Institute for Systems Research Technical Reports
Permanent URI for this collectionhttp://hdl.handle.net/1903/4376
This archive contains a collection of reports generated by the faculty and students of the Institute for Systems Research (ISR), a permanent, interdisciplinary research unit in the A. James Clark School of Engineering at the University of Maryland. ISR-based projects are conducted through partnerships with industry and government, bringing together faculty and students from multiple academic departments and colleges across the university.
Browse
Search Results
Item Designing a Learning Historian for Manufacturing Processes(2000) Reaves, Lakeisha A.; Herrmann, Jeffrey; Plaisant, Catherine; ISRIn all aspects of life, reviewing history has proven to have some influence on future decisions. Review of past events is not new to our society. Basketball coaches often review videotapes of games to see what worked well and what could be improved upon (Plaisant et. al page 1). Black boxes in airplanes also provide a record of the conversation held by the pilot and co-pilot prior to a plane crash (Plaisant et. al page 1.) Allowing users to have some record of their actions gives them the opportunity to review these actions and perhaps decide what to do next.Providing a way to review history may also prove beneficial in the manufacturing environment. Simulations provide a means of modeling a "system to reproduce the dynamic behavior of the system" (Herrmann page 11).
While simulations are excellent tools for creating these models, they may lack in helping the user to understand the relationships that exist in manufacturing processes. For example, they may lack in facilitating learning that would help the user to understand the relationship that exist between such measures such as the capacity (the number of machines), rate (part/time), through-put (number of completed parts), net profit and cycle time (average time per part).
Understanding the relationship held between these measures is the key to understanding the model itself.
The Institute for Systems Research at the University of Maryland in a joint effort with the Human Computer Interface Lab (HCIL) at the University of Maryland endeavored to provide a solution to helping the user understand these relationships. Their objective was to help students understand the relationship held between the following performance measures: capacity, throughput and cycle time. Once this relationship is understood, the student could use this knowledge to optimize system design. It is believed that providing a link between the student and the simulation that would facilitate learning and understanding would accomplish this objective.
The Learning Historian had the capabilities of providing such a tool. The following course of action was followed in designing a Learning Historian for a manufacturing process:
ﵠDevelop a simple simulation of a manufacturing process using Arena
ﵠUse a Learning Historian that is able to read the Arena file
ﵠSelect the input and output configuration files to be displayed in the Historian
ﵠDevelop a study that would test the usability of the Historian as a user interface
ﵠTest the usability of the Historian on users by means of an informal study
ﵠObserve and record users comments and suggestions
ﵠImplement minor changes to Historian based on frequency of suggestion or comment
ﵠAfter initial testing of historian is complete collate all studies and look for trends in suggestions, comments and problems encountered by users
Item A Tool Optimization Interface for a Semiconductor Manufacturing System(2000) Thomas, Ryan; Herrmann, Jeffrey W.; ISRThis paper will serve as the documentation for the Tool Optimization codeof the HSE software. The purpose of the software is, simply, to enable auser to optimize a factory's tool selection. This will be added to theexisting Factory Administrator which enables users to understand theeffects of changes in many parts of the manufacturing process (i.e. Temperatures, Pressures, etc.).To accomplish this an interface was designed via the DELPHI programminglanguage that can take inputs from a user as well as factory details froman Excel spreadsheet, run simulations, determine an optimal toolconfiguration, and output this data as easily as possible to the user.
The Interface will guide the Simulation as many times as needed to performits gradient analysis. After the program is complete, it determines a bestcase tool configuration that meets the user's throughput while maintainingto his budget. The interface will output how many of each tool to purchaseas well the best possible tool allocation (usage) for each tool.
Item Improving Cluster Tool Performance by Finding the Optimal Sequence and Cyclic Sequence of Wafer Handler Moves(2000) Nguyen, Manh-Quan Tam; Herrmann, Jeffrey; ISRThe research aims to develop algorithms that can minimize the total lot processing time (makespan) of cluster tools used for semiconductor manufacturing. Previous research focuses on finding an optimal sequence of wafer handler moves in a cluster tool that has one process chamber in each stage. In practice, if the number of chambers in a stage is more than one, either a pre-specified sequence of moves is given in advance or a dispatching rule is applied. No previous work has addressed the problem of finding an optimal sequence of wafer handler moves to improve performance of cluster tools with more than one chamber in a stage. Cluster tools are highly integrated machines that can perform a sequence of semiconductor manufacturing processes. The performance of cluster tools becomes increasingly important as the semiconductor industry produces larger wafers with smaller device geometry. Some factors that motivate the use of cluster tools, instead of stand-alone tools, include increased yield and throughput, less contamination, and less human intervention. In this research, the cluster tool is modeled as a manufacturing system with a material handling system (wafer handler). The model specifies all constraints that a feasible sequence of wafer handler moves must satisfy. The thesis develops two cluster tool scheduling algorithms. Given the lot size, the wafer handler move time, the in-chamber processing times, and the tool configuration the first algorithm, based on a complete forward branch-and-bound algorithm, searches for an optimal solution from the set of all feasible sequences of wafer handler moves. The second algorithm, a truncated branch-and-bound algorithm, quickly searches for the best solution from the set of feasible cyclic sequences of wafer handler moves. For simple tool configurations, analytical makespan models are also derived. The results show that, in many cases, the search algorithms can significantly reduce the total lot processing time. This reduces tool utilization, reduces manufacturing cycle times, and increases tool capacity.Item Sensitivity Analysis and Discrete Stochastic Optimization for Semiconductor Manufacturing Systems(2000) Mellacheruvu, Praveen V.; Herrmann, Jeffrey W.; Fu, Michael C.; ISRThe semiconductor industry is a capital-intensive industry with rapid time-to-market, short product development cycles, complex product flows and other characteristics. These factors make it necessary to utilize equipment efficiently and reduce cycle times. Further, the complexity and highly stochastic nature of these manufacturing systems make it difficult to study their characteristics through analytical models. Hence we resort to simulation-based methodologies to model these systems.This research aims at developing and implementing simulation-based operations research techniques to facilitate System Control (through sensitivity analysis) and System Design (through optimization) for semiconductor manufacturing systems.
Sensitivity analysis for small changes in input parameters is performed using gradient estimation techniques. Gradient estimation methods are evaluated by studying the state of the art and comparing the finite difference method and simultaneous perturbation method by applying them to a stochastic manufacturing system. The results are compared with the gradients obtained through analytical queueing models. The finite difference method is implemented in a heterogeneous simulation environment (HSE)-based decision support tool for process engineers. This tool performs heterogeneous simulations and sensitivity analyses.
The gradient-based techniques used for sensitivity analysis form the building blocks for a gradient-based discrete stochastic optimization procedure. This procedure is applied to the problem of allocating a limited budget to machine purchases to achieve throughput requirements and minimize cycle time. The performance of the algorithm is evaluated by applying the algorithm on a wide range of problem instances.
Item A Geometric Algorithm for Automated Design of Multi-Stage Molds for Manufacturing Multi-Material Objects(2000) Kumar, Malay; Gupta, Satyandra K.; ISRThis paper describes a geometric algorithm for automated design of multi-stage molds for manufacturing multi-material objects.In multi-stage molding process, the desired multi-material object is produced by carrying out multiple molding operations in a sequence, adding one material in the target object in each mold-stage.
We model multi-material objects as an assembly of single-material components. Each mold-stage can only add one type of material. Therefore, we need a sequence of mold-stages such that (1) each mold-stage only adds one single-material component either fully or partially, and (2) the molding sequence completely produces the desired object.
In order to find a feasible mold-stage sequence, our algorithm decomposes the multi-material object into a number of homogeneous components to find a feasible sequence of homogeneous components that can be added in sequence to produce the desired multi-material object.
Our algorithm starts with the final object assembly and considers removing one component either completely or partially from the object one-at-a-time such that it results in the previous state of the object assembly. If the component can be removed from the target object leaving the previous state of the object assembly a connected solid then we consider such decomposition a valid step in the stage sequence. This step is recursively repeated on new states of the object assembly, until the object assembly reaches a state where it only consists of one component.
When an object-decomposition has been found that leads to a feasible stage sequence, the gross mold for each stage is computed and decomposed into two or more pieces to facilitate the molding operation. We expect that our algorithm will provide the necessary foundations for automating the design of multi-stage molds and therefore will help in significantly reducing the mold design lead-time for multi-stage molds.
Item A Geometric Algorithm for Multi-Part Milling Cutter Selection(2000) Yao, Zhiyang; Gupta, Satyandra K.; Nau, Dana S.; ISRMass customization results in smaller batch sizes in manufacturing that require large numbers of setup and tool changes. The traditional process planning that generates plans for one part at a time is no longer applicable.In this paper, we propose the idea of process planning for small batch manufacturing, i.e., we simultaneously consider multiple parts and exploit opportunities for sharing manufacturing resources such that the process plan will be optimized over the entire set of parts. In particular, we discuss a geometric algorithm for multiple part cutter selection in 2-1/2D milling operations.
We define the 2-1/2D milling operations as covering the target region without intersecting with the obstruction region. This definition allows us to handle the open edge problem. Based on this definition, we first discuss the lower and upper bond of cutter sizes that are feasible for given parts. Then we introduce the geometric algorithm to find the coverable area for a given cutter. Following that, we discuss the approach of considering cutter loading time and changing time in multiple cutter selection for multiple parts. We represent the cutter selection problem as shortest path problem and use Dijkstra's algorithm to solve it. By using this algorithm, a set of cutters is selected to achieve the optimum machining cost for multiple parts.
Our research illustrates the multiple parts process planning approach that is suitable for small batch manufacturing. At the same time, the algorithm given in this paper clarifies the 2-1/2D milling problem and can also help in cutter path planning problem.
Item Selecting Flat End Mills for 2-1/2D Milling Operations(2000) Yao, Zhiyang; Gupta, Satyandra K.; Nau, Dana S.; ISRThe size of milling cutter significantly affects the machining time. Therefore, in order to perform milling operations efficiently, we need to select a set of milling cutters with optimal sizes. It is difficult for human process planners to select the optimal or near optimal set of milling cutters due to complex geometric interactions among tools size, part shapes, and tool trajectories.In this paper, we give a geometric algorithm to find the optimal cutters for 2-1/2D milling operations. We define the 2-1/2D milling operations as covering the target region without intersecting with the obstruction region. This definition allows us to handle the open edge problem. Based on this definition, we introduced the offsetting and inverse-offsetting algorithm to find the coverable area for a given cutter. Following that, we represent the cutter selection problem as shortest path problem and discuss the lower and upper bond of cutter sizes that are feasible for given parts. The Dijkstra's algorithm is used to solve the problem and thus a set of cutters is selected in order to achieve the optimum machining cost.
We believe the selection of optimum cutter combination can not only save manufacturing time but also help automatic process planning.
Item A Geometric Algorithm for Finding the Largest Milling Cutter(2000) Yao, Zhiyang; Gupta, Satyandra K.; Nau, Dana S.; ISRIn this paper, we describe a new geometric algorithm to determine the largest feasible cutter size for2-D milling operations to be performed using a single cutter. In particular:1. We give a general definition of the problem as the task of covering a target region without interfering with anobstruction region. This definition encompasses the task of milling a general 2-D profile that includes bothopen and closed edges.
2. We discuss three alternative definitions of what it means for a cutter to be feasible, and explain which of thesedefinitions is most appropriate for the above problem.
3. We present a geometric algorithm for finding the maximal cutter for 2-D milling operations, and we show thatour algorithm is correct.
Item Comparison of Run-to-Run Control Methods in Semiconductor Manufacturing Processes(2000) Zhang, Chang; Deng, Hao; Baras, John S.; Baras, John S.; ISRRun-to Run (RtR) control plays an important role in semiconductor manufacturing.In this paper, RtR control methods are generalized. The set-valued RtR controllers with ellipsoidapproximation are compared with other RtR controllers bysimulation according to the following criteria: A good RtR controller should be able to compensate for variousdisturbances, such as process drifts, process shifts (step disturbance)and model errors; moreover, it should beable to deal with limitations, bounds, cost requirement, multipletargets and time delays that are often encountered in realprocesses.
Preliminary results show the good performance of the set-valued RtRcontroller. Furthermore, this paper shows that it is insufficient to uselinear models to approximate nonlinear processes and it is necessary to developnonlinear model based RtR controllers.
Item The Set-Valued Run-to-Run Controller with Ellipsoid Approximation(2000) Zhang, Chang; Baras, John S.; Baras, John S.; ISRIn order to successfully apply Run-to-Run (RtR) control or real time control ina semiconductor process, it is very important to estimate the processmodel. Traditional semiconductor process control methods neglect theimportance of robustness due to the estimation methods they use.A new approach, namely the set-valued RtR controller with ellipsoidapproximation, is proposed to estimate the process model from acompletely different point of view. Because the set-valued RtRcontroller identifies the process model in the feasible parameter setwhich is insensitive to noises, the controller is robust to theenvironment noises.Ellipsoid approximation can significantly reduce the computation load for the set-valued method.
In this paper, the Modified Optimal Volume Ellipsoid (MOVE) algorithm is used toestimate the process model in each run. Designof the corresponding controller and parameter selection of the controller are introduced.Simulation results showed that the controller is robust toenvironment noises and model errors.