Experimental research on dynamic tensile behavior of full-scale weld-necked flange joints used in transmission steel tubular towers

2.1. Test specimens

Three full-scale WF joints were fabricated and tested in the laboratory to investigate their behavior under axial tensile load. The steel tube, which were welded to the new WF joint by a circular full penetration welding, has an outer diameter of 356 mm and a thickness of 8 mm, as this size is very common in steel tubular towers. The detail geometric parameters of these new WF joints are shown in Table 1.

As shown in Fig. 1(c), the steel tube was connected to the new WF joint through a circular full penetration welding. The stiffness of the flange plate is of importance to reduce the prying force for the new WF joint. The thickness of the flange plate is therefore an important parameter for the new WF joint. As shown in Table 1, three flange plate thicknesses are studied. The plate thickness of F2 is determined due to design results, which will be discussed in detail later. The other two specimens (i.e., specimen F1 and F3) are used for comparative analysis. To avoid local stress concentration, the fillet radius between flange neck and plate was set as 8 mm. The steel grade for the tube is Q460 and the steel grade for the flange is Q420. High strength bolts (grade 8.8) are used to fasten the flange plates.

The diameter of bolt holes in the new WF joint was 2 mm larger than the nominal diameter of steel bolts. The steel bolts were tightened to designed torques by a dynamometric wrench. The applied torque of M24 bolt was 280 N·m [3]. The design preloads were taken as equal to $P_0=T_r/(K_t·d_0)$, in which $T_r$, $K_t$ and $d_0$ are the applied torque, the torque coefficient (= 0.25), and the steel bolt diameter, respectively. All steel bolts were equipped with two nuts and two flat washers to prevent potential loose during loading process. Meanwhile, tensile tests were carried out on the coupons, which were extracted from the tubes and flanges, with three specimens in each group. The mean values of the mechanical parameters of steel are presented in Table 2.

2.2. Test set-up

Two 40 mm-thick end plates, which were welded to the steel tubes, were used to apply tensile loads. In order to avoid local failure of the end plates, twelve 10 mm-thick stiffened plates were welded at the ends of steel tubes. The distance between the end plates and flanges was 600 mm, as depicted in Fig. 3, and the end plates were connected to the reaction beam with anchor bolts (M45) to form a self-balancing loading system. Two hydraulic jacks, with 500 t loading capacity, were employed to apply monotonic tension, and they worked well in synchronism with each other as their source of power was from the same oil channel.

Fig. 3Experimental set-up

a)

b)

To investigate the stress variation of the WF joint, strain gauges were mounted at typical positions, such as flange plate, flange neck, tube wall near the weld toe, and mid-section of the tube. Four gauged bolts were also introduced into each connection, and two displacement transducers were placed at plate edges of each flange to measure the plate deformation. The dimension of strain gauge is 10.0 mm×4.0 mm, with maximum ultimate strain of 2 %-3 %, and the arrangement of strain gauges was shown in Fig. 4. Prior to the tests, the preloads were exerted and strains of the same location were observed to confirm the specimens were concentrically loaded. The monotonic tension was applied in increments, and each increment is equal to 4 % of the yield capacity of the steel tube. At each load increment, a data acquisition system was adopted to record the readings of strains and displacements.

Fig. 4Strain gauge placemen

2.3. Experimental results

2.3.1. Dynamic load-strain behavior of flange and steel tube

In the early stage of loading, the specimens were elastic, and no obvious deformation was observed. With the increase of dynamic tensile load, the dynamic strains at typical positions bifurcated gradually, and relatively obvious tensile deformations occurred in steel tubes. The tests were ceased once any buckling or cracking failure occurred, or the maximum measured strain reached 15000 u, as it is the maximum value the strain gauge can measure. Table 3 lists the yield capacities ($F_{de}$), which are calculated with the actual yield stress obtained from coupon tests, and average test capacity ($F_{te}$) of three WF joints, where the test capacity are the maximum load values of each WF joint. The ratio of $F_{te}$ over $F_{de}$ is 1.143.

Table 3The experimental results of specimens

Specimen	Tube size	$F_{tei}$ (kN)	$F_{te}$ (kN)	$F_{de}$ (kN)	$F_{te}$ / $F_{de}$	$\mathrm{\Delta }l$ (mm)	$\mathrm{\Delta }t$ (mm)
F1	$Ф$356×8 /$Ф$356×8	5500	5466.7	4781.7	1.143	6.2	1.08
F2		5400				9.6	0.94
F3		5500				8.2	0.37

The measured strains represented the local deformation at strain gauge locations, and the strain development tendency can be obtained by comparing load-strain curves of typical locations, as shown in Fig. 4. Fig. 5(a) and (b) show the strain development of mid-section of the steel tube and the tube wall near the weld toe. As the flanges of all three specimens provide much higher stiffness than the steel tube, the strains of the two sections developed consistently and almost overlapped. The tube section at the conjunction with the flange (i.e., the tube wall near the weld toe) was found to be loaded concentrically and the tension strain of the outer skin developed more rapidly than the inner skin. As the centroids of the steel tube and the flange were not collinear, bending moment were developed on this section. Due to the local bending moment, this section is the crucial part of the whole joint.

As shown in Fig. 5(c), the flange neck is also under the local bending moment, and the tension strain of the outer skin (i.e., point B in Fig. 5(c)) developed more rapidly than the compression strain of the inner skin (i.e., point A in Fig. 5(c)). In addition, the tension strain at the outer skin of F1 tends to be larger at the same tension load than that of F2 and F3. The reason lies that the prying force of flange plate of F1 may result in additional moment on flange plate and flange neck. Fig. 5(d) shows the strain development of flange plate, the plate of F2 and F3 remained elastic and the strains developed almost linearly, while the measured strain of F1 developed more rapidly, and the plate have already reached the yield status.

After the tests, the residual elongations of steel tubes ($\mathrm{\Delta }l$) were measured and recorded, as shown in Table 3. The residual deformation of steel tube was 6.2 mm in F1, 9.6 mm in F2 and 8.2 mm in F3, with the mean value of 8.0 mm. The steel tubes tested show good ductility.

Fig. 5The load-strain development of typical positions

a) The load-strain curves of mid-section of steel tube

b) The load-strain curves of tube near weld toe

c) The load-strain curves of flange neck

d) The Load-strain curves of flange plate

2.3.3. Dynamic load-strain behavior of connection bolts

Fig. 7 shows the typical load-strain curve of steel bolts for all specimens, the bolt response was essentially linear, and the strain of inner side of bolt shank developed more rapidly than that of outer side. It is obvious that the connection bolt is under local bending moment. The measured strains of steel bolts could be converted into bolt force ($T_{B1}$), and the equivalent bending moment ($M_{B1}$), as listed in Table 4. As expected, the equivalent moment also tends to be smaller with the increase of plate thickness.

After the tests, no obvious failure or fracture was observed for the steel bolts, except a slight bending deformation in the bolts of F1, as shown in Fig. 6(b). The experimental results show that the new WF joints and the steel bolts are reliable, which can meet the requirements of practical engineering projects.

Table 4The comparison between bolt force and moment obtained from FEA and test

Specimen	Bolt size (mm)	Test result		FEA result			$T_B$/$T_{B1}$	$M_B$/$M_{B1}$
		$T_{B1}$ (kN)	$M_{B1}$ (kN·m)	$T_B$ (kN)	$M_B$ (kN·m)	Contact stress (MPa)
F1	24	335.3	1.08	342.4	0.97	285.7	1.021	0.898
F2		285.7	0.74	283.5	0.67	3.87	0.992	0.905
F3		275.9	0.53	278.4	0.44	0	1.009	0.830

Fig. 6The comparisons of deformation of flange plates and connection bolts

a) The deformation of flange plates

b) Deformation of bolt

Fig. 7Load-strain curves of connection bolts (F2)

Fig. 8The strain development of bolts

3.1. Finite element model

Theoretical analyses were conducted using ANSYS. The steel material of steel tubes and flanges was modeled by an ideal elasto-plastic tri-linear model, which is often used for the simulation of structure steel, with mechanical properties obtained from coupon tests, and the tangent modulus $E_0$ was taken as 10 % of the Young’s modulus $E$. The isotropic hardening model was used in the post yield range.

In the FE model, the bolts, flanges, and steel tubes were modeled by eight-node SOLID185 element. The surface-to-surface contact algorithm was employed to simulate the interaction between all possible contact surfaces of flange plates, bolt nuts, bolt shanks and bolt holes. The target element TARGE170 coupled with contact element CONTA174 were used to simulate contact pairs, and friction coefficient of the contact interfaces was taken as 0.25 [3]. The equivalent temperature method was introduced to simulate the preloads in connection bolts, and the expansion thermal coefficient was set as 1.2×10^-5/°C.

The analytical model was then created, as shown in Fig. 8. Taking advantage of spatial symmetry, the flange could be divided into several sectors in the circumferential direction. The flange separation model was selected to reduce computing time and storage space. The arc-length method, together with modified Newton-Raphson method, was used to accelerate numerical solution convergence.

3.2. FEA results

As the bolt cluster circle and the tube wall didn’t overlap, the extra bending moment existed on flange plates. The contact pressure between flange plates was extracted from analytical models, as shown in Fig. 10, which could not be measured in the full-scale experimental tests. The simulated contact stresses between outer edges of flange plates of the three specimens are 285.7 MPa, 3.87 MPa, and 0. It means that some degree of prying force did exist between the flange plates of F1, and the prying force of F2 is negligible small. These show a similar trend with the experimental deformation of flange plates, as shown in Fig. 6. The prying force distributed uniformly at the outer edges of flange plates, and it leads to additional tension and moment on steel bolts.

Fig. 9The contact status between nut and flange plate

a)

b)

c)

Fig. 10The contact status between flange plates

Fig. 11The dynamic stress distribution on bolt shanks

Fig. 9 displayed the contact status between bolt nut and flange plate, the contact stress between nut inside and flange plate (i.e., zone A in Fig. 9) is much bigger than that of outside (i.e., zone B in Fig. 9). Fig. 11 shows the stress distribution of bolt shanks of three WF joints, and the maximum stimulated stress are 862MPa, 713 MPa, and 636 MPa. The simulated bolt stress could be converted in a linear elastic way to simulated tension ($T_B$) and equivalent moment ($M_B$). As shown in Fig. 12, $T_B$ and $M_B$ of connection bolts tends to be smaller with the increase of plate thickness. The comparisons between test results and FEA results show good agreement as shown in Table 4.

The FEA results were proved to be precise and reliable. With the help of FE models, an in-depth understanding on the tensile behavior of the new WF joints can be achieved, and a wide range of numerical calculations can then be used to create a simplified design model.

Fig. 12The development of simulated tension and moment of each configuration

a) Bolt tension

b) Bolt moment

4.1. Basic assumption

The experimental and numerical investigations presented are with the objective to propose a simplified design model for the new WF joint. The design procedure herein is based on the assumptions listed below:

a) The flange should be designed with higher capacity and rigidity than the steel tube which it splices;

b) In order to mitigate the effects of bending moment, the WF joint adopts inner-slope section, and the bolt cluster circle reduces relative to the steel tube wall.

c) The flange plate of the new WF joint is thicker than the traditional WF joint, and prying force does not exist at plate outer edges. The experimental results indicate that a thicker plate can help to reduce the additional tension due to prying force between flange plates and bending moment on bolts.

d) The connection bolt is assumed to be axially loaded, and the sum of steel bolts’ tensile force is equal to the tension of corresponding tube section.

e) The effect of circular penetration weld is not considered in the design model. It is therefore required that the full penetration weld should be employed between flange and tube, and the weld should provide equal or higher strength than the steel tube.

4.2. Design model

According to the basic assumptions listed above, a simplified design model can be proposed due to the force characteristics of each flange separation. There are three fundamental steps in the design procedures. Firstly, calculate the bolt number and size based on the tension of steel tube. Secondly, determine the geometric dimension of flange plate. Thirdly, determine the geometric dimension of flange neck.

4.2.1. Design of flange plate

For the new WF joint, the flange plate has good local stiffness and works as a cantilevered beam with a concentrated load acted at mid-span. To simplify the calculation formula, the flange plate is divided into several sectors in the circumferential direction, which has a steel bolt hole. Fig. 13 describes the force mechanism of each plate sector schematically. $T_{B0}$, the tension of a connection bolt, can be given by Eq. (2), Where $T$ is the tensile capacity of steel tube, $n$ is the number of the connection bolts:

1

$T_{B0}=\frac{T}{n}.$

The arc length $S_d$ of each tube section can be expressed by Eq. (2), where $d$ is the outer diameter of steel tube:

2

$S_d=\frac{d}{2}\cdot \theta =\frac{d}{2}\cdot \frac{2\pi }{n}=\frac{\pi d}{n}.$

As shown in Fig. 13, each plate separation is mainly subjected to bolt force $T_{B0}$, radial moment $M_r$, and tangent moment $M_t$. The equilibrium moment equation can be established as Eq. (3). Due to calculation results of tiny deformation theory of elastic annular plate [18-23], $M_t$ can be expressed by $r$×$M_r$, where $r$ can be taken as 0.3. Eq. (3) can be converted into Eq. (4):

3

$M_r+2M_t\mathrm{s}\mathrm{i}\mathrm{n}\left(\frac{\theta }{2}\right)=T_{B0}\cdot B,$

4

$M_r=\frac{T_{B0}\cdot B}{1+2r\cdot \mathrm{s}\mathrm{i}\mathrm{n}\left(\frac{\theta }{2}\right)}.$

Then, the thickness of flange plate, $C$, can be determined by:

5

$C=\sqrt{\frac{5M_r}{S_d\cdot f_y}}=\sqrt{\frac{5n{M}_r}{\pi d{f}_y}}.$

Fig. 13The force diagram of flange plate separation

4.2.2. Design of flange neck

To simplify the design procedure, the flange can be divided into two halves, and each half is mainly subjected to tube tension ($T$/2), bolt force ($T$/2) and bending moment ($M_0$), as depicted in Fig. 14. For each flange separation, the equilibrium equation can be expressed as [24-27]:

6

$\sum M_O=0\Rightarrow M_0+0.5T(y_t-y_b)=0,$

where $y_t$ is the force arm of tube tension, and $y_b$ is the force arm of bolt force. They can be given by:

7

$y_t=\frac{R_t\cdot \mathrm{s}\mathrm{i}\mathrm{n}\mathrm{\alpha }}{\alpha }=\frac{2R_t}{\pi }=\frac{d}{\pi },$

8

$y_b=\frac{2R_b}{\pi }=\frac{K}{\pi }.$

$K$ is the diameter of bolt pitch circle, and Eq. (6) can be written as:

9

$M_0=\frac{(K-d)T}{2\pi }.$

Fig. 14The force diagram of flange separation

Fig. 15The sectional centroid of flange cross-section

For the new WF joint, the cross section can be divided into three parts (Fig. 15), and the sectional centroid can be determined as:

10

$y_{O\text{'}}=\frac{\sum A_i\cdot y_{o_i}}{\sum A_i}=\frac{A_1{y}_{o_1}+A_2{y}_{o_2}+A_3{y}_{o_3}}{A_1+A_2+A_3}.$

The sectional inertia moment $I_y$ and sectional modulus $W_y$ can be derived as:

11

$I_y=\sum I_{y_i}+{\sum A_i\cdot (y_{oi}-y_{o\text{'}})}^2,$

12

$W_y=\frac{I_y}{(H-y_{o\text{'}}).}$

Then, the flange sectional stress ${\sigma }_f$ can be written as:

13

${\sigma }_f=\frac{M_0}{2W_y}=\frac{(K-d)\cdot (H-y_{0^{\text{'}}})\cdot T}{4\pi \cdot I_y}\le f_y.$

4.3. Design procedure

Based on the above two simplified design models for flange plate and flange neck, a practical design procedure is proposed here, as shown in Fig. 16. It should be noticed that the procedure is based on the basic assumptions described in Section 4.1. Above all, the new WF joint adopts inner-slope section, and the prying force does not exist at plate outer edges.

For the three new WF joints, the detail dimensions of F2 are calculated according to the design procedure; the other two joints are used for comparative analysis and verified the rationality of the method. The experimental and numerical investigations indicate that when the steel tube reach the yielding status, the plates of F1 were contacted through the test, while the plates out-edge of F2 and F3 were separated, it means the prying force does not exist between the flange plates, proving the accuracy of the model.

Fig. 16Design procedure