Journal of Hydraulic Engineering

The shore attachment of buoyant effluent in a strong crossflow is investigated experimentally. The effluent, in the form of a thermal plume deflected by the crossflow, is created in the laboratory by discharging warm water from a side channel into an open‐channel crossflow of the same depth as the side‐channel discharge. The plume has a tendency to lift off the channel bottom and move away from the shoreline. However, if the crossflow is sufficiently strong, the plume will reattach to the shoreline, and a recirculating flow region will form on the lee of the effluent. Temperature measurements are made in the plumes. The strong crossflow condition required for shore attachment is determined from the experimental investigation. Under the strong crossflow condition, the recirculating region remains attached to the shoreline, and the dilution of the plume is relatively independent of buoyancy effects.

Mathematical modelers of stratified flows in estuaries and fjords need to include in their equations terms representing the transfer of fluid across the density interface. Using a combination of results obtained from theory and laboratory and field measurements, it is found that two distinct types of interfacial mixing can be identified. When the interfacial transition layer is in a subcritical state, mixing takes place in either direction to a greater or lesser extent, depending on the turbulence in each layer. This type of mixing is called entrainment and is due to Kelvin‐Helmholtz instabilities. When the interfacial transition layer is in a critical or supercritical state, a two‐way transfer process called turbulent‐diffusion mixing takes place. In this condition, violent vortex motions, generated by internal wave interference, exchange equal volumes of fluid between the layers. The resulting equations are compared with the results of other researchers and found to have similar characteristics.

The nature of hydraulic transients in rock‐bored tunnels is expected to be frequency‐dependent. Three issues related to the frequency dependence are addressed in this paper. A complex‐valued and frequency‐dependent wave speed in a rock‐bored tunnel filled with water is formulated with consideration of the dynamic effect of the rock mass surrounding the tunnel, and its physical meaning is defined. The theory of hydraulic resonance in pipelines is extended to tunnels with frequency‐dependent wave speeds. Results show significant changes in resonant conditions, illustrate dramatic reductions in response at practical frequencies, and confirm the low probability of the occurrence of high‐frequency resonances in long tunnels. A variation in the impulse response method is applied for transient analysis. It is found that pressure waves propagating in tunnels are subject to dispersion and attenuation, and the complex wave speed may alter a transient to some extent, depending on the frequency spectrum of the transient.

A new methodology for incorporating reliability considerations directly into least‐cost optimization design models for looped water distribution networks is presented. The essence of the methodology is the measurement of reliability and making changes to the distribution system if the reliability is found to be unsatisfactory. The optimization model constrains the probability of pipe failure for each link and the probability of demand exceeding design values at each node for a fixed flow pattern in the network. The probabilities of pipe failure and demand exceedance are combined into a single reliability measure, the probability of no node failure. Due to the relationship between changing pipe breakage rates and pipe capacity, changing the demand exceedance probability also tends to reduce the probability of pipe failure. On the basis of earlier work, a simple reduction in the probability of the node demand exceeding the design values, through increasing the severity of the design flow, is used to achieve improvement in the probability of no node failure. Use of the reliability measure and the network reliability improvement procedure is demonstrated by application to a sample network.

A new methodology is presented for the reliability analysis of pumping stations for water‐supply systems. The methodology considers both mechanical failure and hydraulic failure and models the available capacity of a pumping station as a continuous‐time Markov process, using bivariate analysis and conditional probability approaches in a frequency and duration analysis framework. A supply model, a demand model, and a margin model are developed and used to compute the expected duration of a failure, expected unserved demand due to a failure, expected number of failures in the period of study, expected total duration of failures in the period of study, and expected total unserved demand in the period of study. Five example applications are used to illustrate the new methodology. This methodology can be used to analyze existing pumping systems and to design new systems.

A reliability‐based optimization model for water‐distribution systems has been developed. The model is aimed at the following goals: (1) Design of the pipe network including the number, location, and size of pumps and tanks; (2) design of the pumping system using a reliability‐based procedure considering both hydraulic failures of the entire network and mechanical failure of the pumping system; and (3) determination of the optimal operation of the pumps. The optimization problem is a large mixed‐integer, nonlinear programming problem that is solved using a heuristic algorithm consisting of a master problem and a sub‐problem. The master problem is a pure 0‐1 integer programming model, and the subproblem is a large nonlinear programming model solved in an optimal control framework. The conservation of flow and energy constraints are solved implicitly for each iteration of the nonlinear optimization procedure using a hydraulic simulation model, and the reliability constraints are also solved implicitly using a reliability model. The nonlinear programming problem is solved using a generalized reduced gradient code.

The classical theoretical model of density front propagation is based on a momentum equation for the whole fluid depth and an energy balance for one of the two layers. However, it is not always obvious which layer satisfies the assumption of energy conservation. Wood and Simpson's analysis of internal bores, which assumed energy conservation in the converging ambient fluid layer, is compared with an alternative solution based on energy conservation in the diverging layer. Both solutions are physically realistic because neither predicts an energy gain in the nonconserving layer but the alternative solution gives better agreement with the experimental data.