Term Project
ME586
1
Topics
Related
to Design for Manufacturability
Approval of term project is required.
2
Goals
Oral
presentation
lTechniques for slides
lTechniques for talking
Written
report
lStyle
lContents
3
Presentation
Cue
cards DONT USE THEM
Taguchi DOE with
Interactions Among Variables
ME 586
1
Interactions
One
variable affects another in some
way.
Need a way to compute interactions
It should be no surprise that additional
experiments will be needed to deal with
interactions.
What does i
Taguchi Loss Function
Application
ME 586
1
How to Compare 2 Processes
Which setting, T0 or T0 + 25 will be better for
achieving uniform thickness of polysilicon
with target thickness of 3600 .
Experiment - Run a number of batches under
the two temperature
Guardbands
ME 586
1
Classic US View of Loss
% of product
out of spec
LSL=15
T=30
USL=45
2
Taguchi Loss Function
A
T-
T
x
T+
3
Definitions
product tolerance
A loss if defective part is shipped
K c cost per unit of product (mfg, test, market)
Within T , lo
Reliability
ME 586
1
Some theory first
f(t)
All units fail at some
time after sale
pdf
F(t)
cdf
d
f (t ) = F (t )
dt
t
F (t ) = f (v)dv
0
2
Mean Time To Failure
If b > a > 0, then T (time to failure)
b
P(a<T b)=F(b)-F(a)= f(t)dt
a
MTTF = T = T = E[T ] = t
Developing Design for X Tools
ME586
1
Basic Questions
How
to represent decisions in designing
products, processes, and resources
How to relate these decisions
How to measure decisions and their
intentions
How to collect display data necessary
for the
Failure Modes and Effects
Analysis (FMEA)
1
Motivation
Product
reliability is desirable metric,
but HOW to achieve it is problematic
Desire a procedure that injects
reliability early in design process
Goals are to improve
lproduct quality
lprocess effi
ME 486 / ECE 586: Design for Manufacturability
Spring 2016
January 19, 2016
Instructor
Prof. Samuel John (sajohn@unm.edu)
ME 319, (505) 277-6279
Office hours: Tuesday 2pm-3pm and Thursday 2pm-3pm (tentative)
Course Location and Time
Dane Smith Hall 132, T
Design for Recycling
ME 586
1
Approach
Design
for disassembly and design for
recycling highly related
Economic value could become huge in
the future
lLegislation
lPublic relations
When
is it appropriate to consider DFD?
lDesign phase!
2
Two alternative
Taguchi Loss Function
ME 586
1
Introduction
Based
on Taguchi philosophy, quality is
said to be achieved if:
lAll quality characteristics have minimum,
stable variation around the optimum value.
Quality
control aims at reducing
variations in quality char
Boothroyd and Dewhurst
Automated Assembly
ME 586
1
DFMA
Important
for manual assembly
operations
CRITICAL for automated assembly
Think about symmetry
lIf asymmetric, need vision or other
expensive sensing to determine the
orientation of part
2
Assembly
Taguchi Loss Function
ME 586
1
Taguchi Loss Function
A
T-
T
x
T+
2
Taguchi Loss - Nominal Best
A cost of replacing defective product
tolerance
A
l = 2 ( x T ) 2 Taguchi Loss formula
1 n
A n
E(l ) = l i = 2 ( xi T ) 2
n i =1
n i =1
A
E(l ) = 2
n
n
2
cfw_(
Boothroyd & Dewhurst
Manual Assembly Techniques
ME586
1
Part Handling Guidelines
Parts
should have end-to-end symmetry
and rotational symmetry about axis of
insertion
If parts cannot be made symmetrical,
make them obviously asymmetrical
2
NASA-Parachute
Process Capability
ME 586
1
Probability Density Functions
Fx(x)
x
2
Probability Density Functions
x =
xf
x
( x)dx
mean
x2 =
2
(
x
)
f x ( x) dx variance
x
2
x
=
2
2
(
x
2
x
+
x
x ) f x ( x ) dx
2
x
=
x
2
f x ( x)dx 2 x
xf
x
( x)dx +
2
x
f x ( x) dx
Boothroyd and Dewhurst
Automated Assembly
ME 586
1
DFMA
Important
for manual assembly
operations
CRITICAL for automated assembly
Think about symmetry
lIf asymmetric, need vision or other
expensive sensing to determine the
orientation of part
2
Assembly
Taguchi Design of
Experiments (DOE)
ME 586
1
Taguchi Parameter Design
Complex machines have many controls, e.g.,
n control knobs
Suppose each control has 2 values
2n experiments required
Assume 15 controls
l215 experiments required - all combinations
l
Design for the Environment
ME 586
1
Why is DFE Important?
Technology
sometimes but not always
aware of the consequences, e.g.,
refrigerator compressors
lEarly - ammonia, sulfur dioxide
lModern - chlorinated fluorocarbons (CFCs)
Waste
streams today
l180
Term Paper for Grad Students
ME586
1
Term Paper Grad Students
Graduate
students ONLY
Goal
lWrite a journal paper on the state-of-theart in something
l5-10 pages single spaced
lMust look like a journal paper
I
need to approve the topic.
2
Topics
Anythi
Presentation for Grad Students
ME586
1
Presentation Grad Students
Graduate
students ONLY
Goal
lPrepare a 25 minute PowerPoint
presentation on a topic related to DFMA
that is not taught in class.
I
need to approve the topic.
2
Topics
Anything
related t
Overview
ME 586
1
Logistics
Contact: ME Dept office
lRoom 319
l277-6279
2
Logistics
Lecture notes
lPDF of notes available
lhttps:/learn.unm.edu
lLet me know if there are problems linking in all
sections sometimes an issue
Email used a lot not getting emai
Introduction
ME 586
1
What Do Customers Want?
Separate
customer wants from needs
lBells and whistles - will customer pay for
the feature?
Metrics
lHow to measure what a customer wants
Tolerancing
lReality and cost
2
Types of Tolerancing
NB - Nominal is
B&D Introduction
ME 586
1
Manufacturing - a Definition
Manufacture
lCreate the individual component parts of a
product or assembly
Assemble
lAddition or joining of parts to form the
completed product
Hence
DFM and DFA are different.
Interest is in DFMA
Rules for Design - Hundal
ME 586
1
Fundamental Rules for Design
Clarity
Simplicity
Safety
Ask
questions related to requirements
2
Function
Are
the main function and secondary
functions fulfilled?
What other functions are needed?
Is there an unambig
Modified Westinghouse DFM
ME586
1
Goal
Target
lTOP - Total number of operations
lTAT - Total assembly time
In
manufacturing, simple mistakes cost:
ltime
lwaste material
lmoney
lattitude of workers
2
General Comments
DFA
for mechanical parts is more
adv
Process Capability
ME 586
1
Probability Density Functions
Fx(x)
x
2
Probability Density Functions
x =
xf
x
( x)dx
mean
x2 =
2
(
x
)
f x ( x) dx variance
x
2
x
=
2
2
(
x
2
x
+
x
x ) f x ( x ) dx
2
x
=
x
2
f x ( x)dx 2 x
xf
x
( x)dx +
2
x
f x ( x) dx
Testing
ME 586
1
Testing Reality - Errors Possible
P(A) means probability of A
Fail
MFGR
D F
D F Type I Error
Test
Defect,
Not defect
Not
Fail
D F Type II Error
D F
Type 1 error: You test a part and you state it is defected,
but in reality it is not ( We