Archive:Hong Kong Mass Transport Study/Chapter 5

←

Chapter 4 — Planning Characteristics and Projections

Hong Kong Mass Transport Study, Chapter 5 Travel Characteristics and Projections

Chapter 6 — Recommended System

→

TRAVEL CHARACTERISTICS AND PROJECTIONS

1. A superficial inspection might give the impression that urban traffic movement is entirely random in character. However, consistencies are revealed when large numbers of journeys are reduced to such basic components as trip purpose, travel mode and trip length. All movement expresses the social and economic activities of the community and, once the relationship between travel and these activities is established, a mathematical evaluation of traffic flow can be made. Existing travel patterns in Hong Kong have been analysed intensively to derive these mathematical relationships.

HOME-INTERVIEW SURVEY

2. Two complete home-interview surveys were conducted by the Passenger Transport Survey Unit. One was based on a selective sample of households obtained from Government building and housing records. The other was based on a sample of car and motor cycle-owning households, taken from motor vehicle registration data. A total of 7,853 households was included in the former and 4,161 in the latter. The car-owning households included in the building-unit survey were later removed, to avoid duplication when the data from the two surveys were combined.
3. The home-interview survey data provided most of the in formation upon which this Study is based. A great deal of information about household characteristics, in addition to travel information concerning the members of the households, was gathered.
4. The Passenger Transport Survey Unit applied the initial expansions to the basic data. Additional expansions, adjustments and corrections were then made based on comparisons with other data, including all cross-harbour travel as measured in the ferry surveys. Analysis began after comparison with known full-scale data revealed sufficient correlation to proceed with confidence.

TRIP GENERATION (HOME-BASED)

5. Trip Generation is a term used to describe the number of journeys starting or ending in a zone, in relation to its social and economic characteristics. A study of trip generation does not attempt to cover all the characteristics of trips—direction, length, duration, etc.—but simply to quantify and classify the trip-ends in each zone.
6. Since home-based trips constitute more than 80 per cent of the total and can be related to household characteristics, they are analysed separately from those which are non-home-based. The home-based trip generations from the expanded home-interview survey were stratified according to trip purpose and related to various categories of households. Many categories were considered before settling on house type, car ownership, and family income as the variables to be used. The generations were converted to trip rates per person, as summarized in Tables 38, 39 and 40.
7. The actual trip rates used in the survey were different for each of the 10 primary traffic zones, and those presented in the tables are the Colony averages. The trip rates for squatter households were also produced as part of the process, but since it is assumed that all squatter households will have been relocated in permanent housing by the design year (1986), they are not included as a separate category. The trip rates used in the projections were for total transport, and the public transport portion was obtained from the modal split process described later in this chapter. However, for comparative purposes, Table 41 shows the public transport trips and the trip rates per household as obtained from the home-interview survey. It should be noted that the term "trip" as used in this chapter refers to the entire journey from origin to destination regardless of the number of modes used. Therefore, the totals quoted herein for the base year do not compare with the totals derived from the records of the public transport companies. Public transport and total trip generations, and generations per household are tabulated in the appendix of this report.
8. New Territories Travel — As the home-interview survey was only conducted within the urban area, it was necessary to synthetically calculate data for the New Territories. As primary Zone 8 is the urban area most closely resembling the New Territories, having some concentrations of urban development though primarily rural, the trip generation rates obtained in this zone were used to represent conditions in the New Territories. A completely independent analysis of New Territories travel was also made based on survey information gathered there as part of this Study. A comparison of the results obtained by these two methods revealed very similar trip patterns. (Overall, the independent analysis resulted in 5 per cent more trips; but in the largest zone, Tsuen Wan, it was within 3.6 per cent.) Therefore, the simulated New Territories data based on primary Zone 8 were used for the projections, as they provide more detailed information that can be of value later.

TRIP ATTRACTION (HOME-BASED)

9. An analysis of the various attracting influences exercised by different activities throughout an area becomes quite complex due to their interaction. A person about to make a trip is influenced by many different considerations in his choice of destination. The location cf available jobs, shopping centres, schools, recreational facilities and other activities, largely determines the number of trips attracted to the various zones. Therefore, trip attraction has been developed by regression analysis, using the parameters available for each zone. The best correlation between trip attractions and planning parameters was found in the following equations:

$\begin{array}{l}\text{Attractions from Car-owning Households}\\ =0.416X_1+24.522X_2-37.036& \text{for Manual Work Trips}\\ =2.381X_3-156.276& \text{for Non-manual Work Trips}\\ =1.379X_4+0.070X_5+0.389X_6-218.954& \text{for Other Home-based Trips}\\ \text{Attractions from Non-car-owning Households}\\ =1.392X_7-663.386& \text{for Manual Work Trips}\\ =2.403X_8-1,561.602& \text{for Non-manual Work Trips}\\ =4.235X_4+2.723X_5+0.860X_6-478.547& \text{for Other Home-based Trips}\end{array}$$\begin{array}{l}\text{where:}\\ X_1=\text{Number of manual employees who are residents of households with cars}\\ X_2=\text{Attraction index (attraction rating for each zone)}\\ X_3=\text{Number of non-manual employees who are residents of households with cars}\\ X_4=\text{Cars available}\\ X_5=\text{Retail employment}\\ X_6=\text{Other employment}\\ X_7=\text{Number of manual employees who are residents of households without cars}\\ X_8=\text{Number of non-manual employees who are residents of households without cars}\end{array}$

10. To provide an additional parameter for use in this analysis, each zone was rated with an attraction index number based on the amount of activity in each zone. This parameter was found to be significant in the attraction of manual work trips. Public transport and total attractions by sectors are tabulated in the appendix along with the subjective attraction points for each zone in the base and design years.

NON-HOME-BASED TRIPS

11. Non-home-based trips, like attractions, cannot logically be equated to household characteristics. The factors that motivate this type of trip are varied and complex. Therefore, these trips were developed by regression analysis, and non-home-based origins and destinations were found to correlate with the number of total home-based trip attractions. The following equations were used:
    $$\begin{gathered} \begin{array}{l}\text{Non-home-based trip origins} \\ =\\ 0.220 \\ \text{home-based attractions}-390.368\\ \text{Non-home-based trip destinations} \\ =\\ 0.232 \\ \text{home-based attractions}-726.831\end{array} \end{gathered}$$

SCHOOL TRIPS

12. The Passenger Transport Survey Unit made a separate school survey during the school term, since the home-interview survey was conducted in the summer months when many schools were not in session. The school information obtained in the home-interview survey was therefore removed and a separate analysis was made of the school survey information.
13. The number of school trips in the base year was related to the number of resident students and the school enrolment in each zone. Public transport school trips were projected to the design year based on the resultant equations but the appropriate constants were reduced to produce a smaller proportion by public transport. Overall there was a 28.6 per cent reduction. This was done in recognition of the efforts being made to improve the accessibility of schools in the Colony. This will make it possible for a higher proportion of students to attend school in their own neighbourhood and will thus increase the proportion who walk to school. Table 42 shows the present proportion of school trips by the various modes of travel. School trip data for the base and design years are included in the appendix tables of trip generations and attractions by sector and purpose.

TRIP-END PROJECTIONS

14. The trip rates and regression equations, developed from analyses of travel data for the base year, were applied to the design-year (1986) household categories and planning parameters. This was done to obtain the trip-ends for the design year and is based on the premise that, on an average, the members of households with certain social and economic characteristics will have certain travel habits. In the future the number and distribution of house holds in any given category may change and individual families may move from one category to another but the basic relationship between travel and other characteristics remains the same for each household category.
15. Table 43 is a summary of the total design-year trip-ends according to purpose and geographic area. More detailed tables are included in the appendix.

TRANSPORT SYSTEMS ASSUMED FOR TESTING

16. Up to this point the analysis has only concerned the total number of trip-ends in and out of each zone. No consideration has been given to the mode of travel nor to the length and direction of trips. To take these into account it was necessary to relate the travel to a specific network, or networks. Three complete design-year public transport systems were devised to fill this need and to provide the basis for the study of alternatives. Each of these conceptual systems was sufficiently different from the others to provide information on a wide variety of public transport services and give a clear indication of the value of each. However, all the systems were designed to serve the same living, industrial and commercial areas, so the principal variation was in the type and level of public transport service offered.
17. Since future travel volumes were not known at the time the test systems were devised, it was necessary to anticipate the public transport service that would be needed. To arrive at the system required, hypothetical systems were designed which were above and below estimated needs. The first system tested consisted primarily of improvements and extensions to the existing surface public transport system with no rapid-transit elements. The third system included extensive rapid transit supplemented by bus, train and ferry services. The second system fell between these two, including a smaller proportion of rapid transit and relying more on surface vehicles.
18. All of these systems were converted to coded network form for traffic assignment by computer. The network speeds were based on detailed travel-time studies of existing public transport movements, modified to reflect expected street improvements and growing traffic congestion. The rapid-transit speeds were computed from station spacing and the performance characteristics of modern equipment. Walking, waiting and transfer times were based on field surveys in some instances and on an analysis of anticipated conditions in others.
19. System 1 — This system was designed to test whether the capacity requirements of future urban travel could be met, with minimum capital expenditure, by expanding the existing public transport system of buses, trams, ferries and the passenger services of the Kowloon-Canton Railway. Even this minimum-cost system would require large capital expenditures for additional vehicles and for the replacement of existing ones. The main components of this system are shown in Figure 24 and briefly described below.
20. The majority of passenger movements would continue to be accommodated by ordinary surface bus lines, expanded in coverage and capacity to meet increased future travel. To reduce the travel times of longer distance journeys, local buses would be supplemented by limited-stop or express bus lines on major thoroughfares. Many street improvements would need to be made to facilitate the movement of buses if this plan were adopted. Several bus terminals were included to provide off-street loading facilities and to facilitate transfers between express and local services.
21. Major improvements in the tramway service included relocating the portion of the line through the Central District, between the Naval Dockyard and the Western Market, into an underground tunnel. This would require re-equipping the operation with single-decked cars of either articulated or multiple-unit type. Improvements would also include extending the tram line to Chai Wan and changes in the Wan Chai area to increase the speed of tram operation. The portion of the line that now runs on Johnston Road would be relocated to Hennessy Road, to eliminate the need for trams to turn across the Hennessy Road traffic at each end.
22. The cross-harbour and outlying area ferries would continue to form vital links in the transport system and in many instances would have to be expanded to meet new demands. All existing cross-harbour ferry routes would be retained except the Hung Hom—Edinburgh Place and Hung Hom—Wan Chai lines, which it was assumed would be replaced by buses using the cross-harbour tunnel. A hovercraft ferry, operating between the Western District of Hong Kong Island and Castle Peak, was also included as part of this system, supplementing an express bus operating to Castle Peak from Kowloon.
23. The Kowloon-Canton Railway was assumed to be double-tracked to Sha Tin to provide added passenger capacity to this future New Town. Access to the existing Yau Ma Tei Station would need to be improved to provide for transfer with buses, and for better pedestrian connection with Mong Kok. A bus-railway transfer station was also included at the new Hung Hom railway station. Individually powered diesel passenger cars operating between Hung Hom and Sha Tin were envisaged as a major component of this system.
24. System 2 — This system was designed to provide a basic high-capacity rapid-transit service in the main travel corridors, including the cross-harbour movement; it also allowed for a complete network of surface bus transport for those areas and short trips not accommodated by rapid transit. This system is shown in Figure 25. The rapid transit was envisaged as predominantly an overhead system, but with underground segments at least through the Central and Western Districts on the Island, across the harbour and extending approximately one mile north in the Nathan Road corridor of Kowloon. There was rapid transit within the extended urban area, including Kwun Tong, Tsuen Wan and Sha Tin, with direct connection to express and local buses running to suburban communities.
25. The Kowloon-Canton Railway was assumed to be electrified to Sha Tin and equipped with rapid-transit type vehicles to make it an integral part of the urban system. It was tested as a double-track system from Hung Hom to Sha Tin, including the tunnel north of Kowloon. It was assumed that all goods movements would be accommodated in the late night and early morning hours to avoid interference with day-time passenger services. A new station was included in Kowloon Tong to provide direct connection with other parts of the rapid-transit system. The overall length of the rapid-transit lines, including the Kowloon-Canton Railway section, was 33 miles and there were 45 stations. A few cross-harbour ferry routes were included to supplement the cross-harbour bus and rapid-transit services. A hovercraft ferry operating between the western part of Hong Kong Island and Castle Peak was also included.
26. System 3 — This system assumed maximum development of rapid-transit routes and is shown in Figures 26 and 27. It was designed to divert as many trips as possible from surface buses and trams, as well as from automobiles and taxis, and to accommodate these trips efficiently. It would provide direct rapid-transit service to Junk Bay, Kwun Tong, Sha Tin, Tai Po, Tsuen Wan, Castle Peak and Aberdeen. It included two rapid-transit harbour crossings. The overall length of the rapid-transit system would be 70 miles and there would be 68 stations.
27. The improvements to the Kowloon-Canton Railway would be similar to those for System 2, except that an additional connection between the railway and the urban rapid-transit system would be provided at the new Hung Hom railway terminus. In this system, electrification and double tracking was extended to Tai Po.
28. There would be a network of local surface buses operating for the short trips and in areas not close to rapid-transit stations. It is doubtful whether passenger ferry service within the harbour could be sustained with this system. However, the ferry services to the outer islands would have to continue and there might still be a need for vehicular ferries to supplement the cross-harbour vehicle tunnel. A hovercraft service from Kennedy Town to Castle Peak was also included in the network.
29. Trams were not specifically included in Systems 2 and 3. However, the assignment of volumes to the many bus lines included along the north side of the Island, provides an estimate of the total movement by surface transport, regardless of the type of vehicle.

• Figure 24 — Test System One
  Figure 24 — Test System One
• Figure 25 — Test System Two
  Figure 25 — Test System Two
• Figure 26 — Test System Three
  Figure 26 — Test System Three
• Figure 27 — Test System Three - New Territories
  Figure 27 — Test System Three - New Territories

MODAL DISTRIBUTION

30. At some time every person making a vehicular trip must choose from different forms of transport and the factors which influence a person's choice need to be studied in making an analysis of modal distribution.
31. Car Ownership — Analyses of modal distributions show that the most significant variable is car ownership. The results of the analysis for Hong Kong also indicate its important effect on modal choice. This is apparent from an examination of Figure 29, which shows two distinct patterns of public transport journeys, one in car-owning, and the other in non-car-owning households. The differences are due to the fact that members of non-car-owning households have a more limited choice in methods of travel. The effect of owning a car is to increase the total number of trips per day, and to reduce the number by public transport.
32. Income and Family Composition — The other household characteristics influencing the rate of trip generation by different travel modes are household income, family size and the sex and age composition of the family. Certainly household income is important in modal distribution for it determines car ownership; but beyond this, higher incomes increase trip generation by all modes, rather than by any particular one. Likewise, the family size and its sex and age distribution affect its total number of journeys and their purpose distribution more than the modal distribution.
33. Conditions at Each Trip-End — Every trip has two ends, and conditions at each end influence the choice of mode. The most important of these conditions are:
    1. Net residential density. This influences car ownership and modal distribution. The incentive to own a car in high density residential areas is less since such areas tend to have better public transport facilities (including taxis) and less garaging space, and are better served by local shopping facilities, etc.
    2. Parking space. This obviously has a direct influence on modal choice since inability to find a parking space at the attracting end of a trip discourages the use of a car.
    3. Public transport facilities. The proximity of public transport facilities to the generating and attracting ends of the trip influences the mode chosen. Long walking distances to the nearest public transport stop will discourage its use.
34. When more than one mode is available the various possibilities "compete" for the trip. People are then influenced by the comparative quality and cost of each service. Here travel time, cost, frequency of service, number of transfers, transfer times, convenience and comfort are most important. Unlike the other characteristics all these factors vary for different zone-to-zone movements.
35. Travel time is the only one of these parameters which can be calculated objectively. Previous studies have shown, however, that it is not sufficient to use travel time alone in modal-distribution analyses. The time required for walking, waiting and transferring must be taken into account to obtain a proper measure of public transport service.

36. File:MTS Fig28.png

    Figure 28 — Base-year Public Transport Accessibility

    Accessibility Ratings — The analyses of modal distribution were made by the use of accessibility ratings. There are many ways of defining the "accessibility" of a traffic zone and any definition will be somewhat arbitrary. One could consider (a) the number of public transport routes passing through the zone; (b) the frequency of the transport services; (c) the number of stops or stations; (d) the distribution of the routes among the various places that people want to reach; and (e) the number of trip attractions within a given distance or time. The best definition is one that is simple, rational, easy to calculate and reliable. To provide this, an accessibility rating was defined for each traffic zone. These ratings reflect the travel time from each zone to all the others. Individual studies were made for each mode. Travel time included walking, waiting and transfer time to allow for public transport conditions. The general equation can be stated as follows:
    

    $$\begin{gathered} \text{Accessibility of zone} \\ z=\sum _{x=1}^{x=n}\frac{\text{employment in zone} \\ x}{\text{time from zone} \\ x \\ \text{to zone} \\ z} \end{gathered}$$
    

    The accessibility factors were then stratified and assigned numbers between one and six. A low number (rating) indicates poor accessibility and a high rating indicates good accessibility. Figure 28 shows the public transport accessibility ratings for each zone in the base year.

37. File:MTS Fig29.png

    Figure 29 — Public Transport Trips by Accessibility Rating

    Correlation was revealed between the percentage of public transport trips and the accessibility ratings. This is shown in Figure 29 and Table 44. Therefore, the modal distribution between public and private transport was made on this basis, using separate distributions for each mode and each primary zone, and for households according to car ownership.

38. Trip Distribution by Car Ownership — Table 45 is a summary, according to car ownership, of the percentage distribution of trips between the various modes. It shows that over 90 per cent of the trips from non-car-owning households are by public transport, while nearly 77 per cent of those from car-owning households are by private transport. (It should be noted that "other public transport" in this and the following table includes taxi, pak pai and public car trips.) Because of the high proportion of non-car-owning house holds, over 78 per cent of all trips are by public transport.

39. Trips by Mode and Purpose — Table 46 shows the percentage distribution of public transport trips by mode and purpose. It shows that buses are the most frequently used of all public transport modes.

40. Design-Year Modal Distribution — Two different sets of public transport trip percentages were used for modal distribution in the design year. For System 1, the base-year percentages of trips by bus, tram and ferry were used since this system is similar in character to the present one. However, the accessibility ratings of many zones were altered due to changes in travel time and employment. The percentages of bus, tram and ferry trip generations and attractions by purpose, accessibility rating and car ownership are included in the appendix.

41. File:MTS Fig30.png

    Figure 30 — Design-year Public Transport Accessibility

    Since Systems 2 and 3 contain rapid transit, a slightly different approach was used. There is, of course, no existing information available to directly determine the proportion of persons who would use rapid transit for their daily travel. However, it is known that people will use the faster of two modes if the price is the same or if the extra cost of the faster mode is considered reasonable in terms of the time saved. Therefore, the split between public and private transport travel for Systems 2 and 3 was made in the modal distribution phase of the analysis. The division of the resulting public transport travel between rapid transit and surface vehicles was made subsequently in the trip assignment process. Tables showing the percentage of public transport trip generations and attractions for two rapid-transit conditions are included in the appendix. Basically the public transport percentage for zones that are not within walk ing distance of a rapid-transit station is equivalent to the combined bus, tram and ferry percentages. The percentage for zones within walking distance of a rapid-transit station includes both these, and most "other public transport" (taxis, pak pais and public cars). However, separate sets of percentages were used for each primary zone and some subjective adjustments were made where special conditions intervened. Figure 30 shows the public transport accessibility ratings for each zone in the design year. These are the ratings developed for System 2 and used for Systems 2 and 2A which is described later.
42. The modal distribution procedure produced the following numbers of average daily public transport trips:

System 1	6,526,630
System 2	7,403,653
System 3	8,011,510

43. Design-Year Patronage on Recommended System — The recommended system, as described in Chapter 6, includes more rapid transit than System 2 but less than System 3. As it most closely resembles System 2, the figure of 7,403,653 average daily public transport trips derived from the analysis of System 2 is used for the recommended system. This conservative figure is used in the presentation of all design-year data, including the public transport trips shown in the appendix tables.

BASE AND DESIGN-YEAR TRIP COMPARISONS

44. Many comparisons can be made to show the results of projecting and distributing design-year trips. These indicate that the design-year projections are reasonable when considered in relation to Hong Kong growth trends.
45. Distribution by Trip Purpose — Table 47 and Figure 31 show a comparison, in terms of purpose, of base-year with design-year public transport trips. Although the number of trips increases for every purpose, the percentage of manual work and school trips diminishes. Other home-based trips (shopping, social, recreation, etc.) account for the majority of trips. However, combining the two work trip categories shows that work trips account for slightly over 40 per cent of total trips in both years.
46. Geographic Distribution — Table 48 shows the public transport trip-ends by geographic areas. As with the previous table, a numerical increase is shown for every area, but the percentages decrease in each of the urban areas. This is the result of a more than four-fold increase in the New Territories. These are trip-ends (generations and attractions) so the numbers are twice what they would be in a tabulation of trips.
47. Trips Per Household — The home-based trips were projected on a trip-per-person basis. Table 49 shows the results on a trip-per-household basis according to major geographic areas. The number of trips per household will increase, with the projected increase in the overall standard of living. The largest increase will be in the Kowloon portion of the urban area.
48. Distribution by Income — Figure 32 and Table 50 show the distribution of trips by travel mode and household income. The modes shown are public transport, taxi and private automobile. As would be expected the proportion of public transport trips diminishes and that of private car trips increases with rising incomes. The taxi trips stay relatively constant throughout the income range. Here, "taxi" includes pak pais and public cars.
49. Work Trips — Figure 33 and Table 51 show the relationship between the work trip generations of a household and its income. Public transport and total trips are shown. Work trips per household increase very rapidly with rising income in the low income ranges. However, the public transport portion tends to drop where the income is more than $1,500 per month. Members of households earning more than this can apparently afford to use taxis or pak pais and some can purchase motor-cycles or cars.

• Figure 31 — Public Transport Trips by Purpose
  Figure 31 — Public Transport Trips by Purpose
• Figure 32 — Trips by Mode and Income
  Figure 32 — Trips by Mode and Income
• Figure 33 — Work Trips Per Household by Income
  Figure 33 — Work Trips Per Household by Income

50. Trip-Ends Per Acre — Figures 34 and 35 show the geographic distribution of public transport trip-ends in the urban area for the base and design years. Since the sectors are of unequal size, this information is presented on a basis of trip-ends per acre. All the New Territories zones have less than 100 trips per acre in the base year. In the design year, Tsuen Wan (Zones 921 and 924) is in the 450 to 1,000 trips-per-acre category and Sha Tin (Zone 941) is in the 200 to 450 trip category. All other New Territories zones have less than 100 trips per acre in the design year.
51. Maximum and Minimum Design-year Trip Generations — Relating trip generations per household to various parameters such as income, population density and car ownership results in a wide variety of generation rates. If these rates were applied to the estimated number of design-year households, a wide variety of total design-year trips would result. This simplified method of projection can only be expected to give a rough approximation for the total trip-ends. However, such an approximation was made to test how the projection of public transport trips compared with an estimate of maximum and minimum design-year trips. It was considered that a practical minimum would result from a direct application of present trip rates, as calculated from the base-year data. This assumed that there would be no increase in the standard of living, and that all factors (except the numbers of households) affecting travel would remain static for the next 20 years. To obtain a maximum figure, the trip rates produced from a comparison including population density and household income were used. The maximum rate was found to occur in high income (over $3,000), high density (over 500 persons per acre) households, but there are only a small number of these and it is unlikely that they will predominate in the design year. Therefore, the rate (7.8 trips per household) for medium density (80 to 500 persons per acre) and an income range of $1,500 to $2,000 was used to estimate the maximum number of trip generations in the design year. The results of this comparison are shown in Figure 36 and Table 52. The projections used are conservative in that they are about half way between the maximum and minimum on Hong Kong Island and relatively closer to the minimum in Kowloon, the New Territories and the Colony as a whole.

• Figure 34 — Density of Base-year Public Transport Trip-ends
  Figure 34 — Density of Base-year Public Transport Trip-ends
• Figure 35 — Density of Design-year Public Transport Trip-ends
  Figure 35 — Density of Design-year Public Transport Trip-ends
• Figure 36 — Maximum and Minimum Design-year Trip Generations
  Figure 36 — Maximum and Minimum Design-year Trip Generations

TRIP DISTRIBUTION

52. The results of the trip-end and modal distribution analyses indicate the magnitude of the expected travel demand in the projection year; but these estimates give no details of the geographical distribution of trips in the form of zone-to-zone traffic movements necessary for a study of the alternative transportation systems.
53. Trip Distribution Model — An interactance trip distribution model was chosen for projecting zone-to-zone movements. The model relates the numbers of individual zone-to-zone movements to the numbers of trip generations and attractions in each zone, and the travel time between each pair of zones. This type of model has proved reliable for trip distribution in many other cities of diverse characteristics and is a member of the family of gravity models.
54. The form of the interactance trip distribution model, as used in this study, may be expressed mathematically by the following equation:
    解析失败 (未知函数“\begin{array}”): {\displaystyle \begin{array}{l} \qquad T_{ij} = a_j G_i A_j K_ij F ( t_{ij} ) \\ \text{where:} \\ \qquad T_{ij} \ \text{is the number of trips generated in zone "i" and attracted to zone "j";} \\ \qquad a_j \ \text{is a constant applicable to the attracting zone required to make attracted trips equal to attraction;} \\ \qquad G_j \ \text{is the total number of trips generated in zone "i";} \\ \qquad A_j \ \text{is the total number of trips attracted to zone "j";} \\ \qquad F ( t_{ij} ) \ \text{is the distribution factor (a function of the travel time between zones "i" and "j";} \\ \qquad \text{and} \ K_{ij} \ \text{is an adjustment factor which may be incorporated into the equation} \\ \qquad \qquad \text{where necessary to reflect particular circumstances} \\ \qquad \qquad \text{affecting trips between two sets of zones (such as trips} \\ \qquad \qquad \text{across a harbour or other major topographical barrier).} \\ \end{array} }
55. The distribution factor was observed to vary according to trip purposes and it also depended on traffic zone boundaries. The factors were therefore developed and the model calibrated for each trip purpose. Distribution factors were established from the zone-to-zone movements, the numbers of trip generations and attractions and zone-to-zone travel times in the base year. Calibration of the model was then completed by synthesizing the base-year travel pattern, comparing the synthesized with the measured pattern and making minor adjustments to the distribution factors to allow for the constants.
56. The interactance trip distribution model was used in projection by substituting, in the model formula for each trip purpose, the expected numbers of trip generations, trip attractions, zone-to-zone travel times and distribution factors. The result was an estimate of each zone-to-zone movement for each purpose. This estimate allows for the fact that, as travel facilities are improved over a period of years, many people will make trips farther afield.
57. The calibration of the interactance trip distribution model was a process of successive approximation which was considered complete when the model gave a pattern adequately representative of the base-year survey data.

58. Once the models agreed with the origin-destination data, the survey trips and the synthetic model trips were loaded by purpose onto a simplified network. Cross-harbour trips and corridors both on the north side of Hong Kong Island and north of Kai Tak Airport were checked. Table 53 shows the comparisons for bus trips.
59. The bus volume comparisons shown in Table 53 were considered to be satisfactory and at this point the bus model was assumed to be calibrated.
60. The tram model however was distributing tram trips throughout the mainland, so a two-minute time value was inserted in the network for all intra-mainland movements. The distribution factor for two minutes was then made equal to zero and this procedure prevented the model from distributing tram trips in Kowloon. Another run of the model showed that the calibration was not adversely affected and when the new trip tables were loaded on the network the cross-harbour volumes checked very closely with the survey volumes. This showed that the intra-mainland tram trips had been effectively suppressed.
61. The number of ferry trips was not of sufficient magnitude to warrant a separate model. It was therefore decided to combine the ferry trip-ends with the bus trip-ends and run the model using the bus distribution factors. This, as expected, resulted in an imbalance between survey trips and model trips. Application of "K" factors balanced the cross-harbour trips without adversely affecting the calibration or the other corridor volumes. The bus-ferry transportation model is made up of the bus distribution factors plus the "K" factors. These factors were only used for System 1 bus and ferry distributions, not for the trams or for Systems 2 and 3.

• Figure 37 — Bus Trip Distribution Curves
  Figure 37 — Bus Trip Distribution Curves
• Figure 38 — Tram Trip Distribution Curves
  Figure 38 — Tram Trip Distribution Curves

63. Desire Lines — Figures 39 and 40 are "desire line" drawings showing the volume of zone-to-zone public transport trips for System 2. Figure 39 shows the urban area desires on a sector basis. The large volumes in the north and north-east areas are due to very large, heavily populated sectors. Figure 40 includes the travel de sires within the New Territories and between the New Territories and the urban area.
64. As part of the travel analysis, synthetic or schematic networks were prepared to determine the location of main corridors of traffic flow. These networks are developed by joining all the zone centroids with straight lines. Because of their appearance they are called "spider web" networks. Figures 41 and 42 show the System 2 public transport volumes in spider web network form for the urban area and the New Territories.

• Figure 39 — Design-year Urban Area Travel Desires
  Figure 39 — Design-year Urban Area Travel Desires
• Figure 40 — Design-year New Territories Travel Desires
  Figure 40 — Design-year New Territories Travel Desires
• Figure 41 — Design-year Urban Area Spider Web Network
  Figure 41 — Design-year Urban Area Spider Web Network
• Figure 42 — Design-year New Territories Spider Web Network
  Figure 42 — Design-year New Territories Spider Web Network

TRAVEL ASSIGNMENTS

65. The process of travel assignment consists of loading, by computer, the trips from the trip-distribution phase on the transport networks. The networks consist of a numerical description of the routes in the form of walking, waiting and travel times on the various segments or links. The individual zone-to-zone movements were assigned along the shortest time paths and were accumulated to produce the volume on each segment.
66. Public transport networks were prepared for the base year and for Systems 1, 2 and 3 in the design year, and trips were assigned to each. An initial examination of the results revealed that the base-year assigned volumes were higher than the observed volumes at all seven check points and their average was too high by approximately 8 per cent.
67. A complete review of all procedures was made to determine the reason for the high volumes in the base-year assignment. It was concluded that the numbers of trips were correct but the trip length was too high. This made all the trips too long and therefore resulted in high volumes throughout the network. To correct this, all the travel times were increased on the design-year networks and the trips were reassigned. The urban area travel times were increased by 6 per cent and those in the New Territories by 20 per cent, since the investigation revealed that the New Territories trip lengths in the design-year assignments were abnormally long. An investigation of base-year cross-harbour trips revealed that there is no measurable psychological restraint to cross-harbour travel at present. The number of trips across the harbour is commensurate with the very long time that it takes to make the crossing. However, it must be assumed that a psychological restraint will develop when improved traffic facilities shorten the crossing time. As this sort of restraint is found in most urban areas divided by harbours, rivers or other severe topographic restrictions, a factor of 0.87 was applied to all cross-harbour links.
68. A test system designated as "2A" was developed by adding to System 2 a rail rapid-transit line on the east side of Kowloon. System 2 travel volumes were assigned to this network in order to test an additional "system" between 2 and 3.

59. No changes were made in System 1 and no additional assignments were made to this system. Table 54 shows a comparison between the results of the original assignments and those of the adjusted assignments for System 2A. The principal difference is that the average trip length has been reduced and the numbers in nearly every category decrease accordingly. It should be noted that this information applies to the entire Colony and the trip length is therefore influenced by the many inter-urban trips expected in the New Territories.
60. System Comparisons — Table 55 is a comparison of the statistical data relating to the four systems tested. It can be seen by examining this tabulation that better public transport service increases its use. However, it also should be noted that System 3 represents a public transport service much superior to that of System 1, and only attracts about 13 per cent more of the total trips. The 72.5 per cent of public transport trips in System 3 is slightly less than the present percentage. Rapid-transit trips account for between 31 and 41 per cent of the total. This figure varies between 20 and 50 per cent in the rapid-transit systems of the world and depends, of course, on the extent of rapid-transit services offered in relation to other transport facilities in use.
61. The relatively high average trip length shown for System 1 is due to the fact that this represents the original assignment. If System 1 had been adjusted this figure would be somewhat reduced. A comparison between the average trip time and average riding time shows that approximately one half of the public transport travel time is spent in walking and waiting. The relatively high average speed on ferries is due to the inclusion in all systems of a hovercraft ferry between Hong Kong Island and Castle Peak.
62. Urban Area Comparisons — Table 56 shows a comparison between base and design-year travel within the urban area. The design-year System 2 travel statistics relate to the original assignments, since no re-assessment was made on an urban area basis. A comparison between the trip length information in this table and that contained in Table 55 shows a marked difference between the length of travel in the urban area and that in the Colony as a whole.
63. Traffic Flow Maps — Figures 43 to 48 show the design-year public transport volumes in the form of person trips per day. Although some manual adjustments were made to the volumes obtained from the computer to take account of network peculiarities, there has been no attempt to fit the volumes to the capacity of various roadways. The volumes in the New Territories are similar for Systems 1, 2 and 2A, so Figure 46 serves as a traffic flow diagram for all three systems. The implications of the traffic volumes shown in these figures are discussed in Chapter 6.
64. Fare Differential Assignment — The distribution and assignment procedure described split public transport trips into surface transport and rapid transit. This division was based on the relative travel times for each zone-to-zone movement. This assumes that the cost is the same on both modes or that the cost differential would be so slight as to have no effect on the travel patterns. It is felt that this is the correct approach for system planning purposes. However, for feasibility considerations it is necessary to ascertain what inhibiting effect higher cost on rapid transit will have on travel by that mode. Therefore a separate assignment was made to System 2 with increased time on the rapid-transit lines to represent the effect of this increased cost. The time increase was developed from assuming a value of time equal to $2.00 per hour. It was assumed that the effect on traffic that would result from other cost differences could be interpolated from the results of the "equal fare" and "fare differential" assignments. Seven-tenths of a minute was added to each waiting link and 0.4 minutes per mile was added to all travel links on the rapid-transit system. This caused the movements of one mile or less to be reduced by 100 per cent, two miles by 48 per cent, three miles by 28 per cent, ranging down to a 5 per cent reduction for trips of eight or more miles.
65. The overall effect of the fare differential can be seen in Table 57, where the total rapid-transit trips are 24.2 per cent of 1 public transport. This may be compared with the "equal fare" assignment, which had 32.5 per cent rapid-transit trips.

• Figure 43 — System 1 Traffic Volumes
  Figure 43 — System 1 Traffic Volumes
• Figure 44 — System 2 Traffic Volumes
  Figure 44 — System 2 Traffic Volumes
• Figure 45 — System 2A Traffic Volumes
  Figure 45 — System 2A Traffic Volumes
• Figure 46 — System 2A New Territories Traffic Volumes
  Figure 46 — System 2A New Territories Traffic Volumes
• Figure 47 — System 3 Traffic Volumes
  Figure 47 — System 3 Traffic Volumes
• Figure 48 — System 3 New Territories Traffic Volumes
  Figure 48 — System 3 New Territories Traffic Volumes