Journal of Manufacturing Systems 37 (2015) 149–163

Contents lists available at ScienceDirect

Journal of Manufacturing Systems

journal homepage: www.elsevier.com/locate/jmansys

Technical Paper

KBE rules oriented to resources management in coordinates

inspection by contact
S. Martínez-Pellitero a,∗ , J. Barreiro a , E. Cuesta b , A.I. Fernández-Abia a
a
University of León, Department of Mechanical, Informatics and Aerospace Engineering, Campus de Vegazana, 24071 León, Spain
b
University of Oviedo, Department of Manufacturing Engineering, Campus de Gijón, 33204 Gijón, Asturias, Spain

a r t i c l e i n f o a b s t r a c t

Article history: Different proposals have been presented along last years with the purpose of automating the process of
Received 16 December 2014 dimensional inspection planning with coordinate measuring machines (CMM) and touch probe technol-
Received in revised form 24 April 2015 ogy. These proposals have focused mainly on particular items such as analysis of accessibility, orientation
Accepted 28 July 2015
of the part or trajectory paths among surfaces or inspection points. The main limitation of these proposals
Available online 17 August 2015
is in the use of a predefined probe configuration, although current technology allows automatic probe
change for improving the inspection process. In this paper, detailed rules are explained for developing a
Keywords:
knowledge based engineering (KBE) application. These rules make it possible to incorporate the selection
Knowledge
KBE of the most appropriate configuration of sensor group with the basis of precision.
Inspection planning © 2015 The Society of Manufacturing Engineers. Published by Elsevier Ltd. All rights reserved.
Coordinates measuring
Contact inspection

1. Introduction for inspection process planning [1]. The Onto-Process ontology is

One of the most important activities when carrying out inspec-
tion process planning is the selection of necessary resources to
achieve results with the best possible precision. Once the capac-
ity of a coordinated measuring machine (CMM) has been proved
for inspecting tolerances associated to a particular part (uncer-
tainty/tolerance correlation), the first task is to select the sensor
configuration to be used. Then, the most appropriate probe ori-
entations are determined for this configuration. However, sensor
selection should be linked to linked to orientation of sensor group,
since there is a direct connection among precision achieved in
measurement, probe orientation and type of probe. In most cases,
these relationships are not always well-known for CMMs opera-
tors and it is common to prioritize operation speed and easiness
when selecting CMM sensors. In other cases, tacit knowledge of
skillful operators is difficult to formalize and represent; in this cir-
cumstance it is a decision of CMM operator without aid of expert
systems, routing sheets or orders derived from planning.
In this paper CMM operator knowledge is identified for storing
it in the form of knowledge rules. An ontology called Onto-Process
is used, specifically defined for process planning and, in particular,
∗ Corresponding author. Tel.: +34 987 291000x5358.
E-mail addresses: smarp@unileon.es (S. Martínez-Pellitero), jbarg@unileon.es
(J. Barreiro), ecuesta@uniovi.es (E. Cuesta), aifera@unileon.es (A.I. Fernández-Abia).
necessary for implementing a KBE system so that the process of
inspection planning can be automated as much as possible, both
for simple or complex parts.
Inspection activities are classified into two levels in the pro-
posed model of knowledge:
• Inspection macro-plan. This plan includes high-level activities
in the planning, such as the definition of the scope of inspection
(for example, contact or no contact inspection, hybrid inspection
with different inspection technologies, etc.), selection of CMM
and the right head and probe body, identification of part position-
orientation-fitting on the machine, classification of inspection
elements and high level sequence of operations.
• Inspection micro-plan. Micro-plan includes low-level activities
to specify with more details of the inspection operations to be car-
ried out for each inspection element (sampling strategies, probes
and probe modules selection, orientations, trajectories, number
of points, etc.).
The work in this paper is focused on one of the planning activi-
ties, the planning of resources necessary for inspecting mechanical
parts. In this paper developments belong to the micro-plan level.
Many CMM operators use always a common and practical sensor
configuration, without making a standalone approach to the mat-
ter. This is due to lack of time, resources and even the dispersion and
absence of knowledge. So, although very much knowledge exists in

http://dx.doi.org/10.1016/j.jmsy.2015.07.005
0278-6125/© 2015 The Society of Manufacturing Engineers. Published by Elsevier Ltd. All rights reserved.
150 S. Martínez-Pellitero et al. / Journal of Manufacturing Systems 37 (2015) 149–163
Fig. 1. Configuration of straight contact probes.
this scope, it is not taken usually into account for planning execu-
tion. The scope of the proposal is mechanical parts composed of
typical elements such as planes, cylinders, spheres and cones and
theirs combination. In this type of parts the recommended inspec-
tion is by contact, as concluded in analysis done in previous research
[2].
The sensor group is defined as the assembly of probe head,
probe and stylus. As the main objective of the research is to auto-
mate inspection processes, it is required that all components of
the sensor group be easily and automatically interchangeable. For
this reason, only automatic CMMs with motorized probe heads and
interchangeable probes will be taken into consideration. Most of
the probes that enable automatic change (cinematic or electronic
type) are composed of at least two modules: main module or body
of the probe, and an additional module or stylus module. These
modules are interchangeable and configurable to use different trig-
ger forces in X–Y–Z axes. The most common thread used in this
type of devices is M2 or M3 metric. In inspection by contact the
probe is the main link with the part to inspect and, therefore, it
is critical to make the best selection in order to achieve the maxi-
mum precision. In the case of machines with motorized head, the
most common probe configuration is straight configuration (Fig. 1).
The straight configuration is generally composed of a single sty-
lus ended in a spherical tip, which has replaced the star stylus
configuration widely used years ago. The reason is based on recom-
mendations for achieving better precision: whenever possible the
simplest composition of straight probe would be chosen instead of
using a more complex composition [3]. Accordingly, this paper only
focuses in selecting straight probes and probe modules for discrete
contact inspection.
Few researchers have considered the automatic selection of
the sensor assembly. Although some of them have considered
the configuration of stylus dimension and type of tip, they have
not deepened in other essential features such as the probe type
(discrete or continuous type), the probe technology (resistance cin-
ematic, electronic or piezoelectric) or the stylus tip type (shape,
material and dimensions). Corrigal and Bell [4] proposed a method-
ology for determining the configuration of a ball stylus so that it
could carry out all the inspection cycle. They considered a standard
stylus configuration of 20 mm stem length and 2 mm ball diame-
ter with a probe of 35 mm length and 13 mm diameter. They used
the concept of approach orientations (PAD) to analyze the risk of
collision; in case of collision the first action is to add extensions to
the probe, the second action is to use a larger stylus. Moroni et al.
[5] considered a star assembly configuration defining its orienta-
tion, dimensions and structure. They presented an expert system
Fig. 2. Brief summary of tasks necessary for inspection resources selection.
Fig. 3. Representation of the test part.
to generate the probes configuration that allow to inspect all the
part starting from a series of geometric restrictions and heuristic
rules that condense the human experience in relation to this prob-
lem. The base of knowledge is composed of five groups of related
rules (islands of knowledge) and an inference motor. The inference
motor calls the five islands consecutively, so that each island con-
figures one element of the sensor group, always for a star probe
configuration.
With regard to probe orientation, some authors have analyzed
this parameter to carry out the inspection of mechanical parts.
However, several differences are remarkable with regard to the
proposal presented in this paper. Some authors have focused on
determining all possible orientations to inspect an element [6] or
simply one contact point on an inspection element [7,8]. In all cases,
works in the state of the art use a predefined configuration of the
sensor assembly for inspecting the complete part. Nevertheless, to
use the same probe configuration for all the part is not always possi-
ble. Reasons are clear: firstly, due to the differential features of each
inspection element that may require different strategies; secondly,
using the same configuration for all part inspection would imply to
identify and change the probe orientation to avoid collisions. This
way of operating presents two main drawbacks:
(a) Changes in probe orientation reduce repeatability of measure-
ment and, therefore, it is not advisable when acquiring points
on the same inspection element.
(b) A probe orientation can be the most appropriate for accessing
an inspection element, but maybe it is not the most favorable in
terms of precision, mainly when stylus is neither perpendicular
nor parallel to probe body.
Therefore, in our work the main criterion for selecting sensor
assembly and its orientation is to achieve the maximum precision
in the measurement of every element. The main focus of this paper
is to collect and model the knowledge related to the automated
decision tasks about the best probe selection into the CMM scope
(material, length, type of sensor, configuration assembly, etc.). This
knowledge has been elicitated from several sources as expert’s
opinion, research projects, manufacturer’s guides or handbooks. In
the case that there are not available rules or explicit knowledge,
new rules are proposed in an automated decision environment.
2. Determination of knowledge rules
First of all, it is important to indicate the scope and limitations
of the rules which will be identified and modeled in this section.
The scope is inspection by discrete contact using indexed head and
straight stylus. Also, it is mainly oriented to traditional machined
parts, with any mixture of prismatic and/or cylindrical features but
 S. Martínez-Pellitero et al. / Journal of Manufacturing Systems 37 (2015) 149–163 151

Fig. 4. IDEF0 diagram for A212 activity: Determine the contact strategy for each inspection element.

without free form surfaces. This kind of machined parts and this
sensor configuration represent a very common case at the man-
ufacturing of mechanical components at metalworking industry.
Otherwise, it can be said that most of the free form surfaces do not
require high precision inspection and they are used in applications
more esthetic than functional.
Every dimensional or geometrical indication in the CAD model
is considered an inspection item as expressed in reference [9]. An
inspection item contains information of tolerances and affected
elements. These affected elements will be called in advance inspec-
tion elements. The part is composed of simple and compound
features. A simple feature is based on a single canonical surface
(plane, cylinder, sphere, cone or free form surface) whereas a com-
pound feature is composed of two or more canonical surfaces. In
consequence, an initial decomposition of the part into elementary
geometries is required. Only a simple surface affected by some type
of dimensional/geometrical indication or tolerance is considered
as an inspection element. Fig. 2 shows a brief summary of activ-
ities necessary to afford selection of inspection resources. In this
figure basic elements are canonical or simple geometry in which
the part can be decomposed, regardless they are affected or not by
tolerances.
Fig. 3 shows the test part for using as demonstrator of the
proposed methodology. This part was defined with elements that

Table 1
Items and inspection elements.

Inspection item Type Description Nominal value Tolerance zone Inspection elements Datums

Item001 Geometric Shape (Flatness) – 0.05 E1 –

Item002 Geometric Shape (cylindricity) – 0.15 E8 –
Item003 Geometric Location (position) – 0.05 E16 –
Item004 Geometric Location (position) – 0.05 E17 E1
Item005 Geometric Orientation (angularity) 45◦ – E11 E1
Item006 Geometric Orientation (perpendicularity) – 0.05 E4 E2
Item007 Geometric Orientation (parallelism) – 0.05 E6 E1
Item008 Dimensional Distance 20 0.15 E5, E7 –
Item009 Dimensional Distance 70 0.02 E7, E10 –
Item010 Dimensional Diameter 10 H7 E10 –
Item011 Dimensional Diameter 40 – E8 –
Item012 Dimensional Diameter 25 0.10 E12–
Item013 Dimensional Distance 35 0.10 E17, E19 –
Item014 Dimensional Diameter 6.65 – E16 –
Item015 Dimensional Diameter 6.65 – E17 –
Item016 Dimensional Distance 35 0.10 E16, E3 –
Item017 Dimensional Distance 25 0.10 E16, E18 –
Item018 Dimensional Distance 25 0.10 E17, E18 –
Item019 Dimensional Distance 5 0.05 E11, E13 –
152 S. Martínez-Pellitero et al. / Journal of Manufacturing Systems 37 (2015) 149–163
Fig. 5. Dimensional attributes of a straight stylus.
force to change the probe configuration. The part has different
features as grooves, holes or threaded holes. These features must
be split up into their elements as shown in Fig. 4. Sixteen of
the twenty elements have dimensional or geometrical specifica-
tions. The elements not affected by tolerances are E9, E14, E15 and
E20 as indicated in the figure. Table 1 summarizes the nineteen
inspection items with the sixteen inspection elements and their
attributes.
Fig. 6. Proposal for selecting stylus tip type (images of Renishaw).
Each inspection element has been assigned to an internal pre-
cision as part of the proposed methodology. Values for precision
are quantitative and correspond to low, good or high precision. The
value to apply to each element is based on the element tolerances
and the CMM precision. The assignment of precision to each ele-
ment is necessary for subsequent selection of the probe tip and
for determining the number and distribution of inspection points,
and also for selecting the technology of inspection to use. In refer-
ences [10,7,1,11] some rules and algorithms are proposed for this
task. Table 2 shows the classification of elements according to the
assigned internal precision.
The PH10M Renishaw indexed head was already selected in
prior activities of the macro plan. This head has an angular res-
olution of 7.5◦ between consecutive positions. It consists of two
axes A and B. Axis A rotates in the range 0–105◦ whereas axis B
does in the range 0–180◦ . Since the test part has elements cat-
egorized with high precision, the probe chosen is an electronic
 S. Martínez-Pellitero et al. / Journal of Manufacturing Systems 37 (2015) 149–163 153

Table 2
Classification of inspection elements related to the assigned precision. Planar ele-
ment (PLE), cylindrical element inner-outer (ICyE-OCyE).

Element Type of element Precision Element Type of element Precision

E1 PLE High E10 ICyE Good

E2 PLE High E11 PLE Good
E3 PLE High E12 ICyE Low
E4 PLE Low E13 PLE Low
E5 PLE Low E16 ICyE Good
E6 PLE Low E17 ICyE Good
E7 PLE Good E18 PLE Low
E8 OCyE Good E19 PLE Low

Fig. 7. Machining marks in turning and milling processes.

type TP200, from Renishaw. This type can use probe modules that Therefore, a simplification to the methods used by other authors
support larger stylus than cinematic probes with more precision. is proposed in order to use the knowledge that relates probe ori-
Besides geometric and dimensional probe attributes, there is the entation to sensor configuration. This relationship is not taken into
Stylus Maximum Length that can support (100 mm), the metric of account in other works since authors use predefined-fixed sensor
stylus-probe joint (M2) or styli storing stations and the number configurations. This fact does not reflect the industrial practice with
of available docks. These attributes are necessary when selecting automated CMMs, where several tips and styli or even the probe
probes/styli and when defining their orientations. are automatically changed during the execution of an inspection
Probe selection is performed in the activity of the Micro-plan program. Our proposal considers inspection planning from a prac-
called “Determine the contact strategy for each inspection element” tical point of view, so that results can be applied to the industry. It
(Fig. 4). The proposed methodology is divided into two parts: (i) is therefore close to inspection practice.
first the stylus tip is determined, its type, material and dimensions
(activity A2121); (ii) then the length of stem is selected (activity
2.1. Rules for determining probe parameters for each inspection
A2126), which allows determining the probe module necessary
element
for inspecting each element. As indicated in Fig. 4, these two sub-
activities are fully related to other decisions such as the number
A straight stylus is composed of a tip and a stem. This section
and distribution of contact points in each element [11], the most
explains rules for determining the tip type and its dimensions as
adequate probe/stylus orientation or the approach and retraction
well as the stem characteristics and the probe module (A2121 and
distances. These parameters are determined in sub-activities of
A2126 activities).
A212.
Fig. 5 represents the dimensional attributes used by manufac-
Also, determination of stylus length and, therefore, configura-
turers to describe the design of a straight stylus. These attributes
tion of the sensor assembly has a direct influence on the potential
will be referenced in the following subsections.
probe orientations for inspecting each element. A difference is
Manufacturers provide general rules for determining sensor
made between approach directions (always it should be the most
assembly specifications, but when applying to practical cases some
perpendicular as possible to the contact point on the inspection
problems appear related to actual dimensions and configuration of
surface) and probe orientations. The last ones have been designated
inspection features. In this paper, to define practical rules consider-
with the PAO term (probe approach orientation), which is a more
ing both manufacturer’s recommendations and feature dimensions
adequate name that the used by Corrigal (PAD, probe approach
is tackled. This is the main objective of the following subsections
direction) [4] and used later on by other authors [9,12,13]. All these
and figures.
authors consider that each element can be inspected using at least
one of the 5 directions of the CMM reference system (X+, X−, Y+,
2.1.1. Determination of the stylus tip type (Activity A2121)
Y−, Z+), which are called PADs, trying to reduce the number of final
The main tip types are ball, disk, cylindrical and hollow.
orientations when inspecting prismatic parts. The proposal in this
Although the ball tip is adequate for inspecting most of surfaces of
paper follows the same principle, but additionally to these five ori-
a prismatic part, in special circumstances a tip change is necessary.
entations we consider the possibility of adding new orientations
In general, the following rules are applied:
when the element is not oriented with the CMM reference system.
In the test part the element 11 is an inclined plane, therefore it is a
• Cylindrical: for measuring sheets or thin parts where precise con-
candidate for using additional orientations.
The probe selection proposal is directly linked with the orien- tact point cannot be assured with a ball probe. It is also suitable
tations of the probe when approaching each contact point on the to measuring inner diameters and axis of threaded elements and
element. It follows the procedure: holes.
• Disk: for measuring grooves, notches and slots in holes where a
star probe cannot get into.
1. First, analysis and pre-selection of probe orientations (PAO) is
done following precision criteria and minimization of orienta-
tion changes. In this sense, an additional criterion is settled down
in order to make contact on all points of the same element using
the same probe orientation.
2. Second, selection of the most adequate probe configuration for
acquiring a set of points on each element. This configuration
includes stylus tip type, tip dimensions and probe length.
3. Third, checking of orientations and configurations. If necessary,
change of configuration or orientation. This step is explained in
Section 2.2.1. Fig. 8. Selection of tip quality grade for ball tips.
154 S. Martínez-Pellitero et al. / Journal of Manufacturing Systems 37 (2015) 149–163

Fig. 9. Proposal for determining dimension of cylindrical and ball tips for measuring planes and cylinders.

• Hollow ball: for measuring deep holes or non uniform and rough
surfaces.
Therefore, a series of parameters can be identified for determin-
ing tip type and its dimensions:
• Height or thickness of the element to measure. 3 mm thickness is
considered as the minimum value to assure an element can be
inspected with a ball stylus. Most of the sheet manufacturers
establish a value between 2.1 mm and 3.5 mm to discriminate
thin sheet from thick sheet and, therefore, an intermediate value
can be used as reference. If an element is less than 3 mm thick-
ness or height, it would be appropriate to use a cylindrical tip or
even, in the case of a groove, a disk tip.
• Superficial roughness of the element to be measured. Roughness
arithmetic average (Ra) is the most widely used parameter [14]
to determine if a surface can be described as smooth or rough. Ra
value is classified in intervals or roughness classes between N1
and N12 [15]. Based on this criterion and considering that there is
no recommendation, experience in the field indicates that finish
 S. Martínez-Pellitero et al. / Journal of Manufacturing Systems 37 (2015) 149–163 155

Table 3
Stylus tip characteristics for each inspection element.

Element Tip type Tip diameter Tip material Tip grade Element Tip type Tip diameter Tip material Tip grade

E1 Ball 4 Ruby 3 E10 Ball 4 Ruby 5

E2 Ball 4 Ruby 5 E11 Ball 4 Ruby 5
E3 Ball 4 Ruby 5 E12 Ball 2.5 Ruby 5
E4 Ball 4 Ruby 5 E13 Ball 2.5 Ruby 5
E5 Ball 4 Ruby 5 E16 Cylindrical 3 Ruby –
E6 Ball 4 Ruby 5 E17 Cylindrical 3 Ruby –
E7 Ball 4 Ruby 5 E18 Ball 4 Ruby 5
E8 Ball 4 Ruby 5 E19 Ball 4 Ruby 5

is very good up to the N7 class, good up to N9 class and from N10
(Ra = 12.5 ␮m) it would be advisable to use a hollow ball stylus
due to irregularities in the surface. Values of superficial rough-
ness are not always known, but there are characteristic values
associated to every manufacturing process as a function of sev-
eral factors, as indicated in Oberg et al [16]. These values can be
used as reference for selecting the most adequate stylus tip.
• Deep hole element. A hole and similar features can be considered
deep when its length is higher than 5 times its diameter [17,18].
• Geometry and dimension of special stylus tips. There is limited vari-
ability for special tips in typical M2/M3 metrics as the ones used
in common CMMs. For hollow ball tips, most of manufacturers
offer only a 18 mm diameter stylus for M2 metric and a 30 mm
diameter for M3 metric. Consequently, when selecting a hollow
ball tip to filter the effect of roughness, the minimum diameter
of a hole or cylinder has to be greater than 20 mm. For disk tips,
disk diameter is related to disk height. Although there are disks
up to 6 mm height for M3 metric, their diameter increases consid-
erably around 64 mm. The election of a disk of these dimensions
is only justified in case of measuring large cylinders having inte-
rior grooves with high depth. For cylindrical tips, diameter goes
from 1 mm to 4 mm for M2 and M3 metrics, according to different
manufacturers such as Renishaw [19] and Zeiss [20].
The rules for determining the most adequate tip type are shown
in Fig. 6 in a flow chart. As observed in this figure, selection of tip
type is based on precision and geometry criteria. When using these
criteria there is no possible to select a specific tip type, the only
chance is to choose the standard ball tip type.
According to these rules, two different tips are necessary for
inspecting the test part: ball and cylindrical.
2.1.2. Determination of stylus tip dimensions (Activity A2121)
Tip size influences selection of the stem. When stem diameter is
twofold the measuring error diminishes by sixteen [21]. Therefore,
choosing the maximum stylus tip size is important for the following
reasons:
Fig. 10. Determination of stylus tip material.
(a) To minimize the impact of superficial roughness in the precision
of measurement.
(b) To reduce the probability of collisions between the stem and
the part (shanking effect).
(c) To increase the stylus rigidity, since higher tips allow stems of
higher diameter.
The following parameters have been identified to establish the
rules in this item:
• Tip type and availability. Manufacturers provide ball diameters
that go from 0.3 mm to 8 mm for M2 metric, and from 0.5 mm to
12 mm for M3 metric. To unify criteria and to reduce the num-
ber of possible configurations, it can be established that for single
elements not in conflict with other elements in the part, the most
appropriate ball diameter is 4 mm, which is justified by two rea-
sons:
(a) It is a very widely diameter in CMMs calibration activities
(usually between 4 and 5 mm).
(b) Main manufacturers provide this diameter for all stylus
lengths up to 100 mm, although with different stem materials
for improving rigidity. Consequently, the final configuration
does not require additional extensions and it is simpler.
Later selection of stem material is important to choose
maximum lengths without introducing additional deflexion
elements.
As already commented, in the case of specific styli the man-
ufacturer’s variability is minimum. For the case of M2 or M3
metric, cylindrical tips have a maximum diameter of 4 mm, with
an effective working length that is function of diameter. There-
fore, selection of the tip diameter for inspecting a threaded hole
is predetermined by the effective working length of the stylus.
Disk tips and hollow ball tips have specific sizes for diameter and
height. All the manufacturers provide a reduced set of very similar
options.
One aspect to consider is the effect of feedrate in machined
parts. If the feedrate used when machining a surface is known,
then the stylus tip radius should be greater than the feedrate
value in order to avoid measurement errors [3]. Although there
are several parameters that affect the final quality of machined
surfaces, cutting speed (Vc ) and feedrate (fn ) are those with the
most influence on feed marks on the machined surface (Fig. 7).
Therefore, these parameters (Vc and fn ) must be taken into
account when defining this rule.
As indicated in [16] for several materials, in turning the feed
rate limit used to be 0.9 mm/rev whereas in milling operations
the maximum value used to be 1 mm/tooth. Consequently, the
influence of feedrate on measurement only will take place in case
of ball tips with very small diameters.
• Dimensions and geometry of the element to be measured. Besides
the stylus tip type, it is also necessary to consider the dimensions
of the element to be measured. In many occasions dimen-
sions of the elements to be measured and their orientation and
156 S. Martínez-Pellitero et al. / Journal of Manufacturing Systems 37 (2015) 149–163

Fig. 11. Determination of valid PAO orientations and probe length following precision criteria for each inspection element.

location in the part constrain the size of the styli. For example,
in the case of measuring an inner cylinder, the stylus tip diam-
eter will be constrained by the cylinder diameter. In the case
of a groove the opening or groove thickness will constrain the
maximum tip size.
Considering an automatic criterion to assure results, the fol-
lowing rule has been determined. For cylinder diameters equal
or greater than 8 mm the stylus tip diameter will be 4 mm. For
lower cylinder diameters, the main criterion is to constrain the
stylus tip diameter to a value half the cylinder diameter; in the
 S. Martínez-Pellitero et al. / Journal of Manufacturing Systems 37 (2015) 149–163
Fig. 12. Results of applying stylus length rules to the test part.
most unfavorable case, constrain the maximum tip diameter to
three quarters. This way, an automatic rule can also be defined
for approach and retract distances.
• Precision of the element to be measured. On the other hand, the
most common ball tips are made of ruby and they are manufac-
tured with different levels of precision or “grade”, which indicates
the ball shape deviation with regard to an ideal sphere. The grade
varies between 48 (the lowest grade of precision) and 3 (the
highest) according to DIN-5401-11:1993 standard. As more pre-
cise the CMM as more significant the effect of the ball grade.
Manufacturers use to work with balls of grade 3 and 5. The rule
here is to choose balls of grade 5 as common solution; for spe-
cial precision applications, i.e. when the element is classified with
high precision as aforementioned and when it also has assigned
shape tolerances, the rule is to choose a grade 3 ball if available.
Fig. 8 shows the proposal for selecting the quality grade for the
case of ball tips.
• Cost and time criteria. A rule is defined in such a way that inspec-
tion elements in the same feature are measured with the same
probe configuration. For example, in the measuring of three
planes that compose a groove, dimension of the most critical tip
is chosen to measuring the three planes.
Fig. 9 shows the rules for determining the size of cylindrical and
ball tips, both for measuring planes and cylinders.
Tip dimensions obtained according to the proposal do not
always match with commercial components. If this is the case, for
157
example for a ball tip of 2.7 mm diameter which does not exist
in the market, the selected tip would be the closest inferior from
those offered by manufacturers or available in the plant. Selecting
a superior tip dimension would require the operator supervision
and it limits the system autonomy.
2.1.3. Determination of tip material (Activity A2121)
Ruby is the most common material for stylus balls, mainly due
to its high hardness and rigidity. However, there are some appli-
cations in which other materials are required [19]. It is the case of
continuous contact inspection of parts made of aluminum alloys
that lead to adhesive wear, or other abrasive materials that pro-
duce abrasive wear. In these situations, tungsten carbide or ceramic
balls are used instead of ruby, but only for particular diameters and
lengths.
In the case of unusual styli there are few options for choosing
tip material. For example, in the case of hollow ball tips the only
available material is ceramic. In the case of cylindrical and disk tips
the materials are steel or ruby but only for specific dimensions.
Therefore, bearing in mind that in CMM inspection using dis-
crete contact points the stylus tip only gets in contact with the part
surface during a very short time, the rules expressed in Fig. 10 are
defined. Since variability of ruby tips are very much wider than for
other materials, selection of this material is of priority.
Table 3 shows the characteristics for stylus tips (type, diameter,
material and grade) for the test part considering the aforemen-
tioned rules.
158 S. Martínez-Pellitero et al. / Journal of Manufacturing Systems 37 (2015) 149–163

Fig. 13. Determination of stem material and/or stylus extension.

2.1.4. Determination of stylus length (Activity A2126) Two new related criteria have been taken into account to calcu-
Identification of stylus length and analysis of potential PAOs late the stylus/probe and its orientation:
have been carried out for each inspection element. Five directions
of the CMM reference system (X+, X−, Y+, Y−, Z+) together with 1. On one hand, as larger the probe as higher its bending during
geometry and orientation of the inspection element have been con- the contact in the measurement and the delay of trigger signal;
sidered for it. The approach direction always should be normal hence the error related to measurement is also higher. This fact is
to the surface of the element on the contact point. However, the especially important when using kinematics probes larger than
approach direction must not be confused with the probe orienta-
tion during the approaching movement.
Subsequently, a prior and exhaustive analysis has been accom-
plished about the possible PAOs for each element and a new
concept called “PAO dimension” has been introduced. The PAO
dimension is defined for each inspection element – PAO pair. It
represents the minimum length of the sensor assembly to avoid
collisions with the inspection element or adjacent elements for
a particular orientation. Each PAO dimension is calculated from
the PAO-CAD dimension, which is derived from the CAD model of
the part (dimension of the element in each PAO). The attributes
considered for this analysis include dimensions of probe, head
and CMM shank, all of them extracted from the resources already
identified in the macro-plan. Then, considering the final length of
the sensor assembly the following rule is checked: trace a semi-
infinite straight line with origin in each contact point on the
element (considering the approach and retraction distances) fol-
lowing each PAO; then check is there is intersection between this
line and the other elements in the part. In case of intersection, this
PAO orientation is deleted. In the end, a final set of valid PAOs is
obtained. Fig. 14. Parameterization of sensor assembly.
 S. Martínez-Pellitero et al. / Journal of Manufacturing Systems 37 (2015) 149–163
Fig. 15. Options to configuration/orientation setting to avoid collisions.
30 mm. The reason is that the force required for triggering the
sensor during the contact buckles the stem and this buckling is
proportional to the length. A half reduction in length reduces
by four the measurement errors [3]. Consequently, to choose the
minimum possible length is crucial.
2. On the other hand, when working with other type of probes
which allow larger styli, the probe orientation will take pref-
erence over the length, so that approach direction to the contact
point will be normal to the probe body. The second option for
approach orientation will be parallel to the probe body and
finally the rest of orientations.
PAO dimension has an equivalence in manufacturer catalogs
and, therefore, in resource databases of any laboratory. Commer-
cial styli with ball tip are classified in intervals of length. Length is
measured from the ball center to the threaded joint as seen in Fig. 5.
In some occasions the total length is coincident with the effective
working length (EWL), but not ever, it is higher in many occasions.
Length interval defined by manufacturers is related to ball diam-
eters. For ball diameters lower than 3 mm there are only styli up
159
to 50 mm length for M2 and M3 metric. For ball diameters greater
than 4 mm the range of lengths is wider, running up to 100 mm
using more rigid stem materials.
The calculated PAO-CAD dimension must correspond with
EWLmin. The task is then to look for available styli in the resource
database that match a specific tip diameter having the same or
higher EWL than EWLmin. The selected stylus will have a Total
Length attribute that will be the equivalent PAO dimension. For
example, for a 4 mm ball stylus, if the PAO-CAD dimension is 33 mm
this value will be the minimum effective working length that should
have the stylus (EWLmin). As this value is not found in commercial
catalogs, the immediately superior value will be chosen as EWL,
in this case 35 mm, and the corresponding total length (40 mm) as
PAO dimension for this element – probe orientation pair.
For disk or hollow ball tip styli, limitations are more severe
with regard to length, therefore constraining available configu-
rations. Hollow ball and disk styli used not to include the stem
in their configuration. Consequently, it is always necessary to
add extensions to complete the PAO-CAD dimension. For cylin-
drical tip styli, it is possible to find geometrical and dimensional
160 S. Martínez-Pellitero et al. / Journal of Manufacturing Systems 37 (2015) 149–163
Fig. 16. Results of applying Activity A213.
configurations within the very reduced availability range to avoid
the use of extensions.
Fig. 11 shows the proposal for selecting the most adequate PAO
orientation for each element. The selection depends on the probe
type used for the inspection.
The rules up to now are valid regardless of probe manufactur-
ers. For the test part used in this paper the selected manufacturer is
Renishaw, leader in the sector with a wide catalog. For threaded ele-
ments E16 and E17, a cylindrical tip stylus of dimension PAO-CAD
equal or higher than 25 mm is not available in manufacturer cata-
log. For diameters of cylinder of 3 mm, the maximum EWL length
found is 22.5 mm, inferior to the one required, for what the mini-
mum extension is looked for to adding to the stylus. In this case, an
extension of 5 mm length is chosen.
The maximum stylus length supported by the selected probe
body is 100 mm (including extensions). Therefore, element number
E10 cannot be measured with this probe body. On the other hand,
the PAO can be neither changed since diameter of the element is
too small. In this particular case operator intervention would be
required, for example to decide to use a shorter stylus and use two
different orientations for inspecting the element. Fig. 12 shows the
results of applying these rules to the test part.
2.1.5. Determination of stem material and stylus extensions
(Activity A2126)
In some occasions a selected stylus can have different stem
materials. In this case additional criteria should be applied for
selecting the most appropriate material, as shown in Fig. 13.
The selection of stem material is a commitment between stiff-
ness and weight. An excess of weight in stylus can produce false
triggering due to inertial mass during acceleration phase in CMM
movement [3]. For selecting the most adequate materials, manu-
facturers usually use the material specific stiffness (rate between
Young’s module and density, E/). Tungsten carbide is the most
rigid but its high density leads to a very low specific stiffness. Con-
sequently, tungsten carbide stems are only advisable for lengths
inferior to 50 mm which allow small ball tip diameters. On the
other hand, ceramic materials offer high specific stiffness and
good thermal stability. This material is advisable when using styli
lengths between 30 and 50 mm and when temperature fluctuation
is possible, but they are only valid for ball diameters higher than
3 mm.
The best material is carbon fiber because it is more rigid and
lighter than ceramic. However they have an important limita-
tion in stem diameter, since they are manufactured only with
diameters higher than 3 mm. This inferior limit leads to use them
with ball diameters superior to 4 mm. Carbon fiber is stable with
temperature and therefore suitable for measuring in production
environment. Steel is the worst of all considered materials due to its
high density and low stiffness, which would cause important bend-
ing when using large styli. This material should be used with stylus
lengths inferior to 20 mm, although range of ball diameters is wide
in this case. Criteria for choosing stylus extensions, if necessary, are
also included in Fig. 13.
 S. Martínez-Pellitero et al. / Journal of Manufacturing Systems 37 (2015) 149–163
Fig. 17. Final configurations for inspecting the test part.
2.2. Selection of probes for full part inspection
The Renishaw database has been used as reference with M2
metric, which is the one required for TP200 probe without adding
couplers, with would reduce accuracy of the sensor assembly. Fur-
thermore, the CMM is supposed to be located in a laboratory with
controlled environmental conditions (temperature and humidity).
Validation of selected configurations is included in this section.
Collision with part elements is checked and optimization of probe
configuration groups is indicated.
2.2.1. Validation of probe assembly configuration free of collisions
An analysis of collisions is performed in the CAD for each
inspection item without considering the sequence of intermediate
movements among elements. The starting point is the set of probe
orientations valid for each inspection element, selected by criteria
of accuracy and availability. If a collision is detected for a particu-
lar probe orientation, it could be solved using different procedures
depending on the type of collision:
1. Go back and choose another PAO in the set of valid PAOs which
were discarded using the accuracy criterion.
2. Consider to change stylus tip size.
3. Add probe extensions.
4. Add stylus extensions to increase the length.
5. Change probe orientation or ask for operator mediation.
161
The first option is the most recommended provided there are
valid PAOs perpendicular or parallel to the normal vector. The
change of tip dimension may also solve a collision with the stem; for
example, an increment in ball diameter allows a higher clearance
with the stem. If these former options are not available, manu-
facturers recommend first add extensions to the probe since they
have no measurable effect on the performance. On the contrary, the
increment of stem length affects to stylus stiffness and, hence, to
probe sensitivity. Finally, the last option is to change probe orien-
tation, which can affect the repeatability.
The necessity to use probe and stylus extensions requires
parameterizing the sensor assembly as shown in Fig. 14, where
LP = length of probe (total length); LPB = length of probe body;
LPM = length of probe module; LPE = length of probe extension;
LS = length of stylus (total length); LSS = length of stylus stem;
LSE = length of stylus extension; LH = reference length of head;
LSA = length of sensor assembly. These lengths are of importance
to detect and avoid collisions and to identify the corrective action
to apply, as it is indicated in the next paragraphs.
In order to avoid collisions, firstly it is necessary to determine the
element of the sensor assembly which is colliding with the inspec-
tion part (Fig. 15). If collision occurs with the assembly head, the
proposal is to add a probe extension to avoid it. In the unusual
case that no probe extension is available, a request for changing
the head should be sent feedback to the Macro Plan. If collision
occurs with the probe, the proposal is to add stylus extensions
to avoid it, whenever the total length keeps under the maximum
length allowed by the probe. If collision persists, a probe change
162 S. Martínez-Pellitero et al. / Journal of Manufacturing Systems 37 (2015) 149–163
Fig. 18. Diagram of UML classes for inspection resources in onto-process.
is proposed and the last option is to try a new orientation. Finally,
collision can occur with the stylus. If collision is with the stem, the
proposal is to increase the tip diameter. In case of collision with
the tip, the proposal is to reduce tip diameter and to reduce touch
speed.
In this case, the collision checking performed in each inspec-
tion point detects collision in the E11 element. Consequently, the
rules set is used again and the next PAO valid orientation is cho-
sen for this element, A45 B-90, which also has a shorter probe. This
configuration does not produce collision; consequently it is cho-
sen for inspecting E11 element and a new list of selected targets is
generated, as shown in Fig. 16.
Next, selection of final probe configurations and orientations
is carried out for the full part. So, inspection items are grouped
using as criterion the identified probe configurations. Probe groups
are ordered from the smallest to the highest in size and from the
shortest to the largest in length probe, starting with ball tips and fol-
lowing with special tips. Each probe configuration, Cp, {p = 1 . . . n},
leads to a measuring operation (MOm), where m represents the
operation execution order, n is the number of configurations such
that m = {1 . . . n}. There will be so much measuring operations
(MOm) as probe configurations (Cp) and each MOm begins with
a configuration change.
For the test part there are 6 different probe configurations
(Cp = 6), as seen in Fig. 17. As the automatic changer device has
6 docks, these 6 configurations are considered valid and planning
process continues.
3. Structuring of knowledge
Using the identified knowledge related to selection of these
inspection resources, onto-process ontology can store this knowl-
edge orderly. Firstly, in an informal way using developed forms
associated to the ontology [1]. Secondly, in a formal way using
UML as representation language. Onto-process has been devel-
oped using MOKA [22] as reference, but in the same manner
that in the informal representation were introduced modifications,
the formal representation also requires of a series of improve-
ments. In MOKA there are two models: Product Model and Design
Process Model. These models are unsatisfactory for representing
knowledge related to production processes. In consequence, a mod-
ification is proposed to add a new model called Resource Model and
to rename the Design Process Model as Process Model, so that for-
mal knowledge is better structured. In each model the knowledge
is seen from different points of view, which helps to organize better
its representation.
The Product Model is adequate for representing the WHAT of a
product; the Process Model represents the HOW and the Resource
Model represents the WITH. It is true that the Product Model can be
used also to represent resources, but in many occasions selection of
a suitable resource is not a design concern and even it has not any
related process of production. In this sense to extract the resources
out of the Product Model was considered since it contributes to
knowledge structuring. Each model uses graphical representations
of the elements that compose it. The diagrams are classified in views
 S. Martínez-Pellitero et al. / Journal of Manufacturing Systems 37 (2015) 149–163
or packages of UML. Each view shows a particularized focus of the
model and it is composed of several diagrams.
Both the Product Model and the Resource Model represent static
knowledge and UML class diagrams are used for the formal struc-
ture definition. On the other hand, the Process Model represents
the dynamic knowledge of planning and it requires another type of
diagrams, the UML activity diagrams, although it is also necessary
a static representation of the process. In this sense, the inputs and
outputs of each activity are considered the static part of the process
and they are defined through the attributes of each activity class.
This static part of the process is directly related to the static parts
of the Product and Resource Models, as can be seen in Fig. 18.
Fig. 18 shows a glimpse of the UML models. It is not pos-
sible to include the complete models due to their extension.
However, it can be appreciated semantically the difference of
static classes in the Process Model and the Resource Model.
Whereas in the Process Model classes are related to operations
(i.e. Oper select probe configuration ei is the class related to the
operation of configuring the sensor assembly) in the Resource
Model classes are related to the elements of sensor assembly (i.e.
Ball tip stylus or is the class that contains the properties of a stylus
ended in a ball, which is a son of abstract class Straight stylus an so
on).
Some details can be observed. In the case of the classes in the
Process Model they include methods, which are associated to the
rules described trough the paper. Also, public(−)/private(+) prop-
erty has been added to every attribute in all the classes in the
model. This attribute indicates if only the owner class has access
to it or, on the contrary, all classes have access to the attribute. Car-
dinality between classes are also defined. For example, a Star stylus
will have at least one Straigth stylus (cardinality 1..*), which will
be determined through one of their son classes since Straight stylus
class is an abstract class that is not possible to instantiate.
4. Conclusions
A proposal has been carried out to automate and improve
selection of probes in the scope of dimensional and geometrical
inspection of prismatic parts, using the technology of discrete con-
tact. The proposed model of knowledge takes into account several
aspects related to precision, which operators sometimes ignore or
obviate with the purpose of making easier the inspection process.
A detailed planning of required probes has been explained to carry
out the inspection with the best precision, considering that a direct
relationship exists between orientation and configuration of the
probe assembly. Valid free-collision probe orientations have also
been determined for each inspection element. It has been consid-
ered that the part is oriented according to CMM axes, and that most
of the inspection elements in a prismatic type part can be measured
following one of the five directions of the reference system. The
proposal has been checked using a test part with a geometry that
leads to necessary probe changes. The proposed methodology is in
163
certain aspects simpler than other methodologies, especially those
in the topic of collision analysis, and it could be solved quickly by
computer applications. The knowledge model is so prepared to be
included in a knowledge based engineering system together with
the rest of inspection planning activities.
Acknowledgment
We gratefully acknowledge the financial support provided by
the Spanish Minister of Economy and Competitiveness through
project DPI2012-36642-C02-01.
References
[1] Martínez-Pellitero S, Barreiro J, Cuesta E, Álvarez BJ. A new process-based
ontology for KBE system implementation: application to inspection process
planning. Int J Adv Manuf Technol 2011;57:325–39.
[2] Martínez-Pellitero S, Cuesta E, Barreiro J, Álvarez BJ. Analysis of laser scanning
and strategies for dimensional and geometrical control. Int J Adv Manuf Technol
2010;46:612–29.
[3] Flack D. Measurement good practice guide, No. 43: CMM probing. National
Physical Laboratory (NPL); 2001. ISSN 1368-6650.
[4] Corrigal MJ, Bell R. Probe and component set-up planning for coordinate mea-
suring machines. Int J Comput Integr Manuf 1991;4(1):34–44.
[5] Moroni G, Polini W, Semeraro Q. Knowledge based method for touch probe
configuration in an automated inspection system. J Mater Process Technol
1998;76:153–60.
[6] Spyridi AJ, Requicha AAG. Accessibility analysis for the automatic inspection
of mechanical parts by coordinate measuring machines. In: IEEE Intl Conf on
Robotics & Automation, vol. 2. 1990. p. 1284–9.
[7] Álvarez BJ, Fernández P, Rico JC, Mateos S, Suárez CM. Accessibility analysis
for automatic inspection in CMMs by using bounding volume hierarchies. Int J
Prod Res 2008;46(20):5797–826.
[8] Spitz SN [Ph.D. thesis] Dimensional inspection planning for coordinate mea-
suring machines. USA: University of Southern California; 1999.
[9] Hwang C-Y, Tsai C, Chang C. Efficient inspection planning for coordinate mea-
suring machines. Int J Adv Technol 2004;23:732–42.
[10] Barreiro J, Martínez S, Labarga JE, Cuesta E. Validation of an information model
for inspection with CMM. Int J Mach Tools Manuf 2005;45(7–8):819–29.
[11] Álvarez BJ, Cuesta E, Martínez S, Barreiro J, Fernández P.36th Intl MATADOR
Conf. 2011.
[12] Beg J, Shunmugam SM. An object oriented planner for inspection of prismatic
parts – OOPIPP. Int J Manuf Technol 2002;19(12):905–16.
[13] Cho M-W, Lee H, Yoon G-S, Choi J. A feature-based inspection planning system
for coordinate measuring machines. Int J Adv Manuf Technol 2005;26:1078–87.
[14] ISO/CD 4287. Geometrical product specifications (GPS) – surface texture: pro-
file method – terms, definitions and surface texture parameters; 2014.
[15] UNE 1037.1983. Technical drawings – method of indicating surface texture on
drawings; 1983.
[16] Oberg E. Machinerys Handbook. 29th ed. Industrial Press; 2012. ISBN 978-
0831129002.
[17] Abele E, Ellermeier A, Hohenstein J, Tschannerl M. Tool length influence on
wear behavior of twisted carbide drills. Prod Eng Res Dev 2007;1(1):51–6.
[18] AB Sandvik Coromant. Modern metal cutting: a practical handbook. Sandvik
Coromant; 1996.
[19] Renishaw, http://www.renishaw.es/es/en-primera-linea-guia-para-la-
seleccion-del-palpador-de-una-mmc-10927 [accessed 04.12.14].
[20] Zeiss, https://world.probes.zeiss.com/?country=GB [accessed 04.12.14].
[21] Flack D. Measurement good practice guide, No. 41: CMM measurement strate-
gies. National Physical Laboratory (NPL); 2014. ISSN 1368-6550.
[22] Stokes M. Managing engineering knowledge: Moka methodology for knowl-
edge based engineering applications. Professional Engineering Publishing
Limited; 2001. ISBN 978-1860582950.