7/21/2019 ppt 4 spc http://slidepdf.com/reader/full/ppt-4-spc 1/53 S6 – 1 Statistical Process Con trol (SPC) Dr. R K Sing h
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 1/53
S6 – 1
Stat ist ical Process Con trol(SPC)
Dr. R K Singh
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 2/53
S6 – 2
Variab i l i ty is inherentin every p rocess
Natural or commoncauses
Special or assignable causes
Provides a statist ical signal whenassignab le causes are presen t
Detect and elim inate ass ignab lecauses o f var iat ion
Stat ist ical Process Con trol
(SPC)
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 3/53
S6 – 3
Qual ity Assu rance us ing SPC
Design ed Standard
• Centre of s pecif icat ion
l im its (Target)
• Upper Specif icat ion Lim it
(USL)
• Lower Speci f icat ion Limi t
(LSL)
• (USL – LSL ): Desiredtolerance
This represents the voice
of the customer
Status o f process
• Centre of the process
(Process Av erage)
• Upper Contro l Lim it (UCL)
• Low er Contro l Limit (LCL)
• (UCL – LCL ): Spread o f the
process
This represents th e voice
of the process
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 4/53
S6 – 4
Variabi l i ty
• Random
– common causes
– inherent in a
process
– can be el imin ated
on ly through
improv ements in
the system
• Non-Random
– special causes
– due to identi f iable
factors
– can be modi f ied
through operator or
management act ion
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 5/53
S6 – 5
Qual i ty Measures
• Attr ibute
– a produc t character is t ic that can be
evaluated w ith a disc rete response
– good – bad; yes - no
• Variable
– a product character is t ic that is
con t inuous and can be measured
– weigh t - leng th
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 6/53
S6 – 6
App lying SPC to Serv ice (con t.)
•Hospi ta ls – t imel iness and quic kness of care, staff respons es to
requests, accu racy o f lab tests , cleanl iness, cou rtesy,accuracy o f paperwork , speed of adm ittance andcheckouts
•
Grocery Stores – wait ing t im e to check out , frequency of o ut-of-stock
i tems, quali ty of fo od items, cleanl iness, customercom plaints, checkout register errors
• Air l ines
–f l ight delays , lost lug gage and lu ggage handl ing , wait ingt ime at t icket coun ters and check- in, agent and fl ightattendant cou rtesy, accu rate f l ight info rmation ,passenger cabin c leanl iness and m aintenance
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 7/53S6 – 7
App lying SPC to Serv ice (cont.)
• Fast-Food Restaurants
– wait ing t ime for serv ice, customer complaints,
cleanl iness, food qual ity, order accu racy,
employee cour tesy
• Insu rance Companies
– bi l l ing accuracy, t imel iness of claims pro cessin g,
agent avai labi l i ty and respon se t ime
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 8/53S6 – 8
Contro l Charts
• A g raph that
establ ishes contro l
l imi ts of a process
• Control l im i ts
–
upper and lower bandsof a con tro l char t
• Types o f charts
– Attr ibutes
• p-chart
• c-chart
– Variables
• range (R-chart)
• mean (x bar – chart)
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 9/53S6 – 9
Setting up a process contro l
sys temChoose th e character is t ic
for process cont ro l
Choos e the
Measurement m ethod
Choose an appro pr iate
Sampl ing procedure
Choose the type of
Contro l Chart
Calculate
cont ro l l imi ts
Plot the data & Analyse
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 10/53S6 – 10
Character is t ics for p rocess contro lSome examplesSl. No. Type of Applications Characteristic for Measurement
1 Component Manufacturing Conformance of physical measurements ofcomponents and sub-assemblies tospecifications
Conformance to operating characteristics ofmachines and other resources involved in theprocess
2 Final Assembly Number of defects in the product
Conformance to test specifications Number of missing elements
3 Process Industries Temperature, Pressure and Heat specifications Conformance to product specifications Conformance to equipment specifications Vibrations and other variations in equipments
and sub-systems Conformance to specifications of the
automation & control system
4 Service Systems Number of defects in various businessprocesses
Errors in processing documents Conformance to waiting time/lead time related
specifications
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 11/53S6 – 11
Populat ion and Samp l ingDist r ibut ions
Dist r ibut ion ofsample means
Standarddeviat ion ofthe samplemeans
= sx =s
n
Mean of sample means = x
| | | | | | |
- 3sx - 2sx - 1sx x + 1sx + 2sx + 3sx
99.73% of al l xfal l w i th in ± 3
sx
95.45% fall within ± 2sx
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 12/53S6 – 12
Con trol Charts fo r Variab les
For variables that have
cont inuous dimens ions
Weight , speed, leng th ,streng th, etc.
x-charts are to con trol
the central tendency o f the pro cess
R-charts are to contro l the dispers ion ofthe process
These two charts must be used together
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 13/53S6 – 13
Sett ing Chart Lim its
For x-Charts when we knows
Upper contro l l imi t (UCL) = x + z sx
Lower contro l l im i t (LCL) = x - z sx
where x = mean of the sample means or a target
value set for the pro cess
z = num ber of norm al standard deviat ionssx = standard deviat ion of the samp le means
= s / n
s = population standard deviat ion
n = sample size
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 14/53S6 – 14
Sett ing Con tro l Lim i ts
Hour 1
Sample Weight of
Number Oat Flakes
1 17
2 133 16
4 18
5 17
6 16
7 15
8 17
9 16
Mean 16.1
s = 1
Hour Mean Hour Mean
1 16.1 7 15.2
2 16.8 8 16.4
3 15.5 9 16.3
4 16.5 10 14.8
5 16.5 11 14.2
6 16.4 12 17.3n = 9
LCLx = x - z sx = 16 - 3(1/3) = 15
For 99.73% control l im its, z = 3
UCLx = x + z sx = 16 + 3(1/3) = 17
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 15/53S6 – 15
17 = UCL
15 = LCL
16 = Mean
Sett ing Con tro l Lim i ts
Contro l Chartfor sample of9 boxes
Samp le number
| | | | | | | | | | | |
1 2 3 4 5 6 7 8 9 10 11 12
Variat ion dueto assignable
causes
Variat ion du eto assig nable
causes
Var iat ion due tonatural causes
Out ofcontro l
Out ofcontro l
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 16/53S6 – 16
Sett ing Chart Lim its
For x - Charts when we don’t knows
Lower contro l l im i t (LCL) = x - A 2R
Upper contro l l imi t (UCL) = x + A 2R
where R = average range of the samples
A 2 = contro l chart factor found in Table.1
x = mean of the samp le means
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 17/53S6 – 17
Con tro l Chart Facto rs
Table 1
Samp le Size Mean Factor Upper Range Lower Range n A 2 D 4 D 3
2 1.880 3.268 0
3 1.023 2.574 0
4 .729 2.282 0
5 .577 2.115 0
6 .483 2.004 0
7 .419 1.924 0.076
8 .373 1.864 0.1369 .337 1.816 0.184
10 .308 1.777 0.223
12 .266 1.716 0.284
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 18/53S6 – 18
Sett ing Con tro l Lim i ts
Process average x = 12 ounces
Average range R = .25
Sample size n = 5
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 19/53
S6 – 19
Sett ing Con tro l Lim i ts
UCLx = x + A 2R
= 12 + (.577)(.25)
= 12 + .144
= 12.144 ounces
Process average x = 12 ounces
Average range R = .25
Sample size n = 5
FromTable.1
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 20/53
S6 – 20
Sett ing Con tro l Lim i ts
UCLx = x + A 2R
= 12 + (.577)(.25)
= 12 + .144
= 12.144 ounces
LCLx = x - A 2R
= 12 - .144
= 11.857 ounces
Process average x = 12 ounces
Average range R = .25
Sample size n = 5
UCL = 12.144
Mean = 12
LCL = 11.857
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 21/53
S6 – 21
R – Chart
Type of var iables contro l chart
Shows sample ranges over time
Difference between smallest and
largest values in sample
Mon itors process var iabi l i ty
Independent from process mean
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 22/53
S6 – 22
R – Chart
For R-Charts
Lower contro l l im i t (LCLR ) = D 3R
Upper contro l l im i t (UCLR ) = D 4R
where
R = average range of the samples
D 3 and D 4 = contro l chart factors from Table.1
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 23/53
S6 – 23
Sett ing Con tro l Lim i ts
UCLR = D 4R
= (2.115)(5.3)
= 11.2 pounds
LCLR = D 3R
= (0)(5.3)
= 0 pounds
Average range R = 5.3 pounds
Sample size n = 5
From Table S6.1 D 4 = 2.115, D 3 = 0
UCL = 11.2
Mean = 5.3
LCL = 0
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 24/53
S6 – 24
Mean and Range Charts
(a)
These
sampl ing
dist r ibut ions
result in the
char ts below
(Samplin g mean is
sh i f t ing u pward but
range is cons istent)
R-chart
(R-chart does not
detect change in
mean)
UCL
LCL
x-chart
(x-chart detects
sh if t in central
tendency)
UCL
LCL
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 25/53
S6 – 25
Mean and Range Charts
R-chart
(R-ch art detects
increase in
dispers ion)
UCL
LCL
(b)
These
sampl ing
dist r ibut ions
result in the
char ts below
(Sampling m ean
is constant but
d ispers ion is
increasing)
x-chart
(x-chart does no t
detect the inc rease
in d ispers ion)
UCL
LCL
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 26/53
S6 – 26
Con tro l Charts fo r A tt r ibu tes
For variab les that are categor ical
Good/bad, yes /no ,
acceptable/unacceptable
Measu rement is typically count ing
defect ives
Charts may measu rePercent defect iv e (p -chart)
Number o f defects (c-chart)
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 27/53
S6 – 27
Contro l Lim i ts fo r p-Charts
Popu lat ion w i l l be a binom ial dist r ibu t ion,
bu t app ly ing the Central Limit Theorem
al low s us to assume a no rmal dis t r ibu t ion
for the sample stat is t ics
UCLp = p + z sp ^
LCLp = p - z s
p ^
where p = mean fract ion defect ive in the sample
z = num ber of standard deviat ions
sp = standard deviat ion of the sampl ing dist r ibut ion
n = samp le size
^
p (1 - p )
np =^
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 28/53
S6 – 28
p-Chart fo r Data Entry
Sample Number Fract ion Sample Number Fract ion
Number of Errors Defect ive Number of Errors Defect ive
1 6 .06 11 6 .06
2 5 .05 12 1 .01
3 0 .00 13 8 .08
4 1 .01 14 7 .075 4 .04 15 5 .05
6 2 .02 16 4 .04
7 5 .05 17 11 .11
8 3 .03 18 3 .03
9 3 .03 19 0 .0010 2 .02 20 4 .04
Total = 80
(.04)(1 - .04)
100p = = .02 ^
p = = .0480
(100)(20)
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 29/53
S6 – 29
.11 –
.10 –
.09 –
.08 –
.07 –
.06 –
.05 –
.04 –
.03 –
.02 –
.01 –
.00 –
Samp le number
F r a c
t i o n
d e
f e c
t i v e
| | | | | | | | | |
2 4 6 8 10 12 14 16 18 20
p-Chart fo r Data Entry
UCLp = p + z sp = .04 + 3(.02) = .10^
LCLp = p - z sp = .04 - 3(.02) = 0 ^
UCLp = 0.10
LCLp = 0.00
p = 0.04
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 30/53
S6 – 30
.11 –
.10 –
.09 –
.08 –
.07 –
.06 –
.05 –
.04 –
.03 –
.02 –
.01 –
.00 –
Samp le number
F r a c
t i o n
d e
f e c
t i v e
| | | | | | | | | |
2 4 6 8 10 12 14 16 18 20
UCLp = p + z sp = .04 + 3(.02) = .10^
LCLp = p - z sp = .04 - 3(.02) = 0 ^
UCLp = 0.10
LCLp = 0.00
p = 0.04
p-Chart fo r Data Entry
Possib le
assignable
causes p resent
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 31/53
S6 – 31
Con tro l Lim i ts fo r c-Charts
Popu lat ion w i l l be a Poisson d istr ibut ion ,
bu t app ly ing the Central Limit Theorem
al low s us to assume a no rmal dis t r ibu t ion
for the sample stat is t ics
where c = mean number defect ive in the samp le
UCLc = c + 3 c LCLc = c - 3 c
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 32/53
S6 – 32
c-Chart for Cab Company
c = 54 complaints /9 days = 6 complaints /day
|
1
|
2
|
3
|
4
|
5
|
6
|
7
|
8
|
9
Day
N u
m b e r
d e
f e c
t i v e14 –
12 –
10 –
8 –
6 –
4 – 2 –
0 –
UCLc = c + 3 c
= 6 + 3 6= 13.35
LCLc = c - 3 c= 6 - 3 6
= 0
UCLc = 13.35
LCLc = 0
c = 6
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 33/53
S6 – 33
Which Con tro l Chart to Use
Using an x-chart and R-chart :
Observat ion s are variables
Collect 20 - 25 samp les o f n = 4, or n =5, or more, each from a stable pro cess
and compu te the mean for the x-chart
and range for the R-chart
Track samp les of n observat ion s each
Variables Data
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 34/53
S6 – 34
Which Con tro l Chart to Use
Using the p-chart :
Observat ions are attr ibutes that canbe catego r ized in two states
We deal w ith fract ion , propo rt ion, orpercent defect ives
Have several samples, each w ithmany observations
Attr ibu te Data
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 35/53
S6 – 35
Which Con tro l Chart to Use
Using a c-Chart:
Observations are att r ibutes whose
defects per un i t of ou tpu t can becounted
The number counted is a smal l part ofthe poss ib le occurrences
Defects su ch as number of blem isheson a desk, number of typos in a pageof text , f laws in a bol t of c loth
Attr ibu te Data
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 36/53
S6 – 36
Process Capabi l ity
The natural var iation of a p rocess
shou ld be small enough to produce
products that meet the standards
required
A process in stat is t ical contro l does not
necessar ily meet the design
speci f icat ions
Process capabi l i ty is a measu re of therelat ionsh ip between the natural
var iat ion o f the pro cess and the design
speci f icat ions
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 37/53
S6 – 37
Process Capab i l ity Ratio
Cp =Upper Speci f icat ion - Lower Speci f icat ion
6s
A capable process mus t have a Cp of at
least 1.0
Does not look at how wel l the process
is centered in the speci f icat ion range Often a target value of Cp = 1.33 is used
to al low for o f f -center p rocesses
Six Sigma qual ity requ ires a Cp
= 2.0
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 38/53
S6 – 38
Process Capab i l ity Ratio
Cp =Upper Specif icat ion - Lower Specif icat ion
6s
Insurance claims process
Process mean x = 210.0 m inutes
Process s tandard deviat ions = .516 m inutes
Design speci f icat ion = 210 ± 3 m inutes
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 39/53
S6 – 39
Process Capab i l ity Ratio
Cp =Upper Specif icat ion - Lower Specif icat ion
6s
Insurance claims process
Process mean x = 210.0 m inutes
Process s tandard deviat ions = .516 m inutes
Design speci f icat ion = 210 ± 3 m inutes
= = 1.938213 - 207
6(.516)
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 40/53
S6 – 40
Process Capab i l ity Ratio
Cp =Upper Specif icat ion - Lower Specif icat ion
6s
Insurance claims process
Process mean x = 210.0 m inutes
Process s tandard deviat ions = .516 m inutes
Design speci f icat ion = 210 ± 3 m inutes
= = 1.938213 - 207
6(.516)Process is
capable
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 41/53
S6 – 41
Process Capab i l ity Index
A capable process mus t have a Cpk o f at
least 1.0 A capable process is no t necessar i ly in the
center of the speci f icat ion, but i t fal ls w ithin
the speci f icat ion l imi t at both extremes
Cpk = min imum of ,
Upper
Specif icat ion - x
Limi t
s
Lower
x - Specif icat ion
Limi t
s
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 42/53
S6 – 42
Process Capab i l ity Index
New Cutt ing Mach ine
New p rocess mean x = .250 inches
Process s tandard deviat ion s = .0005 inches
Upper Speci f ication Lim it = .251 inchesLow er Speci f ication L imit = .249 inches
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 43/53
S6 – 43
Process Capab i l ity Index
New Cutt ing Mach ine
New p rocess mean x = .250 inches
Process s tandard deviat ion s = .0005 inches
Upper Speci f ication Lim it = .251 inchesLow er Speci f ication L imit = .249 inches
Cpk = min imum of ,(.251) - .250
(3).0005
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 44/53
S6 – 44
Process Capab i l ity Index
New Cutt ing Mach ine
New p rocess mean x = .250 inches
Process s tandard deviat ion s = .0005 inches
Upper Speci f ication Lim it = .251 inchesLow er Speci f ication L imit = .249 inches
Cpk = = 0.67.001
.0015
New mach ine is
NOT capable
Cpk = min imum of ,(.251) - .250
(3).0005
.250 - (.249)
(3).0005
Both calcu lat ions resul t in
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 45/53
S6 – 45
Interpret ing Cpk
Cpk = negat ive number
Cpk = zero
Cpk = between 0 and 1
Cpk = 1
Cpk > 1
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 46/53
S6 – 46
Acceptance Samp l ing
Form of quali ty test ing used for
incom ing mater ials or f in ished goods
Take samp les at random from a lot
(sh ipment) of i tems
Inspect each of the items in the samp le
Decide whether to reject the whole lot
based on the inspect ion resul ts
Only sc reens lots; does not d r ive
qual ity improvement effor ts
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 47/53
S6 – 47
Acceptance Samp l ing
Form of quali ty test ing used for
incom ing mater ials or f in ished goods
Take samp les at random from a lot
(sh ipment) of i tems
Inspect each of the items in the samp le
Decide whether to reject the whole lot
based on the inspect ion resul ts
Only sc reens lots; does not d r ive
qual ity improvement effor ts
Rejected lo ts can be:
Returned to the
suppl ier
Cul led fordefect ives
(100% in spect ion )
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 48/53
S6 – 48
Operat ing Charac ter is t icCurve
Shows how wel l a sampl ing plan
discr im inates between good and
bad lots (sh ipments )
Shows the relat ionsh ip between
the probabi l i ty o f accept ing a lot
and its qual i ty level
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 49/53
S6 – 49
Return whole
sh ipment
The “Perfect” OC Curve
% Defect ive in Lot
P ( A c c
e p
t W h o
l e S h
i p m e n
t )
100 –
75 –
50 –
25 –
0 – | | | | | | | | | | |
0 10 20 30 40 50 60 70 80 90 100
Cut-Off
Keep whole
sh ipment
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 50/53
S6 – 50
An OC Curve
Probabi l i tyo f
Acceptance
Percentdefect ive
| | | | | | | | |
0 1 2 3 4 5 6 7 8
100 –
95 –
75 –
50 –
25 –
10 –
0 –
= 0.05 producer’s risk for AQL
= 0.10
Consumer’sr isk for LTPD
LTPDAQL
Bad lotsIndifference
zoneGoodlots
Figure S6.9
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 51/53
S6 – 51
AQL and LTPD
Accep tab le Qual i ty Level (AQL)
Poorest level of quali ty we are
w i l ling to accept
Lo t To lerance Percent Defectiv e
(LTPD)
Quali ty level we cons ider bad
Consum er (buyer) does not want to
accept lots w ith more defects than
LTPD
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 52/53
S6 – 52
Producer’s and Consumer’s
Risks Producer 's r isk ()
Probabi l i ty o f reject ing a good lot
Probabi l i ty o f reject ing a lot w hen thefract ion defect ive is at or above theAQL
Consumer 's r isk ()
Probabi l i ty o f accept ing a bad lot
Probabi l i ty o f accept ing a lot whenfract ion defect ive is below the LTPD
7/21/2019 ppt 4 spc
http://slidepdf.com/reader/full/ppt-4-spc 53/53
SPC and Process Variab i l ity
(a) Acceptance
sampl ing (Some
bad un its accepted)
(b) Statist ical pro cess
control (Keep the
process in c ontro l )
(c) Cpk >1 (Designa process that
is in contro l )
Lower
speci f icat ion
l imit
Upper
speci f icat ion
l imit
Process m ean, m