GB1375917A - Method of using a digital computer for determing the start and finish times for activities contained within a precedence network - Google Patents

Abstract

1375917 Digital computers: precedence networks BURROUGHS CORP 23 Nov 1972 [6 Dec 1971] 54131/72 Heading G4A A digital computer is used for determining the start and finish times for each activity contained in a precedence network, e.g. using PERT and CPM for building an aircraft carrier or planning a business merger. The precedence network comprises activities (i.e. operation steps) each having a time duration associated therewith and vectors, each vector representing the relationship between two activities (e.g. the first activity must finish before the second one starts or one starts after the other has started). Each vector can carry a time indication, e.g. the second activity starts a predetermined time after the previous activity has ended. All the activities and their vectors are loaded into a general purpose computer (Fig. 7, not shown) including a core memory, a disc memory, an 1/0 multiplexer, a card reader and a printer. Then, for each activity, a list of predecessor and successor activities (Figs. 5 and 6, not shown) is created in step 26 by reading each vector in turn. Then all the activities without predecessors are loaded into the top of a stack memory, the validity of the topmost activity is checked and the start and finish times for it are calculated in step 36 (described in detail with reference to Fig. 3, not shown). Then any constraints on the current activity's successors are calculated (step 46), these are loaded into the bottom of the stack, the current activity is removed from the top of the stack (step 32) and the next activity evaluated in the same way. For each activity it is calculated how many times it returns to the top of the stack and this is compared with the number of its predecessor activities. If the former is larger, an error has occurred and the loop is terminated (step 42); if the former is smaller, step 32 is entered and the current activity is removed from the top of the stack. If they are equal this would have been detected in step 38 and step 42 would have been executed. This is repeated until step 46 has been calculated for all the activities.