The Application Of Theory Of Constraint In Logistics

This paper proposes the Three–Zone Buffer Management algorithm, it find solutions for practical applications with multi-constraints. The algorithm has made a full analysis of optimization algorithms and the characters for solving multi-constrained problems, and use the Theory of Constraints (TOC) and DBR (Drum-Buffer-Rope), put forward by Goldratt Dr., as reference. TZBM algorithm is used to solve multi-constrained problems, such as Container Loading Problem (CLP); the problems have a common character: Problems systems have at least one bottleneck restricting the operation of the entire system. This algorithm recognizes bottlenecks limiting the system operations of system, and stressed the full utilization of bottleneck. Through the planning of bottleneck utilities, it can find most suitable implementation progress, and through reasonable management of the bottleneck, it can promote the optimization of the entire system.In TZBM algorithm, the final solution composed of individuals. The orientation of TZBM is not emphasizing on the performance of single individual, instead, it emphasize on optimization of overall solution. Every individual takes competitive behavior to achieve its own goal, and takes concessive behavior for the interests of the overall system.The TZBM algorithm is:TZBM algorithm divides its constraints into hard constraints and soft constraints. The hard constraints are must to meet, and can't be relaxed under any circumstances. The soft constraints can meet conditionally, and can be relaxed according to the actual situations. First, the algorithm considers only hard constraints, and uses a simple algorithm to obtain the initial solution in the shortest possible time and this solution is composed of individuals. Next, sets the values of relaxed soft constraints'parameters to the smallest, and then uses evaluation function to evaluate the individual and zoning. In which, the individuals in Green-Zone do not violate all constraints, including hard and soft constraints; the individuals in Yellow-Zone do not violate relaxed soft constraints; and the individuals in Red-Zone violate the relaxed constraints. Then the method negotiates the individuals of Red-Zone, and if there are no individuals in Red-Zone, the method can negotiate the individuals of Yellow-Zone. If all individuals finish the negotiation, the method can relax the parameters of soft constraints, and re-zoning, negotiating.The primary role of TZBM method is to identify constraints and improve the conflict of constraints in the process of negotiation based on the DBR. In practical problems, the method can avoid system falling into deadlock and reduce the individuals of Red Zone, with a reasonable application of the experience and knowledge of experts, such as the relaxation parameters, changes the working conditions of TZBM. Competition and concessions mechanism can improve the system constraints to a large extent, which is applicable to deal with complex practical problems having bottlenecks.The implementation of the algorithm is divided into the following steps:Step1: Make sure the constrained conditions. According to the requirements of actual problem solution, it divides the conditions into hard constraints and soft constraints. The hard constraints can't be relaxed, and soft ones can make a certain relaxation according requirements.Step2: Make sure the objective function and the evaluation function. The evaluation function is defined, according soft constraints, to evaluate the individuals. The objective function is defined, according to actual requirements, to make choices of the solutions in solving progress.Step3: Work out the initial solution. It obtains the initial solution using an existing simple algorithm, and this solution must meet the hard constraints above. The usage of simple algorithm is to ensure find out the initial solution in shortest possible time.Step4: Select a soft constraint to relax, and divide the zones based on the evaluation function. According to the objective function, the individual of current solution are assigned to three zones: GZ, YZ and RZ respectively.Step5: Select the individual from RZ, according to evaluation equation, to negotiate. The individual can be negotiate through a variety of means, such as self-negotiating, inserting and exchanging with other individual selected according evaluation equation. At the same time, choose the solution using objective equation, and make comparation with the best solution preserved, and save the better as optimal solution. Next, change the parameter of the relaxed constraint, and decide whether to continue negotiation.Compare with common optimization algorithms, this algorithm's main characteristics are following:(1) When generating the initial solution, it only considers the hard constraints. And can use a simple heuristic algorithm to shorten the time of initial solution as far as possible.(2) Because in the process of the initial solution and the optimization process of solutions, the best solution preserved all meet hard constraints, it can output the solution at any time; the output is the optimal solution until now.(3) Because the algorithm uses the buffer size to control the changing buffer parameters, it can adjust optimizing time by adjusting buffer size.(4) Since the algorithm uses individual negotiation strategy, it can select suitable negotiate means according to practical problems. The buffer parameters and sizes can be set in accordance with experts'experiences and knowledge, and the algorithm has great flexibility and diversity, and can prevent the system derive to deadlock to the largest extent.(5) It can decide the relaxed order of the conditions. Using flexible relaxed order, it is more easily to deal with multiple conditions. For the actual applications and most of requirements of customers, this paper establishes mathematical model and solutions for mutli-constrainted problems. The algorithm has also been applied to the"Logistics Loading System"of TYBB, and the system has been used by many logistics companies.According to customers'feedback information, TZBM reflects the following distinct characteristics in the application of the actual system:(1) It can improve the constraints and enhance the performances of systems.(2) It can easily get system out of the deadlock and the deadlock situation has been reduced.(3) Three different buffers are propitious to decide the size of the buffer zones.For mutli-constrainted problems like logistics, finding out optimal solution is a complicated process. Many of the optimization algorithms have yet to be further optimized.