Infrastructure systems are interdependent, and disruption continues to develop after the initial event. Electricity failures can affect water and wastewater services. Restoration of one part of a network can depend on other services being available. The duration and sequencing of recovery therefore shape both the total burden of an event and where that burden falls. 

In a recent paper, we developed a way to estimate these whole-of-event consequences by following disruption and recovery across electricity, water and wastewater networks. Applied to a coastal flooding scenario, we found that around 4,500 properties were directly exposed to flooding, but once cascading infrastructure failures were included, almost 14,000 properties experienced at least one utility outage. Following those outages through recovery produced more than 94,000 property-days where at least one essential service was unavailable. 

The important point for planners is that this can change the decision: the places that appear highest risk at the point of impact are not always the places with the greatest cumulative burden.

Consequences accumulate through recovery

A conventional exposure assessment captures where the hazard intersects with people or infrastructure. Adding cascading failures improves this by identifying where disruption spreads through connected networks.

Both approaches still treat consequence as a point-in-time measure. They do not capture how long disruption lasts or how quickly services recover.

I argued for this in my 2021 paper, and it’s useful to now have quantified it for interconnected infrastructure and communities. In our recent paper, we apply what we’ve called a “whole-of-event” assessment, tracking how long services remain unavailable, how recovery progresses through connected systems, and how those consequences accumulate across communities.

We noticed that even the burden attributed to different utilities can vary significantly. At the start of the modelled event, around 7,000 properties were without electricity and a similar number were without water. Once outage duration was included, the resulting burdens were very different: approximately 63,000 property-days without electricity compared with 29,000 without water.

The priority areas can also change

Recovery also changes the geography of risk. Some places that appear highly impacted immediately after the event recover relatively quickly, while others experience prolonged disruption because of their position within interconnected infrastructure networks. When we compared property rankings using static cascading consequences (i.e. initial outages) with rankings based on cumulative recovery burden (i.e. outage duration), the priorities changed. 

If a business case estimates the benefit of an intervention from avoided consequences, omitting cascading failures and recovery can understate those benefits. If investment priorities are based on the spatial distribution of risk, a snapshot assessment can also direct attention toward different assets or communities than a whole-of-event assessment would.

The same applies to emergency management: where to direct resources, which communities may need temporary support, and which infrastructure should be restored first all depend on outage duration as well as initial disruption.

Dependencies matter

These results also reinforce why infrastructure dependencies need to be represented explicitly. Electricity, water and wastewater are usually managed as separate systems, but communities experience them together. A household with water restored but no electricity may still face substantial disruption. An infrastructure asset may be physically undamaged but unable to operate because an upstream service has failed. Those dependencies affect both the initial cascade and the recovery trajectory.

Understanding cascading risk therefore requires looking at the system of dependencies through which consequences propagate, as well as the hazard itself.

What this means for planning

For planners, infrastructure providers and policymakers, recovery should be treated as part of consequence estimation. Measures such as property-days without service, outage duration, the proportion of properties still without service after a defined period, and time to restoration provide a closer representation of what communities experience than exposure or initial outage alone.

Where risk estimates inform infrastructure investment, adaptation prioritisation or emergency planning, the choice of consequence metric can change the answer.

Read more: Brunner, L.G., Skipper, D., & Logan, T.M. (2026). Estimating whole-of-event consequences in interdependent infrastructure systems. Reliability Engineering and System Safety, 276, 112894. https://doi.org/10.1016/j.ress.2026.112894