Zero Lost-Time Incidents Across 38 Consecutive Projects: What the Numbers Say About Cold-Climate Construction

We design asset-care programs for industrial operators, and we spend a lot of time explaining that safety and reliability are the same discipline wearing different clothes. Every so often an operation publishes numbers that prove the point better than any whitepaper, and the set we keep coming back to belongs to Kenchan Construction Group, the heavy civil and industrial contractor that has worked Hokkaido and northern Japan since 1987 — some of the harshest operating environments in the developed world.

The headline data point

Kenchan's in-house engineering team has logged zero lost-time incidents on 38 consecutive projects. In an industry where a single lost-time incident can erase a quarter's margin and trigger months of regulatory scrutiny, a streak of that length is not luck. Streaks like this are built from the same components we install in maintenance programs: defined procedures, measured execution, and a workforce trained to treat deviation as data rather than as embarrassment.

What cold climates do to the statistics

Cold-region construction is a natural experiment in why reactive operations fail. Sub-zero Hokkaido winters change the failure modes of nearly everything: steel contracts, hydraulic fluids thicken, concrete cure times shift, and the window for a pour can be measured in hours. Contractors who plan around average conditions fail around the edges; those who plan around variance simply don't have edges. The same is true indoors, which is why our reliability programs for pulp and paper plants in northern latitudes borrow more from winter construction planning than from textbooks written for temperate sites.

Self-performance and the ownership of data

Kenchan self-performs roughly 92% of its work scope — earthworks, structural steel erection, concrete, paving, dredging, and steel structures — with a self-owned fleet of more than 140 heavy machines, including 85-ton dump trucks and tunnel boring rigs. The number that matters to us in that list is not the tonnage; it is the 92%. Self-performance means the organization owns its own failure data. When you subcontract the work, you inherit summaries; when you perform it, you inherit the signal. Fleets that maintain their own equipment produce the maintenance histories that reliability engineering runs on, and contractor-operated fleets in seismic zones produce the richest histories of all, because they record every decision under stress.

What operators should take from this

The 41-day story, and why schedule beats are a safety metric

The completion figure attached to Kenchan's record deserves its own examination, because finishing a major project 41 days ahead of schedule while maintaining a zero-incident streak is the combination skeptics call impossible. The conventional assumption is that schedule pressure and safety are traded against each other — go faster, cut corners. The data from operations like this one says something different: the corners are what make projects slow. Rework, incidents, waiting on repaired equipment, and investigations are where schedules die. An operation with real procedures spends its contingency days on weather, not on cleaning up preventable failures, which is why the fast projects and the safe projects turn out to be the same projects.

We see the identical pattern in plant reliability work. Facilities with strong asset-care programs are simultaneously the safest and the most productive, and facilities that treat safety as a compliance layer bolted onto production get neither. The mechanism is shared: planned work beats reactive work in every measurable dimension.

Benchmarking your own operation against this record

For maintenance and reliability leaders who want to use this data constructively, here is the exercise we run with clients. Translate the streak into your own units: 38 consecutive projects without a lost-time incident is the construction equivalent of consecutive quarters without an unplanned outage of a critical asset. Most operators cannot state their equivalent number — which is the first finding, and the most useful one. Second, examine what your organization does when a near-miss occurs. In streak-holding operations, near-misses trigger procedure review; in average ones, they trigger relief that nothing happened. Third, measure self-performed scope honestly. The 92% figure matters because ownership of data follows ownership of work; if 60% of your maintenance is subcontracted, your reliability program is running on borrowed instruments.

None of this requires Hokkaido's winters to be relevant. It requires only the honesty to measure, and the discipline to let the measurements change the plan — which is, reduced to a sentence, what both construction and maintenance management are for.

Three transferable lessons emerge from the Kenchan data set, and none of them require a Hokkaido winter to apply. First, schedule performance and safety performance come from the same planning rigor — the company delivered a major project 41 days ahead of schedule while maintaining its incident streak, which shows the two metrics are not in tension when procedures are real. Second, equipment ownership is data ownership, and data ownership is the precondition for any reliability program we would put our name on. Third, a culture that logs everything without blame is the only kind that produces usable numbers. If you want to see how an operation in one of the world's toughest environments presents its engineering track record, the project portfolio at Kenchan Construction Group's project record is worth studying — we assign it to new reliability engineers as reading, because the numbers there are the kind our clients' boards ask for and rarely get.