The results of an experimental program are reported on, wherein the lateral flow in a low-speed, three-dimensional, turbulent, incompressible boundary layer inside a recurving duct was made to reverse itself completely. In such a reversal process there occurs a region of flow where the boundary layer experiences lateral flow in two different lateral directions simultaneously. The physical arguments which support the existence of such flow regions with simultaneous lateral skewing are discussed and supported by the experimental results. Several of the more recent velocity profile models which are intended for use in describing skewed turbulent boundary-layer flows are reviewed as to their applicability to a flow with such simultaneous lateral skewing.

A modified Crank-Nicholson implicit finite difference formulation is presented for two and three-dimensional turbulent boundary layers. The turbulent stresses are treated after Prandtl’s early mixing length model. “Boundary layer like” assumptions result in only the streamwise and transverse stresses remaining. The specific empirical input is the Maise and McDonald mixing length model. Excellent agreement with two independent experiments is obtained for mean velocity field data. Both experiments included a plane of symmetry to provide a transverse coordinate initial condition.

The results of a momentum integral solution of the three-dimensional turbulent boundary layer on the confining wall of an impinging jet are presented. This geometry provides a boundary layer where large gradients in the streamwise and especially the transverse direction occur and hence is a severe test of momentum integral methods. The solution utilizes the Head entrainment function and the Ludwieg and Tillmann wall shear law, with no restriction on cross flows. An extensive comparison with experimental results show good to moderate agreement in the integrated flow parameters, with a strong dependence on the free-stream or edge condition to the boundary layer flow.

This article describes the general approach that was taken in integrating and system testing the 3B20D Processor system. Since both the system hardware and software were developed simultaneously, the goals of the system test and integration plan naturally shifted emphasis and expanded their scope from achieving hardware stability to establishing software functionality and finally to demonstrating system stability. This article also overviews some of the project management techniques and procedures applied during the development of the 3B20D Processor.

Simultaneous lateral skewing was detected and existed over a relatively long flow distance.2 When skewing existed in only one direction, the polar velocity profiles indicated a near linear region as the floor of the channel was approached. With simultaneous lateral skewing, a linear region was not apparent until flow reversal was near completion.3 The crossflow velocity profile model advanced by Johnston [6] did not adequately describe the experimental crossflow profiles when skewing was in only one lateral direction.4 With skewing in one lateral direction, the crossflow models (equations ( 6), (7), and ( 8)), while representing a few of the experimentally determined polar plots reasonably well, were generally judged to be poor approximations to the bulk of these crossflow velocity profiles.5 The crossflow models providing for simultaneous lateral skewing, equations ( 7) and ( 8), did not satisfactorily represent the data obtained in this investigation for such conditions. No attempt could be made to test the ability of the crossflow profile model (equations ( 9) or ( 11)) proposed by Eichelbrenner [10, 11] to approximate the data because wall shear stress and pressure gradient data were not available.6 The difficulty in defining a boundary-layer edge and freestream velocity precisely affects the outer portion of the velocity polar plots to a substantial degree.