The above analyzes and findings give us some implications for practice, application and methodology as shown in the following discussions.
6.1 Implications for practice and applications
I have observed an amount of walking energy expenditure and plotted them as a histogram to estimate a specific distribution for the WEEs. The above surveys could enrich the traditional walking distance with more content and a wider application. The surveyed WEEs show a right-skewed distribution, suggesting that when people walk when commuting they do so with a focus on saving effort. This characteristic is actually quite easily determined, since walking distances have been demonstrated to be right-skewed distributions and WEE is associated with walking time. However, WEE cannot be estimated if we do not model the effects of the pedestrian environment. As a result, effort-saving cannot be demonstrated without evidence of the right-skewed distribution of WEE.
It should be worthwhile to explore applying iso-energy in planning. For example, in a community, a local walking path system offers a pedestrian a set of alternative routes within an O-D pair. Each alternative route can be plotted as a point in Figure 4, which presents a bundle of distances and street amenities. We can then draw the energy curve by linking each point that consumes the same WEE. After repeating this process for all alternative routes per the levels of WEE, a set of curves similar to Figure 4 can be created. Let us assume the outer energy curve gives the pedestrian a lower utility. There is a constraint line BC for a pedestrian commuting from home to the transit station. To pursue a less energy-consuming route, the pedestrian would choose point C instead of point B. However, this idea needs more future study, as pedestrians may get used to their daily walk on specific routes regardless whether that route maximizes utility.
The energy-based approach supports considering more individual characteristics for setting walking distance. A new trend in planning theory is to change the current top-down planning perspective to a bottom-up one (Batty, 2001). Our proposed WEE approach improved the traditional aggregate walking distance by taking more individual characteristics into consideration (such as speed, body weight, gender, etc.) and by classifying the WEEs into levels based on the surveyed WEE data. In addition, energy reflects physical effort, and using WEE as a walking ability is suitable for enhancing the rational duration during agent-based simulation.
In addition to physical WTW, the psychological WTW was further estimated for a higher
63
level of SA that can aid street improvement evaluation. The above analyses and findings should interest planners for several reasons. Sometimes a project site is very small or narrow, so planners may have few alternatives because it is only permissible to improve or rebuild one or two street facilities. Even so, they still hope the chosen alternative will prove effective. Referring to Table 12, planners can choose a set of SA attributes as targets to enhance WTW. There is scope for our results to be further developed as a criterion to evaluate the degree of achieving WTW enhancement.
These findings can improve the traditional land use theory. First, the proposed WEE approach could enrich the current planning theory in policy implications on land use patterns. “Concentric” is a general land use pattern for neighborhood or community planning where the levels of distance are set as radiuses of circles surrounding a service center (Banai, 1998). The bid-rent theory, inspired by the von Thünen model, supports the concentric pattern with a space economics analysis (Alonso, 1960). However, the circular area drawn using a constant distance does not take into account the effects of PEQ. For example, walking distance should be shorter in blocks without a walkway where pedestrian-vehicle conflict frequently happens. According to this research, a service area should be larger in an area with better PEQ but smaller in one with worse PEQ.
Consequently, the concentric land use pattern should be modified into a contour pattern, as shown in Figure 17.
Figure 17 The service area with cosideration of WEE
捷運站
服務範圍
Transit Station
Service Area
64
Not only the shape, service area can also be broadened in the better environment. In our estimation, an improved street would increase pedestrians’ utility to the point where they can spend more time to maintain the same level of utility. The increase in walking time implies that people would walk slowly or extend his/her distance. For example, street retailing is improved to provide pedestrians with a variety of stores (RF→RV), and this yields a 1.715(SP)/2.61(RP) min increment in WTW (t). When walking speed is set as 1.44m/s (), the change in distance is equivalent to 148(SP)/226(RP) m (st). For a TOD planning, the result implies that the street improvement around a transit station would broaden its service area, as shown in Figure 18.
Figure 18 Street improvement around a transit station would broaden the walking accessible area.
6.2 Implications for methodology
This study applied two utility function types to determine the difference between the RP and SP models. The results show little discrepancy in the significant variables in equation 13, in contrast to equation 14, which considers individual characteristics, meaning that the participants do not take individual characteristics into account in the process of SP. Based on this finding, it should be noted that if a street improvement project is planned for a specific group, such as the elderly, planners should not be solely reliant on SP data, but should collect both SP and RP data to examine users’ behavioral intentions.
As shown earlier, the result of the test conducted by Swait and Louviere (1993) rejects the estimation of an increase of WTW using a joint model. This result may not only occur in
65
this study. Previous studies have explored the possibilities of combining various procedures and conditions (Huang, Haab, and Whitehead, 1997; Swait and Louviere, 1993), or sought another combing technique (Earnhart, 2002). However, reasons for and implications of this result are worth to be explored, which provide future research the direction to improve the preference equality or to explain the discrepancy between the SP and RP models.
In agent-based experiment, I found that there is a limitation when I analyzed the flow rate greater than 23 p/min/ft. As shown in Table 18, when the number of elderly pedestrians exceeds 5%, the mean speed is 1.44m/s and when the flow rate is less than 23 p/min/ft, the simulation can well fit Fruin’s standards. However, it failed when the flow rate was greater than 23 p/min/ft; even when the elderly increased to 20%. It is likely that the nature of the extreme high density flow is best represented with the models of Helbing et al. (Helbing et al., 1995; Helbing wt al., 2000). In fact, people in highly congested flows may frequently end up touching each other. Although pushing might not happen, more detailed motion modeling than ours will be required to represent this situation. I designed direction-choosing and stopping as a bypass movement of agents, but readers will find that Helbing et al. used several forces, e.g., sliding friction force, to describe jammed pedestrians. Thus, extreme high density flows are out of the scope of this study.
Because the corridor is assumed only to serve as a passing corridor, the psychological effect of corridor design is not taken into account. Planting boxes and benches were the elements used to separate the counter flow or the difference in mobility between pedestrians, but in the real world they act both as a barrier and an amenity. Planting boxes provide greenery and raises the aesthetics of the corridor environment. Some pedestrians may avoid them by some distance but some may be attracted to them and slow down to admire them. Benches are usually placed on the side of a walkway to provide a means to take a rest. It also attracts pedestrians to remain in the area longer. However, this research assumed that the corridor serves mainly as a passing corridor, and that pedestrians tend not to stay for the facilities and amenities. This type of corridor exists mainly in bus, train and rapid transit terminals, and is not suitable for a corridor with different uses.
For uneven flows, the description of the LOS should be redefined to fit the real feel and need of the pedestrians. For example, when the flow rate is 10~15 p/min/ft, all the observed LOS were at level E. According to the original definitions of Fruin, this means that virtually all pedestrians would have their normal walking speeds restricted. However, the LOS of the younger pedestrians was higher, up to D (1.21~1.143 m/s). I did not use Fruin’s original
66
description to describe this situation, because the younger pedestrians may not be of the opinion that there is insufficient area available to maintain their buffer zone and desired speed. They may have a mindset that the only thing they need is a frequent change of direction or adjustment of gait if they wish to bypass slower-moving pedestrians. Thus it is suggested that the description of LOS be redefined to reflect the characteristics of uneven flow.
67