Optimization of geometric features of circular saw blades and

parameters of the manufacturing process aided by optiSLang

S. Weiland 1, C. Birenbaum 2*

1
Institute for machine Tools – University of Stuttgart
2
Fraunhofer Institute for Manufacturing, Engineering and Automation IPA

Abstract

This article is dedicated to the numerical optimization of circular saw tools. Because of their
geometrical shape, resembling a thin spinning disc, they are susceptible to vibrations and only
possess a low static and dynamic stiffness. The main elements to achieve high-quality
machining results with circular saw blades are their geometrical features and manufacturing
parameters, The typical behaviour of these tools can be influenced by specific variation of the
geometry of expansion slots and damping slots. On the part of the manufacturing process, a
massive influence on the static and dynamic behaviour is exercised by the roll tensioning
parameters. The target values of circular saw blades are high critical speeds and maximum
static and dynamic stiffness. Previous attempts at optimizing and designing circular saw
blades have been absolutely empiric and are not based on continuously parametrized models.
This article shows in examples, that optimization using consequently parametrized models of
circular saw blades is feasible. The calculated results are contrasted with and rated against
conclusions of the latest state of technology. The results show, that the CAD-models of
circular saw blades can be parametrized during the concept-phase using the presented
approach and that building on these models, optimized products with high stiffness and
optimum dynamic behaviour can be developed.

Keywords: Circular saw, optimization, roll tensioning, slots, natural frequencies, stiffness,
critical speed

*
contact: Dr.-Ing Christoph Birenbaum, Fraunhofer-Institut für Produktionstechnik und Automatisierung IPA,
Holzgartenstrasse 17, 70174 Stuttgart, [email protected]

11. Weimarer Optimierungs- und Stochastiktage – 06.-07. November 2014 1

1 Introduction
The initial stage of many industrial production processes is a sawing operation. Raw materials
- such as a rod-shaped mass product for instance - will have to be cut to length before further
machining in a modern production process of which many are highly or even fully automated.
The economic efficiency of automated cutting processes directly correlates with their cutting
performance. Thus, at high degrees of automation, circular saws offer a key advantage over
other saw types such as band or hack saws. The importance of circular sawing is further
exemplified by the ubiquity of small stationary circular saws or hand-held power tools used in
workshops or in the craft trade sector for a multitude of materials ranging from wood or metal
to plastics. Thus, circular sawing can be considered one of the most important preproduction
processes. In a figurative sense, the circular saw tool establishes a link between machine and
work piece. Since - just as with other production processes - the tool's mechanical properties
ultimately define machining quality, the circular sawing process requires saw blades with
optimized static and dynamic characteristics. This need is further emphasized by the blade's
round, disc-like geometry and its centre mounting hole which makes it prone to oscillations
and easily excitable by lateral force. The general tendency to smaller kerfs and therefore
thinner tools places increased weight on optimization methods with which a blade's static and
dynamic characteristics can be optimized. The spectrum of circular saw blades in the field of
production technology is shown in Figure 1.

Figure 1: Spectrum of circular saw blades. Left: Machines for high output rates in mass-
production (automation), Right: Battery driven handheld tool (Handcraft)

All these factors constitute a demand for research of optimization methods for circular saw
blades. Both static and dynamic properties of a circular saw blade depend on several factors
including the tool's rotation, its temperature profile, the slots and the stress condition of the
blade due to roll tensioning. The influence of the thermal effects is not presented here.

11. Weimarer Optimierungs- und Stochastiktage – 06.-07. November 2014 2

Figure 2: Main influences on saw blade dynamics and statics

A key part of this article focuses on the roll tensioning process which is one of the most
crucial stages in the production of circular saw blades. Two rolls apply a contact pressing
force and thereby plastically deform the blade (Figure 4). Internal residual stress, resulting
from these deformations, improves the blades static and dynamic characteristics.
This method is well accepted throughout the scientific community, even though the
underlying physics as well as its precise interrelation to the internal stress distribution are not
yet fully disclosed. The influence of residual stress is successfully simulated using linear
finite element models whereas conventional indirect models only describe the roll tensioning
outcome, while lacking any information on precise stress distribution and actual roll
tensioning process parameters. In [1] a model is presented which is able to calculate the real
stress conditions within the tool using a nonlinear finite element model.

On another subject, this article covers expansion and damping slots which saw blades are
usually equipped with in order to improve their static and dynamic behaviour. The number of
slots necessary along with their shape and location is, in current industrial practice, commonly
still being determined by empirical methods. In this work however, finite element calculations
help identify significant slot parameters and allow quantifying their effects on blade
performance. The presented algorithm can also be used to obtain an optimum slot
configuration parameter set including slot length, orientation, location and quantity given
certain base blade geometry.

Ultimately, this article provides simulation and optimization algorithms with which the
relevant parameters of circular saw blades under conditions of operation can be predicted with
a high level of accuracy. The results aim to give new impetus to the development, design and
future advances of circular saw blades.

11. Weimarer Optimierungs- und Stochastiktage – 06.-07. November 2014 3

2 Theoretical background
2.1 Natural frequencies and critical speeds and stiffness of circular
saw blades
2.1.1 Fundamentals
The global static and dynamic behaviour of circular saw blades can be approximately
described by circular plates. The dynamic behaviour of a circular plate is characterized by
nodal circles (n) und nodal diameters (k). Figure 3 gives some examples of characteristic
mode shapes of a circular plate. As a general rule, mode shapes with few nodal diameters are
most problematic for circular saws during operation. The statics and dynamics of a basic
circular plate with no slots and not roll tensioning treatment can be solved using the
Kirchhoff-Love theory [2,3], which can be considered a two dimensional bending line. Real
circular saw blades with a complex stress state and slots cannot be calculated by such basic
analytic approaches. Instead numerical solution approaches are necessary, like the finite-
elements-method used in this article.

k = 0, n = 0 k = 0, n = 1 k = 1, n = 0 k = 3, n = 0 k = 2, n = 0 k = 1, n = 1

Figure 3: Characteristic mode shapes of a circular plate

2.1.2 Campbell-diagram and stability of circular saw blades

Long since the dynamic performance and the stability of spinning structures have been
analysed. The authoritative works to this field of research were contributed by Southwell [4].
The fundamental result of his research is the fact, that natural frequencies of spinning plates
can be deduced from natural frequencies of motionless blades. The centripetal and centrifugal
forces which are affecting a spinning structure, cause stiffening of the structure and therewith
an increase in eigenfrequencies in the stationary frame of reference. If speed dependant
eigenfrequencies are plotted to a diagram, a frequency-speed diagram or Campbell diagram
will be obtained. It describes the relation between eigenfrequencies of nonrotating saw blades
in comparison to rotating ones. The so-called critical rotation speeds – critical speeds in
short - can be extracted from the frequency-speed diagram. To illustrate the following
circumstances a simple example of the frequency-speed diagram has been calculated and is
shown in Figure 4.

11. Weimarer Optimierungs- und Stochastiktage – 06.-07. November 2014 4

 Figure 4: Schematic frequency-speed diagram (Campbell diagram)

In Figure 4 the dynamic circumstances for two mode shapes with one and three nodal
diameters and no nodal circles are shown. To completely understand the frequency-speed
diagram, the choice of reference systems is crucial. There are three lines plotted for each
mode respectively. The so-called speed-dependant eigenfrequency as it would be recognized
by a moving observer situated on the disc is represented by the dashed line in the middle. The
two other lines for each mode, one of which is sloping up the other downwards, reveal
themselves to an immobile observer, watching the blade.
Mathematically they can be deduced as the sum respectively the difference of the
speed-dependant eigenfrequency and the product of the number of nodal diameters and the
rotational speed according to Equation (3) According to [4] the speed-dependant
eigenfrequencies can be approximated by the following analytical equation
f eω2  f e2  λf n2 (1)
In Equation (1)feω denotes the rotation-speed-dependant eigenfrequency of a specific mode, fe
denotes the eigenfrequencies of the concerned mode of the immobile blade and fn the
rotational speed. The factor  accounts for the mode shape and can be calculated using the
number of nodal diameters k and the poisson ratio ν as followed:
1 ν 2 3  ν
λ k  k (2)
4 4
The dimensionless poisson ratio ν is an elastic material property, which is the negative ratio of
transverse to axial strain of a body. For metallic materials the value of ν is approximately 0.3,
which is the value used for calculations in this paper.
It should be emphasized, that the approximation (2) is only valid for mode shapes with
exclusive nodal diameters and no nodal circles. Equation Fehler! Verweisquelle konnte
nicht gefunden werden. reproduces the true behaviour of unslotted circular saw blades
satisfactory. If there are slots however, the factor  has to be modified. In [1] this coefficient
is quantified for annular plates and slotted plates.
Starting with Equation (1) the other two curves can be calculated as followed:
f1,2  f eω  k  f n (3)
The curves can be imagined as a forward and a backward running wave on the blade. Both are
moving waves of the same frequency and speed, though they are propagating in different
directions. One wave moves in the direction of the rotation of the saw blade (f1) the other

11. Weimarer Optimierungs- und Stochastiktage – 06.-07. November 2014 5

opposite to it (f2). This results in the mode split shown in Figure 4. The positive term in (3)
accounts for the forward whirl, the negative one for the backwards whirl. Using the Campbell
diagram critical states can be detected.

The most important scenario resulting in instabilities of the saw blade can be read directly
from Figure 4. The abscissa of the first intersection of a backwards whirl with the axis of the
rotational speed is generally called the first critical speed ncrit,1 of the saw blade. At this
rotational speed the circular saw blade loses its stability and an observer in the stationary
frame of reference can register a standing wave with very large amplitudes. The blade seems
to freeze in its shape. In this state very small lateral forces cause large deflections. This leads
to a bad machining quality. To avoid this state of resonance in praxis, saw blade
manufacturers usually specify the maximum permitted rotational speed with 0.7 to 0.85 times
ncrit,1. This first critical speed is often caused by mode shapes without nodal circles and with
only few nodal diameters. Increasing the first critical speed ncrit,1 is a fundamental aim of the
optimization measures of circular saw blades next to the general elevation of their
eigenfrequencies, maximizing their damping properties and static stiffness.

The critical speeds ncrit,k, measured in min-1 can be deduced from Equation (3) by postulating
that the frequencies of backwards traveling waves have to be zero, i.e. f2 = 0:
60  f e,k
ncrit,k  (4)
k2  λ
Here the speed-dependant eigenfrequency of a mode with k nodal diameters is represented by
fe,k .

2.1.3 Plate buckling and stability

Influenced by internal stress thin plates can show buckling effects similar to buckling
columns [5]. If the internal stress exceeds a critical value this will lead to so-called plate
buckling. In this state the circular saw blade loses its stability and it will be deflected
perpendicularly to the plane of the plate. For circular saw blades buckling can be caused by
unfavourable thermal loads or high roll tensioning forces. Caused by thermal influences saw
blades tend to buckle in the 2/0 or the 3/0 mode shape. Excessive roll tensioning forces leads
to buckling in the 0/0 or 1/0 mode. In this paper only buckling due to roll tensioning effects is
taken into consideration. When the critical load case is reached the corresponding
eigenfrequencies take on the value of zero. Assuming a linear-elastic behaviour it is possible
to solve the eigenvalue problem, which allows to determine the critical loads and buckling
mode shapes.
(K   K SS )u  0 Det(K   K SS )  0 (5)
In these equations KSS represents the stress stiffness matrix or the geometrical stiffness
matrix, which is described in detail in [6]. The matrix K is the regular stiffness matrix and the
vector u stands for the displacements. The eigenvalue  is called load factor. The effective
load has to be multiplied by this load factor to calculate the critical loads Fw,cr and Tmax,cr.
It has to be noted, that the calculated values are purely theoretical. Effects like structural
imperfections, plasticities, contacts and geometric non-linearities are disregarded in this
calculation. To take these into consideration a non-linear calculation has to be conducted. The
critical tensioning forces Fw,cr effective in this case can be determined by successively
increasing the exerted loads using FE-analyses including the foresaid influences until the

11. Weimarer Optimierungs- und Stochastiktage – 06.-07. November 2014 6

structure reaches an unstable state. Because of imperfections, which in practice always exist,
the critical loads calculated using the linear buckling analysis are only non-conservative
values. Failure will occur sooner than calculated. This paper will only regard the linear
buckling load analysis.

2.2 Roll tensioning

The natural resonance of the tool can be de-tuned by the internal stresses induced during roll
tensioning, raising specific eigenmodes or lowering others. The critical speeds for modes with
none or one nodal diameter are reduced; those with more than one nodal diameter are raised.
The increase of the natural frequency is accompanied by an improvement of the static
behaviour of the tool. Tensioning loads which are too high lead to buckling of the structure,
similar to buckling columns. The fundamental process parameters of the roll tensioning
procedure are tensioning force, the tensioning radius and the roll crown. Within this article
only the influence of the tensioning radius is analysed. By application of the presented
procedure optimal parameter values can be determined, based on given boundary conditions.

Based on the finite element method, the optiSLang optimization corroborates the findings of
earlier works, which had proven a roll radius to blade diameter ratio of two thirds to be most
appropriate. The optimization’s objective function chosen in this work is the critical rotation
speed. The optimization tool can be used to determine optimum values for the other
parameters too.

Figure 5: Natural frequencies dependent on the amount of pretension for different vibration
modes (left), kinematics of the roll tensioning process (right)

The schematic progression of the eigenfrequencies, dependant on the tensioning force and the
kinematics of roll tensioning process are presented in Figure 5. Countless papers exist, which
describe the positive effects of roll tensioning on static and dynamic characteristics of circular
saw blades. Among others the results are documented in [7-18].

11. Weimarer Optimierungs- und Stochastiktage – 06.-07. November 2014 7

2.3 Slots and circular saw blades
Experimental and analytic examinations of circular saw blades were often part of studies
made in science and industry in further times. They often were accompanied by examinations
of the effect of slots on static, dynamic and acoustical behaviour of circular saw blades [19-
28]. Basically there are two types of slots: slots on the outer edge of the saw blade and slots in
the inner area, the integration of slots at the edge counteracts the effect of thermal expansion,
as the segments gain the possibility of tangential expansion. Two examples of the basic
configuration of damping and expansion slots are shown in Figure 6. Slots can basically be
oriented radially or tangentially on the saw blade or they can be designed in complex shapes
and arbitrary directions as shown in Figure 6. Often viscoelastic materials or solid dampers,
for example copper rivets, are integrated into the interior slots or the slot ends to increase the
systemic damping.

visco-elastic
Damping material

outer slots
inner slots Solid damper

Figure 6: Slotted wood-cutting circular saw blades and attenuation medium (sources: Robert
Bosch GmbH, AKE Knebel GmbH & Co. KG)

Even after many years of research, the modes of action aren’t accurately recorded yet. The
positive effect of slots is attributed to the interference to respectively the displacement of the
natural frequencies of the saw blade. It was speculated, that structural weakening of the saw
blade due to the slots allows for relative movement between its segments, which facilitates
energy dissipation. Other interpretations assume that the reduction of internal compression
stress at the outline of tool is the source of the positive effect of slots. Another positive impact
is that standing waves are prevented at critical speed due to asymmetric conditions on the
blade, which is established by integration of slots. Slots can reduce the oscillation amplitude
because the vibrational energy is spread more widely in the spectrum. Many literary
references point out, that because of the slot geometry eigenmodes may split into two distinct
eigenfrequencies if the slot configuration fits the mode shape.
However literature also mentions that slots have a negative effect on the static stiffness,
because they always weaken the structure. In context of optimization-problems, an optimum
always represents a compromise between stiffness and dynamic behaviour. Finally, the
analysed studies have shown that only a proper orientation of the slots leads to a raise of the
critical rotation speeds, or a small minimization of the static stiffness respectively. One
example for such a slot configuration is presented and analysed in chapter 3.1.3. Alongside

11. Weimarer Optimierungs- und Stochastiktage – 06.-07. November 2014 8

the many scientific studies in the area of slot geometries of circular saw blades, practical
examinations were conducted by the industry, which point out the advantage of slotted over
unslotted saw blades, for example [29] and [30]. There are also many independent
developments and patent applications which exist on the part of the industry. An impression
of the variations of slot geometries and positioning is given in Figure 7.

A B C D

Figure 7: Examples of slotted circular saw blades (sources: Robert Bosch GmbH (A,B), Leitz
GmbH & Co. KG (C), EDESSÖ Tools and More GmbH & Co. KG (D)

3 Optimization of circular saw blades: State of the art

3.1.1 Fundamentals
As was presented in the previous chapter, an extensive knowledge about roll tensioning and
about slots is available. The significant modes of action aren’t completely known yet, despite
all the scientific work which was done by now. For this reason roll tensioning seems to be
more like craftsmanship than a modern automated industrial production process. At almost all
saw blade manufacturers roll tensioning and the handling of corresponding machines, is the
task of long-time employees, which choose the process parameters under empirical aspects,
building on their professional experience. For an outside audience roll tensioning as well as
slot configuration seem like “alchemy”. In most cases slot configurations are tailored to
clients’ requests based on experience values. For new applications and changed base
conditions slot configuration are identified in empirical test series. The introduced results
show, that modern calculation methods, accompanying the design process are able to generate
optimized products. Using the suggested procedure, the engineer designing circular saw
blades has a tool at his disposal, to automatically identify optimized slot configurations and
process parameters for roll tensioning procedure.

3.1.2 Targets for optimization

Target values for optimized circular saw blades are manifold. In general a high static stiffness
for circular saw blades is to be pursued. Especially in metal working, tools are exposed to
great loads because of high machining forces. Therefore an increased stiffness at the tools
outer rim is needed. Founded in the much lower rotation speeds in comparison to wood-
working, a high critical rotation speed is not mandatory for metal working tools. To be able to
tolerate the prevailing high temperature gradients the endurable buckling loads have to be as
high as possible. This is true for both metal machining as well as wood-working, as high
temperatures can occur at the teeth. Usually circular saw blades are designed to be fatigue-
endurable. Especially in the slotted areas the tension has to be minimized, by choosing

11. Weimarer Optimierungs- und Stochastiktage – 06.-07. November 2014 9

appropriate slot geometries. Ultimately it is obvious, that circular saw blades are exposed to
complex stresses and strains which necessitate a multi-objective optimization.
In the following part exemplary calculations are made for the optimization of circular saw
blades. A sensitivity analysis and basic problems relating to the roll tensioning process are
defined and analysed. The chosen procedure is presented and the results are compared with
actual conclusions and measurements.

3.1.3 Example 1: Sensitivity analysis of a slot configuration

The first example will show how a sensitivity analysis can be applied to identify the relevant
geometric parameters of an exemplary slot configuration on the most important technical
properties of the saw blade base body. The slot configuration used for these calculations is
based on the works of [31]. The geometry of the blade and its slots is depicted in Figure 1.

Figure 8: Slot configuration in compliance with [31]

The blade base and the slot configuration are defined by the tooth base radius RG, the mean
slot diameter RM, the length of the circular arc LT of a single slot without the overlap φo, the
radial slot length LR as well as the slot width s and the blades thickness t. In compliance with
the cited patent the lengths respectively radii are combined into the dimensionless aspect ratio
QL (11) and the radius or diameter ratio QD (12).
L
QL  T (6)
LR

11. Weimarer Optimierungs- und Stochastiktage – 06.-07. November 2014 10

 RM DM
QD   (7)
RG DG
For the sensitivity analysis the aspect ratio QL and diameter ratio QD as well as the slot width
s and the overlapping angle φo are defined as model parameters. The ranges of the parameter
values are set to allow only feasible configurations, to prevent merging of slots for example.
The blade diameter is constant at 300 mm as are the blade thickness (2.2 mm) and the number
of slots (14).
As explained earlier there are several objectives that are relevant to optimizing saw blades:
 The static stiffness cst. The ratio of a lateral force F, applied to the outer edge of the blade,
to the corresponding displacement w. It represents the blade’s resistance to transverse
disturbances (8).
 The maximum equivalent stress σv at the ends of the slots.
 The margin between the rotational frequency of the blade and its closest
eigenfrequency Δfmin. This parameter substitutes for the critical speed, as the saw blade in
this scenario is intended for a specific rotational speed (n 0 = 3000min-1) (9).
 And the sum of the first 10 eigenfrequencies at operational speed Δfsum, as a characteristic
for the blades overall dynamic behaviour (10).
F
cst  (8)
w
f min  min f 0  f n,i (n0 ) (9)
10
f sum  min  f n,i (n0 ) (10)
i 1

Figure 9: Workflow in ANSYS Workbench

The FE-simulation used to calculate the objective function follows the workflow shown in
Figure 9. APDL scripts in each block distil the objective functions out of the separate
analyses.

11. Weimarer Optimierungs- und Stochastiktage – 06.-07. November 2014 11

Figure 10: Extended correlation matrix

In Figure 10 row-wise opposite algebraic signs of the correlation coefficients show that the
operation specific objective Δfmin and the more general objectives cst and Δfsum are conflicting
with each other. The contrary trends of static stiffness and frequency margin of the blade
won’t allow finding an optimum slot configuration for every application. Depending on the
application of the saw blade (wood-working or metal machining) the objectives have to be
weighted differently in future optimizations.

11. Weimarer Optimierungs- und Stochastiktage – 06.-07. November 2014 12

Figure 11: Coefficients of importance

Representations of the coefficients of importance for each objective reveal at first glance that
the slot width and the aspect ratio seem to have little impact on either of the objectives
(Figure 11). This will allow future optimizations to omit these parameters in order to save
computing time.

3.1.4 Example 2: Optimization of roll tensioning radius

The first example didn’t consider the effects based on roll tensioning. This example will show
that it is possible to identify ideal tensioning parameters by consequently parameterizing the
model. The roll tensioning process is incorporated in ANSYS using a linear prestress analysis.
Within the Workbench environment this is realised by an APDL-Script. The residual stress
effects are modelled via different thermal expansion coefficients in the area of the roll
tensioning path and the rest of the blade. For these procedures the optimization parameters is
the roll tensioning temperature T. This is a non-mechanical substitute for the physical
tensioning force F. The tensioning temperature can be converted to the physical domain
“force” by aid of corresponding characteristic curves. The chosen approach delivers
qualitative information concerning the magnitude of the pretension.

11. Weimarer Optimierungs- und Stochastiktage – 06.-07. November 2014 13

This example will determine the radius rt which offers a maximum the first critical speed.
This optimization problem is typical for wood-working applications. When calculating
optimal tensioning radii the effect described in chapter 3.1.3 has to be considered. Tensioning
forces or temperatures Tt,crit which are too high lead to buckling of the saw blade, so
tensioning temperatures higher than Tt,crit should be eliminated in advance. For this
examination a procedure was chosen, which identifies the maximum permissible tensioning
temperature corresponding to a particular radius in an initial calculation. These critical
temperatures Tt,crit are transmitted to the optimizer for every design m via the parameter
manager of ANSYS. The typical approach is illustrated in Figure 12. This procedure rules
out, that invalid, physically unfeasible values are given to the optimizer. In this scenario the
parameter is no independent parameter, but rather a range for variations between its lower
bound and the maximum temperature for each respective design m. For the implementation of
this approach, an auxiliary variable WBhelp has been established, for which a range of values
between 0 and 1 was defined. This value is committed to Excel, where the temperature value
Tmax,crit,m for the current radius is interpolated.

Figure 12: Basic procedure of the initialization step for the roll tensioning optimization

This yields the following procedure:

1. Variation of the parameters WBhelp und rt,m for the current design m by optiSLang
within the global range and transfer to Excel
2. Computation of the valid maximum value Tmax,crit,m on basis of rt,m.
3. Calculation of the temperature for the design m and transfer of the value to the
parameter manager for further use in ANSYS:
Tt=Tmin+WBhelp∙(Tmax,crit,m-Tmin) (11)

The minimal temperature was set to a constant value. Roll tensioning can lead to transposition
of modes. For this reason the eigenmodes have to be tracked using an automated process. As
shown in Equation (11) the knowledge of nodal diameters is necessary for the identification
of mode shapes and the establishment of the Campbell diagram. Mode tracking and the
extraction of nodal diameters and circles is performed by an APDL-script. The basic
procedure of building a Campbell diagram is shown in Figure 13:

11. Weimarer Optimierungs- und Stochastiktage – 06.-07. November 2014 14

Figure 13: Basic procedure of the implemented mode tracking algorithm and example based
on the 3/0-modeshape

The knowledge, that mode shapes of circular saw blades are made up of nodal circles and
nodal diameters, which can be represented as a combination of analytic sine waves, is used
for mode tracking. An analytic test model is generated, within which nodal diameters and
circles are known. This model is compared to the FE-model of the circular saw blade.
At the outer edge of the circular saw blade the nodal solutions of the FE-model are selected
for a specific eigenvector i. Then they are compared to the analytical signal.

1 360
r    sin    k  sin      k
 
N  0 reference   (12)
(FEM) analytical test signal
mode shape i

By shifting the test signal over the phase angle for all possible combinations of nodal
diameters and circles correlation values for the current configuration are generated, which
measure the conformity of the modes. By iteratively comparing the analytical test signal with
each calculated mode shape of the FE-model for every nodal diameter and circle the identity
of the eigenmode and its shape can be determined. Correlation values near to 1 indicate a
match, values of 0 a mismatch, as shown in Figure 13.

The results of the optimization run are contrasted to experimental data and experience values.
In practice circular saw blades are roll tensioned at a radius rt at 2/3 of the outer diameter.
This ratio is independent of the tool geometry, but varies depending on the flange, which

11. Weimarer Optimierungs- und Stochastiktage – 06.-07. November 2014 15

clamps the blade in its center. To identify the eigenmodes an experimental modal analysis
was conducted. The tools were analysed under free-free boundary conditions. Annular plates
were used, which were roll tensioned at varying diameters. The experimental setup and the
results are depicted in the following figure. The outer diameter of the plates is 175 mm, the
central bore 30 mm and the plate’s thickness 2.4 mm

Figure 14: Measurement setup and frequency response functions of circular plates with
different roll tensioning radii

Figure 14 shows clearly, that the highest eigenfrequencies of the relevant eigenmodes are
attained for a ratio of tensioning radius to outer radius of 0.62. Extrapolating the results by use
of Equation (3) to the domain of the rotational speed, the highest critical speeds are attained at
this ratio as well. As a substitute for a fully detailed circular saw blade the optimum
tensioning radius of an annular plate is calculated with optiSLang using the previously
described calculation procedure. The results are depicted in Figure 15. The geometry of the
plate is in accordance with the experimental setup. Up to a diameter of 110 mm the inner area
of the plate is clamped, to account for the flange. The degrees of freedom of the
corresponding nodes are locked.

Figure 15: Calculation results (Optimization of the roll tensioning radius)

The left-hand representation in Figure 15 shows the parameter and its bounds. An optimum
tensioning radius of 109 mm was calculated. Divided by the outer diameter of 175 mm this
results in a ratio of 0.62, which conforms to the experimental data shown in Figure 15 and the
educated guess use in practice. The middle representation depicts the objective function

11. Weimarer Optimierungs- und Stochastiktage – 06.-07. November 2014 16

which is converging quickly and clearly after only a few designs. The characteristic quadratic
curve of the right-hand graph in Figure 15, which shows the critical speed as a function of the
tensioning radius, compares well to literary sources, i.e. [9,10].

4 Conclusions
By means of concrete practical application it was proven that circular saw blades can be
improved using parametric optimization techniques. The results concerning tensioning
parameters and slot configurations show a good match with conducted experiments, empirical
values and the state of technology. In future the analyses will be expanded to further
applications. Especially the examination of rotating tolls during a machining process. To add
this feature and more to the procedural method will be supplemented by the involved research
institutions. With the sketched method the design engineer eventually will have a simple tool
to analyse and refine circular saw blades during the design process.

11. Weimarer Optimierungs- und Stochastiktage – 06.-07. November 2014 17

5 References
BIRENBAUM, C.: Beitrag zur rechnerischen und experimentellen Auslegung von Kreissäge-
Stammblättern für die Holzbearbeitung. Universität Stuttgart, Dissertation, 2013

KIRCHHOFF, G.R.: Gesammelte Abhandlungen. Saarbrücken: VDM Müller, 1881, Reprint

2006.

GROSS, D.; HAUGER, W.; SCHNELL, W.; WRIGGERS, P.: Technische Mechanik 4. Berlin:,
Springer-Verlag, 2002.

SOUTHWELL, R.V.: On the free transverse vibrations of a uniform circular disc clamped at its
centre, and on the effects of rotation. In: Proceedings of the royal society of London, series
A 101, 1922.

GOGU, G.: Die Verwölbung von Kreissägeblättern unter dem Einfluß von Mem-
branspannungen. In: Holz als Roh- und Werkstoff 46 (1988) 4, S. 135-140

RUST, W.: Nichtlineare Finite-Elemente-Berechnungen. Wiesbaden: Vieweg+Teubner

Verlag, 2009

EKEVAD, M.; CRISTOVAO, L.; GRÖNLUND, A.: Different methods for monitoring flatness and
tensioning in circular-saw blades. In: Proceedings of the 19th IWMS, Nanjing, China, 2009

FENDELEUR, D.; AUBRY, E.; KAISER, E.; RENNER, M.: Dynamical measurements and
evaluation of the tension of circular saw blades. In: Proceedings of the 14th IWMS, Paris,
Epinal, Cluny, Frankreich, 1999

FRIEBE, E.V.: Steifheit und Schwingungsverhalten von Kreissägeblättern für die

Holzbearbeitung. Universität Braunschweig, Dissertation, 1973

HACKENBERG, P.: Spannungen in mechanisch und thermisch vorgespannten

Kreissägeblättern. Universität Aachen, Dissertation, 1974

MÜNZ, U.V.: Mechanisch und thermisch erzeugte Spannungen in Kreissägeblättern. In: Holz
als Roh- und Werkstoff 38 (1980), S. 201-208

MÜNZ, U.V.: Untersuchung des Einflusses unterschiedlicher Spannungsverteilungen im

Bereich der Walzspur auf den gesamten Vorspannungszustand und das Arbeitsverhalten
von Kreissägeblättern. Forschungsberichte des Landes Nordrhein-Westfalen. Opladen:
Westdeutscher Verlag, 1981

NICOLETTI, N.; FENDELEUR, D.; NILLY, L.; RENNER, M.: Using finite elements to model
circular saw roll tensioning. In: Holz als Roh- und Werkstoff 54 (1996), S. 99-104

11. Weimarer Optimierungs- und Stochastiktage – 06.-07. November 2014 18

NICOLETTI, N.; AUBRY, E.; FENDELEUR, D.; RENNER, M.: Finite element analysis for
burnishing residual strains in an annular plate. In: Holz als Roh- und Werkstoff 54 (1996),
S. 273-277

NICOLETTI, N.; AUBRY, E.; FENDELEUR, D.; RENNER, M.: A finite element model for the
analysis of roll burnishing. In: Holz als Roh- und Werkstoff 55 (1997), S. 183-187

SCHAJER, G.S.; MOTE, C.D. JR.: Analysis of Roll Tensioning and its Influence on Circular
Saw Stability. In: Wood Science and Technology 17 (1983), S. 287-302

SCHAJER, G.S.; MOTE, C.D. JR.: Analysis of Optimal Roll Tensioning For Circular Saw
Stability. In: Wood and Fibre, 16(3), 1984, S. 323-338

SZYMANI, R.; MOTE, J.D.: Theoretical and experimental analysis of circular saw tensioning.
In: Wood Science and Technology 13 (1979), S. 211-237

BELJO-LUČIĆ, R.; GOGLIA, V.: Some possibilities for reducing circular saw idling noise. In:
Journal of Wood Science 47 (2001), S. 389-393

HOLØYEN, S.: Vibrations and natural frequencies of angular slot circular saws. In: Holz als
Roh- und Werkstoff 45 (1987), S. 101-104

HU, Y.N.; WANG, C.Y.; DING, H.N.: The Mechanical Performance of Diamond Saw Blades
with Special Structure. In: Key Engineering Materials Vols. 259-260 (2004), S. 141-145

NISHIO, S.; MARUI, E.: Effects of slots on the lateral vibration of a circular saw blade. In:
International Journal Machining Tools Manufacturing Vol 36, No. 7, 1996

NISHIO, S.: Hyper Critical Sawblade. In: Proceedings of the 16th IWMS, Matsue, Japan, 2003

PLESTER, J.: Geräuschentstehung und Maßnahmen zur Geräuschminderung beim Sägen mit
Kreissägeblättern. Universität Braunschweig, Dissertation, 1986

RAY-CHUNG, Y.; MOTE, C.D.: Vibration of circular saws containing slots. In: Holz als Roh-
und Werkstoff (1987) 45, S. 155-160

SINGH, R.: Case History: The Effect of Radial Slots on the Noise of Idling Circular Saws. In:
Noise Control Engineering Journal Volume 31, No. 3, 1988, S. 167 - 172

SVOREŇ, J.; MURIN, L.: The effect of the shape of the compensating slots in the body of a
circular saw blade on noise level of a circular saw in the cutting process. In: Proceedings
of the ACOUSTIC High Tatras 34th Int. Acoustical Conference-EAA Symposium, 2009

TOBIAS, S.A.; ARNOLD, D.: The influence of dynamical imperfections on the vibration of
rotating discs. In: Proceedings of the Institution of Mechanical Engineers Volume 171
(1957), S. 669-690

HAFNER, T.: Maßnahmen zur Schallreduktion von Hartmetallkreissägen bei der Zerspanung
von Stahl. Schriftenreihe der Bundesanstalt für Arbeitsschutz, 1990

11. Weimarer Optimierungs- und Stochastiktage – 06.-07. November 2014 19

N.N.: Geräuschgeminderte Sägeblätter für Holz, Kunststoff und Aluminium-Marktübersicht,
Schalldruckpegel in Labor und Praxis. Hauptverband der gewerblichen
Berufsgenossenschaften, Lärmschutzarbeitsblatt Ausgabe April 2006.

NISHIO, S.; ILNUMA, T.; NAKAJIMA, Y.: Disk-like tool. Patent specification, European
Patent Office, EP 1 679 165 A1, 2004.

[1] BIRENBAUM, C.: Beitrag zur rechnerischen und experimentellen Auslegung von
Kreissäge-Stammblättern für die Holzbearbeitung. Universität Stuttgart,
Dissertation, 2013
[2] Kirchhoff, G.R.: Gesammelte Abhandlungen. VDM Müller, Saarbrücken, 1881,
Reprint 2006
[3] Gross, D.; Hauger, W.; Schnell, W.; Wriggers, P.: Technische Mechanik 4. .
Springer-Verlag, Berlin, 2002
[4] Southwell, R.V.: On the free transverse vibrations of a uniform circular disc
clamped at its centre, and on the effects of rotation. Proceedings of the royal society
of London, series A 101, 1922
[5] Gogu, G.: Die Verwölbung von Kreissägeblättern unter dem Einfluß von Mem-
branspannungen. In: Holz als Roh- und Werkstoff 46 (1988) 4, S. 135-140.
[6] Rust, W.: Nichtlineare Finite-Elemente-Berechnungen. Vieweg+Teubner Verlag,
Wiesbaden, 2009
[7] Ekevad, M.; Cristovao, L.; Grönlund, A.: Different methods for monitoring flatness
and tensioning in circular-saw blades. 19th IWMS, Nanjing, China, 2009.
[8] Fendeleur, D.; Aubry, E.; Kaiser, E.; Renner, M.: Dynamical measurements and
evaluation of the tension of circular saw blades. 14th IWMS, Paris, Epinal, Cluny –
Frankreich, 1999.
[9] Friebe, E.V.: Steifheit und Schwingungsverhalten von Kreissägeblättern für die
Holzbearbeitung. Dissertation Universität Braunschweig, 1973.
[10] Hackenberg, P.: Spannungen in mechanisch und thermisch vorgespannten
Kreissägeblättern. Dissertation Universität Aachen, 1974.
[11] Münz, U.V.: Mechanisch und thermisch erzeugte Spannungen in Kreissägeblättern.
In: Holz als Roh- und Werkstoff 38 (1980), S. 201-208
[12] Münz, U.V.: Untersuchung des Einflusses unterschiedlicher Spannungsverteilungen
im Bereich der Walzspur auf den gesamten Vorspannungszustand und das
Arbeitsverhalten von Kreissägeblättern. Forschungsberichte des Landes Nordrhein-
Westfalen. Opladen: Westdeutscher Verlag, 1981
[13] Nicoletti, N.; Fendeleur, D.; Nilly, L.; Renner, M.: Using finite elements to model
circular saw roll tensioning. In: Holz als Roh- und Werkstoff 54 (1996), S. 99-104
[14] Nicoletti, N.; Aubry, E.; Fendeleur, D.; Renner, M.: Finite element analysis for
burnishing residual strains in an annular plate. In: Holz als Roh- und Werkstoff 54
(1996), S. 273-277
[15] Nicoletti, N.; Aubry, E.; Fendeleur, D.; Renner, M.: A finite element model for the
analysis of roll burnishing. In: Holz als Roh- und Werkstoff 55 (1997), S. 183-187

11. Weimarer Optimierungs- und Stochastiktage – 06.-07. November 2014 20

[16] Schajer, G.S.; Mote, C.D. Jr.: Analysis of Roll Tensioning and its Influence on
Circular Saw Stability. In: Wood Science and Technology 17 (1983), S. 287-302.
[17] Schajer, G.S.; Mote, C.D. Jr.: Analysis of Optimal Roll Tensioning For Circular
Saw Stability. In: Wood and Fibre, 16(3), 1984, S. 323-338.
[18] Szymani, R.; Mote, J.D.: Theoretical and experimental analysis of circular saw
tensioning. In: Wood Science and Technology 13 (1979), S. 211-237.
[19] Beljo-Lučić, R.; Goglia, V.: Some possibilities for reducing circular saw idling
noise. In: Journal of Wood Science 47 (2001), S. 389-393.
[20] Holøyen, S.: Vibrations and natural frequencies of angular slot circular saws. In:
Holz als Roh- und Werkstoff 45 (1987), S. 101-104.
[21] Hu, Y.N.; Wang, C.Y.; Ding, H.N.: The Mechanical Performance of Diamond Saw
Blades with Special Structure. In: Key Engineering Materials Vols. 259-260 (2004),
S. 141-145.
[22] Nishio, S.; Marui, E.: Effects of slots on the lateral vibration of a circular saw
blade. In: Int. Journal Machining Tools Manufacturing Vol 36, No. 7, 1996
[23] Nishio, S.: Hyper Critical Sawblade. 16th IWMS, Matsue, Japan, 2003
[24] Plester, J.: Geräuschentstehung und Maßnahmen zur Geräuschminderung beim
Sägen mit Kreissägeblättern. Universität Braunschweig, Dissertation, 1986
[25] Ray-Chung, Y.; Mote, C.D.: Vibration of circular saws containing slots. In: Holz
als Roh- und Werkstoff (1987) 45, S. 155-160
[26] Singh, R.: Case History: The Effect of Radial Slots on the Noise of Idling Circular
Saws. In: Noise Control Engineering Journal. Volume 31, No. 3, 1988, S. 167 - 172
[27] Svoreň, J.; Murin, L.: The effect of the shape of the compensating slots in the body
of a circular saw blade on noise level of a circular saw in the cutting process. In:
Proceedings of the ACOUSTIC High Tatras, 34th Int. Acoustical Conference-EAA
Symposium, 2009
[28] Tobias, S.A.; Arnold, D.: The influence of dynamical imperfections on the
vibration of rotating discs. In: Proc. Inst. Mech. Eng. Volume 171 (1957), S. 669-
690
[29] Hafner, T.: Maßnahmen zur Schallreduktion von Hartmetallkreissägen bei der
Zerspanung von Stahl. Schriftenreihe der Bundesanstalt für Arbeitschutz, 1990
[30] N.N.: Geräuschgeminderte Sägeblätter für Holz, Kunststoff und Aluminium-
Marktübersicht, Schalldruckpegel in Labor und Praxis. Hauptverband der
gewerblichen Berufsgenossenschaften, Lärmschutzarbeitsblatt Ausgabe April 2006.
[31] NISHIO, S.; ILNUMA, T.; NAKAJIMA, Y.: Disk-like tool. Patent specification,
European Patent Office, EP 1 679 165 A1, 2004.

11. Weimarer Optimierungs- und Stochastiktage – 06.-07. November 2014 21