Experimental and Numerical Investigation of Dam Break Effects on Sedimentary Bed Morphology

Dams are one of the fundamental infrastructures in every country, which have been used and exploited for purposes such as water storage, flood prevention, and energy production. In the design and construction of large dams, various safety coefficients are always applied based on guidelines and codes, to reduce the risk of dam failure. However, dams are always subject to failure and the probability of dam failure cannot be reduced to zero only by considering these coefficients. The dam's failure can be based on various factors, including water leakage, cavitation (internal erosion), overpass due to insufficient overflow capacity, and liquefaction caused by earthquakes. Due to the amount of water stored in the reservoir of the dams and sometimes their proximity to the residential areas, the failure of the dams can lead to many human and financial losses. Dam failure mainly causes the transfer of bed sediments and morphological changes downstream. According to the presented statistics, the loss of lives in the failure of concrete dams (immediate dam failure) is twice of gravel dams (gradual failure). The main cause of this issue is the different nature and mechanism of dam failure based on its type, such that concrete dam failure is explosive-sudden and earth dam failure is gradual. Therefore, the type of dam break has been considered one of the effective parameters in evaluating bed sediment transport and bed morphology.

In this research, experimental and numerical studies have been carried out in order to investigate the phenomenon of bed sediment transport under two types of instantaneous and gradual dam failure.The experimental model has been implemented in the flume of Babol University Hydraulic Research Center and the instantaneous failure of the dam has been modeled by considering the sediment layer in the dam reservoir and downstream. Also, in these experiments, three types of sediment materials (sand d50=4mm, coarse sand d50=10mm, gravel d50=20mm) have been used as effective variables in bed morphology. Numerical modeling has been implemented in ANSYS software using the Fluent model. In the process of numerical modeling, the experimental models are simulated and two variables of the meshing size and the turbulence model are studied. After the sensitivity measurement of the numerical model and the model calibration, the performance of the numerical model is examined and the error rate of the numerical model is compared to the experimental results. Finally, scenarios have been defined for modeling the phenomenon of sediment transport under instantaneous and gradual dam breaks.

Experimental modeling has been carried out in the laboratory flume of the Hydraulic Research Center of Babol University, which has a flume of 10 meters in length, width and height of 0.5 meters. In order to perform numerical modeling, Fluent is used and it is calibrated based on two parameters of the mesh size (0.5, 1, 2 cm) and the turbulence model ((k-ɛ / Standard) and (k-ɛ / RNG)). In addition, the performance of the numerical model has been evaluated based on the results of the experimental model. Experimental modeling has been done based on two instantaneous dam failure models (models A1, B1) with the same thickness of the sediment layer and with different types of sediment materials (types A, B). The results show that in the area of the dam failure (around the position of the opening gate (level 1.5 meters long)) in both models, there was first a decrease in the sediment layer and then an increase in the sediment layer. It is also clear that in model A1, the amount of reduction and increase in the thickness of the sediment layer around the failure zone has greater values than in model B1.Based on the results of the sensitivity analysis of the numerical model with respect to the meshing size, it has been found that the 1 cm meshing has a modeling error of 2.41% compared to the laboratory model. In addition, the results of the sensitivity analysis of the numerical model with respect to turbulence models ( (k-ɛ / Standard) and (k-ɛ / RNG)) showed that the k-ɛ / Standard turbulence model had a modeling error of 2.75% and has been performed better. By examining the results of the numerical modeling, it has been found that the average modeling error is less than 5% (CA1 model has an error of 2.75%, CB1 model has a modeling error of 4.71%), which indicates a high accuracy of the numerical model.In order to numerically model the gradual dam failure in the Fluent model, a movable gate has been used. For this purpose, the movable gate, which has the boundary condition of an impermeable wall, is placed in front of the water reservoir of the dam, and based on passaging time, it leads to an upward movement. In the Fluent model, the moving mesh technique is used to implement the movement of the gate.The results show that the rate of bed sediment transfer and thus the reduction of the thickness of the sediment layer in the gradual failure of the dam had greater values than the instantaneous failure under the same modeling conditions. In other words, the thickness of the sediment layer under the same modeling conditions in Gradual failure has been reduced more than in instantaneous failure. In the gradual dam failure with the movement of the gate, the simulation has been done gradually with the passage of time, so the duration of water discharge as well as the interactions between water and sediment particles is more than the instantaneous failure.By evaluating the results of numerical models, it has been determined that the size of sediment particles (type of sediment material) has been an effective parameter in the changes of bed morphology in both instantaneous and gradual dam failures. The percentage of sediment layer changes in conditions of instantaneous failure compared to gradual failure for gravel materials (Type A material) has been equivalent to 7.47%, for sand materials (Type B material) equivalent to 19.35%, and for intermediate materials (Type C material) equivalent to 11.52%.