Climate, Uncertainty, and the Engineer’s Obligation to Prepare
Climate has changed throughout Earth’s history, long before industrial civilization existed. This post looks at the engineer’s obligation to prepare for conditions that are already shifting. Ice cores, sediment layers, and tree rings document warming and cooling cycles driven by natural forces. Milankovitch cycles, variations in Earth’s orbital path around the sun, have produced ice ages and interglacial periods over hundreds of thousands of years.
Solar output has varied too. Volcanic eruptions have cooled the planet for years at a time by injecting aerosols into the stratosphere. Ocean circulation patterns have shifted, redistributing heat across hemispheres on timescales that dwarf recorded human history. None of this is in dispute by serious people on any side of this conversation.
Why I Approach This With Caution
I have worked in mechanical and plumbing engineering for over four decades. In that time, I have lived through several rounds of climate predictions that did not play out as forecast.
In the 1970s, peer reviewed literature warned of a coming ice age driven by aerosol pollution blocking solar radiation. By the 1980s and 1990s, specific predictions about sea level rise, hurricane frequency, and regional temperature increases were made with confidence the data did not fully support later.
That track record matters. It does not mean the science is wrong. It means the science is hard, and confidence intervals deserve more respect than they typically get in public discourse.
Climate is one of the most complex systems humans attempt to model. Feedback loops, ocean currents, solar cycles, volcanic activity, land use changes, and atmospheric chemistry all interact in ways that remain difficult to isolate from one another, however engineers have an obligation to prepare. The models driving policy recommendations are sophisticated, and the scientists building them are serious professionals. Even so, they work with incomplete data, imperfect parameterizations, and systems that behave nonlinearly, meaning small input changes can produce disproportionately large output changes. That is not a political observation. It is a mathematical one.
To be clear, I am not denying the planet is warming. Measured temperature records, satellite data, and ocean heat content all point the same direction. Nor am I denying that human activity contributes to that warming. The basic physics of CO2 as a greenhouse gas has been understood since the 19th century, and no one with a physical science background seriously contests it.
What I question is the confidence assigned to any single driver, given that natural variability has produced larger swings in Earth’s climate before any of us were born. I also question it because the models projecting specific outcomes have a documented history of running warmer than measured results over multi-decade periods.
That is not denial. That is engineering discipline applied to an evidentiary question.
The Practical Problem for the Built Environment
Here is what I know from experience. Buildings are designed to last 30 to 50 years. The HVAC systems inside them are designed for the climate conditions at the time of design, not the conditions that may exist at the end of the building’s useful life.
When those conditions shift significantly during a building’s service life, the systems underperform. Occupants grow uncomfortable, energy consumption rises, and mechanical equipment wears out faster than projected.
This is not a theoretical concern. In my experience, design conditions in many regions have shifted measurably over the past 30 years. Summer design dry bulb temperatures in parts of the Southeast and Southwest now run higher than the values engineers used in the 1990s. Cooling degree days in many metropolitan areas have increased as well. The design conditions used for buildings built in 1985 or 1995 may not reflect the conditions those same buildings face today, let alone the conditions ahead in 2040 or 2050.
ASHRAE recognized this and responded appropriately. The 2025 edition of ANSI/ASHRAE Standard 169, Climatic Data for Building Design Standards, was fully revised and updated, expanding to 12,424 global station locations, a 35 percent increase over previous editions, with updated climatic design condition tables for every station. ASHRAE updates these tables on a four-year cycle using the most recent measured weather data available. The 2021 update used hourly observations from 1994 to 2018. The 2025 update incorporated the most current data available.
This is the correct engineering response to a changing climate, regardless of what you believe about the cause. The data shows what conditions buildings actually operate in. Standards should reflect measured reality, not historical assumptions that no longer apply. Engineers who keep using outdated design conditions are not being conservative. They are being negligent.
Adaptation Is Not Surrender
This debate tends to treat any position short of full endorsement of the consensus narrative as climate denial, and treats adaptation as a concession that mitigation has failed. Both framings are wrong, and both are worth pushing back on directly.
Adaptation is what engineers do. We do not control the loads buildings experience. We design systems capable of handling those loads within acceptable parameters. When the loads change, we update the design. That is not surrender. That is competence.
Consider a building designed for a 95 degree Fahrenheit summer design condition that now regularly experiences 102 degree peak temperatures. That building is under-designed. The occupants feel it. The utility bills reflect it. The equipment failures confirm it. The solution is not to debate whether the temperature increase is natural or human caused. The solution is to size the equipment for the conditions the building will actually experience.
The same logic applies to infrastructure broadly. Stormwater systems designed for historical precipitation intensity curves have been overwhelmed by rain events exceeding their design parameters. Electrical grid infrastructure designed for historical peak demand is stressed by cooling loads that exceed original projections. Road surfaces specified for historical temperature ranges are experiencing thermal expansion and contraction outside their design envelope. These are infrastructure performance arguments grounded in measured data.
On the Limits of Predictive Certainty
It is worth being precise about what I am, and am not, arguing here.
I am not arguing that CO2 emissions are harmless, or that environmental stewardship is unimportant. Reducing unnecessary emissions is a reasonable goal on multiple grounds that have nothing to do with climate projections, including air quality, energy efficiency, and resource conservation.
I am not arguing the planet is not warming. The measured data is clear on that point.
I am arguing that the confidence assigned to long range projections often exceeds what the underlying models can support. Building policy on worst-case projections as though they were certainties produces outcomes that are difficult to reverse once the projections turn out to be overstated.
In my experience, overconfidence in a single model is often a precursor to expensive failures. Running sensitivity analyses helps us check our assumptions. Built-in safety factors exist because we know our models are imperfect representations of complex systems. That same discipline should apply to climate policy.
What Should Actually Happen
Several practical steps follow from this analysis, regardless of where anyone stands on the underlying science.
Building codes should be updated to reflect current and projected design conditions. Engineers should design to the 2025 ASHRAE Standard 169 tables, not tables from 2001 or 2009. Existing buildings undergoing significant renovation should have their HVAC systems evaluated against current design conditions, and upgraded where the gap is material.
Infrastructure planning should incorporate scenario analysis that accounts for a range of possible climate trajectories, rather than a single projected outcome. Designing stormwater systems, road surfaces, and electrical infrastructure to handle conditions somewhat more demanding than historical norms is a reasonable hedge against uncertainty, and one that pays off under multiple scenarios.
Energy efficiency must be pursued aggressively on its own merits. A building that uses less energy to maintain comfort is less vulnerable to utility cost increases, less dependent on grid reliability, and less expensive to operate, regardless of what the climate does over the next 30 years.
Finally, engineers should be honest with clients and building owners about the uncertainty in long-range projections. after all, engineers do have an obligation to prepare and this includes informing their clients, while being clear that the uncertainty cuts both ways. Conditions could turn out better than projected, or worse. Designing for the middle of the range while understanding the tails is standard engineering practice, applied to a problem that deserves it.
The Point of Agreement
We may not agree on cause, on the precise magnitude of warming over the next century, or on the relative contributions of human activity and natural variability to the changes already observed. Even so, we can likely agree on this: buildings and infrastructure designed for conditions that no longer exist are underperforming, and the gap between design assumptions and current reality is measurable, documented, and growing.
Closing that gap through updated standards, better data, and honest engineering practice is the correct response, under any theory of what is causing the problem.
Adaptation is not a political position. It is an engineering obligation. The conditions we are actually experiencing should drive our designs. That has always been true, and it remains true today.