Data-driven dike reinforcements
In addition to traditionally used data sources such as drillings, cone penetration tests, and statistics of extreme water levels, recently interest has increased for long-term monitoring of groundwater levels, historical data, performance information, and combining data from different data sources. One of the data sources that have not been used so far is archaeological data of the dike buildup. By using the properties of historical dikes, an estimate can be made of the dike buildup of current dikes. This information can subsequently also be used for groundwater and dike stability schematisation, which is crucial to assess dike safety in more detail. Another component is the use of information on the observed performance of dikes to improve failure probability estimates. Examples of performance information are survived loads such as high water or test loads and measurements during such load situations.
Participants responses to the question: What terms come to mind when you think of "data-driven dike reinforcement"?
Furthermore, there is already a large toolbox of existing techniques to interpret different data sources. A combination of all types of data would probably give a complete picture, but this is an impossible task in practice. Thus, priority must be given to the most effective data types.
Reflections on the use of two new data sources: archaeological dike cross-sections and test loads
Two new data sources are discussed: Archaeological dike cross-sections, to understand the dike structure better, and proven strength by test loading of dikes to reduce uncertainty in strength.
Participants responses to the question: To what extent do you find the following aspects a limitation to the use of archaeological data in dike assessment?
According to the participants, the use of archaeological dike cross-sections can be implemented in the Wettelijk BeoordelingsInstrumentarion (WBI, for assessment). The WBI is mainly seen as a basis for safety assessment, a kind of recipe. People already look much further than the WBI rules in many situations, and historical data also fits in. In addition, it would be nice if standard classification tests were done more often when soil research is done. If these tests are used in a wide range of types of research, such as geotechnical and historical research, different types of research can be combined more easily, and better conclusions can be drawn about the behaviour of soil. Such a classification standard of Dutch soils based on basic tests has already been developed and would be a good addition to the historical data.
Many participants feel that test loads are a promising method to determine dike safety better but are concerned about possible risks such as the ‘damage’ that a test load may cause. This concern is remarkable because a test load’s cost-benefit analysis may explicitly include this risk. A risk-neutral decision-maker would always decide to carry out a test load if the benefits of reinforcement cost reduction are larger than the costs and risk related concerns to the test load.
The fact that participants label test loads ‘undesirable’ and identify ‘damage’ as the greatest risk may indicate that decision-makers are slightly risk-averse. Many participants also find it difficult (administratively) to explain when a test load results in some damage. In addition, they find it difficult to have to explain that the dike is weaker than previously assumed. This concern is also remarkable because finding a weak spot is very valuable, especially if that spot would otherwise have gone unnoticed.
Optimisation of data use in dike reinforcements
The new data sources, historical dike data and test loads are just two of the many possibilities for gathering more information about our dikes. When we present these various possibilities to the attendees, long-term monitoring is seen as the best option for optimising dike reinforcements. However, we often see that all the information after a dike reinforcement disappears into storage and is not used to adjust the safety image continuously.
Long-term pore water pressures are in particular seen as data that is almost always lacking at present, making good uncertainty estimates very difficult. However, the influence of the uncertainty of pore pressures is not necessarily the greatest uncertainty in calculations.
The participants clearly prioritise how to improve flood risk management (see Figure). (Long-term) monitoring is very important, as well as full probabilistic safety assessments. Full-scale experiments are also important. The use of historical data is less attractive than risk-driven research because it is hampered by its availability and difficulties regarding adequate data descriptions for further reuse.
Conclusion and outlook
Whatever data sources have been used, there is a unanimous view that data should be (re)used as much as possible in the broadest possible application of projects. The ideal picture of the future is that current and historical data is used to continuously adjust the probability of failure and a residual lifetime of flood defences, and thus to be ‘in control’ regarding flood safety. To this end, it is desirable to have all data on current and historical dike reinforcements publicly available. An important point of attention here is proper access to the data to be present and easily accessible. The improved data management requires good agreements between the client (Water Authorities) and the contractor (consortium/contractor) about which data are relevant or can become relevant in the future.
Participants responses to the question: What do you think are the most important steps to improve flood risk management?
Since it is expected that we can benefit a lot from long-term monitoring of dikes, this aspect deserves extra attention. Relevant knowledge questions are:
- How long do we have to monitor to derive reliable characteristics? and
- How can we specify in advance how new data and insights will affect the dike reinforcements? so that this vision remains constant over the longer term, independent of changing project managers.
A misconception is that the behaviour can be captured generically in simple calculation rules with this long-term data. Although monitoring will clarify the behaviour, it remains more important to understand and interpret the situation in geotechnical engineering.
To continuously update the remaining lifetime of dikes, we also need a good understanding of how soil and soil behaviour change over time and the effect on the probability of failure. For example, an increasing probability of failure due to cracking and a decreasing probability of failure due to ageing of soil and proven strength. The impact of future climate change also plays an important role. Together with the existing monitoring methods and new developments, the dike safety specialist’s toolbox will contain more and more possibilities to use data to estimate the reliability of dikes better.
- Use innovative data to assess dike safety. This can provide many new insights and improve the assessment.
- Commit to long-term monitoring of dikes.
- Make dike data available for wider (re)use.
Last modified: 07/01/2022