Get updated about news, events and contributors experiences

Reflection: Risk-based inspection and interactions between failure mechanisms

Posted at 04/10/2021 by Matthijs Kok

Safety standards of flood defences are based on probabilities of flooding. Using a variety of failure models and statistical methods, we try to assess these probabilities as good as possible, in order to take effective measures for flood risk reduction. The question is how well we actually have insight in the flooding probability, and whether we take all relevant factors into account.

Contact details Matthijs Kok

Delft University of Technology

In this webinar we consider 2 aspects that, up to now, have not been explicitly accounted for in flood probability estimates: influence damage of the revetment due to for instance animal burrowing and interactions between failure modes that might lead to failure behaviour that deviates from currently considered behaviour. How can we account for these effects? And how does it influence the practical action perspective?

This reflection emerged from the discussion between the webinar team and 24 participants to the All-Risk webinar organised in October 4, 2021.

Would you like to view the presentations of the webinar team? You can view them in the above video.


The presentation discussed the accuracy of flood defence inspections, which has been quantified in a field test with inspectors from the Rivierenland Water Authority. This clearly shows that not all damages are observed in inspections and that this has to be accounted for when estimating failure probabilities of flood defences. In the considered case, the influence of damage on the system failure probability was quantified: this results in a significant increase and demonstrates that existing inspection policies are often insufficient to meet the requirements. With targeted interventions, the influence of damage on flood defence reliability can be reduced significantly. This has been substantiated using a multi-objective analysis of total cost and structural robustness. The latter is a new and useful indicator to quantify how design, inspection and maintenance influence the impact of damage of revetments on flood defence reliability.

What comes to your mind when you think about inspections?

What comes to your mind when you think about inspections?

There was a broad agreement among participants on the importance of accounting for damage to flood defences. This extends to design considerations, although it is recognized that this is often difficult. The proposition that every flood defence design has a (risk-based) inspection and maintenance manual has broad support, although it is not yet clear what such a manual should contain exactly. There are important uncertainties, amongst others on the exact influence of damage on strength. The proposition on doing physical tests (e.g., in the Delta flume) gained mixed support, primarily because the starting point should be the risk contribution of the uncertainties and not the physical unknowns. It is widely recognized that there is not much knowledge on the influence of damage to flood defences yet.

Future research on the impact of damage should primarily focus on large-scale burrowing related to internal erosion and slope instability, and damage to grass revetments. The key reason is that this is relevant for a very large part of the flood defences, which makes the potential implications on a national scale very large. Similar reasons hold for pattern-placed revetments, which are seen as important as well. The relatively small portfolio of asphalt revetments and toe structures leads to a lower priority for research.

What about the interactions between failure mechanisms?

The presentation first explained what interactions between failure mechanisms are, and in which cases they may affect the flooding probability. This is mainly the case when the residual resistance is eliminated (or reduced) by the occurrence of a more frequent initial failure process. Thus research on interactions appears to be related to research on so-called ‘residual resistance’ or ‘residual strength’, that is strength that is not yet taken into account. An analysis of residual resistance (and thus interactions) is always tailored, for instance using failure scenarios or integral reliability analysis.

What comes to your mind when you think about the interactions between failure mechanisms?

During the discussion it was recognised that this kind of analysis is sometimes needed to move from somewhat conservative analyses based on failure initiation towards more accurate reliability estimates. About 2/3 wants to work with interactions; from keeping it in mind during residual resistance analyses to a full quantification. On the other hand, people note that it involves more work than standard analyses and that there is still much unknown about the underlying processes.

Would you like to do something with interactions in the assessment of flood defences?

Therefore, there is a particular need for knowledge development: how exactly do these interactions work? That is difficult to imagine as few such cases have been reported in detail. It would be useful to have a description of failure cases in practice where multiple mechanisms play a role. In addition, part of the group could use a practical guide or calculation examples.


All-Risk recommendations:

  • Explicitly take into account the accuracy of inspections and maintenance in design (choices).
  • Further research into the effect of, for example, damage to the grass cover must be risk-based: only when significant risk reduction is expected, experimental research (e.g. in the Delta flume) make sense.
  • An important part of the above risk reduction is the size of the area affected by the damage: it is therefore recommended to look primarily at the consequences of large burrows and damage to grass and revetments.
  • Take interactions into account when applying residual strength, quantitatively or to inform decisions to which degree residual resistance should be included.
  • Further knowledge development is needed in the form of case descriptions of breaches (e.g. via TU Delft ILPD data base), knowledge of failure processes after initiation, and quantification of the effect of interactions on the failure probability for several practical cases.

Related updates

Last modified: 13/12/2021