Transcript Developing An Expert System for GP Implementation (Powerpoint

DEVELOPING AN EXPERT
SYSTEM FOR GP
IMPLEMENTATION
RUBY PINEDA-HENSON
Department of Industrial Engineering
Holy Angel University-Angeles City, Philippines
[email protected]
ALVIN B. CULABA
Department of Mechanical Engieering
De La Salle University-Manila, Philippines
[email protected]
OUTLINE OF
PRESENTATION





INTRODUCTION
RATIONALE FOR GP MODEL
EXPERT SYSTEMS METHODOLOGY
GP MODEL DEVELOPMENT
APPLICATION TO GP ANALYSIS OF
SEMICONDUCTOR
ASSEMBLY/PACKAGING
 CONCLUSION/ RECOMMENDATION
GREEN PRODUCTIVITY
PARADIGM
FRAMEWORK
 FOR
IMPROVEMENT
CONTINUOUS
IMPROVEMENT
FOUNDATION
ENVIRONMENTAL  FOR
PRODUCTIVITY
PERFORMANCE
SUSTAINABLE
DEVELOPMENT
RATIONALE FOR GP
MODEL
 LIFE CYCLE ASSESSMENT - THE
TECHNICAL FRAMEWORK
 ANALYTIC HIERARCHY PROCESSMULTICRITERIA DECISION
MAKING(MCDM) MODEL AND TOOL
LIFE CYCLE ASSESSMENT
 Streamlined LCA
 Process-Based
 Phased Approach
Inventory Analysis
Impact Assessment
Improvement Assessment
INVENTORY ANALYSIS
INPUTS
UNIT

PROCESSES
OUTPUTS
IMPACT ANALYSIS

IMPACTS
IMPROVEMENT
Analysis

IMPROVEMENT TECHNIQUES 
GREEN PRODUCTIVITY INDICATORS
IMPACT 1
OPTION 1
EMISSIONS TO AIR
IMPACT 2
1
OPTION 2
Raw Materials
EMISSIONS TO WATER
IMPACT 3
GREEN PRODUCTIVITY
PERFORMANCE
2
OPTION 3
Energy
EMISSIONS TO LAND
IMPACT 4
3
Ancillary Materials
OPTION 4
OTHER
RELEASES
IMPACT 5
n
OPTION j
PRODUCTS/ COPRODUCTS
IMPACT i
ANALYTIC HIERARCHY
PROCESS

Pairwise Comparison
 Mechanism For Consistency Check
 A Panel Of Experts May Be Utilized
 Geometric Means Of Comparison
Ratings
EXPERT SYSTEMS
TECHNOLOGY
 The potential of expert system technology is
explored to develop a software that
emulates how human experts diagnose GP
performance of manufacturing processes.
 Expert systems (ES) are computer programs
that use expert knowledge and heuristics or
rules of thumb to solve problems in a
specific domain.
Complex decision analysis may involve an
intricate combination of facts and heuristic
knowledge which is organized into three
distinct components:
 Knowledge Base
 Working Memory
 Inference Engine
GP DIAGNOSTIC SOFTWARE
 Front-end database system (Visual FoxPro)
 Windows shell program/interface
 CLIPS (C Language Integrated Production
System) expert system
 The shell program embeds the ES. The
Dynamic Data Exchange (DDE) feature of
Windows operating environment is used to
transmit data to and from the two program
ends.
GP MODEL DEVELOPMENT




Sub-Models
Inventory Analysis
Impact Analysis
Improvement Analysis
Green Productivity Assessment
Knowledge Base
Inventory
Analysis
------------------------Input Data
Output Data
Environmental
Impact
Analysis
------------------------Classification
Valuation
Productivity
Improvement
Analysis
------------------------Classification
Valuation
Multicriteria Decision Analysis
(Analytic Hierarchy Process)
Input - Output Analysis
Figure 2.
Green Productivity ES Model Structure
Green Productivity
(GP) Assessment
------------------------GP Ratios
GP Indices
DIAGNOSTIC MODEL
FEATURES
 The inventory module prompts the user for
inventory data on the manufacturing
process.
 The diagnostic module, through an
embedded expert system program, performs
impact classification on the inventory data.
EXAMPLE
 In impact classification, pseudo-rules which are asserted
as facts in the knowledge base links an input or output
indicator substance found in the inventory to an impact
category or classification. For example:
IF [process input deionized water] and [deionized
water >0]
THEN [environmental impact water resource
depletion]
IF [process output mold runners] and [mold runners
>0]
THEN [environmental impact terrestrial ecotoxicity]
Table 1. Environmental Impact Factors for Semiconductor Assembly Packaging
Impact Factors
Water Resource Depletion (WRD)
Energy Resource Depletion (ERD)
Human Toxicity in Air (HTA)
Human Toxicity in Water (HTW)
Human Toxicity in Land (HTL)
Aquatic Ecotoxicity (ETA)
Terrestrial Ecotoxicity (ETT)
Indicator
Cooling Water, Deionized Water
Electricity, Thermal Energy
SO2, NO2, CO,VOCs, Arsine, Phosphine
Metallic (Pb) vapors
Sulfuric acid, hydrochloric acid, phosphoric acid,
nitric acid, acetic acid, methanesulfonic acid
Heavy metals: Pb, Al, Ni, Cd, Cr, As, Sn
Isopropanol,
acetone,
N-butyl
acetate,
trichloroethylene, xylene, petroleum distillates,
halocarbons, Methylene chloride
Plastics, epoxies, glues, flux, off-specification
products or rejects, molding compound, mold
runners, melamine, waste plastics, mold runners
 Reads environmental impact and
improvement priority weights from AHP
calculations as well as green productivity
performance ratios and indices.
 Using an interface program between the
database and the expert system, knowledge
processing is performed on the passed
parameters
 The output consists of diagnostic advice on
the result of inventory analysis, impact
assessment, improvement assessment and
green productivity assessment.
APPLICATION TO GP
ANALYSIS OF
SEMICONDUCTOR
ASSEMBLY/PACKAGING
PROCESS INVENTORY
ANALYSIS
INPUTS
UNIT
 OUTPUTS
PROCESSES
IMPACT
ANALYSIS

IMPACTS
IMPROVEMENT
ANALYSIS

IMPROVEMENT TECHNIQUES

PERFORMANCE
INDICATORS
WATER
RESOURCE
DEPLETION
EMISSIONS
TO AIR
1
Raw
Materials
EMISSIONS
TO WATER
EMISSIONS
TO LAND
3
Ancillary
Materials
HUMAN
TOXICITY:
Land emission
OTHER
RELEASES
ENERGY BASED
PROCESS BASED
GREEN
PRODUCTIVITY
PERFORMANCE
PRODUCT BASED
HUMAN TOXICITY:
Water
emission
n
MATERIAL BASED
HUMAN
TOXICITY:
Air emission
2
Energy
ENERGY
RESOURCE
DEPLETION
MANAGEMENT
-BASED
PRODUCTS/
COPRODUCTS
ECOTOXICITY:
-Aquatic
ECOTOXICITY
- Terrestrial
Conceptual Framework for Green Productivity Analysis Applied to Semiconductor Assembly/Packaging
Wafer
FIRST LEVEL
ASSEMBLY
DI Water
DIE
PREPARATION
Leadframe
LEGEND
Input/ Product
Waste/ emission
Reuse/ Recycle
Die
Waste
Water
DIE ATTACH
Ancillary
Processes
DEIONIZED
WATER
PRODUCTION
WASTEWATER
TREATMENT
Used Solvent
FLUX CLEAN
Reuse
MOLD/ POST
MOLD
Waste
Water
SOLDER PLATE
/ POST SOLDER
CLEAN
DI Water
FINAL TEST,
MARK,
PACK
Semiconductor Assembly/ Packaging Process Flowchart
Semiconductor
product
PROCESSES







DIE PREPARATION
FIRST LEVEL ASSEMBLY
DIE ATTACH
FLUX CLEANING
MOLDING/POSTMOLD CURE
SOLDER/POST SOLDER CLEAN
TESTING
INVENTORY DATA
 SCENARIO 1 : BASE PERIOD
 SCENARIO 2 : PLC MODIFICATION
IN THE MOLDING PROCESS
Input - Output Analysis
Massin = Massout
(1)
Energyin = Energyout
(2)
The total amount of a specific material m for i unit processes
n
M =

m
is:
(3)
i=1
The total amount of specific energy e for i unit processes is:
n
E =

e
(4)
i=1
The total amount of a specific waste or emission from a unit process (i) to medium (j ),
where j = 1 to 3 corresponding to air, water or land and for n unit processes is:
Eij =
n
3
i 1
j 1
 
Eij
(5)
SEVEN IMPACT
CLASSIFICATION







WATER RESOURCE DEPLETION -WRD
ENERGY RESOURCE DEPLETION-ERD
HUMAN TOXICITY ON AIR
- HTA
HUMAN TOXICITY ON WATER - HTW
HUMAN TOXICITY ON LAND - HTL
AQUATIC ECOTOXICITY
- ETA
TERRESTRIAL ECOTOXICITY - ETT
IMPROVEMENT
TECHNIQUES
 MATERIAL-BASED
 ENERGY-BASED
 PROCESS OR
EQUIPMENT-BASED
 PRODUCT-BASED
 MANAGEMENT-BASED
(MBT)
(EBT)
(PET)
(PBT)
(MGMT)
Level 1
Goal : Green Productivity
Level 2
Factors:
Impact
Level 3
Alternative /Schemes:
Improvement
Techniques
GREEN PRODUCTIVITY OF SEMICONDUCTOR
ASSEMBLY / PACKAGING
Water
Resource
Depletion
Energy
Resource
Depletion
Material
Based
Human Toxicity
Air
Energy
Based
Human Toxicity
Land
Process Based
Human
Toxicity
Water
Ecotoxicity
Aquatic
Product Based
Decision Hierarchy Structure for Green Productivity Analysis of Semiconductor
Assembly/Packaging
Ecotoxicity
Terrestrial
Management
Based
• Aj
•
=  Wi Kij
i = 1, 2, …n impact factors
j = 1, 2, …m options
• where Wi = the relative weight of
impact factor i with respect to the
over-all goal
•
Kij
=
relative weight of
option j with respect to impact i
•
Aj
=
priority weight of
option j.
Table 2
Relative Weights of Options (Aj) to Improve Green Productivity Performance
of Semiconductor Assembly/Packaging
Aggregate
RESULT
Impacts
Relative weight of impacts, Wi
Options
MBT
EBT
PET
PBT
MGMT
WRD
0.14
ERD
0.13
HTA
0.13
HTL
0.15
HTW
0.12
Relative weight of options with reference to impacts, Kij
0.19
0.15
0.24
0.42
0.26
0.17
0.32
0.16
0.15
0.15
0.38
0.30
0.27
0.24
0.25
0.12
0.10
0.13
0.12
0.13
0.14
0.13
0.19
0.14
0.21
ETA
0.12
0.31
0.16
0.22
0.14
0.17
ETT
0.21
0.34
0.14
0.24
0.11
0.18
Aj
0.28
0.18
0.27
0.12
0.17
GREEN PRODUCTIVITY
INDICATORS
BASED ON MATERIAL/ENERGY
UTILIZATION:
 Water Utilization Ratio (MUR) =
kg product /kg water input
BASED ON ENERGY UTILIZATION:
 Energy Utilization Ratio (EUR) =
kg product/kWh energy input
SPECIFIC WASTE OR
EMISSION RATIOS
BASED ON WASTE MINIMIZATION:
 Waste Ratio or Emission Ratio (WR/ER) =
kg waste or emission/kg total material
input
GREEN PRODUCTIVITY
INDEX
 GP INDEX OF “1” IS ASSIGNED TO
THE BASE PERIOD AND GP INDEX
FOR TEST SCENARIO IS
DETERMINED
FOR TEST SCENARIO




FOR MATERIAL/ENERGY
PRODUCTIVITY:
IF GP INDEX > 1 , GP IMPROVEMENT
IF GP INDEX <1, GP DECLINE
FOR WASTE OR EMISSION INDICES:
IF GP INDEX > 1 , GP DECLINE
IF GP INDEX < 1 , GP IMPROVEMENT
Table 3. Green Productivity Ratios and Indices
GP Water Utilization Ratio
0.001869 1.000000 0.001871
1.001070
>1
GP Energy Utilization Ratio
0.043383 1.000000 0.045046
1.038333
>1
GP Human Toxicity - Land Waste Ratio 0.000508 1.000000 0.000508
1.000000
GP Terrestrial Ecotoxicity Waste Ratio
0.000726 1.000000 0.000723
0.995868
GP Human Toxicity -Water Waste Ratio 0.184547 1.000000 0.184547
1.000000
GP Aquatic Ecotoxicity Waste Ratio
0.000312 1.000000 0.000312
1.000000
GP Human Toxicity - Air Emission Ratio 0.020410 1.000000 0.020410
1.000000
Test Scenario : With PLC Modification
<1
Improvement in GP/ Water
Utilization Ratio
Improvement in GP/ Energy
Utilization Ratio
Constant GP/Human ToxicityLand Waste Ratio
Improvement in
GP/Terrestrial Ecotoxicity
Constant GP/ Human
Toxicity-Water Waste Ratio
Constant GP/ Aquatic
Ecotoxicity Waste Ratio
Constant GP/Human ToxicityAir Emission Ratio
CONCLUSION/
RECOMMENDATION
 The assessment methodology and computerized
diagnostic prototype may be utilized as an internal
management or self-assessment tool by companies
in their continuous GP improvement strategies.
 The application of expert systems technology is
particularly appropriate to provide flexibility in
testing assumptions and in preserving valuable
human expertise on green productivity
implementation in the manufacturing industry.
 Enhancements may be made in future
versions with more powerful analysis
engine, sufficient database and
comprehensive scope of GP analysis to
include all life cycle stages.
ACKNOWLEDGEMENT
 Asian Productivity Organization (APO) for the
materials on Green Productivity
 Semiconductor and Electronics Industries of the
Philippines (SEIPI) and the Association of
Electronics and Semiconductors for Safety and
Environment Protection (AESSEP) for
their
favorable endorsement of the study to some
member-semiconductor
companies
which
provided the necessary data and information for
this research.