Analytical and numerical investigation of bolted steel ring

flange connection for offshore wind monopile foundations

C A Madsen1*, J-C Kragh-Poulsen2, K J Thage3 and M J Andreassen4

1,2
COWI A/S, Department of Marine & Foundation Engineering, Parallelvej 2, DK-
2800 Kgs. Lyngby, Denmark
3COWI A/S, Department of Major Bridges International, Parallelvej 2, DK-2800 Kgs.

Lyngby, Denmark
4
Technical University of Denmark, Department of Civil Engineering, Brovej, Building
118, DK-2800 Kgs. Lyngby, Denmark

*Contact author: [email protected]

Abstract. The monopile foundation is the dominant solution for support of wind turbines in
offshore wind farms. It is normally grouted to the transition piece which connects the
foundation to the turbine. Currently, the bolted steel ring flange connection is investigated as
an alternative. The monopile--transition piece connection has specific problems, such as out-
of-verticality and installation damage from driving the MP into the seabed and it is not fully
known how to design for these. This paper presents the status of the ongoing development
work and an estimate of what still needs to be covered in order to use the connection in
practice. This involves presentation of an analytical and non-linear FE analysis procedure for
the monopile-transition piece connection composed of two L flanges connected with preloaded
bolts. The connection is verified for ultimate and fatigue limit states based on an integrated
load simulation carried out by the turbine manufacturer.

1. Introduction
Monopile foundations are the prevailing substructure for offshore wind turbine generators (WTG)
supporting 81% of the WTGs in operation at the end of 2016 [1]. The monopile (MP) substructure
consists of circular steel tubes welded together and driven to target penetration by hydraulic
hammering. The bottom of the WTG tower section is bolted directly to the MP or to the top of the
transition piece (TP) by the use of flanges pre-welded to the tubular shells. The majority of the
installed MPs have a TP, as shown in Figure 1. The loads between the MP and TP are typically
transferred by a grouted connection. Settlements of the TP on several offshore wind farms (OWF) in
2010 [2] with the traditional cylindrical grouted connection led to the development of today's most
widely used conical grouted connection and alternatives such as the bolted ring flange connection or
slip/double slip joints. Current tendencies are that an increased number of offshore wind farms are
moving towards the bolted flange solution between the MP-TP [3].
 Figure 1. Conceptual sketch of the MP foundation incl. a TP joined to the bottom of the WTG tower.
The top of the monopile is joined to the transition piece and the bottom of the monopile is fixed in the
seabed. Figure: © COWI A/S.
This paper presents the ongoing work of extending the established design procedures of the bolted
steel ring flange connection for the WTG tower to the requirements of the MP-TP connection. The
WTG tower flange is typically located as shown in 0, pos. A-D. The position of the MP-TP
connection, pos. E, leads to specific considerations of the MP-TP flange connection:
 During installation the MP is driven into the seabed by direct hammering on the pile top. This may
cause deformation of the flange exceeding limiting values not taken into account in established
design procedures.
 Driving the MP into the seabed can result in the MP not being perfectly vertical and the WTG
mounting flange being inclined. This is not covered in existing design procedures.

B
C

D
E

Figure 2. Conceptual and idealized sketch of the MP-TP foundation indicating

potential positions for application of a bolted steel ring flange connection. Figure: © COWI A/S.
 The work described in this paper is limited to investigation of the ultimate (ULS) and fatigue (FLS)
limit states of the flange connection during operation of the WTG. ULS and FLS are investigated by
well-known analytical methods considering the shells and the two flanges of the MP-TP. The
analytical analysis is accompanied by definition of similar numerical models. This is done to compare
the methods and to gain further knowledge of the safety levels included in the analytical approach. An
ideal MP-TP flange geometry is assumed in the numerical study, and local and global tolerances are
not considered. ULS and FLS loads applied for the investigation are obtained from an offshore wind
farm designed by COWI A/S.

2. Analytical approach
The tensile ULS and FLS resistance of the circular flange connection used in present paper are
approximated by the analytical methods of Petersen-Seidel and Schmidt-Neuper [4]-[6]. The methods
have already been studied by several researchers in relation to the design of the WTG tower or similar
tubular flange connections. The review presented in the following, highlights the existing methods and
assumptions. Furthermore, it will form the basis for evaluating the key issues associated with the
specific requirements of the MP-TP connection.

Figure 3. Bolted ring flange connection and equivalent

single bolt segment. Figure from [7].

2.1. Ultimate limit state

The tensile ULS resistance of the flange connection by Petersen and Seidel consider a segment cut out
of the entire flange connection (L segment) including a single bolt, see Figure 3. The approach assume
linear bolt-load behaviour as preload is not considered. Furthermore, perfectly aligned flanges are
assumed.
The L segment is investigated for the equivalent tensile normal force, Fz specified in Equation (1).
The expression is derived from Navier's formula considering the foundations global bending moment,
MEd and axial force, NEd at the position of the circular flange connection. Additional verification is
needed for torsional and shear forces which are not considered in present paper. The loads are scaled
to the L segment by the number of bolts, nb in the flange connection and the mid-thickness diameter,
Dsh,m of the tubular shells.
4 ⋅ M Ed N
Fz = + Ed (1)
nb ⋅ Dsh ,m nb
The tensile resistance of the L segment, Fult, Equation (2) is found from static force equilibrium
considering the three failure modes defined in Figure 4.
Fz ≤ Fult = min( FU,A ; FU,B ; FU,C1 ; FU,C2 ) (2)
 Figure 4. Approximated failure modes for L flange (segment) connection
according to Petersen’s method [5]. Figure modified from [7].
2.1.1 Failure of the bolt: The tensile resistance for failure mode A is linked to tensile failure of the bolt.
The bending deformation of the flange caused by the eccentric tensile load is assumed small combined
with the assumption of a sufficiently large flange thickness resulting in absence of prying forces, Q.
The resistance of the segment flange for failure mode A, FU,A is therefore equal to the design tensile
resistance of the bolt, Ft,RD according to Equations (3) and (4). fy,bolt and fu,bolt is the design yield and
ultimate tensile strength of the bolt. As is the tensile stress area of the bolt and γM2 and γM7 partial
safety factors.
FU , A = Ft , Rd (3)

 f y ,bolt ⋅ As 0.9 ⋅ f u ,bolt ⋅ As 

Ft , Rd = min ;  (4)
 γM2 γ M7 
Failure mode A assumes that the shell is able to resist the moment, ME,Sh according to Equation (5),
M E ,Sh
< f yd ,Shell (5)
WE ,Shell
where WE,Shell is the elastic moment of resistance and fyd,Shell the design yield strength of the shell.
2.1.2 Failure of the bolt combined with plastic hinge (partial yielding) of the shell: Failure mode B considers
combined failure due to plastic yielding of the shell and failure of the bolt in tension. The resistance is
obtained from force equilibrium according to Equation (6) assuming the flange can resist the moment,
ME,Fl Equation (7).
Ft , Rd ⋅ a + M pl ,3
FU , B = (6)
a+b

M E , Fl ≤ WE , Flange ⋅ f yd , Flange (7)

2.1.3 Plastic hinge in shell and flange: Failure mode C examines the tensile resistance, FU,C for the
occurrence of failure of the shell and flange due to yielding. This supposes that the considered bolt has
sufficient capacity to resist the axial tension load, Ft,Ed including effects of prying forces. The
resistance is determined from Equation (8). Petersen's method outlined in [5] is different from [4] as it
further divides failure mode C into (C1) plastic hinge in the bolt hole center plane and (C2) plastic
hinges next to the bolt hole.
M pl , 2 + M Pl ,3
FU ,C = (8)
b
2.2. Fatigue limit state
The FLS verification of the circular flange connection is in this paper limited to investigating the
resistance of the bolts. The Schmidt-Neuper method [4] links the cyclic load in the form of a
max./min. equivalent normal force in the shell, Fz (max/min) to the bolt force, Fb. The stress range in the
bolt is determined from the bolt force from which the fatigue capacity of the bolt can be calculated.
The method is selected as it considers the effect of preload and resilience of the joint. The method,
however, does not consider bending stresses in the bolt which are taken into account during selection
of the fatigue reference strength (FAT class) for the bolt. The Schmidt-Neuper method is devised to be
valid for fabrication/installation imperfections of the flange within the specified range defined in Table
1 and illustrated in Figure 5. This implies that the bolt loads determined by the Schmidt-Neuper
method include additional prying forces as a consequence of the gap, k. Furthermore, it is assumed
that additional bending stresses in the bolts due to the inclination of the flange, αs are sufficiently
small.
The Schmidt-Neuper method approximates the non-linear relation between the shell and bolt forces
due to preload by a piecewise trilinear function. The three straight lines are derived by Equations (10)-
(12) and presented in Figure 6 as three regions:
Region 1: The bolt load for region 1, Fb1, Equation (10) is determined for the flange zone under
compression due to the preload. During the increase of the shell load the compressive zone moves in
the direction of the edge of the flanges. The bolt load is for region 1 a function of the tensile shell
force multiplied by the load factor for the total stiffness of the bolt, p.

Figure 5. Specification of flange tolerances. Figure from [9].

Table 1. Allowable flange tolerances. According to [9].

Criteria Limiting value
Flatness deviation, k per flange around
2.0 mm
the entire circumference
Flatness deviation, k per flange over a
1.0 mm
segment of 30°
Taper, αs to the inside of the connecting
0.0° to 0.7°
surface of each flange
Outer flange surfaces limit, αs 2°

The preload, FP,C* is added, representing the initial condition before the external load is imposed on
the joint. As a consequence, the bolt force relative to the shell force increases in region 1 according to
the total stiffness of the joint.
Fb1 = FP ,C* + p ⋅ FZ (10)
 Region 2: The tensile zone increases in region 2 and the flange connection begins to open, leading
to the bolt load, Fb2 given by Equation (11), to increase more rapidly than the shell force. The first part
of (11) represents the maximum bolt force for region 1, and the last part the increase in magnitude of
the bolt force relative to the increase in shell force.

Fb 2 = FP ,C* + p ⋅ FZ ,I ,max +
[λ ⋅ F
*
Z , II ,max ]
− ( FP ,C* + p ⋅ FZ ,I ,max ) ⋅ (11)
Fz − FZ ,I ,max
FZ ,II ,max − FZ ,I ,max
Region 3: Region 3 is where the tensile shell load has exceeded the compressive preload, Fz,IImax
and the flanges are in contact only at their edges. The bolt force, Fb3 given by Equation (12), increases
as a function of the shell force multiplied by the auxiliary parameter λ* (16). λ* represents the
additional contribution from prying forces due to eccentric load of the connection.
Fb 3 = λ* ⋅ FZ (12)
It is decided in the present paper to reduce the characteristic preload, Fp,C to 90% according to [8],
see Equation (13), to account for relaxation of the connection. Assuming a 90% reduction requires re-
tightening of the bolts 6 months after commissioning. The torque applied to the bolt should minimum
be 70% of the ultimate tensile strength of the bolt.

Figure 6. Trilinear approximation of the non-linear relation between

the shell force and the bolt force. Figure modified from [7].

FP ,C* = 0.7 ⋅ As ⋅ f u ,bolt ⋅ 0.9 (13)

The maximum limit shell force for region 1 and 2 is given in (14) and (15).
a − 0.5 ⋅ b
FZ ,I ,max = ⋅ FP ,C* (14)
a+b

1
FZ ,II ,max = ⋅F (15)
λ ⋅ q P ,C*
*

0.7 ⋅ a + b
λ* = (16)
0.7 ⋅ a
 The load factor for the total stiffness of the bolt, p, and the flanges, q, are obtained from Equation
(17) and (18) where Cs and CD is the spring stiffness of the tension- and compression spring. δp and δs
represents the elastic resilience of the clamped parts and the bolt. The spring characteristics of the
members in the connection are derived according to methods outlined in [10].
Cs CD
p= q= (17)
CS + CD CS + CD

1 1
CS = CD = (18)
δs δp

3. Numerical approach
The ULS and FLS resistance of the one bolt segment is in the present paper investigated by using
Abaqus CAE. A geometry equivalent to the analytical approaches presented in Section 2.1 and 2.2 is
used and no tolerances are included. Non-linear material properties are defined for the shell, flange,
and bolt assembly along with contact interaction specified between the parts according to Table 2. The
bolt and nut geometry is simplified where an equivalent shank diameter is specified for the bolt. Linear
brick elements (C3D8R) are assigned to the geometry. Different element sizes are used as part of the
refinement study. Two types of loads are defined, see Figure 7; a bolt load representing the preload
and a pressure load applied at the TP shell representing the equivalent normal force.
The mechanical boundary conditions specified for the L flange segment are shown in Figure 8.
This includes cylindrical symmetry conditions applied on both sides of the section cut of the segment
and a pinned support at the bottom of the MP shell.

Figure 7. Definition of loads. Left: Bolt load. Figure 8. Mechanical boundary conditions
Right: Segment shell load. Figure from [7]. specified for the segment model. Left: Pinned
support. Right: Cylindrical symmetry
condition. Figure from [7].
 Table 2. Contact surface properties.
Definition Flange to flange, Bolt to washer
Flange to washer
Friction coefficient 0.2
Behaviour Small sliding
Formulation Augmented Lagrange
Interface treatment Adjust to remove overclosure

4. Case study
The connection is designed as a one sided preloaded flange connection located on the inside of the
foundation structure. The connection has 140 bolts specified as hot dip galvanized M64, 10.9 grade
(HV M64-10.9 tZn) bolts. The bolt size exceeds M36, hence [11] is consulted for specification of the
geometry of the bolt, nut and washer. The flange and shell are specified as S355ML. No tolerances are
included in the case study. Non-linear material properties are estimated based on values defined in
[12].

4.1. Joint geometry and material parameters

The main geometrical and material parameters of the connection are defined in Figure 9 and listed in
Table 3. The material properties including strain hardening, considered in the numerical analysis, are
given in Figure 10.

Figure 9. Definition of joint geometry.

Figure 10. Properties of bolt, head, nut, washers

(bolt assembly), shell and flange materials.
 Table 3. Geometrical and material parameters.
Description Symbol Value Unit
Outer diameter of shell OD 6 [m]
Nominal shell thickness ts 0.08 [m]
Nominal flange thickness tf 0.18 [m]
Nominal flange width bges 0.31 [m]
Dist. bolt center to flange
a 0.15 [m]
edge
Dist. bolt center to shell
b 0.12 [m]
center
Inner radius ri 15 [mm]
Number of bolts nb 140 [-]
Nominal bolt size dnom 64 [mm]
Nominal diameter of washer dw 115 [mm]
Nominal washer thickness h 10 [mm]
Grip length lg 380 [mm]
Nominal bolt length lnom 450 [mm]
Shank length lsch 332 [mm]
Cross section area of shank Asch 3,217 [mm2]
Cross section area of bolt
Abd3 2,520 [mm2]
thread
Nominal diameter of washer dw 115 [mm]
Characteristic yield strength
fyk,sh 325 [MPa]
of shell (t=80mm)

4.2. Analytical approach

ULS and FLS resistance of the L segment are found analytically in this section.
4.2.1 Ultimate limit state: The analytical ULS determination of the L segment joint is carried out
according to Section 2.1. The found tensile resistance of the L segment and bolt configuration using
the corresponding equilibrium equations is summarized in Table 4. The L segment is seen to exhibit a
load deformation characteristic leading to failure mode B. Indicating development of a plastic hinge in
the flange at the transition to the shell and subsequently tensile failure of the bolt. Equivalent design
ULS loads at the MP-TP joint from a COWI A/S OWF project are presented in Table 5 including the
tensile equivalent normal force determined from Equation (1). The utilization ratio representing the
ratio between the equivalent tensile normal force and the resistance of the L segment is presented in
Equation (19).
4.2.2 Fatigue limit state: The analytical FLS determination of the L segment according to Section 2.2
is limited to an investigation of the cyclic loads in the bolt during operation of the WTG. Parameters
for establishment of the Schmidt-Neuper curve reflecting the bolt force as a function of the tensile
force in the shell wall are presented in Table 6. The spring stiffness of the bolt assembly and L-
segment under compression are specified according to procedures in [10].
 Table 4. Tensile resistance for failure mode a-c for the l flange single bolt connection.
Description Symbol Value Unit
Plastic limiting moment of flange without hole Mpl2 334 [kNm]
Plastic limiting moment of flange with hole M'pl2 150 [kNm]
Plastic limiting moment of tubular section Mpl3 56 [kNm]
Design value of limit tension force Ft,Rd 1,927 [kN]
Limit tension force failure mode A FU,A 1,927 [kN]
Limit tension force failure mode B FU,B 1,279 [kN]
Limit tension force failure mode C1 FU,C1 1,980 [kN]
Limit tension force failure mode C2 FU,C2 2,448 [kN]
Table 5. Resulting ULS design loads for full flange and segment model
Global design bending Global design Equivalent normal force
moment Axial force
MEd [kNm] NEd [kN] Fz [kN]
248,475 10,674 1,123
Fz
URZ = = 0.88 (19)
Fult

Table 6. Parameters for definition of bolt force as function of tensile force in shell wall.
Description Symbol Value Unit
Spring stiffness, bolt p 0.17 [-]
Spring stiffness, clamped flange q 0.83 [-]
Design preload Fp,C* 1,680 [kN]
Shell force max. limit (region 1) Fz,I,max 568 [kN]
Shell force max. limit (region 2) Fz,II,max 954 [kN]
Bolt force max. limit (region 1) Fb,I,max 1,775 [kN]
Bolt force max. limit (region 2) Fb,II,max 2,021 [kN]
Auxiliary parameter λ* 2.1 [-]

Figure 11. Characteristic tensile bolt force as a function of the shell force by the Schmidt-Neuper
approach. Figure modified from [7].
 The trilinear relation between shell- and bolt force for the segment flange is presented in Figure 11.
The tensile shell force, Fz, is determined from Markov matrices containing information on the moment
load cycles at the MP/TP joint due to the cyclic actions on the foundation. The Markov matrix
containing the governing directional cyclic loads relative to north is used as basis for the verification.
The moment loads in the Markov matrix are sorted into bins, j, based on the magnitude of the moment
range. Each bin contains ni number of load cycles. The extent of the Markov matrix used in this paper
is given in the format presented in Table 7. The Markov matrix includes information on min. and max.
moment ranges and min. and max. mean moment values for the load cycles included in each bin [13].
The Markov matrix contains 9.95∙108 load cycles divided into j = 5,373 bins which in this study are
assumed to reflect the expected load history for the foundation.
From the information in the Markov matrix the min. and max. values of the shell force are obtained
by Equation (20) to (25). Ashell is the cross sectional area of the shell for the L segment, WE,shell the
section modulus of shell, and Abd3 the minor cross section area of the bolt at the thread-root section.
Table 7. Markov matrix incl. bending moments and number of cycles sorted into bins. Example values
given in top row.
Min. moment Max. moment Min. mean Max. mean No. of
range range moment moment cycles
ΔMmin,j [kNm] ΔMmax,j [kNm] Mmean,min,j Mmean,max,j nj,I [-]
[kNm] [kNm]
0 1.398∙103 -1.053∙105 -1.036∙105 9
ΔMmin,2 ΔMmax,2 Mmean,min,2 Mmean,max,2 n2
ΔMmin,3 ΔMmax,3 Mmean,min,3 Mmean,max,3 n3
ΔMmin,4 ΔMmax,4 Mmean,min,4 Mmean,max,4 n3

∆M min, j + ∆M max, j
∆M mean, j = (20)
2

∆M mean, min, j + M mean, max, j

M m, j = (21)
2

∆M mean, j
∆Fz , j = Ashell ⋅ (22)
WE , Shell

M m, j
Fz , mean, j = Ashell ⋅ (23)
WE , Shell

Fz , max, j = Fz , mean, j + 0.5 ⋅ ∆Fz (24)

Fz , min, j = Fz , mean, j − 0.5 ⋅ ∆Fz (25)

The non-linear dependency of the shell force and the bolt force illustrated in Figure 11 result in the
maximum and minimum tensile bolt force being determined for each bin. The stress range in the bolt
is found according to Equation (26) to (28) for the minor cross section area of bolt, Abd3 considering
the reduction of the bolt diameter due to the thread.
∆Fb , j = Fb , max, j − Fb , min, j (26)

Fb,max, j Fb,min, j
Δσ b,max, j = Δσ b,min, j = (27)
A bd3 A bd3
 ∆σ b , j = ∆σ b , max, j − ∆σ b , min, j (28)

The sum of the fatigue damage in the bolt is based on the S-N approach in [9] under assumption of
linear cumulative damage (Palmgren-Miner rule):
j
ni
D = ∑( ) ≤ 1.0 (29)
i =1 Ni
As bending moments in the bolt are not considered, the detail category reference strength of the
bolt, 36* (40 MPa) is used for damage accumulation according to [5].
Available S-N curves are limited to nominal bolt sizes up to 30mm. The M64 bolts studied in the
present paper exceeds this value, hence the additional size effect leading to a reduction of the fatigue
strength, is to be considered. Multiplying the reduction factor, ks defined in Equation (30) with the
reference strength of the bolt account for this effect according to [5].
0.25
 30  (30)
k s =  
 d nom 
The obtained fatigue damage for the bolt exposed to the loads presented in the Markov matrix
result in a summarized fatigue damage in Equation (31).

∑D bolt = 0.83 ≤ 1.0 (31)

4.3. Numerical approach

The ULS and FLS capacity for the numerical L flange model are determined. The model includes the
flange with connected shell, bolt and washer. The equivalent normal force is defined as a variable
pressure load applied at the top of the TP shell.
4.3.1 Ultimate limit state: The maximum allowable tensile load, fult, according to the analytical
approach in Section 2.1 is applied as reference in the numerical analysis to obtain the corresponding
stresses and plastic strains. Using this approach result in the safety level inherent in the analytical
solution are ensured in the numerical segment model. Positions A-E in Figure 12 are selected for the
analysis as they represent the points of interest for failure mode B.
The calculated axial stresses in the bolt and the flange segment, for the segment force equal the
analytically derived ultimate resistance, are presented as contour plots in 0 and 0. Stresses and plastic
strains as a function of the segment shell force for positions A-E are given in Figure 15 and Figure 16.
The obtained maximum allowable plastic strains are listed in Table 8.
The development of plastic strain shown in Figure 16 indicates that the elastic limit is exceeded for
a tensile segment force equal to 1,100kN and consequently a overutilization according to Equation
(32).
F
URelastic = z ,elastic ,max = 1.16 (32)
Fult
Figure 12. L segment model geometry Figure 13. Axial stresses in Figure 14. Axial stresses in
incl. idealized MP/TP flanges, the bolt for F = F the shell and flange for
z ult
equivalent bolt diameter, identical bolt Fz = Fult
head and nut, and washers. Pos. A-E
indicate investigated positions.

Figure 15. Normalized axial stresses at pos. A-E Figure 16. Plastic strain at pos. A-E as function
as function of tensile shell segment force. of tensile shell segment force.

Table 8. Max. allowable plastic strains for positions A-E.

Position Max. Plastic strain [%]
A 0.003
B 0.088
C-D 0
E 0.026

4.3.2 Fatigue Limit State: The fatigue capacity of the bolt is investigated by a finite element analysis
(FEA) of the bolt center, positions C-D, and near the transition from bolt head to shank, position B
shown in Figure 12. The segment shell force is defined in the numerical model for a range of -1,000kN
to 1,500kN, covering the full load range of the Markov matrix.
 Figure 17. Output stresses for Fz = 600kN for a mesh size of 8mm and 10mm. Figure from [7].
The stresses obtained from the FEA at the bolt center and near the bolt head are extracted for the
defined segment load range and for two mesh configurations as illustrated in Figure 17.
The position-specific bolt stresses as functions of the shell force are utilized to determine the stress
range for each load bin in the Markov matrix according to (28). The stresses are hereby a function of
the obtained relation between the segment and bolt force, shown in Figure 18. Damage accumulation
is calculated equivalent to Section 2.2, except that a detail category reference strength of 50 MPa is
chosen. This is a consequence of the numerical model considers both axial and bending stresses. The
fatigue damage sum is given in Table 9.

Figure 18. Characteristic tensile bolt force as a function of the shell force in
the segment shell wall. The normal force range covers all load values
derived from the Markov matrix.

Table 9. Accumulated fatigue damage in the bolt for two element sizes
Fatigue damage Fatigue damage
(element = 8 mm) (element = 10 mm)
Bolt head 0.36 0.04
Bolt Shank 0.041 0.028
5. Comparison and discussion
The analytical and numerical results of the L flange segment are compared and discussed in the
following.

5.1. Ultimate limit state

The minimum derived tensile resistance in Table 4 is obtained for failure mode B. The computed
stresses shown in Figure 13 and Figure 14 confirm this behavior. As seen at the transition of the pile
wall to the flange, local yielding develops and subsequently load transfer via the inner edge of the
flanges at the limit tension force of the bolt. As a result positions A-E, defined in Figure 12, are
specified as hot spots for establishing a link between the tensile shell resistance and the corresponding
stresses and plastic strains. From the comparison of the analytically and numerically obtained results
in positions A-E the following is highlighted:
• Maximum stresses and corresponding development of plastic strains are obtained at point B in
Figure 15 and Figure 16. The presented result indicates that a more detailed study of the stresses
in this area is needed along the entire circumference.
• As the tensile segment force shown in Figure 15 increases the slope of the compressive stresses,
representing the prying force, at (pos. E) develop more rapidly. As result, the slope of the graphs
for the tensile stresses in the bolt (pos. B-D) becomes steeper. As the analytical approach does not
include any tolerances, the prying force presented in Figure 15 is seen as minimum value.
• The difference in stress found at the two sides of the bolt C-D in Figure 15 is seen to increase with
increasing tensile segment load. The difference is explained by the fact that the secondary
bending moment in the bolt is not considered in the analytical approach.
• The plastic strains obtained in Table 8 are the answer to an extended requirement that ensures the
integrity of the connection according to the analytical approach. As the analytical L flange
concept ignores the fact that opening of the connection and plastic deformation will lead to some
redistribution of forces, this requirement is to be maintained for verification of a full flange
model.

5.2. Fatigue limit state

The numerically determined accumulated fatigue damage sum in the bolt is found by the numerical
approach to be approximately 50% lower than the analytically obtained value. Furthermore, the
numerical model indicates that the maximum fatigue damage in the bolt occurs in the bolt shank near
the bolt head. The difference between the analytically and numerically obtained fatigue damage is seen
to be caused by:
• Figure 18 shows the numerically and analytically derived bolt forces as a function of the segment
shell force. For a given tensile shell force it can be seen that the numerically predicted bolt force is
significantly lower than the analytically predicted.
• The result shows the effect of the built-in fabrication tolerances in the Schmidt-Neuper method
which are not included in the numerical model.
• An initial study of the fabrication tolerances identifies the estimated contact surface between two
flanges as a crucial parameter for the slope of the curve.
Using the numerical model as an alternative to the analytical approach includes consideration of
following:
• The numerically obtained damage values are seen to be very sensitive to the mesh size. In order to
directly apply the stresses obtained in the numerical model further investigations are required. In
addition, the impact of simplifying the model bolt geometry should be clarified.
• The numerical procedure is as consequence found only to be applied for specification of the
relation between segment shell force and the bolt force. Following this approach the integrity of
the bolt can be verified for fabrication tolerances exceeding the values included in the Schmidt-
Neuper method or sought optimized by consideration of the full flange model.
6. Conclusion
An analytical and numerical ULS and FLS examination of the L flange segment is presented as the
basis for the ongoing development work on the design of the bolted MP/TP steel ring flange
connection. The L flange segment is investigated due to the availability of analytical approaches
making it possible to establish numerical criteria applicable to a full flange model. The analytically
determined failure mode B in the analytical ULS determination is confirmed by the identical
numerical model leading to the conclusion that the derived tensile resistance can be applied to specify
plastic strain criteria. However, as the analytical approach does not consider tolerances, the tolerances
influence on the plastic strain values is required before the criteria can be applied to a full flange
model. Fabrication tolerances are included in the analytical FLS determination by the trilinear curve
relating the segment force with the tensile bolt force. Comparison of the analytically and numerically
determined accumulated fatigue damage shows the damage found by the numerical approach (which
does not include tolerances) too low. Furthermore, the local obtained stress values in the numerical
model are very sensitive to the choice of element size as well as the level of detailing of the bolt. The
numerical FLS approach is therefore suggested to be preferred for determining the relation between
the segment load and the tensile bolt force. Implementation of the numerical FLS model to a full
flange model will require that sufficient fabrication tolerances can be included in the numerical model.
The presented results indicate that focus for future development work must include handling
fabrication and installation tolerances in the numerical models.

References
[1] Wind Europe 2017 The European Offshore Wind Industry – Key Trends and Statistics 2016 vol
1, ed Pineda I and Tardieu P (Belgium: Wind Europe) p 7
[2] Vries D 2010 Offshore Monopile Failure – A solution may be in sight Windpower [Online].
Available: http://www.windpowermonthly.com/article/1011507/offshore-monopile-failure---
solution-may-sight
[3] Offshore Wind Industry 2016 Waiting for new tenders number 2, ed Buddensiek V (Germany:
Offshore Wind Inustry) p 24
[4] Damiani R R 2016 Offshore Wind Farms: Technologies, Design and Operation vol 1, ed Chong
Ng and Ran L (United Kingdom: Woodhead Publishing) chapter 10 pp 263-357
[5] IEC 61400-6 2016 Wind Energy Generation Systems – Part 6: Tower and Foundation Design
Requirements
[6] Petersen C 2013 Grundlagen der Berechnung und baulichen Ausbildung von Stahlbauten vol 4,
ed Harms R and Prenzer A (Munich: Springer Vieweg) chapter 16 pp 1165-1260
[7] Thage K J 2017 Investigation of Flange Connection Capacity for Offshore Wind Turbine
Foundations M.Sc. Eng. Thesis (Lyngby: Technical University of Denmark)
[8] DiBt 2012 Richtlinie für Windenergieanlagen Einwirknungen und Standsicherheitsnachweise
für Turm und Gründung
[9] GL COWT 2012 Guideline for the Certification of Offshore Wind Turbines, Rules and
Guidelines, IV-Industrial Services, Part 2, Guideline for the Certification of Offshore Wind
Turbines
[10] VDI 2230-1 2015 Systematic Calculation of High Duty Bolted Joints – Joints with One
Cylindrical Bolt
[11] DASt 021 2013 Schraubenverbindungen aus Feuerverzinkten Garnituren M39 bis M72
entsprechend DIN 14399-4, DIN EN 14399-6
[12] DNVGL-RP-C208 2016 Determination of Structural Capacity by Non-linear Finite Element
Analysis Methods
[13] Hobbacher A F 2016 IIW-2259-15 2016: Recommendations for Fatigue Design of Welded
Joints and Components vol 2, ed Mayer C (Switzerland: Springer International Publishing)
chapter 10 pp 113-138