CHAPTER FIVE PRESENTATION AND ANALYSIS OF ...paginas.fe.up.pt/~pala/MSc_Dissertation/Ch5_Pres...34 CHAPTER FIVE PRESENTATION AND ANALYSIS OF RESULTS 1 5.1. Do-minimum forecast Table

34

CHAPTER FIVE

PRESENTATION AND ANALYSIS OF RESULTS1

5.1. Do-minimum forecast

Table 5.1 shows the variation in aggregate transport indicators between the base year and year

15 as predicted by the model.

The results show only a modest

increase in the total number of

trips (+0.7%), in comparison with

the growth in population and

employment (4.3 and 12.4%). Car

trips, however, grow by nearly the

same proportion as population

(+3.8%) whereas bus trips suffer a

small decline (-0.8%) and walking

trips fall noticeably (-4.6%). These

changes in trips reflect on the

modal split. The car reinforces its

dominance with a 53.7% modal

share, while bus overtakes walking

as the second most important

mode.

While the number of trips grows

only slightly, distance travelled

shows a much more marked

increase (+4.7%) as the result of

an increase in trip length for both car and bus trips (4% and 1.7%). At the same time, average

travel speed falls for all modes (-1.7% and -4.3% for car and bus, respectively). These results

help explain why the growth in the total number of trips has not been greater, bearing in mind

the constant travel time budget constraint.

1 This chapter is structured in accordance with the numbered tests described in section 4.6

Table 5.1: Change in aggregate indicators, do-minimum LU0-T0 LU0-T0 % Year 1 Year 15 change Trips total 1,543,097 1,554,204 0.72%Car 804,279 835,147 3.84%Bus 366,298 363,566 -0.75%walk 372,519 355,491 -4.57%modal shares Car 52.1% 53.7% 3.10%Bus 23.7% 23.4% -1.46%walk 24.1% 22.9% -5.25%passenger-kms total 10,749,097 11,252,844 4.69%Car 7,223,239 7,631,019 5.65%Bus 3,097,461 3,213,010 3.73%walk 428,397 408,815 -4.57%CO2 emissions total 495,155 524,914 6.01%Trip length (kms) total 7.0 7.2 3.94%Car 9.0 9.1 1.74%bus 8.3 8.8 6.29%Average speed (kph) average 20.9 20.8 -0.45%Car 25.3 24.9 -1.66%Bus 11.5 11.0 -4.26%Demographics population 727700 758798 4.27%employment 352100 395698 12.38%

35

These transport indicators reflect the changes in land-use structure prompted by the Leeds

UDP. As population is allowed to move to more distant suburbs, trip lengths increase, the car

becomes more attractive, and car ownership and use also rise. With a greater number of cars

on the road, congestion is also likely to increase and speeds fall. Moreover, employment is

kept relatively concentrated in the city centre and that reinforces the increase in congestion.

5.2. Impact of land-use scenarios on travel patterns and transport system performance

Tables 5.2 and 5.3 show the change in aggregate transport indicators for the alternative land

use scenarios relative to the do-minimum. According to these results, travel patterns and

overall transport system performance are considerably affected by changes in the urban

structure, even when these are relatively small in magnitude (population growth = 4.2%,

employment growth = 12.3%). Table 5.4 shows the results of the cost-benefit analysis.

Table 5.2: Percentage change in aggregate transport indicators for the alternative land use scenarios, relative to do-minimum (year 15 forecasts) LU1 LU2 LU3 LU4 T0 T0 T0 T0 Trips total 1.6% -0.3% 1.2% -12.1% Car -4.0% 0.3% -3.1% -10.2% Bus 4.3% -1.7% 1.7% -15.0% Walk 11.9% -0.4% 11.0% -13.9% modal shares Car -5.5% 0.6% -4.3% 2.3% Bus 2.7% -1.4% 0.4% -3.3% Walk 10.1% -0.1% 9.6% -1.9% passenger-kms total -4.5% 0.3% -4.5% -13.6% Car -6.2% 0.8% -5.6% -12.6% Bus -2.6% -0.7% -3.7% -16.1% Walk 11.9% -0.4% 11.0% -13.9% CO2 emissions total -5.8% 0.6% -5.1% -11.4% Car -5.8% 0.6% -5.1% -11.4% Trip length (kms) total -6.0% 0.7% -5.6% -1.7% Car -2.3% 0.5% -2.6% -2.7% bus -6.6% 0.9% -5.3% -1.2% Average speed (kph) Average -2.8% 0.4% -2.9% -17.8% Car -1.6% -0.1% -2.4% -19.2% Bus 3.0% 6.2% 2.9% -9.7%

36

5.2.1. Centralisation (LU1). Scenario LU1 produces the most significant and, arguably, the

most positive changes of all land-use scenarios. Car modal share is reduced by 5.5%, while

Table 5.3: Percentage change in trips for the alternative land use scenarios, relative to do-minimum (year 15 forecasts) car-peak car-opeak bus-peak Bus-opeak ped-peak ped-opeak L1_T0 -4% -4% -3% 8% 17% 10% L2_T0 0% 0% -2% -2% 3% -1% L3_T0 -3% -3% -4% 4% 15% 9% L4_T0 -4% -13% -5% -20% -1% -18%

Table 5.4: Summary of the cost-benefit analysis for the alternative l.u. scenarios Money savings Time savings LU benefits2 Users PT ops Govt External H'way PT Walk H'holds Ops OF L1_T0 21 50 -161 91 160 -85 -159 240 124 415 L2_T0 6 -30 18 5 -25 49 15 735 -385 379 L3_T0 16 -29 -153 93 -34 -140 -127 -664 -882 -1853L4_T0 40 -517 -459 264 -1591 -939 -198 675 -775 -3338

bus modal share is increased by a less impressive 2.7%. The most remarkable change,

however, is the 10% increase in the share of walking trips. At a more fundamental level, this

scenario produces a decrease in average trip length for all modes. For car-based trips, average

journey length falls by 2.3%. Combined with an overall 4% reduction in car trips, this means

that car passenger-kms are 6.2% below the do-minimum forecast. Public transport trips also

become shorter. Despite an increase in bus trips of 4%, the 6.6% reduction in trip length

means that total bus passenger-kms actually fall. Given the overall reduction in distance

travelled it is not surprising that CO2 emissions fall by 6%. The only negative impact of this

scenario vis-à-vis the do-minimum is the slight reduction in car speed (-1.6%). In contrast,

bus speeds rise by 3%.

The reduction in trip length (-2.3% for cars and -6.6% for buses) shows that bus is relatively

more attractive in relation to car in the innermost suburbs. Table 5.2 also shows that, whereas

2 The reason why land use benefits for households and landlords do not cancel out is that they are calculated in

a different way. Household benefits are calculated using the rule-of-a-half, whereas landlord benefits are

calculated as the total change in revenue. This explains why both groups may have positive benefits.

Nonetheless, it remains unclear as to what these benefits stand for and whether the way in which the model

calculates them remains valid for such large changes in land-use structure.

37

-100000

-80000

-60000

-40000

-20000

0

20000

40000

L1_T0 L2_T0 L3_T0 L4_T0

car-peak

car-opeak

bus-peak

bus-opeak

ped-peak

ped-opeak

car speed falls, bus speed actually rises. Car speeds fall because more car trips are now made

in the more congested central areas. Bus speeds rise, on average, because there is a high

proportion of segregated bus ways in the central zones and buses are therefore less affected by

increased central area congestion.

At the same time, many bus routes

in the outer suburbs can now

operate under less congested

conditions.

Figure 5.1, which shows the change

in number of trips by mode and

time of day relative to the do-

minimum, helps to illustrate how

these results come about. The chart

shows that there has been a reduction

in car and bus-based peak-time trips (-4% and -3% trips, respectively), while peak-time

pedestrian trips increased by approximately the same absolute amount. One can hypothesize

that, as more jobs and residents move into the central area, walking becomes a viable

alternative for the journey to work for an increasing number of people. At the same time, the

central area becomes increasingly congested because of the higher concentration of activities

within it. As a result, vehicle speeds drop and motorized modes become less attractive.

While the trip matrices included in appendix 3 show that the number of peak car trips destined

for the two innermost zones is above the do-minimum level, the relative increase is below the

corresponding relative increase in population and employment within the central area. At the

same time, the number of car trips within the two outermost rings is well below the do-

minimum level. Therefore, it can be argued that the overall decrease in peak-time car trips is

due to population and jobs moving to the central area, where alternative modes are able to

attract a higher proportion of trips.

With regard to the decrease in peak-time bus trips, the trip matrices show that despite an

increase in bus trips within rings 1 and 2, the number of bus trips originating in other areas

falls considerably. As explained with respect to car trips, as activities become more

Figure 5.1: Change in trips for the alternative l.u. scenarios, relative to do-minimum (year 15)

38

concentrated in the central area, walking becomes the mode of choice for a higher proportion

of trips and the overall number of bus trips falls.

Figure 5.1 shows an increase in the overall number of off-peak trips despite a decrease in off-

peak car-based trips. As explained in the previous paragraphs, the decrease in car trips and

increase in pedestrian trips is due to the relocation of activities to areas where walking is

relatively more attractive. The number of bus trips actually increases in the off-peak, which is

contrary to what occurs in the peak. Since the increase in bus trips occurs mostly in the two

innermost rings, it is likely that these trips are being diverted from walking. One reason for

this is that bus fares are cheaper in the off-peak. Another reason is that lower congestion in

the off-peak makes buses more attractive than they are in the peak.

According to table 5.4, both car users and public transport operators benefit from money

savings in comparison with the do-minimum (+21 and +50 thousand euros, respectively). This

can be interpreted as the result of average lower vehicle running costs. It was mentioned in

the previous paragraphs that peak trips have increased for all modes within the two innermost

rings, while in the outer zones the opposite has happened. Running costs would therefore be

expected to increase, unless the effects of reduced congestion in the outer zones outweighed

the increased congestion in the central area. Another possible explanation is that the overall

reduction in peak-time motorized trips in favour of walking trips has led to an aggregate

reduction in vehicle running costs even though running costs per vehicle may have risen.

The increase in money savings for public transport operators is also likely to be a result of the

increase in public transport trips and therefore a boost in revenues.

Despite the aggregate reduction in costs for transport users and operators, public authorities

receive fewer benefits. This is probably due to increased maintenance expenditure of central

area roads because of higher wear and tear, and possibly a drop in fuel consumption because

of fewer motorized trips. On the other hand, parking revenues have increased because there

are now more vehicles travelling to the two central zones. Another result in this scenario is

the decrease in external costs consistent with the reduction in CO2 emissions shown in table

5.2.

39

With regard to time savings, the results show an increase for car users and a reduction for

public transport users and pedestrians. This indicates that the reduction in congestion in the

outer zones (where the majority of trips are by car) affects more drivers than the increase in

congestion in the inner zones. On the contrary, walking and bus trips have increased the most

within the central zones where congestion has increased. Furthermore, the reduction in

pedestrian time savings shows the increased interaction of pedestrians with motorized modes.

5.2.2. Dispersal (LU2). The dispersal scenario produces very little change relative to the do-

minimum in terms of transport performance. One possible reason for this is that the Leeds

UDP is itself conducive to a dispersal of activities.

In terms of total trips, the results show a slight reduction (-0.3%), mostly fuelled by a 1.7%

reduction in bus trips. Pedestrian trips also decline, even if only by 0.4%. On the other hand,

car trips see a small increase (+0.3%). Despite a fall in trip numbers, passenger-kms actually

increase (+0.3%). This is brought about by an increase in average trip length of 0.7%. CO2

emissions increase by 0.6%, probably as a consequence of longer car journeys. More

surprising perhaps is the 6.2% increase in average bus speeds. Despite the increase in car trips,

car speeds remain virtually the same.

Again, disaggregate trip data is helpful in interpreting these results. Figure 5.1 shows that, in

the peak, car trips remain static, bus trips fall by 2% but pedestrian trips actually increase by

3.5%. The reduction in bus trips seems consistent with the fall in central area employment.

Indeed, public transport is much more attractive for radial trips than for trips within the

suburbs, where all the growth in population and employment has actually taken place.

Furthermore, the reduction in central area car traffic explains the noticeable increase in bus

speeds. As for pedestrian trips, they decrease in the two inner rings but suffer a remarkable

increase for trips within ring 3 (+22%). This shows that the re-location of employment to the

same suburbs as population can make slow modes more viable for the journey to work,

especially where public transport cannot provide a frequent, high-quality service. However,

this move to the suburbs is also responsible for higher car trip lengths. Car users are now able

to reach wider areas within the same travel time since outer areas roads are relatively

uncongested.

40

With regard to the off-peak, figure 5.1 shows a small overall increase in car trips, nevertheless

outweighed by a reduction in bus and walking trips. Despite the increase in peak-time

pedestrian trips, the car seems to become dominant in the off-peak because of reduced

congestion. Indeed, whereas peak pedestrian trips within ring 3 increased by 22% relative to

the do-minimum, off-peak trips only increased by 4%.

The results in table 5.4 reflect the analysis in the previous paragraphs. Car users benefit from

marginal reductions in running costs, probably because there are fewer trips to the central area

than in the do-minimum. Public transport operators actually lose money (-30 thousand euros),

probably because of a fall in the revenue from trips to the central area. The small increase in

government revenues is probably a combined result of lower revenues from central area

parking and higher revenues from increased fuel consumption. The results show external

benefits despite an increase in CO2 emissions. This is probably due to a reduction in accident

and noise-related costs since traffic has been displaced from the most densely populated areas.

As regards time savings, car users suffer a reduction, while public transport users and

pedestrians benefit from an increase. The time loss for car users is probably due to the

concurrent increase in car trips and travel times in the outer areas. Time savings to bus and

walking trips are probably a consequence of a reduction in central area traffic.

In terms of land use benefits, the net result is similar to that obtained by the centralisation

scenario. However, the results favour households relatively to landlords. This is reasonable

since, as people move from dense suburbs to outer suburbs or rural zones, rents tend to fall.

5.2.3. Compact City (LU3). This scenario produces the same type of results as LU1. As

mentioned earlier (see methodology chapter), its main difference in terms of land-use

structure is that employment and population growth are not as strongly concentrated in the

two central zones but rather spread homogeneously across rings 1 and 2.

Relatively to the do-minimum, the total number of trips increases, car trips fall, and both bus

and pedestrian trips increase. However, both the changes in car and bus trips are smaller than

occurs for LU1. Average trip length falls across all modes (-5.6% overall). In comparison to

LU1, however, there is a larger reduction for car trips (-2.6%) and a smaller reduction for bus

trips (-5.3%). Average car speed falls by 2.4%, which is a considerably higher drop than for

41

LU1. Average bus speeds increase by 2.9%. CO2 emissions fall by 5.1%, probably because of

the decrease in overall distance travelled.

Figure 5.1 and table 5.3 show that there is only a small difference between scenarios LU1 and

LU3 in terms of peak-time trips. Car has around 1% more trips and both bus and walking

have each around 1% less. An inspection of the trip matrices shows that this change is due to

a relative displacement of population (and trips) to rings 2 and 3, along with a decentralisation

of employment (and trip destinations) to ring 2. As a result, bus and walking become

relatively less attractive for the journey to work (the two central zones have the highest public

transport accessibility and have relatively small internal trip distances). Furthermore, car

becomes relatively more attractive as employment decentralises and thus becomes closer to

residences in the inner and outer suburbs. One positive outcome from this trend is that

average car trip length falls even further than for scenario LU1.

With regard to off-peak trips, there is a drop in bus trips (-4%) matched by an increase in car

trips, in relation to LU1. This is probably due to the fact that the car becomes more attractive

for trips between suburbs in the off-peak, when bus frequencies drop. However, off-peak

pedestrian trips remain at about the same level as LU1, which shows that scenario LU3 still

creates enough local opportunities for short-distance trips by slow modes.

The results of the CBA are again similar to those of scenario LU1 with a few exceptions.

Firstly, this scenario produces negative money savings for bus operators. Secondly, it

produces negative time savings for car users and double the value of negative time savings for

public transport users in comparison to LU13. And finally, it produces negative benefits for

both households and landlords4.

5.2.4. Decentralized concentration (LU4). It is important to begin the analysis of this scenario

by underlining that average travel speeds are 18% below the do-minimum level. This explains

why total trips fall by 12%, total passenger-kms by 13.6% and CO2 emissions by 11.4%. As

travel speeds decrease, peak-time trips take more time. In order to maintain the constant travel

time budget hypothesis, individuals then have less time for non-essential trips and the total 3 Although a reduction in money and time savings relative to LU1 would be warranted by decreasing bus patronage and increasing congestion, these values should not be negative (which means they are worse than for LU0). There is no apparent explanation for this phenomenon. 4 It is also unclear why this is so.

42

number of trips falls. It seems reasonable to wonder whether this scenario goes beyond the

model’s capabilities, since such a rise in congestion is likely to bring about changes in

behaviour which the model is not programmed to consider (for example, a change in work

times, resulting in a shift in trips from the peak to the off-peak).

One can argue that this large decrease in speed is due to the spatial structure proposed by the

scenario, in which a large proportion of population and jobs are concentrated in three

suburban zones. As a result, these three suburbs begin to attract large numbers of workers

from surrounding zones and gridlock ensues as road capacity is insufficient to cope with such

high demand. This type of scenario is bound to perform poorly unless adequate transport

supply provisions are made, such as new high-capacity public transport and road links.

However, studying the effect of new transport infrastructure on this type of scenario goes

beyond the scope of this dissertation.

This scenario achieves the largest reduction in mean car trip length (-2.7%). It is however

unclear whether this is due to the urban structure proposed (jobs are closer to residences) or to

factors related to the reduction in the number of trips (need to cut trip distance so as to be able

to make essential trips within the allowed travel time budget).

Another feature of this scenario is that car’s modal share actually increases (+2.3%). One

reason for this is that public transport services (constant for all scenarios) have not kept up

with changes in urban structure. This supports the earlier suggestion that such large changes

to urban structure need to be complemented by changes in transport supply.

In terms of the CBA, this scenario performs much worse than any of the others. Public

transport operators lose large sums because of much lower patronage, especially in the off-

peak. The government also loses much of its revenue, probably because lower trips imply

lower parking revenues and lower fuel consumption, even if many trips are now undertaken

under more congested conditions. As a result of great reductions in speed, all transport users

lose considerable amounts of time. However, there are lower external costs than in any of the

other scenarios since lower traffic produces less pollution and creates fewer accidents. As

regards land use benefits, they favour households more than landlords, possibly because

population is made to move to cheaper suburban locations.

43

5.3. Impact of transport strategies on travel patterns and transport system performance

5.3.1. Optimal transport strategies. Table 5.5 shows the optimal transport strategies for each

land-use scenario. According to these results, land-use structure has only a small bearing on

the best transport strategy achievable.

Table 5.5: Optimal transport strategies (TAS) Fare freq5p freq15p rch5p rch15p r.cap freq5op freq15op rch5op rch15op LU1 -50% 200% 75% 4.50 4.50 20% 200% 200% 1.00 1.00 LU2 -50% 200% 75% 3.00 3.00 20% 200% 150% 0.50 0.50 LU3 -50% 200% 75% 3.00 3.00 20% 200% 200% 0.50 0.50 LU4 -50% 200% 75% 5.00 0.00 20% 200% 100% 0.00 0.00 LU0 -50% 200% 75% 3.50 3.50 20% 200% 150% 0.40 0.40

It was found that large reductions in public transport fares (-50% - the maximum value

allowed) should be part of the optimal transport strategy regardless of the land-use structure in

place. Likewise, small road capacity improvements (traffic management measures) were also

found to be optimal at their maximum level (+20%) across scenarios.

As for public transport frequency, there are small variations in intensity between scenarios but

the results invariably point in the same direction. In the peak, frequency should be increased

by the maximum amount allowed (+200%) in year 5 and then gradually decreased until year

15, to finish at +75% of the do-minimum level. In the off-peak, frequency should equally be

increased by 200% in year 5. However, there are differences in the optimal level it should

assume by year 15. In the concentration scenarios (LU1 and LU3), off-peak frequency should

remain at its maximum level throughout the planning period. In the dispersal scenarios (LU0

and LU2), frequency should fall slightly over time to end at +150% of do-minimum. And for

scenario LU4 frequency should end at +100% of its do-minimum level.

44

It seems reasonable to suggest that the differences over time in optimal public transport

frequency are related to the travel patterns induced by each land-use scenario. LU1 and LU3

produced the highest number of off-peak bus trips. As a result it makes sense to keep

frequency high because there are more potential users who will benefit. On the other hand,

LU4 produced the lowest number of off-peak bus trips. Not surprisingly, it has the lowest

increase in off-peak public transport frequency. This reasoning also applies to the peak period.

All scenarios produce approximately the same number of peak bus trips and therefore the

optimal increase in frequency does not vary amongst them.

Similarly to bus frequency, the optimal level of road charging also varies slightly between

scenarios. In general, however, it was found that the optimum peak charge would be between

3 and 4.5 euros, and the corresponding off-peak charge would be between 0 and 1 euro. With

the exception of LU4, it was found that a constant charge produced similar results to an

equivalent overall charge that varied over time.

One can argue that the optimal road charge level is directly linked to the urban structure

proposed in each scenario. For example, LU1 has the highest charge probably because it

produces the highest concentration of employment in the central zone. As a result, a high road

charge serves to dampen the high levels of congestion that would probably arise. At the same

time, public transport offers a competitive alternative to car for trips to the central zone. Those

users that do decide to switch modes will therefore lose less time relatively to the transport

do-minimum scenario. By reducing congestion, this measure can therefore bring large time

savings to both car drivers and public transport users.

Following this reasoning, those scenarios that concentrate less activities in the central zone –

LU0, LU2, LU3 and LU4 in decreasing order – have lower optimal road charges because the

marginal benefits from a reduction in central area congestion are smaller. LU4 is the extreme

example. Although its initial optimal road charge is fairly high (5 euros in the implementation

year), it drops to zero in the long-run. This is possibly because, as congestion quickly mounts

in the outer suburbs where activities are concentrated, there are benefits to be had from

diverting as many trips as possible to other areas, including the central zone. Although this

rationale is able to explain the road charge level for scenarios LU1 and LU4, it fails to clarify

why LU3 does not have a higher charge than LU2.

45

5.3.1.1. The impact of optimal transport strategies on transport system performance. Table 5.6

shows the percentage change in trips, by mode and time period, from transport do-minimum

to the optimal transport strategy for each scenario. The changes are generally of the same sign

and order of magnitude across scenarios. In the peak, car trips fall by between 6 and 11%, bus

trips increase by between 30 and 38% and walking trips fall by between 24 and 28%. In the

off-peak, car trips increase by between 15 and 17%, bus trips increase by between 70 and

90% and walking trips fall by between 23 and 28%. Although these small differences across

scenarios may imply that the interaction between land-use structure and transport measures

may not be very significant, there are a few relevant differences worth reporting.

Table 5.6: Percentage change in trips from transport do-minimum to the optimal transport strategy (year 15 forecasts) car-peak Car-op bus-peak bus-opeak Ped-peak ped-opeak L1_TAS -9.5% 15.8% 37.7% 89.1% -24.4% -26.8% L2_TAS -7.9% 16.2% 35.2% 81.9% -28.2% -26.8% L3_TAS -8.6% 15.3% 36.5% 89.2% -24.9% -27.6% L4_TAS -6.2% 17.0% 30.9% 72.7% -26.9% -23.7% L0_TAS -10.6% 17.1% 34.8% 83.0% -28.1% -26.3%

Starting with the peak period, scenario LU0 shows the largest percentage reduction in car

trips (-10.6%). It is not clear why this is so. The nest largest reduction occurs for scenario

LU1 (-9.5%), followed by LU3 and LU2 (-8.6 and -7.9%). As for scenario LU1, it is clear

that a high degree of employment centralisation, combined with a high road charge, lower bus

fares and increased bus frequency, is likely to produce a significant modal shift. In effect,

LU1 also produces the highest percentage increase in bus trips of all scenarios. Furthermore,

an inspection of trip matrices (appendix 3) shows that LU1’s optimal transport strategy

produces the largest percentage reduction in peak car trips to the central ring.

With respect to bus trips, LU3, LU2 and LU0 follow LU1 as the scenarios with the largest

percentage increase in trips. It is surprising that LU0 is below LU2 in this ranking, given that

LU0’s optimal road charge was higher and that its land-use structure is apparently more

conducive to public transport use. LU1 manages the smallest reduction in walking trips (-

24.4%), closely followed by LU3 (-24.9%). The fact that the gap between these two scenarios

is narrower for this mode highlights the fact that the compact city scenario may actually be

more favourable to walking than centralisation, as argued in previous sections. Both LU0 and

46

LU2 suffer a much larger reduction in this type of trip (-28%), indicating that the bus is a

closer substitute for walking in these scenarios.

The change in off-peak trips follows a different pattern from the peak. Pedestrian trips

decrease by approximately the same proportion but all other trips increase significantly. This

increase in the total number of off-peak trips is made possible by the time savings in peak

trips that the optimal transport strategies enabled (less car trips and higher capacity lead to

less road congestion; and higher bus frequency results in lower waiting times). It is perhaps

surprising that it was bus rather than car that had the largest increase in trips. Yet the

difference in generalised cost between the two modes is likely to have narrowed considerably

because of much lower waiting times (note that value of time is considered to be higher for

waiting time than for in-vehicle time) and lower fares.

Table 5.7 shows the change in average trip length and speed produced by the optimal

transport strategies in relation to transport do-minimum. As you would expect trip length

increased for both car and bus. As travel costs come down, individuals are able to travel

longer distances within the same amount of time and the ‘time-space action zone’

(Hagerstrand, 1970) expands. It is nonetheless surprising that trip length increases by such a

small percentage (3.3% for car and 1.5% for bus trips) given the large speed increases (over

20% for cars and between 20 and 40% for buses). This is probably due to the rigidity of the

land-use scenarios introduced. If more land was made available then trip lengths would

probably increase even further. On the other hand, individuals must consider a trade off

between variety of opportunities (proportional to distance travelled) and frequency of trips. In

this case, individuals seem to prefer making more trips.

Table 5.7: Percentage change in mean trip length and travel speed from transport do-min to optimal transport strategies (year 15 forecasts) LU0 LU1 LU2 LU3 LU4 Trip length (kms) Total 11.7% 11.8% 11.7% 11.9% 10.7% Car 3.5% 3.3% 3.4% 3.2% 2.9% bus 1.6% 1.2% 1.7% 1.1% 2.0% Average speed (kph) average 16.9% 19.2% 15.7% 2.8% 47.4% Car 27.6% 27.3% 23.3% 27.1% 27.5% Bus 39.8% 34.9% 20.5% 32.0% 26.2%

47

Table 5.8 shows the results of the cost-benefit analysis. In terms of the OF value, the

combined scenario of LU1 and its optimal transport strategy ranks at the top of the list and is

followed, respectively, by LU2, LU0, LU3 and LU4. The top three scenarios appear to

perform much better than the last two.

Table 5.8: CBA results for the optimal transport strategies money savings Time savings Assets Users PT ops toll govt external h'way pt walk

L0_TAS 549 1655 -1880 523 -177 -178 2181 5414 -847 L1_TAS 415 1427 -2006 803 -359 4 2261 5735 -1087 L2_TAS 549 1693 -1914 469 -159 -160 2117 5327 -810 L3_TAS 415 1711 -2085 487 -344 10 2038 5450 -1024 L4_TAS 1592 1767 -2146 164 -579 249 636 3054 -970

LU benefits (rent)

h'holds Ops OF OF (net of lu)

L0_TAS -107 110 7200 7197 L1_TAS 136 222 7637 7279 L2_TAS 658 -303 7417 7062 L3_TAS -788 -758 5137 6683 L4_TAS 532 -625 3801 3894

It is useful to look in more detail at the different components bearing on the final OF value.

One first point to highlight is that time savings to public transport users are the predominant

source of benefits. Money savings to transport users (in terms of fuel, parking charges, road

charges, bus fares) and time savings to car users are also sizeable benefits but together form

less than 70% of the time savings attributed to public transport users. LU1 produces the

largest time savings for car and bus users but the least money savings.

In relation to costs, public transport operators bear the largest proportion due to the provision

of additional frequency and reduction in public transport fares. However, these costs are

generally less than the benefits associated with time savings for public transport users. The

other main contribution to costs comes from increases in travel time for pedestrians (around

half the magnitude of time savings for car users). These arise from the large increases in

vehicle-kms (especially bus: +70% from do-minimum) and travel speeds (+16% from do-

minimum) for all motorized modes5.

5 Note that, in SPM, walking conditions are affected by the level of motorized traffic

48

Given this broad configuration of costs and benefits it becomes clearer how the optimal

transport strategies came about. Firstly, road capacity will always be optimal at its maximum

level because it contributes to time and money savings, the two most important sources of

benefits. Moreover, inspection of the CBA tables shows that it produces a negligible fraction

of capital costs.

Large increases to public transport frequency are also likely to be part of any optimal strategy

since they produce time savings of their own and are also responsible for a significant modal

shift (from both car and walking). As a result, they contribute to reductions in congestion and

further time savings to both car and bus users. However, this measure has a considerable cost

and may therefore not be optimal at its maximum level (assuming diminishing returns on

investment in higher public transport frequency).

Large reductions in fares are also optimal because they reinforce the modal shift to public

transport. As a result, congestion falls further and there are more bus users who can benefit

from increased frequency. Road charging is likely to have similar effects to fare reductions.

However, its optimal value is around the middle of the interval considered, rather than at one

extreme. The reason for this may be that as the charge increases, car users may shift to public

transport or may instead choose to travel longer distances and so the marginal benefits of

higher charges decrease (in the case of LU1, the central zone is so dominant that it is optimal

to push up the cordon charge).

In the light of these arguments, it is reasonable to expect LU1 to achieve the highest OF value

since its land-use structure is most conducive to a modal shift to public transport.

It is relevant to comment briefly on two other elements of the CBA. Firstly, it is notable that

capital assets have positive values for all scenarios despite the fact that all scenarios require an

investment in public transport capacity (because of increased frequency). Careful inspection

of the user-defined parameters in SPM revealed that the price that can be recovered from

selling public transport vehicles is set at a higher level than their initial cost. This counter-

intuitive finding seems to be responsible for the reduction in public transport frequency over

time in the optimal transport strategy.

49

A second point to note is that external costs have a very small weight in the overall OF value

(less than 2% in scenarios LU0 to LU3). This explains why optimal strategies were found to

lead to an increase in CO2 emissions and in total motorized trips.

5.3.2. Sensitivity of transport system performance to transport strategy intensity. Figures 5.2,

5.3 and 5.4 illustrate the variation in three aggregate indicators (objective function, car-

passenger-kms and CO2 emissions) for all scenarios as the ‘intensity’ of a standard transport

strategy is varied. The standard transport strategy (corresponding to index 100 on the x-axis)

comprises a 50% reduction in bus fares, a 20% reduction in road capacity, a 200% increase in

bus frequency, a 6 euro peak road charge and a 1 euro off-peak road charge.

Figure 5.2 shows that the objective

function responds in a similar manner to

changes in intensity across all scenarios:

OF is an increasing function of strategy

intensity but subject to a declining rate

of growth. It is unclear at this point

whether this declining rate of growth is

due to diminishing marginal returns (for

example, if the elasticity of demand with

respect to fares falls as fares increase), or

whether it arises because individual

transport measures begin pulling in different directions as intensity increases (for example, if

there is a threshold after which

increases in bus frequency begin

generating more costs than benefits).

Figures 5.3 and 5.4 show very

similar results to figure 5.2. With the

exception of scenario LU4, car-pax-

kms and CO2 emissions are

increasing functions of intensity. The

reason behind this general trend in

-4000

-2000

0

2000

4000

6000

8000

10000

0 50 100 150

transport strategy intensity

OF

D-minL1L2L3L4

Figure 5.2: OF value versus transport strategy intensity (year 15 forecasts)

y = 10456x + 8E+06

y = 9175.3x + 7E+06

6,000,000

6,500,000

7,000,000

7,500,000

8,000,000

8,500,000

9,000,000

9,500,000

0 50 100 150transport strategy intensity

car-

pax.

kms

d-minL1

L2

L3

L4

Figure 5.3: Car passenger kms versus transport strategy intensity (year 15 forecasts)

50

distance travelled is that, because of the constant travel time budget hypothesis, transport

measures which reduce travel time will lead to an increase in the number or length of trips.

With regard to CO2 emissions the increase is due to the way public transport emissions are

calculated: they begin at zero and start increasing only once public transport frequency has

gone over a certain threshold.

If one excludes the first point in the

CO2 series (intensity index 20), then

CO2 and car-pax-kms are both

approximately linear functions of

intensity. Linear regression was

applied to scenarios LU1 and LU2

and it turned out that scenario LU1

has a lower gradient than LU2 by

about 10%. This means that in spite

of responding broadly in the same

way to changes in transport intensity,

the sensitivity of that response varies with land-use structure. In other words, different

synergies develop between land-use and transport, depending on the type of land-use structure.

This is not so surprising if one bears in mind that, as has been greatly emphasized in previous

sections, concentration scenarios are able to produce higher modal shifts to public transport.

The reason why LU4 behaves differently from the other scenarios is due to the way the model

calculates CO2 emissions and to the higher relative weight of public transport frequency in

this scenario. It is a relatively incidental result which would take too long to clarify and is

therefore out of this report.

5.3.3. Elasticities to individual transport measures. Table 5.9 shows the elasticity of OF, car-

pax-kms and CO2 emissions with respect to changes in individual measures, for scenarios

LU1 and LU2.

Starting with LU1, the results show that the value of the OF is determined almost entirely by

variations in public transport frequency and road capacity. This indicates that these measures

are likely to induce great time and money savings in a very cost-effective way.

y = 411.03x + 483517

y = 463.35x + 515485

430,000

450,000

470,000

490,000

510,000

530,000

550,000

570,000

590,000

0 50 100 150transport strategy intensity

CO

2 em

issi

ons

d-min

L1

L2

L3

L4

Figure 5.4: CO2 emissions versus transport strategy intensity (year 15 forecasts)

51

On the other hand, public transport fare reductions have only a marginal impact on the OF.

This suggests that fare reductions are not very cost-effective. Indeed, this measure is not

likely to attract significant numbers of car drivers to public transport. Instead, it will mostly

reduce the number of pedestrian trips. Since it has little bearing on the level of road

congestion, the benefits brought about by this measure will be limited to time savings for

those pedestrians who will now travel by bus.

The effect of road charging is

practically negligible. This comes as

less of a surprise if one bears in mind

that the cordon charge only affects trips

to and from zone 1, and that the off-

peak charge is very small. Even in

scenario LU1, the number of peak time

car trips affected by the charge comes

to only 8% of total trips. And although

this measure is likely to reduce

congestion and therefore generate travel time savings for car users, it is also likely to produce

time losses for those car users who shift to public transport. Similarly, the toll may produce

revenue for the government but represents expenditure for car users. These reasons may

justify why road charging tends to be neutral in cost-benefit terms.

With regard to car-pax-kms, road capacity is by far the most influential transport measure

(e=0.158). This highlights the direct causal link between increased accessibility (due to

greater road capacity and therefore higher speed) and greater distance travelled.

Public transport frequency has the opposite effect on distance travelled. This is because an

improved bus service is likely to attract some car users and therefore reduce the number of car

trips. On the other hand, the direct time savings stemming from lower waiting times allow

individuals to undertake a larger number of off-peak trips (according to the constant travel

time budget hypothesis). Likewise, a modal shift towards bus is likely to reduce road

congestion and produce time savings for car users as well. The opposing direction of these

two forces (modal shift and increased travel distance because of time savings) provides an

Table 5.9: Elasticity of OF, car-pax-kms and CO2 emissions with respect to changes to public transport fares and frequency, road charge and road capacity LU1 w.r.t. fares frequency rch rcap OF -0.039 0.178 -0.003 0.321 C.p.kms 0.022 -0.018 -0.005 0.158 CO2 0.030 0.040 -0.008 0.046 LU2 w.r.t. fares frequency rch rcap OF -0.044 0.235 0.000 0.324 C.p.kms 0.016 -0.016 -0.002 0.159 CO2 0.024 0.025 -0.004 0.048

52

explanation for the relatively low sensitivity of car-pax-kms to variations in bus frequency

(e=-0.018).

The low sensitivity to bus fares (e=0.022) is in line with the common belief that the cross-

elasticity between bus costs and car trips is relatively low, but positive. The elasticity with

respect to road charging is of the opposite sign but of even lower magnitude. Similarly to the

case of bus frequency, there are several dynamics at work. On the one hand, a higher charge

will cause a modal shift towards public transport. On the other hand, it may cause a

lengthening of trips as car users choose to drive to more distant destinations in order to avoid

the charge. In any case, road charging only affects a small fraction of all trips and its effect is

therefore likely to be low.

CO2 emissions show a higher, positive elasticity with respect to fares and frequency: a

slightly higher, but negative elasticity with respect to road charging and a lower elasticity

with respect to road capacity. The fact that the elasticity w.r.t. frequency is positive is due to

the fact that the growth in bus CO2 emissions is higher than the corresponding decrease in car

emissions arising from modal shift. The higher elasticities w.r.t. bus fares and road charging

are probably due to the fact that these two measures tackle emissions where congestion is

most severe. In those areas, any improvements are likely to greatly reduce emissions. The

relatively lower sensitivity to changes in road capacity may, in turn, arise from the fact that

this measure is applied homogeneously across the city, with no particular emphasis on those

areas where it may have higher returns (the most congested areas).

Turning now to the elasticities for scenario LU2, they are broadly in agreement with those

obtained for LU1 in terms of sign and magnitude. All elasticities w.r.t. road capacity are

slightly higher for LU2, probably because it has a higher proportion of car drivers and a more

car-oriented land-use structure. As regards road charging, elasticities are about half their

corresponding value for LU1. This is to be expected given that the proportion of car trips

affected by the cordon is much lower in scenario LU2. With respect to car-pax-kms and CO2

emissions, these indicators are less responsive to bus fare and frequency changes. This may

occur because of the higher proportion of car trips and the wider gap in generalised cost

between car and bus. More surprising however, is that the OF value is actually more

responsive to changes in public transport costs for scenario LU2. This may have to do with

the complex interactions between different costs and benefits, for example those associated to

53

time and money savings for different groups. However, it is hard to see why exactly this

unexpected results occurs.

5.4. Impact of transport strategies on land use patterns

Figures 5.5 to 5.8 show the difference in population (year 15) between optimal transport

strategy and transport do-minimum for each land-use scenario.

LU1. About 1% of the total number of households move from outer suburbs in ring 3 (Pudsey

South, Bramley, Moortown, Seacroft and Cookridge) and into Beeston (ring 2) and

Weetwood (ring 3). The introduction of the transport strategy seems to have led to a re-

population of an inner suburb close to the central zone, which do-minimum results had shown

to be particularly undesirable as a place to live. It is unclear why Weetwood suffers a small

population increase, since it is no more accessible than any of the other outer suburbs. It may

be that congestion would have limited the extent of its growth. Some households would

decide to move there when conditions improved.

LU2. About 2% of total households move

out of Harehills (ring 2), Middleton,

Roundhay and Seacroft (ring 3), and

Barwick (ring 4), and move into Armley,

Chapel Allerton, Burmantofts (ring 2), and

Aireborough (ring 4). Most zones with a

population increase are within ring 2. The

transport strategy has also lead to a re-

concentration of population in some inner

suburbs, possibly because of a reduction in

central area congestion and the increase in

relative public transport accessibility of

those zones. The large increase in

population in Aireborough may be due to a

reduction in congestion (which will have

allowed for any extra latent growth that

would have been otherwise curtailed) and

-200

-100

0

100

200

300

400

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33

population

Figure 5.5: Change in spatial distribution of population as a result of the optimal transport strategy – scenario LU1 (year 15)

-600

-400

-200

0

200

400

600

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33

population

Figure 5.6: Change in spatial distribution of population as a result of the optimal transport strategy – scenario LU2 (year 15)

54

the fact that it is relatively accessible by public transport.

LU3. About 1% of total households move out of Wortley and Harehills (ring 2) and into

Armley, Chappel Allerton, Beeston and Headingley (ring 2). Lower congestion allows re-

location to more desirable areas.

LU4. About 1.5% of the population move

away from Bramley, Moortown, Seacroft,

Harehills (ring 3) and Kirkstall (ring 2), and

move into University (ring 1) and

Burmantofts (ring 2). The move to

University may have to do with availability

of land, reduced congestion and improved

public transport.

It is difficult to make full sense of these

results, especially because the development

constraints are different in each scenario. If

land was freely available in all zones of the

city then one would expect population to

move according to changes in transport

accessibility. In reality, some zones are

always fully developed by year 15

regardless of changes in accessibility. This

makes it difficult to appreciate all the

dynamics at play without looking at every

zone in detail to see whether its development capacity is exhausted.

In any case, the addition of transport strategies seems to have had two main effects. On the

one hand, it generated a re-concentration of population in the central area (rings 1 and 2),

where the cost of public transport was greatly reduced. On the other hand, by reducing

congestion it allowed residents to move to more desirable areas (in the case of LU2,

Aireborough; in the case of LU3, most of the inner suburbs). Whatever the case, these effects

are very small in magnitude.

-400

-300

-200

-100

0

100

200

300

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33

population

Figure 5.7: Change in spatial distribution of population as a result of the optimal transport strategy – scenario LU3 (year 15)

-600

-400

-200

0

200

400

600

800

1000

1200

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33

population

Figure 5.8: Change in spatial distribution of population as a result of the optimal transport strategy – scenario LU4 (year 15)

55

5.5. Synergy between land use and transport strategies

Table 5.10 compares the OF value obtained by implementing together each land use scenario

and the respective optimal transport strategy, against the value obtained from the separate

implementation of the land use and transport measures. The bottom row of the table calculates

the ‘synergy’ effect, i.e., whether the integrated strategy achieves a higher OF than the sum of

the OF values achieved by the two types of measure implemented separately – a positive

value indicates a synergy effect. Since the separate implementation of the transport measures

is actually done against the do-minimum land use scenario, what is really being measured is

whether the synergy between a given land-use scenario and its optimal transport strategy is

higher than the synergy between the do-minimum land use scenario and that same transport

strategy.

Table 5.10: Synergy between land use and transport policies LU1 LU2 LU3 LU4 1 Combined LUi+TASi 7,637.05 7,416.71 5,136.69 3,800.80 2 single LUi 415.15 378.92 -1,852.82 -3,337.56 3 single TASi 7,196.31 7,195.11 7,197.94 6,971.15 1-(2+3) 25.58 -157.31 -208.44 167.21

The results are somewhat contradictory but, on the whole, show few signs of significant

synergy effects (the maximum synergy value – LU4 – represents less than 3% of the OF value

achieved by LU0 and the respective optimal transport strategy). LU4 shows the greatest level

of synergy, followed by LU1. LU2 and LU3, however, show negative synergy values, which

means that LU0 develops a stronger level of synergy with their respective transport strategies.

These results are somewhat difficult to interpret given that the OF is a relatively complex

indicator. In any case, it is not surprising that there are no particularly large synergy effects

since the optimal transport strategies do not differ significantly between land use scenarios.

Furthermore, the OF value is most sensitive to road capacity enhancements (as shown in point

5.3.3) whose optimal value does not vary at all between optimal transport strategies.

These results probably warrant a deeper investigation but because of the time constraints on

this project and because the synergy estimates are so small, it was decided not to take this line

of research any further.

56

5.6. Optimal land-use structure

5.6.1. Spatial distribution of employment and population. Charts 5.9, 5.11 and 5.13 show the

changes with respect to the do-minimum in the spatial distribution of population for scenarios

EMP1, EMP2 and EMP3 (for no change in fuel tax and a 500% increase). Charts 5.10, 5.12

and 5.14 represent the corresponding changes in the spatial distribution of population.

These results show that there is a strong tendency for population to move to the most central

zones even when a significant proportion of jobs decentralize to the more peripheral areas. In

scenario EMP2, for example, circa 10% of all jobs move out of ring 1 to rings 3 and 4, while

zones 1 and 2 (City Centre and Beeston) experience the largest gains in population.

Most inner suburbs within ring 2

also experience an increase in

population, though more modest

than the central zones. This

increase is most accentuated for

zones 9, 10, 11 and 12, to the

west of the city centre. On the

other hand, zones 3 and 4

(Wortley and Armley), to the

southeast and east of the city,

suffer either smaller population

increases or actually lose

inhabitants relatively to the do-

minimum (which is the case for

zone 3).

As for the zones within rings 3

and 4, they generally suffer a

population decline, which is

more noticeable under employment scenario EMP1 (centralisation) and less so under scenario

EMP2 (dispersal). Zones 19, 20, 21, 23 and 30, situated to the north of the city centre within

Figure 5.9: Change in spatial distribution of population – scenario EMP1 (year 15)

-20000

-10000

0

10000

20000

30000

40000

50000

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33

fuel 0

fuel +500%

Figure 5.10: Change in spatial distribution of employment – scenario EMP1 (year 15)

-10000

0

10000

20000

30000

40000

50000

60000

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33

fuel 0

fuel +500%

Population – EMP1

Employment – EMP1

57

ring 3, are the exception. They all see a relative population growth, which is greater for

scenario EMP2.

Although these results show a general movement of population towards the centre of the

study area, the intensity of this phenomenon depends significantly on the direction in which

employment is moving. If 50,000 jobs are assumed to disperse towards the edge of the city

(EMP2) then the relative

increase in the population of

zones 1 and 2 comes to 25000

people and to around 17000 in

the remaining inner suburbs. If,

on the other hand, those jobs are

assumed to be equally

distributed between the central

area and the surrounding suburbs

(EMP3), then the corresponding

increases in population are

30,000 and 28,000. However, if

employment is concentrated

within the central area alone,

then the population of zones 1

and 2 grows by about 58,000,

and the surrounding suburbs by

17,000.

Another interesting feature of

these results is that they seem to show that there is a limit to how much concentration is

desirable. Indeed, the model only moved a relatively small proportion of the total number of

inhabitants, which means that many peripheral suburbs kept the majority of their base

population. On the other hand, many households were moved to inner suburbs rather than to

the most central area.

It is not completely clear why this result came about. However, three reasons come to mind.

Firstly, many home-to-work trips are likely to take place within the same zone. As a result,

-20000

-10000

0

10000

20000

30000

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33

fuel 0

fuel +500%

-30000

-20000

-10000

0

10000

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33

fuel 0

fuel +500%

Figure 5.11: Change in spatial distribution of population – scenario EMP2 (year 15)

Figure 5.12: Change in spatial distribution of employment – scenario EMP2 (year 15)

Population – EMP2

Employment – EMP2

58

for those individuals who happen to work in a peripheral zone, accessibility will be

maximised by living near to work and far from the centre. SPM does not model such

individual behaviour explicitly but it does so implicitly by using a gravity-type trip

distribution model.

Secondly, if individuals seek to maximise their accessibility by private car and some jobs

remain within suburban locations, then it may make sense to relocate to an inner suburb with

good road connections to the whole city (this is most clear for scenario EMP3). In this way,

people are able to easily access both the main employment centre (zone 1) and a wider job

pool scattered throughout the

remaining parts of the city. This

explains why population growth

outside the central area occurs in

the zones to the north and east of

the city centre, which are at the

geographic centre of the city, but

not in zones 3 and 4, which are

on the border of the study area.

The final reason has to do with

congestion. As more people and

jobs locate within the central area

congestion is likely to grow. At

some point, accessibility from

the central area will fall below

the accessibility from the

surrounding suburbs. As a result,

population growth in the central

area will tend to taper off.

Increasing fuel tax is a measure that tends to reinforce the concentration of population within

the central area at the expense of most other zones. However, this outcome is unlikely to be

-20000

-10000

0

10000

20000

30000

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33

fuel 0

fuel +500%Population – EMP3

-30000

-20000

-10000

0

10000

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33

fuel 0

fuel +500%Employment – EMP3

Figure 5.14: Change in spatial distribution of employment – scenario EMP3 (year 15)

Figure 5.13: Change in spatial distribution of population – scenario EMP3 (year 15)

59

due to a need to reduce travel costs6. A more likely explanation is that higher fuel costs lead

to a reduction in car travel and therefore in road congestion. According to the reasoning in the

previous paragraph, this then allows for the central population to keep increasing without

compromising accessibility.

5.6.2. Aggregate transport indicators. Table 5.11 shows the change from do-minimum in

aggregate transport system indicators. These results are useful in judging whether a land-use

structure in which individuals seek to maximise their accessibility does bring about

improvements in the performance of the transport system (in other words, is the ‘optimal’7

land-use structure a system optimum or a selfish/user optimum?).

All scenarios achieve a reduction in peak time car trips. For scenarios EMP2 and EMP3 there

is also a reduction in bus trips. Walking trips rise significantly for all scenarios. The fall in

average vehicle speeds (more significant than those reported in section 5.2) explains why the

car and, in some cases, the bus have become relatively less attractive than other modes. This

result suggests that these ‘optimal’ land use structures actually lead to reduced traffic

efficiency. On the other hand, by bringing jobs and residences closer together, they make

walking relatively more attractive, thus contributing to greater equity. EMP3 had the largest

percentage increase in walking trips.

All employment scenarios achieve a reduction in average trip length, which is greatest for

EMP1 (-10% for all motorized trips). Furthermore, this reduction is about twice as great as

that achieved for any of the scenarios reported in section 5.2. This shows that the proposed

optimisation method can automatically produce a land-use structure in which mean trip length

is reduced. It is difficult to prove conclusively that this fall in trip length is due to the land-use

structure proposed, rather than to the increase in congestion and consequent shift to slower

modes. Nevertheless, it is important to notice that the reduction in trip length is not actually

correlated with the percentage change in walking trips across scenarios.

When fuel tax is increased by 500%, average speed rises and trip length falls further,

relatively to the previous scenarios. This is made possible by the shift to bus8 and walking that

6 In SPM, accessibility is actually not influenced by travel costs but only by car travel times. 7 Accessibility-maximising 8 In SPM, fuel tax increases do not affect bus travel costs

60

occurs for all scenarios. The results show that the increase in fuel tax achieves the greatest

decrease in trip length for scenario EMP1, probably due to the remarkable rise in walking

trips.

Table 5.11: Percentage change in aggregate transport indicators for optimal land use scenarios, relative to do-minimum (year 15 forecasts) F0 F+500% EMP1 EMP2 EMP3 EMP1 EMP2 EMP3 Trips -7% -2% -1% -2% 2% 3% Car -16% -6% -8% -23% -8% -12% Bus -1% -5% -1% 15% 10% 14% Walk 6% 9% 15% 29% 20% 26% Peak trips total 0% 0% 0% 0% 0% 0% Car -7% -4% -6% -22% -15% -17% Bus 5% -5% -7% 19% 13% 13% Walk 13% 19% 27% 32% 21% 29% CO2 emissions total -19% -7% -10% -31% -13% -19% Trip length (kms) total -10% -5% -9% -17% -8% -12% Car -5% -2% -4% -7% -4% -6% bus -11% -2% -8% -13% -2% -7% Average speed (kph) average -15% -7% -8% -12% 1% -3% Car -12% -7% -8% -5% 4% 1% Bus -13% 1% -2% -7% 9% 4%

These results show that, although fuel cost does not affect location decisions per se, it does

affect travel patterns and may possibly generate synergies with those land use structures

which enable greater use of walking and public transport.

5.6.3. CBA indicators. Table 5.12 shows the OF values (net of land use benefits) for the

‘optimal’ land-use scenarios. EMP1 achieves the lowest OF value. An inspection of the full

CBA table (see appendix 4) shows this to be due to significant time losses to road users, and

to lost fuel revenues. This result is consistent with the results in point 5.6.2, where it was seen

that EMP1 generated the greatest reduction in average speed. EMP2 is the only scenario

which produces a positive OF value.

61

The increase in fuel tax improves the value of the OF

significantly, especially by bringing about large time savings

to car and public transport users while being relatively neutral

in fiscal terms. Yet, the difference between the performances

of the different scenarios remains approximately the same.

Table 5.12: OF values for optimal land use scenarios

Of (net of lu)

F0 EMP1 -2243 EMP2 119 EMP3 -937 F5 EMP1 -1987 EMP2 1159 EMP3 -395