Decongest the Railway Terminus
No P&L, no client revenue — structure a public-operations problem.
The Prompt
A major metro-city railway terminus sees dangerous platform crowding between 7:30–10:00 am. The divisional railway manager — your client — cannot add platforms or land for at least five years. Reduce peak crowding meaningfully within 12 months, with a modest budget.
Opening exchange
Crowding is density: people on platforms ÷ platform area over time. With area fixed, I can reduce the people, the time they spend standing there, or smooth the distribution across platforms and minutes. Quick facts: what share of platform occupants are passengers versus accompaniers/vendors? And how early do passengers arrive before their train?
Unconventional cases reward defining the governing quantity (here: person-minutes per square metre) before any solutions.
Surveys say: 70% passengers, 18% people seeing someone off, 12% vendors and staff. Long-distance passengers arrive 45–70 minutes early on average. Suburban passengers arrive ~8 minutes early. Both share the same concourse and some platforms.
Two findings jump out: a fifth of the crowd isn't travelling at all, and long-distance passengers spend nearly an hour occupying platform space they only need for five minutes of boarding. The structure: reduce non-travellers, shift waiting off-platform, and stagger/redistribute flows. Capacity expansion is off-limits — good; this is a flow problem, not a capacity problem.
Structure & Hypothesis
Analysis & Data
Estimate the impact of the airport-gate model alone.
Say 12 long-distance departures in the peak band, ~1,200 passengers each, arriving ~55 minutes early on average. Today that's 12 × 1,200 × 55 ≈ 7.9 lakh passenger-minutes on platforms. Holding them in a concourse hall until 15 minutes pre-departure cuts platform dwell to ~15 minutes: 12 × 1,200 × 15 ≈ 2.2 lakh. That removes ~5.8 lakh passenger-minutes — roughly 70% of long-distance platform load — using waiting halls that already exist but are unattractive and unenforced.
Quantifies with rough numbers and states the operational requirement (enforcement + displays), not just the idea.
What breaks this in practice?
Three things: passengers don't trust announcements, so they camp at platforms — fix with reliable coach-position displays and SMS gates; the halls are unpleasant — fix with fans, seating, and vendor stalls (which also relocates vendors off platforms); and enforcement at platform entry needs gating staff for the first months until behaviour settles. None requires capex beyond a few crore.
Recommendation
Recommend to the DRM
- Phase 1 (months 1–3): airport-gate waiting for long-distance trains + live coach displays — the single biggest density lever, near-zero capex.
- Phase 2 (months 2–6): peak-hour platform-ticket suspension and vendor restocking windows moved off-peak — removes ~25% of non-traveller bodies.
- Phase 3 (months 6–12): re-time 3–4 long-distance departures out of the 8–9 am band and rebalance suburban platform assignments.
- Measure with a density dashboard (CCTV people-counting per platform per 5 minutes) — set the target as person-minutes/m², not anecdotes.
Key Takeaway
What this case teaches
Unconventional cases want you to define the governing metric (here, person-minutes per square metre) and decompose it like a P&L. Most "capacity" crises are flow problems — the system has slack, just at the wrong place or time.