Thanks for sending me the CPTU soundings. I found the profile to consist of a one metre surficial sand layer on a 32 m thick layer of soft clay deposited directly on a sand layer at the end of the sounding. Because the cone stress is small, the friction ratio gets to exceed 10 % for some data points. The 10 % is the upper boundary for the Robertson classification chart. Presumably, Robertson did not have CPTU data with a friction ratio larger than 10 %. If the data are plotted in a Robertson chart with the friction ratio axis extended beyond 10 %, it looks logical to simply extrapolate Robertson's boundary lines for the different classification areas.
However, you can avoid the quandary by using the Eslami-Fellenius classification approach. The E-F classification is essentially the same as in the Robertson 1990 method, which supports the mentioned “logical” extrapolation of the boundaries.