Quality: 7 Tools

Antonio Moschitta
Introduction

◼ Company-Wide Quality Control (CWQC)

➢ Principles:
1. Customer satisfaction
2. Quality improvement
3. Application of quality tools
4. Education and training
5. Leadership
Introduction

◼ Company-Wide Quality Control (CWQC):

➢ Method PDCA

➢ Basic Tools Seven Tools

➢ Advanced tools i ANOVA, DOE…

Introduction

◼ Data collection: "analyze data and speak with

facts"
Sampling
Significance:
Are the data collected sufficient to represent
reality with adequate accuracy?
Representativity:
Are the criteria for their collection such as to
guarantee us a correct interpretation of the
facts?
Both requirements must be met
Introduction

◼ Data collection: "analyze data and speak

with facts“
Data types
▪ Measurement data
▪ Data based on judgments of merit
▪ Data based on sequential ordering
▪ Data based on scoring scales
▪ …
Introduction

◼ Data collection: ”analyze data and speak

with facts“
Requirements
▪ Clarity (know the purpose of the collection)
▪ Ease (supports structured in a simple, organic, clear
way and completed with all the information necessary
to define the analysis)
Introduction

◼ Data collection: ”analyze data and speak

with facts“
Recommendations
▪ Correct mental attitude
▪ Clear purpose of data collection
▪ Will, ingenuity, commitment and use of scientific
methods
SEVEN TOOLS

1. Check Sheet
2. Stratification – Flowchart – run-chart
3. Pareto Diagram
4. Cause-Effect Diagram
5. Scatter Diagram
6. Histogram
7. Control Chart
NOTES
◼ The Seven Tools are a heterogeneous set of techniques, which include
methods based on measurement data, prior knowledge, brainstorming
techniques and the application of statistical techniques.
◼ The Seven Tools are structured in such a way as to require a level of
technical knowledge, in particular statistics, which depends on the point of
view of those who approach them.
◼ Planning their use in a given context may require a rather high level of
knowledge (Ex: what parameters need to be measured? How accurately?
How often? How many measurements need to be taken?).
◼ Conversely, once the planning phase is complete, actual use requires an
average level of knowledge, suitable for use by personnel directly involved
in production
1. Check Sheet

◼ Support to collect data

◼ Suitably designed
◼ Reference model
◼ Completed with more information
 1. Check Sheet
◼ Distribution of discrete quantities
◼ Measurement results are ranked by membership of assigned ranges.
◼ The use of the check mark is intended to facilitate the counting of measurement
results
1. Check Sheet

◼ Distribution of
continuous
quantities
1. Check Sheet

◼ Defect position

https://qctraininginc.com/7-quality-tools/
1. Check Sheet

◼ Synthesis
1. Check Sheet

◼ Synthesis
1. Check Sheet

◼ Confirmation
of execution
1. Check Sheet

◼ Confirmation
of execution
2. Stratification
◼ What is it?
 Method of classifying data into homogeneous
groups
 Powerful tool for determining how a factor in
the manufacturing process affects product
characteristics
 Logical subdivisions = stratification factors
(time, operators, machines, material, controls)
 Purpose: to identify the weight of certain
factors on the overall result, through
comparison between global and stratified data
2. Stratification
◼ Layering application steps:
 Define the phenomenon/characteristic to be
analyzed
 Represent the overall relative data
 Identify the most appropriate stratification
factors
 Classify existing data into homogeneous
groups
 Graph each group of data
 Compare homogeneous groups within each
stratification factor to detect differences
2. Stratification

◼ The presence of differences is indicative of

successful stratification
◼ Understanding of a phenomenon increases as
the number of stratification factors increase
◼ Stratification provides useful information to
define a phenomenon and therefore to identify
the causes that have the greatest negative
influence on the phenomenon itself
2. Flowchart

◼ A flowchart, also called process map, is a

graphical representation of the structure of a
process or system
◼ A flowchart identifies and describes the
sequence of activities (or the flow of materials
and information) occurring in a process.
◼ Flowcharts help understand the process more
objectively, since they provide a picture of the
steps needed to accomplish a task
2. Flowchart, semanthic

https://www.conceptdraw.com/examples/flowchart-definition
2. A Flowchart example: Order Process

https://www.conceptdraw.com/examples/flowchart-definition
 2. Run chart
◼ A run chart, also known as run-sequence plot, is
a graph that displays observed data in a time
sequence.
◼ A run charts can show the performance of a
process (e.g. quality or productivity indicators)
over time in a graphical fashion

https://upload.wik
imedia.org/wikipe
dia/commons/d/d
f/SimpleRunChar
t.jpg
 2. Run chart
◼ Run charts also help identifying process
changes and trends over time and show the
effects of corrective actions.
◼ Control Charts extend the capabilities of the Run
Chart

https://upload.wik
imedia.org/wikipe
dia/commons/d/d
f/SimpleRunChar
t.jpg
3. Pareto Diagram

◼ The Universal Law of Priorities (distinguishes

priority from negligible aspects)
"In every situation there are always a few
important things and many others of little
importance"
◼ Empirical law, derived from the observation of a
significant number of practical cases Pareto
analysis: decision-making basis on which to
develop actions through the Pareto chart
3. Pareto Diagram

◼ Also known as “Pareto Principle”, or as “80-20

Rule”, “The Law of the Vital Few” or “The
principle of Factor Sparsity”
◼ 80% of effects arise from 20% of the causes

◼ 20% of your action may account for 80% of the

effects
3. Pareto Diagram
▪ Example:
3. Pareto Diagram
▪ Example:
3. Pareto Diagram

https://www.qimacros.com/pareto-chart-excel/pareto-analysis/
3. Pareto Diagram
 3. Pareto Diagram

https://support.sas.com/documentation/cdl/en/qcug/63922/HTML/default/viewer.htm#q
cug_pareto_a0000000961.htm
4. Cause-Effect Diagram

 Graphical representation of the measurement

results, designed to determine:

◼ Correlation between a non-quality effect and

possible causes
◼ Correlation of possible causes with each other
4. Cause-Effect Diagram

◼ Construction (1)
 Three methods:
1) Classification of causes
2) Steps in the process
3) List of causes
4. Cause-Effect Diagram

◼ Construction (2)
1.Classification of causes:
1. Step 1 → choose the effect you want to analyze
(circumscribed, well known)
2. Step 2 → identify causes

The engine of my car

does not start
4. Cause-Effect Diagram

◼ Construction (3)
1.Classification of causes:
4M: Methods, Materials, Machines, Manpower
5M: Methods, Materials, Machines, Manpower, Milieu
8M: Methods, Materials, Machines, Manpower, Milieu,
Management, Measurements, Money
4. Cause-Effect Diagram

◼ Construction (4)
1. Classification of causes:
Step 3 → Identification of major categories of causes

milieu method material

The engine of my car

does not start

man machine
 . Cause-Effect Diagram
4.

.
◼ Construction (5)
1. Classification of causes:
Step 4 → Identification of causes for each major category

milieu method material

The engine of my car

does not start

Electrical system
man machine
dirty
Spark plugs
4. Cause-Effect Diagram

◼ Construction (6)
1.Classification of causes:
Step 5 → include in the diagram all the factors that can
affect the effect
 4. Cause-Effect Diagram

https://images.saymedia-content.com/.image/t_share/MTc2MzE1MzI4MDM1MzY2MDYx/fishbone-diagram-ishikawa-diagram-
continuous-process-improvement.gif
 4. Cause-Effect Diagram

https://www.gov.scot/binaries/content/gallery/publications/independent-report/2020/11/root-cause-analysis-care-home-
outbreaks/SCT10204672081_g00.gif
4. Cause-Effect Diagram

◼ Construction (7)
2. Classification by process steps:
• Identify sequential steps and identify them with block
diagram
• Analyze all the phases to find the causes
• Useful for sequential process-dependent effects
4. Cause-Effect Diagram

◼ Construction (8)
3. List of causes:
• List of all possible causes
• Structuring Cases to Highlight Relationships
4. Cause-Effect Diagram

◼ Preparation
1. Collaboration with other people interested in the
problem
2. Don’t forget factors belonging to Major Categories
of Causes
3. Consider errors as causes
4. Separate Diagrams for specific effects
4. Cause-Effect Diagram

◼ Analysis
1) Identify the most likely causes
2) Define the most important causes
3) Organize the most important cases
4) Verify the causes starting from the most important
5) Design of verifications

▪ Corrective actions if the cause is found to be

influential
▪ Verification of the second cause in order of
importance if the cause is not influential
4. Cause-Effect Diagram

◼ Use
1. Find the effect you want to analyze
2. Choose the most suitable method for construction
3. Diagram Construction
4. Classification of causes
5. Verification of the validity of causes
6. Corrective actions
4. Cause-Effect Diagram

◼ Fields of application

1. Process Improvement
2. Corrective actions
3. Development of standard operating procedures
4. Educational Use
5. Discussion Guide
6. Verification of the technical level of knowledge
7. Knowledge Sharing
4. Cause-Effect Diagram

◼ Pros
 Stimulates creativity in the search for causes
 Categorizes possible causes
 Diagramming in teams opens up new perspectives
 Causes can be represented
 Encourage employees to come up with ideas
 Simplifies the problem
 It is easy to apply and learn
4. Cause-Effect Diagram

◼ Cons
 Requires discipline and simplification to ensure that the
diagram retains its usefulness
 Predefined categories (e.g. 5 M) can limit creative approaches
to problem-solving
 Complex connections cannot be represented in the Ishikawa
diagram
 Doesn't provide solutions, but only displays causes
5. Scatter Plot

◼ What is it?

◼ A tool that allows you to highlight the link between

measurable quantities
◼ An instrument that allows you to quantify the
strength of the bond
5. Scatter Plot
Example: a cutting machine operating on a conveyor belt

◼ Preparation (1):
1. Gather and order data
Velocity Cut Velocity Cut
Number Number
[cm/s] Length [cm/s] Length
5. Scatter Plot

◼ Preparation (2):
1. Plot data

1055
1050
1045
1040
lungh. tagliata
Cut Length

1035
1030
1025
1020
1015
1010
1005
0 2 4 6 8 10
velocità del trasportatore
Velocity [cm/s]
5. Scatter Plot

◼ Interpretation (1):
1. Positive correlation

2. There may be a positive correlation

5. Scatter Plot

◼ Interpretation (2):
3. No correlation
4. There may be a negative correlation
5. Negative Correlation
5. Scatter Plot
◼ Evaluation (1):
Sign-test: Hypothesis testing to decide whether one random variable is
greater than another. Deciding whether there is a positive or negative
correlation is the same as deciding whether the sum of the points
contained in quadrants I-III exceeds the sum of the points contained in
quadrants II-IV. The coordinates of the axes are given by the medians of
the horizontal and vertical coordinate sets
1055
1050
1045 II I Area Points
1040
I 19
tagliata

1035
Cut Length

1030
II 4
III 20
lungh.

1025
1020 IV 5
1015 On lines 2
1010 III IV ____
1005
0 2 4 6 8 10
Total 50
Velocity
velocità [cm/s]
del trasportatore

Since most of the samples are contained in quadrants I-III, it can be

concluded that there is a probable positive correlation between the two
quantities represented
5. Scatter Plot
◼ Evaluation (2):
Sign-test: By iterating the procedure in the sub-quadrants, it is possible to
estimate the level of correlation between the two quantities by drawing the line
that joins the centers of symmetry. A more rigorous alternative is the
evaluation of the correlation coefficient by linear regression on the data
1055
1050
1045
1040
lungh. tagliata
Cut Length

1035
1030
1025
1020
1015
1010
1005
0 2 4 6 8 10
velocità del trasportatore
Velocity [cm/s]
5. Scatter Plot
◼ Evaluation (2):
Sign-test: By iterating the procedure in the sub-quadrants, it is possible to
estimate the level of correlation between the two quantities by drawing the line
that joins the centers of symmetry. A more rigorous alternative is the
evaluation of the correlation coefficient by linear regression on the data
1055
1050
1045
1040
Pearson correlation coefficient
lungh. tagliata
Cut Length

1035
1030 σ𝑁
𝑛=1 𝑥𝑛 − 𝑥ҧ 𝑦𝑛 − 𝑦
ത
1025
𝑟𝑥𝑦 =
1020 σ𝑁
𝑛=1 𝑥𝑛 − 𝑥ҧ
2 𝑦𝑛 − 𝑦ത 2
1015
1010 1 1
1005 𝑥ҧ = 𝑁 σ𝑁
𝑛=1 𝑥𝑛 , 𝑦ത = 𝑁 σ𝑁
𝑛=1 𝑦𝑛
0 2 4 6 8 10
velocità del trasportatore
Velocity [cm/s]
5. Scatter Plot

◼ How to use the Scatter Plot (1):

1. Layering is an important tool for interpreting the correlation
diagram. In the figure, the stratification makes it possible to
highlight two subsets characterized by strong correlation.

Without Stratification With Stratification

5. Scatter Plot
◼ How to use the Scatter Plot (2):
2. Determine the range in which correlation exists

Range of experimental values

Range of production
values
Product characteristic

Process parameter
5. Scatter Plot

Limits:
◼ The correlation diagram does not allow us to infer cause-and-effect
relationships (In the absence of other information, it can be concluded
either that parameter 1 influences parameter 2, or that parameter 2
influences parameter 1, or that there are one or more additional factors
that influence both parameters 1 and 2)
◼ The correlation plot can lead to incorrect conclusions if used
inappropriately (if there is a third quantity influencing parameters 1 and 2,
the results of the correlation diagram can be misleading)
5. Scatter Plot

Ex.: plot the x1- x3 e x2- x3 cross-plots for the data shown in the table.
What kind of correlation is highlighted by the two cross-plots?

Data point x1 x2 x3
1 -2 1 10

2 -1 -2 5

3 0 -5 0
5. Scatter Plot
Ex. (continued): plot the x1-x3 e x2-x3 cross-plots for the data shown in the table
(extends the previous table)
Are the conclusions the same as before?

Data point x1 x2 x3
1 -2 1 10
2 -1 -2 5
3 0 -5 0
4 1 2 15
5 2 -1 10
6 1 -2 7
7 2 1 14
8 0 0 10
9 3 2 17
10 1 -3 5
11 1 -5 1
5. Scatter Plot

Ex. (continued): perform stratification to draw more accurate conclusions

5. Scatter Plot

Es. (final): data were obtained using the formula x3=x1+2x2+10

The ambiguity of the results is due both to the insignificant size of the
sample represented by the first table, which leads to an incorrect
conclusion, and to an inadequate use of the tool, as the estimation of the
correlation between two variables is influenced from time to time by the
third variable
Possible stratification criterion: pairs (x1, x3) characterized by the same
value assumed by x2
PRINCIPLES OF OPERATION OF A
SYSTEM

◼ Concept of system = complex set of parts related to each

other by interdependence
◼ Every system is aimed at a specific goal = to carry out a
transformation (process)
◼ Nature of a system's performance:
 by measuring a parameter of a given process, a set of different
data is obtained, which can be characterized by:
◼ Average
◼ Dispersion
PRINCIPLES OF OPERATION OF A
SYSTEM
◼ Statistical view (1): "Statistics is the fundamental
discipline for interpreting data variability"
◼ It is important to collect a sufficient amount of data,
with appropriate criteria
◼ Ex: E. H. Simpson Paradox (1951)
◼ Two plants produce the same object, with a process
that is not yet well understood. An analysis of the
outgoing defects is carried out, and it is found that the
overall success rate is 11% out of 10100 products. By
applying a process change in both plants, the overall
success rate is 46% out of 11000 products. What
conclusions can be drawn from this?
PRINCIPLES OF OPERATION OF A
SYSTEM

Plant 1 Plant 2

Successes 1000 95
method 1
Failures method 9000 5
1
Successes 50 5000
method 2
Failures method 950 5000
2
PRINCIPLES OF OPERATION OF A
SYSTEM
◼ The second method is inferior to the first in both
implants!
◼ The paradox is because the numerousness of data
sets collected in the two plants were not chosen
adequately
PRINCIPLES OF OPERATION OF A
SYSTEM
◼ Statistical view (2):
◼ "Statistics is the fundamental discipline for interpreting
data variability"
◼ What drives data variability?
 Natural Variability (System under Statistical Control)
 Accidental Variability (System out of Statistical Control)
PRINCIPLES OF OPERATION OF A
SYSTEM
◼ Statistical view (3):
◼ Natural variability:
 It depends on how the process was designed and
implemented (inherently imperfect knowledge of the factors
involved)
 It is stable over time (steady-state context)
 The magnitude of the data dispersion can be estimated It is an
intrinsic aspect of the process, which does not require
corrective action
 Can be reduced through improvements
PRINCIPLES OF OPERATION OF A
SYSTEM
◼ Statistical view (4):
◼ Accidental variability:
 System disturbance due to sudden and undesirable changes
in one or more influencing factors (process inputs)
 The mean and dispersion of measurement data are generally
not stationary, and may exhibit undesirable variations
 It is the cause of frequent non-conformities in the output of a
process
 Requires corrective action to identify and eliminate the cause
PRINCIPLES OF OPERATION OF A
SYSTEM
◼ Statistical view (5):
 A process is in a state of statistical control when there are no
causes of accidental variability
 In the context of process control, it is essential to identify the
occurrence of accidental causes of variability at an early
stage, in order to take corrective action
 The occurrence of an accidental cause of variability can be
detected by analyzing the variations induced by it in the
distribution of measurement data and related statistics (mean,
standard deviation, etc.)
PRINCIPLES OF OPERATION OF A
SYSTEM

◼ Statistical view (6):

Distinction between natural and accidental causes:
 Important knowledge to set up effective control
actions
 Reduces the risk of taking unnecessary corrective
actions or not taking them when necessary
 Possible tools: histogram, control chart
 Best Tool: Control Card (Upper Control Limit/Lower
Control Limit)
 Deviation from control limits indicative of the
presence of accidental causes
HISTOGRAM

◼ What is it?
 Graphic tool
 Tool used to represent data
 Data Probability Density Estimator
 It has fundamental elements
6. HISTOGRAM

◼ Fundamental elements
▪ Class = bar
▪ Class Limits = Minimum and Maximum
▪ Value Class Amplitude = Max-Min Range
▪ Frequency = number of observations
40 class
35
30
25
20 amplitude
15
10 frequency
5
0

limits
6. HISTOGRAM

◼ Construction (1)
Determine the maximum data excursion “R”

Thickness of metal parts [mm]

Largest value in group

Smallest value in group
6. HISTOGRAM

◼ Construction (2)
◼ Define the number of classes, the class range (h),
the class limits:
◼ The excursion R is divided into classes of equal
amplitude
◼ The number of classes is denoted by k and is
chosen according to the number of data N
Example: 𝑘 = 𝑁 𝑘 = 100 = 10
◼ The class range is obtained as (xmax-xmin)/k
Example h=0.38/10=0.038 (unwieldy range, you
can choose h=0.05 and then k=0.38/0.05= 8)
6. HISTOGRAM

◼ Construction (3)
◼ The histogram estimates the probability density, and the
number of classes is a trade-off between two conflicting
requirements
◼ Given the number of measurements, too many classes
leads to "jagged" histograms, i.e., the classes tend to be
populated in a way that can greatly differ from the
probabilities they estimate (the variance of the estimator is
too high)
◼ Conversely, a small number of classes leads to reliable
estimates, but reduces the amount of information obtained
6. HISTOGRAM

◼ Construction (4)
Define the number of classes, the class range (h),
the class limits:
 Define class boundaries starting with the minimum value
(3.30) and increasing it by 0.05 for each class
 Class I = 3.30 – 3.35

 Class II = 3.35 – 3.40

 Class III = 3.40 – 3.45

 Class IV = 3.45 – 3.50

6. HISTOGRAM

◼ Construction (5)
Classify data in a frequency table:
 The data is recorded in the relevant classes
 Attendance is calculated for each class

Plot the histogram

 X-axis = class limits
 Y-Axis = Frequency
6. HISTOGRAM

◼ Construction (6)
Plot the histogram

Specifications
6. HISTOGRAM
◼ Interpretation (1)
 Use to learn about data structure
 Analyze shape and placement against specifications:

▪ Questions about the form of distribution

▪ Questions about positioning against specifications
6. HISTOGRAM
◼ Interpretation (2)
Shape analysis:

Island Diagram Comb tooth diagram

40 30
35 25
30
20
25
20 15
15 10
10
5
5
0 0

◼ Island diagram: possible deviation of the observed process average,

defects in the measuring instrumentation
◼ Comb tooth diagram: number of classes too high compared to the
number of measurements taken
6. HISTOGRAM
◼ Interpretation (3)
Shape Analysis:
Asymmetric Cliff
35 35
30 30

25 25
20
20
15
15
10
10
5
5 0
0

◼ Possible interpretation: average and dispersion of the process such as to exceed the
measuring range of the instrumentation used
6. HISTOGRAM
40
35
30
◼ Interpretation (4) 25
20
Shape Analysis: 15
10
5
0

◼ Bimodal diagram: deviation of the mean of the observed process, which

occurred during the observation period
6. HISTOGRAM

◼ Interpretation (5)
Positioning against specifications: adequate

35
30
25
20
15
10
5
0
6. HISTOGRAM

◼ Interpretation (6)
Positioning against specifications: mean shift
6. HISTOGRAM

◼ Interpretation (7)
Positioning against specifications: dispersion too wide
A stratification may be needed
35
30
25
20
15
10
5
0
1
6. HISTOGRAM
◼ Limitations (1)
 The histogram is a powerful tool for accurately determining the
probabilistic distribution of an observed parameter, as long as the
context is stationary.
 The onset of an accidental cause, however, is a typical event of a
non-stationary context
 The histogram estimates the distribution of the phenomenon after
acquiring all the data and does not use the information brought
by the instant of acquisition of each measurement.
 The same data, captured in a different order, produces the same
histogram.
 As a result, the histogram is an instrument that is not very
sensitive to the onset of accidental causes, which are not
detected in a timely manner
6. HISTOGRAM
◼ Limitations (2)
 Any deviation of the observed process mean from an m1 value to
an m2 value can be manifested by a bimodal histogram.
 However, the histogram does not allow us to conclude which of
m1 and m2 was the initial mean, or even at what point in time the
mean of the process began to deviate from one value to another.
 Similarly, an increase in the dispersion of data due to an
accidental cause results in a histogram that is less concentrated
about its axis of symmetry and does not even allow the
conclusion that the process is not in a state of statistical control.
 These limits are exceeded by the Control Chart, which allows you
to determine the status of a process by making a series of short-
term averages
7. CONTROL CHART

◼ What is it?
 Dynamic Representation Tool
 A tool that allows you to follow the temporal evolution of a
parameter
 A tool that allows you to assess whether a process is in a state
of statistical control (values within limits)
 A tool that allows you to evaluate whether a process is not in a
statistical control state (values outside the limits)
7. CONTROL CHART

◼ How do you quantitatively determine whether a

process is in a statistical control state?
 The time course of the process average is analyzed
 The time course of the standard deviation of the process (or of
a parameter related to it) is analyzed
7. CONTROL CHART

◼ Example: 30
Upper Control Limit

29

28
Lower Control Limit

2 Upper Control Limit

0 10 20 30 40
1

20 days 0
5 data per day Lower Control Limit
7. CONTROL CHART

◼ Types:
1. Continuous Data Charts
 𝑋ത − 𝑅 charts (measurement-based data)
 Average values and excursions are displayed

2. Discrete Data Cards

 Control charts for defective element percentage (p)
 Control charts for the number of defective elements (pn)
 Control charts for the number of defects per unit of
product (u)
7. CONTROL CHART

◼ Use (1):
◼ Process Analysis
 Instrument used for a defined time
 Requires at least 100 data points for a total of 20-25
subgroups
 Never deliberately change the operating conditions
 Find out if the process is or not under statistical control
 Identify special causes responsible for a process that is
not under statistical control
 Remove special causes to bring the process under
statistical control
 Define control chart limits and therefore the natural
variability of the process
7. CONTROL CHART

◼ Use (2):
◼ Process Control
 After analysis, process control begins
 Comparison of natural limits and process behavior

 If the points remain within the boundaries, the process

remains under control
 If the points go out of bounds, it is likely that there has
been a change in the factors influencing the process
 Investigation of the cause

 Corrective Action
7. CONTROL CHART

◼ Use (3):
◼ Process Control
 Conclusions drawn from the Control Charts are valid at less
than a degree of confidence
  : Probability of Error of the First Kind (False Positive,
False Alarm)
  : Probability of Error of the Second Kind (False Negative,
No Alarm)
 The selection of control limits corresponds to a compromise
solution between the two types of error.
7. CONTROL CHART

◼ Use (4):
◼ Some parameters
 Central Line (CL), Upper Control Limit (UCL), Lower
Control Limit (LCL), Upper Warning Limit (UWL), Lower
Warning Limit (LWL)
 Average Run Length

.
7. CONTROL CHART

◼ Construction of the 𝑋ത − 𝑅 chart (1):

 Show trend of the mean value 𝑋ത
 Show the progress of the excursion/dispersion R

 It is used for continuous values

 It consists of two graphs together:

▪ Top highlights changes over time of the process

mean 𝑋ത
▪ Lower shows the trend of dispersion
 Connect two aspects of the process
7. CONTROL CHART

◼ Construction of the 𝑋ത − 𝑅 chart (2):

Data collection and subgroup formation (at least 100
data points and approx. 25 subgroups)
 Subgroup conditions: Constant or presumably constant
operating conditions
 Data within the subgroup are homogeneous

 The elements are equal in number (N) in all subgroups

and generally range from 3 to 10
 The total number of subgroups is called K

 Example: K=25, N=4

7. CONTROL CHART

◼ Construction of the 𝑋ത − 𝑅 chart (3):

Sort data into tables and process:
◼ Sort data into tables
◼ Calculate the data
Average 𝑋𝑖 of the data 𝑋𝑖,1 … 𝑋𝑖,𝑁 belonging to the i-th sub-group:
𝑁
1
𝑋𝑖 = ෍ 𝑋𝑖,𝑛
𝑁
𝑛=1

𝑋1 + 𝑋1 + 𝑋3 + ⋯ + 𝑋𝑘
𝑋ത =
𝑘
7. CONTROL CHART

◼ Construction of the 𝑋ത − 𝑅 chart (3):

Sort data into tables and process:
◼ Sort data into tables
◼ Calculate the data

Average Excursion :
for the i-th group, evaluate 𝑅𝑖 = 𝑋𝑚𝑎𝑥,𝑖 − 𝑋𝑚𝑖𝑛,𝑖

𝑋𝑚𝑎𝑥,𝑖 = 𝑚𝑎𝑥 𝑋𝑖,1 … 𝑋𝑖,𝑁

𝑋𝑚𝑖𝑛,𝑖 = 𝑚𝑖𝑛 𝑋𝑖,1 … 𝑋𝑖,𝑁

𝑅1 +𝑅1 +𝑅3 +⋯+𝑅𝑘

𝑅= 𝑘
7. CONTROL CHART

◼ Construction of the 𝑋ത − 𝑅 chart (4):

Calculation of control limits with the appropriate tabulated
coefficients A2, D4, D3
ത the following formulas are used:
For control chart 𝑋,
1) Centerline 𝑋ത
2) Lower Control Limit LCL = 𝑋ത − 𝐴2 𝑅
3) Upper Control Limit UCL = 𝑋ത + 𝐴2 𝑅
For the control chart R, the following formulas are used:
1) Centerline R
2) Lower Control Limit LCL = 𝐷3 𝑅
3) Upper Control Limit UCL = 𝐷4 𝑅
7. CONTROL CHART
◼ Construction of the 𝑋ത − 𝑅 chart (5):
Tabulated coefficients
Numerousness Constants for Constants
of the n sub- the X Chart for the R
groups Chart
A2 d2 D3 D4
1 2.660 1.128 0.000 3.276
2 1.880 1.128 0.000 3.276
3 1.023 1.693 0.000 2.575
4 0.729 2.059 0.000 2.282
5 0.577 2.326 0.000 2.115
6 0.483 2.534 0.000 2.004
7 0.419 2.704 0.076 1.924
8 0.373 2.847 0.136 1.864
9 0.337 2.970 0.184 1.816
10 0.308 3.078 0.223 1.777
7. CONTROL CHART

◼ Construction of the 𝑋ത − 𝑅 chart (6):

Chart Construction:
◼ Draw the axes, centerline, and boundaries of both the 𝑋ത and
R plots to an appropriate scale
◼ By convention the centerline is continuous, the boundary line
is dashed
◼ The numerical values of the limits are indicated

◼ The X and R values of the subgroups are reported in

sequence
◼ By convention, the values of X are indicated by a dot and
those of R by a cross
◼ Out-of-bounds values are highlighted

◼ The distance between one point and another along the

horizontal axis must be proportionate to the scale used for
the vertical axis
 7. CONTROL CHART

◼ Construction of the 𝑋ത − 𝑅 chart (7):

Plot: UCL = 13.72
14

X = 12.94
13
𝑋ത
LCL = 12.161

12

UCL = 2.86
3

2 R = 1.35
R
1

5 10 15
0 Sub-groups
7. CONTROL CHART

◼ Interpretation of the 𝑋ത − 𝑅 chart (1):

1. One point equal to or exceeding the Control Limit

Chart Meaning Possible causes (some examples)

Shift of the process mean 1. Operator error (distraction)

X 2. Machine defect

Variation of the process 1. Operator error (distraction)

R dispersion 2. Machine defect
7. CONTROL CHART
◼ Interpretation of the 𝑋ത − 𝑅 chart (2):
2. Consecutive series of points on the same side
7 consecutive 10 out of 11 12 out of 14

Chart Meaning Possible causes (some examples)

Shift of the process mean 1. Different raw materials

X 2. Inexperienced new operator
3. A defect in a part of the machine
4. A change in the set-up of the machine

Significantly increased or 1. Inexperienced operator

R decreased dispersion of values 2. Sudden variation in the play of a mechanism
3. Great variation in material
7. CONTROL CHART

◼ Interpretation of the 𝑋ത − 𝑅 chart (3):

3. Alternating points near the two boundaries

Chart Meaning Possible causes (some examples)

The dispersion of the average 1. Big difference in the quality of materials

X values of the process has 2. Two or more machines on the same control
changed chart
Variable dispersion amplitude 1. Different operators use the same control
R over a wider range card
2. Raw material from different suppliers
7. CONTROL CHART

◼ Interpretation of the 𝑋ത − 𝑅 chart (4):

4. Sequence of increasing or decreasing points

Chart Meaning Possible causes (some examples)

The average of the process 1. Tool wear

X follows a continuous trend of 2. Gradual deterioration of a part
increase or decrease 3. Gradual change in temperature or humidity
4. Change in viscosity in a chemical process
Dispersion amplitude follows 1. Improvement in operator skill
R a continuous trend of 2. Decrease in operator skill due to fatigue
increasing or decreasing 3. Improvement (or deterioration) in material
homogeneity
7. CONTROL CHART

◼ Interpretation of the 𝑋ത − 𝑅 chart (5):

5. Cyclic sequence

Chart Meaning Possible causes (some examples)

The values of the process 1. Seasonal Effects of Materials

X average have a periodic 2. Periodic rotation of operators
trend 3. Periodic effects of temperature and humidity
The values of the dispersion 1. Operator fatigue and "rest" resulting from intervals
R have a periodic trend 2. Lubrication cycles
7. CONTROL CHART

◼ Interpretation of the 𝑋ത − 𝑅 chart (6):

6. Shift of the process mean

Chart Meaning Possible causes (some examples)

Shift of the process mean 1. Different raw materials

X 2. Inexperienced new operator
3. A defect in a part of the machine
4. A change in the set-up of the machine
Significantly increased or 1. Inexperienced operator
R decreased dispersion of 2. Sudden variation in the play of a mechanism
values 3. Great variation in material
7. CONTROL CHART

◼ Discrete Data Control Chart:

 p Chart: shows the percentage of non-conforming
elements
 pn Chart: shows the number of non-conforming
elements
7. CONTROL CHART

◼ Construction of the p-Chart (1):

a. Collect data into a set of subgroups that should be at least
20-25
b. Sort the data into tables to be processed and calculate:
 p-value of each subgroup

𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑛𝑜𝑛 𝑐𝑜𝑚𝑝𝑙𝑖𝑎𝑛𝑡 𝑒𝑙𝑒𝑚𝑒𝑛𝑡𝑠

𝑝= ∙ 100
𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑒𝑙𝑒𝑚𝑒𝑛𝑡𝑠 𝑓𝑜𝑟𝑚𝑖𝑛𝑔 𝑡ℎ𝑒 𝑠𝑢𝑏𝑔𝑟𝑜𝑢𝑝

 p-value of the non-compliance of all the elements examined

𝑡𝑜𝑡𝑎𝑙 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑛𝑜𝑛 𝑐𝑜𝑚𝑝𝑙𝑖𝑎𝑛𝑡 𝑒𝑙𝑒𝑚𝑒𝑛𝑡𝑠

𝑝= ∙ 100
𝑡𝑜𝑡𝑎𝑙 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑖𝑛𝑠𝑝𝑒𝑐𝑡𝑒𝑑 𝑒𝑙𝑒𝑚𝑒𝑛𝑡𝑠
7. CONTROL CHART

◼ Construction of the p-Chart (2):

c. Calculation of limits (three possible cases):
1) Subgroups with constant numerosity N
3 𝑝(100 − 𝑝)
𝑈𝐶𝐿 = 𝑝 +
𝑁
3 𝑝(100 − 𝑝)
𝐿𝐶𝐿 = 𝑝 −
𝑁

2) Subgroups with quasi constant numerosity

3 𝑝(100 − 𝑝)
𝑈𝐶𝐿 = 𝑝 +
𝑛
3 𝑝(100 − 𝑝)
𝐿𝐶𝐿 = 𝑝 −
𝑛

𝑁1 +𝑁2 +𝑁3 +⋯+𝑁𝑘

where 𝑛= is the average sub-group numerosity
𝑘
7. CONTROL CHART

◼ Construction of the p-Chart (3):

c. Calculation of limits (three possible cases):
3. Sub-groups with variable numerosity
3 𝑝(100 − 𝑝)
𝑈𝐶𝐿 = 𝑝 +
𝑛𝑖
3 𝑝(100 − 𝑝)
𝐿𝐶𝐿 = 𝑝 −
𝑛𝑖
Where 𝑛𝑖 is the numerosity of the i-th sub-group
Graph construction
◼ Decide on the appropriate scale of representation
◼ Report Centerline and Control Limits
◼ Report the individual p-values
7. CONTROL CHART

◼ Interpretation of the p and pn Charts:

▪ They do not reflect any characteristics of the process
deployment
▪ No indication of mean, shape and dispersion can be
derived
▪ Indications only on the distribution mode between
compliant and non-compliant elements
7. CONTROL CHART

◼ Considerations for the use of cards:

 Decide what is the problem you want to address
 Choose the right type of control paper
 Construct the Control Chart for Process Analysis
 Build the Control Chart for Process Control
 Check the process
 Redefine control boundaries
The PDCA METHOD

◼ Introduction:
 “Control" for the CWQC
◼ Maintenance
◼ Improvement
 PDCA methodological approach to ensure effective
control
The PDCA METHOD

◼ 4 Steps:

I. P = Plan = thorough planning

II. D = Do = do what you decided in the plan phase
III. C = Check = check the results against what you
have planned
IV. A = Action = decide to keep or correct
The PDCA METHOD

◼ Three sub-cycles:
1. Maintenance cycle

yes
CHECK ACTION:
MAINTAIN AS IT IS

FOLLOW-UP
The PDCA METHOD
◼ Three sub-cycles :
2. Corrective action cycle

PLAN DO CHECK

ACTION:
REMEDY NO

ACTION:
RECURRENT
PREVENTION
The PDCA METHOD
◼ Three sub-cycles :
3. Improvement cycle

PLAN DO CHECK ACTION:

MAINTAIN AS IT IS

FOLLOW - UP

IMPROVEMENT
The PDCA METHOD

◼ Improvement project (1):

 PLAN phase
1. Identify possible themes
2. Select your theme
3. Define your goals
4. Observe and document the real situation
5. Analyze the current situation
6. Identify possible causes
7. Determine the Actual Causes
8. Determine change actions
The PDCA METHOD

◼ Improvement project (2):

 DO phase
1.Training of those who have to perform the actions
2.Implementation of countermeasures
3.Verification of the application of countermeasures
4.Recording of work done and results obtained
The PDCA METHOD

◼ Improvement project (3):

 CHECK phase
1.Definition of the parameters to be checked
2.Execution of verification and data logging
3.Comparison of the results with the initial situation
and with the set objectives
4.Assessment of whether or not the set results and
objectives have been achieved
The PDCA METHOD

◼ Improvement project (4):

 ACTION phase
1.Implementation of different activities depending on
the set objective
2.Standardization of countermeasures
3.Continuous verification of the effectiveness of
countermeasures
REF
1) https://www.conceptdraw.com/How-To-Guide/flowchart-definition
2) https://www.juran.com/blog/a-guide-to-the-pareto-principle-80-20-rule-pareto-analysis/
3) https://qctraininginc.com/7-quality-tools/
4) https://www.managementstudyguide.com/choosing-correct-control-chart-continuous-
data.htm
5) https://econoshift.com/en/types-of-control-chart-2/