DEVELOPING AN EXPERT SYSTEM FOR GP IMPLEMENTATION RUBY PINEDA-HENSON Department of Industrial Engineering Holy Angel University-Angeles City, Philippines.

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 PRESENTATIONINTRODUCTIONRATIONALE FOR GP MODELEXPERT SYSTEMS METHODOLOGYGP MODEL DEVELOPMENTAPPLICATION TO GP ANALYSIS OF

SEMICONDUCTOR ASSEMBLY/PACKAGING

CONCLUSION/ RECOMMENDATION

PRODUCTIVITY

IMPROVEMENT

FRAMEWORK FOR CONTINUOUS IMPROVEMENT

ENVIRONMENTAL PERFORMANCE

GREEN PRODUCTIVITY PARADIGM

FOUNDATION FOR

SUSTAINABLE DEVELOPMENT

RATIONALE FOR GP MODEL

LIFE CYCLE ASSESSMENT - THE TECHNICAL FRAMEWORK

ANALYTIC HIERARCHY PROCESS- MULTICRITERIA DECISION MAKING(MCDM) MODEL AND TOOL

LIFE CYCLE ASSESSMENT

Streamlined LCAProcess-Based Phased Approach Inventory Analysis Impact Assessment Improvement Assessment

IMPACT ANALYSIS IMPROVEMENT Analysis

INVENTORY ANALYSIS

PRODUCTS/ COPRODUCTS

EMISSIONS TO LAND

OTHER RELEASES

EMISSIONS TO WATER

Ancillary Materials

Energy

Raw Materials

EMISSIONS TO AIR

1

n

2

3

OPTION 1

OPTION 2

OPTION 3

OPTION 4

OPTION j

GREEN PRODUCTIVITY PERFORMANCE

IMPACT 5

IMPACT 1

IMPACT 2

IMPACT 3

IMPACT 4

IMPACT i

UNIT PROCESSESINPUTS OUTPUTS IMPACTS IMPROVEMENT TECHNIQUES

GREEN PRODUCTIVITY INDICATORS

ANALYTIC HIERARCHY PROCESS Pairwise ComparisonMechanism For Consistency CheckA 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 BaseWorking MemoryInference Engine

GP DIAGNOSTIC SOFTWARE Front-end database system (Visual FoxPro) Windows shell program/interface CLIPS (C Language Integrated Production

System) expert systemThe 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-ModelsInventory AnalysisImpact AnalysisImprovement AnalysisGreen Productivity Assessment

Figure 2. Green Productivity ES Model Structure

InventoryAnalysis

-------------------------Input Data

Output Data

EnvironmentalImpact

Analysis-------------------------

ClassificationValuation

ProductivityImprovement

Analysis-------------------------

Classification Valuation

Green Productivity (GP) Assessment

-------------------------GP RatiosGP Indices

Knowledge Base

Multicriteria Decision Analysis (Analytic Hierarchy Process)

Input - Output Analysis

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 IndicatorWater Resource Depletion (WRD) Cooling Water, Deionized WaterEnergy Resource Depletion (ERD) Electricity, Thermal EnergyHuman Toxicity in Air (HTA) SO2, NO2, CO,VOCs, Arsine, Phosphine

Metallic (Pb) vaporsHuman Toxicity in Water (HTW) Sulfuric acid, hydrochloric acid, phosphoric acid,

nitric acid, acetic acid, methanesulfonic acidHuman Toxicity in Land (HTL) Heavy metals: Pb, Al, Ni, Cd, Cr, As, SnAquatic Ecotoxicity (ETA) Isopropanol, acetone, N-butyl acetate,

trichloroethylene, xylene, petroleum distillates,halocarbons, Methylene chloride

Terrestrial Ecotoxicity (ETT) Plastics, epoxies, glues, flux, off-specificationproducts or rejects, molding compound, moldrunners, 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

PRODUCTS/ COPRODUCTS

EMISSIONS TO LAND

OTHER RELEASES

EMISSIONS TO WATER

Ancillary Materials

Energy

Raw Materials

EMISSIONS TO AIR

1

n

2

3

MATERIAL -BASED

ENERGY -BASED

PROCESS -BASED

PRODUCT -BASED

MANAGEMENT -BASED

GREEN PRODUCTIVITY PERFORMANCE

HUMAN TOXICITY: Water

emission

WATER RESOURCE DEPLETION

ENERGY RESOURCEDEPLETION

HUMAN TOXICITY: Air emission

HUMAN TOXICITY:

Land emission

ECOTOXICITY: -Aquatic

ECOTOXICITY- Terrestrial

UNIT PROCESSES

INPUTS OUTPUTS IMPACTS IMPROVEMENT TECHNIQUES

PERFORMANCE INDICATORS

IMPACT ANALYSIS

IMPROVEMENT ANALYSIS

PROCESS INVENTORY ANALYSIS

Conceptual Framework for Green Productivity Analysis Applied to Semiconductor Assembly/Packaging

FIRST LEVEL ASSEMBLY

DIE PREPARATION

DIE ATTACH

FLUX CLEAN

MOLD/ POST MOLD

WASTEWATER TREATMENT

DEIONIZED WATER

PRODUCTION

SOLDER PLATE / POST

SOLDER CLEAN

FINAL TEST, MARK,PACK

Semiconductor product

DieLeadframe

Used Solvent

Reuse

Waste Water

Wafer

Ancillary Processes

DI Water

LEGEND Input/ ProductWaste/ emissionReuse/ Recycle

DI Water

Waste Water

Semiconductor Assembly/ Packaging Process Flowchart

PROCESSES

DIE PREPARATIONFIRST LEVEL ASSEMBLYDIE ATTACHFLUX CLEANINGMOLDING/POSTMOLD CURESOLDER/POST SOLDER CLEANTESTING

INVENTORY DATA

SCENARIO 1 : BASE PERIOD

SCENARIO 2 : PLC MODIFICATION IN THE MOLDING PROCESS

Input - Output AnalysisMassin = Massout (1)

Energyin = Energyout (2)

The total amount of a specific material m for i unit processes is: n

M = m (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:

EijEij = (5)

n

i 1

3

1j

SEVEN IMPACT CLASSIFICATION

WATER RESOURCE DEPLETION -WRDENERGY RESOURCE DEPLETION-ERDHUMAN TOXICITY ON AIR - HTAHUMAN TOXICITY ON WATER - HTWHUMAN TOXICITY ON LAND - HTLAQUATIC ECOTOXICITY - ETATERRESTRIAL ECOTOXICITY - ETT

IMPROVEMENT TECHNIQUES

MATERIAL-BASED (MBT)ENERGY-BASED (EBT)PROCESS OR EQUIPMENT-BASED (PET)PRODUCT-BASED (PBT)MANAGEMENT-BASED (MGMT)

GREEN PRODUCTIVITY OF SEMICONDUCTOR ASSEMBLY / PACKAGING

WaterResource Depletion

Level 1Goal : Green Productivity

Level 2Factors:Impact

Level 3Alternative /Schemes:Improvement Techniques

Energy Resource Depletion

EcotoxicityAquatic

HumanToxicityWater

Human ToxicityLand

Human ToxicityAir

EcotoxicityTerrestrial

Management Based

Product BasedProcess BasedEnergyBased

MaterialBased

Decision Hierarchy Structure for Green Productivity Analysis of Semiconductor Assembly/Packaging

• 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

AggregateRESULTImpacts WRD ERD HTA HTL HTW ETA ETT

Relative weight of impacts, Wi 0.14 0.13 0.13 0.15 0.12 0.12 0.21

Options Relative weight of options with reference to impacts, Kij Aj

MBT 0.19 0.15 0.24 0.42 0.26 0.31 0.34 0.28

EBT 0.17 0.32 0.16 0.15 0.15 0.16 0.14 0.18

PET 0.38 0.30 0.27 0.24 0.25 0.22 0.24 0.27

PBT 0.12 0.10 0.13 0.12 0.13 0.14 0.11 0.12

MGMT 0.14 0.13 0.19 0.14 0.21 0.17 0.18 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 Improvement in GP/ Water Utilization Ratio

GP Energy Utilization Ratio 0.043383 1.000000 0.045046 1.038333 >1 Improvement in GP/ Energy Utilization Ratio

GP Human Toxicity - Land Waste Ratio 0.000508 1.000000 0.000508 1.000000 Constant GP/Human Toxicity-Land Waste Ratio

GP Terrestrial Ecotoxicity Waste Ratio 0.000726 1.000000 0.000723 0.995868 <1 Improvement in GP/Terrestrial Ecotoxicity

GP Human Toxicity -Water Waste Ratio 0.184547 1.000000 0.184547 1.000000 Constant GP/ Human Toxicity-Water Waste Ratio

GP Aquatic Ecotoxicity Waste Ratio 0.000312 1.000000 0.000312 1.000000 Constant GP/ Aquatic Ecotoxicity Waste Ratio

GP Human Toxicity - Air Emission Ratio 0.020410 1.000000 0.020410 1.000000 Constant GP/Human Toxicity-Air Emission Ratio

Test Scenario : With PLC Modification

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.