Abstract
This note presents two observations on using the group efficacy measure of Kumar and Chandrasekharan (1990). First, an example is presented where striving to maximize grouping efficacy results in a solution to the part family-machine cell formation problem with illogical part or machine assignments. A restriction on admissible solutions corrects the flaw. Secondly, a reassignment of parts to families and/or machines to cells often results in a decrease to one variable, such as the number of voids, and increases the value of the other variable (number of exceptional elements). Upper bounds are developed for the positively impacted variable. When the positively impacted variable is below this upper bound, the result is an improvement to the grouping efficacy measure.