Analysis of Plan, Design and Simulation of AMHS System and Fab Layout

Automatic delivery systems and Fab layout planning, design and simulation analyses are described as common examples, such as “300mm semiconductor Fab” and “Automatic transport systems in FPD Fab,” as discussed earlier. The ground limits are too limited to approach theoretically, and the reduction cannot express the contents in a continuous manner, so examples are given in the order in which the automatic return system and the Fab layout are prepared in a typical production line. Although it consists of examples, it contains a great deal of information depending on the level of understanding, as it is based on phenomena that may appear in the actual production line.

Introduction
  Establishing a plan for the layout and utilization of factory facilities using an automatic delivery system is critical for efficient use of all possible resources. Since many factors can occur and affect the production line between each automatic return system, process equipment, and the operating system of the Fab and the operating personnel of the production site when actual production is started inside the Fab, it will be possible to understand the importance of forecasting and pre-planning to achieve the desired output and yield by absorbing these factors.

The preparation of the planning, design and layout of these production lines and automation systems and the analysis of the inter process logistics are essential considerations, which can lead to efficient operation of each system. In addition to consideration in new production lines, the status and capacity of logistics facilities for existing production lines can be analyzed to identify problems and derive improvement plans through experiments.

Basic calculations can be made through mathematical and statistical simulations of each facility for the purpose of production and the conveying systems responsible for transport between these facilities in order to plan and design new production plants and to derive improvement plans for each facility for existing production lines. However, simulations using mathematical models have limitations in reflecting the realistic phenomenon of each system (each facility in the production line and automatic return system). By using computer simulations, a scenario for facility operation can be prepared and programmed in advance to the production line to verify that each system is used in a balanced manner (Utilization) and transport of production work (WIP) is smoothly supplied. The advantages of simulation are that it provides a practical approach to productivity improvement through the layout of the actual installation, etc., and route analysis of the work space and production line logistics. The basis for these manufacturing systems will be based on the Queuing theory and the SCM (Supply Chain Management) theory. In addition, this article will be developed as part of the verification and optimization of the adequacy of each initial written facility layout, using basic mathematical techniques and several computer simulation programs to illustrate.

1. Production Line Design Overview
The meaning of designing a production line for the first time approaches what to do with the framework of the production line. Depending on geographical and spatial constraints for production lines, single-layer or multi-layered production sites can be constructed, and the types of automatic delivery systems that can be used may vary depending on the composition of single-layer or multi-layered production sites. This basic definition of the frame of the production line is to define more detail.

Since each production plant has different production objectives, the approach may be slightly different depending on the production objective, but the fundamental approach theory or method is the same.
To analyze the feasibility through simulation and the capacity of the system when a basic plan for the composition of the production line is drawn up, the following three main steps can be approached.

First, with the theoretical analysis approach of Statistic Analysis, this is a quantitative analysis and optimization of Work In Process (WIP) occurring within the production line, based on the Queuing Theory, Jackson Network and Little’s Law, calculate the sizing of each Process Tools and Tool Group and Stocker based on the basic operation inventory. Calculation of inventory quantities can lead to an increase in cycle time due to an increase in workmanship (WIP) that exists within the production line, which can cause huge inventory costs.
Traffic volume analysis is performed based on Process Data (Flow and Production Volume) and Layout to determine the density of the material flow of work that occurs within each production line, and these two analyses are used as the basis for the next step (Discrete Event Simulation) and can determine the direction of analysis.

Second, by using Discrete Event Simulation Model, Simulation Model is used as the basic data for development based on data analyzed through Statistic Analysis. The development of simulation models should be carried out to reflect basic operating algorithms for AGV, OHS/T and Stocker, all automated material handling systems within the production line.

Third, the data extracted from the Discrete Event Model is statistically analyzed, and the simulation results are visualized so that the suitability of the system can be determined.

2. Simulation Assumption
Simulations are simulated actual production lines, so many simplifications or assumptions are defined and used to an extent that does not violate the purpose and scope of the analysis based on layout, industrial engineering theory, and simulation theory that reflects realistic constraints.

Static Analysis Assumption
Based on the Queuing theory to determine the storage requirements of the buffer through quantitative analysis of the amount of work (WIP) generated within the production line, each Stoker is considered to be a buffer placed in front of equipment where the Arrival Rate and Service Rate follow the Exponential Distribution. Therefore, the process equipment supplied with work from a particular stocker on the Process Flow is considered a tool group, and the tool group’s track time is calculated by considering the target time, number of process repetitions, and number of equipment in the Tool Group.

Event Simulation Model Assumption
To develop and experiment an Event Simulation Model based on the results of a Static Analysis analysis, each event identifies the distribution type of the system depending on which system is simulated during Simulation and applies it to the simulation model. Furthermore, depending on the nature of the production line, Warm-up Time should be defined to avoid bias in the simulation results (Date) from the start of the initial simulation until the simulation model reaches a steady state.

Layout Assumption
Based on the final written layout, the available location or quantity of the other systems shall be defined for each system to which the systems are accessed. In addition, each transport system shall be defined on the layout for use as input data for simulations, as well as for paths that can be used for commercial use or bypass.

3. Static Analysis
Quantitative analysis is based on the following basic data.
 Process Flow Information
 Production Mix
 Special Move Requests
 System Expectations and Requirements
 Factory Layout

[Table 1] Criteria of Production Line Design

Using the process flow information, calculate the transfer requirements of the supplies according to the process flow through the process flow information. In this case, the flow of non-normal workpieces in the production base data is reflected.

[Table 2] From-To Matrix by Production Plan

When a basic From-To Matrix is created, data is generated to yield a certain quantity of quantitative buffer. From-To matrix shall be any quantity transferred for product production in the production line, where the location of the buffer available in the layout should be defined. Using the number of buffers defined by the basic data and layout and the service rate per unit hour that the system can handle, the following formula is based on the Jackson Network in the Queuing theory as an example of the following formula for the validity of buffers in the layout.

[Table 3] Calculation of Buffer Robot’s Processing Requirements

Perform a quantitative calculation of the amount that the buffer can handle for each process equipment. In [Table 3], the amount the Buffer has to deal with in Bay 2 is 133% more than the capacity the system has. In such cases, the layout should be re-examined to distribute the buffer or to increase the number of systems to be processed.

The basic supply chain management (SCM) in the above example, the management of logistics and supply chain, can predict the time of the manufacturing process and the amount of WIP analysis and logistics transfer according to the logistics transfer between schedule operation.

The planning, design and analysis of automation of production lines is based on WIP, Cycle Time and Throughput, which are performance indicators of production system for efficient manufacturing process operation. Here, WIP and CR, a measure of performance, are better when lower and Throughput is better when higher, and these three indicators can be expressed as WIP=Cycle Time X Throughput.

By understanding the transfer of WIP and logistics according to Little’Law’s theory, the impact on WIP, operation rate, Cycle Time, and Throughput can be summarized as follows.

 Increasing the Input Amount when Cycle Time is equal = More WIP WIP(Increase) = CT * Start Rate(Increase) = Increase of Required Storage(Bad)
 Increased Cycle Time when input is equal = More WIP WIP(Increase) = CT(Increase) * Start Rate = Increase of Required Storage(Bad)
 Simultaneous Increase of Input and Cycle Time = Much More WIP WIP(Increase a Lot) = CT * Start Rate = Increase of Required Storage(Very Bad)
 Decrease of Cycle Time when Input is equal = WIP OK WIP = CT(Decrease) * Start Rate(Increase) = Good

4. Discrete Event Simulation
Analyzing Method and Simulation

Analyzing methods can be expressed by mathematical calculations and need to be simplified through assumptions about many variables. However, in complex cases, such as production lines, these analyzing methods are often not mathematically solvable. If the answer to the problem is not available by this analyzing method, the simulation technique is applied.

 Analyzing Method: x+a=b – Expressed as x=b-a in mathematical expressions of Analyzing Methods
 Simulation: x+a=b – The simulation repeats the value of x to repeat all possible values that satisfy the expression.

Analytical methods are used to extract simulation initials using Markov Chain and probabilistic methods, which require a lot of time, such as when performed repeatedly, but can minimize simulation time if initial input values are obtained, so analyzing methods and simulation methods cannot be considered separately and should be applied step by step.

What is Simulation Modeling?
Simulation cannot solve the problem for phenomena that may occur at all production lines. However, they can respond to specific systems or phenomena, provide answers to the cases of operating different production lines, and express the following simulation and simulation models.

 Simulation cannot solve problems.
 Simulation is a Tool used for the following.
-Tests and Responds for Future Systems
-Analysis of possible options
-Evaluation of Changes
-Determination of System Suitability
 Simulation can provide the answer to the question ‘What if’ scenario for the system.
 Based on an understanding of the operation of the system, the simulation model abstracts the system using programming to predict changes.
-Test new concepts or systems before running
– Use established theories or hypotheses to predict the future shape of the system to predict the effects of changes in the system or changes in the operating methods of the system.
 In case of simple systems, mathematical models can solve the problem.
– The mathematical model gives you an answer that is close to accuracy,
– It can be simplified by using too many assumptions to get the answer.
 Use complex models for complex systems to approach accuracy.

The following [Figure 1] illustrates the proximity of the simulation model to the actual system. Such simulations can be said to have enough accuracy to simulate the actual system and obtain an answer to the problem.

[Figure 1] Simulation Model

Simulation vs. Real System
Given the nature of the model, it is generally assumed that the real system is assumed, so it is common to apply a safety factor to the simulation model to give stress to the model. [Figure 2] shows how much safety factor the simulation should have, by analyzing the log of the actual system and comparing the characteristics of the return demand of the simulation model with the index distribution and applying the normal distribution.

Typically, when an event simulation is performed, most exponential distributions are applied to generate events. Although it takes a lot of consideration to explain why, typically when conducting a Discrete Event simulation,

The probability density function of probability variable X is as follows:

X, when given, follows the exponential distribution with parameter λ.
In other words, if the average number of events in a unit of time occurs as a Poisson probability,
It is shown in [Figure 2] below that the probability distribution for time interval X after one event is followed by the exponential distribution is appropriate for the production system and that the index distribution can indicate a distribution closer to the actual system.

[Figure 2] Real System vs. Simulation

Warm-up Time
Simulation is an imitation of Real System using a computer. Even if the production line is constructed with Real System, the planned performance (stability: Steady State) can be reached over a period of time for each system to perform its initial planned performance.

Computer simulation also requires preparation for this process. Simulation of an initial modelling system causes each system to start at the same time and at the start of the simulation, the simulation model is not properly handled and places a sudden stress on the system due to the high processing demands. This phenomenon should be eliminated in the process of deriving results, and the simulation theory defines it as ‘warm-up time’. Warm-up Time shall be applied at different intervals to the characteristics of each system, and [Figure 3] shows an example of the transport system in the production line.
Simulation is an imitation of Real System using a computer. Even if the production line is constructed with Real System, the planned performance (stability: Steady State) can be reached over a period of time for each system to perform its initial planned performance.
Computer simulation also requires preparation for this process. Simulation of an initial modelling system causes each system to start at the same time and at the start of the simulation, the simulation model is not properly handled and places a sudden stress on the system due to the high processing demands. This phenomenon should be eliminated in the process of deriving results, and the simulation theory defines it as ‘warm-up time’. Warm-up Time shall be applied at different intervals to the characteristics of each system, and [Figure 3] shows an example of the transport system in the production line.

 System not serviced and started in Idle State
 All return requests are initiated simultaneously by all initial systems.
 Therefore, the initial feed time rises rapidly and then reaches the steady state.
 Use the Steady State section after 20 days of warm-up.

[Figure 3] Example of Warm-up Time of Simulation

If the model is completed through analyzing methods and simulations as shown above and the analysis is performed by expressing the actual production line, the most important thing is accurate input data. In the case of first-time production lines, initial input data can be obtained by prediction and calculation, but the most accurate response is to use the production information of the actual production lines when attempting to derive alternatives for problems with existing production lines.

Derivation of Alternatives through Simulation
How the final conclusions should be expressed in the Discrete Event simulations should be addressed by a number of studies depending on the derivation of alternatives to which systems. [Figure 5] represents the final conclusion of the alternative to the transport system at the production line.

[Table 4] shows the average time required by the return request of the transport system.
The simulation shows the average time for the number of events to be moved purely by the system according to the return request from the process equipment in a discrete probability, while the ‘Transfer Time’ indicates the time taken for the process equipment to be returned, and the ‘Wait Time’ indicates that each transport system is serviced by the return request of the equipment.
Here, the smaller the ‘Wait Time’ is, the better it is and can be serviced on time whenever there is a return request. [Table 4] and [Figure 5] express the same conclusion differently. In [Figure 5], [Table 4] shows the average amount of time taken by all of the event simulations shown on the Y-axis, and [Figure 5] shows the distributed values and completion rates of each event.

[Figure 4] Conveying System using AutoModTM 10.0
[Table 4] Simulation Result of Path Move System A-Line

[Figure 5] simulated two to four systems to calculate the appropriate number of systems required when a return request from process equipment occurred at a particular line in a transport system, such as an AGV system or an OHS system, and the performance of the system was performed by determining the capability of the transfer system to respond when 97% of the return request was made by the process equipment.
It is known that 97% of the yellow line can be completed when the first two systems are operated is between 570 and 600 seconds (9.5 to 10 minutes), and 4 of the 270 to 300 seconds (4.5 to 5 minutes) when operating the three systems, for a return requirement of 97% between 210 and 240 seconds (3.5 to 4 minutes).
Is it best to introduce four of the best performance models in this simulation? The answer to this is unknown because the performance of the three and four systems are not significantly different, although two systems are significantly behind. It cannot be said that the time difference between three and four one-minute units is more efficient than the cost of introducing the system. In any case, it should be considered that the performance of the appropriate system is correlated with the characteristics and importance of the system to be applied, and in most cases too many systems are introduced which can degrade the performance of the system due to traffic congestion between many systems.

[Figure 5] Simulation Result of Discrete Event of Path Move System A-Line

4. Evaluation of the Layout
If the actual system is started after the basic concept of the production line has been established and the design has been completed, the feed system at each live line will transport the WIP (Work In Process). The basic path of this large transport system is defined in the layout, and [Figure 6] is an example of traffic density analysis using EDS Corporation FactoryFlowTM Version 7.1, which indicates that traffic increase occurs in certain areas. For these parts, it is necessary to change the path of the logistics transfer or distribute the route through the preparation of the bypass.

[Figure 6] Traffic Intensity Analysis using FactoryFlowTM

Conclusion
When planning and executing the layout of the automatic delivery system and the Fab, or performing an analysis through simulation to derive problems for existing production lines, it is not possible to proceed only with specific personnel.
The basis of simulation and analysis can be more precisely modeled the actual line, depending on how detailed it is to represent all cases, such as the setting of the basic frame of the initial production line, many variables, realistic constraints and feasible ideas, and how much it can be introduced or simplified into the simulation model. Modeling through simulation may not reflect all cases, but should be simplified based on mathematical modeling and statistical theory at each stage, and numerous processes should be repeated to validate the validity of each step.
Participation and active participation in each required department may be the basis for many of these processes.

[Figure 7] Steps of Design and Layout of the Production Line

The application of simulation techniques to the operation of existing domestic manufacturing lines or to the composition of new lines may be considered to be a different field in practice from the field in which domestic manufacturing businesses used to be operated and may in most cases be executed by experience in introducing manufacturing lines. In the event that verification procedures for the manufacturing process and the elicitation of appropriate alternatives are carried out through these objective assessments, the determination to recognize and implement the need for objective analysis by the chief executive, as well as the person who runs the manufacturing line, that the production efficiency of the production line can be shown by a high increase.

[Reference]

J.Hopp, Mark L. Spearman : FACTORY PHYSICS-Foundations of Manufacturing Management, Second Edition, McGraw-Hill

Averill M.Law, W.David Kelton : SIMULATION MODELING AND ANALYSIS, Third Edition, McGraw-Hill

Hillier/Lieberman : INTRODUCTION TO OPERATIONS RESEARCH, Seventh Edition, McGraw-Hill

Katsundo Hitomi, 조규갑,차성운 공역 : 생산시스템공학, SiTech

JT Shin / CEO

Close Menu