Вы находитесь на странице: 1из 21

II/ALE 2012

An Operation Study for a Semiconductor Equipment Production


Using Discrete Simulation

Group No : C-1
Adviser : Wei-Jung, PhD, Professor
Mentor : Joanne Chiou
Rimson Encanto Junio Mapua-5
-- SEU-3
Sean Wang--(10074040) -- CYCU-22
Zi-Yi Li--(10074033) -- CYCU-23
Roy Lai--(10074023) -- CYCU-24

CHUNG YUAN CHRISTIAN UNIVERSITY

Table of contents
Introduction......................................................................3
Motivation and Purpose...................................................4
Steps of the project...........................................................5
Methodology....................................................................6
Conclusion.....................................................................15
List of figures
Fig 1 Steps of the Project.................................................5
Fig 2 Part 1 Processing Station........................................6
Fig 3 Part 2,3, and 4 Processing Stations.........................7
Fig 4 WIP1 Processing Station........................................8
Fig 5 Decision Area.........................................................9
Fig 6 Disassembly Line.................................................10

Introduction
Kelton, et al. describes system as a facility or process which could be actual or
planned. A system can be said to contain a set of well-defined methods or procedures

to carry out certain tasks.


Most of the things in this world can be considered as systems, some maybe
simple or some maybe complex. Examples of these are manufacturing plants, banks,
computer networks, and restaurants, each designed to perform various tasks.
Systems are often studied so as to measure how they perform or to improve their
performance if they do not satisfy certain requirements. One may also need to design
new systems that can provide better efficiency and satisfaction.
To study these systems, simulation and models are used to mimic their
characteristics. With the advent of technology, modeling these systems is now an easy
task as what is done in this study.
In this study, Arena simulation software, an advanced simulation software for
modeling real-world systems especially in the industry, is used to model a
semiconductor equipment production. Through this software, the key characteristics,
such as the assembly and the rework processes, of the semiconductor equipment
manufacturing process are imitated and the production is simulated with consideration
of various parameters for data analyses.

Motivation and Purpose


Semiconductor industry is highly affected by globalization, and the
customers of semiconductor equipment products have become diverse as they require
products with different specifications. Therefore, customization of equipment

products had more and more variations, and sales forecasting had produced less
accurate results.
This project is intended to use discrete event simulation software to develop a
production model of a machine in a semiconductor equipment factory, and to study
the effect of input variable variations on some output indices. In order to solve
inaccurate forecasting or excess demand problems, the cost model based on the
customer order decoupling point (CODP) with simulation software was developed to
minimize production cost and fulfill customer satisfaction. This model would
determine the customer order decoupling point, and decide whether to disassemble
similar products to meet the customer requirements.

Steps of project

Fig 1 Steps of the Project

Methodology
Due to the impact of the global market and changes of customer needs,

enterprises are facing a variety of challenges. Under the influence of the short product
life cycle and diverse small amount of fluctuations in demand, a wide range of
production management strategy were adopted by enterprise in order to solve the
different production management problem.
Kingsman (1996) mentioned that based on the response time of customer order,
factory production patterns can be divided into Make-to-Stock (MTS) and Make-toOrder (MTO). MTS means that the products required by customers have been
completed and can be immediately provided to the customer when the customer
orders arrive while MTO means that the production is based on the order of the
customer so it is customized and flexible. In other words, MTS stresses the
importance of immediate delivery, excellent quality, reasonable price and standard
product while the MTO emphasizes the significance of providing technical
capabilities to a variety of customized products.
Jan Olhager (2003) defines Customer Order Decoupling Point (CODP) as the
key disagreement between the Make-to-Stock and the Make-to-Order or cut-off point
of the forecast stock production and actual customer orders in mass customized
production mode. In the general production process, the upstream flows of CODP are
usually based on forecasts of production, while the downstream flows are actual
demand-oriented and conduct-differentiated operations in order to reduce the
prediction error cost of inventory or shortage and to reduce the customer delivery lead
time so as to improve the level of customer satisfaction.

Comparison between Customer Order Decoupling Point


(CODP) strategies

Fig the manufacturing strategy for CODP


When the production mode is make-to-stock (MTS), the CODP is at the end of
production process as seen in the previous figure. In this mode, products are produced
ahead of time and the amount of production is usually based on demand forecasts or
demand history. This strategy yields the least waiting time to receive the product but
may provide no customization.
Assemble-to-order (ATO) is yet another production strategy wherein the basic
parts that constitute the product are manufactured ahead of time but are not yet
assembled to form the final product. The assembly of the parts is done once an order
has been received. In this strategy, waiting time is more compared to the previous
strategy but provides customization.
In make-to-order (MTO) production strategy, on the other hand, the end
product, as well as its key items to make it, is manufactured only once an order has
been made. An advantage of this production strategy is customization to satisfy the
requirements imposed but may incur additional waiting time to receive the product
because of the nature of this strategy.
Finally, in engineer-to-order (ETO) strategy, the customer is heavily involved
in the design and the production process for customization. This strategy is
characterized by longer lead times but may provide the best satisfaction as

customization is extremely employed.


As what can be seen in the previous figures, the location of the CODP changes
depending on the market strategy. These changes may provide advantages and
disadvantages as tradeoff is noticeable between customization and waiting time as
what is explained in the succeeding section.
Characteristics of CODP
Moving the CODP towards the customer or the company provides certain
advantages and disadvantages as summarized in the tables below. Customization, lead
time, economy, and efficiency are some of the factors considered in the two tables.
Advantages and Disadvantages of CODP Moving Towards the Customer
Advantages

Reasons for CODP to move

Disadvantages

closer towards the customer


1.Delivery
efficiency

2. Delivery
reliability

1. Decrease in the delivery lead 1. Needs more accuracy


time

2. Less product customization

2.Increase in the economy and 3. More inventory and Workefficiency for manufacturing

in-process

3.Product price

Advantages and Disadvantages of CODP Moving Towards the Company


Advantages

Reasons for CODP to move

Disadvantages

closer towards the company


1.Product variety

1.Better product customization

1.Needs more lead time

2.Product

2. Lesser need for forecasting

2. Decrease in the economy

flexibility

3.Decrease in the risk of and

3.Product quality

inventory and Work-in-process

efficiency

manufacturing

for

Part Processing
The Part Processing Area has four processing stations namely Part 1,2,3, and 4
stations, and as the name implies, produces the various parts to be assembled to form
the end product.

Fig 2 Part 1 Processing Station

Part 1 processing station consists of three Create blocks that serve as the blocks
for inputting materials. One Create block is for the materials to make the forecasted
orders, another create block is used for inputting materials to make up for the
machines that are disassembled to meet the number of forecast order machines
required and the third Create block is for input materials to make the emergency
ordered machines if the number of disassembled machines is less than the number of
required rush ordered machines.

Fig 3 Part 2/3/4 Processing Station


In Parts 2, 3, and 4 processing stations, the general structure is similar to part 1
processing station with minor modifications. In each station there is a station block to
capture the disassembled parts from the Disassembly Area and a decision block to sort
out those parts which come from the Disassembly Area and those that are not. Those
parts that come from the Disassembly Area go to the rework line while those that are

not go to the original process line. The Part 2 process line waits for the WIP 1
Processing Station to finish its job and then the output of WIP 1 Processing Station
will be batched together with part 2 and then will be assembled in the WIP 2
Processing Station. The same is happening in the other part processing stations until
the end product is produced at the WIP4 Processing Station.

Fig 4 WIP1 processing station

WIP 1 Station receives part 1 and sorts out whether it comes from a
disassembled machine or not and directs the part into one of two different processing
lines depending on the nature of the part.
The output of this processing station then is transferred to the next Part
Processing Station for batching and sent to the WIP 2 station for assembly as
described in the previous paragraphs. The same happens to the other WIP Processing
Stations until the final product is form at WIP4 Station.

Decision Area

Fig 5 Decision Area

Figure 5 shows the portion where decisions are made regarding the production of
the preordered machines and the production of rush-ordered machines which is the
main concern of this study.
As the machines from WIP4 Processing Station enter this module division, each
enters the first decision block to categorize whether the machine that passes through is
a reworked machine (for the rushed order)that goes to the emergency order disposal
area, or not. The purpose of this decision block is to prevent reassembled machines to
enter the disassembly line for the second time.
Machines that enter the second decision block could be a mixture of machines
that are preordered machines or machines that are made for the rushed orders, i.e.,
machines which are built with new input materials. The machines that are made from
new input materials are then segregated and go directly to the rushed order disposal
area while those machines that are preordered are ready to go to the third decision

block.
In the third decision block, the machines are sorted to two types: those that are
needed to be disassembled and reworked to satisfy the newly ordered machine
products or those that just go directly to the disposal area for the preordered machines.
Those that will need rework will then go to the disassemble station.
One can see that in this module division there are a lot other blocks other than
decision blocks but they are merely used as records for various parameters in the
module such as delay times necessary for calculating a variety of costs that may have
incurred due to disassembly and reassembly process.
DISASSEMBLY LINE

Fig 6 Disassembly Line

Machines that require rework to satisfy the emergency orders enter the

disassembly division. In this division, the machine will be separated into part 4 and
WIP3 which is done by the separate block. After they are separated, the first decision
block sorts and segregates the two. The part4 object will then go to the part 4 station
and the WIP3 will then enter other blocks for further disassembly.
In the same manner, the WIP 3 will be separated into part 3 and WIP2 and again
the decision block will be used for sorting the two objects. After which, part 3 will go
to the part 3 station and WIP 2 will enter other blocks to be disassembled further.
The same thing happens all throughout this module division. To cut it short, the
original machines are separated to part 1, part 2, part3, and part 4 and will just enter
the Part Processing zones for the rework process.
In this module division, assign and record blocks are used as well to record
processing times and other important parameters necessary for statistical analysis of
the data that will be gathered.
Definition of Costs
(1) Total cost=Cost of Manufacturing + Work-in-process (WIP) Cost +
Disassembly cost + Delay cost
(2) Cost of Manufacturing= Labor cost time for units number of machines
(3) WIP Cost= WIP Cost of Part + WIP Cost of product
WIP Cost of component =cost of component number of component
holding time of component(hr)

WIP Cost of product =cost of product number of product holding time


of product(hr)
(4) Disassemble cost=cost of disassembly per laborer (150 dollars /hr.) time of
disassembly(hr.)
(5) Delay cost = delay cost of forecasting order+ delay cost of emergency order
delay cost of forecasting order = number of forecasting order delay cost

of forecasting order per hour(100 dollars /hr) Max[(delivery time for


forecasting order- time of forecasting order (250hr)-lead time for
forecasting order),0]
delay cost of emergency order =number of emergency order delay cost

of emergency order per hour(50dollars/hr) Max[(delivery time for


emergency order - time of emergency order (180hr)- lead time for
emergency order),0]

Costs of Various Parameters


Name of parameter

Cost per
unit(dollars/hr)

WIP Cost of
component

WIP Cost of
product

Station 1

Station 2

Station 3

Station 4

Station 1

Station 2

Station 3

10

Station 4
cost of disassembly of a

15
150

machine
cost of manufacturing a

150

machine
delay cost for customer A

100

(forecast order)
delay cost for customer B

50

(emergency order)

The assumption and limitation of the research


This research seeks to find the best CODP in terms of cost. There are many
uncontrollable things or unforeseen events that may happen when the production
process is in operation such as when a machine fails, materials delay, or a worker fails
to arrive which are beyond the scope of this study. The research also does not take
into account the varying operation times between workers which, of course, will
affect the resulting cost of the CODP.
operation process and limitation of worker
The product will delivery after n operation.
The time of manufactory and disassemble for every station is constant
but there are not the same.
The manufactory number is not infinite.
Does not take into account the time spent in moving components from
station to station5
customer and order
All order for product is similar.
The number and kind of product for forecasting order match the fact.
the number of each order is constant.
the number of forecasting order is more than emergency order

the lead time for emergency order is smaller then time of manufactory.
the company cannot forecast order time accurate
assemble of component and product
No product on the process line before working.
Put new material for forecasting order when assemble the product for
forecasting order
Assemble the product after assembling the component

The assemble time of component and product


Operation name

Operation station

Assemble time
(hr)

Assemble component

Assemble product

Disassemble product

Rework product

Station1

Station 2

Station 3

Station 4

Station1

Station 2

Station 3

Station 4

Station1

Station 2

Station 3

Station 4

Station1

Station 2

Station 3

Station 4

Therefore, by the CODP the concept, the use of simulation software Arena tools, to
build a model to known the optimal number of machines. This research just discusses
the CODP which disassemble from station 4 to station 1 for ARENA model.
Therefore, the model is built, the emergency order arrival time is set to 130 hours, 180
hours and 230 hours , the number of demolition machine is set to 0~10, repeat the
simulation 33 times, the output of the report data into Minitab for ANOVA analysis.
Then by analyzing emergency orders the time and the number of demolition machine
to understand for any impact on the total cost. After using Excel to draw a chart of the

cost and the number of demolition machine, infer the optimal number of machines.

Analyze

We can see the No and times p-value is 0.000, lower than 0.05. So we can say
the No and time have a significant effect to total cost. The R-Sq is mean the
correctness of data, higher value mean high correctness. And we can see that R-Sq is
99.82%, it was very high value mean the data is more correctness.

Residual Plots for Value


Normal Probability Plot of the Residuals

Residuals Versus the Fitted Values

99

6000
3000
Residual

Percent

90
50
10
1
-8000

-6000
-4000

0
Residual

4000

8000

200000

Histogram of the Residuals

250000

300000 350000
Fitted Value

400000

Residuals Versus the Order of the Data

6000

3000
Residual

Frequency

0
-3000

4
2

0
-3000
-6000

-6000

-3000
0
Residual

3000

6000

10
15
20
25
Observation Order

30

This picture is talking about the appropriateness of the statistical hypothesis.


Figure Normal Probability Plot and Histogram of residual is show that residual
distributed on normal, only the value in red circle was the outliner. The second figure
next to Figure Normal Probability Plot is Residuals Versus the Fitted Values. This

figure result was random distribution on 0 (central line). In fourth figure is Residual
versus order, there should be random and didnt have any mode.

Table The input data for ANOVA


time

No

Value

130

233560

130

233721

130

235310

130

242411

130

253468

130

268168

130

287793

130

312350

130

343170

130

382510

130

10

422751

180

216160

180

216498

180

220775

180

227278

180

238792

180

254481

180

278661

180

305320

180

335794

180

371017

180

10

412360

230

216160

230

220198

230

227084

230

237778

230

252104

230

269331

230

290593

230

314948

230

343114

230

377407

230

10

418215

Conclusion
The follow tables are the data about simulation resulted with two variable, Rush
order arrival time (ROAT) and Quantity of disassembly(QOD).
Table Resulted of ROAT = 130 hr simulation
Rush order arrival time = 130 hr

Disassembly No.

Total
Cost

Cost of
Manufactured

delay
cost

Disassembly
cost

Part
Cost

WIP
Cost

216160

130500

10500

22860

52300

216498

131700

8750

1650

23283

51115

220775

132900

7000

6525

23759

50591

227278

134100

6200

14625

24246

48107

238792

135300

6150

25950

24758

46634

254481

136500

7600

40500

25410

44471

278661

137700

13600

58275

26376

42710

305320

138900

19050

79275

27216

40879

335794

140100

24950

103500

28130

39114

371017

141300

31950

130950

29304

37513

10

412360

142500

41250

161625

30920

36065

Part
Cost

WIP
Cost

Table Resulted of ROAT = 180 hr simulation


Rush order arrival time = 180 hr
Total
Cost

Cost of
Manufactured

delay
cost

Disassembly
cost

Disassembly No.

216160

130500

10500

22860

52300

216498

131700

8750

1650

23283

51115

220775

132900

7000

6525

23759

50591

227278

134100

6200

14625

24246

48107

238792

135300

6150

25950

24758

46634

254481

136500

7600

40500

25410

44471

278661

137700

13600

58275

26376

42710

305320

138900

19050

79275

27216

40879

335794

140100

24950

103500

28130

39114

371017

141300

31950

130950

29304

37513

10

412360

142500

41250

161625

30920

36065

Table Resulted of ROAT = 230 hr simulation


Rush order arrival time = 230 hr

Disassembly No.

Total
Cost

Cost of
Manufactured

delay
cost

Disassembly
cost

Part
Cost

WIP
Cost

216160

130500

10500

22860

52300

220198

131700

12450

1650

23283

51115

227084

132900

14200

6525

23759

49700

237778

134100

16700

14625

24246

48107

252104

135300

19750

25950

24758

46346

269331

136500

22600

40500

25410

44321

290593

137700

26000

58275

26376

42242

314948

138900

29650

79275

27216

39907

343114

140100

33950

103500

28130

37434

377407

141300

40950

130950

29304

34903

10

418215

142500

50750

161625

30920

32420

Вам также может понравиться