Available

Available online
online at
at www.sciencedirect.com
www.sciencedirect.com

Available
ScienceDirect
ScienceDirect
online atonline
www.sciencedirect.com
Available at www.sciencedirect.com
Procedia
Procedia CIRP
CIRP 00
00 (2019)
(2019) 000–000
ScienceDirect
ScienceDirect
000–000
www.elsevier.com/locate/procedia
www.elsevier.com/locate/procedia

Procedia Procedia
CIRP 00 CIRP
(2017)102
000–000
(2021) 156–161
www.elsevier.com/locate/procedia

18th
18th CIRP
CIRP Conference
Conference on
on Modeling
Modeling of
of Machining
Machining Operations
Operations

Advances
Advances in in Modeling
28th CIRPof
Modeling the
the Kerf
ofDesign Kerf Formation
May 2018,considering
Formation
Conference, Nantes, France the
considering the Primary
Primary andand
Deflection
Deflection JetsJets for
for the
the Abrasive
Abrasive Water
Water JetJet Technology
Technology
A new methodology to analyze the functional and physical architecture of
Eckart
existing
Eckart Uhlmanna,b
products
Uhlmann a,b, Harald Kruggel-Emdencc, Constantin Männela,
for an assembly
, Harald oriented
Kruggel-Emden product
, Constantin a,*, Enrico Barthaa,
family
Männel identification
*, Enrico Barth ,
cc
Darius Markauskas
Darius Markauskas
a
aTechnische PaulBerlin,
Technische Universität
Universität Stief
Berlin, *, Jean-Yves
Institute
Institute for Machine
for Dantan,
Machine Tools
Tools and FactoryAlain
and Factory Etienne,
Management
Management Ali Siadat
(IWF), Pascalstr.
(IWF), Pascalstr. 8-9, 10587
8-9, 10587 Berlin,
Berlin, Germany
Germany
bFraunhofer Institute for Production Systems and Design Technology (IPK) , Pascalstr. 8-9, 10587 Berlin, Germany
b
Fraunhofer Institute for Production Systems and Design Technology (IPK) , Pascalstr. 8-9, 10587 Berlin, Germany
École Nationale
TechnischeSupérieure d’Arts et Métiers, ArtsProcess
et Métiers ParisTech,and
LCFC EAProcessing,
4495, 4 Rue Augustin Fresnel, 1,
Metz 57078, France
Technische Universität
Universität Berlin,
Berlin, Mechanical
Mechanical Process Engineering
Engineering and Solids
Solids Processing, Ernst-Reuter-Platz
Ernst-Reuter-Platz 1, 10587
10587 Berlin
c
c
Berlin
** Corresponding
Corresponding author.
author. Tel.:
Tel.: +49-30-314-21791;
+49-30-314-21791; fax:
fax: +49-30-314-25895.
+49-30-314-25895. E-mail [email protected]
address: [email protected]
E-mail address:
* Corresponding author. Tel.: +33 3 87 37 54 30; E-mail address: [email protected]

Abstract
Abstract
Abstract
Processing
Processing of of difficult
difficult to
to machine
machine materials
materials is is aa promising
promising application
application for
for abrasive
abrasive water
water jet
jet kerf
kerf cutting
cutting and
and milling.
milling. However,
However, due due to
to the
the large
large
Innumber
today’sofbusiness environment, the trend towards more product variety and customization is removal
unbroken. Due to for
thisitsdevelopment, the need of
number of interactions of an energy-bound cutting process, a precise prediction of the material removal is crucial for its application. Based on
interactions of an energy-bound cutting process, a precise prediction of the material is crucial application. Based on an
an
agile and reconfigurable
analytical approach, a production
material systems
removal emergedmodel
simulation to copeis with various
introduced products and
considering the product
primary families.
and To design
deflecting jet and optimize
impacts to production
rapidly predict
analytical approach, a material removal simulation model is introduced considering the primary and deflecting jet impacts to rapidly predict
systems
various as
various well situations.
cutting
cutting as to choose
situations. Thethe
The optimal
model
model product
describes
describes matches, cutting
fundamental
fundamental productmechanisms
cutting analysis methods
mechanisms are needed.
considering
considering the
the waterIndeed,
water jet’s most ofremoval
jet’s material
material the known
removal rate methods
rate and
and is aim to
is calibrated
calibrated
analyze
for titanium aluminides. The model allows for a comprehensive kerf prediction and thus potentially accelerates the process design improvingand
for a product
titanium or one
aluminides. product
The family
model on
allows the
for physical
a level.
comprehensive Different
kerf product
prediction families,
and thus however,
potentially may differ
accelerates largely
the in terms
process of
design the number
improving the
the
nature of components.
productivity and quality This
for fact impedes
abrasive water an efficient
jet kerf comparison
cutting
productivity and quality for abrasive water jet kerf cutting and milling. and and
milling. choice of appropriate product family combinations for the production
system. A new methodology is proposed to analyze existing products in view of their functional and physical architecture. The aim is to cluster
© 2021
these
© 2021 The Authors.
products
The Published
in new assembly
Authors. Published by Elsevier
oriented
by B.V.B.V.
product families for the optimization of existing assembly lines and the creation of future reconfigurable
© 2021
This The
is an
an openAuthors.
access Published
article by ELSEVIER
under ELSEVIER
the CC
CC B.V.
BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0)
assembly
This
This is systems.
open
is an open Based
access
access on
articleDatum
under
article under Flow
the Chain, the physical
BY-NC-ND
thescientific
CC BY-NC-ND license
licensestructure of the products is analyzed. Functional subassemblies are identified, and
(https://creativecommons.org/licenses/by-nc-nd/4.0)
(https://creativecommons.org/licenses/by-nc-nd/4.0)
Peer-review under responsibility of the committee of the 18th CIRP Conference on Modeling of Machining Operation.
aPeer-review under responsibility of the scientific committee of
Peer-review under responsibility of the scientific committee of the 18th physical
functional analysis is performed. Moreover, a hybrid functionalthe 18th
and CIRP
CIRP Conference
architecture
Conference on
on Modeling of
of Machining
graph (HyFPAG)
Modeling is the Operation
Machining output which depicts the
Operation
Keywords:
similarity Waterjet
Keywords:between machining;
Waterjet product
machining; Titanium
families aluminides;
Titaniumbyaluminides;
providingAWJ AWJ milling;
design Modeling;
support
milling; AWJ
to both,
Modeling; deflection
AWJproduction
deflection system planners and product designers. An illustrative
example of a nail-clipper is used to explain the proposed methodology. An industrial case study on two product families of steering columns of
thyssenkrupp Presta France is then carried out to give a first industrial evaluation of the proposed approach.
© 2017 The Authors. Published by Elsevier B.V.
1.
1. Introduction
Peer-review under responsibility of the scientific committee of the 28th CIRP
Introduction have aa strong
Design
have effect
effect on
Conference
strong the
the cutting
2018.
on cutting tool
tool causing
causing highhigh wear
wear and
and
aa short
short life
life time
time of
of the
the insert.
insert.
1.1. Motivation
Motivation
Keywords:
1.1. Assembly; Design method; Family identification Abrasive
Abrasive water jet
water jet (AWJ)
(AWJ) machining
machining utilizesutilizes highly
highly
accelerated
accelerated abrasives to cut almost all kinds of materials.
abrasives to cut almost all kinds of materials. DueDue
Economic considerations,
Economic considerations, environmental
environmental awareness awareness and and to
to the
the absence
absence of of aa workpiece-tool
workpiece-tool contact,
contact, thethe AWJ
AWJ process
process
increasingly political frameworks are just some of the reasons offers an
an ever-sharp tool, low
low andand material independent tool
1.increasingly
whyIntroduction political frameworks are just some of the reasons
efficiency improvements
improvements for for all
all kinds
kinds ofof systems
systems and and ofoffers
wear and
ever-sharp
the product little
tool,
range effects
and characteristics
on
material
the
independent
manufactured
workpiece
tool
and/or
surface
why efficiency wear
assembled and little
in this effects
system. on the
In thismachining workpiece
context, theprocesses
main challenge surface in
processes
processes are desirable. An approach to increase the efficiency characteristics [1]. Thus,
Thus,isAWJ offers an
ofDue toarethe
many technical
desirable.
technical fast An
systems
approach to increase
development
is to
to apply thein
apply the mostthe
the
suitable
efficiency
domainmaterialof characteristics
modelling
alternative and
for
[1].
analysis AWJ
now machining
not only toprocesses
cope offers
with an
single
of many
communication andsystems
an is
ongoing trend most
of suitable
digitization material
and products, for the
the manufacturing
alternativea limited manufacturing
product range
of difficult
of existing
or
to
to cut
cut materials,
difficult product materials,
families,
for aa task.
for task. Consequently, more
more resilient,
Consequently, enterprises resilient, durable
durable and
and Fig.
Fig. 1.
1. However,
However, due
due to
to the
the energy-bound
energy-bound properties
properties of
of the
the jet,
jet,
digitalization,
light-weight manufacturing
materials such as are facing
metal important
matrix but
the also
AWJ to be able
process to
doesanalyze
not and
produceto compare
clearly products
predictable to define
shapes
light-weight
challenges in materials
today’sfiber such
market environments: as metal matrix
a continuing the
new AWJ process
product families. does not produce
It can tobee.g. clearly
observed predictable
thatHence, shapes
classical existing
composites
composites (MMC),
(MMC), fiber ofreinforced
reinforced polymers
polymers (FRP) (FRP) or
or of
of the
the workpiece
workpiece compared
compared to in
e.g. an
an insert.
insert. Hence,orespecially
especially
tendency
titanium towards
aluminides reduction
(TiAl) areproduct
developed development
and used. times
However, and product
AWJ families
milling are regrouped
operations, Fig. 1b, function
and more of clients
efficient features.
AWJ kerf
titanium
shortened aluminides
product (TiAl) are developed
lifecycles. In addition, and
there used.
isbe However,
anlimited
increasing AWJ
However, milling operations,
assembly oriented Fig. 1b,
productand more
families efficient
are AWJ
hardly kerf
to find.
the
the use
use of
of these
these high-performance
high-performance materials
materials can
can be inlimited due
due milling
milling with
with additional
additional material
material removal,
removal, Fig.
Fig. 1c,
1c, are
are difficult
difficult
demand
to of customization,
difficulties occurring being
in at the
manufacturingsame time
processes. a global
The to On the product family level, productssimulative
differ mainly in two
to difficulties
competition with occurring
competitors in manufacturing
all over g. processes.
themilling,
world. can This trend,The to predict.
main predict. For
For this
this reason,
characteristics: reason,
(i) the
aa number
numberofof
number simulative approaches
ofcomponents approaches
and (ii) the
conventional
conventional production
production processes,
processes, e. e. g. milling, cantodamage
damage have
have been
been pursued
pursued to
to understand
understand and
and predict
predict the
the kerf
kerf formation
formation
which
the FRP is inducing
material, on the
the development
other side MMC from macro
and TiAl micro
materials type
and of components
thus (e.g. mechanical, electrical, electronical).
the FRP material, on the other side MMC
markets, results in diminished lot sizes due to augmenting and TiAl materials thus improve
andClassical
improve the
the capability
methodologiescapability of
of AWJ machining.
AWJmainly
considering machining.single products
product varieties (high-volume to low-volume production) [1]. or solitary, already existing product families analyze the
To cope with this augmenting variety as well as to be able to product structure on a physical level (components level) which
2212-8271 ©
© 2021
2021 The Authors.
Authors. Published by
by ELSEVIER
identify
2212-8271 possible The optimizationPublishedpotentials
ELSEVIER in B.V.the existing
B.V. causes difficulties regarding an efficient definition and
This
This is
is an
an open
open access
access article
article under
under the
the CC
CC BY-NC-ND
BY-NC-ND license
license (https://creativecommons.org/licenses/by-nc-nd/4.0)
(https://creativecommons.org/licenses/by-nc-nd/4.0)
production system, it is important to have a precise knowledge
Peer-review under responsibility of the scientific committee of the 18th CIRP Conference comparison of of
on Modeling different
Machiningproduct
Operation families. Addressing this
Peer-review under responsibility of the scientific committee of the 18th CIRP Conference on Modeling of Machining Operation
2212-8271 © 2021 The Authors. Published by Elsevier B.V.
This is an©open
2212-8271 2017access article Published
The Authors. under theby CC BY-NC-ND
Elsevier B.V. license (https://creativecommons.org/licenses/by-nc-nd/4.0)
Peer-reviewunder
Peer-review under responsibility
responsibility of the
of the scientific
scientific committee
committee of theof28th
the CIRP
18th CIRP
DesignConference
Conferenceon Modeling of Machining Operation.
2018.
10.1016/j.procir.2021.09.027
 Eckart Uhlmann et al. / Procedia CIRP 102 (2021) 156–161 157
2 Author name / Procedia CIRP 00 (2019) 000–000

a AWJ kerf cutting b AWJ milling predictions for one pass with the AWJ based on this approach
vf have been introduced [6, 7]. Furthermore, empirical models [8]
vf, nP and approaches regarding the energy conservation have been
employed [9, 10]. The models described above are designed for
kerf cutting operations, Fig 1a, with one pass of the AWJ.
C
W + WD for    0 (1)
c AWJ kerf milling d Additional WB =  C1
material removal WC 2 + WD for    0
Milling 90° Milling 45°
vf, nP vf, nP MC (v p sin( ) − K ) 2  MC (v p sin( ) − K ) 2  (2)
=WC1  v p cos( ) − 
B α 0.5 v p sin( )  0.5 v p sin( ) 

A M (v 2p cos 2 ( ) − K1 (v p sin( ) − K )3/2
WC 2 = (3)
2
Fig. 1. AWJ machining processes: (a) AWJ kerf cutting; (b) AWJ milling; M (v p sin( ) − K ) 2 (4)
WD =
(c) AWJ kerf milling; (d) Additional material removal 2
If milling operations Fig 1b (C) are considered often higher
Nomenclature feed rates, lower water pressures and multiple passes are
applied [11, 12, 13]. Consequently, adapted models are
α jet impact angle required. A model to predict AWJ milling by expressing the
αd deflection angle MR through an etching rate E, which is a material removal rate
ε deformation wear factor (MRR) associated to the jet for one material has been
ς cutting wear factor introduced [11]. The etching rate E can be related to the particle
C constant by Bitter velocity vp and can be calibrated considering a known kerf
df diameter of the focus tube profile Z0. An extension of the approach was presented to
dO diameter of the orifice include ductile materials [12]. The model considers various
dK kerf depth cutting mechanisms by differentiating between shallow and
E material removal rate (MRR) deeper kerfs. In another milling model [13] the superposition
eK deviations of the kerf depth and the path distance between the trajectories is discussed. So
h smoothing length far, the presented models only predict the 2D kerf profile or the
K constant by Bitter kerf depth rather than the MR on a 3D part. Although the jet
lf length of the focus tube deflection is known to have an effect on the kerf that cannot be
lW length of the workpiece neglected, none of the models actively considers the jet
ṁA abrasive mass flow rate deflection [14]. Thus, most of the models are not able to
M particle mass reproduce waterjet specific artefacts e.g. undercuts that can
n surface normal vector appear during the cutting of a corner. Consequently, the ability
np number of AWJ passes of the models to predict transition conditions, Fig 1c (A, B), for
pW pressure of the water AWJ kerf milling are also limited.
p1, p3 kerf profile constants Numerical models using combinations of smoothed particle
r radius of the AWJ hydrodynamics (SPH) and finite element method (FEM) to
resi workpiece resolution for the coordinates model the AWJ solve this problem [15, 16, 17]. However, the
tc time increment models are limited regarding the amount of predictable
vf feed rate material removal. Another numerical approach using a cellular
vP particle velocity automat has been successfully implemented in a
W material removal (MR) 2D environment [18, 19]. This approach has the potential to be
x, y, z coordinates used in a 3D environment and predict undercuts. Though, the
Z0 kerf profile model has not been applied for 3D kerf cutting or milling.
Due to the lack of 3D models which are able to efficiently
predict AWJ kerf profiles with characteristics such as
1.2. Simulation approaches
undercuts, an analytical model considering the effects of the
secondary jet has been introduced in a previous
The fundamental cutting mechanisms of abrasive particles
investigation [20]. For this model, the approach by Axinte [11]
impacting a solid body have been described by Finnie [2]
has been utilized to develop a 3D kerf prediction for any
through micro-cutting and micro-ploughing (cutting) and by
movement of the AWJ considering the jet deflection. This
Bitter through crack-initiation and propagation (deformation)
model can be applied for milling operations and has proven to
[3]. The combination of the cutting material removal (MR) WC1
reproduce some typical AWJ characteristics. However, so far,
and WC2 and MR through deformation WD can be used to
the model does not consider changes of the MR and varying
predict the overall MR through Eq. 1 to 4. [4, 5]. Sufficient
degrees of the secondary MRR depending on the impact angle.
158 Eckart Uhlmann et al. / Procedia CIRP 102 (2021) 156–161
Author name / Procedia CIRP 00 (2019) 000–000 3

2. Simulation models This approach considers the secondary and tertiary jet
related MR. During the calculation the following steps are
To enhance and improve the analytical model presented conducted. Firstly, the entire primary MR is applied for a time
above two major improvements have been made. First, the increment tc. Secondly, the additional MRRs and MR are
modeling approach of describing the jet through its MRR is conducted pointwise for the points within the target zone. This
combined with fundamental cutting mechanisms. Furthermore, procedure takes place until the jet deflections leave the
the model’s limitation to only one jet deflection is resolved by workpiece area or until the jet’s MRR is used up. In order to
following one finite jet element through the kerf. Furthermore, calibrate the analytical model, experimental results on the MR
a numerical model is introduced to evaluate possible prediction presented in a previous investigation were considered [21]. The
improvements. investigation provides the primary Wp and the secondary Ws
MR for various impact angles α for a parameter range from
2.1. Analytical model of the primary and deflecting jets pw = 100 MPa to pw = 200 MPa. The primary MR Wp is used to
calibrate the coefficients of Bitters equation, Eq. 1 to 4. Thus,
To model the AWJ material removal (MR) process Matlab the experimental results are connected to the fundamental
Release 2020b by The MathWorks Inc., Natick, Massachusetts, material removal mechanisms cutting and deformation. The
USA, is utilized. For this model, the workpiece is represented combination of the jet’s MRR E with the MR mechanisms is
by solid elements which state the percentage of present achieved by considering the ratio of the MR by Bitters equation
material. The number of elements is adjusted depending on the for a given impact angle WB(α) to the MR WB,90 at an impact
workpiece dimensions (x, y, z) and a given resolution resi. The angle of α = 90°, Eq. 7. Applying this approach, the MRR E1(r)
MRR of the jet E1(r) (Fig. 2) is calculated by Eq. 5 and Eq. 6 can still be calibrated by a kerf profile Z0(x) created with high
which were derived in a previous investigation [20]. Eq. 5 feed rates vf at impact angles of approximately α = 90°, Eq. 6.
considers the AWJ radius r and parameters from a known kerf The secondary MR Ws was used to identify the jet’s
profile Z0. The kerf profile Z0 is approximated by two constants secondary MRR E2. To identify the MRR of the secondary
p1 and p2 using Eq. 6. The MRR of the jet E1(r) (Fig. 2) is fixed jet E2, the ratio of the MR at the secondary target material for a
to a virtual focus tube moving above a workpiece following given impact angle W2(α) to the MR W2,0° at the impact
simplified G-Code instructions. angle α = 0° is considered, Eq. 8. Thus, the ratio of the primary
p1 − p3 − 2 p1r 2 MRR at a point E1,i to the primary MR W1,i and the ratio of the
E1 (r ) = (5)
primary MRR E1,i to secondary jet MRR E2,i are defined. The
 1− r2
same Eq. can also be used for the secondary or the tertiary
(r ) p1r 2 + p3
Z 0= (6)
material removal. To test the model’s capabilities to reproduce
The MR of the primary jet at one element W1,i is calculated waterjet specific kerf characteristics, several simulations were
by Eq. 7 through the MRR E1(r) and the cutting mechanisms conducted. A series of feed rates vf were simulated to identify
depending on the impact angle α, Fig. 2. The impact angle α is the application area of the model. Furthermore, the simulations
derived from the local normal vector n1,i. In addition, a jet include slow feed rate vf conditions to create a jet with
deflection of the secondary jet is created at this point and a considerable deflections to observe the kerf formation during
secondary MRR E2,i is assigned to the deflection. The MRR E2,i the cut, undercutting and the exit behavior. In addition,
of the deflection is calculated by Eq. 8. The direction of the multiple passes of the waterjet np are tested at high feed rates vf
secondary jet is derived by the normal vector n1,i assuming a conditions to investigate the jet entry and exit behaviors for
deflection angle αd. With the directional vector of the AWJ kerf milling, Fig. 1c (A).
secondary jet v2, the next material element in the secondary jet
path can be determined. Based on this procedure further 2.2. Jet flow simulation
deflections and MRs can be determined equally.
In addition to the analytical model a numerical model was
vf df r(x, y) developed in order to achieve a second perspective with
Focus tube
α improved physical constraints of the AWJ kerf milling process.
x y E1(r) n1,i The model is based on the smoothed particle hydrodynamics
(SPH) method [16]. Here, small water droplets are modeled as
v2,i, E2,i αd
z W1,i
separate smoothed SPH-particles with a given mass M. SPH-
v3,i, E3,i particles interact with each other and walls. Resultant
W3,i
n3,i n2,i interaction forces are summed on each SPH-particle which
W2,i causes the particles to move according to Newton’s second law.

3. Machining experiments
Fig. 2. Analytical model setup applying an MRR and jet deflections

WB ( ) To evaluate the accuracy of the analytical model,

Wn = En (r ) t (7)
WB ,90 ( ) experiments were carried out on titanium aluminide type
W ( ) Ti-43,5Al-4Nb-1Mo0,1B (TNM-B1) with a waterjet cutting
En +1,i = En ,i 2 (8)
W2,0
 Eckart Uhlmann et al. / Procedia CIRP 102 (2021) 156–161 159
4 Author name / Procedia CIRP 00 (2019) 000–000

machine type HRX 160 L from STM Germany GmbH, with a coefficient of determination of R2 = 96 %. Furthermore,
Schweinfurt, Germany. The machine was equipped with a a deflection angle of αd = 20° was derived from experimental
focus tube having a length of lf = 76.2 mm, a diameter of results [22]. The calibrated functions for the MR by Bitter WB
df = 0.76 mm and an orifice with a diameter of do = 0.25 mm. and MR of a deflected jet W2 are given in Fig. 3.
Garnet Mesh 120 by GMA Garnet Europe GmbH, Hamburg,
Germany, was used as abrasive material. Since the model Table 1. Calibration parameters
addresses milling operations, a water pressure of Parameter Abb. Unit Value
pW = 100 MPa was chosen. Kerf depth dK values were Bitters constant K m/s 30
measured using the optical surface measurement device FRT Bitters constant K1 (m/s)1/2 1
MicroProf 100 by Fries Research & Technology GmbH,
particle velocity vP m/s 403
Bergisch Gladbach, Germany. MR is measured by a LP 620P
Bitters constant C m1/2s5/2/kg 0.13
precision scale by Sartorius AG, Göttingen, Germany.
deformation wear factor ε N/m 2
0.17
4. Results and Discussion cutting wear factor ς N/m2 2.3

4.1. Influence of the impact angle on the jet properties 4.2. Comparison of the kerf depths dK

Fig. 3 shows the average material removal (MR) Wp, Ws Fig. 4 shows the measured and the calculated kerf depths dK
depending on the impact angle α [21]. The measured primary for a broad range of feed rates vf. The predicted maximum kerf
MR Wp shows a typical progression corresponding to the depth dK, max is differentiated by the share of the primary,
results from Bitter [2, 3]. The development of the MR reflects secondary and tertiary material removal. The results show that,
the high hardness of the TiAl material, which causes a shift of considering only the primary material removal, the kerf depth
the MR’s maximum towards higher impacting angles [4]. The prediction is satisfactory for high feed rates.
measured results were used to calibrate the coefficients for Proce ss: Calculation parameters: Workpiece:
Bitter’s Eq. AWJ kerf cutting tc = 0.08 ms T NM-B1 γ-TiAl
Proce ss: Tools: Proce ss parameters: Tools: αd = 20 Process parameters:
AWJ kerf cutting Garnet, Mesh 120, GMA p w, a = 100-200 MPa Garnet, Mesh 120, resx = 0.05 mm p w = 100 MPa
Workpiece: dO = 0.25 mm ṁ A = 250 g/min GMA resy = 0.05 mm ṁ A = 250 g/min
T NM-B1 γ-TiAl dF = 0.76 mm vf, a = 50-83.3 mm/s d O = 0.25 mm resz = 0.01 mm np = 1
lW = 66.6 mm lF = 76.2 mm np = 1 d F = 0.76 mm = 34.7 µm vf
Material removal lF = 76.2 mm = -34.7 µm
x α vf
primary Wp y Maximum kerf depth:
secondary Ws z Measured
combined Wp+s Predicted by the primary jet
by Bitter WB ST M Predicted by the primary and secondary jet
secondary W2 PT M Predicted by the primary, secondary and tertiary jet
1 8
maximum kerf depth d K, max

mg mm
material removal W

0.5 4

0.25 2

0 0
0 22.5 45 90 1 2 5 10 mm/s 100
impact angle α feed rate vf

Fig. 3. Measured material removal [21] and adapted material removal for the Fig. 4. Maximum kerf depth dK depending on the feed rate vf
primary (PTM) and secondary (STM) target material
For feed rates of vf < 10 mm/s, the difference between
The coefficients were determined by minimizing the measured and calculated kerf depth is greater than 10 % and
squared residuals of Eq. 1. Table 1 contains the coefficients, thus not satisfactory anymore. Consequently, it is necessary to
which describe the measured results with a coefficient of consider the secondary and tertiary MR for feed rates of
determination of R2 = 99 %. Besides the primary MR, a Eq. to vf < 10 mm/s. Applying the same 10 % criterium, the model
calculate the secondary MR was strived. To determine the correlates well with the measured kerf depth for feed rates of
secondary material removal, a trigonometric function was vf > 2 mm/s. For feed rates of vf < 2 mm/s the consideration for
implemented. As a result, the experiments can be described additional deflections might be required.
160 Eckart Uhlmann et al. / Procedia CIRP 102 (2021) 156–161
Author name / Procedia CIRP 00 (2019) 000–000 5

4.3. Material removal characteristics for AWJ kerf cutting the kerf depth between eK = 10 % and eK = 35 % [21]. The
deviation calculated by the model is about eK = 31 %. Thus, the
One of the main aims of the model is to be capable of model seems to be capable to predict this transition phenomena
reproducing water jet specific kerf characteristics such as an sufficiently well.
undercut, or the remaining material triangle when the AWJ Proce ss: Calculation parameters:
leaves the workpiece. To test the accuracy of the model, a kerf AWJ milling tc = 0.08 ms
Workpiece: resx = 0.05 mm B B
including those elements was simulated and manufactured. A vf
T NM-B1 γ-TiAl resy = 0.05 mm
cross-section view of the kerf prediction and a measurement of Proce ss parameters: resz = 0.01 mm y
the kerf depth is depicted in Fig. 5. The cross-section view p W = 100 MPa = 34.7 µm
shows that indeed both characteristics are replicated by the ṁ A = 250 g/min = -34.7 µm x
-z
model. Although the shape of the undercut fits well with the vf = 90 mm/s
n p = 15
measured observations, the measured results show the undercut Cross-section B-B
at lower x-values. The behavior of the waterjet exiting the 0
workpiece is a well-known characteristic [6] which is well
reproduced by the model. The kerf depth decrease starts about
mm
0.4 mm before the end of the workpiece. This value shows the
degree of the jet deflection and depends on the parameter

kerf depth d K
settings. This value and the reduction of the kerf depth of about 3
50 % is in good agreement with the experiments.
Process: Calculation parameters:
AWJ kerf cutting tc = 0.08 ms 4.5
Workpiece: resx = 0.05 mm
T NM-B1 γ-TiAl resy = 0.05 mm A vf A
Process parameters: resz = 0.01 mm 6
p W = 100 MPa = 34.7 µm 0 1 2 mm 4
y x x-axis
ṁ A = 250 g/min = -34.7 µm
vf = 3 mm/s
np = 1 -z Fig. 6. AWJ kerf characteristics for transition problems
Cross-section A-A
measured
0 4.5. Comparison of the analytical approach with the flow
simulation

mm Besides the modeling of the AWJ through the MRR and the
jet deflections, a SPH model of the water jet was set up and
kerf depth d K

3
tested. Both models have been applied to the kerf given in
Fig. 7 for a time increment of tc = 30 µs.
Proce ss: Workpiece:
4.5 AWJ milling T NM-B1 γ-TiAl xpos
Proce ss parameters: Etching rate model ypos
p W = 100 MPa parameters: vf
y x
6 ṁ A = 250 g/min tc = 0.03 ms
vf = 5 mm/s resx = 0.05 mm -z
0 0.75 1.5 mm 3
x-axis np = 1 resy = 0.05 mm SPH parameters:
xpos = 2.5 mm resz = 2 µm vp = 403 m/s
Fig. 5. AWJ kerf characteristics for a low feed rate ypos = 0.5 mm = 34.7 µm M = 6.25e-12 kg
= -34.7 µm h = 2.76e-5 m
3.5
4.4. Prediction of transition phenomena for AWJ milling a b

mm
Besides the model’s capabilities to reproduce AWJ specific
characteristics for kerf cutting with low feed rates, the model is
E Pressure p
supposed to help understand transition phenomena Fig. 1c (A,
y-axis

1.75 1 15
B) for AWJ milling operations applying high feed rates. Fig. 6 mg/s
C MPa
shows the kerf depth for the transition of the waterjet into and
0.875 0.5 B 7.5
out of the workpiece for np = 15 passes. Firstly, it can be D
0.25 A 3.75
observed that the variation on the kerf depth, the kerf ground
0 0
waviness, is neglectable small compared to lower feed rates, 0
Fig. 5. This observation corresponds well with other 0 0.75 1.5 mm 3 0 0.75 1.5 mm 3
x-axis x-axis
investigations [13, 14]. In addition, at the beginning and the end
of the workpiece a deepening of the kerf depth is visible. Such
Fig. 7. Model comparison: (a) Material removal of the analytical approach;
phenomena have been observed in reality with a deviation of (b) SPH-flow simulation results
 Eckart Uhlmann et al. / Procedia CIRP 102 (2021) 156–161 161
6 Author name / Procedia CIRP 00 (2019) 000–000

As a result, Fig. 7a shows the MRR calculated by the to further include the physics of the AWJ while being able to
primary and deflection jets model. Furthermore, Fig. 7b depicts predict a reasonable amount of material removal.
the pressure that the SPH model exposes on the kerf. Assuming
that the pressure correlates with the material removal, it is Acknowledgements
apparent that most exposed areas differ significantly. The
primary and deflection jets model shows a lower MRR with This paper is based on results acquired within the project
increasing AWJ radius r from the center of the jet, Fig. 7a No. 317070629, which is kindly supported by the Deutsche
detail A. An exception is the region in opposition to the feed Forschungsgemeinschaft (DFG).
rate direction where the highest MR appears, Fig. 7a detail B.
This is the area where the secondary jet impacts the material. References
The resulting MR of the tertiary jet deflections are also visible
within the kerf, Fig. 7a detail C. The SPH model predicts the [1] Axinte DA, Karpuschewski B, Kong MC, Beaucamp AT, Anwar S, Miller
strongest impact at maximum AWJ radius, Fig. 7b detail D. D, Petzel M. High Energy Fluid Jet Machining (HEFJet-Mach). CIRP Ann
Manuf Technol 2014;63:751-71.
This result is probably caused due to the neglection of the [2] Finnie I. The Mechanism of Erosion of ductile Metals. Proceedings of the
velocity profile within the water jet in this model. However, the third U. S. National Congress 1958:27-32.
more interesting attribute of the SPH model is the jet deflection [3] Bitter JGA. A study of erosion phenomina, Part I. Wear 1963;6:5-21.
which indicates that the effects of the jet reach up until the end [4] Bitter JGA. A study of erosion phenomina, Part II. Wear 1963;6:169-90.
of the kerf. [5] Neilson JH, Gilchrist A. Erosion by a stream of solid particles. Wear
1968;2:111-22.
[6] Hashish M. A Modeling Study of Metal Cutting With Abrasive Waterjets.
5. Conclusion and Outlook J Eng Mater Technol 1984;106:88-100.
[7] EITobgy M, Ng EG, Elbestawi MA. Modelling of Abrasive Waterjet
In this paper an analytical model based on the AWJ’s Machining. CIRP Ann Manuf Technol 2005;1:285-8.
material removal rate considering multiple jet deflections was [8] Zeng J, Kim TJ. An erosion model for abrasive waterjet milling of
polycrystalline ceramics. Wear 1996;199:275-82.
combined with fundamental material removal mechanisms. [9] Hoogstrate AM. Towards High-Definition Abrasive Waterjet Cutting.
The model was calibrated using experimental results and was Delft: Technische Universiteit Delft; 2000.
tested for its capability to predict AWJ kerf characteristics. In [10]Pi VN. Performance enhancement of abrasive waterjet cutting. Delft:
addition, a jet flow simulation applying a SPH model has been Technische Universiteit Delft; 2008.
introduced. The main findings can be summarized as follows. [11]Axinte DA, Srinivasu DS, Billingham J, Cooper M. Geometrical modelling
of abrasive waterjet footprints: A study for 90° jet impact angle. CIRP Ann
Manuf Technol 2010;59:341-6.
• The model based on the primary and deflection jets enables [12]Kong, M. C.; Anwar, S.; Billingham, J.; Axinte, D. A. Mathematical
a suitable prediction of the kerf depth for feed rates of modelling of abrasive waterjet footprints for arbitrarily moving jets:
vf ≥ 2 mm/s. The model has been qualified for a pressure of Part I - single straight paths. Int J of Mach Tools Manuf 2012;53:56-68.
pw = 100 MPa and the given focus tube. Due to the [13]Van Bui H, Gilles P, Sultan T, Cohen G, Rubio W. Adaptive speed control
for waterjet milling in pocket corners. Int J Adv Manuf Technol
fundamental approach of the simulation a transfer to 2019;103:77-89.
different parameter settings should be feasible. [14]Fowler G, Shipway P, Pashby I. Abrasive water-jet controlled depth
• The model qualitatively predicts the AWJ kerf cutting milling of Ti6Al4V alloy - an investigation of the role of jet-workpiece
characteristics undercut, kerf ground waviness and the traverse speed and abrasive grit size on the characteristics of the milled
exiting behavior of the cutting process. material. J Mater Process Technol 2005;161:407-14.
[15]Dong X, Li Z, Jiang C, Liu Y. Smoothed particle hydrodynamics (SPH)
• The presented approach enables the prediction of the AWJ simulation of impinging jet flows containing abrasive rigid bodies. Comp
kerf milling characteristics. The entry and the exiting Part Mech 2019;3:479-501.
behavior of the cutting process are well reproduced by the [16]Guo L, Deng S, Yang X. Numerical simulation of abrasive water jet cutting
simulation. chemical pipeline based on SPH coupled FEM. Chem Eng Trans
• The jet flow simulation through the SPH approach implies, 2016;51:73-8.
[17]Jianming W, Na G, Wenjun G. Abrasive waterjet machining simulation by
that the effects of the jet deflection reach considerably SPH method. Int J Adv Manuf Technol 2010;50:227-34.
further than the prediction by the model based on the [18]Jerman M, Valentinčič J, Lebar A, Orbanić H. The Study of Abrasive
primary and deflection jets. Water Jet Cutting Front Development using a Two-Dimensional Cellular
Automata Model. J Mech Eng 2015;5:292-302.
Although some measured kerf parameters are in good [19]Orbanic H, Junkar M. Simulation of abrasive water jet cutting process: Part
2. Cellular automata approach. Exp Mech 2004;6:1171-84.
quantitative agreement with the prediction, it is currently [20]Uhlmann E, Männel C. Modelling of the Kerf Formation through Primary
assumed that the model of the primary and deflection jets only and Secondary Jet Energy for the Abrasive Waterjet. WJTA Conference
qualitatively predicts these characteristics. Nevertheless, the and Expo New Orleans 2019:A1.
model can be used to test compensation methods to achieve [21]Uhlmann E, Männel C. Effects of the Cutting Angle on the Kerf Formation
better results for AWJ kerf cutting and milling. Furthermore, During Near-Net-Shape Fabrication with the Abrasive Water Jet. Lect
Notes Mech Eng 2021;178:252-61.
the model fosters a better understanding of the AWJ material [22]Uhlmann E, Männel C, Braun T. Efficient abrasive water jet milling for
removal process. For the future, a combined approach of near-net-shape fabrication of difficult-to-cut materials. Int J Adv Manuf
material removal mechanisms with SPH should be considered Technol 2020;111:685-9