Sensitivities associated with the example problem:



In plan view:



We notice positive sensitivity to T1 on the left and negative on the right, while the opposite is true for T2.

Sensitivity is the derivative of the simulated function with respect to the parameter value, OR for UCODE which takes derivatives as perturbations, it is the change in the simulated value divided by the change in the parameter value. Consequently, a negative sensitivity indicates an inverse relationship between the simulated value and the parameter value whereas a positive sensitivity indicates a direct relationship.

Observe below that an increase in T1 causes less head loss across the first zone, thus raising heads there (a positive sensitivity), but more head loss occurs across the second zone due to the greater contrast in T1 and T2 such that beyond the midpoint, heads are lower for the higher T1 and the sensitivity is negative.



Carry that same type of reasoning out on your own to understand the impact of lowering T1, as well as raising and lowering T2.