Well there are a lot of possibilities. The easiest one is to use the ofat method: one factor at time. You vary your input variables one at time and you record the impact/variation on your output variable D. You can check out saltelli's books/articles on this topic. Be aware that the ofat method does not take into account combined effect of input variables.
Given that you have a simple linear additive model without interaction (with few values for each factor) you could use sigma-normalized derivatives or even a linear regression. See GSA the Primer by Saltelli et al. chapter 1 section 1.2.2.