Journal of Hydraulic Engineering

The concept of unification, the opposite of branching theories used in biological sciences, is introduced as a new avenue to investigate the behavior of bed features in alluvial channels under unidirectional flow. The approach starts with the theoretical concepts initiated by Exner, according to which the features propagate at speeds inversely proportional to their heights. This leads to coalescence and rearrangement of the general pattern of bed features. Numerical experiments show that an initial large number of arbitrary bed disturbances rapidly reduces to a small number. If one feature reaches the maximum height for given flow conditions, smaller features pass through it. An initial uniform distribution of heights of disturbances slowly changes into a broad distribution of heights with some at the maximum height, similar to what is observed in nature.

A conventional numerical river model that failed to adequately reproduce the erosion of a 20‐km stretch of a sand‐bed river is modified to allow a substantial proportion of the eroded material to be transported as wash load. This reflects the formation of the river plain from proglacial lake sediments, which commonly have a large silt fraction. Because wash load has practically no effect on the transport capacity, except at very high concentrations, less of the transport capacity is devoted to transport of already eroded material, and more to further erosion. By assuming the proportion of silt to be 40%, the observed erosion could be satisfactorily reproduced. The average wash‐load concentration that follows from this assumption is quite small.

A three‐dimensional (3D) flow model is developed to improve the defects of a two‐dimensional (2D) model proposed by Shimizu and Itakura. Calculated results are compared with experiments as well as with results calculated by the 2D model. The flow field is predicted more correctly by the 3D model than the 2D model. A simplification of the full 3D model is made, in which the depth‐wise profile of the velocity is assumed to be logarithmic. This approach is favorably tested in the calculation of flow in a meandering bend. The simplified 3D model is applied to the computation of bed deformation by bed load as well as suspended load transport in meandering channels. Good agreements are found in comparison with the experimental results. A definite difference between the 3D and 2D models is found in the predicted bed deformation with suspended load. Evaluations of the flow and bed deformation by the proposed model are clearly demonstrated through application examples, and the validity of the new calculation model is verified.

It is well known that in steep channels, the flow becomes unstable when the Froude number exceeds some critical value. Instead of obtaining uniform flow, a short distance from the channel inlet, waves of various lengths, amplitudes, and phase velocities appear. These waves, traveling downstream and occasionally overtaking each other, are called roll waves. The value of the friction factor in open channels in unstable flow conditions is affected by the Froude number. Some formulas for the calculation of the friction factor in unstable flow conditions have been suggested by other authors. After these formulas are checked for smooth channels, the formulas for rough channels are compared with measurements. The critical Froude number, which is presented as a function of the Reynolds number, the channel roughness, the vertical velocity distribution coefficient, and the channel width, is an important parameter. A generalized formula is given to compute the friction factor both in smooth and rough channels in unstable supercritical flow conditions. The experiments carried out confirm the increase of the friction factor in unstable flow in rough channels.

The hydraulic reliability of a water distribution system can be defined as the probability that the system can provide the demanded flowrate at the required pressure head. Due to the random nature of future water demands, required pressure heads, and pipe roughness, the estimation of water distribution system reliability for the future is subject to uncertainty. A methodology is presented to estimate the nodal and system hydraulic reliabilities of water distribution systems that accounts for the uncertainties. The framework for the methodology is based upon a Monte Carlo simulation consisting of three major components: random number generation, hydraulic network simulation, and computation of reliability. For purposes of illustration, an example water distribution system consisting of 17 pipes and 12 demand nodes is used. This new methodology can be used in the analysis or expansion of existing systems, or the design of new systems.

The advective transport of a scalar is simulated by using the minimax‐characteristics method, which is an explicit and efficient finite difference scheme derived from the local minimax approximation of the exact solution of the pure advection equation. Fourier mode analysis shows that the method is unconditionally stable, and produces relatively small celerity error and little amplitude dissipation. The scheme compares favorably with the other commonly used backward characteristics schemes: it is better than the scheme using quadratic interpolation (higher accuracy, approximately equal computational effort), and is better than the schemes using cubic interpolation (approximately equal accuracy, less computational effort); it is less accurate than the scheme using Hermitian cubic interpolation, but requires only one‐half the computational effort. By interpreting the scheme as a backward characteristics scheme with quadratic approximation of the exact solution over four nodes, the extension of the method in a split‐operator approach for advection‐dispersion and hydrodynamics modeling in two or three dimensions is straightforward.

Nonlinear flow‐type straight‐line expressions are derived on the basis of exponential flow law instead of the linear Darcy law. A procedure of hypothesis testing for the flow‐regime linearity is explained step‐by‐step. It is concluded that any straight line for late time‐drawdown data on semilogarithmic paper does not necessarily guarantee the accurate application of the classical Jacob methods but confirms that the flow is radial. A dimensionless time‐drawdown plot is proposed for the identification of flow regime. If the dimensionless time‐drawdown plot for the late time‐drawdown data fits a straight line with the same slope as the Jacob straight line, then the underlying flow regime is linear. Otherwise the flow regime is nonlinear. Once the flow regime is nonlinear, the necessary equations are given for the estimation of parameters such as the storability, turbulence exponent, and transmissivity. Finally, the application of the methodology has been performed for some field data available in the literature.