Several cases of backward erosion/liquefaction piping from Hungary

Experience to date (see e.g., [1 to 18]) shows that during major floods, hundreds of sand boils may develop, but only a small percentage may need significant response measures. Only a small fraction of these have caused breach disaster. We still not know these processes very precisely. The backward erosion/liquefaction pipings may develop in fine-grained soil with no cohesion, which is poorly graded and extremely loose. It is called „liquefiable sand” or „liquid sand" in Hungary. The current name of it is fine silty sand according to soil classification; earlier, it was called as sand flour (Mo). It can be noted that this sand is simultaneously internally unstable on the basis of the grading entropy criterion and liquefiable on the basis of the well-known criteria [10, 18]. In thin layers of poorly graded sand, the increasing flood water level may lead to the usual sand boils, which can be mitigated by counterpressure. The event tree can be seen in Fig. 3-35. Remedial measures, sandbags ringing the sand boil can be effective as emergency response measure on condition a protocol is followed to keep counter pressure. At the end of the flood, first, the sand material of the ringing sandbags are used to fill the pipe, some additional measures are made afterwards. In case of thick layers of loose, poorly graded sand, the flood may lead to ‘fast’ piping, failure within seconds minutes after the observation of a sand mud geyser, and a simultaneously appearing vortex funnel in the river which is moving towards the dyke. The reason of the breach is liquefaction. The pipe in the waterside is formed along with the least resistance under the minimum energy principle before the breach. Szepessy, J.(1983) [11] gave first the interpretation of the fast piping events based on case studies 4 and 5 (happened in 1954, 1965, 1980), suggesting the hypotheses that liquefaction may take place. In these cases, the breach occurred within minutes/seconds after the appearance of a mud geyser and a simultaneously appearing vortex funnel moving towards the dyke in the river. In this work, it is found that case study 3 (happened in 1926, Fig. 1) has a similar failure scenario and failure path (see Fig. 3-35) as in case studies 4 and 5, except that in the 1926 case study, there is an early stage of the flood with usual sand boils, too (which probably may occur if no plastic soil cover is present).