In cellular manufacturing system formation of machine-part families is an important task. Cellular manufacturing systems deals with various clustering algorithms which are particularly relevant to the problem of machine component group formation. In this paper, case study from well-known gearbox manufacturing company is considered where there is scope for using cellular manufacturing system. Company is manufacturing the gearbox varying from single stage reduction to four stage reduction. A modified direct clustering algorithm is proposed and is applied after the initial incidence matrix derived from direct clustering algorithm. This algorithm overcomes the problem arises during application of direct clustering algorithm and gives the optimum solution. The algorithm is specifically designed to deal with large amount of data during realistic situations. The results of proposed algorithm are compared with well-known existing algorithms and it is found that the proposed algorithm gives the better solution than the existing algorithms under consideration.