Managing infrastructure resilience and adaptation

ABSTRACT

The paper contributes to strategic thinking about infrastructure resilience and adaptation. A technique is presented in a case study for analysing options on the timing of protection against climate change. Predetermined trigger points for responses to resilience weaknesses are discussed, showing how flowcharts can act as prompts to the need for action. The roles of national governments in improving resilience in key areas are discussed. The paper introduces the concept of a ‘chain of resilience’ for interdependent infrastructure systems and identifies some key questions for national governments to ask about the linkages between systems before requiring appropriate remedial actions. Practical steps for private owners and investors to increase resilience in their own infrastructure are identified. Some suggestions for tackling the increasingly important area of cyber resilience are presented. The use of actuarial techniques is discussed.

KEYWORDS:

• Adaptive pathways
• climate risk mitigation
• remedial action triggers
• interdependent systems
• cyber resilience

1. Introduction

The subject of resilience is highly complex, and this paper aims to make a useful contribution to strategic thinking about its analysis and management by Government and other public sector officials, planners, senior executives, infrastructure analysts, engineers, actuaries and investors. Hopefully they will find that the ways in which the paper brings together and discusses key resilience issues will stimulate their own thought processes when deciding or advising on the challenges they meet in practice.

Infrastructure exists to provide a service to end users and resilience ensures that this service continues as far as possible despite adverse events or circumstances which may arise and impair performance or cause operation to cease altogether. The factors affecting resilience include climate change, severe storms, landslips, floods, earthquakes, fire, structural failure, poor maintenance, changes in safety regulations, disruption of supplies, shortages of skilled staff, gradual deterioration of performance, enemy action, vandalism, changes in the level of demand for the service, major computer issues, and a lack of robust recovery plans. Strategic thinking about resilience necessitates the development of a framework which envisages as many as possible of these adverse events and circumstances, and the steps which can be taken, either in advance or at the time performance deteriorates, to mitigate their impact and maintain or restore a standard of service which is acceptable to users. This paper discusses the components of such a framework.

Managing resilience needs good recovery plans for when things go wrong. It also needs a long-term plan that calls for the ongoing monitoring of performance, which will indicate when remedial actions need to be taken, or a completely new approach adopted, to maintain services at the standard required. Resilience planning should recognise the time dimension of resilience and provide an ongoing mechanism which will enable infrastructure assets and systems to be adapted if performance deteriorates to the extent that predetermined trigger points are reached. The paper outlines a suggested mechanism for this purpose.

Account should be taken of the minimum standard of performance which users will find acceptable, not necessarily a higher and more costly standard.

By ‘performance’ a simple measure would be ‘economic performance’, i.e., the percentage of occasions on which a promised service fails to be delivered or is delivered late. However, this is not the whole story. If the service user can easily substitute an alternative service, he or she might be content with a lower standard of service from the first provider, and resilience planning needs to take this into account.

Another measure of performance would be ‘environmental performance’, i.e., the extent to which the service and the resources necessary to provide it is meeting society’s environmental expectations, for example in terms of natural landscapes, carbon consumption or pollution. These expectations are likely to rise significantly in future, so environmental performance which is currently regarded as acceptable may have to be raised as time goes on. This paper will concentrate on economic performance for the sake of clarity, but it should be understood throughout that similar steps should be taken to optimise the resilience of environmental performance.

One of the biggest challenges to resilience is climate change. That it is happening is certain, but there is not a clear expectation of its speed, intensity or geographical distribution, and even experts have only a vague idea of how infrastructure assets will be affected. Should infrastructure designers aim to look at ‘worst case’ scenarios and build in the maximum degree of protection they can? Unfortunately, this is likely in many cases to be perceived as just too expensive, and an alternative approach to the issue may be sought. One option that this paper discusses is to take an ‘adaptive pathway’ approach to climate change, which builds in only a reasonable minimum of protection initially at relatively low cost but has the option to build in more protection or rebuild later if climate risks increase.

One complication is that many infrastructure systems are interdependent, i.e., they depend on inputs from other systems or outside sources. If the underlying systems fail, the system they supply will also fail, though not immediately if some of these inputs have been stored as buffer stocks. In effect there is a chain of resilience, extending through two or more systems. For example, if a local authority operates a refuse collection and treatment system which takes household waste by road to recycling plants, the resilience of that system depends on the continued operation of the recycling plants, which in turn depends on the continuity of a supply of electricity and water, and also on the road system not being blocked by bad weather, particularly near the entrances to the vehicle depots and recycling plants. It will also be essential to have a continuing fuel supply for the vehicles, though a buffer stock could cover temporary shortages. All these elements – electricity, water, roads and fuel – are supplied by separate systems which form a chain of resilience for the recycling system. The paper discusses what actions are needed from governments and infrastructure managements to identify and tackle such chains.

Much of a country’s infrastructure is usually managed by private-sector owners, including pension funds and insurance companies, and the paper suggests that they too need to decide on resilience options, to protect their revenues and ensure the continuation of services which are of importance to the community.

The operation of many infrastructure assets and systems is becoming increasingly dependent on the correct ongoing performance of their cyber systems. The paper therefore discusses how resilience in this important area can be achieved.

1.1. Acceptable levels of service

The main aim of maintaining an infrastructure system’s resilience is to provide end users with a continuing service which is at an acceptable level, even if performance is not necessarily perfect. This minimum standard of service will differ from one type of service to another, and may differ at different times according to changes in society.

Understanding the ever-changing minimum levels of service required by end users and the extent to which they will tolerate poor performance is a key to the degree of resilience to be targeted in each system – the aim need not always be 100% performance.

Of course, the managers of infrastructure systems should aim to provide higher standards of service than the minimum wherever this can be done at a reasonable cost which can be afforded. If, however, the provision of these higher standards would involve significant extra cost, it may be decided at a political level that the aim should be restricted to the provision of the minimum acceptable standard of service. It would be worthwhile in this case to devote considerable effort to understanding the minimum service standard required from time to time, since if the service provided were to fall consistently below this minimum standard, users would complain strongly and there could be far-reaching consequences.

Some of the practical influences on the minimum standards of service required by users are:

• If there is an increasing frequency of extreme climate events, the system in question may not be the only infrastructure to suffer degradation in its service. In this case, there might be a somewhat greater degree of tolerance.

• Comparisons will be made with the performance of similar systems in other countries.

• Users’ levels of tolerance are likely to be greater if the managers of the system have gradually built-up trust over a long period by sharing with users the full details of incidents and remedial actions taken.

• If the system in question is providing inputs for other systems, any under-performance is less likely to be tolerated.

2. The meaning of resilience

A useful practical definition of resilience is given by the Intergovernmental Panel on Climate Change (IPCC) (2011, as cited in Happold, 2020, p. 11): ‘The ability of a system and its component parts to anticipate, absorb, accommodate, or recover from the effects of a hazardous event in a timely and efficient manner, including through ensuring the preservation, restoration or improvement of its essential basic structures and functions’. However, for the purposes of the present paper a proviso should be added that resilience requires that there is not too much harm in periods while the service is down.

The resilience terms which are used in this paper have the following meanings:

• Infrastructure assets – physical assets fixed to the ground

• System – one or more infrastructure assets which are intended to work together

• Feeder systems – the systems which supply inputs to other systems

• Resources – the equipment, supplies and people necessary to operate a system

• Service – benefits provided by a system for people or for another system

• Resilience – a system or an asset can be said to be ‘resilient’ if, when its service ceases or deteriorates to a level unacceptable to its users, it returns to an acceptable level within a reasonable period of time without having caused too much harm to users or the wider community in the meantime.

A ‘reasonable’ period of time is subjective and will depend on the nature of the service – an interruption to the electricity supply which is restored after half a day might be considered reasonable by most people but an interruption lasting a week may not. The requirement of ‘without having caused too much harm in the meantime’ is also subjective, though a lack of electric power which resulted in deaths would not be consistent with resilience.

As another example, a railway system could be regarded as resilient if, upon trains being stopped due to the discovery of an unsafe bridge, pre-arrangements with bus companies enable passengers to continue their journeys after a delay that they see as being of reasonable length. It would be consistent with resilience if bus services were available on subsequent days until the bridge had been repaired, but the more days this takes and the longer the journey times, the less likely it is that resilience of the railway will be seen to have been achieved.

Resilient infrastructure should be able to withstand, at least to some extent, natural calamities like earthquakes, storms, excessive heat, excessive cold, landslips and flood. In 2021 there were 432 disastrous events related to natural hazards worldwide, accounting for 10,492 deaths, affecting 101.8 million people and causing approximately 252.1 billion US$ of economic losses (Centre for Research on the Epidemiology of Disasters CRED, 2022).

The financial resilience of infrastructure is also important, to ensure funds are available for adequate maintenance and the adoption of remedial actions when needed.

People are themselves naturally adaptable and resilient if adjustments can be made without too much disruption to their lives. For example, when public transport ceased during the Covid pandemic, many people demonstrated their adaptability by working from home using the internet and some actually found the experience more enjoyable than commuting. This leads to the general proposition that the extra cost which would be involved in making a system more resilient may not be justifiable if it is clear that in the event of failure of the system, many people would just adapt and adjust their lives accordingly.

For a wider perspective on resilience, it is worth looking at a comprehensive report prepared in 2020 by The UK National Infrastructure Commission (NIC) and a separate report on the case studies they examined (2020b).

Some of the reasons for the lack of resilience of an infrastructure asset or system include:

• Physical damage to it, for example by a storm, flood or wildfire;

• A gradual deterioration in its performance to the point where it reaches a level unacceptable to its users which cannot be easily or quickly remedied;

• Social changes which mean that it falls out of use to such an extent that it is no longer regarded by its owners as economic to continue to operate it;

• Changes in social attitudes which mean that previous performance standards are no longer acceptable;

• Price increases for the ongoing resources required for its operation or maintenance, resulting in higher charges to users which render the continuation of the service uneconomic;

• Continued inability to obtain ongoing resources, which mean that the service has to be discontinued for long periods;

• Shortages of trained staff;

• Discovery of serious construction faults which cannot quickly or economically be repaired;

• Inability to recover quickly from a cyber attack or a machine breakdown.

3. The meaning of adaptability

Adaptability is the extent to which an infrastructure asset or system, or the service it provides, can be modified in order to improve resilience. These modifications can be achieved by remedial actions designed to mitigate the probability or impact of a risk, or to restore a service more quickly. Sometimes major changes may be required before resilience can be returned to an acceptable standard, in which case there will have to be a new pathway which aims for the continuance of the service but in an entirely different way.

When the managers of an infrastructure asset or system are faced with a growing resilience issue, taking a relatively modest remedial action (such as the updating of recovery plans, the introduction of staff retraining programmes, the establishment of more frequent maintenance inspections, or the installation of more fire doors and escape routes) would not usually be significant enough on its own to count as embarking on a new pathway. When it commences, a new pathway is usually associated with one or more major changes but thereafter the infrastructure will continue on its journey along the new pathway and further remedial actions may occur along the way. ‘The pathways perspective implies an iterative and ongoing approach … that enables … learning so that choices along pathways can be altered in response to predefined triggers’. Examples of major changes which would normally mark the start of a new pathway include the rebuilding of important structures in new locations, drastic modernisation of the methods of operation, extensive structural alterations to cope with sudden big increases occurring from time to time in the numbers of people physically present, or the introduction of significant modifications to the services provided to end-users.

4. Planning for adaptive pathways

Climate change is the principal uncertainty for which adaptive pathways must be planned. The ‘City of London Adaptive Pathways Study’ (Happold, 2020) was a major study of the climate challenges facing the city up to 2050, including warmer and drier summers, less rainfall, milder and wetter winters, longer heatwaves and longer droughts. The study proposed an adaptive pathway, the purpose of which was to develop a risk-based approach to climate adaptation and resilience and to identify trigger points by which times actions must be taken to ensure continued resilience. The insights obtained included a realisation that some risks – flooding and water stress – may need major infrastructure interventions in due course, triggered by reaching certain thresholds. In the meantime, a short-term action plan was formulated. This study provides an important benchmark for similar planning related to other cities around the world.

This paper supports the conclusion of a recent study (International Coalition for Sustainable Infrastructure ICSI, The Resilience Shift, & Arup, 2022) that new infrastructure assets and systems need to be planned, designed, built and operated for resilience to more frequent and more severe climate events. The existing built environment, a complex network of independent and connected assets, must also adapt. Climate hazards include heat and cold, drought, flooding, wildfires, windstorms, snowstorms, landslides and coastal erosion. Infrastructure ought also to be able to respond to the indirect impacts of climate change, such as transition risks, supply chain disruptions, workforce and lifestyle changes, and population movements.

Although climate change is the most important reason for planning adaptive pathways, it is not the only reason. A pathway may have trigger points for reductions in performance due to any cause, whether climate-related or not.

When considering whether to make an asset or system more resilient or adaptable, the first point to think about is whether this can be done without incurring any significant extra capital cost. It may sometimes be found that this is possible, in ways such as the following:

• Setting up permanent arrangements which enable other services for end users to be substituted quickly and easily if necessary, for example the provision of bottled water to substitute for tap water if a water main bursts or buses to substitute for trains if a railway line is damaged;

• Making more space for the storage of resources to enable a bigger buffer stock to be maintained;

• Setting up permanent arrangements which enable the resources necessary for operations or maintenance to be sourced from elsewhere in the event of supply difficulties;

• Establishing crisis-management plans and insisting on regular practice exercises;

• Ensuring that regular structural surveys actually take place and that there are suitable arrangements for acting on the results where necessary;

• Preparing in advance a range of emergency operating plans for use in different scenarios when the service is partially interrupted or degraded, to prevent chaotic consequences – for example, reduced timetables for a railway system or rationing plans for an electricity system;

• Preparing in advance a number of recovery plans for various scenarios, to facilitate speedy recovery.

However, there are many circumstances where providing extra resilience or adaptability will involve extra cost. For example, it might be decided to build information systems to give early warning of developing resilience problems and enable adaptation to take place while there is still time – such systems might monitor trends in the flooding levels in nearby rivers or data about emerging structural weaknesses.

If there are growing doubts about the resilience of an existing system, it may be worthwhile to spend money on improving the degree of resilience, for example by installing flood barriers or storm protection. However, all possible causes of a lack of resilience must be considered, not just the few which are easy to mitigate. It may not be worth protecting against floods if the installation cannot be protected from wild fires, for example.

Extra costs will also often have to be incurred in order to mitigate disaster risks involving potential loss of life as far as possible, which must always be a priority in resilience planning.

A study (Tellman, Bausch, Eakin, et al., 2018) of Mexico City’s seven centuries of adaptation to water risks demonstrates that the adoption of adaptive pathways may carry its own risks in the longer term, as the changes may themselves create new problems.

The introduction of advanced technology or artificial intelligence (AI) to facilitate the operation of an infrastructure system or asset may often mark the start of a new pathway, though this would carry new risks and would necessitate new kinds of remedial action when necessary.

The design and construction of infrastructure assets may need to be modified in order to facilitate the introduction of a new pathway later if it proves to be necessary. One example would be to use a modular construction approach. Buildings are constructed offsite in a factory setting, and then transported and assembled onsite at the building locations. Such an approach is versatile, adaptive and customizable, allowing for more rapid feedback that enables dynamic management of the project and the possibility of making significant structural changes if these prove necessary in future years.

Modular buildings can be disassembled and relocated or refurbished if necessary. Therefore, part of a building can continue to serve the community whilst other parts are being upgraded offsite if this proves necessary in the future. Constructing a building in a modular way can also allow management to test the demand and revenue forecast assumptions used for the project, before the project is complete, so that capacity does not go unused; or if necessary, so that additional capacity can be added (Flyvbjerg, 2021). This type of responsiveness to feedback and iterative learning by management is not possible with monolithic infrastructure projects like the Channel Tunnel.

This paper now presents, through an example, a simple technique which can help in the analysis of the options for an adaptable pathway in relation to an infrastructure project. The case study included in the Appendix at the end of this paper shows for illustration purposes the kind of analysis the planning team would carry out, by using well-established discounted cash-flow techniques, for three different scenarios of climate change. The technique displayed in the case study takes account of time and cost dimensions: costs and benefits which will occur in the near future carry a heavier weighting than those which will not arise for many years. It can be used to explore a wide range of future scenarios, including those which are based on different assumptions about the speed and impact of climate change, as well as scenarios not related to climate change. The technique is not in itself new but many people may not have realised how useful it can be when applied to scenario analysis for infrastructure.

5. Trigger points and remedial actions

Whenever a massive event occurs which has a disastrous impact on a system’s infrastructure, it is bound to trigger a review and a search for remedial actions or a new pathway. Such events are often nature-related and the remedial actions may have to be confined to reducing the impact if the event were to recur for example, by legislating for new buildings to be earthquake-proofed or for disaster recovery plans to be strengthened. However, this section of the paper discusses how to plan trigger points related to less extreme deteriorations in performance.

Plans for adaptive pathways should normally contain predetermined trigger points on performance which, when reached, will indicate the need for remedial actions to be taken. If little or nothing is done, the risks may increase and disorderly functional failure of the asset, system or service becomes more likely, resulting in strong complaints.

When planning an adaptive pathway for an asset or system, one of the key questions is what trigger points should be built into the plan. As time passes and new operational experience of the asset or system emerges, various specified aspects of the operational performance should be closely monitored and when predetermined trigger points are reached the plan will require at least a review and possibly the commencement of some remedial actions or even the adoption of a new pathway. A flowchart such as that shown in Figure 1 can guide managers on what should be done when trigger points are reached – the flowchart can be modified to fit particular circumstances. At the outset, when the adaptive plan is being prepared, values will be fixed for the trigger points; these values can be adjusted after a period of operation has elapsed if considered necessary.

Figure 1. Resilience tool.

The flowchart in Figure 1 helps identify when the ongoing performance of an infrastructure asset or system has fallen below its trigger point to such an extent that remedial actions or a new adaptive pathway may be needed. The numerical values on which such a flowchart are based are determined at the outset of a project and are based on the perceived tolerance of end-users to poor performance. These values can be revised occasionally from time to time in the light of emerging changes in end-users’ tolerance, for example as determined by opinion surveys.

To use the flowchart, a measure of tolerance for underperformance beyond the trigger point is defined (variable X in Figure 1). The left-hand branch of the flow chart applies to situations where the degree of under-performance is relatively small and less than X%. In this case the first step is to test whether the underperformance is consistent with peer underperformance, for example whether there has been a similar degree of underperformance in comparable systems in the home country or overseas, and if so there may be no need for further action at this stage. However, if that is not the case, and the under-performance is specific to the particular system, the question is whether it is due to external factors, for example climate change, in which case remedial actions should be considered. The right-hand branch of the flow chart applies to situations where the underperformance is relatively serious and greater than X%, so remedial action and possibly even a new adaptive pathway need to be considered. After checking that the trigger point is still appropriate, the next step is to identify whether the underperformance is due to previously identified external factors like climate change and, if so, to consider implementing a new pathway to cope with it.

It may be unnecessary to take immediate remedial action or adopt a new pathway if the system’s underperformance is within the tolerance level and possibly temporary. For example, should excessive heat experienced during summer months lead to reduced working hours, then any tolerable service delivery disruption experienced might be consistent with that of peer service providers and no immediate remedial action would be required. However, if excessive heat temperatures were to persist and service delivery disruptions became intolerable, for example if the heat resulted in the deaths of staff or end-users, then remedial action or a new pathway would need to be undertaken at that time, or the service might even have to be closed down altogether while the heat persists.

Trigger points may not always remain appropriate over time, and the suitability of these needs to be reassessed on an ongoing basis. If an uncertain event causes a system-wide change to occur (for example, a rise in sea level may result in the occasional flooding of coastal railways and service delivery interruptions), then it is possible that instead of taking remedial action to try and restore the system’s operational capability to prior levels (say, by building sea wall protection and trying to make the coastal tracks more resilient to future floods and ongoing erosion), it may be more sensible and economical to implement a new adaptive pathway, manage stakeholder expectations and change the trigger point for future performance measurement (for instance, by permanently moving the railways inland and normalising the longer journey time).

Typically, when performance becomes intolerable it can be expected that a new adaptive pathway will be implemented where remedial action is not available or cannot be taken immediately; or if an external factor that is being monitored (for example, the change in average temperatures over time) moves beyond a predetermined point (Z say). Here it is understood that if the new pathway is not implemented, the system cannot be expected to continue operating optimally in its current form.

Flow charts could perhaps be further developed by assigning probability values to the uncertain events and determining the payoff profiles associated with each of the possible outcomes, such that the expected economic value of a decision can be estimated. This would help to optimise the decision-making process.

Since the adoption of a new pathway at a future date is likely to involve significant upheaval and cost, an important planning question is how long a period of time will elapse before it becomes necessary. There can be no definitive answer to this question, since much will depend on the occurrence of uncertain events, but it may be possible to carry out a stochastic analysis, based on varying the rates of change in overall performance each year, and to estimate theoretical probability distributions of the performance levels that would trigger the adoption of new pathways over a period of time. The authors have not done this but can see that it may be a possibility.

A difficulty with triggers is that they are likely to require action at a time when the organisation is less able to afford it than previously, because revenues and profits have dropped. There is perhaps some merit in setting money aside each year into a contingencies fund before a trigger point is reached, so that a return to resilience can be financed without too much pain and delay.

The triggering process can be illustrated by considering a hypothetical railway system. Suppose at least 90% of the trains are expected to arrive at their destination on time every year, i.e., a 90% punctuality target. A trigger point for action could be set at 80% punctuality. If in a particular year only 85% of trains were punctual, that would be unsatisfactory but not so serious that it necessitated remedial action. On the other hand, if the following year only 75% were punctual, this would trigger the predetermined limit of 80% and would almost automatically lead to remedial action being taken. Remedial actions might have to include some climate protections but also some changes not related to climate, for example a change in management, new operating procedures, or stricter measures to prevent cable theft.

Alternatively, it might be considered at that time that more drastic action was needed and that a new adaptive pathway, possibly one which incorporated a whole new set of climate protections, should be introduced. The new pathway might include, for example, major works to rebuild stations and install flood barriers, and the automatic closure of the line whenever severe weather was forecast – the precise mix would not need to have been determined in advance, but would depend on the emerging experience of climate-related incidents. Hence activating the predetermined trigger would stimulate action, but the nature of the action would be determined at the time. Hopefully the action would return the railway to at least 90% punctuality so that it was resilient again, but a few years later punctuality might again fall below 80% and a new set of remedial actions would have to be introduced or even a new pathway.

A broader trigger could be used, depending not just on operational performance but instead on the degree of satisfaction of users. If climate change had caused many other infrastructure systems to experience a worsening performance, then users might still give our system a rating of unchanged satisfaction, even if its operational performance had dropped below 80%, so remedial action would not necessarily be triggered at that point.

In addition to the overall trigger point for a review leading to a new pathway, which would be retained, it might be sensible for the hypothetical railway to have some mini-triggers for particular causes of a loss of punctuality due to climate change. Four causes might be identified (see Table 1, which also shows some of the possible remedial actions).

Table 1. Possible remedial actions for specified causes.

Cause	Possible Remedial Actions
Extreme cold or snow	Install points heaters, improve heating in waiting rooms, buy snowploughs
Extreme heat, lineside fires, rails buckling	Paint some rails white, issue bottled water to staff and passengers, improve fire-fighting capability
Structural damage due to storms or wind	Strengthen station roofs and other structures, remove lineside trees
Flooding	Re-route streams, change gradients, clean drains more often

Each climate cause would be monitored separately to assess its contribution to the loss of punctuality, and if it started to exceed a particular level, say resulting in 2% of trains being late, then that would trigger the commencement of remedial action for that particular cause. A similar system could be used alongside for causes of delay not related to climate change.

The advantage of such a system is that the remedial work and its costs could be spread out, rather than coming as one big expenditure when a new pathway had to be adopted. If the gradual approach improved resilience to such an extent that a new pathway was not required after all, the resulting saving in cost could be substantial.

Reaching a trigger point should not be the only event which requires remedial actions to commence. It is just as important to monitor trends before any trigger point is reached and if it becomes clear that the situation is worsening due to a particular cause, it may sometimes be appropriate to commence remedial actions for that cause straight away without waiting for the trigger point.

The routine collection of accurate and consistent data on the causes of incidents of poor performance is essential for any infrastructure system, not just for railways, in order to establish trends and consider what kinds of remedial action are necessary, but one important complication is that each individual performance failure may have several underlying causes. For example, is a power cut in a violent thunderstorm due to climate change or to poor cable or pylon maintenance? If a performance failure is due to a combination of events (whether climate events or otherwise) they should all be recorded and it may be desirable to take remedial actions to mitigate each of the underlying causes. In cases where a performance failure is due to the failure of a feeder system, any remedial action considered necessary should include discussions with the managers of the feeder system to see if its resilience can be improved, as well as consideration of alternative feeder systems.

As an example of data collection, a railway operator might require the recording of the arrival time at destination for every train in the timetable, showing whether it was early, on time, or late (and by how many minutes if late), and the reasons stated by the train crew and by the line controllers for any arrival more than (say) 10 minutes late. The reason for lateness recorded by the train crew would be based on their own perspective (for example ‘late running of train in front’), while the reason recorded by the line controller would take account of wider underlying causes (for example, ‘failure of points number … affecting many trains’). Multiple causes would be recorded for each lateness where known. Codes allocated to the principal causes of lateness would facilitate data analysis, which could show trends and indicate which underlying causes of lateness needed to be tackled by management as priorities to improve resilience, either because of their frequency or because of the extent of the lateness which could arise from them. A similar data collection system would be used for train cancellations and their causes.

6. Role of central governments

There are some actions which should be taken by central government in order to improve resilience right across the country. It is a government’s duty to ‘think big’.

Governments sometimes expect new builds to incorporate climate-change resilience in their planning, design and construction process. For example, there will be parks in Shanghai that serve as drainage (Arup, n.d.) and already there is a road tunnel in Kuala Lumpur that serves as a flood tunnel (Institution of Civil Engineers ICE, n.d.). Many other imaginative solutions to the issue of long-term resilience may have to be considered when designing new builds in future.

An international resilience problem which might become increasingly common and worrying is the possibility of prolonged heatwaves at very extreme temperatures, perhaps beyond any which have occurred so far in any country. Such heatwaves could cause deaths or serious health problems for staff and users of buildings and infrastructure, which may mean that they can only be used for part of the year or only at night. It may therefore be worth thinking seriously about building some buildings and structures underground, even though the expense may far exceed the costs of construction on the surface. As an example, Montreal has a city which is partially underground, stretching for 1½ square miles (Wikipedia, Underground City, Montreal). Another option is to insulate the roofs of new and existing buildings to a far higher standard than at present may appear worthwhile, to keep properties cooler. Greater resilience could also be provided at a national level by investing in fire engines, water pumps and water-carrying aeroplanes which could be moved around the country easily.

In the UK another possible consequence of rising temperatures may be more frequent and more prolonged droughts. It should be a national priority to study possible strategic remedies, including the construction of more reservoirs and even a national water grid, and more effective actions to remove leaks from water pipes.

One set of actions which the UK Government has already been taking to increase national resilience is the introduction of numerous flood protection schemes. However, judgements will increasingly be needed about whether residences in certain communities may have to be abandoned when trigger points are reached, rather than being further protected in this way. This particularly applies to some low-level coastal areas. For those communities which will continue to have flood protection, adaptive pathways should be planned when developing schemes, because of the uncertainties about the timing and impact of climate changes. For an illustrative worked example of the analysis required, refer to a case study by the Institute and Faculty of Actuaries (IFoA), Infrastructure Working Party, 2019).

Of prime importance in any country is the resilience of the national electricity generation and distribution system, because electricity is an input for many other systems which depend on it to provide their end services. The failure of the electricity system could mean the simultaneous failure of many other systems. Governments should consider various ways to increase the resilience of the electricity system, such as:

• Building and maintaining redundant generating capacity which can be brought into action quickly if the main generators fail;

• Ensuring adequate maintenance is carried out each year on distribution systems;

• Preparing emergency plans to share or ration electricity and prioritise certain types of users such as hospitals;

• Establishing treaty arrangements and cables with neighbouring countries to provide electricity from their own systems if necessary;

• Deciding on the best systems for storing electricity where practicable and activating these when needed;

• Offering financial incentives to home owners to install their own solar panels or wind turbines.

In some countries, like the UK, many schemes of this type have already been introduced.

The National Infrastructure Commission (NIC), 2020a) recommended that the UK Government should establish standards every five years for the resilience of energy, water, digital, road and rail services. It is too soon to say what form these standards will take but hopefully they will allow adaptive pathway approaches to be taken where appropriate.

One of the most important areas for governments is food resilience, i.e., the need to ensure that the population has enough food to avoid starvation. The food infrastructure is to a large extent in the private sector, consisting mainly of farms, processing plants and supermarkets, but supplies could easily be disrupted by extreme climate events. The sector has in the past already shown a high degree of flexibility and adaptability, and there have been many alternative sources of supply and consumption options. However, there may still be actions which the government needs to take to ensure that there is sufficient resilience against such events as prolonged heatwaves, droughts, wildfires and very heavy rainfalls. The actions required may include nature-based solutions, the introduction of new technology, and contractual arrangements to ensure that food supplies can be obtained from outside the country in an emergency.

When the central government is considering extra spending on resilience for any type of infrastructure, it will be very conscious that this will mean less spending elsewhere or higher taxes or borrowing. This is a very delicate political balance and much will depend on the extent of public knowledge about existing vulnerabilities in the country’s infrastructure systems. The government should not try to hide these vulnerabilities but should consider actions designed to get public support for a strengthening of infrastructure resilience to the point where it is consistent with end-users’ expectations.

7. Interdependent systems

Four different types of interdependencies between infrastructure systems can be identified, adapted from Beraud and Ahmad (2011):

• Physical, where a commodity produced by one system as output is used by another as input;

• Geographical, where elements of different systems are in close spatial proximity, so that a local environmental event affects all of them;

• Cyber, when there are important information flows between different systems;

• Logical, where different systems will be affected by the same external changes, including climate events and social and political developments.

It is, of course, possible that two (or more) systems will be linked by all four types of interdependencies. The more interdependency there is, the more risks to the delivery of the service there are which need to be investigated. Mitigation of the risks might include reductions in the degree of interdependency, for example by:

• Finding new sources of supply of required inputs;

• Moving a system’s control centre to a new geographical location;

• Seeking back-up sources and dedicated cables for the information flows which are needed;

• Joint planning by relevant authorities for action to mitigate climate events by installing back-up processes which can take over if necessary – this emphasises the importance of stakeholder participation in resilience discussions.

Some of the mitigation actions required may involve such drastic changes that they constitute a new pathway.

It may be possible to gather some useful hints about the kind of analysis required from Mao and Li (2018), who have assessed the impact of interdependencies on the resilience of networked critical infrastructure in urban areas. Their case study of three systems in a middle-sized city in Eastern China covered electric power, telecommunications, and water supply, and emphasised the importance of post-disaster performance restoration.

The central government of a country has a very important duty in relation to its interdependent systems, by ensuring that the links between them are sufficiently identified, including links which are not immediately obvious, and then implementing adaptations which reduce dependencies.

As an example of the complexity of interdependent systems, at the time of writing the logistical system of England’s national health service is performing poorly. There are delays in discharging patients from hospital due to a lack of sufficient social care to enable them to go back into the community and long delays in the ambulance service for taking people suffering emergencies to hospital, because ambulances are having to wait for hours outside hospitals until space can be found to admit the people they are carrying. Patients who have been diagnosed by their own general practitioners as needing non-emergency hospital treatment are facing long waits before hospitals can admit them, since not enough treated patients have been discharged to make room for the newcomers. Thus there are four linked systems: the ambulance service, the network of general practitioners, the hospital service and the social care network.

In looking at possible new pathways, there is a wide measure of agreement that logistics performance could be improved if treated patients could be discharged to social care more quickly, but this is easier said than done, partly because of insufficient numbers of staff in the social care sector. Even if quicker discharges could be accomplished, there would remain the significant challenge of dealing with a large backlog of patients who have been diagnosed but not yet treated.

As the above example demonstrates, the identification and implementation of actions to improve the resilience of interdependent systems may sometimes be a difficult task. However, there may often be actions which can be taken to reduce the resilience risks for delivery of services to end-users. The first stage is to identify the linkages between the various systems and the scope for reducing the dependency of one system on another, by asking questions such as those set out in Table 2.

Table 2. Key questions for interdependent systems.

Key Questions for each interdependent System
1. Has there been an analysis of the inputs from feeder Systems or from outside sources which are critical for enabling the System to operate at the required standard? Which of these inputs could be sourced from elsewhere at short notice? Are contractual arrangements in place to enable these replacement inputs to flow as soon as they are needed?
2. If inputs had to be sourced from elsewhere, are the arrangements for transporting these inputs to the System likely to be robust and timely?
3. Are stocks of all the critical inputs held and for how long would each stock last if the supply failed?
4. If the System would become inoperative when the first stock runs out, what is an acceptable period for it to remain inoperative?
5. If it seems possible that the System might remain inoperative beyond the acceptable period, are there options at present for increasing the size of stocks, and at what initial and ongoing cost?
6. Are there ways in which users of the output from the System might be able to switch to alternative sources to supply their needs, if and as soon as the acceptable period for the System to remain inoperative is exceeded?
7. Could the geographical location of the System be changed, if necessary to spread the climate risks of over-concentration in a place where feeder Systems are also located?
8. Can action be taken to improve the resilience of the System’s feeder Systems?

Once the answers to these key questions are known, the government should undertake a strategic planning process for each system and its linkages to identify an adaptive pathway which would lead to improvements in resilience if this is considered necessary. For example, the following actions might be considered as part of a new pathway:

• Establish contractual arrangements for obtaining inputs from elsewhere promptly if needed;

• For systems where inputs can be stored, increase the level of stocks of any inputs where current stocks would not last for at least (say) three months;

• Develop a back-up system which could take over if necessary;

• Ensure there is sufficient transport capacity to bring in supplies from other sources;

• Establish a review system for keeping track of stocks and any changes in the answers to the key questions;

• Strengthen the structures of the system and any back-up and feeder systems, to protect against extreme climate events.

• Consider whether the geographical distribution of infrastructure systems, back-up systems and feeder sources, and in particular their control locations, is sufficiently dispersed from a risk viewpoint.

• Take action to improve the operational efficiency of at least one of the linked systems.

Considering a system and its feeder systems together, there is often a ‘chain of resilience’ influencing the resilience of the system. Failure of the system’s performance may be triggered by a failure of one of its feeder systems, and failure of a feeder system may be triggered by a failure in one of its own feeder systems. There is a saying, ‘A chain is only as strong as its weakest link’ and that is certainly true of chains of resilience. The identification of these weak links can point to where remedial action is most needed.

If chains of resilience have been identified before the occurrence of a major disaster, the knowledge of the linkages between different systems should be of assistance when decisions are made about which systems should be restored first when recovering from the disaster.

It is suggested that governments should encourage the managers of interdependent systems and their feeder systems and other suppliers to collaborate, with the aim of establishing and maintaining clear communications, data flows and emergency arrangements, and joint plans for actions to reduce the dependence of one system on another.

8. Privately-owned infrastructure

Where infrastructure is controlled by companies or investors in the private sector, its output may be of great importance to the lives of ordinary people. Moreover, the output may sometimes be of critical importance as an input to infrastructure systems or assets controlled and operated by other parties in the public or private sector. It is therefore vital that those who control infrastructure in the private sector should strive as much as managers in the public sector to make their infrastructure as resilient as possible. Moreover, there will usually be a revenue stream from the end users which needs to be protected.

The following paragraphs set out some of the practical steps which investors like insurance companies and pension funds can take, bearing in mind that it is often (though not always) the case that the management of the infrastructure may be in the hands of third-party asset managers rather than the investors themselves:

• When purchasing the freehold of an existing building, investors would normally get a structural survey first. They could ask for this survey to look at resilience risks and specifically at extreme flood risks, extreme storms and extreme heat, as well as the risk mitigation measures which the investor could take and when they should be taken.

• There may also be buildings or structures which the investor already owns, which would be worth getting surveyed in a similar way.

• When entering into a contract to build a new property or infrastructure asset, the investor should first consider, before the design is settled, whether the asset may be exposed to extreme climate risks, and ask the contractor about the options for mitigating these risks, either from the outset or later under an adaptive pathway approach.

• The users of a building which the investor is considering purchasing or constructing may be exposed to risks in the roads and paths giving access to the building if the climate worsens. The investor could consider whether to enter into negotiations with the authorities responsible for those roads and paths to get remedial measures taken before a problem arises.

• Careful consideration should be given to whether the operation of the building or infrastructure to be owned by the investor is dependent on the outputs from systems owned by public authorities or other parties in the private sector. The question should be asked, suppose these outputs from elsewhere were to fail, is there an acceptable alternative source of supply? An answer in the negative might be a barrier to investment.

• Resilience against premature obsolescence should be considered by the investor, thinking particularly of whether there are possible alternative uses for the asset if the risk were to materialise. Possible causes of premature obsolescence might include technological improvements, or sustained price rises or shortages in the raw materials required for successful operation, social changes resulting in reduced uses and revenues, prolonged staff shortages, or the discontinuance of a critical service provided by a public authority or a commercial company.

• In order to protect the asset and its users as far as possible against all kinds of disasters, the investor should consider instructing the asset manager to hold mock disaster exercises at least once every two years.

• In thinking about insurance arrangements for the asset, investors should beware of force majeure clauses which could prevent payouts if extreme climatic events were to occur.

• In looking at possible options for achieving greater resilience, the investor will want to carry out various forms of risk modelling, including scenario analysis, with the aid of actuarial assistance. Techniques similar to those set out in our case study (see Appendix) may be useful. The Bank of England’s climate scenarios will be a useful resource for UK projects. UK financial institutions might be assisted in assessing adaptive pathway options by the Climate Narrative Tool currently being developed by the Climate Financial Risk Forum (CFRF) (2022).

• Financial resilience is a big topic, yet it is of crucial importance, since financial failure leading to lack of maintenance, the inability to afford remedial actions or the closure of the asset or system will disadvantage not only the investor but also users of the service. One way an investor can achieve a measure of financial resilience is to obtain a guarantee from a public authority that the revenue will not fall below a specified level; this is the principle behind rail franchises in the UK, and the new UK infrastructure bank has the power to issue more guarantees for future projects. However, such a guarantee may not achieve permanent financial resilience if there are exit clauses which the public authority can exercise. Investors might sometimes be happy to accept a lower degree of financial resilience than could be obtained with a guarantee, in order to increase their expected returns.

• Just as in the public sector, continued resilience depends on the continuance of satisfactory performance and user satisfaction, and it would be worthwhile to set trigger points which would stimulate an early discussion between the owner and manager about the remedial action which may need to be taken at the owner’s expense. The owner should insist on having a data flow which enables changes in performance to be monitored closely.

9. Cyber resilience

Cybersecurity is often considered only as an afterthought by developers, and they believe it can be included in any kind of development at a later stage. However, with the massive increase of the attack surface, it is vital to develop a security mind-set at early stages of a ‘smart’ development, in order to avoid immature products that fail to satisfy the security requirements of the target operation. Cyber resilience is becoming very important for the overall resilience of infrastructure systems and assets.

In the ‘90s, ‘Internet of People’, was the way for people to communicate fast and inexpensively, search for information and entertain. Over the last decade, the Internet of Things (IoT) is the way to remotely control machines within a household or a building. The number of smart buildings is increasing and the IoT technology is implemented in critical infrastructure, gradually creating smart cities that increasingly depend on the IoT technology. According to studies (Globe Newswire, 2020), the global smart building market is forecast to grow to $127.09 billion by 2027, with a compound annual growth rate of 12.5%. As a result, the dependency on the Internet and the IoT technology will increase, along with the risk of attacks. Industry reports (Unit 42, 2020) state that 57% of the IoT devices are vulnerable to medium or high severity attacks, 72% of healthcare networks mix IoT and Information Technology (IT) assets, allowing malware to spread from individual to vulnerable IoT devices on the same network, and 98% of all IoT device traffic is unencrypted, exposing personal and confidential data on the network to exploitation.

Smart buildings use automated processes to control operations such as heating, ventilation, air conditioning, lighting and security. With the existing buildings being responsible for 40% of the energy consumption (Sandfort, 2021), the energy cost and the environmental effects of production and consumption, smart buildings offer ways to reduce energy consumption levels. Virtual Power Plants (VPP) take energy consumption concepts several steps further, where smart technology brings buildings, building blocks or even entire communities together, sharing the excess energy supply from one building to benefit the excess energy demand from another.

Automation and artificial intelligence are already being used in operations. Smart buildings and cities are supposed to be resilient by definition, as these are considered adaptable; however, their adaptability can also become their weak point. Infrastructure systems can be put out of action by cyber failures from causes such as programming errors, malware attacks, or overloading due to unexpected demands. When these failures occur, the reputational and financial costs for the organisation concerned can be substantial, particularly if confidential data relating to individuals has been compromised or lost, or if a ransom has to be paid. If the failure results in the infrastructure having to be out of action for a prolonged period, this may have widespread impacts on users and cause deep dissatisfaction. Failure can have even more catastrophic effects, when it causes physical damage apart from data theft or denial of services. Cyber resilience is therefore becoming an ever more important strategic issue.

One option that can mitigate both external and internal threats is to construct a cyber system based on the segregation of networks. This is a process that separates critical network elements from the internet and other less sensitive networks. Even if a network is breached, the attacker will be able to attack only that precise asset, rather than the assets of the entire network. If the system stores critical data, they must be kept in a particular data module of the system and protected with adequate security controls.

As circumstances change, the cyber system may need to be adapted along a new pathway, for example to cope with a heavier workload, to carry out new kinds of operation with longer processing times, or to produce outputs in different formats. It is at this time of adaptation that security weaknesses or programming errors might inadvertently be introduced into the system, so the greatest care needs to be taken.

The use of artificial intelligence which enables a cyber system to learn from its experience presents its own resilience challenges. In particular the nature of the learning process itself will need to be thoroughly tested in simulations before it is introduced in practice. Because the process in effect adapts itself, there must be regular human monitoring of the adaptations which are taking place.

The other important requirement for cyber resilience is that, if there is a main control centre of the system, there should be adequate physical protection for it, so that the centre is not destroyed or damaged by natural and/or environmental events. Ideally there should be a back-up plan to operate the cyber system on alternative equipment at a different location, just in case the worst happens at the primary location.

Internal threats within an organization may compromise performance. Individuals who have been previously authorised to work on information systems may now pose a threat due to their acquired knowledge and access to the organization’s systems. They may become less inclined to observe security precautions and occasionally they may be tempted into dishonesty. The necessary remedial actions include staff vetting and regular security checks.

Possibly the infrastructure may be able to continue operating for a limited period or on a limited scale after its cyber system fails, and staff will need to be trained beforehand in how to achieve this continuation safely. This ability to carry on could make a big difference to the resilience of the service provided to users. It might be necessary to operate the service for a while using old-fashioned methods not involving a cyber element, and appropriate equipment may need to be held in reserve to facilitate this at short notice.

Cyber systems will have a continuing influence on the life cycle of an infrastructure asset from its design and development and throughout its lifetime, so the resilience of cyber systems is of major importance. However, it might not be possible to adapt a cyber system indefinitely to take account of new requirements and the time may come when it needs to be replaced, a risky process which may signify the start of a new pathway for the infrastructure asset it serves. Unfortunately, with the present state of knowlege it cannot be forecast when this replacement will occur.

10. Actuarial techniques

There are more uncertainties and complexities in infrastructure analysis than used to be the case before climate change was taken seriously, which means that a wider range of future scenarios now needs to be explored, to gain a greater understanding of possible outcomes before important decisions are taken. For this reason, actuaries, who are accustomed to dealing with the intricacies of finance, risk and investment extending over periods of many years, may now be able to make an increasingly useful contribution in relation to infrastructure also.

The actuarial techniques which can be used in infrastructure work include discounted cash flow analysis, model building, scenario analysis, and the extrapolation of risk trends. These techniques were originally developed for use in the financial services industry but are just as appropriate for analysing infrastructure issues, to ensure that decisions are soundly based. Actuaries have also developed techniques for analysing and managing complex risk situations, such as the CRisALIS^TM step-by-step approach (Milliman, 2022). Such analyses do not lead to a single ‘right’ answer, but they do provide helpful insights into a wide range of possible future outcomes.

Actuaries may more often be included in multi-disciplinary project teams in future, or in some cases teams might obtain advice from firms of consulting actuaries.

11. Conclusion

Increasing levels of uncertainty in the world about the big risks to our way of life require the greatest possible flexibility and adaptability in infrastructure, to take account of changing circumstances in future. There is a continuing search for greater resilience, but it may make financial sense to aim for a lower level of resilience straight away and wait to see how circumstances, particularly climate change, develop. Planning for adaptive pathways will become increasingly common, to keep initial capital costs down.

Pre-planned trigger points for remedial actions, good data flows on incident causation and performance, active planning by central government on big issues and interdependent systems, resilience planning by private investors, and increasing attention to cyber resilience – all are necessary ingredients of a robust resilience and adaptation strategy.

The right choices will make infrastructure more reliable, assist a country’s economic growth and help its inhabitants to be more content.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Notes on contributors

Chris Lewin

Chris Lewin is an actuary, who was formerly the CEO of several of the UK’s largest pension funds and a Public Member of Network Rail. He has worked with members of the Institution of Civil Engineers on guides to risk management and front-end thinking in major infrastructure projects. Currently he leads the Infrastructure Working Party of the Institute and Faculty of Actuaries, which studies infrastructure as an investment for insurance companies and pension funds.

Monica Rossi

Monica Rossi is an investment professional with over 10 years of experience in investment product development and portfolio management in emerging markets. She holds the Financial Risk Manager (FRM) certification and is a member of the Infrastructure Working Party of the Institute and Faculty of Actuaries (UK).

Evangelia Soultani

Evangelia Soultani is a life assurance actuary with experience in regulatory frameworks and financial modelling. She has worked in different projects within Europe for the implementation of capital regimes and profitability standards. With quantitative modelling and risk management expertise, Evangelia is interested in the research of applications of those techniques in different fields. She has co-founded an information security start-up and she currently develops a quantitative model for the risk assessment of cybersecurity and operational risk. She is a member of the Infrastructure Working Party of the Institute and Faculty of Actuaries (UK).

Kumar Sudheer Raj

Kumar Sudheer Raj is an Assistant Professor in Actuarial Science area at Institute of Insurance and Risk Management promoted by IRDAI (Insurance Regulatory and Development Authority of India). He has work experience of more than 12 years and specializes in teaching Financial and Actuarial Mathematics. He holds MSc Actuarial Science degree and is a member of the Infrastructure Working Party of the Institute and Faculty of Actuaries (UK).

Appendix

Analysing Adaptive Pathway Options – Hypothetical Hospital Case Study

This is a highly simplified example designed to illustrate how a discounted cash flow method can be used to analyse adaptive pathway options extending over a period of years. The scenarios presented here relate solely to climate events but the technique can extend to other scenarios as well, for example social changes with major impacts.

It is proposed to build a hypothetical new hospital at a capital cost of £500 m. Although it will be reasonably well constructed, it might not be able to withstand very severe storms, which could cause major structural damage and flooding; currently such storms are expected only once every 100 years, based on historic weather patterns. It has been proposed that an extra £200 m should be spent at the outset, to make the roof and structure much more robust and install self-contained flood protection, which should enable the building to withstand even the most severe storms envisaged as possible. Opponents of this extra expenditure have pointed out that the £200 m could be spent instead on building another worthwhile project, which might result in getting better overall value for money.

The project team investigates whether an adaptable pathway solution to this dilemma should be considered. This would necessitate some modifications to the hospital’s design to enable extra protection to be built in later if necessary. These modifications would involve extra design and construction costs at the outset of £25 m in total. Three different climate scenarios are being considered for the purpose of analysis:

• Scenario (a) – the climate remains unchanged, with major storms having a 1% p.a. chance of occurrence each year, necessitating repairs costing £20 m each time (i.e., an average annual repair bill of £0.2 m).

• Scenario (b) - the climate worsens and the chance of major storms at this location rises gradually over the next 20 years, so that by year 20 there is a 20% p.a. chance of occurrence each year, necessitating repairs costing £40 m each time (i.e., an average annual repair bill of £4.1m rising from £0.2 m to £8 m by year 20). After year 20 the climate stabilises so that the average annual repair bill remains at £8 m per annum thereafter.

• Scenario (c) – the climate worsens as in scenario (b) for the first 20 years, and by year 20 appears to be worsening further with a wide margin of uncertainty, so that the possibility of a major storm destroying the hospital during the next few years cannot be excluded, unless full protection has been built in.

It is assumed that the hospital will provide the community with benefits worth £100 m pa. All the figures are expressed in today’s values with no allowance for inflation.

There are 4 alternative options to be considered:

Option (1) assumes that the capital cost at the outset is minimised at £500 m, with no option later to move to a new pathway – this means that after the first 20 years the hospital would have to close under scenario (c) but could remain open under (a) and (b).

Option (2) assumes that there will be an option later to move to a new pathway, and the capital cost is therefore £525 m, but that a new pathway will not be chosen at year 20. The hospital would have to close at year 20 under scenario (c) but could remain open under (a) and (b).

Option (3) is as in option (2) but under all three scenarios full protection will be built in as a new pathway at year 20, at a cost of £230 m. Note that the cost reflects an assumption about the cost of temporary partial closure while the new protection is installed. The total cost of the protection, with no discounting, would thus be £25 m at the outset plus £230 m in year 20, i.e., a total of £255 m, whereas total protection could have been built in at the outset for only £200 m. However, the value of this cost is much less than £255 m, because of the discounting when calculating a Net Present Value.

Option (4) assumes that full protection is built in immediately, and the capital cost is therefore £700 m.

Here is an analysis of the four options under each climate scenario, based on Net Present Values obtained by discounting future benefits and costs at 4% pa.; in practice the calculations might be repeated for alternative discount rates (See Tables A1 to A4).

Table A1. Option (1) scenario analysis.

Option 1 (in £m)	Scen(a)	Scen(b)	Scen(c)
Benefits (see note 1)	2500	2500	1359
Capital cost	500	500	500
Cost of future repairs (see note 2)	5	147	56
Total cost	505	647	556
NPV of project (benefits less total cost)	1995	1853	803
Benefits-costs ratio (benefits/total cost)	4.95	3.86	2.44

Table A2. Option (2) scenario analysis.

Option 2 (in £m)	Scen(a)	Scen(b)	Scen(c)
Benefits (see note 3)	2500	2500	1359
Capital cost	525	525	525
Cost of future repairs (see note 4)	5	147	56
Total cost	530	672	581
NPV of project (benefits less total cost)	1970	1828	778
Benefits-costs ratio (benefits/total cost)	4.72	3.72	2.34

Table A3. Option (3) scenario analysis.

Option 3 (in £m)	Scen(a)	Scen(b)	Scen(c)
Benefits (see note 5)	2500	2500	2500
Capital cost	525	525	525
Cost of future repairs (see note 6)	3	56	56
Capital cost in 20 years’ time (see note 7)	105	105	105
Total cost	633	686	686
NPV of project (benefits less total cost)	1867	1814	1814
Benefits-costs ratio (benefits/total cost)	3.95	3.64	3.64

Table A4. Option (4) scenario analysis.

Option 4 (in £m)	Scen(a)	Scen(b)	Scen(c)
Benefits (see note 8)	2500	2500	2500
Capital cost	700	700	700
NPV of project (benefits less capital cost)	1800	1800	1800
Benefits-costs ratio (benefits/total cost)	3.57	3.57	3.57

Notes

1. Option (1). For (a) and (b), benefits of £100m p.a. in perpetuity are worth 100/.04 = £2500m; for (c), benefits of £100m p.a. for 20 years are worth 100a₂₀ = £1359m.

2. Option (1). For (a), repairs costing £0.2m p.a. in perpetuity are worth).2/.04=£5m; for (b), repairs costing £4.1m p.a. for 20years and £8m p.a. in perpetuity thereafter are worth 4.1 a₂₀+(1/1.04²⁰)(8.0/.04)= £147m; for (c), repairs costing £4.1m for 20 years are worth 4.1a₂₀=£56m

3. Option (2). For (a) and (b), benefits of £100m pa. in perpetuity are worth 100/.04=£2500m; for (c), benefits of £100m p.a. for 20years are worth 100a₂₀=£1359m

4. Option (2). As for option (1), note 2

5. Option (3). Benefits of £100m p.a. in perpetuity are worth 100/.04= £2500m

6. Option (3). For (a), future repairs costing £0.2m p.a. for 20years are worth 0.2a₂₀=£3m; for (b) and (c), future repairs costing £4.1m p.a. for 20years are worth 4.1a₂₀=£56m

7. Option (3). Value of capital cost £230 in 20 years’ time=230/1.04²⁰=£105m

8. Option (4). Benefits of £100m p.a. in perpetuity are worth 100/.04=£2500m

Results

• Option (1) is the best option in scenarios (a) and (b) but a poor option in scenario (c).

• Option (3) is the best option in scenario (c)

• Option (4) is not the best option under any scenario

• In all four options, the benefits are worth much more than the costs in all scenarios

These results work out this way because of the particular figures chosen for illustration purposes, and different figures could lead to entirely different results.

A similar method was used in an illustrative case study about flood protection published by the IFoA, 2019).

Thus, the method can help in planning and analysing adaptive pathways in a variety of situations.