Increased flood safety due to time-dependent pipe growth
Dike safety assessments can be optimised when considering the time that the water flowing underneath the dike takes to erode its foundation.
Piping or the erosion process in the dike foundation takes time and may not evolve into a dike breach and flooding during a high-water event. Quantifying the time-dependent aspects of piping helps the responsible authorities and technical advisors to reduce the scope of reinforcement efforts and that way meet safety standards faster and cheaper
Piping is a gradual erosion process. It starts when water flowing through a sandy dike foundation erodes so much sand that it forms a small (mm size) channel or ‘pipe’ that may grow (or not) into a shortcut towards the river. Dike safety assessments and designs neglect the time piping takes to develop. Without considering the time development of piping, the dike failure calculations may be too conservative, particularly if:
- high-waters are relatively short, or
- the piping erosion can be stopped with timely flood fighting interventions.
Illustration of the piping erosion processes.
Piping is responsible for a large part of the reinforcement costs of the Dutch dikes
Large parts of the Dutch dikes may need to be reinforced with respect to piping for meeting the safety standards (target probability of dike failure shown in the map).
Source: The Dutch primary dikes were retrieved from the National Georegister website. The sand boil observations are available via the https://wellocaties.app/
Historically, sand boils mainly occur in the river area (see the storm surge - river flood area division in the map), raising the question of whether piping is a risk along coasts. Although the number of sand boil observations may be lower in the coastal area, one of the two dike breaches in the Netherlands attributed to piping is located along the coast (Strijenham breach near Tholen).
River vs coastal flood duration
One factor that influences piping is the flood duration. Riverine floods are typically caused by long-lasting high water levels (weeks). Instead, storm surges at the lakes and coast are typically short-lasting (days). In coastal areas, the development of piping is limited by the short events. In river areas, there is more time available for pipe erosion, which in turn may also give extra time for flood fighting by, for example placing sandbags.
If we can predict how fast piping develops, we can optimise reinforcements and inform flood fighting operations. However, considering or (not) the time factor into dike assessments and design practice may depend on the case. Therefore, I made the following research question part of my PhD research at the TUDelft and the All-Risk programme :
How much does dike safety improve for river and coastal dikes, when considering the time piping takes to develop?
Studying piping under short-lasting high water levels and with timely flood-fighting interventions
By using three complementary methods, I investigated the time-dependant aspects of piping:
- Lab experiments at small (50 cm) and large (10 m) scales increase our understanding of the piping processes in the pipe (mm size), reveal which factors influence the time scale of erosion, and help validate the below pipe development models.
- A simplified pipe development model helps predict piping for dike properties and water levels beyond the conditions of the lab experiments.
- Probabilistic methods allow to include uncertainties of, for example, the high-water duration, the soil properties and the seepage length to determine whether a particular dike meets the safety standards.
Experiments on pipe development. Top left: large-scale test in the Flood Proof Holland facility at the TU Delft. Top right: measurements during the large-scale test. Bottom right: sand boil in the large-scale test. Bottom left: small-scale experiments.
The factor time in the piping failure process
The animation on the left illustrates the development of the pipe length during a high water event. Important factors that determine whether a pipe develops into a breach during a high-water event are:
- The pipe length which is already present before a high-water event occurs.
- How early in the event pipe growth starts. This is related to the timing of clay layer cracking (uplift) and sand boil formation (heave).
- The rate of pipe progression, which increases with increasing grain size and higher water level.
- High water level duration (short storm surge vs. long river floods) determines the time available for erosion.
- The effectiveness of flood-fighting interventions by considering their detection error and required time for the sandbags placement.
Six hypothetical cases
To find out whether the time development of piping makes a difference in the river and coastal dikes, I formulated six cases in which I varied the parameters which one might expect to have a significant influence on the calculated safety level and that way on the required dike reinforcement.
The six cases are hypothetical dikes, four for a river dike and two for a coastal dike. In all cases, the seepage length is 50 m, with the conservative assumption that a pipe is already present up to 1/2 the seepage length.
- For both coastal and river cases, I varied the grain size from fine sand (0.180 mm) to medium-fine sand (0.350 mm).
- The high-water duration varies between the extreme cases of storm surge (days) and Rhine river discharge (weeks).
- For the river case, I consider two scenarios for flood fighting effectiveness with sandbags (detection 50% or 90% successful). In coastal cases, flood fighting interventions are not considered as these are hardly possible during a storm.
When does it make a difference?
For each case, I calculated the probability of failure with and without considering the timing of pipe development. The ratio between these probabilities indicates the difference between considering or not the timing of piping development into the dike safety assessment calculations. The higher this factor, the larger the increase in dike safety.
These results indicate that the high water duration has the largest impact. In the river cases, there is hardly a change in failure probability because the high water lasts much longer than the time required for pipe erosion, but in the coastal cases it is highly unlikely that piping leads to a breach during a single storm. The coastal cases show that the slower erosion in fine sand compared to medium fine sand can make a large difference. Regarding the river cases, the results show that a significant safety increase can be achieved in case of effective flood fighting (90% successful detection, 10 hours placement time).
Lessons learned
Considering the time needed for piping development can significantly reduce the failure probability. Not only in coastal areas with short storms but under certain conditions, such as effective flood fighting, also in river areas.
The comparison between the six hypothetical cases with still some conservative assumptions already provide great insights under which conditions the time of piping development is a relevant factor, and by which order of magnitude it affects the failure probability.
Finally, including the factor time for piping requires a shift in thinking: there is not just a critical water level, but combinations of peak water level and high-water duration that gradually lead to failure.
Next steps
As this research is still ongoing, I currently use the experimental data to improve the physical basis of the simple pipe development model. Furthermore, I am studying the effects of pipe development under multiple high-water events. Those findings may allow to loosen the conservative assumption that there is a pipe present already up to 1/2 the seepage length.
A useful step for practice would be to use the research findings to derive simplified rules to include the effects of time in dike safety assessments. For instance, a reduction factor on the computed failure probability, depending on the governing high water duration, seepage length and foundation soil.
Since flood fighting appears to be an important factor in the river area, a discussion is needed when we want to use that information in dike design. If we want to include it, the effectiveness (detection accuracy and required time) of interventions needs to be quantified further and improved.
Interested to know more?
- Progression rate of backward erosion piping in laboratory experiments and reliability analysis (Publication on earlier version of the reliability analysis)
- Temporal Development of Backward Erosion Piping in a Large-Scale Experiment (Publication on large-scale test)
- Please contact me for more information; the work is still in progress and a publication under preparation.
