Chapter 7: Computer Integrated Manufacturing Systems
7.1 Introduction
(1)In the previous chapters, we have studied various aspects of manufacturing, including
-Design: engineering design, design for manufacturing, design for quality and etc.
-Planning: process planning (feature recognition, optimization), and systems planning (scheduling, bill of material, and Material Requirement Planning (MRP))
-Machine tools and control: CNC, PLC, bar code, smart card, and etc.
-Material handling: conveyor, AGV, warehouse and etc.
-Quality control: Taguchi’s quality loss, failure mode and effect analysis (FMEA), statistic quality control (SPC) and etc.
(2)In this (last) chapter, we will study the manufacturing from a system point of view.
Fig. 7.1: The correlation of the chapter to other chapters
(3)This chapter covers the following materials:
-Just-In-Time (JIT) manufacturing (Chapter 11 of the textbook),
-Flexible Manufacturing Systems (FMS) (Chapter 12 of the textbook),
-Computer Integrated Manufacturing (CIM) (Chapter 13 of the textbook), and
-Enterprise integration (Chapter 14 of the textbook).
7.2. Just-In-Time Manufacturing (JIT)
(1)What is JIT?
-In Chapter 3, we studied the Material Requirement Planning (MRP) system, which develops a bill of material and a schedule, and releases the order to the shop. Accordingly, the manufacturing is initiated.
-As shown in Figure 7.2, such a system is a push system, in which the materials (solid line) and information (dash line) flow in the same direction
Fig. 7.2: Illustration of a push system
-The push system is a open-loop system that may create following problems:
-It may lead to starvation or excessive stocks simultaneously at different stages because of imbalances of stocks between the stages.
-It may lead to conditions of having excessive equipment and surplus of workers.
-In Japan, the problem is called the 3 Ms problems:
-Muda (waste): waste for correction, waster of overproduction, waster for processing, waste for inventory, waste of motion, waste for waiting, …
-Mura (unevenness): uneven workload, uneven schedule, …
-Muri (overburden): overburden machine, overburden process, …
-In order to solve these problems, Toyota developed a JIT manufacturing approach, which is a pull system as shown below.
Fig. 7.3: Illustration of pull system
Note that the information flow serves as a feedback loop that minimizes the 3 Ms problem.
-This approach is called the JIT manufacturing approach.
(2)Kanbans - the basic idea of JIT
-In a JIT system, the information feedback is achieved by Kanbans. Kanban is a Japanese word meaning visible records. But, it has been given a boarder meaning including records, orders and the plans of manufacturing.
-More specifically, kanbans are record cards that travel between preceding and subsequent processes, communicating what parts are needed in the subsequent processes.
-Through the Kanbans, the production plan is initiated backwards from the finished product (pull system).
-There are various types of Kanbans, and the following are the most important:
-Withdraw Kanbans, which is used to pass the authorization for movement of parts from one process to another
-Production Kanbans, which is used to release an order to the preceding process to build parts equal to the lot size specified
A typical example is shown below.
Fig. 7.4: Illustration of withdraw kanban and production kanban
-The rules for operating Kanbans
-Rule 1: no withdrawal of parts without a kanban
-Rule 2: the subsequent process comes to withdraw only what is needed
-Rule 3: do not send the defectives to the subsequent process
-Rule 4: the preceding process should produce only the exact quantity of parts withdrawn by the subsequent process
-Rule 5: smoothing of production
-Rule 6: fine tune of production using kanban
- Following figure illustrates how kanbands are worked in a JIT manufacturing system.
Fig. 7.5: Illustration of the paths of Kanbans
(3)Kanban planning and control
-Kanban is the heart of the JIT system. It is very important to determine the number of kanbans needed and hence, determine the JIT system structure.
-The deterministic model for determining the number of Kanbans by Toyota Motor Inc.
-Let y be the number of Kanbans needed,
-The model:
where, D = demand per unit time,
Tw = waiting time of kanban
Tp = processing time
a = container capacity (not more than 10 percent of daily requirement)
= a policy variable and (1 + ) represents the safety factor
-The objective is to reduce the values of a, and lead time (Tw + Tp) continuously.
-Based on this simple model, we can determine:
(a)how to determine the number of kanbans
(b)impact of lead time on the number of kanbans and work-in-process inventory
(c)interactions between the withdrawal and production kanbans.
-An example (Example 11.1 in the textbook):
-XYZ company produces n = 10,000 units per month, the shipping container capacity is 50 items, the production lead time is 0.5 days, and the policy variable is set at = 0.4 (the smaller the , the smaller the safety factor)
-The number of production kanbans
-Daily demand:
D = 10,000 / 20 (working days per month) = 500 parts
-The number of kanbans:
y = (500)(0.5)(1.4) / (50) = 7
-The average inventory is:
(7)(50) = 350 units
This implies that we can cut down the inventory by cut down the safety margin
-Suppose = 0, which implies that a withdrawal kanban must always be delivered on time, whenever parts are needed; then:
y = (500)(0.5)(1) / (50) = 5
at this time, the average inventory is:
(5)(50) = 250 units
This implies that we can cut down the inventory by cut down the manufacturing lead time.
-Suppose the production lead time is changed to 1 day, then the number of kanbans needed is:
-The number of kanbans:
y = (500)(1)(1.4) / (50) = 14
at this time, the average inventory is:
(14)(50) = 700 units
-Example 2 (Example 11.2 in the textbook)
-ABC company produces product Z, which is assembled from two parts, X and Y, also manufactured in the company
-The factory layout is shown in Figure 7.6
Fig. 7.6: factory layout of the ABC company
where,PK-X = production kanban for part X
PK-Y = production kanban for part Y
WK-X = withdrawal kanban for part X
WK-Y = withdrawal kanban for part Y
SA-X = staging area for part X
SA-Y = staging area for part Y
-The processing data are given in the table below.
Table 1: An example of production organization using kanban
Part
/ Demand(units / day) / Lead time (days) / / Container capacity
Assembly stage
X / 2000 / 1.0 / 0.00 / 100Y / 800 / 0.5 / 0.25 / 50
Manufacturing stage
X / 2000 / 0.5 / 0.20 / 100Y / 800 / 1.0 / 0.00 / 50
-determining the withdrawal kanbans (the assembly process pulls from the manufacturing process):
yX = (2000)(1.0)(1) / 100 = 20
yY = (800)(0.5)(1.25) / 50 = 10
-determining the production kanbans (the manufacturing process pulls from the supplier)
yX = (2000)(0.5)(1.2) / 100 = 12
yY = (800)(1)(1) / 50 = 16
-now, let us assume that the assembly process is shipped to mainland. As a result, the lead time increases to 4 days. The withdrawal kanbans will be:
yX = (2000)(4)(1) / 100 = 80
yY = (800)(4)(1.25) / 50 = 80
but the production kanbans remain the same. This implies that in order to keep the same supply, we must increase the process capability a, which in turn, will increase the total production cause.
(4)A probabilistic cost model for determining optimal number of kanbans
-In practice, it is inevitable that the manufacturing systems are affected by various “random” disturbances, such as machine breakdown, supply shortage, labor absent, and etc. Hence, it is more realistic to develop a probabilistic kanban model for JIT systems.
-Denote:
-n = number of kanbans
-p(x) = probability mass function for the number of kanbans required
-ch = holding cost per container per unit time at a work center
-cs = cost of shortage per container per unit time at a work center
-There are two possibilities:
-The actual requirement for the kanbans, x, is less than n, and the expected cost is:
Expected holding cost =
-The actual requirement for the kanbans, x is more than n, and the expected cost is:
Expected shortage cost =
-Hence, the total expected cost, TC(n), is:
-It can be shown (in the textbook) that the optimal value n can be obtained from the following equation:
where,
-An example (Example 11.3 in the textbook)
-Suppose the probability mass function of the number of kanbans is known as follows.
No. of kanbans012345
Probability00.20.30.350.10.05
-the holding cost and shortage cost per container per unit time are $50 and $200 respectively
-Solution:
-The cost ratio: (cs) / (cs + ch) = (200) / (200 + 50) = 0.8
-The probability function:
P(0) = 0, P(1) = 0.2, P(2) = 0.5, P(3) = 0.85, P(4) = 0.95, P(5) = 1
- hence, the optimal number of kanban is n = 3.
(5)Signal Kanban
-In the previous kanban models, it is assumed that the setup time is short and hence, can be neglected. There are however manufacturing processes, such as forging, die-casting, and stamping, in which the setup time is not short. In these cases, a new type of kanban, signal kanban, must be used.
-Figure 7.5 illustrates the process of signal kanbans.
-In general, signal kanbans can be further divided into two types:
-raw material ordering kanban: used to withdraw material from the preceding stage (represented by a rectangular in Figure 7.7).
-production ordering kanban: used to trigger the production of a lot at the work center (represented by a triangular in Figure 7.7).
-Signal kanbans are different from standard kanbans. A comparison between standard kanban and signal kanban is as follows:
-In the standard kanban process, a production kanban is sent back to trigger production after every withdrawal of a container
-In the signal kanban process, the production kanban is not sent back. Instead, a production ordering signal kanban is used to minimize the setup time.
-There are two important aspects of a signal kanban system:
-The determination of lot size
-The position of both production-ordering as well as material-ordering signal kanbans
The calculations in determining these parameters are the same as the withdraw kanband and production kanban.
Fig. 7.7: Illustration of flow of signal kanban
-An example
-A stamping plant runs two shifts: 2 x 8 = 16 hrs. / day
-The press utilization is 80%: 16 x 0.8 = 12.8 hrs. / day
-The rest of the time (20%) is used for setup (die changes): 16 x 0.2 = 3.2 hrs. / day
-Average setup (die change) time is 32 min.: 32 / 60 hrs.
-The maximum number of possible setup is 2
-The demand for the parts, D: 1400 / day
-The safety factor: = 0.2
-The minimal lot size per setup is:
Minimum lot size per setup = (Demand)(safety) / (time of setup)
or
Minimum lot size per setup = (1400)(1.2) / (2) = 840.
-Suppose the size of the container is 100 parts, then the number of containers needed is:
840 / 100 9
-The position of the production-order signal kanban is determined by the kanban cycle time. The kanban cycle time, Tc, consists of several elements such as waiting time, transfer time, and lot processing time. The formula to calculate the position of the production-order signal kanban is as follows:
Production signal kanban position
-Suppose the kanban cycle time is 3 hrs.: 3 / 16 days
-The production signal kanban is:
Production signal kanban position = (1400)(1.2)(3/16) / (100) = 3.15 4 containers.
-Similarly, we can calculate the material-ordering signal kanban.
(6)Other related issues
-The concept of kanban has been extended greatly to carry out various “feedback” control functions in the manufacturing systems. They all use kanbans, such as
-Express kanban
-Emergency kanban
-…
-It is interesting to know that in additional to push and pull, there are other types of manufacturing models such as:
-Periodic pull
-Constant work-in-process
-Long pull
-…
-Finally, there has been effort to combine JIT in purchasing.
7.4 Group Technology
(1)What is Group Technology (GT)
-GT is a philosophy that recognizing and exploiting similarities in three different ways
-By performing like activities together
-By standardizing similar tasks
-By efficiently storing and retrieving information about recurring problems
-GT has a number of advantages, such as
-A family of products can be manufactured with minimum changes
-A complicated product can be decomposed into a number of standard parts (or systems) and manufactured with minimum additional facilities
-GT would result in significant cost saving. It is a basis of cellular manufacturing, which will be discussed in the subsequent section.
(2)Part features: design features and manufacturing features
-In Chapter 3 (process planning), we have discussed the fact that parts have design features (e.g., a hole), and manufacturing features (e.g., drilling). GT is based on recognizing and using these features.
-A number of GT approaches have been developed to decompose a large manufacturing activity into smaller, manageable system based on similarities of design features and manufacturing features.
-GT approaches can be divided into two categories
-Visual inspection method
-Coding method
-Visual inspection method is relatively simple. An experienced engineer can examine a part and determine its basic design features (e.g., a hole) and manufacturing features (e.g., drilling). After all, this was the only approach used in the past, and is still effectively used.
- In this section, however, we will focus on the coding method.
(3)Coding methods
-Coding refers to the process of assigning symbols to the parts. The symbols represent the features (both the design features and manufacturing features) of the parts.
-Coding is for classification – the process of categorization of a set of parts into part families.
-Various coding systems have been developed and these systems can be grouped into three types:
-Monocode or hierarchical code: the structure of these codes is like a tree in which each symbol amplifies the information provided in the previous digit.
-Polycode: also known as chain code in which each digit is independent and describes a specific information
-Mixed-mode code: a combination of monocode and polycode.
-We will present a few commonly used codes below.
(4)Optiz classification system
-The Optiz classification system is developed at the Technical University of Aachen for German Machine Tool Association
-The Optiz system is one of the oldest systems and you can use it a base to develop your own.
-It is a mixed code system (and hence, rather representative)
-The system consists of the following sequence of digits:
123456789ABCD
Form codeSupplementary codeSecondary code
-The code structure and the definition are shown in Figure 12.3 (in the textbook).
-An example
Form code
/ 1 / 3 / 1 / 0 / 6Interpretation:
-First digit = 1: it is a rotation part with 0.5 < L / D < 3 (9.6 / 5 1.9)
-Second digit = 3: the external shape has a functional groove
-Third digit = 1: the internal shape has a through hole
-Fourth digit = 0: the plain surface does not exist
-Fifth digit = 6: there are spur gear teeth on the part.
-The part is shown in Figure 7.8
Fig. 7.8: A spur gear with Optiz code 13106
-Note that the code does not contain detailed engineering design information. Instead, it shows the design features. Based on the design features, the manufacturing plan can be determined.
-The benefit of GT:
-Help engineering design
-Help layout planning
-Help design / selection of equipment, tools, jigs and fixtures
-Help manufacturing process planning
-Help production control
-Help quality control
-Help purchasing
-Help customer services
(5)GT and Cellular Manufacturing
-Cellular manufacturing is an application of GT in manufacturing, in which manufacturing facilities are organized as cells. The parts are similar in their processing requirement, such as operations, tolerances, and machine tool capabilities are manufactured together.
-The objectives of cellular manufacturing are
-Reduce setup time
-Reduce flow time
-Reduce inventories and
-Reduce market response time
-Manufacturing cells are natural candidates for JIT implementation.
-Cell design is rather complicated in which the following issues must be addressed:
-Selection of part families and grouping of parts in families
-Selection of machines and processes and grouping them together
-Selection of tools, fixtures and pallets
-Selection of material-handling equipment
-Choice of equipment layout
-Detailed design of jobs
-Organization of supporting personnel
-Formulation of maintenance and inspection policies
-Design of operation procedures
-Modification of cost control
(6)Cell formation approach
-There are several cell formation approaches. The most commonly used one is the Machine-Component Group Analysis (MCGA) approach.
-The procedure of MCGA
Stage 1: machine classification
Stage 2: checking parts list and production route information
Stage 3: factory flow analysis
Stage 4: machine-component group analysis.
We will focus on Stages 3 and 4 using an example.
-Example: The machines and components information (the output of Stages 1 and 2) is as follows.
Table 2: An example of machine-part production information table
ComponentsMachines / 1 / 2 / 3 / 4 / 5 / 6 / 7 / 8 / 9 / 10
M1 / 1 / 1 / 1 / 1 / 1 / 1 / 1 / 1 / 1
M2 / 1 / 1 / 1 / 1 / 1
M3 / 1 / 1 / 1 / 1
M4 / 1 / 1 / 1 / 1 / 1 / 1
M5 / 1 / 1 / 1 / 1 / 1 / 1 / 1 / 1
where, an “1” implies the correlation between the machines and components.
-We use the Rank Order Clustering (ROC) algorithm for factory flow analysis and use Single-Linkage Cluster Analysis (SLCA) algorithm for machine-component group analysis.
-Step 1 (ROC): assign binary weight and calculate a decimal weight for each row and column using the formula below:
Decimal weight for row
Decimal weight for column
where, bip (and bpj) are binary weights. If the jth machine is used to process pth component, bip = 1; else bip = 0. For the data shown above, the resulting decimal equivalents are as follows.
Table 3: The ROC column weighting of the example above
Components
1 / 2 / 3 / 4 / 5 / 6 / 7 / 8 / 9 / 10Decimal / Binary weight
Machines / equivalent / 29 / 28 / 27 / 26 / 25 / 24 / 23 / 22 / 21 / 20
M1 / 1007 / 1 / 1 / 1 / 1 / 1 / 1 / 1 / 1 / 1
M2 / 451 / 1 / 1 / 1 / 1 / 1
M3 / 568 / 1 / 1 / 1 / 1
M4 / 455 / 1 / 1 / 1 / 1 / 1 / 1
M5 / 1020 / 1 / 1 / 1 / 1 / 1 / 1 / 1 / 1
Question: suppose the order of the components is changed (e.g., instead of 1, 2, 3, …, 10, we have 10, 9, 8, …, 1), will the result be the same? You are encouraged to think about it.
-Step 2 (RCO): re-arrange rows by sorting the decimal weights in decreasing order and then calculate the column decimal equivalent. For the data above, the result is the following matrix:
Table 4: The ROC row weighting of the example above
Components
1 / 2 / 3 / 4 / 5 / 6 / 7 / 8 / 9 / 10Binary / Binary weight
Machines / weight / 29 / 28 / 27 / 26 / 25 / 24 / 23 / 22 / 21 / 20
M5 / 24 / 1 / 1 / 1 / 1 / 1 / 1 / 1 / 1
M1 / 23 / 1 / 1 / 1 / 1 / 1 / 1 / 1 / 1 / 1
M3 / 22 / 1 / 1 / 1 / 1
M4 / 21 / 1 / 1 / 1 / 1 / 1 / 1
M2 / 20 / 1 / 1 / 1 / 1 / 1
Column decimal equivalent / 28 / 27 / 27 / 27 / 28 / 20 / 28 / 26 / 11 / 11
-Step 3 (RCO): re-arrange the columns by sorting the column decimal weights in decreasing order. For the data above, the resulting matrix is as follows:
Table 5: The resulting machine grouping in the above example
Components
1 / 5 / 7 / 2 / 3 / 4 / 8 / 6 / 9 / 10Binary / Binary weight
Machines / Weight / 29 / 28 / 27 / 26 / 25 / 24 / 23 / 22 / 21 / 20
M5 / 24 / 1 / 1 / 1 / 1 / 1 / 1 / 1 / 1
M1 / 23 / 1 / 1 / 1 / 1 / 1 / 1 / 1 / 1 / 1
M3 / 22 / 1 / 1 / 1 / 1
M4 / 21 / 1 / 1 / 1 / 1 / 1 / 1
M2 / 20 / 1 / 1 / 1 / 1 / 1
-Step 4 (RCO): repeat steps 2 and 3 until there are no changes in row and column positions. This completes the factory flow analysis. For the data above, there is no change in further iteration.