The duration variables were chosen so that the most likely or mean value of the off-site sequence is synchronized with the on-site sequence. Thus, the most likely values of Design, Fabricate (ignoring Rework), and Transport add up to have the right number of pipe spools—though not necessarily the right spools—being staged on site for a desired number of days prior to the most likely completion of Field Work and Prereq Work. In addition, the throughput (average number of resources output per time unit) off site matches the throughput on site (40 spools get produced on average in the same amount of time needed to complete prerequisite work in an area). That way, presumably (if all instances of each activity had a duration close to the mean value of that activity and no quality failure such as rework manifested itself), field production should not be delayed by a shortage of materials. Nevertheless, it could be delayed due to mismatches. Table 3 lists the alternative draw sequences used in each probabilistic model (A, B, and C). For the sake of completeness, it also includes those of the deterministic model described previously (model D).