GERT

A procedure for the formulation and evaluation of systems using a network approach. Problem solving with the GERT (graphical evaluation and review technique) procedure utilizes the following steps:

• 1. Convert a qualitative description of a system or problem to a generalized network similar to the critical path method—PERT type of network.
• 2. Collect the data necessary to describe the functions ascribed to the branches of a network.
• 3. Combine the branch functions (the network components) into an equivalent function or functions which describe the network.
• 4. Convert the equivalent function or functions into performance measures for studying the system or solving the problem for which the network was created. These might include either the average or variance of the time or cost to complete the network.
• 5. Make inferences based on the performance measures developed in step 4.

Both analytic and simulation approaches have been used to perform step 4 of the procedure. GERTE was developed to analytically evaluate network models of linear systems through an adaptation of signal flow-graph theory. For nonlinear systems, involving complex logic and queuing situations, Q-GERT was developed. In Q-GERT, a simulation of the network is performed in order to obtain statistical estimates of the performance measures of interest.

GERT networks have been designed, developed, and used to analyze the following situations: claims processing in an insurance company, production lines, quality control in manufacturing systems, assessment of job performance aids, burglary resistance of buildings, capacity of air terminal cargo facilities, judicial court system operation, equipment allocation in construction planning, refueling of military airlift forces, planning and control of marketing research, planning for contract negotiations, risk analysis in pipeline construction, effects of funding and administrative strategies on nuclear fusion power plant development, research and development planning, and system reliability. See PERT, Simulation