Construction Weather Risk Management: Heavy Build Guide

Climate risk management in construction is the process of anticipating rain, wind, lightning and heat with data at the scale of the job site, to protect schedule, field safety and contractual position in delay disputes. It is not a response to force majeure. It is the management of a measurable variable, with whatever lead time each hazard allows.

• In Brazil, climate events generated R$ 184 bi in losses between 2022 and 2024, with 91% of losses uninsured (CNseg/EY, COP30 2025), a national figure covering all sectors.
• A climate claim depends on two distinct records: the forecast that guided the decision before the event and the measurement that proves the condition on site after it.
• International standard contracts already separate the two: the FIDIC Red Book 2017 only grants a time extension for exceptionally adverse climatic conditions if the event was unforeseeable given published climatic data (clause 8.5(c)).
• i4sea monitors more than 100 critical assets in Latin America and Europe with 1 to 3 km resolution and post-processing calibrated per asset.

I have sat in site meetings where the question was not "will it rain?". It was "who pays for the week we lost?". That question shows up after the event, when nothing can be changed anymore.

At i4sea, we have worked with mining, railways, energy and insurance for more than ten years pushing that question earlier. Heavy construction fits here as an extension of the same method, not as a case we already have to show.

Operational climate risk is the chance of a weather hazard interrupting a critical activity, with impact on time, money or safety. On site, it is the concrete pour that cannot take rain, the crane that stops with gusts above the manufacturer's limit, the lightning that closes the site by protocol.

What is climate risk management in construction

Climate risk management in construction is the set of practices for identifying, measuring and anticipating weather hazards that affect a project, protecting schedule, safety and contractual position. It differs from force majeure because it works with probability, not with the event already consummated.

The difference is not subtle. Force majeure presumes an unforeseeable and unavoidable event. Rain during the rainy season, wind in a windy region, lightning in an area of known incidence: none of that is unforeseeable. It is in the site's historical record and, almost always, in the forecast. What varies is the lead time each hazard allows.

When an engineer asks why I insist on this distinction: managing risk means acting before, and force majeure is the argument of whoever missed the window to act.

What are the main climate risks in heavy construction

On heavy construction sites in Brazil, the hazards that most often interrupt work are rain, wind, lightning, extreme heat and prolonged drought. Each has its own threshold, window and effect. In convective compounds, rain, wind and lightning arrive together and halt operations for longer. Managing these risks means defining threshold, decision-making and record.

Rain affects activities by different criteria. In concrete pouring, precipitation above the tolerated level already forces the pour to stop, and aggregate moisture also matters. In earthmoving and slope work, what matters is the 24 and 72 hour accumulated rainfall combined with antecedent soil moisture.

Wind limits crane, tower crane and lifting operations, with the manufacturer's threshold measured in gusts. Lightning forces work to stop in open areas by internal protocol, such as the 30/30 rule, with a stand-down when the discharge is less than 10 km away, and in work at height NR-35 also applies.

Extreme heat reduces productivity and raises the risk of accidents from fatigue, but the indicator is not air temperature. It is heat stress, measured by WBGT (IBUTG) under NR-15, which depends on humidity and radiant load. NBR 14931 sets the limit for concrete placement temperature.

Prolonged drought reduces water available for curing and compaction and lowers earthmoving yield. On coastal sites, swell and high tide affect access, cofferdams and support vessels.

The windows are not all equal. A cold front wind announces itself days in advance. A convective cell's position has useful skill of a few hours to two days: it can pass 20 km from the site. A weather app forecast, made for the city, does not decide a lift.

How weather impacts construction schedules

Weather impacts schedules through unplanned stoppages: an activity interrupted without planning, with rework, idle crew and equipment, and a cascading effect on the following stages. The cost is not the lost day, it is re-sequencing the work.

An unplanned stoppage costs differently from a planned one. Knowing about Thursday's rain, the site pours concrete on Wednesday and uses Thursday for indoor work. Without warning, the crew is already mobilized, the concrete has left the plant, and the loss is total.

We do not have a construction case to show, and I would rather say that than adapt a number from another sector. The method comes from mining, rail and energy, over more than ten years: anticipating the event and reorganizing the sequence.

The math needs to be measured project by project, with the cost of an idle crew per hour, the cost of reassigning crew and the cost of rework. Without that number, the decision to stop becomes a matter of opinion. The path to building that math is covered in how to calculate the hourly cost of a rain stoppage.

Climate risks and contractual claims: what changes in the contractor's position

In contractual practice, weather enters the project through two paths that tend to be confused. Forecast is for deciding. Observation is for proving. A 1 to 3 km model is not a measurement, it is an estimate. A rain gauge does not anticipate anything, it only records what fell.

Useful evidence has three layers. The first is local observation: calibrated station, rain gauge and anemometer, with dated record, log and chain of custody. Without a proprietary instrument, an official station is used, INMET or Cemaden, with the declared distance.

The second is the baseline. On one side, the unproductive-day threshold written in millimeters per period, which governs the day-to-day of the claim. On the other, the site's climatological series, which defines what is exceptional there: historical percentile, project return period.

The third is demonstrating that the event exceeded that baseline and that the activity could not be performed or substituted in that window. Without the second layer there is no claim, only a complaint.

International standard contracts show how this test is applied. In the FIDIC Red and Yellow Book 2017, clause 8.5(c) grants an extension of time for exceptionally adverse climatic conditions, provided the event was unforeseeable given the employer's climatic data and published data.

In the Silver Book this right does not exist. NEC4 is more generous: time and cost reimbursement for an extreme weather event, with a measurement that occurs, on average, less than once every ten years (Freshfields, 25/11/2021).

Both turn weather into a threshold comparable against a baseline. Without it, there is nothing to compare the event to in the claim.

In Brazil, art. 393 of the Civil Code shifts the discussion to the unavoidability of the effects. Force majeure and fortuitous event are not synonyms in legal doctrine, and the burden of proving exceptionality falls on whoever alleges it.

There is a detail that almost no plan considers. A forecast is a double-edged sword. If the record shows that the project knew about the event days in advance and did not reassign crew or notify the client, that same evidence supports the claim of a breach of the duty to mitigate.

The record only helps when it comes with the decision made based on it.

On case law, I reviewed the ruling most cited on this topic. It is from TJSC, 3rd Civil Law Chamber, and concerns an apartment buyer against the construction company: delivery scheduled for March 2011 with a 90-day grace period, project delivered in April 2012, penalty upheld. The court did not disclose the case number.

The reporting judge treated intense rain and a landslide as a foreseeable risk of the activity and cited the 90-day contractual grace period. The reasoning shows what the courts expect in terms of planning. The regime, however, is consumer law.

Construction contracts, EPC and concession agreements have their own risk allocation, with rulings going the opposite way for atypical events, such as the 2024 floods in Rio Grande do Sul.

The pattern that can be stated is more modest and more useful: rain typical of the region and season is rarely accepted as a fortuitous event, and what decides the claim is the contractual and climatological baseline the project had in hand before signing.

The cost of not managing climate risk in construction in Brazil

The cost of not managing climate risk in construction shows up in three lines: an idle crew without a plan, schedule delay and loss that insurance does not cover. In Brazil, climate events generated R$ 184 billion in losses between 2022 and 2024, with 91% uninsured (CNseg/EY, COP30 2025), a national figure covering all sectors.

What insurance covers has a clear limit. Engineering risk coverage covers physical damage to the project, equipment and third parties from rain and wind. What is left out is the cost of the delay, which only enters with a loss-of-profit extension.

Without that extension, lost schedule time is a loss borne by the contractor or the owner. The 91% measures the environment, not the size of the gap.

There is no public figure for losses from climate claims in Brazilian construction. Foreseeable does not mean manageable. It means it can be planned for in advance, and that is a management choice.

On the regulatory front, CVM Resolution 244/2026 revoked CVM 218 and removed the mandatory climate risk reporting requirement for publicly traded companies. It reaches only listed companies, not privately held contractors. The pressure arrives through contracts: a listed client that audits the supply chain, a bank that prices risk into financing, and IFRS S2, currently voluntary until 2027.

How to structure a climate risk management plan on the job site

A climate risk management plan on the job site starts by mapping the hazards relevant to that project and region, sets stop thresholds per activity and is built on monitoring with enough lead time for decision-making, not on a generic daily bulletin.

The first step is identifying, per activity, which hazard interrupts the work: concrete pouring has a rain threshold, lifting has a gust threshold, open areas have a lightning protocol, earthmoving has an accumulated-rainfall threshold. Without this map, the alert becomes noise.

The second is defining who decides what, and with how much lead time. At i4sea, we have applied this for more than ten years in mining, rail and energy: the alert reaches the right person, within the window in which they can still act.

The third is the record, and this is where most plans stop halfway. A useful record has an instrument on site or a declared official station, up-to-date calibration, a continuous log and chain of custody.

And there is a fourth piece that almost no one includes: verification. Comparing forecast with local observation, month by month, shows where the model is wrong at that particular site, whether it overestimates summer rain or delays the cold front. Without this, the alert does not improve and the team loses confidence in it.

Every stop decision and the measured condition that motivated it become historical record. That history supports schedule negotiations, claim defense and the site's risk profile. For day-to-day routine, it is worth looking at how to structure a climate contingency plan.

Hyperlocal forecasting technology for field safety

The data that reduces stoppages and accidents is the alert at the exact point of the work front, within the window when the decision is still possible: forecast gusts measured against the crane's threshold, lightning with minutes of warning, accumulated rainfall compared against the concrete pouring threshold.

Resolution of 1 to 3 km, calibrated, brings the forecast closer to that scale. Resolution, by itself, is not accuracy. Every forecast comes with uncertainty, including about where and when the event forms.

The gain comes from fine mesh, local calibration and verification against observation. Open global models run at 9 to 25 km, Brazilian regional models at 3 to 15 km, and an app delivers an even coarser point.

i4cast monitors 18 hydrometeorological hazards across more than 100 assets in Latin America and Europe, with 1 to 3 km calibration per asset. The transmission-line principle applies to the job site too: anticipate within the operational window and keep a record of what was forecast and what was measured.

Lightning is the case with the shortest window: it shuts down safety within minutes and requires its own alert, not one derived from rainfall volume. It is worth looking at lightning monitoring for unplanned operational stoppages.

Frequently asked questions

References