, a reduction of 11% to 26% in the coefficient of variation of headway, a decrease in bunching (defined as deviations of more than 50% from the target headway) of 13% to 24%, and a decrease in excess waiting time by 38%. Lizana et al. (2014) described a field experiment in Santiago, Chile. In 2012, Transantiago adopted a policy where bus-operating companies are penalized if bus headways at specific con- trol points exceed a threshold. This led to operator interest in headway control strategies. The field experiment implemented the strategy described by Delgado et al. (2012). The boarding limits described by Delgado et al. (2012) were considered impractical, so speed guidance was included instead. Experiments were conducted on the 210 service, operated by Subus Chile S.A., which had a 3- to 4-minute headway in the morning peak period. Buses could be held at 24 out of 135 stops for a maximum of 1 minute. Holding times were calculated using real-time AVL data and historical APC and smartcard data since these were not available in real time. Holding instructions were relayed by text message to field supervisors located at the control points. The penalties assessed to operators were 50% to 60% lower on the days of the experiment, and fare payments were up by 20%. Lizana et al. (2014) noted that fare evasion is common when buses are crowded because the only smartcard reader is located at the front door. The authors suggest that the reason for the increase in fare payment is that passenger loads were more balanced during the experiment, allowing more passengers to board through the front door. Argote-Cabanero et al. (2015) described a field experiment with D-Bus in San Sebastián, Spain, where headway-based operation was implemented on two routes, 5 and 25, that come together into a common corridor. The branches of Routes 5 and 25 are scheduled at 6- to 8-minute and

Literature Review 19   20-minute headways, respectively, which results in a 4- to 6-minute scheduled headway on the common corridor. The field trial installed consumer Android tablets in the driver cabins, which were used to display holding times and cruising speed guidance (with colored bars). The com- bined holding and speed guidance strategy was found to improve the reliability index by 35%. The authors also discuss driver compliance, which was generally good but was observed to be worse when holding times were longer and more variable. Wood et al. (2018) described a decision-support tool developed at NYCT for rail dispatching. The tool, named the Service Intervention Recommendation Engine (SIRE), estimates the passenger benefit of several potential holding and stop skipping control actions, and recom- mends actions that exceed a passenger benefit threshold. NYCT has implemented SIRE in a pilot phase and reports improvements to dispatcher productivity and passenger travel times. Qualitatively, SIRE helped dispatchers take more control actions, intervene earlier (while headway deviations were smaller), and consider different control actions. SIRE has recom- mended new skip stop patterns that dispatchers had not previously used and advised against holding near the end of the line when delays to onboard passengers were estimated to out- weigh the benefits to passengers waiting at downstream stations. A 2019 survey of 26 transit agencies using TSP found that none of the respondents were using headway adherence conditions for buses to request priority (Anderson et al. 2020). Many of the respondents did, however, indicate that TSP had a positive impact on headway adherence and headway reliability. LA Metro uses TSP at all signals on the Metro Orange Line, a BRT route operated by headway (Flynn et al. 2011). This route has a dedicated busway and uses unconditional TSP.

Next: Chapter 3 - Survey Results »
Intelligent Transportation Systems in Headway-Based Bus Service Get This Book
×
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

Intelligent transportation systems and, in particular, computer-aided dispatch and automatic vehicle location (CAD/AVL), have become quasi-universal in urban bus operations and support a variety of functions.

The TRB Transit Cooperative Research Program's TCRP Synthesis 155: Intelligent Transportation Systems in Headway-Based Bus Service synthesizes the current state of the practice of headway-based service operations and focuses on the proactive use of intelligent transportation systems technologies to optimize these services.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!