Determination of Loading and Residual Stresses on Offshore Jacket Structures by X-ray Diffraction

Abstract

As basements of offshore wind turbines (OWTs) in deep water (>50 m), jacket structures are an economic alternative to monopiles. For this reason, the structural durability of jackets has become more important. In such structures, welded tubular joints are weak points for fatigue design. The harmful effect of tensile residual stresses in welding joints is well known. For these reasons, the residual stresses and the loading stresses of offshore jacket structures were determined by X-ray diffraction (XRD) using a mobile diffractometer. This allows us to directly determine the load stress at the fatigue-critical locations, namely at the weld toe at the testing rig. High tensile residual stresses up to 250 MPa were determined in a welded (and unloaded) condition. At a loaded structure (10,000 load cycles), a lower residual stress level was determined. During loading, a local increase in the stress at the welded joint that is between 1.4 and 4 times higher than the applied nominal stress was determined. Furthermore, it is shown that additional treatment (grinding and clean blasting) influences the local stress state significantly.

1. Introduction

Most of the existing offshore wind turbines (OWTs) use monopile foundations and are installed in sea water depths of <50 m. For larger turbines in deeper waters, monopiles become very large and increasingly uneconomical due to the difficulty of fabricating and installing such systems, as well as the consideration of modal requirements. Lattice structures, such as jackets derived from oil rigs, are lighter and yet stiff alternative to monopiles for OWTs, see Figure 1a. Modern offshore jacket structures for supporting wind turbines are exposed to severe environmental conditions. Failure will result in environmental impacts and in significant financial losses [1]. Thus, the structural reliability assessment of these structures gains importance [2]. As in many welded structures, the welded joints at these jackets, see Figure 1b, are structural weak points regarding their low fatigue resistance compared to the unaffected base material [3]. The main reasons are the stress concentration of the geometrical notch (weld toe), the coarse grain microstructure at the transition from the weld material to the base material (heat affected zone, HAZ), and harmful tensile residual stresses at the weld toe [3].

Residual stresses are the consequence of heterogeneous plastic deformation in parts and components. Regarding the fatigue of welded structures, the term residual stress refers to macroscopic residual stresses (I. kind) [4]. These are homogenously distributed over several grains and result from the thermally induced extension and compression of heated material volumes during the passing of the welding heat source. The extension and compression of material are restrained by colder adjacent material resulting in local plastic strains after cooling to ambient temperature. These various materials, e.g., structural steels, show the phenomena of a phase transformation during heating and cooling, which goes along with a change in the packing density and thus leads to restrained expansion [5].

The relationships between residual stresses, material strength, and external load and their influence on the fatigue strength of a welded joint are complex. It is well known that residual tensile stresses decrease the fatigue strength, while residual compressive stress causes an increase in fatigue life [6,7]. However, residual stresses have a lower impact on the fatigue strength at high-stress levels than at low-stress levels [8] due to their relaxation under loading if yielding occurs [9,10]. The weld transition is the fatigue-critical notch that exhibits a multiaxial stress state. The yield criterion depends on this stress state, the material strength, and the residual stress state whereby the residual stress state changes under load. The consideration of the residual stress relaxation is an important factor regarding their influence on the fatigue strength of welded joints [11].

Based on the mentioned reason, the assessment of residual stresses on the welded joints of offshore jacket structures is essential. However, to the best knowledge of the authors, lacking investigations have not been performed or published yet. Furthermore, it is assumed that the assessment of local stresses at the design hot spots in jacket structures under loading are from crucial importance for the fatigue assessment. The local notch effect at the weld toe, as well as global misalignment of the structure, may lead to high local stresses and stress concentrations that cause local crack initiation. For these reasons, the aim of the current study is to investigate the local stress state of jacket structures in manufactured (as-welded) conditions without loading or under loading conditions.

2. Material, Specimens, and the Welding Process

The diffractometer measurements discussed here were made on models of the jacket nodes on a scale of ca. 1:1.5 (type A) and ca. 1:2 (type B). The test specimens are made from round welded hollow structural sections in S355 G10 + M steel (typical material for offshore structures). The material properties are given in

Table 1. Material properties of the base material S355 G10 + M (acc. to certificate).

Yield Strength [MPa]	Tensile Strength [MPa]	Elongation [%]
431–466	502–542	22

. Type A is 3000 mm high, chord 813 × 39.7 mm with L = 2150 mm, and brace 508 × 14 mm. Type B is 2400 mm high, chord 660 × 30 mm with L = 1750 mm, and brace 406 × 12 mm. The technical drawings of the specimen are given in Figure 2. The braces are manually welded to the chords with MAG (metal active gas, process 135 according to DIN EN ISO 4063). The edges are clean blasted, brushed, and grinded for welding preparation. The electrodes T46 2 P C1 H5/T46; Ti 52 T-FD and protective gas M21 ArC 18; 82% Ar/18% CO2 were used. After welding, ultrasonic non-destructive testing according to DIN EN ISO 23624 was performed. X-ray diffraction (XRD) measurements were performed at P1 to P6, see Figure 2. The conditions of the measurements are given in this Figure.

3. Fatigue Tests

The fatigue behavior of these jacket nodes is being investigated in an extensive fatigue testing program at Test Centre Support Structures Hannover (TTH). For this purpose, the test specimens described under 2 are loaded with a sinusoidal load in the tensile range. The test setup is shown in Figure 3. The nodes are cyclically loaded with maximum axial tensile force varying between 400 and 900 kN, leading to failure at 100,000–500,000 cycles.

Before the tests, finite element modeling is performed to determine the most loaded and fatigue critical spots, see Figure 4a. Based on these simulations, the measurement sensors are positioned as close as possible to the hot spot. The stress is recorded via strain gauges and by means of DIC (digital image correlation), see Figure 4b.

4. Residual Stress Measurement

4.1. X-ray Diffraction

X-ray diffraction is probably the most widely used method for the near-surface residual stress (RS) evaluation in polycrystalline materials. The method is based on Bragg’s law and Hook’s law [12,13]. This technique uses the interplanar spacing as the internal strain gauge for the residual strain measurements [14]. According to Bragg’s law, a polycrystalline material will diffract the incident X-ray beam with an angle proportional to the lattice spacing and the beam wavelength [15]:

$$n\lambda =2d sin\theta$$

(1)

where n is the order of interference, λ is the x-ray wavelength, d the lattice spacing, and θ represents the Bragg’s angle. RS changes the interlayer spacing of the crystal and thereby cause shifts in the reflection position, as illustrated in Figure 5a. The shift in the reflection position could therefore be used to calculate the residual strain in the material [16]:

$$\epsilon =\frac{d-d_0}{d_0}$$

(2)

in which d is the lattice spacing of the material with the RS and d₀ is the lattice spacing of the stress-free material. The width of the diffraction pattern at half intensity (full with half maximum, FWHM), see Figure 5a, was evaluated as a second parameter at each measurement position.

4.2. Cos $\alpha$-Method

The stress analysis by the cos $\alpha$-method according to Taira et al. [17] is based on a strain evaluation over the complete Debye-Scherrer-ring based on a 2D-detector (digital image plate, IP). The reliability of this method in combination with the 2D-detector compared to the commonly used sinψ²-method [18] was achieved by Sasaki et al. [19,20,21] for austenitic steels ({311}-lattice plan) as well by [22] for martensitic steels ({211}-lattice plan). Furthermore, Sarmast et. al. [23] show a high agreement between RS-measurement with the cos $\alpha$- and sinψ²-method for a wide range of materials and material conditions if shear stresses are not present or low (this is assumed to be the case for welded joints).

For the RS evaluation according to the cos$\alpha$-method, the strain in circumferential direction is measured by a shift in the diffraction angle ${\theta }_{\alpha }$ or radius $r_{\alpha }$ depending on the $\alpha$-position on the detector, shown in Figure 5b. For the measurement detector, distance to specimen $L$ and tilt angle ${\psi }_0$ are constant. A strain parameter ${\epsilon }_{a1}$ is defined based on four strains from $\alpha$ = 0° to 90°. The stress ${\sigma }_{\varphi }$ is calculated according to Equations (3) and (4). A detailed description of the method was published by [24].

$${\sigma }_{\varphi }=-\frac{E}{1+\upsilon }\frac{1}{2sin\eta sin2{\psi }_0} \frac{\delta {\epsilon }_{a1}}{\delta cos\alpha }$$

(3)

$$\mathrm{with} {\epsilon }_{\alpha 1}=\left[\left({\epsilon }_{\alpha }-{\epsilon }_{\pi +\alpha }\right)+\left({\epsilon }_{-\alpha }-{\epsilon }_{\pi -\alpha }\right)\right]/2$$

(4)

4.3. Measurement Approach

For the XRD-analysis by the cos $\alpha$-method, a Pulstec µ-360 type diffractometer was used assuming a biaxial stress state. According to the recommendation of [25], a collimator diameter of 1 mm was used. The diffractometer was mounted on a magnetic holder, see Figure 5c. The position of the XRD-diffractometer could be adjusted in two directions with micrometer gauges. Due a high flexibility of the diffractometer regarding the measurement distance $L$, the diffractometer was positioned perpendicular to the welded joints, see Figure 5c. This leads to a distance $L$ of 60 mm to 71 mm. The exact tilt angle ${\psi }_0$ at each measurement position P1 to P6 was determined by using a digital water level. To assure that the maximum distance of 71 mm was not exceeded, a tilt angle of ${\psi }_0$ = 20° to 25° was used. The measurement parameters are summarized in

Table 2. Measurement parameters for the determination of residual stresses at both jackets.

Radiation	Lattice Plane	Kollimator $\varnothing$	Exposure Time	Distance $\mathit{L}$	Tilt Angle ${\mathit{\psi }}_0$	Measurement $\varnothing$	Indentation
[-]	[-]	[mm]	[s]	[mm]	[°]	[mm]	[µm]
CrK$\beta$	{211}	1	14	60–71	20–30	$\approx$2.25 *	4–5 **

* Estimated according to [26] ** Estimated according to [27].

. For the evaluation, a Young’s modulus of $E$ = 220 GPa and a Poisson’s ratio of $v$ = 0.29 were used. Before the measurements, the oxid layer at the measurement positions was removed by citric acid. To determine the residual stress state at the fatigue critical location, at the weld toe, multiple measurements were performed in a line from the welded material at the weld bead to the base material. Only the stresses perpendicular to the welding direction were determined in this study.

Figure 5. (a) Principle of residual stress analysis by X-ray diffraction, (b) principle of stress analysis by the cos $\alpha$-method according to Tanaka [24], and (c) measurement equipment in use.

Figure 5. (a) Principle of residual stress analysis by X-ray diffraction, (b) principle of stress analysis by the cos $\alpha$-method according to Tanaka [24], and (c) measurement equipment in use.

5. Results

5.1. Structure in As-Welded (Unloaded) Conditions

The XRD-measurements were performed at the jacket structure (type A) in as-welded conditions, that means no loading or additional treatment was performed after welding. The results for the positions P5 and P6, see Figure 6, were illustrated in Figure 6. As shown, high tensile residual stresses between 200 MPa and 230 MPa were determined at the weld toe at both positions. In a distance of a few millimeters from the weld toe, compressive residual stresses were determined. In this area, clean blasting was performed before welding. As secondary output, the full width half maximum (FWHM), illustrated in Figure 5a, is used to evaluate the positioning of the measurement [28]. The FWHM correlates with dislocation density [29] as well as with mechanical properties such strength and hardness [30,31]. Thus, gradients in the FWHM are related to microstructural and hardness gradients in the heat affected zone (HAZ) between welded material and base material. It is shown that the FWHM values vary at the weld toe, see Figure 6. The mean value at the weld toe is 3.03° at P5 and 3.29° at P6, which is related to different material conditions at these measurement positions.

5.2. Structure under Loading Conditions

XRD-measurements were performed at jacket type B that is mounted on a fatigue testing rig, see Figure 3. The measurements were performed at four positions P1 to P4. P1 and P4 are the structural hot spots of the structure (type B). Position P2 is located directly and P4 is located close to the start–stop position of the final weld layer being removed by manual grinding. The measurement was performed unloaded and under a tensile load of 500 kN without changing the position of the diffractometer.

Figure 7 summarizes the measurement results of jacket type B. As shown, tensile residual stresses up to 125 MPa (P1) and up to 250 MPa (P4) were determined at the weld toe. While no tensile residual stresses were measured at the ground position P2 and at the position P3, for all positions, the highest tensile residual stress was measured at the toe. Interestingly, the highest tensile residual stresses were determined close to the ground location at P4. Moreover, a wide variation in the FWHM values was determined in the range from 3.05° at the weld toe of P4 to 4.07° at the weld toe of P2.

Under loading, an increase in the measured stresses was determined at each position. Thus, the highest increase was not always determined at the structural notch at the weld toe where it might be expected. However, the determined maximum loading stresses ${\sigma }_{load}$ between stresses in an unloaded condition of 47 MPa (P1), 132 MPa (P2), 88 MPa (P3), and 95 MPa (P4) and in loaded conditions were much higher than the nominal stress of $S$ = 33 MPa (based on a load of 500 kN). By comparing loaded and non-loaded conditions, a shift of the FWMH was also determined.

6. Discussion

An important factor regarding the assessment of residual stresses at large structures and relating them to fatigue strength is the estimation of the measurement accuracy. XRD measurements of welded joints are usually performed under laboratory conditions, which was obviously not the case in this study. For example, in the laboratory the tilt angle and orientation of the X-ray diffractometer to the adjusted precisely by a robot. Moreover, the influence of extraneous light on the 2D-dector (IP) could be avoided. However, investigations by Mattes [26] show a comparably low effect on the measurement results at daylight (500 +/− 100 lm/m²). Thus, the influence of extraneous lighting conditions was neglected. The main challenge was the positioning of the measurement device under an exact angle compared to the reference plane. Due the complex shape of the jacket structures and the curvature of the surface, the exact tilt angle changes at each measurement position. Mattes [26] determined a difference of about 8% for a change in the tilt angle of ${\psi }_0=$ +/−3°. It is assumed that a tilt angle can be adjusted within an accuracy of +/−5°. This should be considered regarding the interpretation of the results in this paper.

Former studies have shown that significant residual stresses close to the yield strength of the base material could be present in large welded parts [10]. The measurements thus confirms the recommendation of the IIW Guideline that tensile residual stresses in the range of the yield strength should be assumed for thick-walled welded structures [3]. The results of jacket type A support this assumption as well. However, it is a well-known fact that residual stresses in welded joints relax under cyclic loading [9]. The comparison of the residual stresses at P6 of jacket type A (as-welded condition) with P3 of jacket type B (pre-loaded with 10,000 load cycles) show major differences in the residual stress level close to the weld toe. One reason for this could be the mechanism of residual stress relaxation due to cyclic loading.

In addition, the results show that the welding process not only affects the measured residual stress state but also the surface conditions. Sand and blast cleaning induce compressive residual stresses that are not related to the welding process [32]. It is strongly assumed that this is the reason for compressive residual stresses at the base material (shown at P5 and P6). Grinding of the start–stop positions and locations with a high-stress concentration also changed the local residual stress state significantly (shown at P2 and P4). At position P2, the tensile residual stresses were completely removed. However, at P4 close to the ground location, high tensile residual stresses were determined. It is assumed that these are related to the grinding process as well.

Regarding the practical applications of the performed XRD-analysis for load monitoring or residual stress assessment in large structures, it should be mentioned that the most important limitation is the relatively small measurement area (only single sport measurements). Whereas, other systems such DIC-systems, see Figure 4b, are able to detect the strain at a comparable large area. However, the XRD technique provides absolute stress values (superposition of residual and loading stresses) in comparison to the DIC-technique. For this reason, the XRD analysis is reasonable at components where the weak points are geometrically defined. In the case of jacket structures, the structural load stress reaches the maximum around a length of a few 100 mm, see Figure 4a. The DIC system can easily cover this area within a single measurement field compared to the XRD-system. However, for the residual stress assessment, due to small dimensions and short measurement times, the applied XRD technique seems to be a good method for the residual stress assessment in large structures.

7. Conclusions

The measurement of residual stresses in unloaded conditions and the resulting effective stress (superposition of loading and residual stress) were determined at the welded joints of offshore jackets structures. For this, X-ray diffraction techniques with a mobile diffractometer were used. All stresses were measured perpendicular to the weld toe. The evaluation of the residual stresses was performed by the cos$\alpha$-method. As a secondary output, the full width half maximum (FWHM) was evaluated and found to correlate with the dislocation density and the mechanical properties of the material. The following conclusions are drawn:

• High tensile residual stresses up to 250 MPa were determined in as welded (unloaded) conditions of the jacket structure. The highest tensile residual stresses at each measurement position are always located at the transition of the weld bead to the base material (weld toe);
• The measured loading stresses (difference of effective stress and residual stress) are between 1.4 and 4 times higher than the applied nominal stress. It is assumed that this is based on the local notch effect as well as the global misalignment of the structure;
• Local grinding for removing the start–stop position significantly changes the residual stress state and removes the tensile residual stresses.

It should be mentioned that the quality of the measurement could not be compared to laboratory conditions and the measurement error may be higher compared to these measurements. It is assumed that the additional tilt angle error is around +/−15%, see Section 6. However, the residual stress measurement by mobile diffractometers shows a great potential for the assessment of local residual and loading stresses in welded structures and components. The variable measurement distance $L$ and the short exposure time of 2D-detectors enables an effective assessment of such local stresses and relates them to the fatigue failure of welded joints.

In the next step, the fatigue tests of these structures must show if fatigue cracks initiate or start at positions where high tensile residual stresses or high loading stresses were measured, or even do not start at positions where compressive residual stresses were determined. Multiple factors may be responsible for fatigue crack initiation besides the local residual stress state, for example, variations in the local weld geometry or global misalignment. However, both effects could be detected by the performed in situ XRD analysis during fatigue loading.