| Summary: | To date the influence of bed roughness onl the propagation and dynamics of gravity currents has been largely neglected. A new physical modelling dataset has been compiled, which details the fundamental affects of several bed roughnesses on lock-release gravity currents. Five bed configurations were chosen encompassing 'grain' and 'form' type elements at a range of spacings. 1%, 5% and 10% initial density excesses were studied and the effect of removing the buoyant ambient fluid between the elements examined. Observations due to changing the current depth relative to the element height were also made. Ultrasonic Doppler velocimetry profiling (UDVP) and video capture techniques were used to analyse stream wise and vertical velocity structures and the affects on the front speed and distance travelled by the current. A 10 depth-averaged model solves modified 2-layer shallow water equations using the method of characteristics to obtain temporal velocity and depth evolution for a current under the influence of a general roughness quantity. 2D and 3D depth-resolved CFD simulations use the commercial software FLUENT to solve the RANS equations and transport of a scalar for the dense current with the RNG k - € turbulence model. The CFD predictions were well validated by the new experimental dataset and provide supplementary predictions of concentration, lateral motion and activity in the vicinity of the roughness elements. Comparison of 20 and 30 models resulted in the conclusion that the 3D model is vital for accurate simulation of internal dynamics of gravity current propagation over beam type bed roughness. In general general, the distance that the front travels decreases with any bed roughness present. This reduction increases with element spacing. The stream wise mean velocity profiles show a reduced velocity maximum further from the bed. Decreased entrainment results from breakdown of larger billows. Also observed is a thicker current, a rounder profile and a shorter, diluted head. Areas of increased vertical motion within the current. associated with decreased horizontal motion are observed, indicative of ejections of ambient fluid from between the elements. The presence of this fluid is found to contribute to ~ 50% of the current retardation. There are also similarities with the effects of bed roughness in open channel and pipe flows, most notably there is a critical element spacing (11'/ kr ~ 7) where the effects of roughness are greatest (where w is element spacing and kr is element height). The experimental and numerical results demonstrate that the application of existing models that rely on experimental validation with smooth beds to situations where a rough boundary is present may lead to significant errors.
|
|---|
