Inside the Numbers: How Ottawa’s $7 Billion LRT — and a Black-Box Data Method — Left Riders Waiting 10 Minutes Longer

At 8:17 a.m. on a February morning, the platform at Tunney’s Pasture felt longer than usual. Commuters stared down the tunnel, phones glowing with transit apps that promised a train “arriving now.” Ten minutes passed. Then another. By the time the eastbound finally rolled in, the crowd had thickened into the familiar Ottawa winter posture: shoulders hunched, eyes narrowed, patience gone.

That lost time — roughly ten extra minutes per trip during peak periods — sits at the centre of a growing controversy over Ottawa’s $7‑billion light-rail project. The problem isn’t just steel wheels and frozen switches. It’s numbers. How the city measures wait times. How it models service. And how a largely opaque data method quietly reshaped the experience of tens of thousands of riders without a clear public explanation.

The $7 Billion Promise Meets the Spreadsheet

Ottawa’s LRT, branded O‑Train Line 1, was sold as a generational upgrade: faster trips, shorter waits, and a modern backbone for a growing capital. Stage 1 opened in September 2019 at a cost of about $2.1 billion. Stage 2 — east, west, and south extensions — pushed the total public price tag to roughly $6.6–$7 billion, according to city budget documents approved in 2023.

The promise rested on frequency. Pre‑LRT bus routes on the Transitway ran as often as every 90 seconds downtown during peak hours. The Confederation Line was designed for peak headways of 3–4 minutes, scaling back off‑peak.

Then came derailments, prolonged shutdowns, a months‑long closure in 2023, and a quiet recalibration of how “wait time” gets counted.

Internal OC Transpo performance reports reviewed by council committees in late 2024 showed average peak waits closer to 7–8 minutes across large stretches of the line. Riders experiencing platform congestion reported effective waits of 10 minutes or more when trains were short‑turned or held to “re‑space service.”

The gap between official averages and lived experience widened — and the math behind those averages stayed mostly hidden.

The Black Box: How Wait Times Get Smoothed

At the heart of the issue sits a methodological choice that transit agencies rarely advertise. Ottawa relies on automated passenger counters (APCs), train telemetry, and scheduling software to calculate “average platform wait time.” Those systems don’t measure how long you personally wait. They model it.

OC Transpo uses industry-standard scheduling and analytics tools — platforms similar to Trapeze FX, Clever Devices’ CleverAnalytics, and real-time operations software that ingests vehicle location data, door events, and historical ridership. The exact configuration remains confidential, but city staff have described a process that:

• Aggregates train arrival data over multi-hour windows
• Applies smoothing algorithms to remove “outliers” like stalled trains
• Weights waits by modeled passenger volumes rather than observed crowding

In practice, this means a 15‑minute gap caused by a disabled train can be mathematically diluted if the system assumes fewer riders during that interval — even when the platform tells a different story.

The city’s Auditor General, Nathalie Gougeon, flagged a related concern in her 2024 report on OC Transpo reliability, warning that performance metrics “may not fully reflect customer experience during service disruptions.” The language was diplomatic. The implication wasn’t.

This is where the tech angle sharpens. These systems increasingly rely on machine‑learning models trained on historical patterns — pre‑pandemic ridership, pre‑shutdown service, pre‑everything. When the world changes, models lag. Unless they’re retrained with transparent assumptions, they quietly normalize worse service.

When AI Optimizes for the Wrong Thing

Modern transit scheduling software optimizes for efficiency: fleet utilization, operator hours, on‑time departure from terminals. Rider waiting sits downstream, inferred rather than observed.

An OC Transpo planner, speaking at a 2024 Transportation Committee meeting, acknowledged that “headway management” now takes priority over strict timetables. Translation: trains get held to maintain even spacing, even if passengers wait longer in the moment.

That approach works when infrastructure behaves. Ottawa’s hasn’t.

Alstom’s Citadis Spirit vehicles — the same fleet linked to multiple bearing failures and derailments — impose operational constraints that algorithms struggle to price in. Short trains reduce capacity. Speed restrictions elongate trips. Recovery margins eat into frequency.

AI systems respond rationally: they smooth. They average. They make the data look better than the platform feels.

The result: official dashboards show incremental improvement while riders experience a flatline — or worse.

Riders Notice. Officials Deflect.

Public reaction has hardened. A fall 2025 survey by the Ottawa Transit Riders advocacy group found:

• 62% of frequent LRT users felt wait times had increased compared to pre‑2023
• 48% reported allowing an extra 10–15 minutes for trips involving Line 1
• Only 14% trusted OC Transpo’s published reliability metrics

At City Hall, councillors have grown sharper. During a November 2025 council meeting, Bay Ward councillor Theresa Kavanagh questioned whether “averaging away bad service” amounted to misleading the public. Transit chair Glen Gower pushed back, arguing that metrics follow “industry best practices.”

Best practice, however, isn’t the same as best outcome.

Private-sector transit agencies in Europe increasingly publish raw headway distributions — not just averages — showing the percentage of waits exceeding specific thresholds. Ottawa doesn’t.

The Public Spending Question No One Wants to Answer

Seven billion dollars buys expectations. Faster commutes. Predictability. Transparency.

Instead, Ottawa riders received a system where the definition of “on time” shifted after the fact. Capital spending debates focused on steel, concrete, and contracts. Operating performance drifted into spreadsheets few residents ever see.

The controversy echoes earlier missteps. In 2022, the city acknowledged overstated bus reliability figures after changing how it counted canceled trips. The lesson didn’t stick.

Every minute added to a commute carries an economic cost. Transport Canada estimates the value of commuter time in urban Canada at roughly $18–$22 per hour. Multiply ten lost minutes by tens of thousands of daily riders and Ottawa quietly burns millions in productivity each year — unaccounted for in LRT business cases.

What Transparency Would Actually Look Like

Fixing the numbers won’t fix the trains. But it would restore trust.

Concrete steps the city could take immediately:

• Publish raw headway data by station and time of day, including the 90th percentile wait
• Separate disruption metrics from normal operations instead of smoothing them together
• Release model assumptions used in scheduling and performance dashboards
• Commission an independent data audit alongside the technical LRT inquiry

Cities like London and Stockholm already do versions of this. Ottawa could, too.

Tools Riders Can Use Right Now

While City Hall debates definitions, riders still need to get to work.

Several consumer tools cut through official optimism by relying on crowd‑sourced reality:

• Transit App – Royale Subscription: Offers real-time crowding and vehicle movement based on user data, often spotting gaps before official alerts

• Citymapper: Flags service irregularities using live GPS feeds and rider reports
• Garmin Venu Sq GPS Smartwatch: Tracks door‑to‑door commute times over weeks, giving riders hard evidence of delays when filing complaints

Data isn’t just for agencies. Riders can collect it, too.

The Bigger Lesson Ottawa Keeps Teaching

Infrastructure failures make headlines. Data failures linger.

Ottawa’s LRT saga shows how performance can degrade without formally “getting worse,” simply by changing how success is counted. A ten‑minute wait doesn’t vanish because an algorithm smooths it into an average.

Until the city opens the black box — and accepts that rider experience matters more than model elegance — platforms like Tunney’s Pasture will keep filling with people staring down tunnels, wondering how $7 billion bought them less time, not more.

The numbers already know the answer. The question is whether Ottawa is ready to let the public see them.