top of page

Illustrative Example: Production Line with 20 Workstations

 

!

Simulation model for a production line with 20 workstations.

  • The minimum value for the average cycle time selected as the first best solution (first incumbent solution) (I-1), is identified from the initial experimental design and corresponded to a value of 197.7 minutes for the LCG Design, 194.5 minutes for the Mersenne Twister Design, 197.2 minutes for the D-Optimal Design (R-Project), 195.2 minutes for the Random Design, 197.2 minutes for the Clustering Design using the modified version, and 199.5 for the D-optimal design (JMP).

 

  • With the initial experimental design, a full quadratic regression metamodel was built, and used as the objective function to be minimized to obtain a predicted competitive solution.  In this example, the best solution when the algorithm stopped corresponded to a system time of 175.4 minutes for the LCG Design, 181.7 minutes for the Mersenne Twister Design, 157.5 minutes for the D-Optimal Design (R-Project), 153.5 minutes for the Random Design, 158.2 minutes for the Clustering Design using the modified version, and 162.1 for the D-optimal design (JMP) 

  • When a comparison between the initial incumbent solution (I-1) with the final one (I-9) was performed, the result was that the system time decreased in 22.3 minutes for the LCG Design, 12.8 minutes for the Mersenne Twister Design, 39.7 minutes for the D-Optimal Design (R-Project), 41.7 minutes for the Random Design, 39 minutes for the Clustering Design using the modified version, and 37.4 for the D-optimal design (JMP).

  • Facebook Classic
  • Twitter Classic
bottom of page