Bus operating speed

Using the values of Table 4.5, it is possible to estimate the potential time and cost savings due to upgrading the fare collection system and/or boarding policy. However, before attempting this exercise a couple of considerations need to be made. First, the full realisation of potential time savings depends on how timetables can be adjusted after changing the fare collection system; for instance, the need to synchronise transfers between two or more lines (Ceder et al., 2001; Fleurent et al., 2004) or to publish timetables at bus stops with departure times rounded to entire minutes would set constraints for the translation of potential savings into actual savings. In what follows, potential benefits from having a quicker fare payment method are estimated, with no concern about the adaptation of timetables; in this respect the approach is directly applicable to bus services that are not based on timetables for passengers (e.g., a high frequency route with one bus every five minutes or less).

Second, the composition of the patronage is relevant because if there are passengers exempted from paying a fare (e.g., school students, senior pensioners), the effectiveness of upgrading the fare payment technology as a tool to decrease travel times is reduced. On the other hand there might be differences within the group that pays a fare; for example senior passengers might be slower to board and alight buses than younger

89 passengers, although the time differences with alternative fare payment methods exist regardless of the age of the passengers (Tirachini, 2011)31.

In order to apply the estimated bus travel time model (Equation 4.12), an estimation of the actual number of stops that a bus makes to serve passengers is required. In formal urban public transport systems that face high demand, such as BRT services, it is common that buses have to stop at every designated station along the line, regardless of whether or not there are passengers that actually want to board or alight. Nevertheless, in the case of low-demand bus services, it is usual that buses stop only when they are required to by passengers on board that need to get off at the next stop, or by passengers who signal to the driver while waiting on a bus stop. Intuitively, the number of times that a bus actually stops along a route depends on ridership, as with a low total demand per bus, it is less likely that buses are required to stop at every scheduled bus stop. The relation between the actual number of stops per bus-kilometre and demand per bus- kilometre found in the travel time surveys is shown in Figure 4.5. It is evident that a high proportion of the variation in the actual number of stops is explained by the average bus demand.

Figure 4.5: Actual number of bus stops as a function of passengers per bus

31 _{The figures in Table 4.5 are average boarding and alighting times for all age groups.}

0.0 0.5 1.0 1.5 2.0 2.5 3.0 0 2 4 6 8 10 St ops [s to ps /bus -km ] Demand [pax/bus-km]

90 Numerically, we find that the curve that better fits the scatter plot depicted in Figure 4.5.is the power function shown in equation (4.13), where Skm is the actual number of stops per kilometre and Dkm is the average number of passengers per bus- kilometre. The power 0.528 implies that the actual number of stops per kilometre roughly varies with the square root of demand.

0.528

0.763

km km

S = D (R2=0.736) (4.13)

The operating (or commercial) speed is the average speed along a route including both running time and stops of any sort. We simulate the circulation of buses with two and four doors, assuming that travel time model (4.12) is valid for the two types of buses despite their difference in size, on a route of 16 kilometres of length, with 11 traffic lights and 8 roundabouts (these figures are the average values of the variables in the sample) during the morning peak (8-9 AM). Average demand varies between 1 and 8 pax/bus-km, which in turn determines the actual number of bus stops per kilometre, as given by Equation (4.12). The APST per payment system are obtained from Table 4.5. Figure 4.6 shows how operating speed decreases with demand, and the loss of speed is stronger the more inefficient the fare payment system is. For example, on 2-door buses speed drops from 24.8 to 14.3 km/h for cash payment, while in the same demand range the drop is 26.4 to 19.7 km/h for contactless card with boarding at the front door only. This is a quantification of an expected result, that the benefits of having an efficient fare payment system, with prepaid cards or off-board payment, are greater the larger demand is.

91 (a) 2-door buses

(b) 4-door buses

Figure 4.6: Bus operating speed as a function of demand and fare payment and boarding policy

When boarding is allowed at the front door only (cases T2B1 and T4B1), in both plots there is a noticeable gain in speed when upgrading the fare collection method from cash to magnetic strip, and from magnetic strip to contactless card; nevertheless, the

14 16 18 20 22 24 26 28 1 2 3 4 5 6 7 8 Spe ed [ km /h] Demand [pax/bus-km] Cash T2B1 Magnetic strip T2B1 Contactless card T2B1 Magnetic strip T2B2 Contactless card T2B2 Off-board T2B2 14 16 18 20 22 24 26 28 1 2 3 4 5 6 7 8 Spe ed [ km /h] Demand [pax/bus-km] Cash T4B1 Magnetic strip T4B1 Contactless card T4B1 Magnetic strip T4B4 Contactless card T4B4 Off-board T4B4

92 technology effect on increasing speed (reducing travel time) is weaker when boarding is allowed at all doors (cases T2B2 and T4B4). This is particularly evident in buses with four doors, as the vertical difference between the three T4B4 curves is almost marginal in Figure 4.6b. In other words, upgrading the fare payment technology has a major impact on performance when boarding is allowed at the front door only, but this technology effect diminishes when boarding is allowed at all doors, especially on bigger buses. Note that the TnBn boarding policy is superior to TnB1 in all cases, even if the fare is paid with magnetic strip (with contact) in the former case and with the faster contactless card in the latter case. This indicates that if a bus service is provided with on-board magnetic strip payment and boarding is allowed at the front door only, in order to save travel time it is more effective to allow boarding at the back doors (installing card reading devices) than to upgrade the technology of payment to contactless card keeping the one-door boarding policy. In other words, the door effect can be more powerful than the technology effect.