Irfan Ovacik, Principal Engineer Intel Corporation February 7 12 , - - PowerPoint PPT Presentation

Irfan Ovacik, Principal Engineer Intel Corporation

February 7 – 12 , 2016 Dagstuhl Seminar, Modeling and Analysis of Semiconductor Supply Chains

1

Introduction

Quick Bio

• Ph.D. in Industrial Engineering, Purdue University, 1994
• i2 Technologies(now JDA Software), 1994-2005
• Intel Corporation, 2005-Present
• Career (and some of childhood) dedicated to planning

Introduction

3

• Supply Chain Decision Solution (SCDS) Group
• Part of Customer Fulfillment, Planning and Logistics Group
• We build decision support tools that drive the supply chain planning processes

within Intel

• Common Technical Platform
• Well-integrated to Enterprise/Factory Systems
• Large Scale Math Modeling/Optimization
• Deep bench with wide-range of skills
• Software Engineers, Operations Research Engineers
• Right mix of expertise in Intel business, industry best practices, and math

modeling and optimization techniques

Introduction

The content builds on previously presented/published material

Advanced Planning and Scheduling Systems: The Quest to Leverage ERP for Better Planning, Chapter in Planning in the Extended Enterprise: A State of the Art Handbook; Springer (2011). Emergence and Evolution of Advanced Planning and Scheduling (APS) Systems, Workshop on Planning Production and Inventories in the Extended Enterprise at North Carolina State University (2008)

Mostly from a solution provider’s point of view

Today’s focus is on Intel’s journey in the last 10 years to build a world class Master Production Scheduling solution

5

Agenda

• Supply Chain Planning Decisions
• Master Production Scheduling (MPS)
• Definition and Challenges
• Industry Approach to MPS
• Intel’s Approach to MPS

6

Supply Chain Planning Decisions

Master Production Scheduling Tactical Planning Strategic Planning

What do I build at each factory? How do I react to changes in the factory/demand? How do I align capacity to my demand?

Planning Frequency HIGH LOW Data Granularity Planning Granularity DETAILED AGGREGATE SHIFT/DAY Planning Horizon SHORT MONTH/QUARTER LONG

7

Master Production Scheduling Problem

“Optimize” Supply to Meet Demand Across the Entire Manufacturing Network

• All Stages: Wafer Starts, Assembly/Test Starts, Finish Outs
• All manufacturing locations – internal/external
• Considering capacity constraints, material availability, inventory targets and other supply

chain constraints by stage Unique Challenges

• Trend towards longer Front End (Fab/Sort) and shorter Back End (Assy/Test) lead times
• Increased product complexity
• Increased supply chain complexity
• Evolving customer landscape and expectations
• Varying product mixes (high volume/low margin, low volume/high margin)

Industry Approach

Most large semiconductor companies have chosen their solution providers

• All align to an ERP company for the enterprise backbone (SAP/Oracle)
• Most use a Best of Breed strategy (primarily JDA) for MPS solutions
• Some use offerings from partner ERP companies
• These are typically fabless semiconductor companies with a simpler planning

problem

• Intel is the only exception among the large semiconductor companies
• Develop in-house solutions

Acknowledgment: Thanks to Puneet Saxena, VP Manufacturing Planning, JDA Software, for information on the latest state of the master production scheduling in semiconductor industry.

Wafer Starts

• Fab

Capacity Feasible

A/T Request

• Near-Term

Die Feasible

• Product

Family level A/T Response

• A/T Capacity

Feasible

• Product

Family level FG Response

• A/T Capacity

and Die Feasible

• Product level

9

Intel’s Approach: Breaking the Problem Down

• Breaks the problem between Front End and Back End (different lead times, different strategies)
• Breaks the problem along organizational boundaries
• Breaks the problem to manageable subsets which can be optimized efficiently
• Provides the opportunity for faster planning cycles
• Allows analysis of multiple business scenarios
• Supports the decision making process

10

MPS Technical Platform

Integrated Planning Data MPS Platform Reporting & Analytics

• Explain plans
• Compare scenarios
• Compare historical plans
• Provide visibility into key

business metrics

• Identify opportunities
• Common Planning

Application Framework

• Designed and developed

in-house with strong partnership with business

• Optimized for Intel’s SC

needs

• Integrated planning data
• Ensures all planning data

is complete, current and in-sync

• Provides mechanisms for

proactive, automated data quality

Enterprise Systems Factory Systems Demand Planning Inventory Planning

Product

Master Data Build Plans Demand Inventory Product Health, Capacity

Planning Strategies

BU

Planning Application Framework

GUI Data Algorithms

F/S A/T

Product

Master Data Build Plans Demand Inventory Product Health, Capacity

Planning Strategies Product

Master Data Build Plans Demand Inventory Product Health, Capacity

Planning Strategies Product

Master Data Build Plans Demand Inventory Product Health, Capacity

Planning Strategies Product

Master Data Build Plans Demand Inventory Product Health, Capacity

Planning Strategies

Lessons Learned

• Understand the business needs
• Go at a speed the business can absorb
• Pay attention to ALL components of the planning solution
• Data infrastructure, data availability/quality
• User interfaces/reports, analysis/explain-ability of results
• Mathematical modeling and optimization alone do not solve the problem