The development of a coarse sand barder as an effective measure against piping underneath dikes Le développement d'une barrière de sable gros comme mesure efficace contre Ie renard André Koelewijn, Vera van Beek, Ulrich Förster Dike Technology, Deltares, the Netherlands, andre.koelewljn@deltares.nl Adam Bezuijen Ghent University Belgium & Soils and Structures, Deltares, the NetherlandsZYXWVUTSRQPONMLKJIHGFEDCBA ABSTRACT: Backward erosion piping is a serious threat to dikes founded on sand. Along the main rivers of the Netherlands, it is one of the major issues. For several reasons, including a higher safety demand, this threat has strongly increased recently. Therefore the traditional measure consisting of a landside berm becomes too costly in term s of materials and land use w ith a length of tens of meters. Moreover, behind many dikes people live at a rather close distance to the dike. Vertical measures like sheet pile walls are even more costly, because of the long stretches of dikes that are potentially to be reinforced. W ith a thin but continuous barrier of coarse sand, covered by clay, installed undernedth an existing dike, a cheap and yet effective measure against piping is achieved, w ith the ability to withstand a hydraulic head that is considerably higher than the head that can be withstood without this measure for dikes of sizes common along the rivers in the Netherlands. This paper describes the research to be performed to quantify the increase in safety achieved w ith this measure, before field application at a large scale in the Netherlands can start. Tests at various scales are planned, varying from small-scale tests w ith seepage length of 0.35 m to large-scale tests using the full 9.5 m depth of the Delta Flume. RESUMÉ : Le renard est une menace sérieuse pour les digues fondées sur le sable. C ’ est un des périls plus sérieux le long des grands fleuves des Pays-Bas. Récemment, cette menace a fortement augmenté par augmenter le niveau de sécurité, le changement climatique et 1 ’ adaptation du modèle de Selhneijer. La mesure traditionnelle consistant en une berme devient trop coüteuse avec une longueur de quelques dizaines de mètres. Aussi, souvent des maisons doivent être démolies. Des mesmes verticales sont encore plus coüteuses en raison de la longueur des améliorations. Avec une petite barrière de sable gros installé sous une digue existante, couverte par 1'argile, une mesure pas cher et pourtant efficace contre le renard est atteint, avec la capacité de résister a une pente qui est considérablement plus élevée que sans cette mesure. La recherche pour arriver valider cette mesure est décrite. Des tests a différentes échelles seraient réalisés, variant des essais a petite échelle avec une longueur d'infiltration de 0,35 m a des tests a grande échelle avec une pente maximale de plus de 6 mètres. KEYWORDS: dike safety, backward erosion, piping, coarse sand barrier, filter, scale test, Delta Flume. 1 INTRODUCTION Backward erosion piping is an interna! erosion mechanism for water-retaining structures founded on granular layers. Shallow pipes are formed at the interface of a cohesive layer and a saridy layer, due to the removal of sand particles under the action of water flovv, starting at the downstream side and progressively moving towards the upstream side. This phenomenon is recognized by the formation of sand boils downstream of the structure, where the sand particles are being deposited in a ring around the boil. Ongoing pipe development can lead to severe erosion and finally failure of the water-retaining structure. The foundation that is susceptible to this mechanism, a combination of a uniform sandy layer covered by a cohesive layer, is often encountered below river dikes in deltaic areas. Numerous sand boils have been observed in the past, but failure due to backward erosion piping is quite rare. Nevertheless, several dike failures in the Netherlands, Italy, China and the U.S. are attributed to this mechanism (V rijling et al. 2010, Cirio et al. 2004, Calabresi et al. 2013, Camici et al. 2015, Yao et al. 2009, USACE 1956). In the Netherlands, safety assessment for backward erosion piping is conducted w ith the Selhneijer model (Sellmeijer 1988, Sellmeijer et al. 2011, TAW 1999). The Sellmeijer model predicts the head difference across the structure that once exceeded w ill lead to ongoing pipe formation. The progression of the pipe is considered the critica! stage in the entire process. In other words, once the pipe has progressed along a critical length under the water-retaining structure, it w ill lead to breaching when hydraulic loads sustain. This model predicts the critical head on the basis of the groundwater flow towards the pipe, the viscous flow through the pipe and the lim it-state equilibrium of particles at the pipe bottom. The model has been validated using experiments, but application in the field proves to be complex, as required parameters like partiele size and permeability are difficult to determine and show large fluctuations in the field. The uncertainty w ith respect to input parameters leads to the selection of safe estimates, such that considerable dike reinforcements are due. The more stringent safety standards, due to a recent validation of the model (Sellmeijer et al. 2011), the inclusion of the length-effect (Kanning 2012) and the risk approach, recently embraced in the Netherlands (Kind 2014, Van der Most et al. 2014), lead to a further increase of the dike length to be reinforced. In the Netherlands, reinforcements for backward erosion piping are traditionally conducted using berms or cut-off walls. Berms are becoming less attractive w ith increasing required seepage lengths and when houses are situated closely behind the dikes. Sheet pile walls are an alternative, but are economically unfeasible when it comes to application for long dike stretches. Innovative or alternative piping measures are therefore becoming more attractive. An example of an innovative measure is the vertical sand-retaining geotextile (Bezuijen et al. 2014, Förster et al. 2015). Using this method, nearby the toe of the dike a vertical geotextile is inserted into a trench. Above the sand layer the trench is refilled w ith clay, such that upvvard seepage is not possible. An optimization of this innovative solution is proposed: a coarse sand barrier. In this solution the pipe formation is resisted by coarse sand instead of a geotextile. - 1295 -