Professional Documents
Culture Documents
MAHMUT Stress Biaxiality-Based Residual Stress Assessment in Welded T-Joints Using The Blind-Hole Method
MAHMUT Stress Biaxiality-Based Residual Stress Assessment in Welded T-Joints Using The Blind-Hole Method
MAHMUT Stress Biaxiality-Based Residual Stress Assessment in Welded T-Joints Using The Blind-Hole Method
A R T I C L E I N F O A B S T R A C T
Keywords: The blind-hole method is widely used to measure the welding residual stress (WRS) in welded steel structural
Steel truss bridge connections such as T-joints. To enhance the test accuracy, a rapid WRS evaluation method considering stress
T-joint level and biaxial stress ratio was developed and validated. The proposed method was applied to measure the WRS
Welding residual stress
distributions of T-joints in steel truss bridges, and a finite element model (FEM) simulating the WRS was verified
Blind-hole method
Stress biaxiality
by the test results. On this basis, the effects of welding parameters and geometric configuration of T-joints on
WRS were analyzed by the validated FEM. The results show that the maximum error from ignoring the hole edge
plasticity or stress biaxiality is about 20%, and the proposed method can effectively reduce the WRS measure
ment error caused by the above factors. The results of the experiment and FE analysis demonstrate that the WRSs
in T-joint are in an obvious biaxial stress state, which further proves that it is necessary to consider stress
biaxiality in calibration for the blind-hole method.
* Corresponding author.
E-mail address: bhji@hhu.edu.cn (B. Ji).
https://doi.org/10.1016/j.jcsr.2024.108610
Received 5 December 2023; Received in revised form 22 February 2024; Accepted 9 March 2024
Available online 16 March 2024
0143-974X/© 2024 Published by Elsevier Ltd.
C. Li et al. Journal of Constructional Steel Research 216 (2024) 108610
2
C. Li et al. Journal of Constructional Steel Research 216 (2024) 108610
⎧ ε1 + ε3 Table 1
⎪
⎪ A=
⎪
⎨ 2(σ1 + σ 2 ) Parameters of SRC formula.
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ (2) Designation d (mm) b (mm) a (mm) r1 (mm) r2 (mm)
⎪
⎪ (ε1 − ε3 )2 + (2ε2 − ε1 − ε3 )2
⎪
⎩B = Parameter value 0.75 0.7275 1.5 1.75 3.25
2(σ1 − σ 2 )
In accordance with GB/T 31310–2014 [21], a uniaxial stress field is
imposed on the specimen. Strain gauge 1 is oriented parallel to the di ⎧ ∫ r2 ∫ b
cos2β [ x ]x=r2
⎪
rection of the loading stress, and the maximum principal stress σ1 equals ⎪
⎪
⎪
⎪
S1 =
r 2
dxdy = 2 arctan
b x=r1
⎪
the stress induced by the loading stress. Under these conditions, with σ 1
r1 − b
⎪
⎪
⎪
⎪ ∫ r2 ∫ b [ ]x=r2
= σ, σ 2 = 0, θ = 0, the SRCs calculation formula can be simplified to Eq. ⎨ cos4β x
S2 = dxdy = − 2b (5)
(3): ⎪
⎪ r1 − b r2 x2 + b2 x=r1
⎪
⎪ [ ]x=r2
⎧ ⎪
⎪ ∫ ∫
ε1 + ε3 ⎪
⎪
r2 b
cos4β 2 x
⎨A = ⎪
⎪ S = dxdy = − b ( )
2σ ⎩ 3 2
(3) r1 − b r4 3 x2 + b2 x=r1
⎩ ε1 − ε3
B=
2σ
where E andμ represent Young's modulus and Poisson's ratio of the
material; d represents the radius of the hole; a and b represent the length
2.2. Finite element model
and the half-width of the strain gauge; r1 and r2 represent the distances
from the hole center to both ends of the strain gauge; r = (r1 + r2)/2,
The biaxial stress field can be easily simulated in the FEM, and the
which represents the distance from the center of the strain gauge to the
biaxial stress ratio can also be arbitrarily specified, overcoming the
hole center; κ represents the correction coefficient, which is equal to
limitations of calibration tests. Additionally, the principle, boundary
1.065; S1, S2, and S3 represent the area integral parameters of the
conditions, and loads for calibrating SRCs using the FEM are identical to
released strain based on the rosette gauge. The values and corresponding
those for experimental calibration. Operational errors are also effec
meanings of each parameter in the formula are presented in Table 1 and
tively avoided. Consequently, a FEM established in ABAQUS software
Fig. 1, respectively.
was employed to calibrate the SRCs in the present research. To enhance
Fig. 4 shows that the absolute values of the experimentally calibrated
calculation efficiency, the FEM length was simplified to 100 mm, while
SRCs are smaller than those calibrated by the FEM in this paper, but
maintaining consistent width and thickness with the calibration spec
their growth trends are similar, which preliminarily proves the ratio
imen. The rationality of the simplified model has been proved by
nality of the established FEM. Furthermore, to mitigate the influence of
Ref. [15].
experimental errors on the SRCs calibration results, we introduce
The C3D8R eight-node linear brick element was employed with
Kirsch's theoretical solution and the FE calibration results from
varying mesh densities. Element sizes near the hole were set to 0.15 mm,
Ref. [15].
while sizes in other areas reached as large as 5 mm. The FEM comprised
When the calibration stress f1 is below 0.5fy, the average errors for A
440,612 nodes and 360,762 elements, respectively. The biaxial stress
and B are 1.46% and 3.0%, respectively. This suggests that the SRCs
field was loaded to the four sides of the FEM. Additionally, k (i.e., f2/f1)
obtained by uniaxial stress calibration align well with Kirsch's theoret
was defined as the stress ratio. Both f1 and f2 were the applied external
ical solution. However, when the calibration stress f1 exceeds 0.5fy and
stresses, with f1 exceeding f2 [15]. A fixed constraint (i.e., UX = UY=UZ
continues to increase, due to the influence of plastic deformation at the
= 0) was applied at the midpoint of the model to prevent rigid body
hole edge, the absolute values of SRCs for uniaxial calibration gradually
displacement. The geometric size and arrangement of the rosette gauge
increase and move away from the theoretical solution. Therefore, when
(TJ120–1.5-Φ1.5) in the model mirrored those in the calibration test
the calibration stress f1 exceeds 0.5fy, the SRC calibration results from
(Fig. 3). Q345qD steel with an elastic modulus of 2.06 × 105 MPa and a
Ref. [15] are introduced for comparison with the FE calibration results
Poisson's ratio of 0.3 was used in the FEM, and the stress-strain rela
in this study. It's evident that the SRCs from the FE calibration align with
tionship was detailed in Ref. [15]. A hole with the diameter of 1.5 mm
the trend observed in Ref. [15], with average errors for A and B below
and depth of 1.8 mm was drilled at the center of the strain rosette. After
3%. In conclusion, the established FEM is accurate and suitable for SRC
drilling, the average value of each node was taken as the released strain.
calibration under biaxial stress.
The released strain and stress field parameters were then substituted
into Eq. (2) to obtain the SRCs.
2.4. The SRCs by biaxial stress calibration
2.3. Validation of the model
To study the effect of biaxial stress calibration on the SRCs, cali
bration stresses f1 and f2 were applied to the above FEM. The stress ratio
To validate the accuracy of the FEM, a uniaxial stress field was
k ranged from − 0.8 to 0.8. The biaxial stress magnitude (i.e., stress level)
applied. According to GB/T 31310–2014 [24], the calibration stress f1
was quantified by the von Mises equivalent stress fEq. [23], as shown in
ranged from 0.1fy to 0.9fy, with f2 set to 0. The SRCs calibrated by the FE
Eq. (6). Fig. 5 reveals that when the stress level is below 0.5fy, the SRCs
method were then compared with Kirsch's theoretical solution. For the
remain relatively constant, suggesting little impact from the stress ratio
case of through-hole, when the material was in the linear elastic range,
and stress level. However, when the stress feq exceeds 0.5fy, attributed to
the formulae for calculating SRCs based on Kirsch's theory were [22]:
the plastic deformation of the hole edge, the absolute values of SRCs
⎧
⎪ 1 + μ d2 increase with the stress level. At the same stress level, the absolute
⎪ A = − κ⋅ ⋅
⎪
⎨ 2E r1 r2 values of SRCs increase with an elevated stress ratio, which can be
[ ( ) ] (4) attributed to the plastic strain transferring to ε1 and the relative decrease
⎪
⎪B = − 1
⎪
⎩
3
(1 + μ) d2 S2 − d4 S3 + (1 − μ)d2 S1 of ε1 [15]. Furthermore, the errors between the SRCs obtained through
2abE 2
biaxial stress calibration and uniaxial stress calibration rise with an in
crease in the stress level. For instance, when feq is 0.9fy, the absolute
values of A and B at a stress ratio of 0.8 increase by 7.9% and 51.3%,
respectively, compared to a stress ratio approaching 0.
3
C. Li et al. Journal of Constructional Steel Research 216 (2024) 108610
(a) A (b) B
Fig. 4. Comparative analysis of SRCs calibrated by different methods.
k k
S
k k
k k
k k k=
k
f f k=
f f
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
of the absolute values of SRCs [6]. Therefore, if the SRC theoretical
feq = f1 2 + f2 2 − f1 f2 (6)
solution is still used to calculate the WRS, they will be overestimated. As
depicted in Fig. 6, under the studied working conditions, the difference
2.5. Effect of SRCs calibration methods on residual stress calculation between the estimated stress calculated by the Kirsch theoretical solu
tion (i.e., theoretical value) and the calibration stress reaches up to 20%.
To assess the influence of the calibration method on the WRS, Additionally, the estimated stresses obtained through biaxial stress
additional working conditions were considered for 0.2 ≤ k ≤ 0.8 (in calibration align well with the calibration stresses, but the maximum
increments of 0.2), which were not accounted for in the SRC calibration. error between the estimated stresses obtained through uniaxial stress
Each working condition contained nine sets of data, ranging from 0.1 to calibration and the calibration stresses is close to 11%.
0.9 (in increments of 0.1). Initially, the release strains under the cali In summary, at low-stress levels, the estimated stresses calculated by
brated stress (i.e., the applied external stresses f1 and f2) were calculated. different SRC calibration methods align with the calibration stresses.
Subsequently, employing the stress level and stress ratio of the selected However, at high-stress levels, the estimated stresses calculated by the
working conditions, the SRCs for uniaxial stress and biaxial stress cali Kirsch theoretical solution (neglecting the plastic effect) significantly
bration were derived through two-dimensional linear interpolation exceed the calibration stress, with a difference of up to 20%. In addition,
among the calibrated SRCs (see Fig. 5). Theoretical SRCs were calculated uniaxial calibration also results in a maximum error of nearly 11%.
using Eqs. (4) and (5). Finally, these released strains and SRCs were Therefore, it's crucial to account for biaxial stress and the plastic effect at
inserted into Eq. (1) to compute the estimated stress under different SRC the hole edge when adjusting the SRCs.
calibration methods, as shown in Fig. 6. At low-stress levels, estimated
stresses from different calibration methods align well with the calibra 3. Plastic correction of SRCs considering stress biaxiality
tion stress. As stress level increases, the difference between estimated
stresses and calibration stresses gradually expands. One potential reason 3.1. Proposed evaluation framework for determining the WRS
for this difference is that the magnitude of the biaxial stress affects the
distribution of plastic strain in each direction around the hole edge. During the test, the released strains ε1, ε2, and ε3 can be determined
As stress level increases, the absolute values of the SRC theoretical from the rosette gauge. Nevertheless, the WRS is unknown, making the
solution are gradually smaller than those from FE calibration (see stress level and biaxial stress unavailable. Obtaining the SRC for the
Fig. 4). However, the residual tensile stress decreases with the increase actual WRS field is challenging. On the one hand, the distortion energy
4
C. Li et al. Journal of Constructional Steel Research 216 (2024) 108610
theory in material mechanics helps determine whether a material is model between SRCs, S, and k. (3) Calculate the WRS based on the
yielding [6,12,23]. Specifically, when the distortion energy factor S released strains, SRCs, and Eq. (1).
reaches a critical value Scv, the material begins to yield. As a result, the
hole-edge stress level can be represented by the distortion energy factor
S, which can be calculated directly using the strain release strains ε1, ε2, 3.2. Derivation of the distortion energy factor S
and ε3. On the other hand, according to the generalized formula of
Hooke's law (Eq. (7)), a functional relationship between εx/εy and σ x/σy The von Mises yield condition in plastic theory can be expressed as
is established. This implies that the stress ratio (k) can be predicted by [12]:
the released strains ε1, ε2, and ε3. [ ]/
( )/ (σ1 − σ 2 )2 + (σ 2 − σ3 )2 + (σ 3 − σ1 )2 6 = C (8)
{
εx = ( σ x -μσ y )/E
(7)
εy = σ y -μσ x E where C represents the distortion energy. The material will yield only
when C reaches the yield limit. If the directions of σ x and σ y align with
Based on the analysis above, a rapid evaluation framework for WRS the principal stresses of the WRS field, the von Mises yield condition for
considering the stress biaxiality effect was developed (Fig. 7). The spe the plane stress problem (i.e., σ1 = σ x, σ2 = σ y, and σ 3 = 0) will be
cific steps were as follows: (1) Calculate the distortion energy factor S expressed as:
based on the released strains, and compare it with the critical value Scv.
(2) if S ≤ Scv, indicating no yielding at the hole edge, calculate the SRCs σx 2 − σx σy + σy 2 = σs 2 (9)
using Kirsch's theory (Eqs. (4) and (5)). If S > Scv, signifying yielding at
the hole edge, predict the stress ratio using the function k = k(εx/εy), and where σ s is the yield stress of the material. The functional relationship
calculate the corrected SRC based on the established mathematical between stress and strain is developed based on generalized Hooke's
law. Therefore, Eq. (10) is deduced by Eqs. (7) and (9).
5
C. Li et al. Journal of Constructional Steel Research 216 (2024) 108610
feq/fy is >0.5, the two curves diverge. Hence, the distortion energy factor
S corresponding to feq/fy = 0.5 is defined as the critical value.
Fig. 7. The proposed WRS evaluation framework.
3.3. The SRCs correction
6
C. Li et al. Journal of Constructional Steel Research 216 (2024) 108610
4.1. Specimen design and measurement point distribution The WRSs of the T-joints were measured by the blind-hole method,
with the test device and procedure depicted in Fig. 14. Initially, the
To further validate the proposed WRS evaluation method and measured area was polished and cleaned. The rosette gauge was
determine the WRS distribution of T-joints considering stress biaxiality, mounted with adhesive on the measuring point in the specified direc
two T-joint specimens were designed following the ‘Standard Drawing of tion. Subsequently, a hole with a diameter of 1.5 mm and a depth of 1.8
Steel Truss Composite Girder Bridge’ [24] published by the Ministry of mm was drilled at the center of the rosette gauge. To prevent eccentric
Transport of China in 2019. One of these specimens was utilized to drilling, the drill was calibrated and fixed in advance. Strains were
supplement the test data. T-joint specimens were welded using Q345qD recorded before and after drilling, and the WRS was evaluated by the
steel plates, and their dimensions were shown in Fig. 13. Following the method proposed in Section 3.1.
completion of the first weld, the second welding operation took place
after the specimen temperature returned to its initial level. The welding
4.3. Measurement results
parameters were set at 280 A for current, 30 V for voltage, and a welding
speed of approximately 8 mm/s.
4.3.1. Effect of different evaluation methods on WRS
The WRSs of the T-joint specimens were measured along three
To validate the proposed WRS evaluation method (see Fig. 7) for its
different paths, located on a specific section surface. To avoid the in
applicability in actual welded structures, the WRS of the T-joint spec
fluences of arc starting and extinguishing at both ends of the weld on the
imen was calculated using various SRC calibration methods based on
WRS, measurement sections were chosen away from the specimen ends.
experimental test data, and the results were also compared. Given the
The coordinates of the measurement sections were x = 150 mm (path 1),
significant impact of different SRC calibration methods on LWRS (see
x = 300 mm (path 2), and x = 450 mm (path 3), as shown in Fig. 13. The
Fig. 6), LWRS is solely analyzed here. Furthermore, due to the similarity
measurement paths were distributed on the top and bottom surfaces of
of WRS distributions on different paths, the LWRS distributions under
the flange, as well as the surfaces on both sides of the web. The distance
different calibration methods were analyzed, specifically on the top
between measuring points near the weld was about 5 mm, while the
surface of the flange and the web on Path2 and Path3 (see Fig.13), as
distance between measuring points away from the weld gradually
depicted in Figs. 15 and 16.
increased, as shown in Fig. 13. Each path consisted of 43 measurement
The LWRS distributions on Path2 and Path3 calculated by different
points, including 15 on the top surface of the flange, 14 on the bottom
SRCs calibration methods appear similar, with SRCs calibration methods
surface of the flange, and 14 on both sides of the web. The drilling po
exhibiting a slight influence on LWRS at lower stress levels. As stress
sition and geometric dimensions of the rosette gauge at the measuring
levels increase, the high-value LWRSs calculated with SRCs theoretical
point are shown in Fig. 3.
solution surpass those calculated with uniaxial FEM calibration,
whereas those obtained by uniaxial FEM calibration exceeded those
7
C. Li et al. Journal of Constructional Steel Research 216 (2024) 108610
Fig. 12. Comparison between the estimated results and calibration values.
derived from the proposed evaluation method. This aligns with the deviation in LWRS peak values is that Kirsch's theoretical solution ne
findings of Section 2.5 and Ref. [15], as elaborated in Section 2.5, glects plastic effects and stress biaxiality, while uniaxial FEM calibration
indicating that the test data was reasonable in this paper. overlooks stress biaxiality. Consequently, when plastic deformation
For the top surface of the flange (see Fig. 15), the LWRS calculated occurs at the hole edge, the WRS calculation results ignoring plastic
using Kirsch's theoretical solution and uniaxial FEM calibration are at effects and stress biaxiality are overestimated [15]. This indicates that
most 33 MPa and 13 MPa greater than those calculated using the pro the proposed WRS evaluation method can reduce the high-level WRS
posed evaluation method. Likewise, for the web surface (see Fig. 16), the induced by plastic effects and stress biaxiality. This conclusion, along
LWRS calculated using Kirsch's theoretical solution and uniaxial FEM with the research findings in Section 3.4, collectively demonstrates the
calibration are at most 31 MPa and 12 MPa higher than those obtained validity of the proposed approach.
from the proposed evaluation method. The primary reason for the
8
C. Li et al. Journal of Constructional Steel Research 216 (2024) 108610
Fig. 13. Measurement paths and measurement points on each path (unit: mm).
Fig. 15. LWRS on the flange top surface under different evaluation methods.
9
C. Li et al. Journal of Constructional Steel Research 216 (2024) 108610
Fig. 16. LWRS on the web surface under different evaluation methods.
4.3.2. WRS distribution in T-joints using the proposed method centerline of the flange, the LWRS on the top surface rapidly decreases,
The WRS distribution on the top surface of the flange is depicted in becoming compressive at approximately 40 mm. The TWRS distribution
Fig. 17(a) and (b). The WRS along the welding direction is longitudinal generally mirrors the LWRS distribution, averaging nearly 150 MPa.
WRS (LWRS), while the WRS perpendicular to the welding direction is However, the peak stress position for TWRS is not aligned with the
transverse WRS (TWRS). The peak LWRS appears near the centerline of centerline of the flange, possibly due to inherent uncertainties in the
the flange, averaging about 300 MPa. Continuing away from the flange blind-hole method [25,26]. In summary, tensile WRS is concentrated
10
C. Li et al. Journal of Constructional Steel Research 216 (2024) 108610
11
C. Li et al. Journal of Constructional Steel Research 216 (2024) 108610
between the specimen and the ambient, which includes natural con using the double ellipsoid heat source model depicted in Fig. 21(a).
vection and radiant heat. The environmental temperature was set to However, a coordinate transformation was required for the double
20 ◦ C. To enhance calculation efficiency, the radiative heat was equated ellipsoid heat source model for T-joints, as illustrated in Fig. 21(b). The
to convective heat based on several trial calculations, with the convec transformation coordinates were represented by Eqs. (12) and (13). This
tion coefficient of 15 W/(m2•◦ C). The latent heat of solid-liquid trans heat source model was loaded by the DFLUX subroutine and represented
formation was 260 kJ/kg. The welding heating process was simulated by Eqs. (14) and (15).
12
C. Li et al. Journal of Constructional Steel Research 216 (2024) 108610
Fig. 23. Effect of the welding speed on WRS near the weld.
y1 = (y − y0 )cosθ + (z − z0 )sinθ (12) model to continue stress calculations. The stress calculation utilized the
same FEM as thermal calculations, with the only difference being the
z1 = − (y − y0 )sinθ + (z − z0 )cosθ (13) change in element type from DC3D8 to C3D8R. The mechanical
√̅̅̅( ) ( boundary conditions of the model are illustrated in Fig. 19. To avoid
)
6 3 ff Q 3x2 3y2 3z2 rigid body motion, fixed constraints were applied on one side of the
qf (x, y, z) = √ ̅ exp −
̅̅ − 2 − 2 ,x ≥ 0 (14)
af bcπ π af 2 b c flange, namely UX = UY = UZ = 0, while releasing the constraint in the
y-direction on the opposite side of the flange, that is, UX = UZ = 0.
√̅̅̅ ( )
6 3(fr Q) 3x2 3y2 3z2 Constraint in the z-direction was released at the top of the web, that is,
qr (x, y, z) = √̅̅̅ exp − − 2 − 2 ,x < 0 (15)
ar bcπ π ar 2 b c UX = UY = 0. Phase transformation effect was ignored because of its
little impact on WRS for low-carbon steel [29].
where Q = ηUI represents the magnitude of the heat input, in which η, U,
and I are the efficiency, welding voltage, and current respectively (Q =
10,800 J for this paper); af, ar, b, and c are the geometric parameters of 5.2. Comparison between FEA and experimental measurement
the heat source; ff and fr denote the fractions of the heat deposited in the
front and rear quadrants, which satisfy ff + fr = 2 (ff = 0.7, fr = 1.3 for To verify the rationality of the FEM, simulated welding deformation
this paper). and WRS were compared with the measured values. Fig. 22(a) shows the
flange deformation characteristics and displacement values from nu
5.1.3. Mechanical analysis merical simulation align with the measured results of T-joint specimens
After the temperature field was obtained through thermal calcula in Section 4.1 and Ref. [30]. Fig. 22(b) demonstrates overall agreement
tion, the temperature history was applied as a predefined field to the between the measured and simulated WRS. Deviations at individual
points in the WRS may arise from measurement errors, which can be
13
C. Li et al. Journal of Constructional Steel Research 216 (2024) 108610
Fig. 24. Effect of the flange thickness on the WRS near the weld.
deemed normal. Thus, the established FEM is well-suited for conducting 5.3.2. Component plate thickness
a parametric analysis of WRS. To study the WRS fields near the weld of T-joints at varying plate
thickness, the WRSs were compared at speeds of 18 mm, 20 mm, 22 mm,
and 24 mm in Fig. 24.
5.3. Parametric analysis As the thickness of the flange plate increases from 18 mm to 24 mm,
the peak LWRS near the flange weld toe rises from 305 MPa to 321 MPa,
5.3.1. Welding speed and the peak TWRS increases from 260 MPa to 290 MPa, representing
To study the WRS fields near the weld of T-joints at varying welding increments of 5.2% and 11.5%, respectively. This rise is primarily
speeds, the WRSs were compared at speeds of 8 mm/s, 10 mm/s, 12 attributed to the increased internal constraint resulting from the thicker
mm/s, and 14 mm/s in Fig. 23. F16W14S8 indicates a flange thickness of component plate, leading to higher WRS [32]. However, when the
16 mm, web thickness of 14 mm, and welding speed of 8 mm/s. flange plate thickness increases from 18 mm to 24 mm, the average
As the welding speed increases from 8 mm/s to 14 mm/s, the peak LWRS decreases from 246 MPa to 237 MPa, and the average TWRS
LWRS near the flange weld toe decreases from 270.4 MPa to 216.2 MPa, decreases from 110 MPa to 77 MPa, resulting in decreases of 3.7% and
and the peak LWRS near the web weld toe decreases from 322.4 MPa to 3.0%, respectively. This indicates that the component plate thickness
205.9 MPa, representing a reduction of about 20.3% and 36.2%, has contrasting effects on the WRS near the weld toes of the flange and
respectively. This decrease is primarily attributed to the reduced heat web, aligning with findings in Ref. [33]. Additionally, the WRS fields
source loading time and heat input per unit length, leading to a decrease remain in a biaxial stress state at varying plate thickness, with the TWRS
in the maximum temperature in the welding area [31]. Moreover, the significantly smaller than the LWRS, and the maximum stress ratio k
TWRSs near the weld toes of the flange and web undergo minimal being around 0.50.
changes at different welding speeds. The average values are about 150
MPa and 130 MPa, respectively. Additionally, the TWRS is notably 6. Conclusions
smaller than the LWRS, and the stress ratio k falls between 0.50 and
0.70, indicating a biaxial stress state near the T-joint weld. To enhance the accuracy of the blind-hole method, a rapid WRS
14
C. Li et al. Journal of Constructional Steel Research 216 (2024) 108610
evaluation method considering the stress biaxiality effect was developed hole method, Trans. Tianjin Univ. 24 (2018) 480–488, https://doi.org/10.1007/
s12209-017-0091-9.
and validated, which utilized the distortion energy to assess the stress
[7] J. Gibmeier, M. Kornmeier, B. Scholtes, Plastic deformation during application of
level at the hole edge and subsequently employed either the SRC theo the hole-drilling method, Mater. Sci. Forum 347-349 (2000) 131–137, https://doi.
retical solution or SRC modified value to determine WRS. Additionally, org/10.4028/WWW.SCIENTIFIC.NET/MSF.347-349.131.
the WRS distribution of T-joints was measured using the proposed [8] J.P. Nobre, M. Kornmeier, B. Scholtes, Plasticity effects in the hole-drilling residual
stress measurement in peened surfaces, Exp. Mech. 58 (2017) 369–380, https://
method, and the stress state of the welds was verified through experi doi.org/10.1007/s11340-017-0352-5.
mental and FEA results. The main conclusions are as follows: [9] B. Qiang, Y.D. Li, Y. Gu, D.D. Boko-haya, Uncertainty of strain release coefficients
for the blind-hole procedure evaluated by Monte Carlo simulation, Mater. Test. 59
(2017) 630–634, https://doi.org/10.3139/120.111052.
(1) Neglecting the plastic effect results in an overestimation of the [10] A.M. Nawwar, K. Mclachlan, J. Shewchuk, A modified hole-drilling technique for
WRS, with a maximum error of 20%. The uniaxial calibration determining residual stresses in thin plates, Exp. Mech. 16 (1976) 226–232,
ignoring stress biaxiality leads to a maximum error of WRS https://doi.org/10.1007/BF02329273.
[11] D. Vangi, S. Tellini, Hole-drilling strain-gauge method: residual stress
calculation results approaching 11%; measurement with plasticity effects, J. Eng. Mater. Technol. 132 (2010) 011003,
(2) We proposed a rapid WRS evaluation method accounting for https://doi.org/10.1115/1.3184030.
stress biaxiality, which could directly predict the WRS by [12] H.Y. Zhao, Y.W. Shi, Y. Pei, Y.P. Lei, Y.J. Chen, On the correction of plasticity effect
at the hole edge when using the Centre hole method for measuring high welding
measured strains. The correlation coefficients between the esti residual stress, Strain 32 (1996) 25–129, https://doi.org/10.1111/J.1475-
mated stresses and the calibrated values are close to 1.0, showing 1305.1996.TB01017.X.
good agreement between the results; [13] Y.B. Wang, G.Q. Li, S.W. Chen, The assessment of residual stresses in welded high
strength steel box sections, J. Constr. Steel Res. 76 (2012) 93–99, https://doi.org/
(3) The TWRS distribution closely mirrored that of the LWRS in T-
10.1016/j.jcsr.2012.03.025.
joints, with tensile WRS concentrated near the weld. The peak [14] A. Giri, C. Pandey, M. Mahapatra, K. Sharma, P.K. Singh, On the estimation of error
TWRS was notably smaller than that of the LWRS, and the WRSs in measuring the residual stress by strain gauge rosette, Measurement 65 (2015)
in the T-joint were in an obvious biaxial stress state, with stress 41–49, https://doi.org/10.1016/J.MEASUREMENT.2014.12.047.
[15] Y.D. Li, J. Wu, B. Qiang, S.T. Zhou, W.Q. Liu, C.R. Yao, Measurements of residual
ratios at some points near the weld reaching approximately 0.7. stresses in a welded orthotropic steel deck by the hole-drilling method considering
(4) The impact of flange thickness on WRS is less significant stress biaxiality, Eng. Struct. 230 (2021) 111690, https://doi.org/10.1016/j.
compared to welding speed. Flange plate thickness has opposite engstruct.2020.111690.
[16] H. Fang, Mechanics of Welding Structure, China Machine Press, Beijing, 2008 (in
effects on the WRS near the weld of the flange and web. Chinese).
[17] A. Hannon, P. Tiernan, A review of planar biaxial tensile test systems for sheet
CRediT authorship contribution statement metal, J. Mater. Process. Technol. 198 (2008) 1–13, https://doi.org/10.1016/j.
jmatprotec.2007.10.015.
[18] J.P. Nobre, A.M. Dias, J. Gibmeier, M. Kornmeier, Local stress-ratio criterion for
Chuanxi Li: Conceptualization, Methodology, Software, Investiga incremental hole-drilling measurements of shot-peening stresses, J. Eng. Mater.
tion, Writing – original draft, Formal analysis. Bohai Ji: Writing – re Technol. Trans. ASME 128 (2006) 193–201, https://doi.org/10.1115/1.2172623.
[19] Y. Ohtake, S. Rokugawa, H. Masumoto, Geometry determination of cruciform-type
view & editing, Project administration, Funding acquisition. Yue Yao: specimen and biaxial tensile test of C/C composites, Key Eng. Mater. 164–165
Writing – review & editing. (1999) 151–154, https://doi.org/10.4028/WWW.SCIENTIFIC.NET/KEM.164-
165.151.
[20] S.P. Timoshenko, J.N. Goodier, Theory of Elasticity, McGraw-Hill, New York, 1970.
Declaration of competing interest [21] GB/T 31310–-2014, Metallic Material-Determination of Residual Stress-Hole
Drilling Strain-Gauge Method, Standards Press of China, Beijing, 2014 (in Chinese).
[22] H.N. Chen, A discussion of calculation of coefficients a and B by blind-hole-drilling
The authors declare that they have no known competing financial method, Mechan. Strength 11 (2) (1989) 31–36 [in Chinese], 10.16579/j.issn.1001
interests or personal relationships that could have appeared to influence .9669.1989.02.009.
[23] D. Vangi, M. Ermini, Plasticity effects in residual stress measurement by the hole
the work reported in this paper.
drilling method, Strain 36 (2000) 55–59, https://doi.org/10.1111/j.1475-
1305.2000.tb01174.x.
Data availability [24] Zhejiang Institute of Communications Company Limited, Standard Drawing of
Steel Truss Composite Girder Bridge, Zhejiang Institute of Communications Co.,
Ltd., Hangzhou, 2018 (in Chinese).
Data will be made available on request. [25] D. Peral, J.D. Vicente, J.A. Porro, J.L. Ocaña, Uncertainty analysis for non-uniform
residual stresses determined by the hole drilling strain gauge method,
Measurement 97 (2017) 51–63, https://doi.org/10.1016/j.
Acknowledgment
measurement.2016.11.010.
[26] G. Schajer, E. Altus, Stress calculation error analysis for incremental hole-drilling
This work is sponsored by the National Natural Science Foundation residual stress measurements, J. Eng. Mater. Technol. Trans. ASME 118 (1) (1996)
of China (No. 52378153). The assistances are gratefully acknowledged. 120–126, https://doi.org/10.1115/1.2805924.
[27] H.Y. Ban, G. Shi, Y. Bai, Y.J. Shi, Y.Q. Wang, Residual stress of 460 MPa high
strength steel welded I section: experimental investigation and modeling, Int. J.
References Steel Struct. 13 (4) (2013) 691–705, https://doi.org/10.1007/s13296-013-4010-1.
[28] X.L. Cao, S.Q. Shen, M. Wu, Z.Q. Tao, et al., Residual stresses of 550 MPa high
strength steel welded cruciform sections: experimental and numerical study,
[1] C. Cui, Q.H. Zhang, Y. Bao, Y.Z. Bu, Y. Luo, Fatigue life evaluation of welded joints
Structures 44 (2022) 579–593, https://doi.org/10.1016/j.istruc.2022.08.024.
in steel bridge considering residual stress, J. Constr. Steel Res. 153 (2018)
[29] D. Deng, FEM prediction of welding residual stress and distortion in carbon steel
509–518, https://doi.org/10.1016/j.jcsr.2018.11.003.
considering phase transformation effects, Mater. Des. 30 (2) (2009) 359–366,
[2] S.D. Liang, Z.Y. Jie, K.N. Wang, Numerical simulation and ultrasonic measurement
https://doi.org/10.1016/j.matdes.2008.04.052.
on welding residual stresses of full penetration U-shaped stiffening ribs, Structures
[30] G.M. Fu, M.I. Lourenco, M.L. Duan, S.F. Estefen, Influence of the welding sequence
46 (2022) 322–335, https://doi.org/10.1016/j.istruc.2022.10.072.
on residual stress and distortion of fillet welded structures, Mar. Struct. 46 (2016)
[3] Ž. Božić, S. Schmauder, H. Wolf, The effect of residual stresses on fatigue crack
30–35, https://doi.org/10.1016/j.marstruc.2015.12.001.
propagation in welded stiffened panels, Eng. Fail. Anal. 84 (2018) 346–357,
[31] N.R. Mandal, Ship Construction and Welding, Springer Nature Singapore Pte Ltd,
https://doi.org/10.1016/j.engfailanal.2017.09.001.
Singapore, 2017.
[4] F. Jiang, Z.Q. Fu, B.H. Ji, L. Wan, Fatigue life evaluation of deck to U-rib welds in
[32] N.X. Ma, Y. Ueda, H. Murakawa, H. Maeda, FEM analysis of 3-D welding residual
orthotropic steel deck integrating weldment size effects on welding residual stress,
stresses and angular distortion in T-type fillet welds, Join. Weld. Res. Inst. 24 (2)
Eng. Fail. Anal. 124 (2021) 105395, https://doi.org/10.1016/j.
(1995) 115–122.
engfailanal.2021.105359.
[33] Q.H. Zhang, Y. Ma, C. Cui, X.Y. Chai, S.H. Han, Experimental investigation and
[5] R. Haghani, M. Al-Emrani, M. Heshmati, Fatigue-prone details in steel bridges,
numerical simulation on welding residual stress of innovative double-side welded
Buildings 2 (2012) 456–476, https://doi.org/10.3390/buildings2040456.
rib-to-deck joints of orthotropic steel decks, J. Constr. Steel Res. 179 (2021)
[6] Q. Zhao, K.S. Chen, M.Z. Chen, B. Briseghella, Z.Y. Guo, G.T. Yang, Use of plastic
106544, https://doi.org/10.1016/j.jcsr.2021.106544.
correction formula to improve accuracy of welding residual stress test with blind-
15