Resilience runs on two clocks

Why climate resilience has become a strategic investment challenge.

Climate resilience is becoming one of the defining strategic challenges facing the UK water sector. Yet many of the planning approaches used today were developed for a more predictable world. As uncertainty grows, organisations need to make better investment decisions across entire systems, rather than optimise individual assets in isolation.

This paper explores why resilience needs to be considered across multiple planning horizons and why integrated, evidence-based planning is becoming increasingly important for future investment decisions. Increasing uncertainty around climate, regulation and investment means organisations need planning approaches that connect operational resilience with long-term infrastructure decisions. Traditional planning methods remain essential, but on their own they are no longer sufficient to support whole-system decision-making.

Water companies need a different way of making investment decisions if they are to build genuinely resilient systems.

One reason this is difficult is that resilience itself means different things to different people.

01An undefined term makes a good shield

One of the defining characteristics of resilience is that it lacks a universally accepted definition. When something does not go to plan, an undefined term lets us fall back on the older metrics we are comfortable with — the ones we can interpret and believe we can control. Those metrics are not necessarily wrong, and I do not think anyone clings to them out of malice, but rather the comfort of familiarity. The result is that responsibility becomes so widely diffused that accountability is weakened. Complex problems are accepted as inherently difficult, rather than owned and resolved.

Many of the metrics used as proxies for resilience assume that more is always better – more interconnection, more storage, more capacity always results in a lower probability of failure. These measures are easy to quantify and easy to justify. However, improving individual metrics does not necessarily improve the resilience of the system as a whole. In practice, all three can improve while overall system resilience deteriorates.

02What do we mean by resilience?

Before organisations can improve resilience, they first need to agree what they mean by it. Different definitions lead to different investment decisions. Within a single meeting, resilience can refer to several different concepts: redundancy (alternative routes and zones not dependent on a single source of supply); buffer (hours of service reservoir storage); recovery time; supply–demand headroom; asset reliability; and adaptive capacity (the extent to which a plan can be adapted if future conditions change). These measures do not necessarily move together, meaning organisations may be optimising different aspects of resilience without realising it.

A second distinction is equally important because it separates two fundamentally different approaches to resilience investment. Organisations can either reduce the probability of a risk occurring or improve the system’s ability to cope when it does. We have seen a preference for making a risky asset less prone to failure rather than making the network able to withstand what that asset’s failure would do. Both are feasible approaches. But hardening the site lowers the likelihood of the risk occurring whilst preserving the structural risk and the cost base that come with it, whereas making the network resilient to it opened up different operating models, and carried a unit cost benefit of roughly 5% on £/MLD as a consequence. That benefit would never appear in a conventional resilience metric. It only becomes visible when the system is considered as a whole.

03The two clocks are one system

Operational resilience and long-term investment planning are often managed separately, yet they are fundamentally the same challenge viewed over different timescales. Resilience also runs on two clocks, and they are usually owned by different people.

Within the day it is service reservoir storage and availability, single source of supply exposure, alternative routes into a zone, leakage, time to recover. Across years and AMPs it is the right capacity at works and pumping stations, trunk main routes, asset maintenance, and leakage and mains renewal strategy.

The important point is that these are not two separate topics. Within-day resilience is almost entirely bought by decisions taken on the long clock. This could refer to the storage that was built, the trunk main route that was chosen, the headroom left at a works, the mains renewed fifteen years ago. And long-clock decisions are only justifiable by reference to within-day performance, which is the same problem just sampled at two different rates. In practice they sit with different teams, in different tools, on different cycles, and get handed between one another in a spreadsheet. That handover is where systemic consequences tend to go missing.

Figure 1. The two planning horizons describe the same physical system viewed at different timescales. Capability is bought on the long clock and consumed (spent within the day) on the short clock. The handover is where most organisations reconcile them, and where systemic consequence is likely to go missing.

04More connectivity is not always more resilience

Investment decisions that appear to reduce risk can sometimes increase it elsewhere in the system. A generalised example: a requirement to remove single source of supply exposure produced a proposal to build cross connectivity between zones. Against the metrics in front of the decision-makers it worked as risk exposure reduced, more customers covered, and it was by a wide margin the cheaper option – roughly an order of magnitude cheaper than the alternative on capital cost.

What it actually did was spread the risk across a much larger population rather than reduce it. The apparent diffusion was lower. It obscured a bigger system risk: had two sources been unavailable at the same time, the consequence would have been catastrophic across millions of customers, and the new connections would not have alleviated it at all.

The alternative cost considerably more. It built genuine surplus capacity where the exposure sat, and then back-fed that surplus to offset risk elsewhere in the system — using local excess resilience in one area to reduce risk across a much wider one. Any metric shaped like cost per customer of risk diffused rewards the first option and cannot see the second.

Figure 2. Option A wins on both of the measures normally placed in front of a decision-maker. It loses on the measure that might not be seen at all. Illustrative; cost figures are indicative orders of magnitude, not a specific scheme.

05Why we deliberately hold less detail

More detail does not always lead to better decisions. For strategic investment planning, organisations often need a broader view of the whole system rather than a more detailed view of individual assets. Looking at a whole system means holding less detail than a hydraulic or hydrological model does, and I am not going to pretend that is consequence free.

Those models are more adept at a simulation-focussed approach, as opposed to an optimisation-centric one. They can afford lots of time periods or very granular detail precisely because the decision space is limited: they interrogate around a specific component of a problem, mostly through a single lens, whether that lens is temporal or spatial. For the questions they were built for, they are the right instrument and nothing else comes close. A system model trades resolution in order to widen the view. This is done to see what other levers exist in the network, where the bottlenecks sit, and how a plan behaves once alternative futures are introduced into the decision-making.

The failure mode is assuming detailed answers aggregate upwards into better system decisions. They do not always do so. Each detailed model carries a local focus, and optimising against it could be done at the cost to the rest of the system. Leakage reduced through pressure management, at the expense of flexibility needed during a transfer. Works capacity optimised against a trunk main constraint that nobody in that workstream owned. Mains renewal ranked on leakage history, which is not the ranking you get from customer supply interruption driven resilience.

None of this is an argument against detail. It is an argument about where detail is deployed, and against the assumption that resolution and scope are the same virtue.

06The spend that buys no resilience

Not all infrastructure investment improves resilience. Understanding where investment adds little system value is just as important as identifying where it should be increased. The reverse case is worth stating too, because it is the part that pays for the rest. Once you understand how a network actually performs, you can also identify spend that does not enhance resilience but drains funds anyway. In one hourly network model we identified service reservoir storage that could be reduced, which released both capital and operating cost – largely because the burden of maintaining a heavily regulated asset came off the books. Here we used an hourly model, because that was the right tool for that particular question. Savings found that way are available for the parts of the network that genuinely need them.

07Implications for future planning

The direction of regulation is increasingly aligned with the ideas explored in this paper. As the sector moves towards more integrated planning, organisations will need stronger evidence for comparing investment options across whole systems, rather than assessing individual assets in isolation. The planning system is being rebuilt around this problem. The January 2026 water white paper commits to abolishing Ofwat and merging water functions from the existing regulators into a single body, explicitly to replace a fragmented system with one capable of integrated management. It also proposes consolidating a patchwork of more than twenty planning instruments into two core frameworks, one for water supply and one for the water environment. The question is no longer whether a system view is wanted.

At the time of writing the 2026 Transition Plan has not appeared, and the Clean Water Bill announced in the King’s Speech in May is still to be formally introduced, so the new regulator remains some way off. Meanwhile the updated water resources planning guideline, published in April 2026, confirmed that WRMP29 proceeds as planned whilst the transition is worked through – draft plans are due early 2028 – and that supplementary guidance is being revised, with adaptive planning and best value metrics named among the topics. Note what those two topics have in common: both are about comparing options rather than testing one.

It is within this window that the framework intended to replace the current plans is being designed, whilst WRMP29 is being built under the existing one, and the methods people take into WRMP29 are the methods the new framework will inherit. The evidential expectation that comes with it is harder than the one before: demonstrating that options were compared against alternatives, not that a preferred option was tested. Simulation-oriented approaches are at risk of being weaker here, since each option has to be simulated and then compared afterwards — the head-to-head does not happen inside the tool.

Which leaves the real problem, and I would rather state it as a problem than as a solution: marrying the elements of simulation and systems planning. We are not there yet. But I do not see how a credible system plan can be assembled by hand from risks identified in simulation, and equally I do not see how a system model alone will surface every risk that granularity would reveal. Both halves are needed. Being explicit about which half you are using, and what you gave up to use it, seems a reasonable place to start.

Climate resilience cannot be achieved through isolated optimisation. It requires organisations to compare options across multiple timescales, understand trade-offs across entire systems, and make investment decisions using evidence that reflects how infrastructure actually operates. The challenge is no longer to build more resilient assets. It is to build more resilient systems.

The value of resilience

The value of resilience is not in agreeing a single definition, but in asking better questions and making better investment decisions. Three questions worth asking in your own organisation:

Sources. A new vision for water (Defra water white paper, CP 1490, January 2026); Independent Water Commission final report (July 2025); water resources planning guideline update and consultation response (April 2026); Clean Water Bill, announced in the King’s Speech, May 2026.

Note on examples. The cases described are drawn from client work and have been generalised. Costs are stated as orders of magnitude and populations as broad ranges; nothing here identifies a specific network, scheme or organisation.

Regulatory position stated as at July 2026. The 2026 Transition Plan and the Clean Water Bill were both outstanding at the time of writing. Check before release.

The principles discussed in this article are reflected in the way Decisio supports integrated, evidence-based investment planning across complex infrastructure systems. Find out more about Decisio’s approach to integrated planning and infrastructure decision intelligence at decisio.ai.