Transportation Engineering — Good-to-Know Rules of Thumb

Transportation Engineering — Good-to-Know Rules of Thumb

Practitioner judgment for transportation engineers and planners. Heuristics, not standards — verify against the governing manual.

Traffic flow and capacity

• Capacity per lane: ~1,900–2,200 pc/hr/ln for freeways, ~1,600–1,800 veh/hr/ln for signalized approaches at 100% green. If your analysis shows 3,000 veh/hr through one lane, the analysis is wrong.
• The 85th percentile rules: speed limits, like designs, follow measured driver behavior — the 85th-percentile speed is the starting point for every speed decision.
• Delay is nonlinear: at v/c = 0.9, small volume changes swing delay wildly; at v/c = 0.6, nobody notices. Design for the knee of the curve, not the average.
• Queues are storage: a turn bay that overflows poisons the through movement — size storage for the 95th-percentile queue, not the average. Unserved queue is the difference between a model and a mess.
• Induced demand is real: added lane capacity fills to roughly the prior congestion level within years (the fundamental law of road congestion, elasticity ≈ 1.0 long-run). Don't promise permanent decongestion from widening.

Signals

• Cycle length: 60–90 s for typical urban signals; 120 s max before pedestrian delay and driver frustration dominate. Webster's optimum is the starting point, coordination is the constraint.
• Lost time: ~4 s per phase (start-up + clearance) — the invisible tax on every cycle. Fewer phases = less lost time = more capacity, which is why split phasing is a capacity killer.
• Progression beats optimization: a well-coordinated 90-s cycle outperforms an "optimal" isolated 60-s cycle on every arterial that matters. Bandwidth is the arterial's currency.
• Detection: video/radar detection fails in fog, snow, and low sun — have a max-recall fallback plan for actuated signals. The signal that goes to flash at the first snowstorm wasn't designed for its climate.
• Pedestrian timing: 3.5 ft/s walking speed for clearance; slower near senior centers and schools. A pedestrian phase nobody can complete is a liability, not a feature.

Safety

• The HSM is a prediction, not a verdict: SPF × CMFs gives the expected crashes for average conditions — your site's geometry, behavior, and enforcement move the real number. Calibrate locally.
• Regression to the mean: high-crash locations were partly unlucky — the Empirical Bayes method exists because naive before-after studies credit treatments for luck. Always EB-adjust hot-spot analyses.
• Roundabouts: the single highest-ROI safety treatment for intersections — injury crashes typically drop 40–80% on signal-to-roundabout conversions. The opposition is political, not technical.
• Road diets: 4-to-3 conversions (two through lanes + center turn lane) typically cut crashes 20–40% with minimal capacity loss on streets under ~20,000 ADT. The most underused treatment in the profession.
• Lighting: nighttime crash rates are 2–3× daytime — lighting CMFs (~0.7–0.9) make illumination one of the cheapest safety investments per crash avoided.
• Speed kills nonlinearly: pedestrian fatality risk is ~10% at 20 mph, ~50% at 30 mph, ~90%+ at 40 mph. Every 5 mph of speed management is a life-safety intervention.

Geometric design

• Sight distance governs: SSD is the minimum design control — if the alignment can't provide it, nothing else about the design matters. Check sight distance before aesthetics.
• Consistency: a single sharp curve in a high-speed corridor kills — design consistency (operating speed vs. design speed alignment) matters more than any individual element's standard compliance.
• Superelevation: e + f = V²/15R — runout and runoff lengths matter as much as the rate; an abrupt superelevation transition is a rollover waiting for rain.
• Clear zones: 30 ft is the ideal, rarely the reality — when you can't get the clear zone, shield it (guardrail) or accept the documented design exception. Undocumented noncompliance is negligence.
• Complete streets: the street that only serves cars serves only one trip purpose badly — multimodal design is the default now, not the amenity.

Planning

• The model is not the forecast: travel demand models encode assumptions about land use, fuel price, and behavior — change the assumptions and the "forecast" changes. Present ranges, not points.
• Fiscal constraint disciplines: the fundable plan beats the visionary plan — every unconstrained project list is a wish list. Learn the revenue before the projects.
• Frequency is freedom: for transit, frequency improvements beat coverage expansions for ridership per dollar, almost always. A bus every 10 minutes that goes somewhere useful outperforms a bus every 60 minutes that goes everywhere.
• Parking is policy: free abundant parking is a subsidy for driving — unbundled, priced, right-sized parking is the cheapest TDM measure available.
• Engagement early: the public accepts what it helped shape and litigates what it first meets at the hearing. Consult at the options stage, not the preferred-alternative stage.

Construction and operations

• Work zones: the taper is the treatment — correct taper length and device spacing prevent more crashes than any amount of signage verbiage. Protect workers with positive protection where speeds are high.
• Incident management: every minute of lane blockage adds ~4–5 minutes of congestion (the 4:1 rule) — quick clearance (tow contracts, push bumpers) is the highest-ROI operations investment.
• Signal retiming: the cheapest capacity project there is — retiming a corridor every 3–5 years typically buys 5–15% delay reduction for the cost of a study. Stale timing is invisible congestion.
• Data before concrete: count, measure, and analyze before proposing construction — the study that kills a bad project is worth more than the design that builds it.

Professional judgment

• Standards are floors: the MUTCD, Green Book, and HCM set minimums — engineering judgment builds the design. "It meets the standard" is necessary, not sufficient.
• Document the judgment: every design exception, every warrant decision, every deviation — write down why. The file is your defense and your successor's guide.
• Multimodal by default: the question is never "cars or bikes" — it's how the street serves all users safely. Single-mode thinking is how agencies get sued.
• Safety over delay: when operations and safety conflict, safety wins — the traveling public forgives delay; it doesn't forgive fatalities.
• Know your lane (geographically): California's VMT framework, the UK's WebTAG, the MUTCD's warrants — practice is jurisdictional. The credential travels; the standards don't.

File: transportation/good-to-know-rules-of-thumb.md — heuristics for study and practice; always verify against the governing manual.

Original ETS reference writing, verified against public sources on 2026-09-30. Thin sourcing is flagged inline. Not legal advice — confirm current editions with the issuing body.