Yon-Chun Chou

A methodology for product mix planning in semiconductor foundry manufacturing

Since a semiconductor foundry plant manufactures a wide range of memory and logic products using the make-to-order business model, the product mix is an important production decision. This paper first describes the characteristics of the product mix planning problem in foundry manufacturing that are attributable to the long flow time and queuing network behaviors. The issues of time bucket selection, mix optimization and bottleneck-based planning are next addressed. A decision software system based on integer linear programming techniques and a heuristic procedure has been implemented for mix planning. Data provided by a wafer plant has been used to study problems related to product mix planning. It was determined that the suitable time bucket of planning is either one week or one month and the lead-time offset factor should be included in the logic of workload calculation. This paper also presents various facets of product mix decisions and how they should be integrated with operations management.

A dynamic system model for proactive control of dynamic events in full-load states of manufacturing chains

Risk management is a major concern in supply chains that have high levels of uncertainty in product demand, manufacturing process or part supply. The uncertainties frequently manifest as dynamic events that pose a threat to interrupting supply chain operation. Depending on the nature and severity of uncertainty, the impact of dynamic events can be distinguished into three categories: deviation, disruption, and disaster. Many studies in literature addressed modelling of deviation events. In this paper, a dynamic system model of supply chains is described which can be applied to managing disruptive events in full-load states of manufacturing chains. An example of disruptive events is given which arises from demand shocks in distribution channel. The procedure to construct full-load production functions of complex manufacturing nodes with internal queuing delay is described. Analytic optimal solution is derived for the dynamic model. Given an unordinary event of demand shock, this model can be used to determine if demand shock can be absorbed by a manufacturing chain and the level of contingent resources that must be synchronously activated in multiple nodes of the chain. This model can be used to reduce what could have been a disruptive event into a deviation event, thus enhancing risk management.

Economic analysis and optimization of tool portfolio in semiconductor manufacturing

The tool portfolio of a plant refers to the makeup, in quantity and type, of processing machines in the plant. It is determined by taking into consideration the future trends of process and machine technologies and the forecasts of product evolution and product demands. Portfolio planning is also a multicriteria decision-making task involving tradeoffs among, investment cost, throughput, cycle time, and risk. Tool portfolio planning is a complex task that has strong bearing on manufacturing efficiency. In the first part of this paper, a multicriteria economic decision model is presented for optimal configuration. of the portfolio and to determine the optimal factory loading. The second and third parts of the paper contain applications of the model. If plants are closely located or have a twin-plant design, portfolio planning at multiple plants can be integrated to enhance the overall effectiveness of portfolios. In the second part, a novel methodology for arbitrating capacity backup between plants is described. Because the economic model is constructed upon a valuation of both cycle time and throughput, it is a suitable method for the evaluation of cycle time reduction projects. The application procedure is outlined in the third part.

Option-based Capacity Planning for Semiconductor Manufacturing

Due to high cost of capacity investment, many semiconductor manufacturing companies have exhibited the need to pursuit innovative capacity plans and planning methods. In this paper, a case study of option-based capacity planning is presented. Three issues are addressed: estimation of production cost parameters, valuation of capacity, and analysis design . It is shown that the option- based approach, in long-term, could generate a capacity plan that requires less investment, but generates higher operating income.

Recency-based storage assignment and warehouse configuration for recurrent demands

Storage assignment is an important decision problem in warehouse operation management. In conventional problem settings of distribution warehouses, stock items are stored in bulk but retrieved in small quantities. Storage assignment methods typically make use of demand attribute information of order quantity, order frequency and correlation between demands. In this paper, we address a different problem in which the request for the same stock items is stochastically recurrent. The problem arises when the items are needed in production and, after production, are returned to warehouses for later reuse. Examples of such items include tooling in factory, books in library and digital objects in data warehouses. Utilizing the recurrent characteristics, a salient recency-based storage assignment policy and an associated cascaded warehouse configuration are proposed and analyzed in this paper. This paper has four parts. In the first part, a model of recurrent demand is described. In the second part, the efficiency of the recency-based policy and a traditional ID-based policy is analyzed. In the third part, a mathematical programming model for optimal configuration of cascaded warehouses is presented. Finally, a case study of hospital visits is presented. This paper concludes with recommendations on cascading and zoning the warehouse for applying the recency-based policy.

A comparative study on the performance of timing and sizing models of capacity expansion under volatile demand growth and finite equipment lifetime

This paper includes an empirical analysis of industry demand data.We show that demand growth can be modeled as a geometric Brownian motion process.The timing capacity expansion model is shown to outperform the sizing model.As a criterion, we show that profitability is as important as demand satisfaction. Competition in global supply chains has become so severe that many suppliers in the high-tech manufacturing industry must shoulder high risk but have negative return on assets. While the literature is abundant with capacity models, there is a need for further research on capacity investment, especially in selecting and correctly using the right model. For a firm with lasting manufacturing operation, capacity expansion has two aspects: the timing and sizing of each expansion. The aim of a sizing method is to determine the scale of capacity expansion and that of a timing method is to determine the right time of the next expansion. The majority of capacity models in the literature can be classified as sizing models. In contrast, timing models have not received as much attention. In this paper, we compare the performance of the two types of models under volatile demand growth in order to find out the more appropriate type for the high-tech manufacturing environment. An empirical analysis of semiconductor demand is first presented. We find that the geometric Brownian motion process is appropriate for characterizing the volatility of demand growth. Based on this finding, simulation is used to compare a canonical timing and a canonical sizing models in various scenarios of demand growth, demand volatility and profit margin. We also advocate using profitability as a capacity investment criterion, in addition to the demand-satisfying criterion that is commonly used in the literature. Simulation results show that the timing model outperforms the sizing model. Finally, the behavior of the timing model is characterized as an aggressive method that can be used to exploit demand volatility for an advantage.

Superior/Inferior Segment-Discriminated Ant System for combinatorial optimization problems

The Ant Colony Optimization method is a heuristic algorithm for solving various optimization problems, particularly the combinatorial optimization problems. Traditional ant-optimization methods might encounter search stagnation owing to a biased pheromone map that is dominated by local optimal trails. To overcome this drawback and lower the number of solution constructions for finding the optima, this paper presents an improving ant-optimization system, the Superior/Inferior Segment-Discriminated Ant System (SDAS). This system proposes a segment-based pheromone update strategy to deposit pheromone on superior segments and withdraw pheromone from inferior ones. The method uses the control-chart technique to define superior and inferior limits to partition the constructed solutions into superior, inferior, and ordinary solutions. Inferior and superior segments are then extracted from the superior and inferior solutions by stochastic set operations. Since the pheromone map is not easily dominated by any local optimal trail, the solution search is more efficient and effective. Several benchmarks from the TSP-LIB and OR-LIB were used as sample problems to test the proposed system against other ant-optimization systems, including the AS, ACS, AS_rank, AS_elite, and MMAS. Numerical results indicated that the SDAS obtains solutions that are similar to or better than others. Maturity index for the pheromone map was discussed and experimental results showed that the proposed method was able to prolong the time for the map to maturity to avoid earlier search stagnation.

A method for aligning serial batching decisions in a two-level organization of factory and distributed workstations

Large factories that manufacture high mixes of complex products are usually composed of a number of workstations and the manufacturing control function is divided between a factory and a workstation level. While the management of individual workstations tends to focus on efficient machine utilization, the top-level factory management is usually concerned with job flow control. Integration of operation decisions between the two organization levels can recover productivity loss stemming from disparate objectives. This paper presents a method for aligning the job batching decision for serial-batch machines that require machine setup to serve stochastic arrivals of multiple job types. The effect of batching on flow time is first analyzed and closed-form formulas for the probability of setup are derived for a time-based batching policy. The misalignment in batching decisions at the two organization levels is next illustrated. Finally, a state-based performance measure is designed for decision integration. Numerical simulation and regression are used to test the proposed method. The main contribution of this paper is on developing a distributed vertical alignment method which compliments the approaches of horizontal coordinated scheduling and vertical functional decomposition in architecture design of distributed manufacturing control.

Integrating machine scheduling and self-healing maintenance by job-mix pull control

In general, machines degrade with use. But, for some manufacturing processes machine ageing can be reversed by processing alternative types of jobs. In the latter case, machines can run longer without breakdowns if job types are balanced and scheduling is optimised. However, when job arrivals are stochastic, even short-term fluctuations in job mixes can increase the risk of breakdown. This paper presents a proof of concept study on job-mix pull control to exploit the age-reversing, healing effect. Because adding pull control will impact the architecture of factory scheduling, three issues are addressed in the proof. First, it is shown that the new architecture would have a better performance than the existing dispatching approach. Second, a method of pulling jobs from upstream to reduce the probability of machine breakdown is developed. The condition of workload imbalance in which job mix control (JMC) should or should not be activated is analysed. Finally, the benefit of JMC is evaluated by using simulation to demonstrate potential improvements that can be achieved. Besides the proof of concept, this study produces an important finding. A lingering cumulative influence of the self-healing effect is discovered, pointing out a new direction for future maintenance scheduling research.

Using capacity as a competition strategy in a manufacturing duopoly

Capacity is a strategic factor of competition in asset-heavy industries. However, when demand is volatile, capacity expansion is hazardous to profits. In this paper, a game theory method is developed for analyzing whether capacity can be used as a competition strategy and for determining its sufficient conditions. We consider a manufacturing service duopoly of differentiated service prices and volatile demand. Sufficient conditions for Nash equilibrium of capacity expansion are derived for lognormal demand. Those conditions specify a choice space for the leader Arm to increase its own profit at the expense of the followers profit by aggressively expanding its capacity.