A proposed tenfold increase in flood-protection standards across the Netherlands was estimated to require €11.5 billion in dike improvements. Research combining economics and optimisation showed that stronger protection could instead be concentrated in three regions, reducing projected investment costs to €3.7 billion while directing investment towards areas where the expected benefits were greatest.
Why did the Netherlands need to rethink its flood-protection standards?
Flood protection affects a large part of the Netherlands. Around 55% of the country’s land area is exposed to flood risk, covering about two-thirds of the population and 70% of GDP.
Many of the legal standards governing that protection could be traced back to decisions made after the 1953 flood disaster. A flood-protection standard sets the maximum acceptable probability of flooding for a protected area. A dike ring is an area surrounded by connected dikes, other flood defences or naturally higher ground.
By the 2000s, population and economic activity had changed substantially, increasing the potential consequences of a flood. In 2008, the Second Delta Committee therefore recommended increasing protection standards at least tenfold throughout the country. Implementing that recommendation was estimated to require €11.5 billion in dike improvements.
The question was whether the same increase was necessary everywhere.
Why was this a prescriptive analytics problem?
Estimating flood risk is only the first part of the problem. Policymakers ultimately need to decide where protection should be strengthened, when investments should take place and how much should be spent.
Those decisions involve competing objectives. Raising a dike reduces expected flood damage, but construction also uses public resources. Different regions have different populations, economic activity, flood probabilities and potential consequences if protection fails.
The project therefore combined cost-benefit analysis, which compares the cost of an investment with the losses it is expected to prevent, with mathematical optimisation. Optimisation allows many possible investment strategies to be compared systematically and identifies those that best balance protection and long-term cost.
Later research extended the same question over time; not simply how safe should an area be, but also when should a dike be raised and by how much?
How did academic, industry engineering experts, and government partners develop the solution?
The first stage brought together the CPB Netherlands Bureau for Economic Policy Analysis, Deltares, Tilburg University, HKV Consultants, Delft University of Technology, the Ministry of Infrastructure and the Environment and the Delta Commissioner’s office.
The collaboration combined economic modelling, flood-risk expertise, optimisation and public-policy knowledge. Instead of treating all protected areas in the same way, the models considered differences in investment costs, potential flood damage and the level of protection needed in individual regions.
The team also tested how sensitive the results were to uncertain assumptions about economic growth, climate change and flood damage. Although the exact economically efficient standard varied under different assumptions, the relative priority of different regions remained comparatively stable.
The research subsequently continued through overlapping collaborations. Work on optimal long-term dike investment involved CPB, Tilburg University and Delft University of Technology, while later research on climate uncertainty brought together Erasmus University Rotterdam, Tilburg University, Deltares and ORTEC.
What was the impact of the projects?
The teams found that increasing protection standards tenfold across every dike-ring area was not economically justified.
Instead, stronger standards were recommended for three regions: parts of the Rhine and Meuse river areas, southern Flevoland around the growing city of Almere, and several dike-ring areas near Rotterdam.
The estimated investment changed from €11.5 billion under the nationwide proposal to €3.7 billion under the targeted strategy — a difference of approximately €7.8 billion.
The three priority regions also accounted for around two-thirds of the expected benefits of the proposed improvements.
For people living in flood-prone areas, the significance is not simply lower public spending. The approach was designed to strengthen protection where the expected consequences of flooding justified greater investment, while avoiding the cost of applying the same increase in places where the additional benefit was much smaller.
How did the research influence Dutch flood policy?
The findings became part of a wider government decision process rather than remaining an academic recommendation.
In 2012, the Dutch state secretary responsible for infrastructure accepted the main results as a basis for legislation and concluded that a tenfold increase was not required across all dike-ring areas. Parliament and government continued developing the standards through the following years.
The subsequent optimisation research was applied across Dutch dike rings. The resulting standards were published in the Delta Programme 2015 and accepted by the House of Parliament in late 2014.
The updated national standards were later accepted by the Dutch government and became legally binding in 2017.
The impact therefore developed through a clear chain. First, analytical research informed flood-protection standards. Those standards entered national policy, and the policy now guides long-term investment in Dutch flood defences.
How did the developed solution evolve from setting standards to informing public expenditure?
Once economically efficient protection standards had been established, another question remained: when should investments actually happen?
The next stage of the research improved an earlier Dutch model developed by economist David van Dantzig after the 1953 flood. The newer approach accounted for economic growth, recognising that the potential consequences of flooding can increase as populations, infrastructure and economic activity grow.
It also calculated both the timing and size of future dike upgrades. For some forms of construction cost, the researchers found that the most efficient strategy followed a periodic pattern. The solution was for dikes to be raised by an optimal amount at calculated intervals rather than through ad hoc decisions.
This provided a connection between long-term economic planning and the legal standards used to determine when local flood protection needs to be improved.
Why does uncertainty about sea-level rise change the decision?
Long-term flood investments face another problem, namely that future sea-level rise is uncertain.
A plan based on one forecast can perform poorly if conditions develop differently. But committing immediately to the most extreme possible scenario can also lead to unnecessary spending.
Researchers from Erasmus University Rotterdam, Tilburg University, Deltares and ORTEC therefore developed an adjustable robust optimisation approach.
In plain language, robust optimisation tests whether a strategy remains workable across a range of possible futures. The adjustable part means that decisions made decades from now do not have to be fixed today. Later investments can change as more information about sea-level rise becomes available.
Applied to the Rhine Estuary–Drechtsteden area, the model found that allowing for substantial uncertainty did not necessarily require a proportionally large increase in spending.
Accounting for 40% uncertainty in sea-level rise increased average total project costs by less than 10% — about €130 million in the reported experiment. Plans that considered uncertainty from the beginning were also less costly than repairing an optimistic plan later when less favourable conditions occurred.
For communities protected by these defences, this means long-term safety does not have to depend on one climate forecast being correct. Investment can be designed to remain safe while retaining the option to adapt as conditions become clearer.
What can this mean for future flood protection?
The research teams showed how analytics can support several stages of the same public decision:
Where should stronger protection be concentrated? How much should be invested? When should infrastructure be upgraded? And how should those plans change when the future is uncertain?
The Dutch models were developed around local geography, risks and policy requirements, so their specific recommendations cannot simply be transferred to another country.
But the underlying approach can be adapted. Other flood-prone regions can combine their own information on populations, infrastructure, flood probability, construction costs and acceptable risk to compare long-term investment strategies.
The earlier research has already prompted interest beyond the Netherlands, including discussions about flood-protection standards in other countries, and there is potential for wider applications in other long-term environmental investments.
The broader shift is from treating flood protection as a fixed engineering target to treating it as an adaptive public-investment decision where resources can be directed where they create the greatest protection today while keeping future choices open as climate conditions change.
