In granular soils (sand and silt) loss of drainage can transform a ductile mechanical behavior into a fragile one. This transformation is usually rapid (minor change in drainage and / or load conditions) and trigger large scale instabilities. This is all too often the case in hydraulic landfills for port, coastal, hydro or mining works. Detecting the potential instability of a soil requires an identification criterion. It has been proven that such a criterion folows considering the soil state, which is defined by combining mean stress and porosity. Thus, potentially unstable situations are given by the distance of the soil state "in situ" to its critical state (fluidization). Measuring "in situ" state quickly and accurately becomes crucial. The most appropriate means of carrying out such a measure is by field testing, since, unfortunately, soils susceptible to instability are practically impossible to sample intact. The piezocone (CPTu) is most suitable due to the wealth of data it offers, its speed and repeatability. There are currently relationships between piezocone measurements and the "in situ" state, but they have important limitations, (eg they are not applicable to silts). This project uses numerical simulation to put such relationships on a firmer foot and extend its field of application. The basis is an advanced numerical method (PFEM) that simulates cone penetration tests on materials described with a model in which soil state is explicitly represented. In the case of silty sands and silts, the granulometry strongly affects both the mechanical properties (stiffness, resistance) and hydraulic properties (permeability). There are no data under controlled conditions that allow to separate both effects on the CPTu. It is proposed to generate such data through discrete simulation (DEM). This will take into account the case where granulometry changes during the test, as fragmentation occurs. This difficult case is not uncommon on unstable soils (eg mining waste). Once the risk of instability has been detected, alternatives must be evaluated. In new works the origin of materials or the construction method can be changed. When assessing a natural terrain or legacy works the situation is more complex. Stabilization works are expensive and generally dangerous, since they can trigger the same problem that is sought to avoid. Having a numerical tool to assess the consequences of the instability that different actions can generate is advantageous. A priori such a tool may be the same PFEM code that is developed in the project to simulate cone penetration. In order to rely on it, it must be validated in a different problem, and this will be also done in the project: first in a case of instability of slopes in the laboratory, with a great abundance of data on the moment of rupture and in a relatively simple material; then in a real case - the rupture of the Prat quay wall. The project is based on our experience with the two numerical methods (PFEM and DEM) and in our general experience in applied geotechnical simulation. The project is supported by external experts who will provide data and specific know-how and with young researchers who will be formed in a project that mixes scientific novelty and industrial interest.