Journal of Hydraulic Engineering

In a previous study, Foufoula‐Georgiou (1989a) investigated a stochastic storm transposition (SST) approach as a possible methodology of assessing the probability of exceedance of extreme precipitation depths over a catchment. The purpose of this paper is to report further methodological advances on the SST approach and present the results of a pilot implementation study in the Midwest. Under the assumptions of an elliptical storm shape and spread function of a given functional form, extreme storms have been described by the joint probability distribution of seven storm parameters (five specifying the magnitude, orientation, shape, and within‐storm spatial variability, and two specifying the location of the storm center). A nonhomogeneous spatial multivariate point process model has been postulated for the joint probability distribution of the storm center depth and storm location and it has been fitted to 65 extreme storms in the nine‐state mid‐western region. The method has been used to estimate the tails of the probability distribution of the average catchment depths over several hypothetical catchments in the Midwest and to describe the spatial variability of the estimates over the studied region.

Retention of diesel fuel in soil material from the Borden aquifer was investigated for five particle‐size fractions over the size range 75–500 μm and for different packing structures. The results showed variation of retention with pore structure especially for small particle size. The variation was, however, nonuniform; retention increased with increasing porosity over some ranges of pore structure, whereas the reverse occurred over other ranges of pore structure. Also, depending upon pore structure, retention showed little variation or decreased with increasing particle size. The retention of diesel fuel was greater than the retention of water under similar packing conditions. The dependency of retention on fluid type and medium structural parameters agrees qualitatively with theoretical predictions for retention in porous media of uniform spherical particles, and can be explained based on the concept of capillary retention.

Numerical results from a computer model for the prediction of the neutrally buoyant flow within sewage treatment plant settling basins are compared with results of several new experiments. The flow field calculations performed using the k‐ε turbulence model agree well with experiments for simple geometries. The quality of the computations, however, deteriorates with increasing flow complexity. It is argued that the effects of flow curvature, which are not included in the standard k‐ε model, are mainly responsible for the differences between computation and experiment. The results of the calculation of dye transport within the basin after pulse injection show that the use of low‐order numerical models results in predictions in which the solution is badly polluted by numerical errors. The quality of the prediction of the dye concentration field is strongly related, to the quality of the flow field prediction. The passing time of the peak output concentration is well predicted if the size of the recirculation regions is well predicted. In all cases, the peak dye concentration is overpredicted.

The effect of stream cross‐sectional geometry and hydraulic roughness data errors on the accuracy of computed steady‐flow water surface profiles is reported. Stream cross‐sectional geometry may be developed by several survey methods. The survey methods studied herein include field surveys, aerial spot elevation surveys, and topographic maps. The different survey methods and associated accuracy specification result in different stream geometry errors. Error in Manning's n‐value is also an important contributing factor to error in computed steady‐flow water surface profiles. A Monte Carlo simulation strategy was applied to develop an array of computed steady‐flow profile errors for the survey methods for 2‐, 5‐, and 10‐ft contour interval accuracy specifications, and Manning's n‐value reliability. Regression equations were derived for estimating computed steady‐flow profile errors. The regression equations relate survey methods, survey accuracy specification, reliability of Manning's n‐value, and stream hydraulic properties to estimated profile error.

Hydrostatic pressures from reservoir and tail water act on the dam and occur within the dam and foundation as internal pressures in the pores, cracks, joints, and seams. Measurements in existing dams confirm that uplift pressures undoubtedly exist, not only at the foundation junction plane, but on other planes within the dam to which water has gained access in the course of time. The distribution of pressure through a horizontal section of the dam is assumed to vary linearly from a full hydrostatic head at the upstream face to zero, or tail water, at the downstream face, provided the dam has no drains. When drains are constructed, the internal pressure within the dam should be modified in accordance with the size, location, and spacing of the drains. The internal pressure distribution through the foundation is dependent on size, depth, location, and spacing of drains, on rock porosity, jointing, faulting, and to some extent the grout curtain. The pore pressures act over the entire area of any section through the concrete. Because of possible penetration of water along construction joints, cracks, and the foundation contact, internal pressure should be considered to act throughout the dam. Internal hydrostatic pressures should be used for analyses of the dam and the overall stability of the dam at its contact with the foundation. In this study, an analytical solution based on seepage theory is presented to determine the average uplift pressure across the section of a gravity dam having a system of equally spaced drains of uniform diameter. The optimal location of the drains for the minimum uplift is also obtained.

Design stream flows are frequently used in water pollution control programs to provide adequate protection against pollutant exposure periods of a given duration. An analysis of the effect that simple stream flow dilution has on the x–day average concentration of a pollutant demonstrates that the x–day harmonic mean flow is a more meaningful statistic to use in computing design flows than is the customary arithmetic mean flow. The significance of this result was examined by computing design flows for 60 rivers throughout the country. Substantial differences were found for the mean daily flow—a design flow suggested for water quality criteria based on protecting human health against lifetime exposures. The harmonic mean daily flow is typically 20–60% of the arithmetic mean daily flow. Data from the 60 rivers showed very good agreement with a theoretical relationship between the arithmetic, harmonic, and geometric means of a lognormal distribution.

A laboratory experiment is done to learn more about the influence of upstream boundary conditions on the hydraulics of overland flow under rainfall. The experiment is done on an impermeable plane 1^-m wide, with a rainfall simulator that generates a uniformly distributed and time-invariant geometrical rainfall pattern over a length L measured along the plane (from x=O to x=L; x=O is defined as the upslope limit of the applied rainfall). Results show the importance of considering upstream boundary conditions, for gentle slopes, other than the most frequently used h(O,t)=O for $t>h$O, where h(x, t) is the flow depth as a function of position x and time t. The observed nonzero water depth at x=O and the consequent increase of water depth over the surface radically change the overland flow process over the slope.