As England, Spain and huge swaths of the U.S. deal with record-shattering high temperatures, the time has come to stop looking at heat and heat waves as temporary inconveniences. As the climate warms, heat waves have become longer, more frequent and more deadly, at their worst killing thousands of people. With warnings that people are unsafe in houses without cooling systems, or that train tracks will buckle, and power and water systems will be compromised, we need to examine how well our infrastructures—the systems we’ve built to deliver critical services such as mobility, energy, water, and access to cooled space—are prepared for these new conditions. A mountain of evidence is emerging indicating that they are not, and that our ability to adapt infrastructures on large scales lags far behind how quickly the climate is changing. Failures are inevitable, and we need to be smarter about how we prepare for them.

As environmental extremes worsen, we must confront the reality that our infrastructures were designed for past conditions that no longer exist. With tremendous uncertainty about future climate, how do we engineer our way out of the challenge? Can the ways that we’ve been designing infrastructures over the past century keep our lights on, homes cooled, and our water running into the future? Designers and engineers intentionally design infrastructures to withstand pressure from water, cold, wind, fire and heat. But how much pressure should a given system be able to withstand? As a society we’ve generally codified how much risk we’re willing to accept when it comes to environmental hazards. When you hear about a 100-year event, an engineer has designed for a 1 percent chance per year of that event happening, versus a 10 percent chance for a 10-year event. Designing past this minimum can be much more expensive, but right now most standards for climate risks are still based on historical data from past climate conditions that appear increasingly obsolete.

The failure of infrastructures can take many different forms. It’s easy to envision failure of infrastructure as catastrophic destruction, for example, a road washed away by a flood or a neighborhood razed to the ground by a wildfire. While it’s important to plan for and mitigate catastrophic failures, heat can have a much more subtle set of effects. In general, as temperatures rise, infrastructure failures increase. Roadways rut and crack; power lines sag, risking short circuits and fire if they come in contact with trees; buildings struggle to keep in cooled air; water pipe reliability decreases; and energy generation is pushed to its limit as demand for AC skyrockets, triggering rolling blackouts during extreme summer heat when households need power most. Lots of small failures quickly snowball into new and larger failures. While failure of any one asset—one buckled stretch of railroad, or a burst water main—is manageable, city- and state-wide failures exacerbated by climate change are beyond what we can handle.

We don’t have a clear indication of how, when—or even if—we’ll curtail global greenhouse gas emissions. So, how do we engineer our way out of hotter days, longer streaks of heat, bigger and more powerful storms and whatever nature will throw at us next, when we are uncertain about our climate’s future? Combining this uncertainty with the limited resources available to refurbish, redesign and replace existing infrastructure, and ongoing political jockeying around climate, means that failures are inevitable. We need to design for the management of failed infrastructures. For centuries, infrastructure design has focused on fail-safe thinking, that is, we design for particular environmental extremes, and if exceeded failure is expected, with consequences (e.g., death, economic disruption) classified as “acts of god.” With increasing intensity and duration of climate hazards, failures will simply become too frequent and significant to dismiss.

So, let’s design with failure in mind. In doing so we’ll open up new design opportunities that allow infrastructures to gracefully fail while mitigating death and economic disruption. Safe-to-fail design is the balancing of community, environment, and infrastructural capabilities towards failing gracefully. We’re seeing the success of safe-to-fail systems already. In the Netherlands, the Room for the River project decided against building and maintaining expensive levees to keep rivers from flooding and instead gave land back to the rivers, accepting and planning for a future of floods. Farmers were allowed to plant in flood-prone land and reimbursed for their crops, which is much cheaper than constructing new levees. We can apply these lessons to heat waves, too. Instead of trying to retrofit all buildings with AC, we can focus on a few strategic community centers to provide more people with a cool place to go without stressing the power supply. We can plan with microgrids as backup power for critical services. How we design our neighborhoods for heat also matters: Density can reduce daytime highs, and in low-rise neighborhoods we can plant and maintain native trees.

In addition to safe-to-fail thinking, here are some ways to approach infrastructure design in the face of worsening climate change:

  1. Adapt and mitigate. Let’s not create an accelerating and increasingly uncertain moving goalpost. There’s an opportunity to adapt our critical services while also reducing greenhouse gas emissions. If it’s necessary to increase AC access, then let’s make sure we do so with renewable energy and efficient technologies. If we’re going to modernize our roads against heat and other hazards, let’s make sure we do so for autonomous and electric vehicles, reducing parking and deploying EV charging that is resilient to hazards.
  2. Armor assets for future climate. If we’re designing and adapting infrastructures for today’s heat, then by the time the systems go live they’re already obsolete. Instead, codes and regulations must be updated to account for climate change and its uncertainty. Governing bodies including professional societies and public agencies should provide guidance on how to make design decisions with deep uncertainty. For example, the American Society of Civil Engineers recently released a manual for designing for climate change. But recognize that resource constraints and time limit armoring as a singular strategy.
  3. Prioritize investment in the most vulnerable and critical systems. Even if there was complete buy-in on climate and a drive to react and save lives threatened by infrastructure outages, we simply will not be able to replace or upgrade all infrastructures fast enough. We need targeted investments that prioritize not only the most vulnerable assets but also the most critical. Two different roads may be equally vulnerable to heat, but if one is the main route into a city, or the road to a fire station or hospital, then it should be first in line for improvements.
  4. Design for heat mitigation. Any decision to modernize infrastructure to better withstand heat must also work to help reduce how hot it gets in urban environments. Too often infrastructure design is driven by legacy performance goals that don’t incorporate the complexities of a changing environment. Today’s climate change reality necessitates new performance goals that allow infrastructure to not only reduce worsening environmental conditions, but more importantly, give all people and the environment an opportunity to thrive.

Infrastructures are the mediators between our society and the environment, and how we choose to design and use their services can affect both people and where they live. For too long, we have accepted that it’s okay for infrastructures to operate with designs and goals rooted in a bygone era, putting communities at risk as the climates we live in move into uncharted territory. Not only is it time for an intervention, we simply have no choice but to intervene. And as we steer infrastructures in new directions, we must make sure that the new goals we establish allow people, natural environments, and economies to thrive in our climate-impacted future.

This is an opinion and analysis article, and the views expressed by the author or authors are not necessarily those of Scientific American.