Analysing NEH and a few other similar permutation flow shop scheduling heuristics using Taillard's and Vallada's benchmarking problems

Abstract

This paper analyses the performance of the NEH and five more similar heuristics proposed by the author using Taillard benchmark problems (120 in numbers) proposed in 1993 and new hard benchmarks (480 in numbers) proposed by Vallada et al. (2015). Two more initial orderings of jobs are considered besides the original, which orders the jobs in non-increasing order of the total processing times. The first one is according to non-increasing order of average processing time and standard deviation of their processing times proposed by Dong et al. (2008) and the other one is pertinent to non-increasing order of average processing time, average machine idle time and the standard deviation of their processing times. The performance of various heuristics and different ordering of jobs are also analysed using real time data collected from a plant manufacturing cold drawn seamless (CDS) tubes. The results indicate that two heuristics proposed by the author perform slightly better than NEH.