manufacturing system design for high product quality n.
Download
Skip this Video
Loading SlideShow in 5 Seconds..
Manufacturing System Design for High Product Quality PowerPoint Presentation
Download Presentation
Manufacturing System Design for High Product Quality

Loading in 2 Seconds...

play fullscreen
1 / 52

Manufacturing System Design for High Product Quality - PowerPoint PPT Presentation


  • 281 Views
  • Uploaded on

Manufacturing System Design for High Product Quality. Jingshan Li, Dennis E. Blumenfeld, Ningjian Huang Robert R. Inman and Samuel P. Marin Manufacturing Systems Research Lab General Motors Research & Development Center Warren, Michigan, USA

loader
I am the owner, or an agent authorized to act on behalf of the owner, of the copyrighted work described.
capcha
Download Presentation

PowerPoint Slideshow about 'Manufacturing System Design for High Product Quality' - jaden


Download Now An Image/Link below is provided (as is) to download presentation

Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author.While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server.


- - - - - - - - - - - - - - - - - - - - - - - - - - E N D - - - - - - - - - - - - - - - - - - - - - - - - - -
Presentation Transcript
manufacturing system design for high product quality

Manufacturing System Design for High Product Quality

Jingshan Li, Dennis E. Blumenfeld, Ningjian Huang

Robert R. Inman and Samuel P. Marin

Manufacturing Systems Research Lab

General Motors Research & Development Center

Warren, Michigan, USA

5th International Conference on Analysis of Manufacturing Systems

May 21 2005

slide2

OUTLINE

1. Motivation

2. Manufacturing system design impacts quality

3. Research opportunities

4. Andon System

5. Repair and Rework System

6. Conclusions

1 motivation
1. MOTIVATION
  • System design and quality management are important elements in manufacturing industry.
  • Substantial research efforts have been devoted to both of them, but independently.
  • Little research attention has been paid to investigate the interactions between manufacturing system design and product quality.
slide4

Manufacturing

Operation

Manufacturing

System Validation

Manufacturing

System Design

Product

Design

TQM

SPC

JIT

Lot sizing

TQM

QFD

DFQ

TQM

Process capability

Tolerancing

?

  • Scarcity of research on attempting to improve quality in manufacturing system design phase
  • Does manufacturing system design impact quality?
2 manufacturing system design impacts quality
2. MANUFACTURING SYSTEM DESIGN IMPACTS QUALITY
  • Evidences from automotive industry:
    • Harbour report – quality and productivity are positively correlated → improving production system can improve quality.
  • American Axles & Manufacturing – quality improvement due to production system changes, e.g., conveyor, inspection, buffers, etc.
  • Ford/Jaquar – improve quality by adopting Toyota production systems.
  • GM – strip out buffers to improve paint quality in paint shops.
  • Toyota – pay attention to production system’s impact on quality, e.g., Andon, additional inspection stations, selected stationary assembly stations, etc.
slide6
Experiments and analysis:
    • Ergonomics – poor workstation layout and high line speeds may hurt quality performance of manual operations.
  • Andon system – stopping the line to fix every problem can improve throughput of good jobs when average repair times are short.
  • Repair and rework system – appropriate design of repair subsystem can improve quality buy rate.
  • Verdict:Manufacturing system design does impact product quality.
  • Largely unexplored area with promising opportunities
3 research opportunities
Strategic issues:

Flexibility

Agility

Level of automation

Modularity

Outsourcing

Scalability

Emerging Technology

Tactic issues:

Andon

Assembly Line Movement and Balancing

Batch Size

Buffer Location and Size

Centralized/Decentralized Equipment

Feedback Loops

Inspection

Line or Machine Speed

Parallel versus Serial lines

Plant Layout

Repair and rework loops

3. RESEARCH OPPORTUNITIES
4 andon system
4. ANDON SYSTEM
  • Andon – a visual control device to monitor quality on assembly line.
  • The worker can pull the Andon cord to trigger a light as a call for help, and stop the line if needed to correct the problem.
slide9

Current literature contains many popular articles that are descriptive or provide qualitative studies of Andon use.

  • It is claimed and taken for granted that, in spite of line stoppages and productivity loss, overall system performance is improved.
  • Why?Under what conditions?
slide10

Two different Andon strategies:

    • Empower workers to stop the line for every problem, so that all jobs are corrected first time, or
    • Encourage workers to reduce the number of Andon calls, so that line only stops for severe problems.
  • Which one is the right way?
slide11

WHY? WHEN? HOW?

  • Need for quantitative model to
    • analyze performance of a transfer production line with Andon.
    • discover the conditions for successful Andon use.
    • investigate trade-offs between productivity and quality.
slide12

mk

m1

m2

Andon cord

Model Formulation

  • Transfer line with k machines (m1, m2,..., mk) linked to one Andon cord
  • Performance index: Throughput of good quality jobs
slide13

Type of systems

    • No Andon:Job moves to next machine at end of cycle, no matter whether job has problem or not.
  • Full Andon: If job has problem or is not complete at end of cycle, Andon cord is pulled and line stops to allow extra time for repair (up to a maximum time tm).
  • Partial Andon: If job has severe problem at end of cycle, Andon cord is pulled and line stops to allow extra time for repair (up to a maximum time tm).If job has minor problem at end of cycle, job moves to next machine.
slide14

Assumptions

    • Machines are synchronized (all jobs start work at the same time) with identical cycle time c.
    • At end of cycle, each machine has fixed probability ithat job has defect or is not complete. A defective job has probability aito have a severe defect.
    • Repair (correspondingly, severe repair) times are independent and exponentially distributed with parameters i(correspondingly, i, and i < i),with truncation at maximum time tm.
    • At most one Andon pull per cycle.
    • Independence of operations and quality failures.
slide15

No Andon:

Full Andon:

Partial Andon:

  • One-machine case
    • Throughput of good quality job, G
illustration
Illustration

c = 1, l = 0.25, a = 0.5

m = 0.9, n = 0.8

Full Andon

Throughput of

good jobs, G

(jobs per unit time)

Partial Andon

No Andon

Maximum extra time for repair, tm

slide17

0.8

0.79

m = 0.7, n = 0.5

m = 0.8, n = 0.7

0.78

0.77

0.76

0.75

Full Andon

0.74

No Andon

0.73

No Andon

0.72

0.71

Full Andon

0.7

Partial Andon

Partial Andon

0

1

2

3

4

5

Illustration

c = 1, l = 0.25, a = 0.5

Throughput of Good Jobs, G

Maximum extra time for repair, tm

slide18

Theorem: Under assumptions,

    • If  + cµ > 1, then

GNo Andon < GPartial Andon < GFull Andon

  • If  + c < 1<  + cµ, then

GPartial Andon < GNo Andon < GFull Andon

  • If  + c <  + cµ < 1, then

GFull Andon < GNo Andon

GPartial Andon < GNo Andon

slide19

Line Condition

Short repair times

Long repair times

Best Strategy

Full Andon

No Andon

  • Insights
    • Implementing Andon can improve throughput of good quality jobs when average repair times are short (i.e., when repair rate is high).
  • Partial Andon is never the best strategy. Even when repair times for severe defects are short, Full Andon is better than Partial Andon.
slide20
Rules of thumb
    • If average repair time is less than the cycle time, then Full Andon will improve throughput of good quality jobs.
  • If average time to repair severe defects is less than the cycle time, then any type of Andon will improve throughput of good quality jobs.
  • It is worth repairing all defects rather than severe ones only.
  • Right way: Stop line for all problems.
  • Toyota: hundreds of Andons per shift with total line stoppage time of 10-15 minutes.
slide21
Multiple machine case
    • Throughput of good quality job, G

No Andon:

Full Andon:

Partial Andon:

slide22
Throughput of good quality job, G

No Andon:

Full Andon:

Partial Andon:

illustration1

k = 5, c = 1, l = 0.1, a = 0.5

m = 0.9, n = 0.8

Full Andon

Throughput of

good jobs, G

(jobs per unit time)

Partial Andon

No Andon

Maximum extra time for repair tm

Illustration
  • Right way:Stop line for all problems.
5 repair and rework system
5. REPAIR AND REWORK SYSTEM
  • Repair and rework systems are often used in many manufacturing industries: automotive, electronics, packaging, process, etc.
  • In automotive assembly plants, product quality is typically characterized by
    • First Time Quality (FTQ): good job ratio of all first time processed jobs
    • Quality Buy Rate (QBR): good job ratio of all jobs, including first time jobs and reworked jobs.
slide25

Inspection

New Jobs

Confirmation

(OK Jobs)

Main Line

Component

Replacement

Rework

Minor

Repair

  • Layout
slide26

Quality buy rate(Q):

where n and nr are the numbers of first time jobs and reworked jobs, respectively, q1 and qr are first time quality and rework quality, respectively.

slide27
Observations:

Minor repair capacity is limited.

  • Whenqr < q1,we obtainQ < q1.
  • Jobs that only need minor repair will be routed to rework when the minor repair capacity is insufficient.
  • Oftenqr < q1.
  • When minor repair capacity is insufficient, rerouting the jobs needing minor repair to rework reduces the quality buy rate of the main line.
  • In addition, it will waste more materials and resources and lead to loss of throughput.
slide28
Need for a quantitative model to analyze quality buy rate as a function of minor repair capacity

Analysis results show that quality buy rate can be improved by appropriate design of minor repair capacity

The study has been applied in an automotive paint shop

slide30

FTQ

QBR

  • Illustration:
6 conclusions
6. CONCLUSIONS
  • Quality is critical. Manufacturing system design does have a significant impact on product quality.
  • Need to fully understand how it impacts quality and how to incorporate quality with productivity and flexibility in making manufacturing system design choices.
  • Lack of research makes it be a largely unexplored area with promising research opportunities, valued and important to industry.
  • Need to motivate research in the interactions between manufacturing system design and product quality. It will open a new area of manufacturing systems engineering.
slide32
Thanks
  • Prof. Chris Papadopoulos
  • Prof. Semyon Meerkov
  • Prof. Stanley Gershwin
slide34
Assembly line movement – how assembly line progress likely affects quality as well as throughput. Synchronous or asynchronous line? Stationary station or continuous moving line?
  • Assembly line balancing – not only from the point of view of worker utilization, but also to identify quality bottlenecks.
  • Plant layout – how layout affects quality? e.g., U - shaped lines produce better quality products.
  • Number and location of inspection stations – integrated quality and productivity model, information feedback, etc.
  • Number and location of rework loops – more rework loops or less? What should capacity of each be?
  • Feedback loops – feedback from inspection, production data analysis, etc.
slide35
Buffer location and size – buffer accommodate variation, lean inventory contributes to quality, what are tradeoffs?
  • Parallel versus serial lines – Parallel line improves productivity, but increases variations. It is difficult to trace root cause, but it may help quality due to slower speed.
  • Centralize versus decentralized equipment – centralized operations benefit from economic scale, better utilization and is easier for quality control, decentralized operations are responsible for dedicated assembly plants, have less logistic cost, less inventory and quicker feedback from assembly.
  • Batch size – large batch may improve quality by avoiding disruptive changeovers, small batch sizes allow quick defect detection but have frequent changeovers.
slide36
Flexibility – e.g.: fixtures on machines (loading/unloading) or on conveyors (improve throughput but more variability and degraded repeatability and reproducibility), need to delineate tradeoffs between cost, flexibility, throughput and quality for different strategies.
  • Agility – producing multiple products add variability which may damage quality, machine maintenance may require highly trained labors to obtain high quality, need to achieve both agility and quality without huge investment.
  • Level of automation – automatic operation provides better quality, manual has more flexibility, need to understand impact of automation on productivity, quality and flexibility.
slide37
Scalability – capacity expansion by speeding up or adding new machines or plants? Single large machines/plants or many small ones?
  • Modularity – easier for final assembler, but difficult to control, what is the impact on quality?
  • Outsourcing – American automakers spin off parts divisions, Toyota rarely hands complex modules to outside suppliers due to quality concerns.
  • Emerging technology – how to take advantage of data collection, communication, analysis capabilities and intelligent agents to design production system for improved quality?
slide38

No Andon

Full Andon

Partial Andon

slide39

No Andon

Full Andon

Partial Andon

slide40

m=0.9

n=1.2

slide41

m=0.85

n=0.9

slide42

m=0.7

n=0.9

slide43

m=0.7

n=0.8

slide44

Theorem: Under assumptions,

    • If (k-1)cµ+ + cµ > 1, then

GNo Andon < GPartial Andon < GFull Andon

  • If (k-1)c+ + c < 1 < (k-1)cµ+ + cµ, then

GPartial Andon < GNo Andon < GFull Andon

  • If (k-1)c+ + c < (k-1)cµ+ + cµ < 1, then

GFull Andon < GNo Andon

GPartial Andon < GNo Andon

slide45
Rules of thumb
    • If average repair time is less than cycle time plus average time within a cycle working on defective jobs, Full Andon improves throughput of good quality jobs.
    • If average time to repair severe defects is less than cycle time plus average time within a cycle working on defective jobs, then any type of Andon will improve throughput of good quality jobs.
    • It is worth repairing all defects rather than severe ones only.
  • Right way:Stop line for all problems.
slide46

Andon cord

Andon cord

Buffer

Extensions

  • Non-identical machines
  • System with multiple Andon cords
slide47
Assumptions:

A job can be reworked/repaired multiple times. No scrap.

Constant percentages of good quality jobs.

All reprocessed jobs have identical good job ratio.

All routing probabilities are constants.

  • Notation:
    • αx,αr,αs: routing probabilities after main line inspection.
    • βsx, βss,βsr, βxs, βxr: routing probabilities after minor repair and component exchange.
    • N: minor repair capacity..
slide48
Need to develop a quantitative model to

analyze quality buy rate as a function of minor repair capacity

design appropriate repair capacity to achieve desired quality buy rate

investigate the trade-offs between investment costs and saving from productivity and quality improvement

slide50
Corollary: Under assumptions, the quality buy rate is monotonically increasing with respect to q, qr and N (when minor repair capacity is insufficient).

FTQ

QBR

slide52

Corollary: Under assumptions, the number of rerouted jobs is monotonically decreasing with respect to q, qr and N (when the minor repair capacity is insufficient).