To obtain an open-loop transfer function from a model, you specify a loop opening. Loop openings affect only how the software recombines linearized blocks, not how the software linearizes each block. In other words, the software ignores openings when determining the input signal levels into each block, which influences how nonlinear blocks are linearized. Consider the following model, where you obtain the transfer function from e2 to y2, with the outer-loop open at y1:
Here, k1, k2, g1, and g2 are nonlinear.
The software linearizes each block at the specified operating point. At this stage, the software does not break the signal flow at y1. Therefore, the block linearizations include the effects of the inner-loop and outer-loop feedback signals.
K1, K2, G1, and G2 are the linearized blocks.
Finally, to compute the transfer function, the software enforces the loop opening at y1, injects an input signal at e2, and measures the output at y2.
The software returns (I+G2K2)-1G2K2 as the transfer function.