NEURAL NETWORKS FOR PREDICTION OF SCOUR DEPTH AROUND BRIDGE PIERS

 NEURAL NETWORKS FOR PREDICTION OF SCOUR DEPTH AROUND BRIDGE PIERS

The depth of scour is an important parameter for determining the minimum depth of foundations as it reduces the lateral capacity of the foundation. It is for this reason that extensive experimental investigation has been conducted in an attempt to understand the complex process of scour and to determine a method of predicting scour depth for various pier situations. To date, no generic formula has been developed that can be applied to all pier cases to determine the extent of scour that will develop.

 The mechanism of flow around a pier structure is so complicated, that it is difficult to establish a general empirical model to provide accurate estimation for scour. Interestingly, each of the proposed empirical formula yields good results for a particular data set, an alternative approach, artificial neural network (ANN) has been extensively used to estimate the equilibrium and time dependent scour depth with numerous reliable database. Numerous ANN models, multi-layer perceptron using back propagation algorithm (MLP/BP) and radial basis function using orthogonal least-squares algorithm (RBF/OLS). Bayesian Neural Network (BNN) and Single Artificial Neural Network (SANN) were used.

 The equilibrium scour depth was modeled as a function of five variables; flow depth, mean velocity, critical flow velocity, mean grain diameter and pier diameter. The time variation of scour depth was also modeled in terms of equilibrium scour depth, equilibrium scour time. Scour time, mean flow velocity and critical flow velocity. In all the published works, the training and testing data were selected from the experiments and from valuable references.

Scour is a local phenomenon that takes place in the vicinity or around a structure (piers, piles, abutments etc.) in flowing water, due to modification of flow pattern, results in increase of local shear stress. This, in effect, dislodges the material on stream bed, results in local scour.

River flow past a pier or an abutment causes three-dimensional flow separations-a system of vortex pairs developed in separated flow. Between them, the primary vortex is more dominating, wraps round the pier in the form of horse shoe-vortex. The magnitude and strength of this vortex depends on the geometry of pier and the magnitude of approaching velocity.

The estimation of correct depth of scour below the stream bed is very important, because it determines the depth of foundation. The phenomenon of bridge pier scour is of paramount concern to hydraulics engineering profession, because without this detailed knowledge, bridge failures can occur. As per National Bridge Register (NBR) of America, out of 577,000 bridges, more than 26,000 of them have been found to be scoured critical, due to erroneous prediction of scour depth during engineering design. In this context, Indo-Gangitic belt of the Indian sub-continent is interwoven with mighty rivers like Indus, Ganga, Brahamaputra and their innumerable tributaries. The alluvial is so deep that in some cases even up to a depth of 100 metres no rock strata are found. Moreover the river beds are highly errodable. In order to protect the bridge piers against scouring the foundations have to be taken very deep. In the bridge across river Ganga at Varanasi, the maximum depth of scour estimated was around 60 m.

Artificial Neural Network (ANN) models are attractive in the domain of estimation of local scour around bridge piers. This is due to their adoptive nature where learning by examples replaces or making functions in search of solutions. This architecture renders computational models more attractive in domains of very little or incomplete understanding of the problem to be solved but where broad training data base is accessible. From the literatures it appears that ANNS provide higher level of accuracy in solving a particular problem in comparison with experimental and theoretical results. ANN may therefore be a viable alternative in the prediction of local scour depth around bridge piers, provided reliable data base is available.

 

Factors Influencing the Local Scour Depth

 

Approach velocity

 

Ø  Under clear water condition, the local scour depth increases with velocity (Ettema 1980) till it reaches a maximum value at critical velocity.

Ø  When the approach velocity exceeds the threshold value, the problem becomes a live-bed problem and scour is about 10% less than the threshold depth (Laursen 1963).

Ø  Meliville (1988) has shown that as the velocity exceeds the threshold value, the local scour depth first decreases and then increases.

 

Flow depth

 

Ø  At shallow depths the local scour around piers increases with flow depth, but as the depth increases, the scour depth becomes independent of flow depth (Laursen 1966, Breuser et al. 1977. Ettema 1980).

Ø  According to Malville (1988), the local scour depth is independent of depth of flow, so long as the two vortex rollers do not interfere with each other.

Ø  Since the vortex strength increases with depth of flow may result in more scour depth.

 

Sediment size

 

Ø  The maximum value of the clear water local scour depth in non-ripple forming sediments is unaffected by sediment size as long as the ratio of the obstruction to the median size of the sediments (D/d50) is greater than or equal to 50 (Ettema 1988).

Ø  However, d50 is limited to single size particle, but as d50/h increases the vortex strength also increases.

 

Pier shape

 

Ø  Blunter the pier greater is the scour, when it faces the flow.

Ø  Very little significance is attached to the shape on the down stream of the pier.

 

Pier size

 

Ø  The relation between pier size and equilibrium scour depth are quite diverging. Keeping all the factors constant, the scour depth varies as DB. B varies from 0.5 to 1. Pier size

 

Angle of attack

 

Ø  As the angle of attack increases so also the scour depth for all shapes expect cylindrical pier.

 

Constriction ratio

 

Ø  The ratio of width of the flume to the size of the (B/D) influences the equilibrium scour depth. Shen et al. (1969) suggested that, for clear water experiments, the flume width should be at least eight times the diameter or the size of the pier. The same ratio for live-bed scour should be at least ten times the pier size (Chiew 1984).

 

Sediment grade

 

Ø Both the sediment grade and the equilibrium scour depth decreases with standard deviation of the particle size distribution (Ettema 1980, & Garde, 1995). This is due to the formation of armour layer at the base of the scour hole.