Int J Adv Manuf Technol

DOI 10.1007/s00170-017-0528-2

ORIGINAL ARTICLE

An analytical model for predicting the machining deformation

of a plate blank considers biaxial initial residual stresses
Hanjun Gao 1 & Yidu Zhang 1 & Qiong Wu 1 & Jing Song 1

Received: 14 November 2016 / Accepted: 7 May 2017

# Springer-Verlag London 2017

Abstract The initial residual stress in a blank is one of 1 Introduction

the main causes of machining deformation that has be-
come a major concern in manufacturing aerospace mono- Monolithic components are widely applied in the aerospace
lithic structures. An analytical prediction model for ma- industry because of their excellent mechanical performance
chining deformation that considers biaxial residual stress- and lightweight. Up to 90% of these materials can be removed
es is proposed to investigate the quantitative relationship from the blank during computer numerical control (CNC) ma-
between deformation and initial residual stress. Finite el- chining due to their thin-walled structure [1]. In the forming
ement method (FEM) simulations and machining experi- process, which includes rolling, forging, and extrusion, resid-
ments are then conducted to validate the accuracy of the ual stress is inevitably generated in the blank [2, 3]. The orig-
analytical model. Results calculated using the analytical inal equilibrium of the blank is disrupted as the material is
model are consistent with the FEM and experiment re- continuously removed, which redistributes internal stresses
sults. Eventually, the quantitative relationships between and deforms the component. Machining deformation, which
deformation and initial residual stress under three typical is the primary cause of the dimensional and geometric devia-
machining strategies are determined. The effects of work- tions of monolithic components, has become a major problem
piece position in the blank on machining deformation are in the aviation manufacturing industry.
summarized based on these quantitative relationships. Machining deformation is basically caused by four factors,
namely, clamping loads, cutting loads (e.g., cutting force and
cutting temperature), machining-induced residual stresses,
Keywords Machining deformation . Analytical model . and the initial residual stresses of the aforementioned blank.
Biaxial initial residual stresses . Plate blank A number of studies on machining deformation that consider
one or several of these factors from various perspectives have
been conducted. The prediction, control, and correction of
machining deformation are the main directions of the present
study.
Electronic supplementary material The online version of this article
(doi:10.1007/s00170-017-0528-2) contains supplementary material,
which is available to authorized users.
2 Literature review
* Qiong Wu
wuqiong@buaa.edu.cn 2.1 Effect of clamping loads

The positions of locators, clamps, and supports affect the ri-

1
State Key Laboratory of Virtual Reality Technology and Systems, gidity of the workpiece in machining. The deformations and
School of Mechanical Engineering and Automation, New Main
Building Room B313 Beijing University of Aeronautics and
surface errors occur due to the low rigidity of the workpiece in
Astronautics, XueYuan Road No.37, Haidian District, local positions. Studies on fixture systems and machining de-
Beijing 100191, People’s Republic of China formation can be summarized as follows: (1) modeling and
 Int J Adv Manuf Technol

analysis of workpiece–fixture systems [4–8] and (2) optimum minimize the tool deflection and constraints of surface rough-
fixture layout design [9–13]. Liao and Hu established a fix- ness and tool life, Saffar and Razfar [26] optimized the ma-
ture–workpiece system model that considered the external chining parameters using a genetic algorithm (GA).
loads, fixture compliances, workpiece and its locators/clamps, Although most of the cutting heat is taken away by the
surface-based contact interaction, and dynamic stiffness of chips, the unevenly distributed temperature field in the work-
such system [4]. To predict machined surface quality, piece will cause the plastic deformation. An FEM model that
Satyanarayana and Melkote [5] proposed accurate considered transient thermo-mechanical loading conditions
methods for modeling spherical–planar and planar–pla- was established by Rai and Xirouchakis [27] to predict tran-
nar locator–workpiece and clamp–workpiece contacts sient temperature distributions that would result in elastic–
using a finite element method (FEM) software. The ef- plastic deformations. To determine the heat flux entering a
fects of various finite element modeling parameters of a workpiece, a mesoscopic scale thermo-elastic FEM model
fixture–workpiece system on workpiece deformation, was developed by Puls et al. [28]. Thereafter, the transient
such as friction and mesh density, were investigated temperature field and thermo-elastic deformations for a case
by Siebenaler and Melkote [8]. study that represented two turning cycles used to manufacture
Liu et al. [9] optimized the positions of the locators of a a gear shaft were obtained via a macroscopic scale FEM
thin-walled workpiece based on an FEM model and a heuristic model.
algorithm. Chen et al. [10] established a multi-objective model
to reduce deformation degree and increase the distribution 2.3 Effect of machining-induced residual stresses
uniformity of deformation based on a genetic algorithm. To
reduce the computation time and cost of finite element analy- In recent years, machining-induced residual stress has been
sis, Lu and Zhao [11] proposed an approach to optimize the extensively researched by scholars. Most studies were con-
fixture layout of deformable sheet metal workpieces based on cerned about the generation mechanism [29] of machining-
the 4–2–1 locating scheme. induced residual stress and the influences of different cutting
parameters [30–34] and cutting tool geometric parameters
2.2 Effects of cutting force and cutting temperature [35–37]. The effect of induced residual stress on fatigue life
[38, 39] was also investigated.
Elastic deflections of workpiece and cutting tools caused by The release of machining-induced residual stress also
the cutting force lead to overcut and undercut in local posi- causes deflections after machining. However, the machining-
tions due to the low rigidity, which affects the geometric and induced residual stress is not a major factor in most cases,
dimensional accuracy. When the rigidity of workpiece is high because the layer depth of the induced stress is generally less
enough, the impact of the cutting force is limited. The impact than 0.2 mm [37]. Some researchers have studied the relation-
of the cutting force should been controlled as one of the main ship between induced residual stress and deformation. Jiang
factors for low rigidity parts, such as parts with thin-walled et al. [40] proposed a mathematical model based on the me-
stiffening ribs. Ratchev et al. contributed significantly to the chanics of material theory and a response surface design meth-
research on cutting force-induced deformation over the past od to predict the distortion caused by machining-induced re-
15 years by conducting theoretical modeling, FEM simula- sidual stresses. Multiple regression analysis was conducted by
tions, and machining experiments [14–20]. Izamshah et al. [41] to determine the correlation between de-
Ning et al. [21] used FEM to achieve machining deforma- flection and six input variables, namely, speed, feed rate, ra-
tion caused by the cutting force. They proposed the numerical dial cut depth, wall thickness, wall height, and wall length. Li
control (NC) compensation method to control deformation et al. [42] analyzed the redistribution law of residual stresses
based on the simulation results. An analytical model and an with different cut depths and provided insights into the cou-
artificial neural network (ANN) model were adopted by pling action of initial residual stress and machining-induced
Benardos et al. [22] to predict elastic deflection under cutting residual stress. Huang et al. [43] investigated the coupling
forces when turning. A theoretical model for calculating the action of initial residual stress and machining-induced residual
deformations of thin-walled plates based on small deforma- stress on machining deformation through experiments.
tion equations was proposed by Tang and Liu [23]. An esti- Machining-induced residual stress was the primary factor of
mation methodology of deflection forces for selecting milling distortion under their research conditions.
tool paths on complex surfaces was presented in literature [24]
to minimize dimensional errors caused by tool defects. In 2.4 Effect of initial residual stresses
literature [25], machining deflection caused by the cutting
force, which was obtained using an ANN method, was pre- Clamping loads, cutting force, and cutting temperature gener-
dicted via FEM modeling. Tool path compensation was gen- ally lead to the plastic deformation in local positions. The
erated based on the FEM results to reduce deflection. To overall deflection is one of the main problems in the
Int J Adv Manuf Technol

machining of plates, frames, and beams [44, 45]. The main obtaining quantitative relationships [57, 58]. An analytical
cause of the overall deflection is the release and redistribution deformation prediction model for the plate blank that consid-
of the initial residual stress, especially for high material re- ered uniaxial residual stress was proposed by Shin [59] in
moval rate [46]. A number of studies have focused on the 1995. This model has been quoted as a theoretical basis by
influence of initial residual stress on machining deformation. many researchers [43, 46, 54].
When initial residual stress is considered, a finite element However, in recent studies on the machining deformation
model is the most commonly used prediction method of a plate blank [49, 53, 55, 56], biaxial initial residual stresses
[44–47]. Wang et al. [45], Dong and Ke [46], and Liu et al. must be considered based on the measurement or simulation
[47] predicted machining deformation using a finite element results. Therefore, an analytical deformation prediction
model with initial residual stresses and conducted verification model for the plate blank that considers biaxial residual
experiments. Guo et al. [48] presented a method called stress is derived in this study based on theory of plates
“house-building frame modeling” for the finite element and shells. Subsequently, the model is validated via
modeling of large monolithic multi-frame components. FEM simulations and machining experiments. The influ-
Cerutti and Mocellin [49] applied an automatic mesh ence coefficients of the initial residual stresses in each
refinement technology to predict the finite element mod- layer are eventually determined, and the effects of
el and parallel technology for improving computational workpiece position in the blank on machining deforma-
efficiency. Richter-Trummer et al. [50] proposed a mod- tion are summarized.
ified layer-removed method to determine through-the-
thickness residual stresses accurately.
Husson et al. [51] found that shaft bending could be de-
3 Machining deformation prediction model
creased significantly by improving the consistency of residual
stress and geometric distribution before heat treatment. To
The following assumptions are made for the modeling
investigate the influence of mortise broaching sequences on
process:
strain energy and deformation, a finite element model for tur-
1. The blank material is isotropic, homogeneous, and
bine disk machining processes that applied the minimum po-
elastic.
tential energy theorem was presented by Liu et al. [2]. Huang
2. The residual stresses in the thickness direction are equal
et al. [3] found that the deformation caused by initial residual
to zero.
stress accounted for 90% of the total deformation of an alu-
3. The residual stresses in the length and width directions
minum alloy 7050-T7451 component, whereas the deforma-
have uniform magnitude at the same depth.
tion caused by machining-induced residual stress accounted
4. The influences of clamping loads, cutting force, cutting
for the remaining 10%. Similarly, Yang et al. [52] showed
temperature, and induced residual stresses on machining de-
that initial residual stress in a blank was the main effect
formation are disregarded.
element of machining distortion for aluminum alloy air-
craft monolithic component, whereas cutting force and
temperature were the main effect elements for titanium 3.1 Theory of plates and shells
alloy.
Zhang et al. [53] used a contour method to acquire the 2D The theory of plates and shells can be used in the case with
cross-sectional residual stress map of an aluminum alloy plate. (1/80–1/100) < h/b < (1/5–1/8), where h represents thickness
Wu et al. [54] adopted a quasi-symmetric machining method and b represents width. When uniform moment is applied to
during the machining of thin-walled components. Maximum the four free edges of a plate (Fig. 1; Mx and My represent the
deformation decreased from 0.3246 to 0.0589 mm compared moment that causes the bending of the plate in the X and Y
with the traditional one-side machining method. Tang et al. directions, respectively), the curvatures in the X and Y direc-
[55] and Rai [56] et al. established a prediction model tions, which are respectively represented by 1/Rx and 1/Ry, can
by considering all the four aforementioned factors. They
validated the accuracy of their model by performing
experiments.

2.5 Summary of literature review

Although many FEM prediction models have been

established, the quantitative relationship between initial resid-
ual stress and machining deformation must be further investi-
gated. An analytical model is an effective approach for Fig. 1 Schematic diagram of the plate boundary conditions
 Int J Adv Manuf Technol

be expressed by a two-order partial derivative of deflection disequilibrium in one direction can be equivalent to the defor-
w(x, y) as follows [60]: mation caused by the uniform moment applied to the two
edges perpendicular to this direction [53]. Therefore, when
1 ∂2 wðx; yÞ the effect of biaxial stress disequilibrium is considered, the
¼ ; ð1Þ
Rx ∂x2 boundary conditions of the blank can be regarded to be the
1 ∂2 wðx; yÞ same as those shown in Fig. 1.
¼ : ð2Þ In this study, the curvature is assumed to be positive if the
Ry ∂y2
plate is bent upward, whereas it is assumed to be negative if
the plate is bent downward. The length, width, and thickness
On the basis of Hooke’s law and the moment equilibrium of a rectangular blank are represented by L, W, and H, respec-
condition, Eqs. (3) and (4) can be derived as follows: tively (Fig. 2). For convenience, the workpiece is divided into
n uniform layers. The X and Y direction average stresses in
1 ∂2 wðx; yÞ M x −μM y each layer are denoted as σx01, σx02…, σx0n and σy01, σy02…,
¼ ¼ ; ð3Þ
Rx ∂x2 Dð1−μ2 Þ σy0n, respectively.
1 ∂2 wðx; yÞ M y −μM x After the first layer is removed, the relationship between
¼ ¼ ; ð4Þ the average stress in the first layer and the equivalent moment
Ry ∂y2 Dð1−μ2 Þ
can be obtained from Eqs. (7) and (8):
Eh3
D¼ ; ð5Þ h1
12ð1−μ2 Þ −σx01  t  dy ¼ M x dy; ð7Þ
2
where D is the bending stiffness of the plate, E is the elastic h1
−σy01  t  dx ¼ M y dx; ð8Þ
modulus, and μ is Poisson’s ratio. 2
The deflection of the plate center is considered as zero.
Therefore, the deflection function can be expressed as where t denotes the thickness of each layer, and h1
denotes the thickness of the workpiece before layer 1
M x −μM y 2 M y −μM x 2 is removed.
wðx; yÞ ¼ x þ y
2Dð1−μ2 Þ 2Dð1−μ2 Þ After Eqs. (7) and (8) are eliminated, these two equations
1 2 1 2 are substituted into Eqs. (3) and (4). The expression of the
¼ x þ y: ð6Þ curvature can be obtained from Eqs. (9) and (10).
2Rx 2Ry


1 1 6th1 σx01 −μσy01
− ¼− ; ð9Þ
Rx1 Rx0 Eh32
3.2 Curvature formed by the redistribution of residual  
stress 1 1 6th1 σy01 −μσx01
− ¼− ; ð10Þ
Ry1 Ry0 Eh32
During the forming process, such as rolling, forging, and ex-
trusion, residual stress is inevitably induced in the blank. To where 1/Rx0 and 1/Ry0 are the curvatures in the X and Y direc-
maintain its shape, the residual stress profile should satisfy the tions, respectively, before layer 1 is removed; 1/Rx1 and 1/Ry1
equilibrium conditions of force and moment. After a part of are the curvatures in the X and Y directions, respectively, after
the material is removed, the original equilibrium is disrupted. layer 1 is removed; and h2 is the thickness of the workpiece
Bending deformation caused by the effect of stress before layer 2 is removed.

Fig. 2 Newly generated stresses

in the plate blank
Int J Adv Manuf Technol

3.3 Redistribution of the residual stress

3.3.1 Generation mechanism analysis

The initial residual stress in the first layer is assumed to

be compressive. Thus, the curvature of the remaining
part is positive after the first layer is removed. Under
these circumstances, the lengths of the layers in the
upper part are decreased, whereas the lengths of the
layers in the lower part are increased. Therefore, a neu-
tral layer must exist, the length of which remains un- Fig. 3 Bending deformation state before and after the removal of the first
changed. The length of each layer, namely, the arc layer in the cross section
length, increases evenly with the radius from top to
bottom, because the bending curvature among the layers
is the same. Thus, the newly generated stresses in the Equations (17) and (18) can be obtained based on the geo-
layers from top to bottom should also change evenly metric relations as follows:
(Fig. 2).
ΔL0 ax0
Equations (11) to (16) can be deduced by considering the εx0 ¼ ¼ ; ð17Þ
L Rx0
equilibrium conditions of force and moment as follows:
ΔL1 ax1
ax1 Sx1 bx1 Sx1
0 εx1 ¼ ¼ ; ð18Þ
σx01 t þ þ ¼ 0; ð11Þ L Rx1
2 2
In the same way, Eqs. (19) and (20) are obtained:
ax1 þ bx1 ¼ h2 ; ð12Þ ay0
εy0 ¼ ; ð19Þ
Ry0
0 bx1 S x1 ay1
−S x1 ¼ ; ð13Þ εy1 ¼ ; ð20Þ
ax1 Ry1
ay1 Sy1 by1 S0 y1
σy01 t þ þ ¼ 0; ð14Þ where ΔL0 and ΔL1 are the arc lengths of the top surface of
2 2
the remaining part before and after the removal of the first
ay1 þ by1 ¼ h2 ; ð15Þ layer, respectively; ax0 and ay0 are the distances between the
top surface and the middle surface of the neutral layer before
by1 S y1
−S 0 y1 ¼ ; ð16Þ the removal of the first layer in the X and Y directions, respec-
ay1
tively; εx0 and εx1 are the strains on the top surface of the
where Sx1 and Sy1 respectively denote the newly generated X remaining part in the X direction before and after the removal,
and Y direction stresses on the top surface, S′x1 and S′y1 re- respectively; εy0 and εy1 are the strains on the top surface of
spectively denote the newly generated X and Y direction the remaining part in the Y direction before and after the re-
stresses on the bottom surface, ax1 and ay1 respectively denote moval, respectively; Rx0 and Rx1 are the curvature radii of the
the distance between the top surface and the middle surface of workpiece in the X direction before and after the removal,
the neutral layer, and bx1 and by1 respectively denote the dis- respectively; and Rx0 and Rx1 are the curvature radius of the
tance between the top surface and the middle surface of the workpiece in the Y direction before and after the removal,
neutral layer. respectively.
When the layer number n is sufficiently large, ax0 can be
approximately equal to ax1. With regard to Eqs. (21) and (22),
3.3.2 Stress redistribution which are deduced from the generalized Hooke’s law, the
relationship between the newly generated stress and the strain
A cross section parallel to the XOZ plane of the workpiece can be expressed in Eqs. (23) and (24):
is shown in Fig. 3. The arc lengths of the layers above the
E

neutral layer are smaller than the length of the workpiece L σx ¼ εx þ με y ; ð21Þ
because of the positive curvature caused by the removal of 1−μ2
the first layer. Although the position of the neutral layer
changes with such removal, the arc length of the neutral E  
σy ¼ εy þ μεx ; ð22Þ
layer is always equal to L. 1−μ2
 Int J Adv Manuf Technol

    
E 1 1 1 1 The calculation formulas for the curvature and stresses in
S x1 ¼ a x1 − þ μa y1 − ; ð23Þ the X direction by layers can be represented in Eqs. (32) to
1−μ2 Rx1 Rx0 Ry1 Ry0
(36):
    

E 1 1 1 1 6th σ −μσ
S y1 ¼ a − þ μa − ; ð24Þ 1 1 i x;i−1;1 y;i−1;1
1−μ2
y1
Rx1 Rx0
x1
Ry1 Ry0 − ¼ 3
; ð32Þ
Rx;i−1 Rx;i Ehiþ1
hiþ1 ð3hi þ hiþ1 Þ
when t, h2, h1, σx01, σy01, E, and μ are regarded as the known axi ¼ ayi ¼ ; ð33Þ
6hi
quantities, and Eqs. (9) and (10) are introduced into Eqs. (22)
ð3hi þ hiþ1 Þtσx;i−1;1
and (23), respectively, an equation set that includes Eqs. (11), S xi ¼ − ; ð34Þ
(12), (13), (14), (15), (16), (23), and (24) and eight unknown h2iþ1
quantities (i.e., ax1, bx1, ay1, by1, Sx1, Sx1′, Sy1, and Sy1′) can be S xi h
t i
established. When the equation set is solved, the unknown S x;i; j ¼ axi − j−i þ ð j ¼ i; i þ 1…; n−1Þ; ð35Þ
axi 2
quantities can be expressed as i
σx;i; j ¼ σx;iþ j − ∑ S x;k;iþ j−1 ð j ¼ 1; 2…; n−iÞ: ð36Þ
h2 ð3h1 þ h2 Þ k¼1
ax1 ¼ ay1 ¼ ; ð25Þ
6h1
For Eqs. (31)–(35), i = 1, 2, 3…, n − 1,
ð3h1 þ h2 Þtσx01 where Rx,i – 1 and Rx,i are the curvatures in the X direction after
S x1 ¼− ; ð26Þ
h22 the (i − 1)th and ith layers are removed, respectively; σx,i − 1,1
ð3h1 þ h2 Þtσy01 and σy,i − 1,1 are the average stresses on the top layer of the
S y1 ¼ − ; ð27Þ remaining part in the X and Y directions, respectively, after the
h22
(i − 1)th layer is removed; hi is the thickness of the workpiece
Thereafter, the newly generated stresses in each layer before the ith layer is removed; axi is the distance between the
produced by the removal of the first layer can be ob- top surface and the middle surface of the neutral layer in the X
tained based on the linear relation of the stresses as direction after the ith layer is removed; Sxi is the newly gen-
follows: erated X direction stress on the top surface of the remaining
part caused by the removal of the ith layer; Sx,i,j is the newly
S x1 h
t i generated X direction stress in the jth layer of the original
S x1; j ¼ ax1 − j−1 þ ð j ¼ 1; 2; …; n−1Þ; ð28Þ
ax1 2 blank caused by the removal of the ith layer; and σx,i,j is the
X direction residual stress in the jth layer of the remaining part
σx1; j ¼ σx0; jþ1 −S x1; j ð j ¼ 1; 2; …; n−1Þ; ð29Þ after the ith layer is removed.
The formulas in the Y direction can be expressed in the
S y1 h
t i same manner. In addition, the deflection function w(x, y) can
S y1; j ¼ ay1 − j−1 þ ð j ¼ 1; 2; …; n−1Þ; ð30Þ be obtained in Eq. (36) based on Eq. (6):
ay1 2
1 2
wi ðx; yÞ ¼ x
σy1; j ¼ σy0; jþ1 −S y1; j ð j ¼ 1; 2; …; n−1Þ; ð31Þ 2Rxi
 
1 2 L L W W
where Sx1,j and Sy1,j are the average newly generated X and Y þ y − ≤ x≤ ; − ≤ y ≤ ; ð37Þ
2Ryi 2 2 2 2
direction stresses, respectively, in the jth layer of the remain-
ing part after the first layer is removed; and σx1,j and σy1,j are where Ry,i is the curvature in the Y direction after the ith layer
the X and Y residual stresses, respectively, in the jth layer of is removed.
the remaining part after the first layer is removed.

3.4 Prediction model by layer 4 Case study

In Sects. 3.2 and 3.3, the calculation formulas for the bending To validate the accuracy of the proposed model, a batch of
curvature and the stress distribution after the removal of the 7075 T6 aluminum alloy extruded plates is selected as speci-
first layer are deduced. When the previous calculation results mens for the experiments. The dimension parameters and ma-
are adopted as the initial conditions of the next removal, the terial properties of the plates are listed in Table 1. In this
expressions of deformations and stresses by layers can be section, the residual stress profile of the plate is measured
obtained. using layer-removed method. Subsequently, analytical model
Int J Adv Manuf Technol

Table 1 Setup of the parameters used in the case study

Parameters Values Description

L (mm) 200 Workpiece length

W (mm) 124 Workpiece width
H (mm) 25 Workpiece height
E (MPa) 71,700 Elastic modulus
μ 0.33 Poisson ratio
n 25 Number of layer
k 20 Number of layer to be removed
t (mm) 1 Thickness of each layer Fig. 5 Initial residual stress profile

For comparison and validation, a corresponding 3D finite

calculations, FEM simulations, and machining experiments element prediction model that considers biaxial residual
are performed. stresses is established using the ANSYS software. The steps
are as follows: (1) selecting SOLID 186 as the element type,
4.1 Measurements of the initial residual stress (2) defining material properties, (3) building the geometric
model, (4) generating the finite element model that consists
In this study, the layer-removed method [61] is employed to of 25 uniform layers along the thickness direction, (5) apply-
obtain the residual stress profile along the thickness direction. ing biaxial residual stresses to each layer, (6) removing the
Prism (Stresstech Group), which is a residual stress measure- material layer-by-layer via element birth and death technolo-
ment device based on the hole-drilling and electronic speckle gy, and (7) outputting the simulation results.
pattern interferometry (ESPI) technique, and a micro-milling Moreover, to investigate the influence of workpiece posi-
cutter (Richards Micro Tool) are used for measuring (Fig. 4). tion in the blank on the final deformation, three removal strat-
Material is removed by a XK7132 vertical milling machine egies (Fig. 7) are adopted in the calculations and simulations
with a three-flute cemented carbide Φ14 end mill. The speci- of the analytical model. The strategies can be described as
men is assumed to be divided into uniform 25 layers (the
thickness of each layer is 1 mm). Thus, the residual stress
profile along the thickness direction is shown in Fig. 5.

4.2 Implementation of the analytical model and FEM

simulations

To obtain the deformation and stress results of the proposed

model, calculation programs are written and implemented
using the MATLAB 2012b software. The parameter setups
are shown in Table 1, and initial residual stresses are assigned
according to Fig. 5. The program flowchart is presented in
Fig. 6.

Fig. 4 Measurement experiments of initial residual stresses using Prism

device Fig. 6 Flowchart of the analytical calculation program
 Int J Adv Manuf Technol

Fig. 7 Three material removal

strategies

follows: (A) removing all the 20 layers from the top of the opposite edge (edge C′D′) because of the bending deformation
blank (layers 1–20), (B) removing 10 layers from the top of the plate. Subsequently, the deflections of vertices C′ and D′
(layers 1–10) and 10 layers from the bottom (layers 16–25), can be determined using a set of feeler gauges that are avail-
and (C) removing 14 layers from the top (layers 1–14) and 6 able within a thickness range of 0.005–1 mm. As indicated in
layers from the bottom (layers 20–25). the geometric relation, the measured deflection of vertex C′ or
D′ under this situation is twice as large as the actual bending
deflection of vertex C′ or D′ in small deformation problems.
4.3 Experiments
The derivation process is omitted in this study.
Machining experiments are also conducted to verify the effec-
tiveness of the proposed model. Three uniform specimens,
which are cut from the same plate as the specimen for residual
stress measurements, are used to measure machining deflec-
tions. Hence, these specimens are considered to have the same
residual stress profile, size, and material properties as the mea-
sured specimen.
The experiments are performed on a XK7132 vertical mill-
ing machine. Cutting tools and cutting parameters are selected
to reduce the effects of machining-induced residual stresses.
The magnitude of machining-induced residual stresses de-
creases with the decrease in cutting depth [42] and feed rate
[35] and the increase in tool diameter [62]. When the cutting
speed reaches a certain value, the cutting speed will not sig-
nificantly affect the generation of residual stresses [63]. Thus,
the low feed rate, small cutting depth, and large tool diameter
are required. Considering machining efficiency, the cutting
speed is set as 132 m/min, the feed rate is 250 mm/min, and
the cutting depth is 1 mm. A three-flute cemented carbide Φ14
end mill is employed. As shown in Fig. 8a, four
20 mm × 20 mm rectangular areas are reserved for clamping
convenience.
Three removal strategies are applied to three specimens A,
B, and C. Ten sets of deflection results are measured for spec-
imen A, whereas only the final deflections of specimens B and
C can be measured because of the change on the machined
surface.
After each removal, the specimen is placed freely on a flat
workbench. To ensure the stability and accuracy of the mea-
surement, edge A′B′, of the specimen (Fig. 8b), is pressed with Fig. 8 Machining experiments. a Machining process. b The deflection
a solid steel block. Consequently, deflections appear at the measurement using feeler gauges
Int J Adv Manuf Technol

Fig. 9 Deformation results. a

The average deformation values
of two measuring points in an
analytical predicted model, FEM
simulations, and experiments
under removal strategy A. b
Residual stresses in the top 1-mm
layer. c Final deformation
comparison among strategies A,
B, and C. d Deformation
comparison between analytical
model and FEM under removal
strategies A, B, and C

5 Results and discussions Although the magnitude is slightly larger, the measured
deformation has the same changing tendency with the analyt-
5.1 Model prediction results ical results. The maximum analytical predicted and measured
deformation is 0.156 and 0.235 mm, respectively, and the final
In Fig. 9a–d, the deflection results of the analytical model, deformation is 0.112 and 0.158 mm, respectively. The analyt-
finite element model, and experiments are compared. The four ical predicted deformation under strategies B and C is −0.001
vertices exhibit the same deflection in the analytical model and 0.001 mm, and the measured deformation is 0.031 and
and finite element model results because of the symmetry of 0.033 mm. The final deformation under removal strategies B
the blank. Nevertheless, the experiment results from the two and C are both significantly smaller than that under strategy A
measurement points are not completely the same because of (Fig. 9c). Analytical results also agree well with FEM results
the errors. Thus, the mean value of the two measuring in the deformation prediction (Fig. 9d). The deformation un-
points is presented as the experiment result in the fig- der strategies B and C begins to decrease after the removal of
ures. The deformation surface of the analytical model the 10th and 14th layer, respectively, because of the reverse
and the bottom surface of the finite element model are machining.
provided for comparison in Fig. 10. The average stress- Besides, the deflections of the analytical model are consis-
es of the center section are selected to compare with the tent with the finite element model on the entire surface of the
analytical results in Fig. 11. workpiece (Fig. 10). Stress redistributions of the analytical
It is found that the deformation predicted by the proposed model are also consistent with those of the FE model
model increases initially and decreases afterwards as the ma- (Fig. 11). The maximum stress deviation is 0.21 MP.
terial is gradually removed under strategy A (Fig. 9a). The comparison shows that the analytical results under all
According to Eqs. (7) and (8), the changing tendency of the three removal strategies exhibit good agreement with the FEM
deformation is determined by the stress state in the top layer. simulation and experimental results. Compared with the ex-
The stresses in the top layer before the removal of the 16th periment results, the maximum relative error of the analytical
layer are tensile, and turn to compressive stress after the 16th model is 53.67%, which occurs after the fourth layer is re-
layer is removed (Fig. 9b). It explains the reason of the defor- moved under strategy A. Compared with the finite element
mation decrease. model results, the average error of the deflection is only
 Int J Adv Manuf Technol

Fig. 10 Comparison of the

deformation surface between
analytical model and finite
element model results under
removal strategy A. a After the
removal of the first 5 layers. b
After the removal of the first 10
layers. c After the removal of the
first 15 layers. d After the removal
of the first 20 layers

2.18%. The main reason for such deviation between the ex- model results is that the analytical and finite element models
periment results and the analytical model or finite element do not consider clamping force cutting loads and machining-

Fig. 11 Residual stress results

calculated using the analytical
model and simulations under
strategy A. a Removal of the first
5 layers. b Removal of the first 10
layers. c Removal of the first 15
layers. d Removal of the first 20
layers
Int J Adv Manuf Technol

induced residual stresses. However, the effects of these factors where Cx,i and Cy,i are the influence coefficients of the
on machining deformation cannot be completely eliminated in initial residual stresses in the ith layer of the blank,
the actual machining process. This situation can also explain which are determined by L, W, H, E, μ, n, k and the
the deviation under strategies B and C in Fig. 9c. material removal strategy.
Thus, the influence coefficients under the given parameters
5.2 Relationship between machining deformation and the three removal strategies used in the case studies can be
and initial residual stresses obtained from Table 2. As indicated in the table, the quantita-
tive relationship between machining deformation and initial
In the previous section, the accuracy of the proposed model is residual stresses can be discussed as follows.
validated via finite element model simulations and machining
experiments. The quantitative relationship between machining 1. The influence coefficients of the layers removed from top
deformations and initial residual stresses can be obtained to bottom are positive, whereas those removed from bot-
based on the analytical model. When parameters such as L, tom to top are negative. The coefficients in the same re-
W, H, E, μ, n, and k are used as constants and a material moval direction exhibit arithmetic progression.
removal strategy is selected, the final deflections of the work- 2. The coefficients in the remaining part, which are all equal
piece can be expressed via the mean values of the initial re- to zero, do not directly influence deformation. That is,
sidual stresses in each layer. The deflection of the vertex is still deformation is only related to the initial residual stresses
used as the object of analysis given its good representation. in the removed part.
Consequently, deflection is represented as follows: 3. Under removal strategy A, the coefficients uniformly de-
  crease from layer 1 to layer 20. The coefficient in layer 1
L W n  
w  ; ¼ ∑ C x;i σx;0;i þ C y;i σy;0;i ; ð38Þ is 7.3 times the coefficient in layer 20. This finding indi-
2 2 i¼1 cates that reducing or eliminating stresses in the upper part

Table 2 Residual stress influence

coefficients of the study case Layer num. Strategy A Strategy B Strategy C

Cx Cy Cx Cy Cx Cy

1 −0.1281 −0.008 −0.0701 −0.0044 −0.0931 −0.0059

2 −0.1223 −0.0076 −0.0642 −0.004 −0.0873 −0.0056
3 −0.1165 −0.0073 −0.0584 −0.0037 −0.0815 −0.0052
4 −0.1106 −0.0069 −0.0525 −0.0033 −0.0756 −0.0048
5 −0.1048 −0.0065 −0.0467 −0.0029 −0.0698 −0.0045
6 −0.099 −0.0062 −0.0409 −0.0026 −0.064 −0.0041
7 −0.0932 −0.0058 −0.035 −0.0022 −0.0582 −0.0037
8 −0.0874 −0.0054 −0.0292 −0.0018 −0.0524 −0.0033
9 −0.0815 −0.0051 −0.0233 −0.0015 −0.0465 −0.003
10 −0.0757 −0.0047 −0.0175 −0.0011 −0.0407 −0.0026
11 −0.0699 −0.0044 0 0 −0.0349 −0.0022
12 −0.0641 −0.004 0 0 −0.0291 −0.0019
13 −0.0583 −0.0036 0 0 −0.0233 −0.0015
14 −0.0524 −0.0033 0 0 −0.0175 −0.0011
15 −0.0466 −0.0029 0 0 0 0
16 −0.0408 −0.0025 0.01753 0.00109 0 0
17 −0.035 −0.0022 0.02337 0.00146 0 0
18 −0.0292 −0.0018 0.02921 0.00182 0 0
19 −0.0233 −0.0015 0.03505 0.00218 0 0
20 −0.0175 −0.0011 0.04089 0.00255 0.01753 0.00109
21 0 0 0.04673 0.00291 0.02335 0.00146
22 0 0 0.05256 0.00328 0.02917 0.00182
23 0 0 0.0584 0.00364 0.03499 0.00218
24 0 0 0.06424 0.004 0.04081 0.00254
25 0 0 0.07008 0.00447 0.04662 0.00307
 Int J Adv Manuf Technol

of the blank is more conducive to controlling deformation 3. Symmetrical machining is an effective method for
when the top-to-bottom machining strategy is used. restraining machining deformation, particularly when
4. Under strategy B, namely, symmetrical machining strate- the initial residual stresses of the blank are symmetric
gy, the coefficients in both directions exhibit good anti- around the middle surface along the thickness direction.
symmetry. Therefore, for the blank with the symmetric The final deflection caused by the symmetrical initial re-
residual stress distribution along the thickness direction, sidual stresses can be theoretically regarded as zero based
the theoretical deflection should be zero under symmetri- on the analytical model. Moreover, semi-symmetrical ma-
cal machining strategy. For the case studies, the final de- chining can also contribute to the reduction of machining
flections calculated using the analytical model and the deformation. In conclusion, the position of the workpiece
finite element model are −0.0012 and −0.0013 mm, in the blank evidently affects the final machining
respectively. deformation.
5. Under strategy C, namely, the semi-symmetrical machin-
ing strategy, the coefficients in layers 9–14 and 20–25 are Acknowledgements This work is supported by the Defense Industrial
Technology Development Program (A0520110009), the State Key
anti-symmetrical, thereby demonstrating that the defor-
Laboratory of Virtual Reality Technology Independent Subject (BUAA-
mation caused by the material removal from the top can VR-16ZZ-07), and the National Science and Technology Major Project
be partly compensated by the material removal from the (2014ZX04001011). The authors thank the referees of this paper for their
bottom. Although the final deflections under strategy C valuable and very helpful comments.
are close to the final deflections under strategy B, their
Compliance with ethical standards
closeness is a coincidence caused by the specific initial
conditions adopted in these cases. The final deflection
Conflict of interest The authors declare that they have no conflict of
under strategy C will not always be close to zero under interest.
different initial conditions.

References

1. Masoudi S, Amini S, Saeidi E, Eslami-Chalander H (2014) Effect

6 Conclusions of machining-induced residual stress on the distortion of thin-
walled parts. Int J Adv Manuf Technol 76:597–608. doi:10.1007/
s00170-014-6281-x
An analytical model based on theory of plates and shells and
2. Liu L, Sun J, Chen W, Zhang J (2016) Finite element analysis of
that considers biaxial initial residual stresses for predicting the machining processes of turbine disk of Inconel 718 high-
machining deformation of a plate blank is proposed in this temperature wrought alloy based on the theorem of minimum po-
study. To validate the accuracy of the analytical model, FEM tential energy. Int J Adv Manuf Technol:1–13. doi:10.1007/s00170-
simulations and machining experiments are conducted. The 016-9026-1
3. Huang X, Sun J, Li J (2015) Finite element simulation and exper-
results calculated using the analytical model are consistent imental investigation on the residual stress-related monolithic com-
with the FEM and experiment results. Finally, the quantitative ponent deformation. Int J Adv Manuf Technol 77:1035–1041. doi:
relationship between machining deformation and initial resid- 10.1007/s00170-014-6533-9
ual stresses is discussed. The conclusions drawn are summa- 4. Liao YG, Hu SJ (2001) An integrated model of a fixture–workpiece
system for surface quality prediction. Mech Eng:810–818. doi:10.
rized as follows. 1007/s001700170108
5. Satyanarayana S, Melkote SN (2004) Finite element modeling of
1. The analytical machining deflection of the plate blank can fixture-workpiece contacts: single contact modeling and experi-
be expressed as the linear sum of the initial residual stress- mental verification. Int J Mach Tools Manuf 44:903–913. doi:10.
1016/j.ijmachtools.2004.02.010
es with the influence coefficients. In addition, the influ-
6. Estrems M, Sánchez HT, Faura F (2003) Influence of fixtures on
ence coefficients of the stresses are determined based on dimensional accuracy in machining processes. Int J Adv Manuf
physical dimension, material properties, and machining Technol 21:384–390. doi:10.1007/s001700300044
strategy. 7. Sánchez HT, Estrems M, Faura F (2006) Analysis and compensa-
2. Under the top-to-bottom machining strategy, the influence tion of positional and deformation errors using integrated fixturing
analysis in flexible machining parts. Int J Adv Manuf Technol 29:
coefficients of the stresses in each layer calculated using 239–252. doi:10.1007/s00170-005-2515-2
the proposed model uniformly decrease from the first to 8. Siebenaler SP, Melkote SN (2006) Prediction of workpiece defor-
the last removed layers. In the case studies, the coeffi- mation in a fixture system using the finite element method. Int J
cients in layer 1 is 7.3 times those in layer 20, thereby Mach Tools Manuf 46:51–58. doi:10.1016/j.ijmachtools.2005.04.
007
indicating that reducing or eliminating stresses in the up- 9. Liu S, Zheng L, Zhang ZH, Wen DH (2006) Optimal fixture design
per part of the plate blank is more conducive to controlling in peripheral milling of thin-walled workpiece. Int J Adv Manuf
deformation. Technol 28:653–658. doi:10.1007/s00170-004-2425-8
Int J Adv Manuf Technol

10. Chen W, Ni L, Xue J (2008) Deformation control through fixture Int J Adv Manuf Technol 42:429–449. doi:10.1007/s00170-008-
layout design and clamping force optimization. Int J Adv Manuf 1610-6
Technol 38:860–867. doi:10.1007/s00170-007-1153-2 28. Puls H, Klocke F, Döbbeler B, Peng B (2016) Multiscale modeling
11. Lu C, Zhao HW (2015) Fixture layout optimization for deformable of thermoelastic workpiece deformation in dry cutting. Procedia
sheet metal workpiece. Int J Adv Manuf Technol 78:85–98. doi:10. CIRP 46:27–30. doi:10.1016/j.procir.2016.03.195
1007/s00170-014-6647-0 29. Fu WE, Cohen PH, Ruud CO (2009) Experimental investigation of
12. Article O (2013) Design and optimization of machining fixture the machining induced residual stress tensor under mechanical
layout using ANN and DOE. 1573–1586. doi:10.1007/s00170- loading. J Manuf Process 11:88–96. doi:10.1016/j.jmapro.2009.
012-4281-2 11.001
13. Sundararaman K, Padmanaban K, Sabareeswaran M (2016) 30. Nasr MNA, Ng EG, Elbestawi MA (2008) A modified time-
Optimization of machining fixture layout using integrated response efficient FE approach for predicting machining-induced residual
surface methodology and evolutionary techniques. Proc Inst Mech stresses. Finite Elem Anal Des 44:149–161. doi:10.1016/j.finel.
Eng Part C J Mech Eng Sci 230:2245–2259. doi:10.1177/ 2007.11.005
0954406215592920 31. Ulutan D, Erdem Alaca B, Lazoglu I (2007) Analytical modelling
14. Ratchev S, Govender E, Nikov S, Phuah K, Tsiklos G (2003) Force of residual stresses in machining. J Mater Process Technol 183:77–
and deflection modelling in milling of low-rigidity complex parts. J 87. doi:10.1016/j.jmatprotec.2006.09.032
Mater Process Technol 143–144:796–801. doi:10.1016/S0924- 32. Huang XM, Sun J, Li JF, Han X, Xiong QC (2013) An experimen-
0136(03)00382-0 tal investigation of residual stresses in high-speed end milling 7050-
15. Ratchev S, Liu S, Huang W, Becker AA (2004) A flexible force T7451 aluminum alloy. Adv Mech Eng. doi:10.1155/2013/592659
model for end milling of low-rigidity parts. J Mater Process Technol 33. Sharman ARC, Hughes JI, Ridgway K (2006) An analysis of the
153–154:134–138. doi:10.1016/j.jmatprotec.2004.04.300 residual stresses generated in Inconel 718™ when turning. J Mater
16. Ratchev S, Nikov S, Moualek I (2004) Material removal simulation Process Technol 173:359–367. doi:10.1016/j.jmatprotec.2005.12.
of peripheral milling of thin wall low-rigidity structures using FEA. 007
Adv Eng Softw 35:481–491. doi:10.1016/j.advengsoft. 2004.06. 34. Outeiro JC, Umbrello D, M’Saoubi R (2006) Experimental and
011 numerical modelling of the residual stresses induced in orthogonal
17. Ratchev S, Liu S, Huang W, Becker AA (2004) Milling error pre- cutting of AISI 316L steel. Int J Mach Tools Manuf 46:1786–1794.
diction and compensation in machining of low-rigidity parts. Int J doi:10.1016/j.ijmachtools.2005.11.013
Mach Tools Manuf 44:1629–1641. doi:10.1016/j.ijmachtools. 35. Jiang XH, Li BZ, Yang JG, Zuo XY, Li K (2013) An approach for
2004.06.001 analyzing and controlling residual stress generation during high-
speed circular milling. Int J Adv Manuf Technol 66:1439–1448.
18. Ratchev S, Huang W, Liu S, Becker AA (2004) Modelling and
doi:10.1007/s00170-012-4421-8
simulation environment for machining of low-rigidity components.
36. Denkena B, Boehnke D, León L (2008) Machining induced resid-
J Mater Process Technol 153–154:67–73. doi:10.1016/j.jmatprotec.
ual stress in structural aluminum parts. Prod Eng 2:247–253. doi:
2004.04.301
10.1007/s11740-008-0097-1
19. Ratchev S, Liu S, Becker AA (2005) Error compensation strategy in
37. Özel T, Zeren E (2007) Finite element modeling the influence of
milling flexible thin-wall parts. J Mater Process Technol 162–163:
edge roundness on the stress and temperature fields induced by
673–681. doi:10.1016/j.jmatprotec.2005.02.192
high-speed machining. Int J Adv Manuf Technol 35:255–267.
20. Ratchev S, Liu S, Huang W, Becker AA (2006) An advanced FEA doi:10.1007/s00170-006-0720-2
based force induced error compensation strategy in milling. Int J 38. Zeng C, Tian W, Liao WH (2016) The effect of residual stress due
Mach Tools Manuf 46:542–551. doi:10.1016/j.ijmachtools.2005. to interference fit on the fatigue behavior of a fastener hole with
06.003 edge cracks. Eng Fail Anal 66:72–87. doi:10.1016/j.engfailanal.
21. Ning H, Zhigang W, Chengyu J, Bing Z (2003) Finite element 2016.04.012
method analysis and control stratagem for machining deformation 39. Choi Y (2009) A study on the effects of machining-induced residual
of thin-walled components. J Mater Process Technol 139:332–336. stress on rolling contact fatigue. Int J Fatigue 31:1517–1523. doi:
doi:10.1016/S0924-0136(03)00550-8 10.1016/j.ijfatigue.2009.05.001
22. Benardos PG, Mosialos S, Vosniakos GC (2006) Prediction of 40. Jiang Z, Liu Y, Li L, Shao W (2014) A novel prediction model for
workpiece elastic deflections under cutting forces in turning. thin plate deflections considering milling residual stresses. Int J Adv
Robot Comput Integr Manuf 22:505–514. doi:10.1016/j.rcim. Manuf Technol 74:37–45. doi:10.1007/s00170-014-5952-y
2005.12.009 41. Izamshah R, Mo JPT, Ding S (2011) Hybrid deflection prediction
23. Aijun T, Zhanqiang L (2008) Deformations of thin-walled plate due on machining thin-wall monolithic aerospace components. Proc
to static end milling force. J Mater Process Technol 206:345–351. Inst Mech Eng Part B J Eng Manuf 226:592–605. doi:10.1177/
doi:10.1016/j.jmatprotec.2007.12.089 0954405411425443
24. López de Lacalle LN, Lamikiz A, Sánchez JA, Salgado MA (2007) 42. Li B, Jiang X, Yang J, Liang SY (2015) Effects of depth of cut on
Toolpath selection based on the minimum deflection cutting forces the redistribution of residual stress and distortion during the milling
in the programming of complex surfaces milling. Int J Mach Tools of thin-walled part. J Mater Process Technol 216:223–233. doi:10.
Manuf 47:388–400. doi:10.1016/j.ijmachtools.2006.03.010 1016/j.jmatprotec.2014.09.016
25. Desai KA, Rao PVM (2008) Effect of direction of parameterization 43. Huang X, Sun J, Li J (2015) Effect of initial residual stress and
on cutting forces and surface error in machining curved geometries. machining-induced residual stress on the deformation of aluminium
Int J Mach Tools Manuf 48:249–259. doi:10.1016/j.ijmachtools. alloy plate. Stroj Vestnik/Journal Mech Eng 61:131–137. doi:10.
2007.08.007 5545/sv-jme.2014.1897
26. Saffar RJ, Razfar MR (2010) Simulation of end milling operation 44. Wei Y, Wang XW (2007) Computer simulation and experimental
for predicting cutting forces to minimize tool deflection by genetic study of machining deflection due to original residual stress of
algorithm. Mach Sci Technol 14:81–101. doi:10.1080/ aerospace thin-walled parts. Int J Adv Manuf Technol 33:260–
10910340903586483 265. doi:10.1007/s00170-006-0470-1
27. Rai JK, Xirouchakis P (2009) FEM-based prediction of workpiece 45. Wang ZJ, Chen WY, Zhang YD, Chen ZT, Liu Q (2005) Study on
transient temperature distribution and deformations during milling. the machining distortion of thin-walled part caused by
 Int J Adv Manuf Technol

redistribution of residual stress. Chinese J Aeronaut 18:175–179. the quasi-symmetric machining method. Metals (Basel) 6:80. doi:
doi:10.1016/S1000-9361(11)60325-7 10.3390/met6040080
46. Dong H, Ke Y (2006) Study on machining deformation of aircraft 55. Tang ZT, Yu T, Xu LQ, Liu ZQ (2013) Machining deformation
monolithic component by FEM and experiment. Chinese J prediction for frame components considering multifactor coupling
Aeronaut 19:247–254. doi:10.1016/S1000-9361(11)60352-X effects. Int J Adv Manuf Technol 68:187–196. doi:10.1007/s00170-
47. Liu L, Sun J, Chen W, Sun P (2015) Study on the machining 012-4718-7
distortion of aluminum alloy parts induced by forging residual 56. Rai JK, Xirouchakis P (2008) Finite element method based machin-
stresses. Proc Inst Mech Eng Part B J Eng Manuf. doi:10.1177/ ing simulation environment for analyzing part errors induced during
0954405415583805 milling of thin-walled components. Int J Mach Tools Manuf 48:
48. Guo H, Zuo DW, Wu HB, Xu F, Tong GQ (2009) Prediction on 629–643. doi:10.1016/j.ijmachtools.2007.11.004
milling distortion for aero-multi-frame parts. Mater Sci Eng A 499: 57. Franchim AS, de Campos VS, Travessa DN, de Moura Neto C
230–233. doi:10.1016/j.msea.2007.11.137 (2009) Analytical modelling for residual stresses produced by shot
49. Cerutti X, Mocellin K (2015) Parallel finite element tool to predict peening. Mater Des 30:1556–1560. doi:10.1016/j.matdes.2008.07.
distortion induced by initial residual stresses during machining of 040
aeronautical parts. Int J Mater Form 8:255–268. doi:10.1007/ 58. Reihanian M, Naseri M (2016) An analytical approach for necking
s12289-014-1164-0 and fracture of hard layer during accumulative roll bonding (ARB)
50. Richter-Trummer V, Koch D, Witte A, dos Santos JF, de Castro of metallic multilayer. Mater Des 89:1213–1222. doi:10.1016/j.
PMST (2013) Methodology for prediction of distortion of work- matdes.2015.10.088
pieces manufactured by high speed machining based on an accurate
59. Shin SHS (1995) Prediction of the dimensional instability resulting
through-the-thickness residual stress determination. Int J Adv
from machining of residually stressed components. Ph.D. disserta-
Manuf Technol 68:2271–2281. doi:10.1007/s00170-013-4828-x
tion, Texas Tech University: 13–14.
51. Husson R, Baudouin C, Bigot R, Sura E (2014) Consideration of
residual stress and geometry during heat treatment to decrease shaft 60. Timoshenko S, Woinosky-Krieger S (1959) Theory of plates and
bending. Int J Adv Manuf Technol 72:1455–1463. doi:10.1007/ shells. Classic, Second edn. McGraw-Hill, New York
s00170-014-5688-8 61. Rossini NS, Dassisti M, Benyounis KY, Olabi AG (2012) Methods
52. Yang Y, Li M, Li KR (2014) Comparison and analysis of main of measuring residual stresses in components. Mater Des 35:572–
effect elements of machining distortion for aluminum alloy and 588. doi:10.1016/j.matdes.2011.08.022
titanium alloy aircraft monolithic component. Int J Adv Manuf 62. Jiang X, Li B, Yang J, Zuo XY (2013) Effects of tool diameters on
Technol 70:1803–1811. doi:10.1007/s00170-013-5431-x the residual stress and distortion induced by milling of thin-walled
53. Zhang Z, Li L, Yang YF, He N, Zhao W (2014) Machining distor- part. Int J Adv Manuf Technol 68:175–186. doi:10.1007/s00170-
tion minimization for the manufacturing of aeronautical structure. 012-4717-8
Int J Adv Manuf Technol 73:1765–1773. doi:10.1007/s00170-014- 63. Yang D, Liu Z, Ren X, Zhuang P (2016) Hybrid modeling with
5994-1 finite element and statistical methods for residual stress prediction
54. Wu Q, Li D-P, Zhang Y-D (2016) Detecting milling deformation in in peripheral milling of titanium alloy Ti-6Al-4V. Int J Mech Sci
7075 aluminum alloy aeronautical monolithic components using 108–109:29–38. doi:10.1016/j.ijmecsci.2016.01.027