Development of Material Control System for Next …...RGV controller application OHV controller...
Transcript of Development of Material Control System for Next …...RGV controller application OHV controller...
85
Vo l . 4 2 N o . 2 2 0 0 9
1. Introduction
In recent years, conventional cathode-ray tube televisions have begun to be replaced by liquid crystal and plasma televisions. Manufacturers of electric appliances are actively investing in new factories. Thus far we have been delivering physical distribution equipment and systems mainly for liquid crystal manufacturing lines. A liquid crystal manufacturing line consists of four main processes; (1) an array process to make a thin film transistor (TFT) array circuit on a glass substrate, (2) a color filter process to generate the three primary colors, RGB, on another glass substrate, (3) a cell process to stick the completed array substrate and the color filter substrate together and inject liquid crystal, and (4) a module process to integrate a driver (driving circuit), backlight and other components into the completed cell to make a display. (1)
The TFT array process flow consists of four or five cycles of (1) a washing process, (2) a photolithography process, (3) an etching process, (4) an inspection process, (5) a repair process, and other processes. The production flow is complicated and the production volume is large. Efficient production requires a material control system (hereinafter called MCS) which controls transfers between process devices and glass substrate stockers in a TFT array process. As the average liquid crystal television gets larger (mainstream television sets are currently 30 to 40 inches), the glass substrates used for liquid crystal displays must also get larger. Because of this, the conventional cassette transfer method that transfers glass substrates in cassettes is approaching its limit, and the single substrate transfer
method, which transfers each glass substrate individually, is being considered in its stead. We developed the MCS that has a function for controlling cassette transfer and single substrate transfer for TFT array processes on liquid crystal manufacturing lines. We also established a technology to verify MCS functions and capacity by using a physical distribution simulator, and verified the functions and capacity of the MCS.
2. Development of the MCS
2.1 System configurationThe MCS is a computer system that receives instructions for transferring glass substrates from a manufacturing execution system (hereinafter called MES) in a factory and controls transfers of glass substrates from the current process to the next process by using cranes, vehicles, conveyors and other transfer devices in the factory. The MES is a computer system that performs (1) production planning, (2) liquid crystal manufacturing process control, (3) process device control, and (4) lot control. The MCS we have developed consists of (1) the MCS that controls cassette transfer (hereinafter called CMCS), (2) the MCS that controls single substrate transfer (hereinafter called SMCS), (3) equipment controllers that control individual transfer facilities, and (4) equipment control panels that control transfer devices. Figure 1 shows the system configuration.2.2 Cassette material control system (CMCS)2.2.1 Main functionsThis section describes the main functions of the CMCS; the transfer control function, the inventory management
Development of Material Control System for Next Generation
Liquid Crystal Glass
HASEGAWA Fumio : Senior Researcher, Control System Project Department, Products Development Center, Corporate Research & Development OKAJIMA Kazumichi : Control System Project Department, Products Development Center, Corporate Research & Development SHIDA Michinori : Control System Project Department, Products Development Center, Corporate Research & Development MURAMATSU Yuya : Electrical System Department, Products Development Center, Corporate Research & Development NAKAMA Mitsuaki : Manager, Quality Assurance Department, Logistics Systems Operations
The material control system ( MCS ) that controls the transportation between process equipment and stockers is indispensable to improve productivity because the process flow is complex and production is high in the LCD production line. It is difficult to transfer the LCD glass with the cassette transfer method, as the LCD glass substrate is becoming larger. Therefore, a single substrate transfer method to transport individual LCD glass substrate is needed. IHI has developed the MCS with optimum cassette transfer function and single substrate transfer function for the TFT array process of the LCD glass production line. The system outline of the developed MCS and the main features are introduced.
英文.indb 85英文.indb 85 09.12.3 11:39:14 AM09.12.3 11:39:14 AM
86
Vo l . 4 2 N o . 2 2 0 0 9
function, and the equipment control function.2.2.1.1 Transfer control functionThe transfer control function mainly consists of three types of processing; transfer route search, transfer command creation, and transfer command execution. Transfer instructions transmitted from the MES to the CMCS designate a starting point, a destination, and a cassette ID in a From-To format, for example, “Transfer cassette ID001 from process A to process G.” To transfer the cassette from process A to process G, the transfer route search process searches for the fastest route to the destination from among several transfer routes. Figure 2 shows a transfer route example.
After a transfer route is determined, the transfer instruction creation process creates transfer instructions for cranes, vehicles, and other transfer facilities on the transfer route. The created transfer instructions are transmitted to individual transfer facilities by the transfer instruction execution process. It is possible to cancel the transfer or change the destination on the way. The transfer states are monitored and displayed in real time. Figure 3 shows a tracking monitor screen that displays transfer states.2.2.1.2 Inventory management functionThis function controls the positions and IDs of cassettes being transferred and those stored in automatic stockers,
SMCS
Buffer CV
Simulation model
Equipment control panel
Device control panel
OHVcontrolpanel
Crane1
G8crane
Actual machine
(Note) CVSTK-CRGV-C
: Conveyer : Stocker crane controller : Rail guided vehicle controller
OHV-CG8STK-C
: Over head vehicle controller: Stocker crane controller of G8
RGV
STK-C1
STK-C2
RGV-C OHV-C G8STK-C
Equipment controller
Transferterminal
Transferterminal
CMCS
Coresystem
Operationsystem
Controlsystem
SMCS application
CMCS application
MES application for verification
MES for verificationWeb server application
Stocker controller application
Stockercontrolpanel
Stocker controller
Simulationapplication
RGV controller application
OHV controllerapplication
OHVcontroller
OHV
Web serverCMCS serverSMCS server DB server (clustering configuration)
Fig. 1 MCS system configuration
Crane2
CraneVehicle (RGV)
Process A
: Examples of available(Note)Vehicle (OHV)
CassetteID001
Fig. 2 Transfer route
Process G
transfer routes
Fig. 3 Tracking monitor
英文.indb 86英文.indb 86 09.12.3 11:39:14 AM09.12.3 11:39:14 AM
87
Vo l . 4 2 N o . 2 2 0 0 9
and the number of empty cassettes stored in automatic stockers. Figure 4 shows the entire-line inventory management monitor and Fig. 5 shows the inventory management monitor for buffer. 2.2.1.3 Equipment control functionThis function controls the working states and operating conditions of cranes, vehicles, and other transfer devices, glass substrate stockers and ports before devices. Equipment working states can be checked on the tracking monitor screen.2.2.2 Features2.2.2.1 Optimum transfer route algorithmTo improve the production efficiency of a factory, it is important to reduce transfer time between processes and improve transfer efficiency. The CMCS has an algorithm that searches for a route to reach the destination in the shortest time by ascertaining equipment states in real time and taking into account the states of process devices and transfer devices in the factory at the moment. Figure 6 shows an example of a search by the optimum transfer route algorithm. 2.2.2.2 High availabilityFor the most part, liquid crystal lines are operated continuously, 24 hours a day, 365 days a year, only stopping production for several days of maintenance work. If a line is stopped due to system trouble, it has
a signif icant impact on production. Therefore, it is important to improve system availability. Availability of a computer system is generally measured in terms of the mean time between failures (MTBF) and the mean time to repair (MTTR) of the system. In the CMCS, the system control process monitors the operation of each process. When it detects a process has gone down, it automatically restores that process. The system is equipped with this function to minimize the frequency and time of system stoppage and improve the availability. Figure 7 shows the concept of process monitoring method.2.2.2.3 Flexibility of system changesOn a liquid crystal manufacturing line, equipment and ports may be added or changed in order to change manufacturing processes and improve the manufacturing capacity. Such system additions and changes should be able to be done without stopping the line. The CMCS collectively controls system data in a database server and the communicat ion between processes also takes place via the database, rather than directly between processes. Such architecture allows processes to be changed or added without stopping the system when a change or addition of equipment or
Fig. 4 Inventory management monitor
Fig. 5 Inventory management monitor for buffer
Current position
Crane
Vehicle (OHV)
Vehicle (RGV)
Destination
: Shortest route
: Route search result
Congestion of vehicles
Duringmaintenance
(Note)
Fig. 6 Example of search result of the optimum transfer route algorithm
Suspendeddue to
abnormality
Operation information
System controlprocess
Transfer controlprocess
Transfer devicecommunication process
MES communicationprocess
Screen managementprocess
Automatic transferprocess
Process restart
Fig. 7 Concept of process monitoring
英文.indb 87英文.indb 87 09.12.3 11:39:15 AM09.12.3 11:39:15 AM
88
Vo l . 4 2 N o . 2 2 0 0 9
2.2.3.2 Optimum transfer route algorithm(1) Modeling of transfer
Build models of cranes, vehicles, transfer units, and other transfer devices and cassette buffer stockers in charts. Figure 9 shows an example of the layout and modeling. A chart consists of points and branches that connect adjacent points. Branches have directions that indicate the directions in which cassettes can be moved.
(2) Determining the weight of a routeDefine weights for all branches in a model. Select
the route that minimizes the sum of the weights of the branches as the optimum route from among the many transfer routes that connect a starting point and a destination. Weight each branch based on the time required for a cassette to go through the branch, and the weight is increased as a function of the volume of transfer. The volume of transfer is expressed as the number of transfer instructions being executed and transfer instructions to be executed in a transfer route. Increased weight means that the route is congested and requires a longer time to complete a transfer. In the example shown in Fig. 10, the lower route is the optimum transfer route if WU > WD .
As shown in the example in Fig. 11, when transfers are concentrated in a particular route, the weight of the route increases due to the weight
process maintenance is required.2.2.3 Optimum transfer route algorithm2.2.3.1 OutlineOptimum route search on a liquid crystal line has the following issues:
(1) The process flow is complicated and contains many processes, so it is virtually impossible to register all transfer routes in the system in advance. It is necessary to determine a transfer route for each transfer.
(2) There are two or more transfer devices that connect processes. Several transfer routes are available to reach the same dest ination. It is necessary to select the optimum transfer route from among several routes.
(3) A process may have devices that are suspended due to trouble or for maintenance, making some routes unavailable or causing congestion in some routes due to concentrated transfer instructions. It is necessary to determine a transfer route that avoids such routes.
In short, it is necessary to have an algorithm that takes the states of process devices and transfer devices in a factory into consideration in real time during transfer and then selects the route from among several that will reach the destination in the shortest time. Therefore, we developed an algorithm that determines routes and transfer devices that will minimize transfer time at each passing point on the transfer routes. It selects the optimum route and devices depending on the states of the line, and if a route has a device that is out of order, it searches for a new optimum route in order to complete the transfer without stopping. It can also avoid any route where congestion occurs. Figure 8 shows the process f low of the optimum transfer route algorithm.
Ascertain the operating state ofeach transfer device in real time
Check transfer instructionscurrently executed
Reflect route informationDetermine weights of routes
Search forthe optimum route
Check for deadlock
Transfer routesavailable?
Determine a transfer route
Optimum transfer route algorithm
Search for a route withthe optimization module
Start transfer → Transferring
Transfer completed
Transfer control processing
Passing point orend point?
Passingpoint
End point
No
Fig. 8 Process flowchart of the optimum transfer route algorithm
Vehicle (OHV)
BranchPoints
Vehicle (RGV)
Crane
Stocker
Transfer unit
: Passing point
: Final destination
(Note)
Fig. 9 Layout and model
Volume of transfer (X)
Sum of weights in the upper route:WU = W1 + W2 + W3 + W4 + W5 + W6 + W7 + W8 + W9Sum of weights in the lower route:WD = W1 + W10 + W11 + W12 + W13 + W14 + W9
Weight (Y)
Weight function : Y = W ( X )
W12
W13
W14
W11
W10W1
W2
W3
W4 W5 W6
W7
W8
W9
Fig. 10 Model and weighting
英文.indb 88英文.indb 88 09.12.3 11:39:16 AM09.12.3 11:39:16 AM
89
Vo l . 4 2 N o . 2 2 0 0 9
function. As a result, another route with a smaller sum of weights is determined to avoid congestion.
It is possible to make a particular route unavailable when a device is suspended for maintenance by increasing the weight of the route sufficiently as shown in Fig. 12.
(3) Check for deadlockFor example, as Fig. 13 shows, when a transfer of
cassette ID1 from point A and a transfer of cassette ID2 from point B are executed at the same time, both cassettes have no space available at the next transfer point and cannot be transferred. Such a state is called deadlock.
To prevent deadlock, the system checks whether or not there is space available for transferring a cassette at the next transfer device at and the final
destination and determines the order of transfer (Fig. 14).
2.3 Single substrate material control system (SMCS)
In the conventional cassette transfer method, all cassette transfer instructions are transmitted from the MES which is a higher-order system of the MCS. If all transfer instructions are received from the MES for each substrate in the single substrate transfer method, the following problems occur:
(1) The volume of communications between the MES and MCS becomes huge.
(2) The MES needs to control t racking of each substrate which causes the volume of processing to increase.
Therefore, we have developed a mechanism that allows the SMCS to receive process information and sor t information for each lot from the MES and utilize the information to control the f low of glass substrates in a single substrate transfer process without receiving any transfer instruction from the MES. Figure 15 shows the SMCS system configuration.2.3.1 Assumed layoutIn comparison to the cassette transfer method, the single substrate transfer method is more effective when it is applied to processes with complicated substrate flow. Our development targeted inspection processes and repair processes where defects are repaired after inspection (hereinafter called inspection and repair processes). In inspection and repair processes, substrates are divided into those to be transferred directly to the next process and those to be repaired before being transferred to the next process, depending on inspection results (this is called a branch point). The next inspection process receives substrates that have passed inspection in the previous process and substrates that have been repaired (this is called a conf luence point). As the number of inspection processes increases, the number of branch and conf luence points increases and the f low of glass substrates becomes more complicated.
Figure 16 shows the layout of a single substrate transfer line that we assumed as the basis for development of the SMCS.
The glass substrate flow is outlined below:(1) Glass substrates are placed into a cassette and
transferred to a carry-in port in the single substrate transfer process.
(2) The loader/unloader device carries out each glass
Congestionof vehicles
Optimum routeavoidingcongestion
Fig. 11 Optimum transfer route to avoid congestion
Suspended
Increasedweight
Increasedweight
This is the optimum route.
This is not the optimum route as the weight is increased due to a suspended device.
Fig. 12 Optimum transfer route to count equipment status
Cassette 2 is not allowed to move from the buffer until cassette 1 moves to the buffer.
Point A
Point B
Buffer/Port
Crane
Deadlock
Shifter
CassetteID1
CassetteID2
Fig. 13 Deadlock condition
Space is available.Space is available.
Buffer/Port
DestinationStartingpoint
Crane
ShifterCassette
ID1
Fig. 14 Checking for free space of equipment
英文.indb 89英文.indb 89 09.12.3 11:39:16 AM09.12.3 11:39:16 AM
90
Vo l . 4 2 N o . 2 2 0 0 9
substrate and transfers it to the single substrate transfer process.
(3) Each transfer device transfers the glass substrates to each process device (inspection or repair device) according to the process flow.
(4) If all process devices in the next process are occupied for processing, the glass substrates are temporarily transferred to a buffer located in front of the device.
(5) After processing, the substrates are transferred to different points, depending on the processing results. For example, a substrate that has passed the processing
inspection at inspection device A is transferred to inspection device C, while a substrate that has been rejected is transferred to repair device B.
(6) The glass substrates are transferred to cassettes at specif ied carry-out por ts, depending on the processing results at each process device. Figure 16 shows the flow of glass substrates with blue arrows.
2.3.2 Main functionsThe SMCS has the following main functions; (1) a transfer control, (2) an inventory management, and (3) an equipment control. The inventory management and equipment control functions are basically the same as those for the CMCS,
ERPPlanning system
MESManufacturing execution system
ControlProcess device control system
ControlTransfer device control system
Scope of this development
(Note) ERP : Enterprise Resource Planning
MCSMaterial control system
Manufacturing instructions (date and time)Manufacturing specifications
Lot information
Work recordsManufacturing statesDefect information
Transfer instructions (for each lot)
Transfer instructions (for each substrate)
Transfer results (for each substrate)
Transfer results(for each lot)
Workinstructions
Workresults
MESDBQuality control Process control
Equipment managementScheduling
Production plans (monthly, weekly, and daily)
Production instructionsManufacturing methods
Manufacturing results, progress, work-in-process and inventory information
Integrated database
MCSDB
The volume of communications in this area increases when the cassette transfer method is changed to the single substrate transfer method.
Some MES functions are integrated into MCS.Develop a mechanism to receive process flow andsort information and allow the MCS to control the process flow in a process.
Fig. 15 SMCS system configuration
Processmanagement
Inventorymanagement
Transferdevice control
Transfer instructioncontrol
Processmanagement
Carry-in port Port B
Port C
Port D
Port E
Port A
Loader/Unloader Inspection
device A1
Inspectiondevice C1
Inspectiondevice C2
Inspectiondevice C3
Repairdevice D1
Repairdevice D2
Lifter A Lifter B Lifter C
Repairdevice B1
Repairdevice B2
Repairdevice B3
Inspectiondevice A2Buffer A Buffer B Buffer C
CV1
CV2
Carry-out port
(Note) CV : Conveyor
Fig. 16 Layout for single substrate transfer line
英文.indb 90英文.indb 90 09.12.3 11:39:17 AM09.12.3 11:39:17 AM
91
Vo l . 4 2 N o . 2 2 0 0 9
individual transfer devices based on work commands generated in the dispatch section. Also manages the states of individual transfer devices, process devices, and buffers and substrate tracking information. Figure 18 shows the SMCS operation block.
2.3.3 Features2.3.3.1 Optimization dispatch algorithmWe developed an algorithm that improves the efficiency of glass substrate flows at branch and confluence points where transfer instructions are centralized in a single substrate transfer process. Single substrate transfer processes may have various line forms, and we created the algorithm by combining simple rules independent from the layouts.2.3.3.2 Process flow controlThe SMCS receives process information and sor t information from the MES and controls process f lows of glass substrates that were conventionally executed by the MES. Table 1 gives description about process
so this section describes the transfer control section only. Figure 17 shows an example of the SMCS monitor. The transfer control function consists of (1) process cont rol processing, (2) d ispatch processing, (3) optimization logic processing, and (4) transfer device control processing. Dispatch here means the processing of assigning work commands generated based on transfer instructions to transfer devices.
(1) Process control processingDetermines the next destination and makes
transfer instructions based on process information and sort information received from the MES and results of processing at process devices.
(2) Dispatch processingMakes work commands and divides transfer
instructions generated in the process control section into work units that can be processed by the transfer device control section.
(3) Optimization logic processingDetermines the order of processing transfer
instructions and destinations in consideration of the current positions and stand-by time for transfer instructions, process device operating states, buffer states, and substrates and transfer device states in a process.
(4) Transfer device control processingOutputs transfer commands to control panels of
Fig. 17 SMCS monitor
Processinformation
Processing results
Process controlsection
Dispatchsection
Transferinstruction
Next transferinstruction tobe executed
Executabletransfer
instruction
Optimizationlogic section
Transferresult
SMCS
Transfer result
Transfer result
Transfer devicecontrol panel
Transfer devicecontrol section
Workcommands
Transfercommand
Sortinformation
MES
Fig. 18 SMCS operation block
Table 1 Process information and sort information
Item Description Example
Process information 1Information on which process a substrate processed by a process device is to be transferred next (Transferred to different destinations depending on processing results).
• Product successfully processed by process device A → Transferred to process device B
• Product not successfully processed by process device A → Transferred to process device C
Process information 2 Information on processing rules for each lot• Lot A is to be processed by process device A• Lot B is not to be processed by process device A
Process information 3 Recipe information for processing with a process device• Lot A is to be processed based on a recipe• Lot B is to be processed based on a recipe
Sort informationInformation as to which carry-out port a substrate is to be transferred depending on the processing results of a process device.
• Product successfully processed by process device C → Transferred to port 3
• Product not successfully processed by process device D→ Transferred to port 4
英文.indb 91英文.indb 91 09.12.3 11:39:17 AM09.12.3 11:39:17 AM
92
Vo l . 4 2 N o . 2 2 0 0 9
information and sort information. The SMCS creates a process f low for each glass substrate based on that information in order to control it.2.3.4 Optimization dispatch algorithm2.3.4.1 OutlineGlass substrates are transferred from several directions at branch and confluence points in a layout, and transfer inst ructions are concentrated at these points. The dispatch algorithm is intended to improve the efficiency of glass flows at these branch and confluence points by optimizing the processing order of transfer devices to which transfers are concentrated. Improved glass f low eff iciency reduces the glass substrate transfer time and minimizes the time from carrying a glass substrate into a process to carrying it out. Figure 19 shows branch and confluence points to which transfer instructions are concentrated.
2.3.4.2 Process flow(1) Checking states in a process Ascertain the operating conditions of process
devices and conveyors and the congestion states of glass substrates in a process.
(2) Making a list of executable transfer instructions Get information on each substrate and device
in the process from a list of transfer instructions currently issued in the process, check whether or not each transfer instruction is executable, and extract executable transfer instructions.
Figure 20 shows an example of executable transfer instructions. Green substrates in the figure correspond to executable transfer instructions. Specif ically, substrates waiting for transfer in buffers and a substrate first in line for a lifter are executable. Note that t ransfer instructions for substrates in buffer C and substrates suspended first
CV1
CV2
Port A
Port B
Port C
Port D
Port E
Carry-in port
Carry-out port
Transfer instructions are concentratedto branch and confluence points.
Loader/Unloader
Inspectiondevice A1
Inspectiondevice C1
Inspectiondevice C2
Inspectiondevice C3
BufferA
Lifter A
Lifter BLifter C
Repairdevice B2
Repairdevice B3
Repairdevice B1
Inspectiondevice A2
BufferB
BufferC
Repairdevice D1
Repairdevice D2
Fig. 19 Branches and confluence points that concentrate transfer instructions
Lifter A Lifter B Lifter C
CV1
CV2
: Glass substrates registered in a list of transfer instructions
: Glass substrates registered in the said list of transfer instructions and at the same time registered in a list of executable transfer instructions
: Glass substrates being processed and not subject to transfer instructions
(Note)
Carry-in port
Carry-out port
Port A
Port B
Port C
Port D
Port E
Loader/Unloader Inspectiondevice A1Buffer A
Repairdevice B1
Inspectiondevice A2 Buffer B
Repairdevice B2
Repairdevice B3 Buffer C
Inspectiondevice C1
Inspectiondevice C2
Inspectiondevice C3
Repairdevice D1
Repairdevice D2
Fig. 20 Example of executable transfer instruction
英文.indb 92英文.indb 92 09.12.3 11:39:18 AM09.12.3 11:39:18 AM
93
Vo l . 4 2 N o . 2 2 0 0 9
in line for lifter C are not included in executable transfer instructions, because all inspection devices in the next process, process C, are occupied for processing.
(3) Determining the priority of transfer instructions by calculationPrioritize transfer instructions in the list of
executable transfer instructions by calculation. Multiply six evaluation items, (1) destination, (2) substrate stand-by time, (3) load on a destination process dev ice, (4) load on a s t a r t ing point process device, (5) the number of substrates in the process, and (6) load on transfer facilities in the transfer direction, by weight, and then multiply the calculated values under conditions by weight.
3. Establishment of a system verification technique
3.1 System verification methodUsing a physical distribution simulator, we have verified functions and capacities of the CMCS and the SMCS. Specifically, we integrated the functions of factory machines and control panels that directly control the machines in simulation models, connect equipment controllers and the SMCS at lower levels of the CMCS to the simulation models, and make an imaginary factory based on these simulation models. We also connected verification MESs at higher levels of the CMCS and the SMCS to verify their functions and capacities.3.1.1 Simulation modelsWe built a TFT array process and a single substrate transfer process for a liquid crystal manufacturing line in the physical distribution simulator. Figures 21, 22 show images of the simulation models.3.1.2 Process flow for system verificationAs an example, the process f low for the CMCS is described below.
(1) The CMCS receives a transfer instruction from the MES.
(2) The CMCS makes t ransfer com mands for individual transfer devices based on the received transfer instruction and transmits the transfer commands to individual equipment controllers.
(3) An equipment cont rol ler breaks down the t r a n s fe r c o m m a nd t o wo r k c o m m a nd s fo r individual transfer devices and transmits them to the simulation model.
(4) The simulation model makes transfer devices in the model operate based on the work commands.
(5) After a transfer is completed, the simulation model transmits a work command completion report to the equipment controller.
(6) The equipment controller gives the CMCS a transfer complete report and in turn the CMCS gives a transfer complete report to the MES.
We use communication messages in the same format as that for actual commands. We use software and devices for the CMCS, SMCS, and equipment controllers that are the same as the real ones. By creating equipment control panels and transfer devices connected to the CMCS and equipment controllers in the simulation model, we are able to check transfer and production states of glass substrates under the same conditions and in the same process f low as in an actual factory. Off line system verification contributes to improved software reliability and faster system startup. Figure 23 shows the process f low of the system verification method.3.2 System verification resultsThis section describes the results of verifying functions and capacities of the CMCS optimum transfer route algorithm and the SMCS transfer efficiency improvement algorithm.3.2.1 CMCS verification results3.2.1.1 Algorithm functions verification results
(1) Avoiding congestionWhen transfer instructions were simultaneously
generated to transfer multiple cassettes from one stocker to another in a factory, the congestion avoidance function prevented transfer instructions from concentrating in the same route.
2F
1F
Stocker
Crane
Traverser
Process device
Vehicle (RGV)
Vehicle (OHV)
Transfer unit
Lifter
Cassettein stocker
Fig. 21 Simulation model of TFT array process
Cassette Buffer LifterCassetteport
Loader/Unloader
Glassconveyor
Goodsubstrate
Non-conformingsubstrate Repair device D Inspection device C
Repair device BInspection device A
Fig. 22 Simulation model of single substrate transfer line
英文.indb 93英文.indb 93 09.12.3 11:39:18 AM09.12.3 11:39:18 AM
94
Vo l . 4 2 N o . 2 2 0 0 9
Figure 24 shows an example of avoid ing congestion. When four transfer instructions were simultaneously generated from the starting point to the destination in the figure, the upper route was assigned to two transfer instructions (with a different route selected for part of the way), and the lower route was assigned to the other two transfer instructions (with the same route).
(2) Avoiding suspended devicesFigure 25 shows an example of avoid ing
suspended devices. Route searching was attempted with several transfer devices suspended due to
trouble or for maintenance on transfer routes from the starting point to the destination. Even when there were suspended transfer devices in the factory, the algorithm selected a route avoiding the suspended devices.
3.2.1.2 Algorithm capacity verification resultsWe verified differences in transfer time between transfers executed in a route searched by the optimum transfer route algorithm and in a route registered in advance (conventional transfer method). In this test, we compared the transfer time when four transfer instructions were simultaneously generated from a starting point to a
VerificationMES
STK-C
Web server
MainMCS server
Transfer terminal
DB servers (clustering configuration)
Stockercontroller
[Imaginary system]Realized with the functions of machines in a factory andof machine control panels that directly control the machinesin the simulation model
OHV-C
OHVcontroller
RGV-C
[Actual system]
Uses actual devices we developed for the main MCS, the equipment controllers, the DB servers, the web server, and the transfer terminal
Material control system
(1) Transfer instruction for the next process
[Imaginary factory]TFT array process model
Crane
Vehicle (OHV)
Process deviceStocker
Cassettein stocker
Lifter Traverser
Vehicle (RGV)
Transfer unit
(2) Transfer command for each transfer device
(3) Control command for each device
(4) Device status and operation results
Fig. 23 Process flow of system verification method
Starting point
(Note) : Crane : Transfer unit : Shifter : Lifter :Vehicle (RGV)
: Vehicle (OHV) : Port or stocker before device
Destination
LF2001
TE2021
TE1021
TR1001
TR1001
CR1023
CR1022
CR1021
CR1024
CR1025
CR1014
CR1013
CR1011
CR1012
PO1011
SF1011
SF1011
SF1013
SF1013
SF1015
SF1015
SF1016
SF1016
SF1014
SF1014
PO1012
PO1022
PO1023
PO1031
PO1041
PO1032
PO1034
PO1035
PO1044
PO1043
PO1042
PO1052
PO1051
PO1061
PO1062
PO1053
PO1054
PO1055
PO1063
PO1012
PO1013
PO1023
PO1025
PO1024
PO1024
SF1012
SF1012
TE1001
TE1002
TE1003
TE1004
TE1005
TE1006
TE1007
TE1008
TE1010
TE1011
TE1012
TE1013
RG1001
RG1002
RG1003
RG1004
RG1005
RG1006
RG1007
RG1008
RG1010
RG1011
RG1012
RG1013
TE1022
TE1023
TE1024
TE1025
TE1027
TE1028
TE1029
TE1030
LE2003
LE1003
LE1004
LE2004
TE2031
TE1031
TE1032
TE1033
TE1035
TE1034
TE1036
TE2036
TF1026
TE2026
LF1002
LF2002
CR1031
CR1032
CR1033
CR1034
CR1043
CR1042
CR1041
CR1051
CR1061
CR1062
CR1063
CR1064
CR1052
CR1053
CR1054
CR1044
CR1035
OH1022
OH1023
OH1024
OH1025
OH1029
OH1030
OH1029
OH1030
OH1998
OH1999
OH1996
OH1997
OH1032
OH1033
OH1034
OH1035
LF1001
Fig. 24 Example of search result of optimum transfer route to avoid congestion
英文.indb 94英文.indb 94 09.12.3 11:39:19 AM09.12.3 11:39:19 AM
95
Vo l . 4 2 N o . 2 2 0 0 9
destination by using the following three patterns:(1) Transfer on the first floor(2) Transfer on the second floor(3) Transfer from the first floor to the second floor
Table 2 shows typical examples of verification results. In verification pattern (2), the conventional transfer method selected the same route to execute all four of the transfer instructions, while the optimum transfer route algorithm divided the transfer instructions into two t ransfer routes, reducing the t ransfer t ime to approximately half. In verification pattern (3), there was no difference in transfer route and transfer time because the transfer route was short and there were no alternatives. The verification results have indicated the superiority of the optimum transfer route algorithm.3.2.2 SMCS verification resultsWe compared simply processing transfers in a congested area in order of ar rival and processing them using the optimum dispatch algorithm. The following three patterns were used for the comparison. Table 3 shows the verification results.
(1) When glass substrates flow normally(2) When a process device is suspended for a certain
period of time(3) When a lift conveyor is suspended for a certain
period of timeProcessing by using the optimum dispatch algorithm
improved the throughput, which is the time from carrying a glass substrate into the process and to carrying it out, by approx. 200 to 300 s (3.5 to 4%) per cassette. This corresponds to saving of about one hour a day in a single substrate transfer process that handles 12 to 15 cassettes a day (on the assumption that the production capacity of the entire factory is 30 000 substrates a month). These results have indicated the superiority of the optimum dispatch algorithm.
4. Conclusion
We have completed development of the MCS as it applies to TFT array processes on liquid crystal manufacturing lines and to single substrate transfer processes expected to be introduced in the future. We have also established an off line system verification technique. The software architecture of this system can be applied to other fields, and we will expand applications to semiconductor lines, general distribution factory automation (FA) lines, and many other lines.
Starting point
(Note) : Crane
: Transfer device suspended due to abnormality or for maintenance
: Transfer unit : Shifter : Lifter : Vehicle (RGV)
: Vehicle (OHV) :Port or stocker before device
Destination
LF2001
TE2021
TE1021
CR1022
CR1021
CR1024
CR1025
CR1014
CR1013
CR1011
CR1012
PO1011
SF1011
SF1011
SF1013
SF1013
SF1015
SF1015
SF1016
SF1016
SF1014
SF1014
PO1012
PO1022
PO1031
PO1041
PO1032
PO1034
PO1035
PO1044
PO1043
PO1042
PO1052
PO1051
PO1061
PO1062
PO1053
PO1054
PO1055
PO1063
PO1012
PO1013
PO1025
PO1024
PO1024
SF1012
SF1012
TE1001
TE1002
TE1003
TE1004
TE1005
TE1006
TE1007
TE1008
TE1010
TE1011
TE1012
TE1013
RG1001
RG1002
RG1003
RG1004
RG1005
RG1006
RG1007
RG1008
RG1010
RG1011
RG1012
RG1013
TE1022
TE1023
TE1024
TE1025
TE1027
TE1028
TE1029
TE1030
LE2003
LE1003
LE1004
LE2004
TE2031
TE1031
TE1032
TE1033
TE1035
TE1034
TE1036
TE2036
TF1026
TE2026
LF1002
LF2002
CR1031
CR1032
CR1034
CR1043
CR1042
CR1041
CR1051
CR1061
CR1062
CR1063
CR1064
CR1052
CR1053
CR1054
CR1044
CR1035
OH1022
OH1023
OH1024
OH1025
OH1029
OH1030
OH1029
OH1030
OH1998
OH1999
OH1996
OH1997
OH1032
OH1033
OH1034
OH1035
LF1001
Fig. 25 Example of search result of optimum transfer route to count equipment status
Table 3 Results of optimum dispatch algorithm
Verification pattern
Throughput (s)Improvement rate
(%)ResultOrder of
arrivalOptimized
transfer
(1) 6 645 6 413 3.5 ○
(2) 6 606 6 377 3.5 ○
(3) 7 113 6 822 4.1 ○
(Note) ○ : Good
Table 2 Results of optimum transfer route algorithm
Verification pattern
Transfer time (h:m:s)Improvement rate
(%)ResultConventional
transferOptimized
transfer
(1) 0:59:11 0:57:37 2.6 ○
(2) 1:43:06 0:55:50 45.8 ◎
(3) 0:33:39 0:33:39 0.0 △
(Note) ◎ : Excellent○ : Good△ : Equivalent
英文.indb 95英文.indb 95 09.12.3 11:39:19 AM09.12.3 11:39:19 AM
96
Vo l . 4 2 N o . 2 2 0 0 9
— Acknowledgements —
We would like to express our sincere appreciation to persons concerned in the Logistics Systems Operations for advising us and cooperating in development of this system, and to IHI Scube Co., Ltd. for cooperating in the software development.
REFERENCES
(1) Y. Nishimura : A separate volume of Electronic Journal 2006, LCD Fab /Equipment /Facilit ies, Electronic Journal corporation (2005.9) pp. 50 - 59
英文.indb 96英文.indb 96 09.12.3 11:39:20 AM09.12.3 11:39:20 AM