On the Evaluation of Indirect Simulations Performance of Multi-parametrical Transient Seepage Models in River Embankments

The design and vulnerability assessment of earth-filled water retainingstructures require the elaboration of reliable seepage analyses. Model predictions,however, can result highly unreliable because of poor parameter estimations. Toaddress this issue, inverse modelling techniques can be implemented, incorporat-ing observations of on-going processes into direct simulations. With reference toriver embankments, time series of the monitored soil water content and pore wa-ter pressure can be helpful for a proper calibration of hydraulic and retentionproperties of the materials involved in the seepage problem. For the present study,the Levenberg-Marquardt algorithm implemented in the commercial code Hy-drus 2D by Pc-Progress has been applied for the optimization procedure of aninstrumented river embankment section along the river Secchia (Modena, Italy),making use of observational data. The seepage analysis is carried out consideringvarious assumptions on the series of monitoring data, weighing distributions andoptimized parameters combinations. The study aims to analyze an effective ap-plication of combined qualitative and quantitative methodologies for the perfor-mance evaluation of indirect simulations in the transient multi-parametricalmodel under investigation, in order to estimate the optimal set of parameters as afunction of the pursued aim of the modelling.