Optimal Tool Module Design Problem for Nc Machine Tools

Abstract

In order tD enhance factory automation and unmanned productien, numerjcal controlled (NC) machine tools are widely used in many rnechanical processing factories, Whenever one machining part is changed to another, the operation of an NC machine tool has to be stopped in order to change some of the cutting teols on the turret too! holder. In this paper, a new concept of ttool module' is usect and the problem of reducing the number of tool changing operations for NC machine tools is diseussed, The problem is formulated as a O-1 integer pr(rgramming (O-1 ILP) problem on a bipartite graph and a branch and bound algorithm is proposed to select an optimal tool module. Since the LP relaxation problem obtained by dropping the integrality condition for the O-1 ILP problern has a special network structure like the well-known transportation problem, this LP is solved by certain network flow algorithm with utilizing its special structure Sorne numerical experiments are reported, and indicate that the algorithrn is reasonably efficient. FurtherTnore, possible extenslons of this method are discussed, which in turn ±ncreases the produc-t±vity. These bring an innovation in the aspect of factory management. However, the machine setup operations which tnclude the ehanging 205

Minimizing the number of tool switches on a flexible machine

“…The problem becomes especially crucial when the time needed to change a tool is significant with respect to the processing times of the parts, or when many small batches of different parts must be processed in succession. These phenomena have been observed in the metalworking industry by Hirabayashi, Suzuki and Tsuchiya (1984), Finke and Kusiak (1987), Bard (1988), Tang and Denardo (1988), Feo (1989), etc. Blazewicz, Finke, Haupt andSchmidt (1988) describe for instance an NC-forging machine equipped with two tool magazines, each of which can handle eight tools.…”

An implicit enumeration scheme for the batch selection problem

“…On the other hand, when the bounding heuristic of Figure 2 is executed, a number of rank inequalities, corresponding to sets with known rank, is implicitly generated. In our implementation, at subproblem , all the locally valid (i.e., no lifting is performed) cuts x(S k ) Յ (S k ), for S k ʦ , where is the collection defined in (11), are added to formulation (1).…”

“…Following the definition by Crama and van de Klundert [6], this is called Batch Selection Problem (BSP). This model was originally proposed by Hirabayashi, Suzuki, and Tsuchiya [11] for loading a single numerically controlled (NC) machine tool, and has been extended to several other NC systems, as reported by Crama [4]. Similar models are also relevant to other applications, such as capital constrained repair kit selection [14], or strip mining [12].…”

Capacitated selection problem

“…A valid inequality y k 1 ϩ y k 2 Ն 1 is then generated so that whenever one of the variables is fixed to 0, the other will be fixed to 1. We note that Hirabayashi, Suzuki, and Tsuchiya [15] and Dietrich, Lee, and Lee [7] implemented preprocessing and cuts, respectively, on problems similar to (CSP). Our computational results show this type of simple cut to be very effective in fixing variables inside the branch-and-bound procedure.…”