The design of the beam is basically two-phase: first, make the spar strong enough to safely withstand working loads, and second, adjust the blade parameters in each panel so that the blade twist is optimal for both helicopter and propeller modes. If ever an adjustment is made that results in spar failure at limit load, revert to the first phase.
The strength testing of the spar is straightforward. A computer program analyses the panels, one by one, beginning with the outer panel. It applies the Tsai-Wu criterion to every ply in every panel. Should a ply fail, the computer program announces to the user that the spar in whatever panel is not strong enough, and quits. The user must adjust the spar parameters (this generally means making the spar thicker) and rerun the program. This cycle is repeated until the spar ceases to fail at limit load.
Once the spar is strong enough, the second phase begins. This is not quite as straightforward. The general procedure is to start on the outside, designing each panel one-by-one to twist appropriately.
Looking at the optimal twist distributions in Figure 1, we see that the helicopter and propeller modes each require a certain amount of twist across the length of the outer panel. The outer panel of the blade is designed so that the difference in panel twist between helicopter and propeller modes equals the difference in optimal twist between helicopter and propeller modes. It is emphatically not important that the actual panel twists equal the optimal twists; only the difference between helicopter and propeller mode twists matters.
To illustrate: suppose the helicopter mode requires 1 degree of twist in the outer panel, while the propeller mode requires 3. Now, suppose a blade panel is designed that twists -1 degree in helicopter mode and +1 degree in propeller mode. This panel can be used to obtain optimal twist, if 2 degrees of twist is built into the panel. (That is, the panel has 2 degrees of twist in an unstressed state.) In helicopter mode, the panel's 2 degrees will untwist 1 degree for a total twist of 1 degree, which is optimal. Likewise, in propeller mode, the panel's built-in 2 degrees, added to the additional 1 degree, yields a total 3 degree twist, which is optimal.
The design proceeds panel-by-panel, generally with an iterative step at each panel. It seems that the orientation of plies in the skin or spar is a good blade parameter to vary in designing the panel.