Question 2 of 5: An ITS Programme for Urban Arterial Signal Systems

Question 2: An ITS Programme for Urban Arterial Signal Systems (20 points)

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.

The premise of the question is sound and worth restating before answering it. On an urban arterial network the signals, not the links, are where the capacity is set and where the delay is incurred: a through movement spends most of its journey time stopped or decelerating at intersections, and the variance of that journey time is dominated by whether the vehicle happens to arrive on green. It follows that the cheapest available improvement to an urban network is almost always better signal operation rather than more pavement, and that a signal system which is designed, deployed and then left unmanaged will decay: turning movements grow, land uses change, and timings that were optimal at commissioning are demonstrably worse than a fresh plan within a few years. Retiming programmes reported in the literature routinely return travel-time reductions of the order of five to fifteen per cent with benefit–cost ratios of an order of ten to one or better, which is a return no capital project on the same corridor can approach. The programme set out below is therefore organised around the three objectives the question names, and it treats operations and measurement as first-class deliverables rather than as things done after the equipment is installed.

The technology set, layer by layer

Field layer. Every signalised approach needs detection, and the choice of detector should follow the function it serves. Stop-bar detection (inductive loops, video image processing or radar) supports phase actuation, queue estimation and the detection of split failures. Advance detection placed in the dilemma zone of a high-speed approach, typically radar so that it can track a vehicle rather than merely occupy a zone, supports safe phase termination. Mid-block or system detection supports plan selection and provides the volume series that drives the whole programme. The controllers themselves must be modern advanced transportation controllers running NTCIP 1202, capable of high-resolution event logging at ten samples per second, because that log is what makes performance measurement possible without any additional field equipment. Pan-tilt-zoom cameras give operators verification, not data. Pedestrian push-buttons, countdown heads and accessible pedestrian signals are field devices in exactly the same sense and belong in the same procurement. Finally, receivers for transit signal priority and emergency vehicle pre-emption are installed at the intersections on the priority routes.

Communications layer. A fibre-optic backbone along the arterials with Ethernet drops at each cabinet is the right long-term investment for a corridor of any size: it removes the recurring cost of leased circuits, carries video as well as control data, and has enough headroom for whatever is added next. Cellular service is the appropriate fallback for isolated intersections. A small number of 5.9 GHz roadside units at the priority and safety-critical intersections provides the connected-vehicle path. Whatever the medium, the requirement that matters is standards conformance: specifying NTCIP 1201 and 1202 for the signal system and TMDD for any centre-to-centre link is what prevents the municipality from being locked to one vendor's controller for the next twenty years.

Central layer. The central signal system holds the timing plans and runs them by time of day, by traffic-responsive plan selection, or adaptively. Alongside it, and at least as important, sits an automated traffic signal performance measures dashboard that consumes the high-resolution controller event log and reports arrivals on green, Purdue coordination diagrams, split failures, pedestrian delay and approach volume without any manual data collection. A travel-time engine fed by Bluetooth or Wi-Fi address re-identification, or by a purchased probe-data feed, supplies the end-to-end measure that the controller logs cannot.

How the technologies serve the three stated objectives

Flow, travel time and reliability. Flow and travel time are improved by coordination: a well-set cycle length, offsets tuned to the platoon travel time between intersections, and splits proportioned to demand. Traffic-responsive plan selection adds the ability to change plan when demand departs from its time-of-day pattern, and adaptive control adds cycle-by-cycle adjustment. Reliability, however, is a different quantity from travel time and is improved by different means: it is the variance rather than the mean that the traveller experiences as unreliability, and the variance is driven by incidents, by demand surges and by the occasional cycle in which a queue fails to clear. Reliability therefore comes from incident detection and rapid response, from adaptive control's tolerance of demand that departs from the plan, and from eliminating split failures. It must be measured as a distribution, most simply through the buffer index

where TT₉₅ is the 95th-percentile travel time over the corridor and the bar denotes the mean. A programme that lowers mean travel time while leaving the buffer index untouched has not delivered reliability, and should be reported as such.

Throughput. Throughput is protected by making sure that green time is actually used by vehicles and that queues do not block upstream intersections. The measures are detector-driven phase termination so that a phase gaps out when its queue has cleared rather than running to a fixed maximum; phase omission for movements with no demand; protected-permissive left-turn operation where the opposing volume allows it, so that left turns are served in gaps rather than only in a dedicated phase; queue detection with spillback protection at critical approaches; and gateway metering to hold arriving demand outside a saturated core rather than allowing it to lock the core's intersections. The direct measure is the proportion of cycles in which an approach experiences a split failure, together with throughput expressed in vehicles served per hour of green.

Safety. The safety measures are largely independent of the mobility ones and must be funded separately or they will be traded away. Advance dilemma-zone detection on approaches with speeds above about 60 km/h allows the controller to avoid terminating a phase while a vehicle is in the zone where it can neither stop comfortably nor clear the intersection, which is the mechanism behind a large share of right-angle crashes. Conversion from permissive to protected left-turn phasing, decided on measured conflict and crash data rather than on complaint volume, addresses the left-turn crash type. On the pedestrian side, countdown heads, leading pedestrian intervals, accessible signals and adequate clearance intervals computed with a walking speed appropriate to the population are the standard measures, and the MUTCDC governs the devices in Canada. Red-light-running monitoring and video conflict analytics let the programme find the problem intersections before the crash record does, which matters because crash records at a single intersection take years to become statistically meaningful.

The implementation plan

The plan follows the systems-engineering process, and it does so for a practical reason rather than a bureaucratic one: ITS projects fail far more often on integration and on unstated expectations than on technology, and the V-model exists to force both to be written down while they are still cheap to change.

The first phase produces a concept of operations. It records who will operate the system, at what hours, with what staff, and what each stakeholder expects it to do; it confirms conformity with the regional ITS architecture; and, critically, it establishes a baseline by measuring corridor travel time, delay and the crash record before anything is installed. A programme that omits the baseline can never demonstrate its own benefit and will lose its operating budget at the first review. The second phase converts the concept of operations into requirements and then into a design: detector type by approach and by function, the communications route and its civil works, the controller upgrade list, the central-system specification written against performance rather than against a product, and the interface specifications naming NTCIP and TMDD conformance.

The third phase deploys a single pilot corridor end to end — fibre, controllers, detection, central system — and retimes it. A pilot is not timidity; it is how the municipality discovers what its cabinets, its power supplies and its staff are really like before it has committed the whole budget, and a measured before-and-after result on one corridor is the most persuasive document available when the next phase goes to council. The fourth phase extends the deployment network-wide and layers on the functions that require the network to exist: adaptive control where demand is genuinely variable, and transit signal priority on the highest-ridership routes. The fifth phase is not a phase at all but a permanent state: operate, maintain and monitor. It requires standing staff, a maintenance contract with defined response times, an annual performance report drawn from the dashboard, and a signal retiming cycle of two to three years. Verification at the end of each phase asks whether the system meets the requirements written for it; validation asks the harder question of whether it delivers what the concept of operations promised.

Measurement and management

The programme's value is defended with numbers, and the dashboard supplies almost all of them at no marginal cost. Against the flow objective: mean corridor travel time, delay per vehicle, number of stops, and arrivals on green,

expressed as a percentage for each coordinated approach. Against reliability: the buffer index and the 95th-percentile travel time, reported as a distribution rather than a mean. Against throughput: split failures per hour and vehicles served per hour of green. Against safety: red-light-running rate, pedestrian delay and conflict rates from video analytics in the short run, and the crash record in the long run. Reporting these annually, against the pre-deployment baseline, is what converts a capital project into a managed system — which is precisely what the premise of the question asserts is the determinant of network efficiency.