Question 1 of 7: Land Use, Trip Production and Travel-Demand Management

Question 1: Land Use, Trip Production and Travel-Demand Management (20 marks)

Question text not reproduced: the examination questions are © Engineers and Geoscientists BC. Open the official past paper (linked at the top of this page) to read the question, then follow the worked solution below.

(a) The residential land development – transit use feedback cycle

Residential land development and transit use are not related by a one-way cause and effect; they form a closed feedback loop that operates over a much longer time scale than the transport model's forecast year. The loop runs in both directions, and a complete answer traces each direction and then names the equilibrium the two directions produce.

Land development drives transit use. Transit is a scheduled, fixed-route service, so its unit cost falls sharply as the number of riders boarding per route-kilometre rises. The variable that controls that number is residential density within the walk catchment of a stop — conventionally 400 m to a bus stop and 800 m to a rapid-transit station. A single-family subdivision at 15 dwelling units per hectare with curvilinear streets and cul-de-sacs places very few dwellings inside that catchment and forces circuitous walking paths, so ridership per route-kilometre is low, headways must be long to keep the operating subsidy tolerable, and the long headway itself suppresses ridership further. The same corridor developed at 60 to 100 units per hectare on a connected grid supports frequent all-day service. Density is not the only lever: land-use mix matters, because a corridor with housing at one end and employment at the other produces a peaked, unidirectional, poorly balanced load that is expensive to serve, whereas mixed use generates two-way and off-peak demand on the same vehicle-hours. Tenure and dwelling type matter too, because apartment households own fewer vehicles, and parking supply matters most of all: abundant free residential and workplace parking is the single strongest predictor of auto use at a given density.

Transit use drives land development. Running the loop the other way, accessibility created by transit is capitalised into land value. A station raises the accessibility of its parcel to the region's jobs and services, developers bid more for that parcel, and the higher land price only pencils out at higher floor-space ratios — so the market itself pushes density up around stations. This is the mechanism behind transit-oriented development, and it is why Canadian regions such as Metro Vancouver (the Frequent Transit Network and its Urban Centres) and the Region of Waterloo (the ION LRT) deliberately pair rapid-transit investment with zoning that permits the density the investment makes viable.

The equilibrium and the policy consequence. Because the loop is self-reinforcing, it has two stable outcomes. A low-density, auto-oriented region settles into an equilibrium of dispersed origins, poor transit productivity, high auto ownership and further dispersal; a transit-supportive region settles into an equilibrium of concentrated origins, high service frequency, low auto ownership and further intensification. The practical consequence for the planner is that transit investment made without matching land-use policy under-performs its forecast — the classic failure mode is rapid transit built to a suburb whose zoning still mandates low density and generous parking. Conversely, the four-step model's usual assumption that land use is a fixed input is only defensible for short-horizon forecasts; over a 20- to 30-year horizon land use is an output of the transport system, which is why integrated land-use / transport models (and, at minimum, an explicit land-use scenario) are needed for long-range planning.

(b) Factors that increase trip production, by level of aggregation

Trip production is estimated at whichever level the data support, and the factor that matters differs at each level because each level aggregates away the variation below it.

(i) Zonal level — the number of households in the zone. A zonal production model regresses total trips produced by a zone on zonal totals, so the dominant variable is simply the size of the zone's population base. Doubling the number of households in a traffic analysis zone roughly doubles its trip production, because trips are generated by households and the zone is nothing more than their sum. Zonal employment or total zonal income work the same way. The reason this is the right variable at this level is also its weakness: a zonal model explains variation between zones but conceals the within-zone composition that actually causes the trips, so it is vulnerable to the ecological fallacy and to changes in average household size over the forecast horizon.

(ii) Household level — vehicle availability (auto ownership). Holding household size constant, each additional vehicle available to a household adds roughly two trips per day in the data of Question 3 (for instance the three-person row runs 7.4, 9.2 and 11.2 trips per household at 0, 1 and 2+ vehicles). A vehicle removes the scheduling and capacity constraint on discretionary travel: shopping, personal-business and social trips that would otherwise be chained onto someone else's trip, or forgone, become separately feasible. Household size is the other classic household-level variable and acts through the same logic — more people means more independent activity agendas — which is why cross-classification models use the two jointly.

(iii) Person level — employment status (being a worker, or a licensed driver). A person-level model predicts trips made by an individual, and the single strongest discriminator is whether that person has a mandatory out-of-home activity. An employed adult makes at least one obligatory home-based work tour per weekday, and the work tour then anchors chained non-work stops, so employed persons produce materially more trips than the non-employed. Driver's-licence holding, life-cycle stage and personal income act at the same level. Person-level factors matter because two households of identical size and vehicle ownership can produce very different travel if one contains two workers and the other two retirees — exactly the variation the zonal model cannot see.

(c) Travel-demand management strategies that raise average vehicle occupancy

Average vehicle occupancy (AVO) is person-trips divided by vehicle-trips. Raising AVO in the commuter peak is attractive because it reduces vehicle-trips without reducing person-trips — the demand is served, but with fewer vehicles on the network. Five strategies, with their effect on travel patterns, travel time and fuel consumption:

1. High-occupancy-vehicle (HOV) and HOT lanes. Reserving a freeway lane for vehicles carrying two or more (or three or more) occupants gives carpools a travel-time advantage that is largest exactly when congestion is worst. Pattern: some solo drivers shift to carpools; a smaller number shift departure time or route to avoid the lost general-purpose lane. Travel time: falls substantially for HOV users, may rise slightly in the general-purpose lanes at first, and falls overall once mode shift matures. Fuel: falls per person-kilometre because the same fuel now moves two or three people, and falls again because HOV lanes operate closer to the 60–90 km/h band where fuel consumption per vehicle-kilometre is minimised, instead of in stop-and-go flow.

2. Employer-based ridesharing and vanpool programs with a guaranteed ride home. Matching services, preferential carpool parking and a taxi voucher for emergencies remove the two real barriers to carpooling — finding a partner and being stranded. Pattern: trips are consolidated at the origin end; some short access trips are added as carpoolers detour to collect partners or drive to a park-and-ride lot. Travel time: door-to-door time rises modestly for the individual carpooler (pick-up detours, less schedule freedom) but falls for everyone on the corridor as volume drops. Fuel: total fuel falls, though not in proportion to the vehicle-trip reduction, because the detour kilometres and the cold-start emissions of the collection trips partly offset the saving.

3. Parking pricing and cash-out of employer-paid parking. Charging the market price for peak-period workplace parking, or offering employees the cash value of a free stall if they do not use it, converts a hidden subsidy to solo driving into a visible price. This is generally the most cost-effective single TDM measure available. Pattern: solo driving falls; carpooling, transit and active modes rise; a small amount of spillover parking appears on nearby streets unless it is managed. Travel time: corridor travel times fall as volumes drop, and the search-for-parking component of urban travel time falls sharply. Fuel: falls both from the removed vehicle-trips and from the elimination of cruising-for-parking kilometres, which can be a significant share of downtown traffic.

4. Congestion pricing (peak-period road or cordon tolls). A toll set at the marginal congestion cost prices the externality that Question 6 quantifies. Pattern: some trips shift to carpool or transit, some to the shoulder of the peak, some to untolled routes, and a few are forgone. Travel time: falls on the priced facility, with the system moving toward the system-optimal assignment rather than the user equilibrium. Fuel: falls, with an extra gain because vehicles operate in free-flow rather than in the accelerate–brake cycles that dominate congested fuel consumption.

5. Compressed work weeks, staggered hours and telework. These do not raise AVO directly — they cut vehicle-trips instead — but they are normally packaged with the measures above and they raise the effective occupancy of the peak-hour road system. Pattern: the peak is flattened and spread; some peak trips disappear entirely. Travel time: falls for everyone still travelling, because the peak-hour demand-to-capacity ratio drops. Fuel: falls roughly in proportion to the removed vehicle-kilometres, plus a further gain from smoother flow.

Two cautions complete the answer. First, all five measures release capacity, and released capacity attracts latent demand — the induced-travel effect — so the long-run travel-time saving is smaller than the short-run saving unless the released capacity is locked up by pricing or reallocated to transit and active modes. Second, pricing measures are regressive unless revenues are recycled into transit service or income-tested rebates, which is a standard condition of Canadian congestion-pricing proposals.