Optimizing the availability of a buffered industrial process

A computer-implemented process for evaluating the fitness of a buffered industrial process performed by a configuration of subsystems and buffers, comprising the steps of:

• collecting statistical operating information on a running buffered industrial process to establish scale and shape parameters for probability density functions associated with operating parameters of subsystems of the buffered industrial process;
• defining design and operation values and operating rules associated with a new buffered industrial process that is expected to operate statistically in a manner related to the running buffered industrial process;
• establishing a schedule of events for the subsystems and buffers;
• initializing a product output rate during a first event cycle and compute a product output, where an event cycle in a subsystem having n lanes, with only p operating lanes, is the time from a (n−p)th lane failure until the time at which (n−1) lanes are again operating;
• determining a second event cycle and determining the product output rate between the second event cycle, computing the product output and accumulating product output;
• continuing to schedule events and accumulating product output until a determined total simulation time is completed;
• computing the product-based availability of the configuration of subsystems and buffers as the ratio of the accumulated product output over the total simulation time to the total product output that would have been resulted in an absence of events associated with subsystem failures.

A computer implemented genetic process for defining sets of design and operation values associated with each one of individual buffered industrial processes that form a population set whose fitness is to be evaluated, comprising the steps of:

• forming an initial population set having members with sets of design and operation values that are randomly selected from available values;
• performing a first fitness evaluation of the product-based availability of each member of the initial population set;
• forming a second population set having members selected from a first set of members identical to the initial population set;
• a second set of members obtained by mutating the initial population set;
• a third set of members obtained by genetic crossover from random pairs of the initial population;
• and a fourth set of members obtained by randomly selecting design and operation values from the available values;
• performing a second fitness evaluation of the product-based availability of each member of the second population set;
• selecting a number of members of the second population set having the highest fitness evaluation to form a new initial population set;
• and
• continuing to perform fitness evaluations and form new population sets until a stop decision is made.