The paper presents a systematic comparison of some widely used turbule
nce models with one another and with experiments. The mathematically s
implest models applicable to three-dimensional steady, incompressible
boundary layers are of primary interest. Besides the more widely used
algebraic formulations, models using simplified transport equations fo
r turbulence quantities, such as the Johnson-King model, are considere
d. These models are applied to a variety of well-established two-dimen
sional and three-dimensional test cases, for which accurate and reliab
le experimental data are available. To avoid differences associated wi
th different numerical integration procedures, all models have been us
ed with the same finite difference method especially developed for thi
s purpose. This method solves the first-order boundary-layer equations
written in terms of streamline coordinates. One broad conclusion of t
his study is that all models considered give reasonable predictions fo
r the gross boundary-layer parameters, but important differences becom
e apparent for certain local values. Specific recommendations for the
choice of turbulence model in practical applications are included.