Journal of Hydraulic Engineering

Field research of the spring, canal, and fountain system at the ancient Inca site of Machu Picchu defined the site's hydrological and hydraulic characteristics. An analysis of the Machu Picchu water system demonstrated that the Inca builders were skilled in empirical hydraulic engineering methods. A field survey also showed that physical constraints of the gravity-driven water system played an important role in defining the design and layout of this ancient Inca mountain-top community. A hydraulic analysis defined several characteristics of the Machu Picchu spring, canal, and fountain system. The flow of the primary natural spring was measured at 26–125 L/min while the stone-lined domestic canal leading from the spring has an estimated maximum operational capacity of 300 L/min. The 749-m-long canal traverses a steep mountain-side terrain on a stable and well-built terrace and passes through the agricultural sector before entering the urban sector of Machu Picchu. There it supplies 16 fountains that served as the domestic water supply for an estimated permanent population of 300 with a total of 1,000 when the Inca Emperor was in residence. The 1.0–4.8% slope of the small cross-section canal provided adequate cleansing velocities throughout its length. The hydraulic design of the 16 fountains provided adequate water jetting to enable easy filling of the ancient Inca water jug known as the aryballo.

A two-dimensional numerical flow model for trapezoidal high-velocity channels is developed. The model is designed specifically for simulation of flow in channels having sloping sidewalls in which the depth is an unknown variable in the governing equations and therefore the plan view of the flow domain is not known a priori. Solutions are obtained by time stepping from specified initial conditions using an implicit Petrov-Galerkin moving finite-element representation of the governing equations. The moving finite-element model produces a simultaneous solution for the boundary displacement and flow variables. This implementation provides stable solutions for supercritical flow even at relatively large Courant numbers. The model is tested by comparison of simulation results with laboratory data. These data sets serve as a basis for evaluation of the numerical model and should also prove useful to researchers in testing other numerical flow models applied to supercritical flow in channels having sloping sidewalls.

An analytical model of the advection-dispersion phenomenon in rivers with a laterally nonuniform axial velocity distribution is presented. The dispersion phenomenon is assumed to be governed by the two-dimensional advection-diffusion equation with constant but anisotropic turbulent diffusion coefficients. An infinitely long river with a prismatic cross section bounded laterally by parallel nondispersive banks is assumed. The velocity distribution is allowed to vary in an arbitrary functional form, but was restricted in this work to a family of power-law velocity distributions. The method of moments is used to derive the important statistical parameters of the concentration distribution. The concentration moment equations are solved analytically using the method of Greens function coupled with the method of images. The moments are then used to construct an approximate model of two-dimensional dispersion for an arbitrary velocity function, an initial distribution, and source injection scenarios. A one-dimensional simplification of the two-dimensional dispersion model with a time-dependent dispersion coefficient is also outlined. The dependence of the concentration moments on the velocity distribution and the shape of the source distribution is analyzed, and numerical results for a plane source and vertical line sources at the centerline and at the side are compared. The effect of the asymmetry of the velocity profile on the mixing length and time and on the skewness of the concentration distribution is shown to be significant, which might partly explain the persistence of the skewness observed in the field. However, the effect of the source injection scenario was not significant at large times. The analytical results can be used to model the fate and movement of pollutants and to better assess the effect of discharge siting on the dispersion of a contaminant. The model can be also used as a simple practical tool in simulating transport in a nearly prismatic river system.

The Swan River is typical of the estuaries in southwest Australia. A shallow (5 m) sill near the entrance controls fluid exchange between the estuary and the ocean. The main estuarine basin is relatively deep and wide and is in turn bounded at the upstream end by a shallower (2 m) secondary sill. An understanding of the dynamics associated with the circulation and response of the deep basin over intertidal time scales was required to predict the dispersal of nutrients and provide data to validate numerical models of the system. The dynamics of the deep basin were documented with several C-T-D-DO transects during intensive summer and winter experiments. A variety of internal features including lee waves, undular bores, and basin scale seiching were observed over a tidal cycle. The circulation pattern and the basin response are shown to depend on the interaction between the tidal dynamics, the sill control, and the stratification produced by the gravitational overflow. A one-dimensional, three-layered analytical model was constructed to explain the basin response.

We present the results of three laboratory experiments in which longitudinally sorted deposits were formed by feeding poorly sorted sediment at the upstream end of a narrow, 45-m-long channel. The input sediment had a median size of 6 mm and included significant amounts of material up to 64 mm and down to 0.2 mm. Water discharge was constant at 49 L/s and sediment discharge varied from 0.048 to 0.19 kg/s. Downstream fining was produced in all three runs; the variation in sediment-feed rate had relatively little effect on the fining profiles. In all three runs, the formation of a longitudinally sorted deposit was mediated by the formation of a coarse surface layer. The surface layer remained at the top of the deposit during aggradation by continually reforming itself at the deposit surface. The coarse surface layer fined by approximately a factor of 2 in D₉₀ and D₅₀ consistently in all three experiments. The deposit (subsurface) fined less, with D₉₀ fining more strongly than D₅₀. The short channel length and the relatively high rate of deposition rule out clast abrasion as the source of fining. The experiments suggest that selective deposition of the coarsest clasts due to unequal mobility is capable of producing fining rates comparable with the highest rates observed in nature.

We compare three-dimensional simulations of two cooling lakes at the Savannah River site to calibrated temperature data derived from high resolution thermal imagery and to data from two dye tracer experiments. The observed temperatures include detailed surface temperature maps created from calibrated thermal imagery and vertical temperature profiles measured at five locations across one of the cooling lakes. The dye tracer data consist of time series of concentrations recorded near the cooling water outlet of one of the cooling lakes. There is generally good agreement between observed and measured quantities, but the agreement is best for the smaller cooling lake that had a relatively higher heat load. We suggest that wind stress variability may have been at least partially responsible for the poorer agreement between observation and simulation for the larger cooling lake. The primary conclusion from this study is that thermal imagery is an excellent additional data source to dye tracer data for cooling lake model verification because imagery provides detailed information about the surface temperatures over the entire cooling lake allowing for a more complete verification of model simulations.

We investigate the penetration and spreading of a round jet into an ambient counterflow using the planar laser-induced fluorescence (LIF) technique. The LIF images enable us to obtain the instantaneous concentration field of the penetrating jet effluent. Large-scale wandering of jet penetration is observed and leads to significant temporal and spatial fluctuations in the location of the dividing interface between jet effluent and ambient fluid. The penetration distance and the lateral spreading width of the jet exhibit fluctuations with covering ranges as large as their corresponding time-averaged values. The degree of fluctuations in these penetration and spreading parameters increases with the jet-to-current velocity ratio, but essentially scales with the mean penetration distance. We also attempt to predict the mean penetration distance by considering the decay of centerline jet velocity under a hypothesis that the counterflow affects the jet via a simple compression of the axial coordinates of jet flow.

An infiltrometer was used at multiple locations at a site in Lee County, Fl. to define the spatial variability in infiltration parameters. Water-level data from a well at this site were collected hourly and used to determine the temporal variability in recharge. These results were used to define recharge in a representative stochastic numerical model of the aquifer. Model results without recharge compare well with existing analytical solutions for spatial head variability. Simulations with representative recharge events indicate that recharge produces a significant to dominant effect on head variability, which creates dispersion of contaminants, and that small-scale spatial and temporal recharge variations are the predominant mechanism causing the head variations.

The characteristics of jets issued by partially filled pipes are studied. The main flow features of such jets are specified, based on systematic observations and accounting for the Froude similarity. The end depth ratio, i.e., the depth at the section of issue normalized to the approach flow depth is predicted with the momentum equation. Further, an expression for the drawdown curve, from the approach flow to the end section is presented. Then, the lower and the upper trajectories of the jet issued into the atmosphere are studied, and expressions are given for the lower trajectory and the axial jet thickness. The analysis accounts for a novel dimensionless axial coordinate in which effects of approach filling and approach Froude number are included. The dimensionless coordinate in the streamwise direction may be considered as the expansion of a familiar term known in hypercritical plane flow. Additional data refer to the modular limit of pipe outflows and to submergence effects. Photographs illustrate the flow features.

Direct measurement of velocities and discharge in turbulent rockbed rivers is dangerous and often impracticable. Float measurements of surface velocity in sediment-free rockbed reaches of rivers are used to test the equation—due to Kennedy—that the mean velocity of a critical wave train is given by v= 1.2495√λ, where λ is the wavelength of the standing waves. Field data collected for wavelengths between 0.5 m and 7 m estimates surface velocity as about 2% significantly higher than the predicted mean velocity. It is concluded that the Kennedy equation successfully represents the mean velocity in rockbed rivers, which are essentially free from sediment, and that measuring the wavelengths of standing wave trains is an accurate and economical method of measuring mean velocity in the field. From estimated wave train width and computed critical depth, minimum discharge can be estimated for the section.

A relatively original point of view is introduced in order to classify sedimentation phenomena according to the spatial scale. Three categories are identified: large-, medium-, and short-scale models. The distinction is made in terms of the relationships between the hydrodynamic adaptation length, the sedimentologic adaptation length, and the study scale. This allows determination of the right type of mathematical modeling for each particular case. The main features of the models involved and of the associated numerical methods are briefly described. Case studies for the different types of models are presented for illustration.