Statistical Process Control SPC: Quality Management in The Bosch Group - Technical Statistics

2021-02-10 - SOCOS
Quality Management in the Bosch Group | Technical Statistics
7. Statistical Process Control SPC
•••••••••
•••••••••
2021-02-10 - SOCOS
•••••••••
•••••••••
Booklet No. 7 ― Statistical Process Control
Quality Management in the Bosch Group

Technical Statistics
Booklet No. 7
2021-02-10 - SOCOS
Statistical Process Control

SPC
Edition 11.2020
© Robert Bosch GmbH | 11.2020
•••••••••
•••••••••
5th Edition, November 2020
1st Edition, 09.1987
All minimum requirements specified in this booklet for capability and performance criteria correspond
to the status at the time of printing (issue date). [CDQ 0301] is relevant for the current definition.
2021-02-10 - SOCOS
•••••••••
•••••••••
Table of Contents
1 Introduction ..................................................................................................................................... 5
2 Quality Control ― Application of the Control Chart ........................................................................ 7
2.1 Basic Rules ............................................................................................................................. 7
2.2 Necessary Documents............................................................................................................ 8
2.3 Short Control Loop ― Directly at the Machine ..................................................................... 8
2.4 Reaction Plan ......................................................................................................................... 9
2.5 Further Measures ................................................................................................................ 11
2.5.1 Sorting Inspection ................................................................................................... 11
2.5.2 Decision about the sorted Parts ............................................................................. 11
2021-02-10 - SOCOS
2.5.3 Risks in Connection with Rework and Repair Processes ........................................ 11

2.6 Long Control Loop ― Long-Term Analysis ........................................................................... 12
2.6.1 Long-Term Monitoring ........................................................................................... 12
2.6.2 Long-term Analysis ― Organizational Questions ................................................... 12
2.6.3 Long-Term Evaluation ― Calculation, Representation, Documentation ............... 13
2.6.4 Communication, Implementation and Pursuit of Measures .................................. 13
2.6.5 Plant-specific Regulations and Solutions ................................................................ 14
2.6.6 Software Support.................................................................................................... 14
3 Planning.......................................................................................................................................... 15
3.1 Inspection Strategy .............................................................................................................. 15
3.2 Definition of SPC Characteristics.......................................................................................... 15
3.3 Definition of the Inspection Method ................................................................................... 16
3.4 Capability of Production Equipment and Processes ............................................................ 17
3.5 Types of Characteristics and Control Charts ........................................................................ 18
3.5.1 Selection of the Control Chart ................................................................................ 19
3.6 Sample Size .......................................................................................................................... 19
3.6.1 Properties of the Different Types of Control Charts .............................................. 20
3.7 Defining the Sampling Interval ............................................................................................ 21
4 Preparation and Use of Control Charts .......................................................................................... 23
4.1 Preparation of the Control Chart ......................................................................................... 23
4.2 Use of Control Charts ........................................................................................................... 23
4.3 Evaluation and Control Criteria ........................................................................................... 24
4.3.1 Control criteria........................................................................................................ 24
4.3.2 Which comparisons can be made? ......................................................................... 25
4.4 Documentation .................................................................................................................... 26
5 Determining Statistical Distribution Parameters ........................................................................... 26
5.1 Pre-Production Run.............................................................................................................. 26
5.2 Disturbances ........................................................................................................................ 26
5.3 Statistical Calculations ......................................................................................................... 26
© Robert Bosch GmbH | 11.2020 3
•••••••••
•••••••••
6 Formulas for Determining the Distribution Parameters ................................................................ 27

6.1 Process Average ................................................................................................................... 27
6.2 Process Variation ................................................................................................................. 27
7 Process-related Control Limits ....................................................................................................... 28
7.1 Natural Control Limits for Stable Processes ........................................................................ 29
7.1.1 Control Limits for Location Control Charts ............................................................. 29
7.1.2 Control charts with moving averages ..................................................................... 31
7.1.3 Control Limits for Variation Control Charts ............................................................ 33
7.2 Control Limits for Processes with Systematic Changes in the Average ............................... 34
8 Tolerance-Related Control Limits ― Acceptance Control Chart.................................................... 35
9 Tables ............................................................................................................................................. 36
2021-02-10 - SOCOS
10 Example of an Event Code for Mechanically Processed Parts ....................................................... 37

11 Reaction Catalog ............................................................................................................................ 38
12 Example of an 𝒙 − 𝒔 Chart ............................................................................................................. 39
13 Responsiveness of a Control Chart and Type-1 Error .................................................................... 40
14 Reviews and Maturity Assessments ............................................................................................... 41
List of Symbols ....................................................................................................................................... 42
Terms and Definitions ........................................................................................................................... 45
Literature ............................................................................................................................................... 57
Index ...................................................................................................................................................... 59
•••••••••
•••••••••
1 Introduction
Statistical Process Control (SPC) is a procedure for open or closed loop control of manufacturing
processes, based on statistical methods.
Random samples of parts are taken from the manufacturing process according to process-specific
sampling rules. Their characteristics are measured and entered in control charts.
According to [CDQ 0301], the use of the Solara® / qs-STAT® / procella® / O-QIS® software package is
prescribed within Bosch. It calculates capability and performance indices and operates control charts
in accordance with the used evaluation strategy.
Statistical indicators are calculated from the measurements and used to assess the current status of
the process. If necessary, the process is corrected with suitable actions. Statistical principles must be
observed when taking random samples.
Corrective actions Take sample and

2021-02-10 - SOCOS
measure characteristic
- General improvement
- Ongoing care Maintain and evaluate quality
control chart
Shortest possible
Production process control loop
influenced by:
Man Machine Material
Method Environment Delivery
Fig. 1.1: Illustration of the SPC control loop
The control chart method was developed by Walter Andrew Shewhart (1891-1967) in the 1920´s and
described in detail in his book “Economic Control of Quality of Manufactured Product” [Shew 1931].
SPC is used at RB in a common manner in all divisions. The procedure is defined in [CD 00301] in
agreement with all business divisions.
Current questions on use of SPC and related topics are discussed in a work group (Functional Council).
Results that are helpful for daily work and of general interest can be summarized and published as QA
Information.
SPC is an application of inductive statistics. Not all parts have been measured, as would be the case for
100% inspection. A small set of data, the random sample measurements, is used to estimate
parameters of the entire population.
In order to correctly interpret results, we have to know which mathematical model to use, where its
limits are and to what extent it can be used for practical reasons, even if it differs from the real situation.
We differentiate between discrete (countable) and continuous (measurable) characteristics. Control

charts can be used for both types of characteristics.
Statistical process control is based on the concept that many inputs can influence a process. The “5 M´s”
– man, machine, material, milieu, method – are the primary groups of inputs. Each “M” can be
subdivided, e.g. milieu in temperature, humidity, vibration, contamination, lighting, ....
•••••••••
•••••••••
Despite careful process control, uncontrolled, random effects of several inputs cause deviation of
actual characteristic values from their targets (usually the middle of the tolerance range).
The random effects of several inputs ideally result in a normal distribution for the characteristic. Many
situations can be well described with a normal distribution for SPC.
A normal distribution is characterized with two parameters, the mean 𝜇 and the standard deviation 𝜎.
Fig. 1.2
The graph of the density function of a normal

2021-02-10 - SOCOS
distribution is the typical bell-shaped curve,

with inflection points at 𝜇 − 𝜎 and 𝜇 + 𝜎.
µ - 3 µ - 2 µ - 1 µ µ + 1 µ + 2 µ + 3
In SPC, the parameters 𝜇 and 𝜎 of the population are estimated based on random sample measure-
ments and these estimates are used to assess the current status of the process.
An essential advantage of quality control charts is the fact that the original data and usually also the
conditions under which they were determined are documented and remain available. In [Shew 1939]
Shewhart explicitely lists the advantages of the original data.
 The measurement results are represented by numerical values (and units of measurement).
 The original data are documented in the order in which they were determined.
 The conditions under which the data were determined and the measurement process are
reproducible.
Especially the chronological order of the data is an essential information which is lost during any kind
of further data compression.
Information given by Number (example)

Original data; complete information individual values 𝑥𝑖 ; 125 values
Means of subgroups 𝑥̅𝑗 25 means
Histogram classification and 7 classes
relative frequencies ℎ𝑗
Distribution model with location and variation 𝑥̅ , 𝑠 2 parameters
Process capability index 𝐶𝑝𝑘 1 index
NOTE: [Booklet 3] shows three different representations of data with increasing values, a random sequence of
values and a sawtooth-like progression, which differ only in their chronological order. The corresponding
distribution model is always the same.
•••••••••
•••••••••
2 Quality Control ― Application of the Control Chart

A prerequisite for SPC incl. regular 𝐶𝑝𝑘 verification is, according to [CDQ 0301], a successfully
completed
 suitability study of the measuring equipment,
 machine capability analysis (resp. short-term capability analysis) and
 initial long-term process capability analysis.
Capability studies are briefly discussed in Section 3.4 and in detail in [Booklet 9]. The investigation of
measurement processes is briefly adressed in Section 3.3.
It cannot be completely avoided that in this chapter also topics are touched, which go clearly beyond
the topic of this Booklet on SPC. Meant are for example
2021-02-10 - SOCOS
 responsibilities, in particular regarding the cooperation of production planning, production and

quality management.
 identification and traceability of products, as well as
 the control of nonconforming or potentially nonconforming products (e.g. also blocking)
These topics are only briefly addressed, but will not be discussed in greater depth.
2.1 Basic Rules

The application of SPC requires some basic rules which will be briefly presented here.
1. Setting within the tolerance zone

With two-sided limited tolerance zone it is sensible, to adjust the machine so that the characteristic’s
values are close to the center point (target value C). If only an upper limit (USL) is given (e.g.
concentricity), a setting to the smallest possible value is selected for reasons of economy.
2. When adjusting the machine or the tool, it is unavoidable to check continuously until the correct
centered location (according to step 1) is achieved. The inspection interval begins then with the entry
of the first measurement results (results of the first sample) in the control chart as first “OK sample”.
Since the parts, produced from the beginning of the intervention or adjustment until achievement of
the center position, can show greater deviations from the target value, they have to be kept separately
and be sorted out, if necessary.
3. After each inspection interval a sample is taken, the considered characteristic is measured and the
measurement results are entered in the control chart.
4. In case of a tool breakage or other reasons for a process intervention one must assume, that the
parts produced since the beginning of this inspection interval are potentially defective/nonconforming.
They are treated according to Section 2.3.
5. In general, after a process intervention it is necessary to proceed with steps 1, 2, and 3.
6. All correctives measures on the process (e.g. according to Chapter 10) and on the parts (Section 2.5)
are documented.
•••••••••
•••••••••
2.2 Necessary Documents

The necesssary documents near the machines and or SPC measuring stations include, e.g.
 process data sheet (information about the production process, the production equipment and
system-specific settings),
 set-up instruction (work instruction for the set-up of one or several machines),
 inspection instruction (instructions for the conduction of certain quality inspections in the
production process),
 regulation for the control of the measuring station (activities for the control of measuring
devices),
 basic rules and reaction plan (s. Sections 2.1 and 2.4)
 action catalog (s. Chapter 11, for instance)
2021-02-10 - SOCOS
2.3 Short Control Loop ― Directly at the Machine

In case of unwanted process results, appropriate measures must be taken to achieve the desired result
(control loop). Reaction plans must be available in case one or more control criteria are met. The
reaction plan must be created specifically for the process, machine or machine type.
The reaction plan must show,

 how the causes can be found and eliminated,
 how the process can be readjusted,
 what to do with the parts produced since the last sample (e.g. sorting instruction).
Thus, the short control loop is essentially about the direct correction measures on the process.
In order to be able to sort out bad parts after an intervention limit has been exceeded, it must be
possible to trace back and check the production quantity since the last sampling.
Hint: Rework and repair processes as well as sorting inspections involve further quality risks, e.g. due
to
• mix-ups of products or components incl. mixing of variants,
• damage of parts by disassembly o rhandling,
• causation of further failure modes, e.g. flashing with wrong software,
• logistics mistakes, quantity deviation, wrong packaging,
• exceedance of maximum storage periods due to violation of the “First In, First Out” (FIFO)
principle.
The creation and maintenance of reaction plans can be supported by providing plant, area and process-
specific catalogs. Within the scope of the obligatory use of Q-DAS software, the maintenance of action
plans/catalogs and the documentation of measures in the software must be ensured.
•••••••••
•••••••••
2.4 Reaction Plan

A reaction plan describes what to do when an intervention is necessary (one of the control criteria is
met). This plan must be created specifically for the process, machine or machine type. It should contain
all measures necessary to
 detect easily recognizable causes of disturbing influences and
 control the process.
The plan should also include clear instructions on what to do with the parts produced since the last
sample. It is advisable to define short designations (codes). See Chapters 10 and 11.
Reaction in SPC
2021-02-10 - SOCOS
SPC, control chart, control criteria:

1) Single value out of tolerance a
no
2) Statistic (𝑥̅ / s) beyond SPC control
Next sample
control limits criterion fulfilled?
3) optional: rule of seven
yes
b
yes SPC,
Single value out of Sort 100 % back control chart,
tolerance zone? to last sample reaction plan
no
c no Correction in SPC,
Statistic beyond case of changed control chart,
alarm limits? process reaction plan
behavior
yes
d Documentation
Further actions
Fig. 2.4: Illustration of the reaction when the control chart responds
a) Intervention necessary?
After each entry in the SPC control chart, it has to be checked whether intervention is necessary. An
"intervention" is an appropriate reaction to a statistically unusual event, i.e. one of the control criteria
listed below is met (e.g. a control limit has been exceeded).
Control criteria:
 A single value is outside the tolerance zone.
 A sample’s statistic (e.g. 𝑥̅ or 𝑠) is outside the control limits.
 Optional: An unusual sequence of points is observed ― rule of seven (i.e. seven points in a row
on one side of the average line, seven measurement values with continuously increasing or
decreasing tendency). For processes with systematic changes of the location ( 𝑥̅ ), e.g. for
processes with a trend, this rule cannot be applied.
•••••••••
•••••••••
In this case, it has to be checked whether the process behavior has actually changed, and, if this is
confirmed, after determination of the cause (according to the reaction plan) the original process state
has to be restored.
NOTE: Due to the operation characteristics of the control charts there is an error probability of about 1 % for
the exceedance of control limits, although the process behavior is unchanged (s. Chapter 13).
b) Single value outside the tolerance zone?

In case of individual values outside the tolerance range, all parts of the production lot manufactured
since the last random sample must be sorted (d).
c) Alarm limit exceeded?

Process-related control limits ( 𝐿𝐶𝐿 , 𝑈𝐶𝐿 ) are only dependent on the process variation and
2021-02-10 - SOCOS
independent of the characteristic’s tolerance (natural control limits).

They are defined so that for a process which is only affected by random causes 99 % of the means or
standard deviations lie within their control limits.
If the control limits are exceeded, it must therefore be assumed that systematic, non-random
influences (non-random causes) have an effect on the process. These influences must be corrected or
removed by appropriate measures (action plan).
Tolerance-related control limits aim at compliance with the tolerance and not at a process
improvement; therefore, they cannot be used for stabilizing and centering a process.
Tolerance-related control limits can be used to support the decision according to Section 2.2. These
limits are then called alarm limits (𝐿𝐴𝐿, 𝑈𝐴𝐿). The formula for their calculation contain the limits 𝐿𝐶𝐿
and 𝑈𝐶𝐿 (s. Chapter 8).
If the monitored statistic (e.g. 𝑥̅ ) is still within the alarm limits, sorting of the parts produced since the
last sample is not necessary. However, it has to be checked whether the process behavior has changed,
and, if this is the case, perform corresponding corrective measures.
d) Sorting inspection
All parts produced since the last sampling are 100 % inspected and sorted (s. Section 2.5).
•••••••••
•••••••••
2.5 Further Measures

As mentioned at the beginning of the chapter, some topics are touched here that go far beyond the
actual scope of this booklet. With the formulation „further measures“ those measures are meant
which cannot be allocated to the short control loop.
2.5.1 Sorting Inspection

A sorting inspection is a 100% inspection and means that products/components are with respect to
defined characteristics and separated according to the individual results (e.g. conforming / noncon-
forming). The inspection can be carried out visually or using technical means, e.g. using measurement
systems or gages.
It is self-evident, that the parts must still be accessible, if a sorting inspection is necessary due to a
quality risk.
However, the measures necessary in this context are highly dependent on where the parts are located
2021-02-10 - SOCOS
and what their status is.

The valid sorting instructions have to be applied, in particular regarding handling and identification of
the parts, equipment and staff briefing/training.
2.5.2 Decision about the sorted Parts

In the framework of a sorting inspection the parts are separated based on the inspection results. As a
further step, it must be decided what has to be done with the parts, produced since the last sample.
This can mean that these parts are
 conforming parts are further processes or deliverd,
 non-conforming parts are reworked or
 scrapped.
The measures are documented. Details are regulated in [CDQ 0503] and [CDQ 0509].
2.5.3 Risks in Connection with Rework and Repair Processes

In general, rework and repair processes as well as sorting inspections involve further quality risks, e. g.
due to
 Mix-ups of products or components incl. mixing of variants

 Damage of parts by disassembly or handling
 Causation of further failure modes
 Logistics mistakes, e.g. quantity deviation, wrong packaging
 Exceedance of maximum storage periods due to violation of the “First In, First Out” (FIFO)
principle
NOTE: In the broadest sense, a sorting inspection is rework of products with unclear status.
To avoid such risks, appropriate safeguarding measures and approvals are necessary. In particular,
effective measures must prevent
 further processing,
 unintended use, and
 the delivery
of nonconforming or incorrectly labeled products or products whose status is not identifiable
(potentially nonconforming products). See [CDQ 0503]. For this purpose, blocking, identification,
sorting and sorting-out must be clearly regulated.
•••••••••
•••••••••
2.6 Long Control Loop ― Long-Term Analysis

An SPC concept always requires the implementation of long-term analyses and measures for quality
improvement. They aim at
• achieving and maintaining capable process results,
• demonstrating the process capability, as well as
• continuously improving the processes.
For this purpose, the process behavior must be evaluated regularly to determine whether it has
changed compared to the initial situation (pre-production run) or to a previous analysis. This includes
the aspects described in the following sections.
Part of the long control loop is also the analysis of causes, i.e. the discovery of the causes of disturbing
influences that lead to an intervention according to the Sections 2.1 and 2.3 and are not easily
recognizable.
2021-02-10 - SOCOS
NOTE: Terms like analysis, tracking and evaluation are often used synonymously in everyday language. In the
context of this booklet they have slightly different meanings.
Analysis: Decomposition, systematic investigation of an issue and its causes
Monitoring: tracking, observation, surveillance, ongoing investigation, activities to find out something
Evaluation: Summary, condensation, presentation of the essentials, interpretation, evaluation of data,
derivation of measures
2.6.1 Long-Term Monitoring

The original specifications for the control chart (type of chart, sampling interval and size, distribution
model, control limits) reflect the understanding of the process behavior at the time of the pre-
production run which is limited due to the small data basis of e.g. 125 values.
Within the framework of long-term monitoring, the conclusions regarding the process behavior that
became apparent in the initial capability study, can be checked using the larger database which is then
usually available.
NOTE: Depending on the inspection plan, there may be exceptional situations where even after a longer period
of time only a rather small database is available.
2.6.2 Long-term Analysis ― Organizational Questions

• Who evaluates the data (completed control chart or after defined time)?
• What does the analysis refer to (e.g. product, process, machine)?
• How often is the evaluation performed (e.g. monthly)?
• Which indices are calculated (𝐶𝑝 , 𝐶𝑝𝑘 or 𝑃𝑝 , 𝑃𝑝𝑘 )?
• Which process-specific data base is used (e.g. per quantity or per time unit)?
• Which method, which tool is used to represent the analysis results as 𝐶𝑝𝑘 overview on a monthly
basis (e.g. M-QIS module of the Q-DAS-Software)?
• Which detailed analysis of the specific process situation is important, e.g. regarding
- the average process location; systematic changes of the process location (trend, stepwise
variation, between-batch variation),
- the distribution model, indices used,
- tool changes, maintenance activities?
•••••••••
•••••••••
2.6.3 Long-Term Evaluation ― Calculation, Representation, Documentation

The following values can be calculated and documented:
 𝑥̅̅ (bzw. 𝑥̃̅ ) Section 6.1
 𝑠̅ (bzw. 𝑅̅ ) Section 6.2
 𝜎̂ Section 6.2
 𝐶𝑝 und 𝐶𝑝𝑘 bzw. s [Booklet 9]
 𝑃𝑝 und 𝑃𝑝𝑘 und s [Booklet 9]
The corresponding requirements (e.g. defined minimum values) have to be observed and it must be
checked whether the distribution parameters and capability indices 𝐶𝑝 and 𝐶𝑝𝑘 or process
performance indices 𝑃𝑝 and 𝑃𝑝𝑘 have changed compared with the pre-production rund or the previous
control charts. Procedures for the calculation of these indices are described in [Booklet 9].
It is recommended to regularly carry out a higher-level evaluation regarding. 𝐶𝑝𝑘 or 𝑃𝑝𝑘 .
2021-02-10 - SOCOS
2.6.4 Communication, Implementation and Pursuit of Measures

The communication, implementation and tracking of measures need organizational regulations which
cover for example the following aspects:
• Function-specific Information and communication (e.g. specific for Production, Production
Planning and Quality Management)
- How are the results provided (e.g. via automatic e-mailing, WorkOn, BGN page)?
- Who gets which information?
- How often is the information submitted?
- Which format is used (e.g. overview of all processes or detailed reports on individual
characteristics)?
• Regular check of the long-term analysis results, decision on measures and their tracking, e.g.
- dealing with non-capable processes
- recalculation of control limits
- review/update of inspection and process planning
• Retention of documented information in accordance with valid central procedures (e.g.
CD 02981)
• Concept for storage and archiving of data and reports in accordance to valid central procedures.
Example:
Management Production Planner Production

Frequency monthly weekly weekly
Content  Histogram of process  Processes relevant  Evaluation of the
indices (capable, for the planner: value progression for
conditionally capable, not cababilities calcu- the individual
capable) lated for the last 4 characteristics
weeks
 For non-capable processes
presentation of the indi-
ces from the last six
months
•••••••••
•••••••••
2.6.5 Plant-specific Regulations and Solutions

Responsibilities
The responsibility for long-term analyses usually lies with Production. Quality Management supports
e.g. regarding the calculations and representation of the evaluation results. It can be sensible to
implement plant-specific solutions which take the local conditions into account. [CD 00301] describes
various roles with tasks regarding
• management of characteristics
• classification of characteristics (incl. special characteristics)
• integrated quality planning
• software applications (software key-user)
Recalculation of Control Limits

Essential changes on the process like technical improvements, maintenance/repair, internal / external
2021-02-10 - SOCOS
relocation or design changes in the produc can have effects on the

• process behavior,
• middle position and variation of the process,
• tolerance definitions.
Then it may be necessary to repeat capability studies and recalculate control limits. Otherwise the
control limits must be kept constant.
Dealing with non-capable Processes

Ensure that each non-capable production process only delivers parts that conform to all specifications.
This can be achieved either by an inspection (e.g. 100% inspection) using a capable measurement
process or other adequate measures (e.g. functional testing during successive process steps, risk
analysis, decision/approval of management), cf. [CDQ 0301] and [IATF 16949].
Review of Inspection Planning / Inspection Strategy

The long control loop also includes the necessity for a regular check/review and adaption of the
inspection planning, e.g. regarding the questions
• Should/must SPC characteristics be added or removed?
• Is it sensible to modify the sample interval or sample size?
• Is it sensible to replace the SPC rules for a characteristic by other measures, e.g. implementation
of 100% inspection?
2.6.6 Software Support

According to [CD 00301], Bosch-internally the current version of the software package Solara® / qs-
STAT® / procella® / O-QIS® is used to evaluate capability studies and maintain control charts. These
modules allow you to maintain control charts, calculate capability indices according to the defined
Bosch evaluation strategy and to evaluate and monitor such data in the long term. The Bosch evalu-
ation strategy fulfills the minimum requirements for statistical calculations described in [CD 00301].
For data transfer and storage the use of the „Automotive Quality Data Exchange Format“ AQDEF is
mandatory.
•••••••••
•••••••••
3 Planning
The planning takes place within the framework of “Management of Characteristics“ according to
CD 00301. Responsibilities are defined there.
3.1 Inspection Strategy

Statististical process control of a characteristic with regular Cpk verification is one of the preventive
inspection strategies in series production.
Reactive measures for failure detection include the application of an acceptance control chart with
regular 𝐶𝑝𝑘 verification and sampling inspection according to a sampling plan, for instance.
The definition of the inspection strategy requires, among others, the analysis of the cause-effect
relationships between process characteristics and product characteristics, taking into account distur-
2021-02-10 - SOCOS
bance / noise and control variables. Knowledge of these relationships is a prerequisite for the process
to be influenced (controlled) with regard to the inspected characteristic (product characteristic).
Usually, production equipment can directly be influenced so that the inspected characteristic (product
characteristic) can be changed in the desired way (short control loop).
If a characteristic cannot be measured directly, it may be possible to determine a substitute inspection
characteristic that is known to be related to it.
3.2 Definition of SPC Characteristics

In the framework of “Management of Characteristics“ a classification is made. Product characteristics
are allocated to classes A, B, C. Essential decision criterion for this allocation is, among others,
robustness.
 Robustness is understood as the capability of a system to maintain its function even with
variations in material properties, manufacturing parameters, as well as ambient, operational and
usage conditions.
 Functional robustness: A product characteristic is “not functionally robust”, if it solely and
immediately leads to a failure when the limits of the characteristic’s tolerance interval are
exceeded.
 Robustness regarding the manufacturing method: A product characteristics is “not manufacturing
method robust”, if it is failure sensitive with respect to the chosen manufacturing method. That
means that there is a high probability of exceeding one or both limits of the characteristic’s
tolerance interval due to the chosen manufacturing method.
In case of an existing process it may prove necessary to add new SPC characteristics. However, there
can be reasons (e.g. change of the production method or introduction of 100% inspection) to substitute
the previous SPC control by other measures.
SPC characteristics can be product or process characteristics.

NOTE 1: The definitions of robustness are only applicable to product characteristics.
NOTE 2: In general, process characteristics are variables that cannot be easily measured and monitored /
controlled "in situ" by sensors, e.g. concentrations of substances in solutions, pH values or proportions of carbon
and accompanying elements in molten steel.
NOTE 3: According to [Booklet 10], Procedure 5, a control chart is also useful for monitoring the long-term
behavior of a measurement process. The evaluation of the measuring system’s stability allows the proof of
consistently correct measurement results. However, a measurement process is not controllable in terms of an
SPC process.
•••••••••
•••••••••
3.3 Definition of the Inspection Method

This planning step includes, among others, the definition of:
 inspection type
 location and personnel
 measurement and test procedure
 inspection timing / sequence and quantity-related definitions
The suitability of the IMT equipment for an inspection task has to be proven by a capability study. In
special cases, a measurement process with known measurement uncertainty can be applied.
• In case of recurrent measurements of the same type, e.g. in the production flow, measurement
process capabilities according to [Booklet 10] are preferably determined.
• In case of varying measuring tasks, e.g. in development and test departments, measurement
2021-02-10 - SOCOS
uncertainties according to [Booklet 8] are preferably determined.

Measuring unit and reference value must correspond to the quantities selected for the measurement
process.
NOTE: IMT equipment according to the definition in “Terms and Definitions” is subject to control. IMT equipment
serves for the proof of conformity in the framework of “Management of Characteristics”, for instance.
Software, used for the creation of measurement results is also subject to this mandatory control. Such software
must be validated in appropriate form.
•••••••••
•••••••••
3.4 Capability of Production Equipment and Processes

Process capability studies must be performed for new or modified production processes (machines)
([IATF 16949], [CD 00301] and [Booklet 9]). Also after extensive repair.
In short-term studies (studies of machine capability), characteristics of product parts are evaluated
which have been manufactured in a continuous production run in an uninterrupted sequence, so that
possibly only the influence of the machine is active.
In contrast, the parts to be measured originate in studies of long-term capability (process capability)
from a larger, more representative period for the series production, so that possibly all influences on
the process, which are to be expected, take effect.
Particularly during series rump-up, there are frequently not enough product parts available nor can
enough parts be taken out of the manufacturing process over a sufficient period of time. Nevertheless,
as an alternative, or in addition to the machine capability, at least a preliminary conclusion about the
2021-02-10 - SOCOS
expected manufacturing process capability can be demanded (see “Initial Process Capability” [AIAG
PPAP] and “Preliminary process capability” [VDA-4]). In this case, a short-term study is conducted,
which can differ from the long-term study in the following points.
 Type of sampling: The parts to be studied can be taken out of the manufacturing process in
shorter intervals, if necessary in extreme cases, one immediately after the other.
 Number of parts: It is permissible if there are not sufficient parts available, to take less than the
required 125 parts for the long-term study.
 Limits for capability and performance indices: The increased limit of 1.67 applies if there are
more than 125 parts. If there are less than 125 parts, the limit is raised depending on the number
of parts at the same value as in the long-term study with reduced quantities.
 Designation of statistical indices: Capability indexes are designated with 𝐶𝑝−𝑆𝑇 und 𝐶𝑝𝑘−𝑆𝑇 and
performance indexes with 𝑃𝑝−𝑆𝑇 und 𝑃𝑝𝑘−𝑆𝑇 (short-term).
NOTE: The general definition of SPC does not presume process capability. However, if capability is not given,
then additional actions are necessary to ensure that the quality requirements for manufactured products are
fulfilled.
•••••••••
•••••••••
3.5 Types of Characteristics and Control Charts

This booklet only deals with continuous characteristics. Refer to Chapter “Terms and Definitions” for
these and other types of characteristics. In measurement technology, physical variables are defined as
continuous characteristics.
According to [CDQ 0301], the use of the Solara® / qs-STAT® / procella® / O-QIS® software package is
prescribed within Bosch. It calculates capability and performance indices and operates control charts
in accordance with the used evaluation strategy.
A control chart consists of a chart-like grid for entering numerical data from measured samples and a
diagram to visualize the statistical indices for the process location and variation calculated from the
data.
Modern SPC software offers the possibility to
 transfer measured data directly to the SPC software,

2021-02-10 - SOCOS
generate the graphical representations automatically and

 be configured according to practical aspects.
Control charts for continuous characteristics

Control charts for Control charts for
process location process variation
𝑥̅ chart 𝑠 chart
𝑥̃ chart 𝑅 chart
𝑥 chart
If a characteristic can be measured, then a control chart for continuous characteristics must be used.
Normally the 𝑥̅ − 𝑠 chart with sample size 𝑛 = 5 is used.
Fig. 3.5: Types of characteristics

Quantitative characteristic
Countable characteristics are special discrete

characteristics. Their resulting values are
Continuous Discrete “counts”. For example, the number of "bad"
characteristic characteristic parts (non OK parts) resulting from an
inspection with a limit gage is a countable
Measured value Counted value
characteristic.
NOTE: In the past, also control charts for discrete characteristics have been used, where the number or
proportion of nonconforming units (np chart, p chart) or the number or proportion of defects per unit (c chart,
u chart) is used as the characteristic to be controlled.
Due to the following disadvantages, control charts for discrete characteristics are no longer up-to-date:
 The existence of defects is a prerequisite for the functioning of these types of charts. Thus, they stabilize
rather than eliminate the undesirable condition.
 For small error proportions, unrealistic sample sizes of several hundred parts are required.
 Due to the large random variation ranges of the characteristics, the response sensitivity is limited, i.e.
until significant changes are detected, more entries are required than with a control chart for continuous
characteristics.
•••••••••
•••••••••
3.5.1 Selection of the Control Chart

Independent of the type of process, the process-related and also the tolerance-related control limits
should be calculated. Knowledge and comparison of both types of control limits is important for correct
selection of the type of chart (refer to the explanation).
Selection of the control chart is performed with the flow chart below.
* For explantion: If tolerance-related control

Selection of the control chart limits were used in this situation, then there
would be more "false alarms", i.e. the control
chart indicating a false need for action.
The attempt to control the process would then
jsut lead to "overcontrol" with still larger
variation.
Are the
2021-02-10 - SOCOS
process-related control no*

limits (per Chapt. 7) within the
tolerance-related control limits
(per Chapt. 8)?
yes
The process cannot be treated as a

capable process.
Is the use of no
tolerance-related control limits
mandated by technical
reasons?
A Shewhart chart with process-

related control limits (per Chapt. 7)
yes can be used parallel to
process improvement actions.
Acceptance Control Chart Shewhart Control Chart Quality assurance with 100 %
inspection or actions to assure
Use of tolerance-related Use of process-related proper function in manufacturing is
control limits control limits absolutely necessary.
(per Chapter 8) (per Chapter 7)
Fig. 3.5.1: Flow chart for selecting the quality control chart
When using the Shewhart chart, tolerance-related control limits can be used to assist in decision
making in the framework of a sorting procedure (refer to Section 2.4). These limits are then referred
to as alarm limits.
It is not permissible to draw tolerance-related control limits (alarm limits) as lines in Shewhart control
charts.
3.6 Sample Size

The appropriate sample size is a compromise between process performance, desired accuracy of the
selected control chart (type I and type II errors, operation characteristic) and the need for an
acceptable amount of testing.
Normally 𝑛 = 5 is selected. Smaller samples should only be selected if absolutely necessary.
•••••••••
•••••••••
3.6.1 Properties of the Different Types of Control Charts
Properties
Average chart  For the same sample size, better sensitivity or selectivity than the 𝑥̃ − 𝑅
and the individual data charts with respect to a shift of the average
location as well as to an increase of the standard deviation 𝜎.
 For 𝑛 ≥ 4 insensitive to deviation of the population from the normal
distribution
 Preferred chart for use with computers
Chart for moving  Can also be used for destructive or complicated testing
indices
 This chart has a delayed reaction to suddenly occurring process
2021-02-10 - SOCOS
changes.
 The control limits are the same as for “normal” average charts
Average chart with  Preferred chart for processes with systematic changes of the average
extended limits
Acceptance control  Due to the tolerance-related control limits, the process is controlled
chart worse than with process-related control limits.
Individual data chart  Presentation of all individual data from the random sample, so the
chart is especially easy to understand and to fill out, lending it suitable
for simple documentation of the manufacturing process.
 No calculation is necessary to use the chart.
 The individual data chart reacts to both a shift of the average location
as well as to an increase of the standard deviation of a characteristic.
 Compared to the 𝑥̅ − 𝑠 chart and the 𝑥̃ − 𝑅 chart less sensitivity or
selectivity with respect to a shift of the average location as well as to an
increase of the standard deviation 𝜎.
 A decrease of 𝜎 is not indicated by going below a limit value.
 More sensitivity to deviations of the population from a normal
distribution than with the average chart.
 Potential confusion of users from the fact that, for example, the 𝑈𝐶𝐿
increases with larger sample sizes n.
 Evaluation, such as calculation of 𝐶𝑝𝑘 , is more complicated, because the
original data is usually not recorded numerically (e.g. only points on the
graph).
•••••••••
•••••••••
3.7 Defining the Sampling Interval

When a control chart triggers action, i.e. when the control limits are exceeded, the root cause must be
determined as described in Section 2.1, reaction to the disturbance initiated with suitable actions
(refer to the reaction catalog in Chapter 11) and a decision made on what to do with the parts produced
since the last random sample was taken. In order to limit the financial “damage” caused by potentially
necessary sorting or rework, the random sample interval ― the time between taking two random
samples ― should not be too long.
The sampling interval must be individually determined for each process and must be modified if the
process performance has permanently changed.
It is not possible to derive or justify the sampling interval from the percentage of defects. A defect level
well below 1 % cannot be detected on a practical basis with random samples. A 100 % test would be
necessary, but this is not the goal of SPC. SPC is used to detect process changes.
The following text lists a few examples of SPC criteria to be followed.
2021-02-10 - SOCOS
1. After setup, elimination of disturbances or after tooling changes or readjustment, measure

continuously (100 % or with random samples) until the process is correctly centered (the average of
several measurements/medians!). The last measurements can be used as the first random sample
for further process monitoring (and entered in the control chart).
2. Sampling intervals for ongoing process control can be defined in the following manner, selecting
the shortest interval appropriate for the process.
 Definition corresponding to the expected average frequency of disturbances (as determined

in the trial run or as is known from previous process experience).
Approximately 10 samples within this time period.
 Definition depending on specified preventive tooling changes or readjustment intervals.
Approximately 3 samples within this time period.
 Specification of tooling changes or readjustment depending on SPC samples.
Approximately 5 samples within the average tooling life or readjustment interval.
But at least once for the production quantity that can still be contained (e.g. delivery lot, transfer to
the next process, defined lots for connected production lines)!
3. Take a final sample at the end of a series, before switching to a different product type, in order
to confirm process capability until the end of the series.
Note: The inspection interval is defined based on quantities (or time periods) in a manner that detects
process changes before defects are produced. More frequent testing is necessary for unstable
processes.
•••••••••
•••••••••
First-off / last-off part inspections

IATF 16949 demands measures to ensure that the products meet the requirements after a planned or
unplanned interruption.
First-off / last-off part inspections, e.g. after set-up processes, are intended to avoid problems in the
context of a batch production or lotwise release (material mixing, incorrect set-up).
The procedure can be easily integrated into the SPC systematics by testing not only a first-off and last
off part, but a group of first and last parts with the defined „sample size“ n. It is important that these
are really the first and last parts and not, for example, any parts from the first and last palette.
It is recommended to take another sample also in the middle of the interval.
NOTE: The first parts produced after the set-up process can hardly be called a random sample. But here it is
rather a matter of proving that the process is mastered and that all products meet the requirements.
2021-02-10 - SOCOS
•••••••••
•••••••••
4 Preparation and Use of Control Charts

4.1 Preparation of the Control Chart
For the function of the control chart and the short control loop, you basically only need a
representation of the
 center point 𝐶 of the tolerance zone and the
 control limits for the statistics used, usually 𝑥̅ and s.
The scale (scaling) for 𝑥̅ (or 𝑥̃ ) and s (or R) as well as the control limits and center lines should be clearly
recognizable in the diagrams.
In most cases, however, further information is immediately visible or at least retrievable, e.g. reference
value and unit as well as the lower and upper specification limit.
2021-02-10 - SOCOS
NOTE: If, in addition to the control limits LCL and UCL, the characteristic limits LSL and USL ("tolerance limits")
are also displayed in the control chart, there is a risk that the statistics are erroneously compared with LSL and
USL (s. also Section 4.3.2).
4.2 Use of Control Charts

It must be determined who is in charge of the control chart and who has to carry out the interventions.
If the chart is maintained using a computer, the measured values may have to be entered via a
keyboard. Modern SPC software offers the possibility to transfer measurement data directly into the
SPC software and to generate the graphical representations automatically.
The steps in detail:

 Entry of the individual measurement values 𝑥𝑖
 Calculation of the statistics 𝑥̅ (or 𝑥̃ ) and s (or 𝑅)
 Representation of the statistics 𝑥̅ (or 𝑥̃ ) and s (or 𝑅 ) in the corresponding diagram and
connection of the point sequence by lines.
If an entry is outside of the control limits, then control actions per the action plan (Section 2.2) must
be initiated.
The measures carried out are noted on the chart (or reference is made to the back).
Note: In order to be able to sort out bad parts after a control limit was exceeded, it is essential to be
able to trace the production quantity manufactured since the last random sample was taken.
•••••••••
•••••••••
4.3 Evaluation and Control Criteria

After each entry, check whether one of the two chart traces indicates a reason for intervention, i.e.
whether a control criteria is fulfilled.
Intervention means, in general, adequate reaction to an event that is unusual in a statistical sense (one
of the control criteria is fulfilled). This means that we have to check whether the process behavior has
actually changed, and if this is confirmed, the original process behavior has to be re-established with
suitable actions after the causes have been determined. Then we have to decide on what to do with
the parts manufactured since the last random sample, e.g. with a sorting instruction. This means that
we must always be able to still access these parts if there is a quality risk.
In order to ensure a simple and effective procedure, a process-specific reaction plan for causes, actions
and dealing with parts is created when the control chart is being prepared. When control is necessary,
the appropriate random sample statistic is marked on the control chart and the causes, actions and
disposition of parts are documented (refer to Chapter 10) on the back of the chart (if appropriate, in
2021-02-10 - SOCOS
coded form).
4.3.1 Control criteria

 The control chart indicates the need for action, i.e. at least one random sample statistic is
outside of the control limits.
 Individual data are out of tolerance (at least one).
NOTE: The control chart does not necessarily indicate the need for action in this case.
 Optional: There is an unusual sequence of points, i.e. at least 7 points in a row on one side of
the average line, or forming a continuously increasing or decreasing trend (rule of 7, refer to the
following figure).
Fig. 4.3.1:
64,0
63,5
Illustration of the “rule of seven”
UCL
63,0
Note: The rule of seven, in the simplest
Mean x_bar
62,5
62,0
case (seven points on one side of the
average line), is based on the assumed
61,5
random process behavior and the at least
61,0
LCL approximately normal distribution of the
60,5
random sample statistics. Of course, it
60,0 cannot be used for processes with
0 5 10 15 20 25
systematic changes of the average.
Sample No.
Since for average charts and stable process behavior, each point has a 50 % probability of being above
or below the average line, independent of the previous point, the probability of seven points in a row
being on one side of the average line is (0,5)7 = 0,008 < 1 %.
This is why it is called an “unusual sequence of points”.
Use of the rule of 7 in this case (the same as comparing a random sample statistic with the control
limits) is a statistical test with a proba-bility of error (type I error) of  1 %.
For the given assumptions, the probability that at least 7 points in a row form a continuously increasing
or decreasing trend is also very low.
•••••••••
•••••••••
4.3.2 Which comparisons can be made?

2021-02-10 - SOCOS
Fig. 4.3.2: Permitted and prohibited comparisons

As the figure in Chapter 7 shows, the distribution of averages 𝑥̅ is significantly “narrower” than the
distribution of the individual data. If a random value average is outside of the control limits but still
inside of the “tolerance limits”, then possibly individual measurements were outside of the tolerance
range. This is why it is not allowed to compare statistical values with characteristic specifications 𝐿𝑆𝐿
and 𝑈𝑆𝐿 (“tolerance limits”).
For the same reason, it is not allowed to compare the difference 𝑈𝐶𝐿 − 𝐿𝐶𝐿 with the tolerance
𝑈𝑆𝐿– 𝐿𝑆𝐿. The distance between the process-related control limits is dependent on the amount of
process variation and may not be interpreted as a “reduced tolerance”.
USL
Fig. 4.3.3
UCL
This figure illustrates the proportions
for an average chart with 𝑛 = 5.
When 𝐶𝑝 = 1.33 the distance of the
control limits 𝑈𝐶𝐿 − 𝐿𝐶𝐿 corres-
LCL ponds to about 30 % of the
tolerance.
LSL
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Comparison of process-related control limits 𝐿𝐶𝐿 and 𝑈𝐶𝐿 with tolerance-related control limits (alarm
limits) per Chapter 8 allows us to assess the amount of variation and the quality capability of a process.
•••••••••
•••••••••
4.4 Documentation
Define when the evaluation results and/or overall summaries are to be submitted. Define the type and
location for archiving. Archiving durations are defined in [CD 02981], for instance.
5 Determining Statistical Distribution Parameters

5.1 Pre-Production Run
Definition of control limits requires knowledge or estimation of the distribution parameters. This is
determined with a pre-production run (trial run) with sample size and interval as specified in Sections
3.6 and 3.7. For an adequate number of parts for initial calculations, take a representative number of
unsorted parts, at least 𝑚 = 25 samples (with 𝑛 = 5 for example), yielding no fewer than 125
measured values.
2021-02-10 - SOCOS
It is important to assess the graphs of the measured values themselves, the means and the standard
deviations. Their curves can often deliver information on process performance characteristics (e.g.
trends, cyclical variations).
In general, the parameters of the resulting process distribution are already known from the capability
study (Section 3.4.).
5.2 Disturbances
If non-random influences (disturbances) occur frequently during the trial run, then the process is not
stable (not in control). The causes of the disturbances must be determined and eliminated before
process control is implemented (repeat the trial run).
5.3 Statistical Calculations

In the framework of currently available statistical software, complicated mathematical procedures are
no longer an obstacle and their application is of course permissible and quite common when using
such programs. According to [CDQ 0301], the use of the Solara® / qs-STAT® / procella® / O-QIS®
software package is prescribed within Bosch. It calculates capability and performance indices and
operates control charts in accordance with the used evaluation strategy.
NOTE: The procedures described below were originally developed to enable calculation with the aid of a pocket
calculator. They are usually also contained in corresponding statistics programs.
Hint: Due to the software support available today, the procedures in connection with the preparation,
management and evaluation of control charts allow a better adaptation to process-specific conditions
(e.g. process models) than is possible with manual calculation procedures. However, they also
inevitably bring along higher requirements regarding statistical knowledge and the handling of
statistical software. The requirements for personnel and training must take this into account.
In every division and every plant a fully trained SPC specialist should be available as contact person.
•••••••••
•••••••••
6 Formulas for Determining the Distribution Parameters

6.1 Process Average
Parameter 𝜇 is estimated by Example (Chapter 12)
1 𝑠𝑢𝑚 𝑜𝑓 𝑥̅ 𝑣𝑎𝑙𝑢𝑒𝑠 62.6+62.8+⋯+62.0
𝜇̂ = 𝑥̅̅ = ∙ ∑𝑚
𝑗=1 𝑥̅𝑗 = 𝜇̂ = 𝑥̅̅ = 25
= 62.3
𝑚 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑠𝑎𝑚𝑝𝑙𝑒𝑠
or
1 𝑠𝑢𝑚 𝑜𝑓 𝑚𝑒𝑑𝑖𝑎𝑛𝑠 63+63+⋯+62
𝜇̂ = 𝑥̃̅ = 𝑚 ∙ ∑𝑚
𝑗=1 𝑥
̃𝑗 = 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑠𝑎𝑚𝑝𝑙𝑒𝑠 𝜇̂ = 𝑥̃̅ = 25
= 62.4
If 𝜇̂ significantly deviates from the center point 𝐶 for a characteristic with two-sided limits, then this
deviation should be corrected by adjusting the machine.
2021-02-10 - SOCOS
6.2 Process Variation

Parameter 𝜎 is estimated by Example (Chapter 12)
a) 1 0.552 +0.452 +⋯+2.552
𝜎̂ = √𝑚 ∙ ∑𝑚 2
𝑗=1 𝑠𝑗 𝜎̂ = √ = 1.41
25
𝑠𝑢𝑚 𝑜𝑓 𝑣𝑎𝑟𝑖𝑎𝑛𝑐𝑒𝑠 Note: When moving averages are used

𝜎̂ = √
𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑠𝑎𝑚𝑝𝑙𝑒𝑠 𝜎̂ = 𝑠 is calculated directly from at least
25 individual measurements.
or
b) 𝑠̅ 1
𝜎̂ = 𝑎 with 𝑠̅ = 𝑚 ∙ ∑𝑚
𝑗=1 𝑠𝑗 0.552 +0.452 +⋯+2.552
𝑛 𝜎̂ = √ = 1.41
25
𝑆𝑢𝑚 𝑜𝑓 𝑠𝑡𝑎𝑛𝑑𝑎𝑟𝑑 𝑑𝑒𝑣𝑖𝑎𝑡𝑖𝑜𝑛𝑠 𝑠̅ 1.27

𝑠̅ = 𝜎̂ = = = 1.35
𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑠𝑎𝑚𝑝𝑙𝑒𝑠 𝑎𝑛 0.94
𝑛 𝑎𝑛
3 0.89
5 0.94
See Chapter 9 for additional values 7 0.96
or
c) 𝑅̅ 1 1+1+⋯+6
𝜎̂ = with 𝑅̅ = ∙ ∑𝑚
𝑗=1 𝑅𝑗 𝑅̅ = 25
=2.96
𝑑𝑛 𝑚
𝑠𝑢𝑚 𝑜𝑓 𝑟𝑎𝑛𝑔𝑒𝑠 𝑅̅ 2.96

𝑠̅ = 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑠𝑎𝑚𝑝𝑙𝑒𝑠 𝜎̂ = 𝑑 = 2.33 = 1.27
𝑛
𝑛 𝑑𝑛
3 1.69
5 2.33
See Chapter 9 for additional values 7 2.70
Note: The use of table values 𝑎𝑛 and 𝑑𝑛 presupposes a normal distribution.
•••••••••
•••••••••
7 Process-related Control Limits

The control limits (lower control limit 𝐿𝐶𝐿 and upper control limit 𝑈𝐶𝐿) are set such that 99 % of all
the values lie within the control limits in the case of a process which is only affected by random
influences (random causes).
If the control limits are exceeded, it must therefore be assumed that systematic, non-random
influences (non-random causes) are affecting the process.
These effects must be corrected or eliminated by taking suitable action (e.g. adjustment).
Relation between the variance (standard deviation 𝜎𝑥 ) of the single values (original values, individuals)
𝜎
and the variance (standard deviation 𝜎𝑥̅ ) of the mean values: 𝜎𝑥̅ = 𝑛𝑥
√
2021-02-10 - SOCOS
80
UCL
70
Characteristic value x
60
LCL
50
40
0 5 10 15 20 25
Fig. 7: Schematic about the functionality of a mean chart with 𝑛 = 5. In order to illustrate the relation-
ships, the mean values (points) as well as the individual values (crosses) from each sample are
represented. The means like the individual values show a variation around the process average 𝜇;
1
however, the variation of the means is smaller by the factor than that of the individual values.
√5
•••••••••
•••••••••
7.1 Natural Control Limits for Stable Processes

7.1.1 Control Limits for Location Control Charts
For two-sided tolerances, the limits for controlling the mean must always be based on the center point
𝐶 . 𝐶 is replaced by the process mean 𝜇̂ = 𝑥̅̅ for processes where the center point 𝐶 cannot be
achieved or for characteristics with one-sided limits.
2.58 2.58
𝑈𝐶𝐿 = 𝐶 + ∙ 𝜎̂ 𝑈𝐶𝐿 = 62 + ∙ 1.35 = 63.6
√𝑛 √5
Average Chart
2.58 2.58
𝐿𝐶𝐿 = 𝐶 − ∙ 𝜎̂ 𝐿𝐶𝐿 = 62 − ∙ 1.35 = 60.4
√𝑛 √5
Also possible for normal distributions:*
𝑈𝐶𝐿 = 𝐶 + 𝐴∗ ∙ 𝑠̅ 𝑈𝐶𝐿 = 62 + 1.23 ∙ 1.27 = 63.6
𝐿𝐶𝐿 = 𝐶 − 𝐴∗ ∙ 𝑠̅ 𝐿𝐶𝐿 = 62 − 1.23 ∙ 1.27 = 60.4
2021-02-10 - SOCOS
2.58 2.58
𝑈𝐶𝐿 = 𝐶 + ∙ 𝑐𝑛 ∙ 𝜎̂ 𝑈𝐶𝐿 = 62 + ∙ 1.2 ∙ 1.35 = 63.9
√𝑛 √5
Median Chart
2.58 2.58
𝐿𝐶𝐿 = 𝐶 − ∙ 𝑐𝑛 ∙ 𝜎̂ 𝐿𝐶𝐿 = 62 − ∙ 1.2 ∙ 1.35 = 60.1
√𝑛 √5
𝑈𝐶𝐿 = 𝐶 + 𝐶𝐸 ∙ 𝑅̅ 𝑈𝐶𝐿 = 62 + 0.59 ∙ 2.96 = 63.7
𝐿𝐶𝐿 = 𝐶 − 𝐶𝐸 ∙ 𝑅̅ 𝐿𝐶𝐿 = 62 − 0.59 ∙ 2.96 = 60.3
Note: Use of the median-R chart is only appropriate when charts are manually filled out,
without computer support.
𝑈𝐶𝐿 = 𝐶 + 𝐸𝐸′ ∙ 𝜎̂ 𝑈𝐶𝐿 = 62 + 3.09 ∙ 1.35 = 66.2
Individual Data
.𝐿𝐶𝐿 = 𝐶 − 𝐸𝐸′ ∙ 𝜎̂ 𝐿𝐶𝐿 = 62 − 3.09 ∙ 1.35 = 57.8

Chart

𝑈𝐶𝐿 = 𝐶 + 𝐸𝐸 ∙ 𝑅̅ 𝑈𝐶𝐿 = 62 + 1.33 ∙ 2.96 = 65.9
𝐿𝐶𝐿 = 𝐶 − 𝐸𝐸 ∙ 𝑅̅ 𝐿𝐶𝐿 = 62 − 1.33 ∙ 2.96 = 58.1
* Do not use for moving calculation of indices!
𝑛 𝐴∗ 𝐶𝐸 𝑐𝑛 𝐸𝐸′ 𝐸𝐸 Estimated values 𝜇̂ and 𝜎̂ are calculated
3 1.68 1.02 1.16 2.93 1.73 per Sections 6.1 and 6.2.
5 1.23 0.59 1.20 3.09 1.33
7 1.02 0.44 1.21 3.19 1.18
See Chapter 9 for additional values
Comments on the individual data chart:

Use an 𝑥̅ − 𝑠 chart to perform the capability study (trial run). Estimate 𝜇̂ and 𝜎̂ for the mean 𝜇 and the
standard deviation 𝜎 of the population from at least 𝑚 = 25 random samples (with, for example, 𝑛 =
5), for a minimum of 125 measurements.
These formulas are based on the assumption of normally distributed individual values. Deviations from
the normal distribution can cause larger fractions nonconforming.
Fill out the chart by periodically taking random samples of size 𝑛 (e.g. 𝑛 = 5), numerically entering the
data in the chart (if possible) and displaying the individual data in the chart graph.
To evaluate the chart (e.g. ongoing calculation of 𝐶𝑝 and 𝐶𝑝𝑘 indices), it is necessary to calculate
jeweils 𝑥̅ and 𝑠 or 𝑥̃ and 𝑅 from the random sample data (𝑛 measurements) for the full control chart
and to continue as with an 𝑥̅ − 𝑠 chart according to [Booklet 9].
•••••••••
•••••••••
Control limits based on Pearson distributions

Due to the central limit theorem of statistics, means 𝑥̅𝑗 of samples, each consisting of 𝑛 ≥ 5 individual
values, can be regarded as approximately normally distributed.
For this reason, the formula for control limits of the average and median chart in the above table
contain the 0.5% quantile -2.58 and the 99.5% quantile +2.58 of the standard normal distribution.
However, [Booklet 9] describes several distributions as potential resulting process distributions which
can occur in practice in the case of unilaterally limited characteristics:
 Lognormal distribution  Rayleigh distribution
 Folded normal distribution  Weibull distribution
The sample means of such skewed distributions and small 𝑛 are no longer necessariy normally
distributed. This also applies to the mixture distribution. It can be sensible then to use a Pearson chart.
2021-02-10 - SOCOS
Compared to the corresponding Shewhart chart, this has the advantage of slightly wider control limits.
They consider the skewness of the distribution of the original values and are therefore asymmetrical
to the mean value. However, the disadvantage is that the control limits are more complicated to
calculate and can practically only be done with the help of a computer.
In this case, a distribution from the family of Pearson distributions is best fitted to the sample means.
The control limits of the Pearson chart then correspond to the 0.5% and 99.5% quantiles of this
distribution, for instance.
•••••••••
•••••••••
7.1.2 Control charts with moving averages

The 𝑥̅ chart with a moving average is a special case of the 𝑥̅ chart. For this chart, only single random
samples are taken (i.e. 𝑛 = 1). This chart can be considered e.g. in case of destructive or expensive
inspection and monitoring of process characteristics (e.g. bath concentration).
𝑛 sample measurements are formally grouped as a random sample and the average of these 𝑛
measurements is calculated as the mean.
For each new measurement from a single random sample that is added to the group, the first
measurement of the last group is deleted, yielding a new group of size 𝑛, for which the new average is
calculated.
Example for 𝑛 = 1(3):

2021-02-10 - SOCOS
3 7 4  𝑥̅1 = 4.7
3 7 4 9  𝑥̅2 = 6.7
3 7 4 9 2  𝑥̅3 = 5.0
3 7 4 9 2 8  𝑥̅4 = 6.3
Of course, moving averages calculated in this manner are not mutually independent. That is why this
chart has a delayed reaction to sudden process changes. The control limits correspond to those for
“normal” average charts:
2.58 2.58
𝐿𝐶𝐿 = 𝐶 − ∙ 𝜎̂ 𝑈𝐶𝐿 = 𝐶 + ∙ 𝜎̂
√𝑛 √𝑛
Calculation of 𝜎̂ according to Section 3.5 a)
Control limits for 𝑛 = 1(3):
𝐿𝐶𝐿 = 𝐶 − 1.5 ∙ 𝜎̂ 𝑈𝐶𝐿 = 𝐶 + 1.5 ∙ 𝜎̂
This approach for moving sample measurements can also be applied to the variation, so that an 𝑥̅ − 𝑠
chart with a moving average and moving standard deviation can be used.
After intervention in the process or process changes, previously obtained measurements may no
longer be used to calculate moving indices.
NOTE: In case of the der 𝑥̅ chart with moving average there is a greater probability for the occurence of “unusual
sequences of points”. This is why the rule of seven must not be applied to this chart type.
•••••••••
•••••••••
EWMA Control Chart

The standard [ISO 7870-6] describes the EWMA control chart with exponentially weighted sample
values (Exponentially Weighted Moving Average). A smoothing factor 𝜆 with 0 < 𝜆 < 1 causes values
further back to have a weaker influence, the last value has the greatest influence.
Area of application:
 Low production rates, small quantities (often the reason to select a control chart with 𝑛 = 1)
 Elaborate inspection (time, costs)
 Small deviations from the target value should be detected (e.g. also with automated 100%
inspection)
The applied test statistic is 𝑦𝑖 = 𝜆 ∙ 𝑥𝑖 + (1 − 𝜆) ∙ 𝑦𝑖−1 . 𝑦0 = 𝐶
Due to this recursive calculation all previous sample results 𝑥𝑖 are contained in the test statistic 𝑦𝑖 . The
2021-02-10 - SOCOS
larger λ is, the less the effect of previous values 𝑥𝑖 on the current test statistic 𝑦𝑖 .
Example: 𝑦4 = 𝜆 ∙ 𝑥4 + (1 − 𝜆) ∙ 𝜆 ∙ 𝑥3 + (1 − 𝜆)2 ∙ 𝜆 ∙ 𝑥2 + (1 − 𝜆)3 ∙ 𝜆 ∙ 𝑥1 + (1 − 𝜆)4 ∙ 𝐶
For 𝜆 = 0.2 is (1 − 𝜆)3 ∙ 𝜆 ≈ 0.1. The third-last sample value is therefore only considered with a
weight of approx. 10 %.
Summarized notation: 𝑦𝑡 = 𝜆 ∙ ∑𝑡−1 𝑖 𝑖

𝑖=0 (1 − 𝜆) 𝑥𝑡−𝑖 + (1 − 𝜆) ∙ 𝐶
The control limits of this chart are recalculated for each sample value.
𝜆 𝜆
𝐿𝐶𝐿 = 𝐶 − 2,58 ∙ √2−𝜆 ∙ [1 − (1 − 𝜆)2∙𝑖 ] ∙ 𝜎̂ 𝑈𝐶𝐿 = 𝐶 + 2.58 ∙ √2−𝜆 ∙ [1 − (1 − 𝜆)2∙𝑖 ] ∙ 𝜎̂
They are therefore variable in time. Their distance from each other initially increases, but asymp-
totically approaches a constant value dependent on 𝜆:
𝜆 𝜆
𝐿𝐶𝐿 = 𝐶 − 2.58 ∙ √2−𝜆 ∙ 𝜎̂ 𝑈𝐶𝐿 = 𝐶 + 2.58 ∙ √2−𝜆 ∙ 𝜎̂
After every intervention the control limits are reset to their initial values and the numbering of the
samples starts again at 𝑖 = 1.
The calculation of the control limits presupposes a known and constant process standard deviation 𝜎̂.
The number 2.58 is the 99.5% quantile of the standard normal distribution.
A typical value of λ is λ = 0.2. Due to the expression (1 − λ)2∙i under the square root the control
λ
limits reach their respective constant values EG = C ± 2.58 ∙ √2−λ ∙ σ
̂ after a few samples, and the
faster, the greater λ is. This is the reason why often these constant control limits are used. On the
other hand, the smaller λ is, the greater the sensitivity of the EWMA card to small mean value
deviations.
Advantage Disadvantages
 The EWMA chart reacts sensitively to  The test statistic is not immediately
small changes of the average location. understandable.
 When variable control limits are used,
this can be irritating for the user.
•••••••••
•••••••••
7.1.3 Control Limits for Variation Control Charts
The control limits to monitor the variation Note: Formula a) must be used in
(depending on 𝑛 ) relate to 𝜎̂ and 𝑠̅ and the case of moving s calculation.
likewise 𝑅̅ (= “Central line”). Calculation of 𝜎̂ per Section 6.2 a).
a) generally applicable formula (also for Example (Chapter 12):

moving 𝑥̅ − 𝑠 chart)
′
𝑈𝐶𝐿 = 𝐵𝐸𝑜𝑏 ∙ 𝜎̂ 𝑈𝐶𝐿 = 1.93 ∙ 1.35 = 2.6
′ 𝐿𝐶𝐿 = 0.23 ∙ 1.35 = 0.3
𝐿𝐶𝐿 = 𝐵𝐸𝑢𝑛 ∙ 𝜎̂
b) for standard 𝑥̅ − 𝑠 chart
∗
𝑈𝐶𝐿 = 𝐵𝐸𝑜𝑏 ∙ 𝑠̅ 𝑈𝐶𝐿 = 2.05 ∙ 1.27 = 2.6
2021-02-10 - SOCOS
∗ 𝐿𝐶𝐿 = 0.24 ∙ 1.27 = 0.3

𝐿𝐶𝐿 = 𝐵𝐸𝑢𝑛 ∙ 𝑠̅
𝑅 chart
𝑈𝐶𝐿 = 𝐷𝐸𝑜𝑏 ∙ 𝑅̅ 𝑈𝐶𝐿 = 2.1 ∙ 2.96 = 6.2

𝐿𝐶𝐿 = 𝐷𝐸𝑢𝑛 ∙ 𝑅̅ 𝐿𝐶𝐿 = 0.24 ∙ 2.96 = 0.7
′ ′ ∗ ∗
𝑛 𝐵𝐸𝑢𝑛 𝐵𝐸𝑜𝑏 𝐵𝐸𝑢𝑛 𝐵𝐸𝑜𝑏 𝐷𝐸𝑢𝑛 𝐷𝐸𝑜𝑏
3 0.07 2.30 0.08 2.60 0.08 2.61
5 0.23 1.93 0.24 2.05 0.24 2.10
7 0.34 1.76 0.35 1.88 0.34 1.91
See Chapter 9 for additional values
The standard deviations s cannot become less than zero. They are subject to a distribution whose
density function is strongly asymmetrical (right-skewed) for small sample sizes. The mean value 𝑠̅
which is entered as a dashed line in the chart diagram is therefore below the middle between 𝐿𝐶𝐿 and
𝑈𝐶𝐿. As a tendency, slightly more values (points) can be expected below this line than above. This also
applies when using the ranges R and the average range 𝑅̅.
•••••••••
•••••••••
7.2 Control Limits for Processes with Systematic Changes in the Average
If changes of the mean need to be considered as a process-specific feature (trend, lot steps) and it is
not economical to prevent such changes of the mean, then it is necessary to extend the “natural
control limits”.
The procedure for calculating an average chart with extended control limits is shown below.
The overall variation consists of both the “inner” variation of the random samples and of the “outer”
variation between the random samples.
Calculation procedure Control limits for the mean

Standard deviation of the means 𝑈𝐶𝐿 = 𝐶 + 2.58 ∙ 𝜎̂𝑥̅
𝐿𝐶𝐿 = 𝐶 − 2.58 ∙ 𝜎̂𝑥̅
1 2 𝐶 is the center point
𝜎̂𝑥̅ = 𝑠𝑥̅ = √ ∙ ∑𝑚 ̅
𝑗=1(𝑥̅𝑗 − 𝑥̅ )
2021-02-10 - SOCOS
𝑚−1
Note: For processes that cannot be

controlled about the mean, replace C with
Calculation of 𝜇̂ according to Section 6.1 the process mean 𝜇̂ .
Determining outer variation with analysis of 2.58

𝑈𝐶𝐿 = 𝜇̂ + ∙ 𝜎̂ + 1.5 ∙ 𝜎̂𝑎𝑑𝑑
√𝑛
variance (ANOVA)
2.58
The variation between random samples 𝜎̂𝑎𝑑𝑑 𝐿𝐶𝐿 = 𝜇̂ − ∙ 𝜎̂ − 1.5 ∙ 𝜎̂𝑎𝑑𝑑
√𝑛
can be determined with analysis of variance
using suitable software (ANOVA model II). Calculate the inner variation (standard
deviation) 𝜎̂ according to Section 6.2
Total standard deviation (standard deviation 2.58
𝑈𝐶𝐿 = 𝜇̂ + ∙ 𝜎̂
√𝑛
of all individual measurements)
2.58
𝐿𝐶𝐿 = 𝜇̂ − ∙ 𝜎̂
√𝑛
1
𝜎̂ = 𝑠𝑡𝑜𝑡𝑎𝑙 = √ ∙ ∑𝑁 ̅ 2
𝑖=1(𝑥𝑖 − 𝑥̅ )
𝑁−1
with 𝑁 = 𝑛 ∙ 𝑚
Calculation of 𝜇̂ according to Section 6.1

Estimating the upper and lower process 2.58
𝑈𝐶𝐿 = 𝜇̂ 𝑚𝑎𝑥 + ∙ 𝜎̂
√𝑛
location limits
2.58
𝐿𝐶𝐿 = 𝜇̂ 𝑚𝑖𝑛 − ∙ 𝜎̂
√𝑛
𝜇̂ 𝑚𝑎𝑥 = Average of the 3 largest 𝑥̅𝑖
𝜇̂ 𝑚𝑖𝑛 = Average of the 3 smallest 𝑥̅𝑖
Calculation of 𝜎̂ according to Section 6.2
•••••••••
•••••••••
8 Tolerance-Related Control Limits ― Acceptance Control Chart

For processes with system-related average changes, the control limits can be based on tolerances in
special cases (e.g. mechanical processing with stepped bores, internal turning tools, form disk). A
prerequisite for this is that the inner variation is small enough compared to the tolerance (𝜎̂ per
Section 6.2), i.e. 𝑇 ≥ 10 ∙ 𝜎̂.
When tolerance-related control limits are used, the process is always controlled worse than when
natural control limits are used, since they focus on staying within the tolerance and not on improving
the process (stabilization, centering).
Control limits for tolerance-related 𝑥̅ or 𝑥̃ charts are calculated using the following formulas:
𝐿𝐶𝐿 = 𝐿𝑆𝐿 + 𝑘𝐴 ∙ 𝜎̂ 𝑈𝐶𝐿 = 𝑈𝑆𝐿 − 𝑘𝐴 ∙ 𝜎̂

2021-02-10 - SOCOS
𝑘𝐴 is called the delimitation factor (for calculating 𝜎̂ refer to the formula below).
Control limits for corresponding 𝑠 or 𝑅 charts are calculated according to Section 7.1.3.
NOTE: Tolerance-related control limits can be used to support the decision according to Section 2.4. These limits
are then called alarm limits (LAL, UAL).
Specifying the delimitation factor

The delimitation factor is specified so that a nonconforming fraction of 𝑝 = 1 % is indicated with a
probability of 𝑃𝐴 = 99 %.
In the following formulas, 𝑢1−𝑝 and 𝑢𝑃𝐴 are the percentage points (quantiles) of the normal distri-
bution of the probabilities 1 − 𝑝 or 𝑃𝐴 or √1 − 𝑃𝐴 .
Average chart
1 𝑝 = 1 % and 𝑃𝐴 = 99% yield:
𝑘𝐴 = 𝑢1−𝑝 + ∙ 𝑢𝑃𝐴
√𝑛
𝑛 𝑘𝐴 𝑛 𝑘𝐴
3 3.67 7 3.21
4 3.49 8 3.15
5 3.37 9 3.10
6 3.28 10 3.06
Individual data chart

𝑘𝐸 = 𝑢1−𝑝 − 𝑢 𝑛√1−𝑃𝐴 𝑝 = 1 % and 𝑃𝐴 = 99% yield:
𝑛 𝑘𝐸 𝑛 𝑘𝐸
3 3.11 7 2.28
4 2.80 8 2.17
5 2.58 9 2.07
6 2.42 10 1.99
•••••••••
•••••••••
9 Tables
Table 1 Table 2a
Constants to estimate the Constants to estimate the control limits of
standard deviation (Section 6.2) the individual data chart (Section 7.1.1)
n 𝑎𝑛 𝑑𝑛 n 𝐸𝐸′
2 0.798 1.128 2 2.807
3 0.886 1.693 3 2.934
4 0.921 2.059 4 3.023
5 0.940 2.326 5 3.089
6 0.952 2.534 6 3.143
7 0.059 2.704 7 3.188
2021-02-10 - SOCOS
8 0.965 2.847 8 3.226

9 0.969 2.970 9 3.260
10 0.973 3.078 10 3.289
Table 2b
Constants to determine the control limits (Section 7.1)
Location Variation
𝑥̃ − 𝑅 𝑥̅ − 𝑠 𝑥−𝑅 s Chart R Chart
∗ ′ ′ ∗ ∗
n 𝐶𝐸 𝐴 𝐸𝐸 𝐵𝐸𝑢𝑛 𝐵𝐸𝑜𝑏 𝐵𝐸𝑢𝑛 𝐵𝐸𝑜𝑏 𝐷𝐸𝑢𝑛 𝐷𝐸𝑜𝑏
2 1.614 2.283 2.487 0.006 2.807 0.008 3.518 0.008 3.518
3 1.019 1.678 1.734 0.071 2.302 0.080 2.597 0.080 2.614
4 0.683 1.398 1.468 0.155 2.069 0.168 2.245 0.166 2.280
5 0.593 1.225 1.328 0.227 1.927 0.242 2.050 0.239 2.100
6 0.471 1.105 1.240 0.287 1.830 0.302 1.924 0.296 1.986
7 0.437 1.015 1.179 0.336 1.758 0.350 1.883 0.341 1.906
8 0.371 0.944 1.133 0.376 1.702 0.390 1.764 0.378 1.846
9 0.354 0.886 1.098 0.410 1.657 0.423 1.709 0.408 1.798
10 0.311 0.837 1.069 0.439 1.619 0.451 1.664 0.434 1.760
′
𝑈𝐶𝐿 = 𝐶 + 𝐶𝐸 ∙ 𝑅̅ 𝑈𝐶𝐿 = 𝐵𝐸𝑜𝑏 ∙ 𝜎̂ 𝑈𝐶𝐿 = 𝐷𝐸𝑜𝑏 ∙ 𝑅̅
𝑥̃ ′
𝐿𝐶𝐿 = 𝐶 − 𝐶𝐸 ∙ 𝑅̅ 𝐿𝐶𝐿 = 𝐵𝐸𝑜𝑏 ∙ 𝜎̂ 𝐿𝐶𝐿 = 𝐷𝐸𝑢𝑛 ∙ 𝑅̅
∗
𝑈𝐶𝐿 = 𝐶 + 𝐴∗ ∙ 𝑠̅ 𝑈𝐶𝐿 = 𝐵𝐸𝑜𝑏 ∙ 𝑠̅
𝑥̅
𝐿𝐶𝐿 = 𝐶 − 𝐴∗ ∙ 𝑠̅ ∗
𝐿𝐶𝐿 = 𝐵𝐸𝑢𝑛 ∙ 𝑠̅
𝑈𝐶𝐿 = 𝐶 + 𝐸𝐸 ∙ 𝑅̅
𝑥
𝐿𝐶𝐿 = 𝐶 − 𝐸𝐸 ∙ 𝑅̅
•••••••••
•••••••••
10 Example of an Event Code for Mechanically Processed Parts

It is necessary to make a distinction between the event code that the machine operator needs and the
event code for his foreman. The event codes for the foreman may not be used by the machine operator,
nor may he make the decisions.
The following plan for the operator level is divided up into causes, action and handling of parts/goods.
Causes
C1 Batch change in the materials S2 Work piece clamping device, breakage
C2 Batch change in processing S3 Work piece limit stop defective
S4 Work piece drive unit defective
K1 Calibration not correct
K2 Reference standard defective V1 Incoming goods are dirty
2021-02-10 - SOCOS
V2 Inspection instrument is dirty

M1 Machine defect, mechanical
V3 Machine is dirty
M2 Machine defect, electrical
V4 Preparation/condition nonconforming
M3 Machine defect, hydraulic
M4 Machine failure in the conveyor W0 Tool defective
system W1 Tool, end of tool life
W2 Tool, breakage
P1 Individual value not plausible
W3 Tool holder defective
P2 Measuring/inspection equipment
defective W4 Tool dressing device defective
S1 Work piece clamping device, tool life Z1 Determine other causes
Action
A1 Termination of the inspection during M6 Machine cleaned
a sample, since the values up to now M7 Machine serviced/repaired
have all been outside the plausibility
area. P3 Inspection equipment changed
P4 Inspection controller changed
A2 Termination of the inspection after
completing a sample, since it is beyond P5 Inspection equipment adjusted
the control limits and it is expected that P6 Inspection equipment cleaned
it will affect subsequent characteristics. P7 Inspection repeated
A3 Termination of the inspection so S5 Tool holder changed

that it can be resumed later. S6 Work piece limit stop corrected
K3 Calibration repeated S7 Work piece drive unit corrected
K4 Calibration repeated with new W5 Tool adjusted
standard
W6 Tool changed
K5 Calibration carried out
W7 Tool holding device corrected
M5 Machine adjusted W8 Tool dressed
W9 Tool dressing device corrected
•••••••••
•••••••••
Handling of the Parts/Goods

T1 Control limits exceeded; a recheck T5 Parts/goods to be scrapped 1)
did not reveal any necessity for sorting;
parts/goods delivered as OK T6 100 % sorted, concession portion delivered 1)
T2 100% sorted, OK portion was T7 100 % sorted, OK or concession portion

delivered delivered, rework portion kept 1)
T3 Parts/goods delivered with Z3 Determine other form of handling

concession 1)
1)
As a rule, the machine operator does not make
T4 Parts/goods blocked; rework the decisions on these matters.
required
2021-02-10 - SOCOS
11 Reaction Catalog
A reaction catalog needs to be created especially for the specific process being controlled (machine,
test location).
The event codes listed in the previous sections are only meant to be examples, which in this case refer
to a mechanical manufacturing process.
Example: Reaction Catalog for SPC

Measuring station no.: 82
If the control chart reacts the following actions must be taken and entered in the chart.
After action on the process take a sample and enter measured values in the control chart!
Goods: Machine:
11. 100% sorting, OK-proportion delivered 41. Adjust machine
12. Goods delivered with concession 42. Machine defect mech./hydr./pneum.
13. Goods held; rework necessary 43. Machine defect electrical
14. Scrap 44. Change dressing tile
45. Change dressing rollers
Measurement system:
Tool:
51. Change of gage caliper
21. Dress grinding wheel
52. Measurement control/compensation
22. Change grinding wheel
General:
Work piece: 61. Contamination
31. Change chuck 62. Preliminary working/delivery
32. Stop dog 63. Other causes (to be documented)
•••••••••
•••••••••
2021-02-10 - SOCOS
•••••••••
•••••••••
Quality Control Chart Part Flange 1 ... ... 027 Department W410
Operation 1130, precision-turn Contol chart by
for Continuous
Quality Management Characteristic Diameter Date 10.09.2019
Characteristics Issue no. 1
x1 63 63 63 61 62 61 62 63 64 63 62 60 62 60 58 61 60 62 62 60 62 58 62 61 59 Machine ..., Spindle
x2 63 63 63 60 62 62 61 62 64 63 64 61 65 62 63 63 61 61 63 61 63 63 63 64 65 Machine no. ... ....
x3 62 62 61 59 63 63 62 63 63 61 64 63 64 65 63 62 62 62 63 65 64 61 64 64 64
x4 63 63 63 63 63 63 61 62 63 60 65 62 63 65 61 64 59 61 63 65 60 62 64 63 62 USL 48.69

x5 62 63 62 62 62 60 62 61 63 63 62 63 62 63 62 62 60 62 63 63 62 64 65 60 60 Center point C 48.62
LSL 48.55
12 Example of an 𝒙
x 62.6 62.8 62.4 61.0 62.4 61.8 61.6 62.2 63.4 62.0 63.4 61.8 63.2 63.0 61.4 62.4 60.4 61.6 62.8 62.8 62.2 61.6 63.6 62.4 62.0 Reference value 48.00
s 0.55 0.45 0.89 1.58 0.55 1.30 0.55 0.84 0.55 1.41 1.34 1.30 1.30 2.12 2.07 1.14 1.14 0.55 0.45 2.28 1.48 2.30 1.14 1.82 2.55 Unit 0.01 mm
65 Sample size 5
64 Frequency every 150 parts
Additional test EII
63 Control limits for o o o
~
̅ − 𝒔 Chart
x x x
62 Process related o
x
39
Tolerance related o
61 UAL 64.6
60 UCL 63.6
Target value 62.0
59 LCL 60.4
0 5 10 15 20 25
LAL 59.4
3
Control limits for o s o R
UCL 2.6000
Central line 1.300
s 2 LCL 0.3000
Evaluation
x 62.3
1 s 1.27
̂ 1.35
Cp 1.73
0 Cpk 1.66
0 5 10 15 20 25
Date 11.10. 11.10. 11.10. 11.10. 12.10. 12.10. 12.10. 12.10. 13.10. 13.10. 13.10. 13.10. 14.10. 14.10. 14.10. 14.10. 15.10. 15.10. 15.10. 15.10. 18.10. 18.10. 18.10. 18.10. 19.10. Corrective action (see overleaf)
Time 6:50 11:40 14:40 20:35 6:25 10:50 14:20 21:05 6:20 10:15 15:00 21:35 7:10 11:50 16:05 21:30 6:50 12:10 16:15 21:20 6:45 12:05 16:20 21:15 6:50 Evaluated
Name Checked
13 Responsiveness of a Control Chart and Type-1 Error

Keeping a control chart corresponds to the repeated application of a statistical test. For example, the
sample mean value 𝑥̅ is used to check whether the process location has changed or not. The further
the real process location moves away from the center point 𝐶, the more likely it is that the control
chart will react to it, i.e. 𝑥̅ lies outside one of the control limits.
If the probability of intervention as a function of the process shift is represented in units of the
standard deviation 𝜎, an s-shaped curve is obtained as shown in the following figure.
This curve is called operation characteristic of the control chart.
1 Fig. 13
0.9
0.8
indicates process shift
Probability that chart
0.7
The curve is steeper at 𝑛 = 5 than at
2021-02-10 - SOCOS
0.6
0.5
𝑛 = 3, so the control chart with a larger
0.4 sample size reacts more sensitively to a
0.3 change in the process location. A high
0.2
responsiveness of the average chart
0.1
0
must therefore be "bought" by the
0 0.5 1 1.5 2 2.5 3 3.5 4
sample size. The curve for the average
Process shift in sigma units
chart is also steeper than that of the
xbar-s; n=3 xbar-s; n=5
median chart.
A special feature is not easily recognizable in the presentation. It is the fact that the curve at the lower
end still indicates an intervention probability of approx. 1 % at zero displacement. Thus, even with a
completely stable centered process, there is a probability of 0.5 % in each case that a sample mean 𝑥̅
is below 𝐿𝐶𝐿 or above 𝑈𝐶𝐿. This error probability is also called a type-1 error.
Maximum number of random violations of control limits

m 100 200 300 400 500 600 700 800 900 1,000
𝑥̅ chart 4 6 8 10 12 13 15 16 18 19
𝑠 chart 3 4 5 6 7 8 9 10 11 12
The above table indicates the maximum number of control limit violations that can occur purely by
chance, although the process has not changed. Where 𝑚 is the number of samples.
EDITORIAL NOTE: Statistically speaking, the situation corresponds to an urn model (bowl model) in which 99 %
of the balls bear the inscription "unchanged process", the remaining 1 % of the balls bear the inscription
"changed process" (see [Booklet 2]). If 100 balls are taken out of the urn, with a probability of 99.5 %, no more
than 4 will have the inscription "changed process". The number 4 is the upper threshold value of the binomial
distribution for a probability of 99.5 %. In the case of the s chart, only the exceeding of the upper control limit is
considered, with the associated intervention probability of 0.5%.
•••••••••
•••••••••
14 Reviews and Maturity Assessments

In the present context of quality management, the word "review" means "to look at things critically
once again; to check". The corresponding noun is identical. According to the [ISO 9000] standard, such
a "review" is about "determination of the suitability, adequacy or effectiveness" of something to
"achieve established objectives". In particular, one attempts to identify, for example, inconsistencies,
deficiencies, errors or gaps in a concept or procedure. This usually results in a need for action and
corrective or improvement measures are suggested.
The term maturity often appears in connection with an assessment model (maturity model) that is
used to assess the "maturity" (quality capability) of an organization, system or process, for example,
by means of a catalog of questions or criteria.
NOTE: The mental link with the development of fruits or living organisms certainly plays a role here.
The assessment is carried out, for example, using a multi-level scale (1, 2, 3 or red, yellow, green). An
2021-02-10 - SOCOS
even number of levels (e.g. four levels: not or merely rudimentary, largely not, largely, almost complete
or complete) has the advantage that there is no "medium" rating. The assessor must therefore decide
in the middle range for the worse or better half of the possibilities. A larger number of steps than about
three or four steps makes it more difficult to differentiate between them. A quantitative scale has the
advantage that the results can be presented quite clearly in the form of a network diagram (radar
chart).
Review
To ensure the consistent realization and effectiveness of the control loops as well as the effectiveness
and efficiency of the inspection processes in production, it makes sense to carry out regular reviews.
Maturity Assessment
Carrying out assessements is recommented, in order to
 check the process of quality planning process regarding its application,
 to assess the maturity of its implementation and
 to derive targeted improvement potentials.
The aspects to be evaluated include, for example, the quality characteristics and the respective
associated inspection strategy, the measurement and test processes as well as the implemetation of
the quality control loops (big, small).
“SPC-Assessment” is just another name for such a maturity assessment. A major difference is the "flight
altitude", i.e. the focus and the level of detail. Typical main and sub criteria are, e.g.
 Personnel and Qualification (qualification of users)
 Planning (measurement/test processes, machine and process characteristics, machine
capability)
 Realization (application of the control chart)
 Documentation (of actions)
 Effectiveness (short-term process capability, long-term evaluation)
•••••••••
•••••••••
List of Symbols
𝐶 center point of the tolerance zone (target value)

𝑐4 factor for the determination of 𝜎̂ from the mean standard deviation 𝑠̅
(in older literature also referred to by 𝑎𝑛 )
𝐶𝑔 , 𝐶𝑔𝑘 potential and critical measurement process capability index (MSA, Procedure 1)
𝐶𝑚 , 𝐶𝑚𝑘 potential and critical process capability index
𝐶𝑝 , 𝐶𝑝𝑘 potential and critical process capability index (long-term)
𝐶𝑝−𝑆𝑇 , 𝐶𝑝𝑘−𝑆𝑇 potential and critical process capability index (short-term)
𝑑2 factor for the determination of 𝜎̂ from the mean range 𝑅̅
𝑖 number (index) of the measurement within a sample; 1 ≤ 𝑖 ≤ 𝑛
2021-02-10 - SOCOS
𝑗 number (index) of the measurement within all measurements; 1 ≤ 𝑗 ≤ 𝑚 ∙ 𝑛

𝑘 number (index) of the sample within all samples; 1 ≤ 𝑘 ≤ 𝑚
𝑘𝐴 delimitation factor in the 𝑥̅ − 𝑠 chart
𝑘𝐸 delimitation factor in the individual data chart
𝜆 smoothing factor for the EWMA chart
𝐿𝐴𝐿 Lower Alarm Limit
𝐿𝐶𝐿 Lower Control Limit
𝐿𝐿 Lower Limit; Lower Specification Limit, minimum value
𝑚 number of samples
𝜇 mean of a population
𝜇̂ estimating value for the mean of a population
𝜇̂ 𝑚𝑎𝑥 estimating value for the greatest mean
𝜇̂ 𝑚𝑖𝑛 estimating value for the smallest mean
𝑛 number of readings per sample (sample size) or in a value set
n number of parts (if different from n)
𝑁 total number of all individual values (𝑁 = 𝑚 ∙ 𝑛) or lot size
𝑃𝑝−𝑆𝑇 , 𝑃𝑝𝑘−𝑆𝑇 potential and critical process performance index (short-term)
𝑅 Range of a value set
𝑅𝑘 Range of sample no. 𝑘
𝑅̅ mean of ranges
𝑠 standard deviation of a sample
𝑠𝑡𝑜𝑡𝑎𝑙 standard deviation of all individual values
𝑠𝑥̅ standard evaluation of the mean values of 𝑚 samples
𝑠̅ mean standard deviation from 𝑚 samples of the same size
̅̅̅
𝑠2 mean variance; mean squared standard deviations
•••••••••
•••••••••
𝜎 standard deviation of the population

𝜎̂ estimate value for the standard deviation of the population
𝜎̂𝑎𝑑𝑑 (additional) variance (standard deviation) between samples
𝑇 tolerance of a characteristic
𝑢1−𝑝 quantile of the standard normal distribution for the probability 1 − 𝑝
𝑈𝐴𝐿 Upper Alarm Limit
𝑈𝐶𝐿 Upper Control Limit
𝑈𝑆𝐿 Upper Limit; Upper Specification Limit (of the tolerance zone)
𝑥𝑖 single value no. 𝑖 of a value set
𝑥𝑖𝑘 single value no. 𝑖 in sample no. 𝑘
2021-02-10 - SOCOS
𝑥𝑚𝑎𝑥 largest individual value of a value set (maximum)

𝑥𝑚𝑖𝑛 smallest individual value of a value set (minimum)
𝑥̅ arithmetric mean
𝑥̅𝑘 arithmetic mean of the individual values in sample no. 𝑘
𝑥̿ mean of mean values (total mean; grand mean)
𝑥̃ median
𝑥̃𝑘 median of the individual values in sample no. 𝑘
𝑥̃̅ mean of medians
′ ′ ∗ ∗ ′
Konstanten: 𝑎𝑛 , 𝑐𝑛 , 𝑑𝑛 , 𝐴, 𝐴′ , 𝐴∗ , 𝐵𝐸𝑢𝑛 , 𝐵𝐸𝑜𝑏 , 𝐵𝐸𝑢𝑛 , 𝐵𝐸𝑜𝑏 , 𝐶𝐸 , 𝐷𝐸𝑢𝑛 , 𝐷𝐸𝑜𝑏 , 𝐷𝑜𝑏 , 𝐸𝐸 , 𝐸𝐸′
Other symbols used only in individual chapters or the use of symbols with different meanings are defined
in the appropriate context.
•••••••••
•••••••••
Variants of usual abbreviations

Depending on source and language version several designations and abbreviations are common in the
framework of SPC.
English German
Lower Limit 𝐿𝐿 or 𝐿 𝑈𝐺𝑊 Unterer Grenzwert

Lower Specification Limit 𝐿𝑆𝐿 𝑈𝑆𝐺 Untere Spezifikationsgrenze
Upper Limit 𝑈𝐿 or 𝑈 𝑂𝐺𝑊 Oberer Grenzwert

Upper Specification Limit 𝑈𝑆𝐿 𝑂𝑆𝐺 Obere Spezifikationsgrenze
2021-02-10 - SOCOS
Lower Control Limit 𝐿𝐶𝐿 𝑈𝐸𝐺 Untere Eingriffsgrenze
Upper Control Limit 𝑈𝐶𝐿 𝑂𝐸𝐺 Obere Eingriffsgrenze
•••••••••
•••••••••
Terms and Definitions

NOTE 1: The following definitions of terms were taken from the respective standards cited in this document.
Corresponding notes were only adopted in single cases if they were considered directly relevant and/or essential
for understanding a standardized term. Otherwise, the respective standard should be referenced for notes and
examples.
NOTE 2: “Editorial notes”are not part of the respective standard.
NOTE 3: In some cases, the same term is listed with several definitions from various standards and directives,
provided that the definitions do not appear entirely consistent.
alarm limits
Tolerance-related control limits whose violation generally requires increased monitoring of the
process.
2021-02-10 - SOCOS
acceptance control chart

Control chart intended primarily to evaluate whether or not the plotted measure can be expected to
satisfy specified tolerances. [ISO 3534-2]
AQDEF (Advanced Quality Data Exchange Format)

The name AQDEF is an officially registered trademark of Q-DAS GmbH & Co. KG, Weinheim. AQDEF is a
standard for the collection and the cross-system exchange of quality information, among others
between measurement systems and and the Q-DAS software products.
determination
Activity to determine one or more characteristics and its characteristic values [ISO 9000]
capability
Suitability of a object (for example product, service, process, person, organization, system, resource)
for realization of a result, which will fullfill the requirements of this result (with reference to [ISO
9000])
capability
Suitability of an organization, a system or an process for realisation of a product, which will fullfill the
requirements of this product [ISO 22514-1]
capability index: see process capability index
center point of the specification interval

Arithmetic mean of lower and upper limiting value
NOTE: In many cases the center point corresponds to the imaginary or to the defined target value.
[DIN 55350-12]
center point of the specification interval

𝐿𝑆𝐿+𝑈𝑆𝐿
Arithmetic mean of lower and upper limiting value: 𝐶 = 2
•••••••••
•••••••••
EDITORIAL NOTE: In the case of characteristics limited on one side above (only USL given), such as roughness
(Rz), shape and position (e.g. roundness, rectangularity), it does not make sense to assume 𝐿𝑆𝐿 = 0 and
𝑈𝑆𝐿
consequently set 𝐶 = .
2
characteristic
Feature for the recognition and distinction of units. [DIN 55350-12]
characteristic
Distinguishing property
NOTE 1: A characteristic can be inherent or assigned.
NOTE 2: A characteristic can be of qualitative or quantitative nature.
NOTE 3: There are different classes of characteristics, e. g.:
 physical, for example mechanical, electrical, chemical or biological characteristics;
2021-02-10 - SOCOS
 sensory, e. g. regarding smell, touch, taste, sight, hearing;

 behavioral, e. g. decency, honesty, truthfulness;
 time-related, for example punctuality, reliability, availability;
 ergonomic, for example physiological or characteristics related to human safety;
 functional, for example top speed of an airplane.
[ISO 3534-2]
conformity
Fulfilling a requirement [ISO 9000]
conformity evaluation
Systematic examination about the degree to which an entity fulfills special requirements
[ISO 3534-2]
continuous characteristic
Characteristic whose characteristic values are the measuring values of a physical quantity (for example,
weight, length, current, temperature)
EDITORIAL NOTE: Often imprecisely called “variable characteristic”; see also “continuous scale” [ISO 3534-2]
continuous characteristic
Quantitative characteristic whose co-domain (range of values) is uncountably infinite. [DIN 55350-12]
EDITORIAL NOTE: The value of such a characteristic is always given as a product of numerical value
and unit.
E XAMPLES:
 Length: 12.54561… m,
 Diameter: 3.532… mm.
The numerical value of the characteristic "length", for example, can take on every value between 12
and 13; mathematically expressed: every real number on the interval between 12 and 13. I n reality, no
value of such a continuous characteristic can be measured with infinite accuracy. The number of
decimal places is always limited by the number of available digits that can be displayed. In addition,
properties of the measurement process can limit the number of useful decimal places.
•••••••••
•••••••••
control chart
Chart on which some statistical measure of a series of samples is plotted in a particular order to steer
the process with respect to that measure and to control and reduce variation. [ISO 3534-2]
countable characteristic
Special discrete characteristic whose co-domain is given by the set of natural numbers including zero
(0, 1, 2, …) or a subset of this set.” [DIN 55350-12]
discrete characteristic
Characteristic whose characteristic values are counted measurands in a countable unit (e. g. good/bad,
right/wrong, red/green/blue)
EDITORIAL NOTE: Often imprecisely called “attributive characteristic”; see also “discrete scale”
[ISO 3534-2].
2021-02-10 - SOCOS
discrete characteristic
Quantitative characteristic whose co-domain is finite or countably infinite. [DIN 55350-12]
EDITORIAL NOTE (E XAMPLES)
 Mass in kg (without decimal places).
 Body weight in cm (without decimal places).
 Number of "6" when rolling a die 100 times. The result can take on all of the values from 0 to 100.
The number is finite.
 Number of lightning strikes in Germany during a given year (one can also locate and, of course,
record atmospheric lightning using antennae). This number can take on a countably infinite
number of values (0, 1, 2, 3, …), although very large values will appear with decreasing probability.
entity
That which can be individually described and considered [ISO 3534-2]
EDITORIAL NOTE: Not to be confused with “unit” (see [VIM])
enumeration
Determining the measurand “number of ellements in a set”, for instance by counting. [DIN 1319-1]
estimator
Statistic used in estimation of the parameter Θ. [ISO 3534-1]
estimation
Procedure that obtains a statistical representation of a population from a random sample drawn from
this population.
NOTE 1: In particular, the procedure involved in progressing from an estimator to a specific estimate constitutes
estimation.
[ISO 3534-1]
estimate
Observed value of an estimator. [ISO 3534-1]
•••••••••
•••••••••
influence quantity
Quantity that, in a direct measurement, does not affect the quantity that is actually measured, but
affects the relation between the indication and the measurement result. [VIM]
inspection
Conformity assessment through observation and assessment, accompanied ― if applicable ― by
measurement, testing or comparison [ISO 3534-2]
lower specification limit

Limit indicating the lower limiting value [ISO 3534-2]
measurement result
2021-02-10 - SOCOS
Set of quantity values that are assigned to a measurand along with any available relevant information
[VIM]
measuring instrument
Device, which is used alone or in conjunction with additional facilities for the performance of
measurements [VIM]
measurand
Quantity which is to be measured [VIM]
measurement
Process in which one or more quantity values, that can be reasonably assigned to a quantity, are
experimentally determined
NOTE 2: A measurement means comparing sizes and includes counting
[VIM]
measurement process
Set of activities for determining a quantity value [ISO 9000]
measurement uncertainty
Not negative parameter that characterizes the variation of values that is attached to the measurand
on the basis of information used [VIM]
measurement value
Quantity value which represents a measurement [VIM]
measuring and test equipment

Measuring equipment for inspections [DIN 1319-2]
EDITORIAL NOTE: Measuring equipment for the proof of conformity regarding defined internal, customer, legal
and or regulatory quality requirements.
measuring system
•••••••••
•••••••••
Combination of measuring devices and often other devices and, if necessary reagents and utilities,
which are arranged and adapted to provide information to obtain readings within certain intervals for
quantities of certain types
NOTE: A measuring system can consist of a single measuring device
[VIM]
median (engl. median)

0.5-quantile
NOTE 1: The median is one of the most commonly applied p-quantiles in practical use. The median of a
continuous univariate distribution is such that half of the population is greater or equal to the median and half
of the population is less than or equal to the median.
[ISO 3534-1]
2021-02-10 - SOCOS
median of a sample
Among the values ordered in ascending numerical order and numbered "1" to "𝑛", for odd 𝑛 the actual
𝑛+1
value with the number 2 , for even 𝑛 usually the mean value of the two actual values with the
𝑛 𝑛
numbers 2
and 2
+ 1.
Example: For a sample of size 5 ordered by ascending numerical value, the median is the middle of the
5 values.
nominal characteristic
Qualitative characteristic whose values have no quantitative relation to each other. [DIN 55350-12]
EDITORIAL NOTE, Example: The characteristic "color" with the values "red", "yellow", "green".
The values of these characteristics can only be classified and differentiated (is equal, is not equal).
“The value of a nominal characteristic is often also called attribute. A nominal characteristic with only two,
mutually exclusive values is called a dichotomous characteristic, a binary characteristic or an alternative
characteristic. It can only take on one of two mutually exclusive values.”
Examples:
 good / bad,
 inside / outside the tolerance range (OK / not OK),
 defective / not defective,
 present / not present.
Such data is, for instance, gained using limit gauges or by visual assessment with boundary samples. It is,
however, also possible to define a “discrete” continuous characteristic by first measuring it and then classifying
the measured value using a category “inside/outside the tolerance range”. However, this is rarely sensible due
to the associated loss of information. If measurement values exist for a continuous characteristic, they should
be used in their original form.
ordinal characteristic
Qualitative characteristic whose values have a quantitative relation to each other. [DIN 55350-12]
EDITORIAL NOTE, Example:
 The characteristic "clothing size" with the values XXS, XS, S, L, XL, XXL, …, 7XL,
 The characteristic "(European) shoe size" with the values 14, 15, 16, …, 48, 49.
The values of these characteristics can be classified and differentiated, e.g. using the relationships “smaller
than”, “is equal” or “greater than”.
•••••••••
•••••••••
population
Population of the considered entities [ISO 3534-2]
EDITORIAL NOTE: The term “population” refers to a limited or unlimited number of observable units that are to
be considered concurrent within the framework of an existing statistical problem. Such observable units can, for
example, come from "observations” or results from “tests” conducted under the same conditions.
Examples of finite populations are the number of
 students in a school,
 eligible voters within a state,
 television viewers who watched the final match of the last Football World Cup,
 parts in a delivery of goods,
 products manufactured within one shift at factory XY.
Examples of (theoretically) infinite populations are the number of
 points observed when rolling dice,
2021-02-10 - SOCOS
 results determined when repeatedly measuring against a standard of length,

 parts that a machine will create, under the assumption that it will retain its current condition for ever.
Above all, the previous examples show that a population does not always have to be real; it can also be fictional.
Moreover, one can recognize that a statistical problem can sometimes be focused at a prognosis (prediction) of
future results.
process
Set of associated or mutually influencing activities, which uses entries to achieve an intended result
[ISO 9000]
EDITORIAL NOTE: In the context of this Booklet, a process is generally understood to be a sequence of activities
and/or procedures in which materials or preprocessed parts/components are further processed and a product
is produced. These can be manufacturing or assembly processes, for example. But also a measuring process is
a sequence of activities and/or sequences in which something is produced, namely measuring results.
parameter
Index of a family of distributions
NOTE 1 The parameter may be one-dimensional or multi-dimensional.
NOTE 2 Parameters are sometimes referred to as location parameters, particularly if the parameter corresponds
directly to the mean of the family of distributions. Some parameters are described as scale parameters,
particularly if they are exactly or proportional to the standard deviation of the distribution. Parameters that are
neither location nor scale parameters are generally referred to as shape parameters.
[ISO 3534-1]
EDITORIAL NOTE 1: Parameters of location (expected value, median), parameters of variability (variance,
standard deviation, coefficient of variation) and shape parameters (skewness, kurtosis, excess) are also called
functional parameters.
EDITORIAL NOTE 2: Colloquially the term process parameter is occasionally used. The standard [ISO 3534-1]
refers this term exclusively to parameters of statistical distributions. In connection with processes [ISO 22514-1]
uses the term process characteristic.
performance index
Parameter that indicates the performance measure with respect to stipulated specifications
[ISO 22514-1]
•••••••••
•••••••••
population parameter
Summary measure of the values of some characteristic of a population
EXAMPLE Population mean = µ; population standard deviation = σ
NOTE Population parameters are usually symbolized by lower case Greek letters in italics.
[ISO 3534-2]
process capability index

Parameter which indicates the capability with respect to given specifications [ISO 22514-1]
process characteristic
Inherent characteristic of a process [ISO 22514-1]
EDITORIAL NOTE: Process characteristics are necessary characteristics to ensure the conformity of the product
2021-02-10 - SOCOS
characteristics. These are characteristics that are transmitted to facilities and equipment, and not on the
product.
For the term “inherent”, see also the editorial note with regard to the term “product characteristic”.
process characteristic in control

Process characteristic, with which the parameters of the distribution of the characteristic values change
practically not or only in known way or within known limits [DIN 55350-11]
process in control (engl. translation of DIN 55350-11 not available)

Process whose main features are process characteristics which are in control [DIN 55350-11]
process quality
Required quantified output from process, which is equal to the quality objective (usually fractions
nonconforming e.g. defined in ppm). [CD 00301]
product
Result of a process [ISO 22514-1], [ISO 3534-2]
product characteristic
Inherent characteristic of a product [ISO 22514-1]
EDITORIAL NOTE: “Inherent” means “an inherent unit” (for example, physical properties such as weight, size,
power consumption of a product); therefore an inherent characteristic may be a quality characteristic, but not
a “mapped” characteristic (such as, e. g., price, the owner).
product characteristic in controlN2)

Product characteristic parameters of the distribution of the characteristic values, which virtually do not
change or only change in a known manner or within known limits.
N2)
National Footnote:
ISO 21747 used the English terms “stable process”and “process in a state of statistical control” interchangeably,
which DIN ISO 21747 translated as “stable process”and “dominated process”.
Deviating from that, ISO 22514-1 designated only the behavior after ISO 21747 as “stable process” and “process
in a state of statistical control”, which DIN ISO 22514-1 translated as “stable process”.
However, the behavior according to ISO 21747 is designated in ISO 22514-1 as a “product characteristic in
control”, which is translated as “dominated product characteristic”. This important change is not yet considered
in DIN ISO 3534-2.
•••••••••
•••••••••
[ISO 22514-1]
qualitative characteristic
Characteristic whose values are allocated to a scale without defined distances.
[DIN 55350-12]
quality capability
Suitability of an organization or parts of an organization (e. g. people, procedures, processes,
equipment) for realizing a result that will meet the quality requirements of this result (with reference
to [ISO 22514-1, 3.3.2] and [ISO 9000, 3.6.12]).
quality capable process

2021-02-10 - SOCOS
A process is quality capable if it is able to fully comply with the specified requirements.
quality characteristic
Inherent characteristic of a product, a process or system related to a requirement [ISO 22514-1]
EDITORIAL NOTE: For the term “inherent”, see also the editorial note for the term “product characteristic”.
quality control chart

Quality control chart (Shewhart chart) for monitoring a parameter of the probability distribution of a
characteristic with the purpose of determining whether the value of the parameter deviates from a
given value.
quantitative characteristic
Characteristic whose values are allocated to a scale with defined distances. [DIN 55350-12]
EDITORIAL NOTE: All physical quantities are quantitative characteristics, for example length, volume,
angle, weight, mass, temperature, tension, current, time, speed.
quantity
Property of a phenomenon, a body or a substance wherein the property has a value which can be
expressed by a number and a reference [VIM]
quantity type
Aspect which is common with comparable quantities [VIM]
quantity value
Numerical value and reference, which together specify a quantity quantitatively [VIM]
random cause
Cause of the variation which is constantly inherent in a process [ISO 22514-1]
random sample
Sample, which has been selected randomly [ISO 3534-1]
•••••••••
•••••••••
range
Greatest minus smallest single value: 𝑅 = 𝑥𝑚𝑎𝑥 − 𝑥𝑚𝑖𝑛
reaction plan
Reaction plan Action or series of steps prescribed in a control plan in the event abnormal or
nonconforming events are detected. [IATF 16949]
resulting process distribution

Time-dependent distribution model that reflects the instantaneous distribution of the characteristic
under consideration, and the changes of its location, dispersion and shape parameters during the time
interval of process observation. According to [ISO 22514-2].
2021-02-10 - SOCOS
requirements
Requirement or expectation that or which is stipulated, commonly provided or mandatory
[ISO 9000]
sample
Subset of a population which consists of one or more selection units.
[ISO 3534-1]
EDITORIAL NOTE: In contrast to the population a sample is a real and therefore finite number of “things” or
events. Examples of this are the set of
 vehicles that passed through the highway tunnel near Leonberg on a specific day,
 results observed when rolling one die 10 times,
 results obtained when conducting 25 measurements against a standard of length,
 50 parts made while testing a machine’s capacity.
By the way, the German term for “sample” (“Stichprobe”) originates from the practice of “piercing” grain sacks
and cotton bales during quality inspection. A sample consists of one or several units that were “drawn” from a
real or fictitious population according to the random principle. The number of these elements is called the
sample size. The properties of the sample are supposed to represent the population. Random sampling
presupposes that each element of the population is given the same chance (same probability) to be picked for
the sample. In general, it is rarely possible to apply the random principle in a nearly ideal manner (flipping a
coin, roulette, drawing the lottery numbers). The idea is especially problematic with regard to fictitious
populations; “drawing” the sample is only possible in a figurative sense.
sample mean
Average, arithmetic mean. Sum of random variables in a random sample divided by the number of
terms in the sum. [ISO 3534-1]
sample size
Number of sampling units in a sample [DIN 55350-14]
sampling unit
One of the individual parts into which a population is divided. [ISO 3534-1]
Shewhart control chart
•••••••••
•••••••••
Control chart with Shewhart control limits intended primarily to distinguish between the variation in
the plotted measure due to random causes and that due to special causes. [ISO 3534-2]
sorting inspection
This is a 100% inspection and means that objects/units/items are inspected (visually or using technical
means) with respect to defined characteristics and sepa-rated according to the individual results (e.g.
conforming / nonconforming).
specification
Document that specifies requirements
NOTE: A specification may refer to activities (for example process document, process specification and test
specification), or products (for example, product specification, performance specification and drawing).
[ISO 9000]
2021-02-10 - SOCOS
specification interval
Area between the limits maximum value and minimum value [ISO 22514-1]
EDITORIAL NOTE: The limits are also referred to as specification limits.
specification limit
Limiting value stated for a characteristic [ISO 3534-2]
stable process; process in a state of statistical control

Process which (with regard to its variation) is only subject to random causes
NOTE 1: A stable process will behave in general as if the samples are random samples at any time with simple
sampling from the same population.
NOTE 4: In some processes, the expected value of the characteristic can change, or the standard deviation can
be increased. The reasons may be, for example, tool wear or the reduction of the concentration in a solution. A
progressive change in the expected value or the standard deviation of such a process is considered as systematic
and not as a random cause. These are then the results of sampling, not simple random samples from the same
population.
[ISO 21747]
EDITORIAL NOTE: The English original version of this term is identically defined in ISO 3534-2 (2006) and ISO
21747 (2006); DIN ISO 21747 contains the older German translation (2007); DIN ISO 3534-2 contains the newer
German translation (2013) and uses only the term “process in a state of statistical control”.
stable process
Process which is only subject to random variation causes
NOTE 2: A stable process behaves in general as if samples drawn from the process are at any time simple
random samples from the same population.
[ISO 22514-1]
EDITORIAL NOTE: The definition of the term from the original English version ISO 3534-2 (2006) was adopted in
modified form in the English version ISO 22514-1 (2014); The German version DIN ISO 22514-1 (2016) uses only
the term “stable process”.
statistic
Completely specified function of random variables
NATIONAL FOOTNOTE: Statistics characterize properties of a frequency distribution
•••••••••
•••••••••
[ISO 3534-1]
standard deviation of a sample

Square root of the variance: 𝑠 = √𝑠 2
Statistical Process Control

Statistical quality control for processes
EDITORIAL NOTE: SPC is a standard method for visualizing and controlling processes, based on measurements
of random samples. The goal of SPC is to ensure that the planned process output is achieved and that
corresponding customer requirements are fulfilled.
SPC is always linked to the (manual or software supported) use of quality control charts with the the goal of
achieving, maintaining and improving stable and capable processes. This is done by recording process or
product data, drawing conclusions from this data and reacting to undesirable data with appropriate actions.
2021-02-10 - SOCOS
statistical quality control

Part of quality control where statistical procedures are applied.
NOTE: A special area of statistical quality control is statistical process control.
[DIN 55350-11]
target value
Preferred value or reference value of characteristic which is specified in a specification
[ISO 3534-2]
(specified) tolerance
Difference between maximum value and minimum value [ISO 3534-2]
tolerance interval: see specification interval
EDITORIAL NOTE 1: The specification of a tolerance consists of a number and a unit of measurement, e.g. 0.01
mm. The term tolerance interval or specification interval refers to the range between the minimum value and
the maximum value. Apart from the limited resolution of the measuring process, the tolerance interval contains
an infinite number of values.
EDITORIAL NOTE 2: Colloquially, phrases such as "exceeding tolerance" or "adhering to tolerance" are often
used. This usually means that a measured value of a characteristic lies within or outside the tolerance interval.
EDITORIAL NOTE 3: Since according to the standard a specification is a document, the term specification interval
is strictly speaking wrong. What is meant is the (specified) tolerance interval defined in the specification for a
characteristic.
tolerance zone see specification interval
tolerance zone
Area of permitted values between lower and upper limiting value. [DIN 55350-12]
upper specification limit

Limit indicating the upper limiting value [ISO 3534-2]
•••••••••
•••••••••
variance of a sample
The deviations of the individual values from the arithmetic mean are summarized and the sum is
1
devided by the number of values reduced by one: 𝑠 2 = 𝑛−1 ∙ ∑𝑛𝑖=1(𝑥𝑖 − 𝑥̅ )2
variation
Difference between values of a characteristic [ISO 22514-1]
2021-02-10 - SOCOS
•••••••••
•••••••••
Literature
Corporate directives, only internally available
[CD 02981] CD 02981 Information Governance
[CD 00301] CDQ 0301 Management of Characteristics
[CD 00501] CDQ 0501 Identification and Traceability
[CD 00503] CDQ 0503 Rework and Repair
[CDQ 0509] CDQ 0509 Concessions
Booklet Series “Quality Management in the Bosch Group”, Technical Statistics

[Booklet 1] Booklet No. 1, Basics Concepts of Technical Statistics, Continuous Characteristics
[Booklet 2] Booklet No. 2, Basics Concepts of Technical Statistics, Discrete Characteristics
2021-02-10 - SOCOS
[Booklet 3] Booklet No. 3, Basics Concepts of Techn. Statistics, Evaluation of Measurement Series
[Booklet 8] Booklet No. 8, Measurement Uncertainty
[Booklet 9] Booklet No. 9, Machine and Capability Capability
[Booklet 10] Booklet No. 10, Capability of Measurement and Test Processes
[AIAG PPAP] AIAG Core Tools, Production Part Approval Process (PPAP)
[AIAG SPC] AIAG Core Tools, Statistical Process Control (SPC)
[AQDEF] Advanced Quality Data Exchange Format of the Automotive Industry, Q-DAS GmbH,
Weinheim, Germany
[IATF 16949] IATF 16949, Automotive Quality Management System Standard ― Quality manage-
ment system requirements for automotive production and relevant service parts
organizations
[DIN 1319-1] Basic concepts in metrology, Part 1: General concepts (under revision)
[DIN 55350-11] DIN 55350-11, Concepts for quality management ― Part 11: Supplement to DIN EN
ISO 9000
[DIN 55350-12] DIN 55350-12, Concepts in the field of quality and statistics ― Concepts relating to
characteristics
[DIN 55350-14] DIN 55350-14, Concepts in the field of quality and statistics ― Concepts of sampling
[ISO 3534-1] DIN ISO 3534-1, Statistics — Vocabulary and symbols — Part 1: General statistical
terms and terms used in probability
[ISO 3534-2] DIN ISO 3534-2, Statistics — Vocabulary and symbols — Part 2: Applied Statistics
[ISO 7870-2] Control charts ― Part 2: Shewhart control charts
[ISO 7870-5] Control charts ― Part 5: Specialized control charts
[ISO 7870-6] Control charts ― Part 6: EWMA control charts
[ISO 7870-8] Control charts ― Part 8: Charting techniques for short runs and small mixed batches
[ISO 9001] DIN EN ISO 9001, Quality management systems — Requirements
[ISO 11462-1] Guidelines for implementation of statistical process control (SPC) — Part 1: Elements
of SPC
[ISO 11462-2] Guidelines for implementation of statistical process control (SPC) — Part 2: Catalogue
of tools and techniques
•••••••••
•••••••••
[ISO 22514] DIN ISO 22514-2, Statistical methods in process management — Capability and
performance — Part 2: Process performance capability and process capability
statistics of time-dependent process models
[Schulze] Dietrich/Schulze: Statistical Procedures for Machine and Process Qualification,

Hanser-Verlag
[Shew 1931] Economic Control of Quality of Manufactured Product, D. Van Nostrand Comp., 1931
[Shew 1939] Statistical Method from the Viewpoint of Quality Control, Washington, The Graduate
School, The Deptartment of Agriculture, 1939
[VDA-4] VDA Volume 4, Securing the quality in the process landscape, methods, economic
process design and process control
[VIM] International Dictionary of Metrology (VIM), German-English version ISO/IEC Guide
99:2007, published by DIN German Institute for Standardization, Beuth Verlag Berlin
2021-02-10 - SOCOS
Vienna Zurich
•••••••••
•••••••••
Index
long ........................................................... 12
5 short ........................................................... 8
5 M .................................................................. 5 countable characteristic ............................... 47
counts ........................................................... 18
A
acceptance control chart......................... 35, 45 D
alarm limits ........................................ 10, 25, 45 delimitation factor ........................................ 35
AQDEF...................................................... 14, 45 determination ............................................... 45
assessment ....................................................41 discrete characteristic................................... 47
attribute ........................................................49 distribution parameters ................................ 26
2021-02-10 - SOCOS
average, moving ............................................31 disturbances ................................................. 26

documentation ............................................. 26
B
basic rules ........................................................ 7 E
entity ............................................................. 47
C
enumeration ................................................. 47
capability .......................................................45 estimate, estimation, estimator ................... 47
IMT equipment .........................................16 evaluation strategy ....................................... 14
index ..........................................................17 event codes............................................. 37, 38
machine.....................................................17
process ......................................................17 F
center point ....................................... 27, 29, 45 first-off / last-off part inspection .................. 22
characteristic ........................................... 46, 52
continuous ................................................46 I
countable ..................................................18 IMT equipment ............................................. 16
discrete......................................................47 individual data chart ............................... 29, 35
nominal .....................................................49 influence quantity ......................................... 48
ordinal .......................................................49 inspection ..................................................... 48
qualitative .................................................52 method ..................................................... 16
quantitative ...............................................52 strategy..................................................... 15
types ..........................................................18 intervention .................................................... 9
communication..............................................13
conformity .....................................................46 L
control limit ............................................................... 44
chart .................................................... 19, 47 long-term monitoring ................................... 12
criteria .......................................................24 lower specification limit ............................... 48
control limits
M
process-related .........................................28
recalculation..............................................14 management of characteristics .................... 15
tolerance-related ......................................35 maturity assessment..................................... 41
control loop measurand .................................................... 48
•••••••••
•••••••••
measurement ................................................48 R
process ......................................................48 random
result .........................................................48 cause......................................................... 52
uncertainty .......................................... 16, 48 sample ...................................................... 52
value ..........................................................48 range ............................................................. 53
measuring reaction plan ............................................. 8, 53
instrument ................................................48 requirements ................................................ 53
measuring and test equipment .................48 responsibilities .............................................. 14
system .......................................................48 responsiveness ............................................. 40
N resulting process distribution ....................... 53
review ........................................................... 41
normal distribution ......................................... 6
risks ............................................................... 11
2021-02-10 - SOCOS
O robustness .................................................... 15
rule of seven ................................................. 24
operation characteristic ................................40
original data..................................................... 6 S
P sample........................................................... 53
mean......................................................... 53
parameter................................................ 50, 51
size ...................................................... 19, 53
Pearson, distribution family ..........................30
sampling
performance index ........................................50
interval ..................................................... 21
population .....................................................50
unit ........................................................... 53
process ..........................................................50
Shewhart ......................................... 5, 6, 19, 53
average......................................................27
smooting factor ............................................ 32
capability index .........................................51
software .................................................. 14, 26
characteristic .............................................51
sorting inspection ................................... 11, 54
characteristic in control ............................51
SPC ............................................................ 5, 55
in control ............................................. 51, 54
characteristics .......................................... 15
quality .......................................................51
specification .................................................. 54
stable .........................................................54
interval ..................................................... 54
product ..........................................................51
limit........................................................... 54
product characteristic ...................................51
standard deviation of a sample .................... 55
Q statistic.......................................................... 54
quality statistical quality control .............................. 55
capability ...................................................52
T
characteristic .............................................52
target value ................................................... 55
quality capable process .................................52
tolerance ....................................................... 55
quality control chart ............................ 5, 23, 52
interval ..................................................... 55
quantity .........................................................52
zone .......................................................... 55
type ...........................................................52
trial run ......................................................... 26
value ..........................................................52
type-1 error................................................... 40
•••••••••
•••••••••
U V
upper specification limit ................................55 variance of a sample ..................................... 56
variation .................................................. 27, 56
variation control chart .................................. 33
2021-02-10 - SOCOS
•••••••••
•••••••••
2021-02-10 - SOCOS
Robert Bosch GmbH

C/QMM Tilsch
Wiener Strasse 42 - 46
70469 Stuttgart
Germany
Phone +49 711 811 - 0

www.bosch.com
•••••••••
•••••••••
2021-02-10 - SOCOS
•••••••••
•••••••••
2021-02-10 - SOCOS
Robert Bosch GmbH

C/QMM Tilsch
Wiener Strasse 42 - 46
70469 Stuttgart
Germany
Phone +49 711 811 - 0
www.bosch.com
•••••••••
•••••••••

Statistical Process Control SPC: Quality Management in The Bosch Group - Technical Statistics

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Statistical Process Control SPC: Quality Management in The Bosch Group - Technical Statistics

Uploaded by

Copyright:

Available Formats

2021-02-10 - SOCOS

Quality Management in the Bosch Group | Technical Statistics

7. Statistical Process Control SPC

Quality Management in the Bosch Group

Statistical Process Control

© Robert Bosch GmbH | 11.2020

5th Edition, November 2020

1st Edition, 09.1987

© Robert Bosch GmbH | 11.2020

2.5.3 Risks in Connection with Rework and Repair Processes ........................................ 11

© Robert Bosch GmbH | 11.2020 3

6 Formulas for Determining the Distribution Parameters ................................................................ 27

10 Example of an Event Code for Mechanically Processed Parts ....................................................... 37

© Robert Bosch GmbH | 11.2020 4

Corrective actions Take sample and

Man Machine Material

Method Environment Delivery

Fig. 1.1: Illustration of the SPC control loop

We differentiate between discrete (countable) and continuous (measurable) characteristics. Control

© Robert Bosch GmbH | 11.2020 5

The graph of the density function of a normal

distribution is the typical bell-shaped curve,

Information given by Number (example)

© Robert Bosch GmbH | 11.2020 6

2 Quality Control ― Application of the Control Chart

 responsibilities, in particular regarding the cooperation of production planning, production and

2.1 Basic Rules

1. Setting within the tolerance zone

5. In general, after a process intervention it is necessary to proceed with steps 1, 2, and 3.

© Robert Bosch GmbH | 11.2020 7

2.2 Necessary Documents

2.3 Short Control Loop ― Directly at the Machine

The reaction plan must show,

© Robert Bosch GmbH | 11.2020 8

2.4 Reaction Plan

SPC, control chart, control criteria:

© Robert Bosch GmbH | 11.2020 9

b) Single value outside the tolerance zone?

c) Alarm limit exceeded?

independent of the characteristic’s tolerance (natural control limits).

© Robert Bosch GmbH | 11.2020 10

2.5 Further Measures

2.5.1 Sorting Inspection

and what their status is.

2.5.2 Decision about the sorted Parts

2.5.3 Risks in Connection with Rework and Repair Processes

 Mix-ups of products or components incl. mixing of variants

© Robert Bosch GmbH | 11.2020 11

2.6 Long Control Loop ― Long-Term Analysis

2.6.1 Long-Term Monitoring

2.6.2 Long-term Analysis ― Organizational Questions

© Robert Bosch GmbH | 11.2020 12

2.6.3 Long-Term Evaluation ― Calculation, Representation, Documentation

2.6.4 Communication, Implementation and Pursuit of Measures

Management Production Planner Production

© Robert Bosch GmbH | 11.2020 13

2.6.5 Plant-specific Regulations and Solutions

Recalculation of Control Limits

relocation or design changes in the produc can have effects on the

Dealing with non-capable Processes

Review of Inspection Planning / Inspection Strategy

2.6.6 Software Support

© Robert Bosch GmbH | 11.2020 14