Introducing prediction for designing climate-resilient infrastructure

There is consensus that climate change is increasing the frequency, duration and severity of weather events that affect the ability of our infrastructure to function as intended. All environmental loads (temperature, wind, ice, snow, rain and combinations thereof) are changing.

Precedence is out of step

For the past century or so, the construction industry has relied on precedence – the use of historic weather records to predict the likelihood of future events. This has served the industry well and much of the knowledge extracted from these records has been condensed into standards and codes of practice,  allowing designers to adopt a consistent approach to designing reliable infrastructure, almost all of which has stood the test of time. So far…

Increasingly, though, the environmental loads extracted from this historic data must be challenged. We can no longer assume that the underlying environmental processes are stationary (in the statistical sense).

In essence, our environment may be changing so quickly that it is no longer safe to rely on historical precedent to predict the loads that our infrastructure may need to resist over its future service life of typically 50 years or more.

Historically, we have focused on extreme values of environmental loads, at the ultimate limit state. In other words, what could mother nature throw at our structures that might lead them to collapse? However, we are now feeling the impact from more frequent events at, or below, the serviceability limit state.

They cause significant disruption through loss of function and amenity, often with minor, yet costly damage that an engineer may not traditionally regard as a structural failure. In contrast, it is worth reflecting on what an insurer or member of the public may consider as a “failure”.

A shift to a prediction-based approach

If the technical precedence that we, as an industry, have become so accustomed to relying upon is not wholly fit for purpose anymore, it is imperative that we begin to explore different methods to better understand our designs’ ability to function as intended and remain safe. If we are to succeed in this endeavour, we need to overcome some key challenges.

While a phenomenal amount of data has been created – for example, to inform the sites chosen for onshore wind turbines – it is often not freely accessible

We currently struggle with relatively limited access to data. While a phenomenal amount of data has been created – for example, to inform the sites chosen for onshore wind turbines – it is often not freely accessible. Our own Meteorological Office now operates as a commercial entity.

Academia can access its data freely for research purposes; industry cannot. We must overcome the issue of commercial value currently attached to data and recognise the potential cost to society if we do not.

In the UK, the committees that write design codes comprise of industry experts who contribute on a pro-bono basis. However, there is a limit to the extent of a such contributions. The magnitude of the climate crisis and the rate at which data should be reviewed will require this process to be professionalised.

For example, the wind map in the UK’s National Annex to EN1991-1-4 was published in 2004. It was based on 30 years of data and prepared by an eminent wind engineer (who has since retired) on an entirely voluntary basis. Repetition of his laudable input will not serve us adequately in the future.

We must also improve the ability of standards and codes of practice to adapt. While the fundamental premise of the Eurocodes is sound – a consistent approach across Europe that pools knowledge and experience, and lifts barriers to trade – their revision cycle is hampered by the process of achieving consensus across the many members of CEN.

Most of the Eurocodes were originally published in the mid-2000s, with the intention of updating them every 10 years. The first round of updates will likely be completed in the mid-2020s. Achieving technical consensus is important, but we need to be more agile in reflecting those elements of our standards that relate to the faster-moving changes in our climate.

Managing change risk and overcoming hesitation

Of course, there are inherent risks in moving completely to a prediction-based approach. Analysis of smaller, more recent datasets from which to inform our designs will come with greater statistical uncertainty.

If predictions are overly conservative, the industry risks increasing costs through overspecification. The key must be a commitment to regular re-evaluation of the data and application of the continually evolving methods of predictive statistical analysis.

If the predictions made by meteorologists are right, it is credible that the environmental loads for design of identical structures may differ in relatively short periods of time. This is a risk and may need to be countered by an acceptance that the structural reliability of infrastructure will change over its service life. Alternatively, infrastructure may need to be repurposed (perhaps with lesser function) over its service life to maintain the same level of reliability.

A statistical approach to evaluating the benefit to cost ratio of measures taken to improve reliability would also be appropriate, which differs from the deterministic limit state requirements currently presented by codes and standards.

Precedence is no longer the reliable option that it once was and the cost of inaction will only continue to escalate

These are big decisions, and the associated risks should be shared. Integrated project delivery models – where the contractor and asset owner share both project risk and decision-making responsibilities – are becoming increasingly popular in America and Asia. Meanwhile in the UK, a lack of risk sharing means that the country has some of the highest cost to build infrastructure in the developed world. Transparency, collaboration and data sharing are all key benefits of the integrated project delivery model and typically lead to better construction outcomes.

While the industry can be forgiven for being hesitant when it comes to implementing such a significant change – it is precedence that sees us continue to use Victorian infrastructure today – the gap between climate change and the industry’s response is only widening. Precedence is no longer the reliable option that it once was and the cost of inaction will only continue to escalate.

The social impact of exceedance leading to disruption can also be considerable. A three-day strike across the UK rail network in 2022 was estimated to cause an output loss of £91 million. While this is a stark example, weather storms have the potential to cause even greater levels of disruption given their likely simultaneous impact on road and rail transport. The insurance industry among others is increasingly interested in understanding the economic loss of events at or below serviceability limit states.

Engineers across the industry will agree that there is an ongoing need to prioritise climate change at the top of infrastructure agendas. And many will also agree that there is a risk in continuing to use precedence as the sole tool for safe and serviceable infrastructure when data analytics has the potential to produce such highly tailored project insights.

With climate change rapidly changing many historic premises on which we have designed infrastructure, a continually evolving prediction-based approach that includes consideration of impacts is vital for closing the gap.

John Rees is director of COWI in the UK. COWI is an international consulting group specialising in engineering, environmental science and economics, based in Lyngby, Denmark.