Computer-Aided Quality Control
INTRODUCTION
The quality control (QC) function has traditionally been performed using manual inspection
methods and statistical sampling procedures.
Manual inspection is generally a time-consuming procedure which involves precise,
yet monotonous work. If often requires that parts be removed from the vicinity of the
production machines to a separate inspection area. This causes delays and often
constitutes a bottleneck in the manufacturing schedule.
Inherent in the use of statistical sampling procedures is acknowledgment of the risk
that some defective parts will slip through. Indeed, statistical quality control attempts
to guarantee that a certain expected or average fraction defect rate will be generated
during the production/inspection process. The nature of traditional statistical QC
procedures is that something less than 100% good quality must be tolerated.
There is another aspect of the traditional QC inspection process which detracts from
its usefulness. It is often performed after the fact. The measurements are taken and the
quality is determined after the parts are already made. If the parts are defective, they
must be scrapped or reworked at a cost which is often greater than their original cost
to manufacture.
All of these various factors are driving the quality control function toward what we are
calling computer-aided quality control (CAQC). Other terms that have been applied to
describe this movement are "computer-aided inspection" (CAI) and "computer-aided testing"
(CAT).
The objectives of computer-aided quality control are ambitious, yet straight forward. They
are:
1. To improve product quality
2. To increase productivity in the inspection process
3. To increase productivity and reduce lead times in manufacturing
The strategy for achieving these objectives is basically to automate the inspection process
through the application of computers combined with advanced sensor technology. Wherever
technically possible and economically feasible, inspection will be done on a 100% basis
rather than sampling.
,TERMINOLOGY IN QUALITY CONTROL
Quality in a manufacturing context can be defined as the degree to which a product or
its components conform to certain standards that have been specified by the designer.
The design standards generally relate to the materials, dimensions and tolerances,
appearance, performance, reliability, and any other measurable characteristic of the
product.
Quality assurance (QA) is concerned with those activities which will maximize the
probability that the product and its components will be manufactured within the
design specifications. These activities should start in the product design area, where
the designer can make decisions among alternatives that might have quality
consequences. QA activities continue in manufacturing planning, where decisions
relative to production equipment, tooling, methods, and motivation of employees will
all have an influence on quality.
Quality control is concerned with those activities related to inspection of product and
component quality, detection of poor quality, and corrective action necessary to
eliminate poor quality. These activities also involve the planning of inspection
procedures and the specification of the gages and measuring instruments needed to
perform the inspections.
Statistical QC is generally divided into two categories: acceptance sampling and
control charts.
Acceptance sampling is a procedure in which a sample is drawn from a batch
of parts in order to assess the quality level of the batch and to determine
whether the batch should be accepted or rejected. Acceptance sampling is
based on the statistical notion that the quality of a random sample drawn from
a larger population will be representative of the quality of that population.
Control charts are used to keep a record over time of certain measured data
colected from a process. A company would use control charts to monitor its
own production processes. The central line indicates the expected quality level
of the process. The upper and lower control limits (UCL and LCL) are
statistical measures of the variation in the process which would be tolerated
without concluding that the process has erred. when these limits are exceeded,
it usually means that something has changed the process, and an investigation
should be initiated to deteimine the cause
, Both acceptance sampling and control charts can be applied to two situations in
quality control: fraction defects and measured variables.
In the fraction-defect case, the objective is to determine what proportion of
the sample (and the population from which it came) are defective. This is often
accomplished by a go/no go gage, which can quickly determine whether a part
is within specification or not.
In the measured-variable case, the object is to determine the value of the
quality characteristic of interest (e.g., dimension, resistance, hardness, etc.).
This requires the use of a measuring instrument of some kind (e.g.,
micrometer, ohmeter, hardness tester, etc.) and is normally a more time-
consuming manual process than the go/no go case.
Inspection is normally used to examine a component of a product in relation to the
design standards specified for it. For a mechanical component, this would probably be
concerned with the dimensions of the part. These might be checked with several go/no
go gages or they might be measured with a micro meter and other instruments. The
corfion situations that warrant inspection are:
Incoming raw materials
At various stages during manufacturing (e.g., when the parts are moved from
one production department to another)
At the completion of processing on the parts
Before shipping the final assembled product to the customer
INTRODUCTION
The quality control (QC) function has traditionally been performed using manual inspection
methods and statistical sampling procedures.
Manual inspection is generally a time-consuming procedure which involves precise,
yet monotonous work. If often requires that parts be removed from the vicinity of the
production machines to a separate inspection area. This causes delays and often
constitutes a bottleneck in the manufacturing schedule.
Inherent in the use of statistical sampling procedures is acknowledgment of the risk
that some defective parts will slip through. Indeed, statistical quality control attempts
to guarantee that a certain expected or average fraction defect rate will be generated
during the production/inspection process. The nature of traditional statistical QC
procedures is that something less than 100% good quality must be tolerated.
There is another aspect of the traditional QC inspection process which detracts from
its usefulness. It is often performed after the fact. The measurements are taken and the
quality is determined after the parts are already made. If the parts are defective, they
must be scrapped or reworked at a cost which is often greater than their original cost
to manufacture.
All of these various factors are driving the quality control function toward what we are
calling computer-aided quality control (CAQC). Other terms that have been applied to
describe this movement are "computer-aided inspection" (CAI) and "computer-aided testing"
(CAT).
The objectives of computer-aided quality control are ambitious, yet straight forward. They
are:
1. To improve product quality
2. To increase productivity in the inspection process
3. To increase productivity and reduce lead times in manufacturing
The strategy for achieving these objectives is basically to automate the inspection process
through the application of computers combined with advanced sensor technology. Wherever
technically possible and economically feasible, inspection will be done on a 100% basis
rather than sampling.
,TERMINOLOGY IN QUALITY CONTROL
Quality in a manufacturing context can be defined as the degree to which a product or
its components conform to certain standards that have been specified by the designer.
The design standards generally relate to the materials, dimensions and tolerances,
appearance, performance, reliability, and any other measurable characteristic of the
product.
Quality assurance (QA) is concerned with those activities which will maximize the
probability that the product and its components will be manufactured within the
design specifications. These activities should start in the product design area, where
the designer can make decisions among alternatives that might have quality
consequences. QA activities continue in manufacturing planning, where decisions
relative to production equipment, tooling, methods, and motivation of employees will
all have an influence on quality.
Quality control is concerned with those activities related to inspection of product and
component quality, detection of poor quality, and corrective action necessary to
eliminate poor quality. These activities also involve the planning of inspection
procedures and the specification of the gages and measuring instruments needed to
perform the inspections.
Statistical QC is generally divided into two categories: acceptance sampling and
control charts.
Acceptance sampling is a procedure in which a sample is drawn from a batch
of parts in order to assess the quality level of the batch and to determine
whether the batch should be accepted or rejected. Acceptance sampling is
based on the statistical notion that the quality of a random sample drawn from
a larger population will be representative of the quality of that population.
Control charts are used to keep a record over time of certain measured data
colected from a process. A company would use control charts to monitor its
own production processes. The central line indicates the expected quality level
of the process. The upper and lower control limits (UCL and LCL) are
statistical measures of the variation in the process which would be tolerated
without concluding that the process has erred. when these limits are exceeded,
it usually means that something has changed the process, and an investigation
should be initiated to deteimine the cause
, Both acceptance sampling and control charts can be applied to two situations in
quality control: fraction defects and measured variables.
In the fraction-defect case, the objective is to determine what proportion of
the sample (and the population from which it came) are defective. This is often
accomplished by a go/no go gage, which can quickly determine whether a part
is within specification or not.
In the measured-variable case, the object is to determine the value of the
quality characteristic of interest (e.g., dimension, resistance, hardness, etc.).
This requires the use of a measuring instrument of some kind (e.g.,
micrometer, ohmeter, hardness tester, etc.) and is normally a more time-
consuming manual process than the go/no go case.
Inspection is normally used to examine a component of a product in relation to the
design standards specified for it. For a mechanical component, this would probably be
concerned with the dimensions of the part. These might be checked with several go/no
go gages or they might be measured with a micro meter and other instruments. The
corfion situations that warrant inspection are:
Incoming raw materials
At various stages during manufacturing (e.g., when the parts are moved from
one production department to another)
At the completion of processing on the parts
Before shipping the final assembled product to the customer
