Introduction

A fundamental objective of introducing Metro Rapid service was to improve service quality, both from a customer perception and actual measurable performance standpoint. Effective service quality can be measured by vehicle headway maintenance or spacing (delivery performance), passenger waiting times (customer experience), and overcrowding of vehicles (from both average delivery performance and customer experience).

There is a significant interrelationship between these measures and with the scheduled service. For instance, the average customer wait times will be a function of both the scheduled intervals and the effective delivery of those intervals, as well as the vehicle loading (which will greatly affect bus bunching and pass-ups). In assessing the service, these relationships will be noted.

The field data was collected by MTA in August and September 2000 during a series of point checks at strategic locations. The use of timepoint data (as opposed to onboard ride check data) is appropriate as the aim is to ascertain a snapshot of the service at particular locations. The data was at various time intervals, as headways/frequencies are different during the day. The timepoints used were Whittier/Soto, Wilshire/Western, and Wilshire La Brea on the Line 720 Wilshire-Whittier corridor; and Ventura/Reseda for the Line 750 Ventura corridor.

It is important to reiterate that the data was from August 2000 prior to the MTA strike and was just 8-10 weeks after the start of a completely new operating strategy. More recent field checks have indicated that the Metro Rapid division line staff together with ongoing improvement in the operating schedules have continued to improve the quality of service and that the loads have continued to grow on Metro Rapid.

Headway Ratio.This ratio is a simple way to measure the variability of headways at a given timepoint, which measures the evenness of vehicle spacing. A headway ratio of 1.0 indicates that vehicles arrived at a stop perfectly spaced, whereas a headway ratio of 2.0 suggests that, on average, vehicles arrived in bunched pairs. In effect, the headway ratio is a measure of the extent of bunching of vehicles. Depending upon the frequency of service, bunching may have a negative effect on the effective level of service delivered to passengers. It results in an actual level of service below that scheduled and may cause overcrowding and unacceptably long passenger waits.

The results on Ventura Boulevard indicate low levels of bunching at Reseda Boulevard, and this generally effective service delivery. The only bunching problems appear to be on the local services, and more so on the westbound local services. The spacing of the Metro Rapid service appears to be very consistent, indicating good headway maintenance success.

On Wilshire Boulevard, the results are mixed. During the midday periods in both directions, headway intervals are adequate but need improvement on both Metro Rapid and local. The average midday passenger waits are consistently less on the Metro Rapid despite operation of the same headway on both Metro Rapid and local. During the peak periods, when the Metro Rapid is operating every 2-to-2½ minutes, many times vehicles are arriving almost in pairs. The problem time and direction for Metro Rapid is westbound during the PM peak where average waits are around 8-minutes with average arrivals in more than pairs. The local service is also experiencing similar problems, but with a shorter route the problems are less acute. While the Metro Rapid service performance looks to be on-par or slightly better during regular demand periods, there is a need to closely monitor spacing during the peak periods with the objective of keeping average wait times below 5-minutes and the measured average load and passenger average load close to one and other.

On Whittier Boulevard, the Metro Rapid and local services are performing similarly with both services delivering similar headway ratios. The exception is eastbound Metro Rapid where interval performance is not satisfactory during the midday with average waits of nearly 8 minutes (ideally they should be 5 minutes) and during the PM peak where almost 2½ buses are arriving together. At Soto Street, the Metro Rapid buses are already some 75 minutes into the eastbound trip; however, the line staff will need to determine why service is bunching significantly after departing the Western Station eastbound with low bunching.

There are two major impacts of higher headway ratios (or higher bunching levels). The first is significantly increased average passenger waiting times over scheduled levels. The second is loading variability, causing overcrowding and poor utilization of available capacity. These impacts are discussed further below.

Average Wait Times. For high-frequency transit service, average wait time would normally be half the scheduled headway, assuming passengers arrive at stops in a random manner (i.e., random walk theory). For example, on a 10-minute frequency, a passenger arriving randomly at a stop could be expected to wait, on average, for five minutes.

However, where service becomes less reliable (due to bunching), average wait times increase. This can be measured as expected average wait time, assuming random arrivals at stops by passengers. This performance measure is, in effect, one of the most powerful and descriptive measures of how effectively the service is being delivered and a good indicator of customer out-of-vehicle wait times. This is because this simply measures how long passengers have to wait for vehicles, as compared to what the schedule suggests. Average wait time is closely tied to the headway ratio – where headway ratios increase, so too will passengers’ average wait times.

Another way to look at average wait time is to use it to calculate the affective level of service being delivered. Simply multiply the average wait time by two, and you have the true level of service that a passenger waiting at that stop would have seen. This can then be compared to the scheduled level of service to measure how effectively the service is being delivered.

On Ventura Boulevard, the average waiting times are in line with the headway ratio. They tend not to be significantly greater than scheduled average waiting times. The only exception is on the local service, westbound in the AM Peak, where average wait time is 4.3 minutes, suggesting the actual level of service delivered is 8.6 minutes, which is well below the scheduled level of service of 5-minutes.

On Wilshire Boulevard, the actual level of service delivered varies (sometimes significantly) from the scheduled level of service. An example is the local service on Wilshire at La Brea, eastbound in the PM Peak. The scheduled level of service is around 7 minutes. Therefore the average wait for a passenger randomly arriving at a stop should be 3.5 minutes. But instead, the average wait was over 5 minutes. The implication is that while the resources expended equal a 7-minute service, from the passengers' perspective, only an 11-12 minute service is being delivered.

Average wait times on Wilshire Boulevard on the Metro Rapid are also, at times, well in excess of scheduled levels. As the headway ratio suggests, the main issues appear to be PM Peak and early evening westbound, where average wait times are over eight minutes, indicating an affective service level of over 16 minutes, again well below scheduled frequencies, and midday eastbound where average waits are around 7 minutes (the scheduled wait is 5-minutes). Overall, however, it appears that the Metro Rapid service is being delivered on-par or slightly better than the local service (i.e., lower headway ratios and lower deviation from the scheduled average wait time), especially when the very high peak direction frequencies are considered.

On Whittier Boulevard, average wait times are much higher than scheduled eastbound, in the off-peak and PM Peak. During the off-peak on the Metro Rapid, the average wait time is nearly eight minutes, suggesting an effective level of service of 15 minutes, while the scheduled level of service is 10 minutes. In the PM Peak (again eastbound), average wait times are 4.6 minutes, indicating an effective actual service level of over nine minutes, which is nearly three times the scheduled service level.

Patron Perceived and Measured Average Loads. This is a measure of the variability of load distribution. Usually, where bunching occurs, some vehicles will be heavily loaded, while some will be relatively empty (particularly close-trailing vehicles). This measure weights the loads according to the actual average customer experience.

In an extreme example, where two buses operate, the first with 60 passengers, and the second with none. The average load is 30, suggesting no capacity issues. However, all passengers saw a load of 60, and therefore the passenger perception is that all buses are overcrowded. In short, this measure considers how many passengers actually experience vehicle crowding. This is also a good measure of loading variability. Loading variability is a measure of service effectiveness, as high loading variability usually means that additional resources are required to provide the necessary capacity. Patron average load experience needs to be measured against the measured average load to measure loading variability.

On Ventura Boulevard there is some sporadic loading variability. However, neither the true average load or patron-perceived average load are close to capacity levels, indicating, if anything, excess capacity on both the local and Metro Rapid services.

On Wilshire Boulevard there are examples of sufficient capacity, but variability of loading causing overcrowding problems. A good example is on the Wilshire Metro Rapid at La Brea, westbound in the PM peak. The average load is 39, indicating no real capacity issues. However, the patron-perceived average load is nearly 52. Therefore, while no average capacity problems exist, there would be a perception of significant overcrowding problems from the passengers themselves. This indicates that there is high loading variability during this time period and during the early evening in the same direction, with some very-heavily loaded buses, and some half-empty buses (almost present on the local service at the same time and direction). The likely outcome would be additional resources, yet there is clearly enough capacity on average.

On Whittier Boulevard, the most significant incidence of loading variability is eastbound in the PM Peak. However, the average load is 27, and the patron-perceived average load is 32 with neither a problem from a customer perception standpoint. The rest of the day, on both the local and Metro Rapid services, there do not appear to be either capacity or overcrowding issues.

Loading Summary: In summary, it appears that there are capacity issues on the Wilshire Metro Rapid westbound throughout the day with significant problems in the AM Peak and midday periods. Eastbound capacity shortfalls are only during the PM Peak and early evening periods. The eastbound loads are evenly distributed between locals and Metro Rapid services at Western, but the Metro Rapid loads are higher at La Brea. Westbound, the Metro Rapid loads are consistently much higher than the local services.

On Whittier Boulevard, the only capacity issue is westbound in the AM Peak, where the average load is 46 passengers. As with the Wilshire corridor, locals and Metro Rapids are similarly loaded eastbound, but the Metro Rapids are averaging somewhat higher loads westbound.

On Ventura Boulevard, the Metro Rapid loads are higher than the locals, except during the afternoons westbound. Overall average loads suggest no capacity issues.

Recommendations:

1. Given the frequency levels and loads on Metro Rapid, continue with plans to introduce higher capacity vehicles on the corridor.
2. Continue to adjust scheduled frequencies and running times to reflect current conditions based on more recent point checks and TOS input.
3. Continue to campaign the bus bunching problems through the deployment of additional capacity where needed, Metro Rapid point checks and ride checks to identify delay issues, strengthen the support of the BOCC to the line TOS in early notice of bunching, and introduce the bunching assistance routines in the LADOT bus signal priority system in a test mode to ascertain the impact of reducing bus bunching on operating speed and resource requirements.
4. As detailed stop level data becomes available, consider the possibility of a short line east of downtown Los Angeles at or before Soto Street. This will add complexity to a simple line and likely strand significant numbers of patrons at the short line terminal. Thus, it should be approached cautiously and have initial and on-going customer notification involved on a real-time basis.
5. The upcoming introduction of the "next-bus" displays will provide early notice to customers and possibly effect customer choice of local versus Metro Rapid. Customer reaction should be monitored for impact on service schedules and delivered performance.

• 1 The unique Metro Rapid operating protocols involved the first time use of traffic signal priority for buses, elimination of timepoints and use of a headway interval spacing to manage vehicles, and separate station stops from local buses.
• 2 Even spacing is very important under most service frequency conditions. However, under extremely frequent service conditions (headways well below 5 minutes), the need to delivery evenly spaced service is unnecessary from a customer wait experience standpoint. The more important objectives under these conditions are to avoid service gaps beyond 4-5 minutes and to provide adequate capacity so that there are no pass-ups.
• 3 Note that this stop is west of the traffic congestion around the I-405 San Diego Freeway interchange - eastbound services will have not yet encountered this point.
• 4 These conditions were present even with the lowest measured average loads of the day for Metro Rapid, but worse from a customer standpoint due to very uneven loading.
• 5 The issue is whether to improve out-of-vehicle wait times (bus bunching) at the expense of in-vehicle travel times (operating speed). This is not an either/or situation; the conventional wisdom is that once the average waits fall well under 5-minutes there is little customer-perceived benefit in further reductions. Thus, bus bunching actions should aim at keeping average waits well below 5-minutes, but recognize that average waits of under 3-minutes have little value in attracting additional customers or retaining current riders.
• 6 MTA Headquarters Operations and Scheduling introduced a weekend shortline at the 6th/Los Angeles station that Division 7 TOS report strand up to 15 customers per trip on Saturdays and Sundays.