All the presently available numerical models have serious limitations 

 and none are recommended for engineering purposes. The earlier versions 

 omitted important terms in the momentum balance equations, and the latest 

 models simply lack accuracy due to excessive truncation errors in the 

 algorithm employed. The lateral (eddy viscosity) mixing stress terms 

 physically smooth the current profiles, circulation patterns, and rip 

 current jets. The present models display evidence of excessive numerical 

 viscosity (Fig. 35) that translates into numerical inaccuracy. Calibration 

 of such models requires use of physically unrealistic closure coefficients 

 that could be a disaster for use as predictive models. The evidence 

 to support this conclusion is vividly displayed in the generally poor 

 comparisons between model and experiment shown in Figures 80, 81, and 

 83. 



As outlined by Vreugdenhil (1980), it is clear that the Delft Hydrau- 

 lic Laboratory has begun a concerted effort to develop a comprehensive 

 numerical model. It is by far the best effort discussed to date because 

 it includes all the physically important terms and proper numerical 

 methods to ensure their accuracy in the simulation. The report by 

 Vreugdenhil (1980) discusses the model requirements, equations, and 

 numerical procedures. Calibrations, tests, and other results will be 

 reported when completed. It is clear that additional comprehensive model- 

 ing projects are desirable as both an alternative and in support of physi- 

 cal model tests and expensive field experiments. 



Previous efforts were hampered by lack of data to calibrate, verify, 

 and test the two-dimensional models. Some additional controlled labora- 

 tory data are now available (Gourlay, 1978; Visser, 1980; Mizuguchi, 

 Oshima, and Horikawa 1978) ' and can be obtained for this purpose. Also, 

 the extensive NSTS field data tapes can be used. 



Some fundamental problems remain with the time-averaged simulation 

 models. The wave height fields must be specified by other means. For 

 this purpose, the model developed by Noda, et al. (1974) for wave shoal- 

 ing and refraction that includes current interactions remains a popular 

 choice. Wave diffraction, reflection, transmission (breakwaters), etc. 

 transformations must also be prescribed. The numerical programs for 

 wave height transformation may require considerable development effort, 

 by themselves, if not already available. In addition, it was seen earlier 

 in this chapter how direct wave refraction calculations by Boussinesq 

 theory produce significant differences with Snell's Law. This indicates 

 that nonlinear refraction and other wave transformation theories should 

 be employed. Finally, the correct numerical simulation of circulations, 

 eddies, and subgrid-scale turbulence is still an active research area 

 in computational hydraulics. 



A verified theory to predict rip current spacing for all possible 

 conditions does not exist. The best engineering estimate available is 

 the semiempirical hypothesis of Sasaki (1977) summarized in Figure 78 

 and equations (167), (168), and (169). No theory exists for rip current 



^''mizuguchi, oshima, and HORIKAWA, op. ait. 



228 



