Na Geng

Stochastic programming based capacity planning for semiconductor wafer fab with uncertain demand and capacity

Capacity planning is a challenging problem in semiconductor manufacturing industry due to high uncertainties both in market and manufacturing systems, short product life cycle, and expensive capital invest. To tackle this problem, this paper proposes a scenario-based stochastic programming model which considers demand and capacity uncertainties via scenarios, where the overall equipment efficiency is employed to describe the uncertain capacity for the first time. Based on the decentralized structure of tool procurement, production, stockout, and inventory decision-making processes, recourse approximation strategies are presented with varying degree of information share. The computational experiments show that the resulting tool set is robust enough to cope with the changes in capacity with the expected profits being maximized for different scenarios, and the scheme can generate pretty good solutions in reasonable computational time.

A Monte Carlo Optimization and Dynamic Programming Approach for Managing MRI Examinations of Stroke Patients

Quick diagnosis is critical to stroke patients, but it relies on expensive and heavily used imaging equipment. This results in long waiting times with potential threats to the patients life. It is important for neurovascular departments treating stroke patients to reduce waiting times for diagnosis. This paper proposes a reservation process of magnetic resonance imaging (MRI) examinations for stroke patients. The neurovascular department reserves a certain number of appropriately distributed contracted time slots (CTS) to ensure quick diagnosis of stroke patients. Additional MRI time slots can also be reserved by regular reservations (RTS). The problem consists in determining the contract and the control policy to assign patients to either CTS or RTS in order to reach the best compromise between the waiting times and unused CTS. Structural properties of the optimal control policy are proved by an average-cost Markov decision process (MDP) approach. The contract is determined by combining a Monte Carlo approximation approach and local search. Extensive numerical experiments are performed to show the efficiency of the proposed approach and to investigate the impact of different parameters.

Capacity Reservation and Cancellation of Critical Resources

This paper addresses the design of contract for reserving the capacity of a shared critical resource from the perspective of a given class of customers. The contract is composed of three parts: contracted time slots (CTSs) reserved by the service provider for the class of customers, advance cancellation of contracted time slots, and requests for regular time slots (RTSs). The problem of CTS cancellation and RTS assignment is formulated as an average cost Markov Decision Process in order to minimize the total cost including customer waiting times, unused CTS, and CTS cancellation. Structural properties of the optimal control policies are established via the discounted cost problem. A local optimization algorithm is proposed to improve a given initial contract. Numerical results show that advance CTS cancellation significantly reduces the ratio of unused CTS with slight increase of customer waiting time.

A decentralised multi-objective scheduling methodology for semiconductor manufacturing

This paper presents a decentralised multi-objective scheduling methodology for semiconductor manufacturing. In this methodology, a new classification method is designed based on utilisations and entropies in order to decentralise global objectives into local ones of work stations. Then, a decentralised multi-objective scheduling policy is proposed to control virtual production lines (VPLs) and machine workload. Results of numerical experiments show that the proposed methodology outperforms common rule-based scheduling policies and a compound scheduling strategy.

Dynamic Surgery Assignment of Multiple Operating Rooms With Planned Surgeon Arrival Times

This paper addresses the dynamic assignment of a given set of surgeries to multiple identical operating rooms (ORs). Surgeries have random durations and planned surgeon arrival times. Surgeries are assigned dynamically to ORs at surgery completion events. The goal is to minimize the total expected cost incurred by surgeon waiting, OR idling, and OR overtime. We first formulate the problem as a multistage stochastic programming model. An efficient algorithm is then proposed by combining a two-stage stochastic programming approximation and some look-ahead strategies. A perfect information-based lower bound of the optimal expected cost is given to evaluate the optimality gap of the dynamic assignment strategy. Numerical results show that the dynamic scheduling and optimization with the proposed approach significantly improve the performance of static scheduling and First Come First Serve (FCFS) strategy.

A memetic algorithm with iterated local search for the capacitated arc routing problem

Abstract The capacitated arc routing problem (CARP) is a difficult vehicle routing problem, where given an undirected graph, the objective is to minimize the total cost of all vehicle tours that serve all required edges under vehicle capacity constraints. In this paper, a memetic algorithm with iterated local search (MAILS) is proposed to solve this problem. The proposed MAILS incorporates a new crossover operator, i.e., the longest common substring crossover (LCSX), an iterated local search (ILS) and a perturbation mechanism into the framework of the memetic algorithm (MA). The proposed MAILS is evaluated on the CARP benchmark instances and computational results show that the MAILS is very competitive.

Optimizing contracted resource capacity with two advance cancelation modes

Critical resources are often shared among different classes of customers. Capacity reservation allows each class of customers to better manage priorities of its customers but might lead to unused capacity. Unused capacity can be avoided or reduced by advance cancelation. This paper addresses the service capacity reservation for a given class of customers. The reservation process is characterized by: contracted time slots (CTS) reserved for the class of customers, requests for lengthy regular time slots (RTS) and two advance cancelation modes to cancel CTS one-period or two-period before. The optimal control under a given contract is formulated as an average cost Markov Decision Process (MDP) in order to minimize customer waiting times, unused CTS and CTS cancelation. Structural properties of optimal control policies are established via the corresponding discounted cost MDP problem. Numerical results show that two-period advance CTS cancelation can significantly improve the contract-based solution.

Combining revenue and equity in capacity allocation of imaging facilities

Because of high procurement and operating costs, imaging facilities (e.g., magnetic resonance imaging (MRI)), are usually critical resources in hospitals. Hospital managers are under high pressure to pursue high utilization of the capacity, which leads to long waiting time for patients. However, different types of patients have different access time targets determined by their priorities according to the urgent levels and payments. The access time target is defined as the maximal amount of time between the appointment date and the examination date. For public hospitals, it is important to manage patient access to critical resources by considering the equity among different types of patients without sacrificing revenue. This paper proposes a nonlinear mixed-integer programming (NMIP) model for allocating the capacity of imaging facilities with the objective of maximizing revenue under the constraints of maintaining equity among different types of patients. The equity constraints are defined as the same access levels for different types of patients and the joint chance constraint for the same service levels in terms of waiting time. To solve this model, each time-slot, rather than the imaging facility, is considered as a server, which leads to an M/D/n queuing model. Based on an analysis of the M/D/n model, an approximation approach is proposed for the NMIP model, and CPLEX is used to solve the approximated model. Extensive numerical experiments based on real data from a large public hospital in Shanghai show the applicability and performance of the proposed model and investigate the impact of different parameters.

Implementation Strategies of A Contract-based MRI Examination Reservation Process for Stroke Patients

Timely imaging examinations are critical for stroke patients due to the potential life threat. We have proposed a contract-based Magnetic Resonance Imaging (MRI) reservation process [1] in order to reduce their waiting time for MRI examinations. Contracted time slots (CTS) are especially reserved for Neural Vascular Department (NVD) treating stroke patients. Patients either wait in a CTS queue for such time slots or are directed to Regular Time Slot (RTS) reservation. This strategy creates “unlucky” patients having to wait for lengthy RTS reservation. This paper proposes and analyzes other contract implementation strategies called RTS reservation strategies. These strategies reserve RTS for NVD but do not direct patients to regular reservations. Patients all wait in the same queue and are served by either CTS or RTS on a FIFO (First In First Out) basis. We prove that RTS reservation strategies are able to reduce the unused time slots and patient waiting time. Extensive numerical results are presented to show the benefits of RTS reservation and to compare various RTS reservation strategies.

MRI reservation for neurovascular patients

Quick diagnosis is critical for neurovascular patients. Diagnosis of these patients needs the assistance of expensive and heavily used imaging equipment. This results in long waiting time which potentially threats patients life. It is clearly very important for the neurovascular department to improve the service level by reducing the waiting time. To deal with this problem, this paper proposes a new reservation process between the neurovascular department and the imaging department. The neurovascular department reserves a certain number of time slots in advance for the imaging techniques, i.e., magnetic resonance imaging (MRI). This ensures that the stroke patients can receive the examination more quickly. This problem is formulated as a stochastic programming model in order to reach the best comprise between patient waiting time and unused time slots. To solve this problem, a two-step Monte Carlo approach is proposed. The problem is first approximated by a deterministic Monte Carlo optimization problem. It is further simplified to determine the contract. Given the contract, we then propose a feasible control policy. Numerical results show that the contract and the control policy proposed in this paper are quite efficient.