Automation in Construction 61 (2016) 98–111

Contents lists available at ScienceDirect

Automation in Construction

journal homepage: www.elsevier.com/locate/autcon

Simulation of load lifting with equalizers used in shipyards

Seung-Ho Ham a, Myung-Il Roh b,c,⁎, Hyewon Lee a
a
Department of Naval Architecture and Ocean Engineering, Seoul National University, Seoul, Republic of Korea
b
Department of the Naval Architecture and Ocean Engineering, Seoul National University, Seoul, Republic of Korea
c
Research Institute of Marine Systems Engineering, Seoul National University, Seoul, Republic of Korea

a r t i c l e i n f o a b s t r a c t

Article history: One of the most commonly used pieces of equipment for lifting a load is the equalizer of a crane in shipbuilding
Received 2 May 2015 production. The equalizer distributes the tension in wire ropes between the load and equipment equally. While
Received in revised form 30 September 2015 the equalizer is composed simply of several fixed and moving pulleys, the mechanism of an equalizer is not easy
Accepted 16 October 2015
to adapt to physics-based simulations because only one wire rope is coiled around all of the pulleys. Thus, instead
Available online 11 November 2015
of the real mechanism of the equalizer, an alternative mechanism, such as stretching or shortening the initial
Keywords:
lengths of the wire ropes, logically, has been used in previous simulations.
Equalizer In this study, a mass-less constraint-based wire rope is adopted to realize the real mechanism of the equalizer.
Constraint-based wire rope The constant length constraint is used to realize one wire rope among the fixed and moving equalizer pulleys.
Discrete Euler–Lagrange equation We derive a discrete Euler–Lagrange (DEL) equation to represent the motion of a multibody system including
Multibody dynamics the constraint wire ropes. The DEL equation is shown to be numerically stable for the case of linear holonomic
Shipbuilding production simulation constraints, such as the constant length constraint.
Crane We test several series of fixed and moving pulleys and model the equalizer based on the real mechanism. Finally,
the equalizer based on the real mechanism is applied to two typical lifting cases in shipbuilding production. One
is lifting a load using a Goliath crane with three equalizers and the other is to use a floating crane with one equal-
izer and two crawler cranes. As a result, the tensions in the wire ropes connected to the load are adjusted equally
as the position of the moving pulleys and the orientation of the equalizer change.
© 2016 Elsevier B.V. All rights reserved.

1. Introduction
1.1. Research background
Ships and offshore plants are divided into several pieces, called
blocks. Basically, they are built by erecting each of these blocks, one by
one. In this study, a block can be regarded as a kind of weight and
thus the term ‘load’ is used for the block. There are many cranes, such
as a Goliath crane and a floating crane, to lift loads in shipyards. A load
has several lifting points that can be lifted, whereas cranes, such as the
floating crane and the Goliath crane, have only limited hooks. Thus, a
special device called by an equalizer is introduced in shipbuilding
production. The first purpose of the equalizer is to connect the lifting
points on the target load with the crane, as shown in Fig. 1.
Because a single load can weigh more than 100 tons, the concentrat-
ed load at one lifting point can cause severe accidents, such as being torn
⁎ Corresponding author at: Department of the Naval Architecture and Ocean
Engineering, and Research Institute of Marine Systems Engineering, Seoul National
University, 1 Gwanak-ro, Gwanak-gu, Seoul 08826, Republic of Korea. Tel.: +82 2 880
7328.
E-mail addresses: hsh0930@snu.ac.kr (S.-H. Ham), miroh@snu.ac.kr (M.-I. Roh),
gpdnjs0215@snu.ac.kr (H. Lee).
out and dropping the load on the ground. Thus, a second purpose of the
equalizer is to distribute tensions to all wires equally. This is done by
several fixed and moving pulleys, which are components of the equaliz-
er and are connected by one wire rope, as shown in Fig. 2.
For a load lifting simulation in shipbuilding, the equalizer should be
modeled properly. However, despite its seemingly simple combination
of pulleys with one wire rope, it is not easy to realize its real mechanism.
This is mainly because of the wire rope, which should be bent around
the sheaves of the pulleys without disconnection. Thus, it is important
to study how to model the wire rope and pulleys for load lifting simula-
tion in shipbuilding.
1.2. Related studies
There have been several related studies about physics-based simula-
tions of shipbuilding production, such as lifting and turn-over.
Cha [1] and Cha et al. [2] adopted multibody dynamics to formulate
the motion of a floating crane and a heavy cargo with full degrees of
freedom. To calculate the tension in the wire ropes between the floating
crane and the cargo, the wire rope was modeled as incompressible
springs. They manually derived the equations of motion. Thus, if the
models or constraints change, the equations should be derived again.
This took much effort to obtain the final form of the equations, because

http://dx.doi.org/10.1016/j.autcon.2015.10.007
0926-5805/© 2016 Elsevier B.V. All rights reserved.
 S.-H. Ham et al. / Automation in Construction 61 (2016) 98–111
Floating
Crane
Hook
Equalizer
Load (Block)
Fig. 1. Load lifting operation using an equalizer of a crane.
the position vectors should be differentiated twice. To overcome this
limitation, Cha et al. [3] used a topological modeling approach to formu-
late the equations of motion automatically, by considering connectivity
between the two bodies. The idea was to find the velocity transforma-
tion matrix from the multiplication of separate sub-matrices, such as
the transformation matrix, connectivity matrix, and joint characteristic
matrix.
The studies mentioned above did not use equalizers in their applica-
tions. They were focused mainly on the equations of motion and the for-
mulation procedures rather than implementation of an equalizer. Thus,
they used incompressible springs as wire ropes for lifting a load.
Cha et al. [4] introduced an approximated mechanism to model an
equalizer. They had no choice but to stretch or to shorten the initial
length of the wire ropes between the equalizer and load logically
Fixed
pulleys
Moving
pulleys
Fig. 2. Components of the equalizer.
99
Goliath Crane
Hook
Equalizer
because wire ropes were implemented as incompressible springs. The
equalizer modeled by the approximated mechanism is shown in Fig. 3.
This may cause several problems. First, the simulation may be nu-
merically unstable due to a sudden change in the length. If the wire
ropes have their real spring coefficients, which are usually very high, a
small variation on one side can cause a large tension. By a length adjust-
ment algorithm, the length of a wire rope that gets a large tension is
shortened and the others should be extended in the next step. Some
of the shortening parts now start to get larger tensions. Thus, the ten-
sions of the wire ropes are repeatedly increased or decreased and finally
induce unstable motion of the system. Second, although the equalizing
is controlled naturally by the pulleys and wire rope, the approximated
mechanism should use an artificial factor that controls the length
adjustment ratio. The shortening length is the same as the total extend-
ed lengths of the other wire ropes. If an artificial factor is equal to 1, the
shortening length is calculated so as to achieve no tension in it. If an
artificial factor is less than 1, the shortening length is set to leave
some tension in it. Thus, the time required for equalizing is controlled
by the artificial coefficient. This can affect the motion of the load before
it is lifted or during lifting. Third, although the total length that can be
extended is limited, the approximated mechanism cannot control the
total length of the equalizer. Moreover, the approximated mechanism
does not calculate the motions of the pulleys of the equalizer. It is also
not possible to connect another pulley at the end of the wire rope for ex-
tension, as shown in Fig. 4.
Some researchers have tried to model the wire ropes and pulleys
mathematically, considering real phenomena. Imanishi et al. [5] pre-
sented a wire rope involving contacts with a winch drum using a finite
element method. Lugris et al. [6] used a semi-analytical method, which
was derived by introducing an analytical model of the rope–sheave in-
teraction in the system. García-Fernández et al. [7] divided the wire
rope into spring segments and calculated the tensions as external forces
in the system. They applied their wire rope to represent exact pulleys.
However, they did not consider in their study how to represent an
equalizer mechanism. Moreover, the constraint-based wire rope is a
simpler way to model the equalizer compared with other methods in-
troduced in previous studies.
1.3. Contribution
The purpose of this study was to model the real mechanism of the
equalizer with fixed and moving pulleys and to apply it to a load lifting
simulation. By developing the equalizer with a real mechanism, it will
prevent predictable problems in advance. Moreover, the simulation
will show more accurate results, especially in terms of the wire rope
100 S.-H. Ham et al. / Automation in Construction 61 (2016) 98–111
Real Mechanism
f0 fi fn
Fig. 3. Approximated mechanism for the equalizer.
tension and the motion of the equalizer and the suspended load. The
motion of the moving pulleys of the equalizer can be also calculated.
In previous studies, the wire ropes [3,4] were modeled as incom-
pressible springs, which is not suitable for modeling the pulleys. Thus,
we used a constraint-based wire rope, as introduced by Servin and
Lacoursière [8]. The constraint-based wire rope represents a kinematic
constraint among the connected objects as the initial length of the
wire rope. When the wire rope with a large spring coefficient is short-
ened, it may cause unstable motion of the system. Thus, we adopt the
discrete Euler–Lagrange (DEL) equation to formulate the equations of
motion. This makes the numerical integration more stable, even though
wire ropes have a large spring coefficient.
This paper is organized as follows: Section 2 explains the real mech-
anism of the equalizer; Section 3 then derives the mathematical model-
ing of the motion equations, based on the DEL equation, and the pulleys
and equalizer based on the constraint-based wire rope; To verify the
Extension Pulleys
Extension Pulleys
Fig. 4. Extension pulleys under the equalizer.
Approximated Mechanism
f0 fi fn
pulleys and the equalizer, Section 4 investigates several test cases;
Section 5 presents two kinds of load lifting simulations, using a Goliath
crane and a floating crane; Section 6 presents our conclusions and sug-
gests future work.
2. Real mechanism of the equalizer
Fig. 5 shows how the fixed and moving pulleys are assembled.
The equalizer is connected under the hook of the crane. Inside the
equalizer, several fixed and moving pulleys are placed by turns. All of
the pulleys are connected by a single wire rope, which should have
the same tension anywhere in its length. Thus, if the tension is denoted
by T, all of the moving pulleys are exerted by 2 T from the equalizing
wire rope.
Fig. 6 shows how the wire rope tensions connected to the moving
pulleys and the load are adjusted. It is assumed that only three moving
pulleys, p0, p1, and p2, are connected and their tensions are denoted
by f0, f1, and f2, respectively. At the first time point, f1 is acting on the
middle moving pulley p1; however, the others are still zero (Fig. 6(a)).
As p1 is the only moving pulley where a force is exerted, p1 starts to
move down. Due to the equalizing wire rope, the other two pulleys –
p0 and p2 – move up simultaneously. Sometime later, f0 which is larger
than f1 is acting on p0. To make the tension equal, p0 moves down and
the other two pulleys move up (Fig. 6(b)). As f1 is larger than f2, the
moving distance of p2 is larger than that of p1. Until all of the wire
rope tensions – f1, f2, and f3 – are exactly same, the pulleys are repeated-
ly moving up and down. Finally, they stop moving when all the wire
ropes have the same tension (Fig. 6(c)).
3. Mathematical modeling
The core technology to realize the mechanism of the equalizer
depends on the wire rope modeling. It is closely related with the equa-
tions of motion. We adopted the discrete Euler–Lagrange equation, as
derived by Wendlandt et al. [9], Marsden et al. [10], and Lew [11]. This
equation makes the numerical integration more stable, even though
wire ropes have a large spring coefficient.
3.1. Discrete Euler–Lagrange equation
The variational principle was introduced in Fowles et al. [12]. To com-
pute a discrete trajectory of a body, the concept of virtual displacement
and virtual work can be considered. According to Hamilton's principle
which means the expenditure of energy in the system during the
 S.-H. Ham et al. / Automation in Construction 61 (2016) 98–111 101

Hook

Equalizer Equalizing
Fixed Pulley Wire Rope

T T

Moving Pulley

Fig. 5. Internal structure of the equalizer.

(a) (b) (c)

p2
p0 p2
p0 p1 p2 p1
p1 p0

f0 f2 f2
f0 f1 f2 f1
f1 f0

f1 > f 0 = f 2 = 0 f 0 > f1 > f 2 = 0 f 0 = f1 = f 2

Fig. 6. Procedure of the adjustment of tensions acting on wire ropes through the equalizer.

motion, the action integral J can be defined as follows: follows:

Z tN Z tN
J¼ Ldt ¼ ðT−V Þdt ð1Þ X
t0 t0
N−1  
Jd ¼ Ld qk ; qkþ1 ; h ð4Þ
k¼0
where T and V are kinetic and potential energy of the particle, and L is
called the Lagrangian. Hamilton's variational principle states that during
a time interval [t0 − tN], the actual motion follows the trajectory that where qk is the position of the particle at time t1 + kh. As mentioned
minimizes the action integral J. This can be expressed as follows: above, according to the variational principle, the particle moves along
Z the trajectory where δJ is zero. The start point and end point should be
tN
δJ ¼ δ L dt ¼ 0 ð2Þ fixed. Thus, the virtual displacements of a particle at the start and end
t0

where δ represents a variation, which is the change in the value of the

Body1 Body1
function f resulting from a change in any or all of its arguments, except
for time t. The variation of generalized coordinate q at time t = t0 and rA A rA A
t = tN should be zero. From Eq. (2), the Euler–Lagrange equation can Y Y
be written as:
  X
ltotal X
ltotal
d ∂L ∂L O O
− ¼ 0: ð3Þ
dt ∂ q ∂q


rB B rB
The action integral J of the Lagrangian can be calculated with the sum Body2
B
Constraint
of the infinitesimal area divided into the time step h. Approximating Body2
Violation


each infinitesimal area as a rectangular shape, and velocity qk as

qkþ1 −qk
h
, the discrete action integral of Lagrangian Jd can be expressed as Fig. 7. Constraint-based wire rope with a single element.
102 S.-H. Ham et al. / Automation in Construction 61 (2016) 98–111

 D2
A 2 A
D D2
E J H E D

K
la lb lc Body3
Body1
11 [kg] C B
Body2
D1 G F 20 [kg]
B C F 20 [kg]
Fig. 10. Modeling of moving and fixed pulleys with a wire rope constraint.
 D1
2 follows:

 
D2 Ld ðqk−1 ; qk ; hÞ þ D1 Ld qk ; qkþ1 ; h ¼ 0 ð6Þ

Constraints are kinematic restrictions on the variables in the system.

Fig. 8. Fixed and moving pulleys using the constraint-based wire rope.
Each body is connected by joints or wire ropes, and they limit the mo-
tion of bodies into certain areas. Considering a system with constraint
m
points are zero. Thus, the formula can be expressed as follows: equations ∑ g j ðqÞ ¼ gðqÞ ¼ 0, we can obtain a modified Lagrangian
j¼1
as follows:
X
N −1  
δJ d ¼ δLd qk ; qkþ1 ; h
k¼0 X
m

X −1 L ¼ L þ LC ¼ L þ λ j g j ðqÞ ¼ L þ λT gðqÞ ð7Þ

N  
¼ D2 Ld ðqk−1 ; qk ; hÞ þ D1 Ld qk ; qkþ1 ; h δqk ¼ 0
k¼1
where, Di is the partial differential operator, which means partial
differentiation by the ith variable. Eq. (5) is always satisfied regardless
of δ qk. Thus, Eq. (5) yields the discrete Euler–Lagrange equations as
ð5Þ j¼1
where λj is a Lagrange multiplier of the jth constraint. The discrete
Euler–Lagrange equation for the system with constraint equations
can be obtained directly from the modified Lagrangian. Through a
discretization process, the discrete Euler–Lagrange equation with

l1 l1 li −1 li −1 li li li +1 li +1 ln ln
m m m m m

f1 f i −1 fi f i +1 fn

Fig. 9. Arrangement of fixed and moving pulleys of the equalizer.

 S.-H. Ham et al. / Automation in Construction 61 (2016) 98–111 103

constraints, is determined as follows:

 
D2 Ld ðqk−1 ; qk ; hÞ þ D1 Ld qk ; qkþ1 ; h þ!hGTk λk ¼ 0
 T
∂g 1 ∂g 2 ∂g m ð8Þ
GTk ¼ ⋯
∂qk ∂qk ∂qk

where GTk is differentiation of constraints by variables. Moreover, using

the Stömer–Verlet method [13], we set the discrete Lagrangian to be
Eq. (9):

1 qk −qk−1 T qk −qk−1
Ld ðqk−1 ; qk ; hÞ ¼ h M −V ðqk Þ ð9Þ
2 h h

where, M is mass matrix. By substituting Eq. (9) into Eq. (8), the follow- 3.3 m
ing equation is obtained:
  
1 q −2qk þ qk−1 ∂V
h − M kþ1 − þ hGTk λk ¼ 0: ð10Þ
h h ∂qk

Using a Taylor series expansion, the constraint equations yields: 8m

   
g qkþ1 ¼ gðqk Þ þ GTk qkþ1 −qk ¼ 0 ð11Þ
1.6 m
qk + 1 and λk are the unknowns. From Eqs. (10) and (11), we can obtain
the matrix form:
Sheave Diameter 5m
2 3 = 0.8 m
  2 ∂V
M −GTk qkþ1 4 Mð2qk −qk−1 Þ−h
2 ¼ ∂qk 5: ð12Þ
Gk 0 h λk
−gðqk Þ þ Gk qk

Eq. (12) is an ill-posed problem, which means it can have many so-
lutions or violate constraints due to the error of approximation. To solve
this, Baumgarte [14] and Eich et al. [15] introduced the regularization
method, which simply adds the product of the Lagrange multiplier λj
and error εj to the constraint equation:
Fig. 12. Modeling of the equalizer.
ελ
0 þ g2ðqÞ ¼ 0 31
ε1 0
B 6 7C 3.2. Constraint-based wire rope
Bε ¼ 6 ε2 7C: ð13Þ
@ 4 ⋱ 5A
0 εm As shown in Fig. 7, Body1 and Body2 are connected to each other
with a wire rope. To simulate stretching with a constraint-based wire
rope, the length of the wire rope should be equal to the constant value
Using a Taylor series expansion, Eq. (13) yields:
ltotal. If the distance between the wire connection points is larger than
ε 2   ltotal, this means that the constraint is violated. If Body2 violates the con-
h λk þ gðqk Þ þ GTk qkþ1 −qk ¼ 0: ð14Þ
h
2 straint, the constraint force is acting on the Body2, which is equal to the
spring force.
Finally, we obtain the regularized DEL equation with constraints in Then, the constraint equation of stretching is given by:
matrix form by substituting Eq. (14) into Eq. (12):
2 3 2 3 g ðqÞ ¼ jrA −rB j−ltotal ¼ 0 ð16Þ
M −GTk  q Mð2qk −qk−1 Þ−h
∂V 2
4 ε 5 2kþ1 ¼ 4 ∂qk 5: ð15Þ where, rA and rB are position vectors from the origin O to point A and B,
Gk h λk
h
2 −gðqk Þ þ Gk qk respectively. The position vector can be expressed using the rotation

t=0.0 t=1.0 t=3.0 t=5.0

Fig. 11. Simulation result of moving and fixed pulleys with the constraint-based wire rope.
104 S.-H. Ham et al. / Automation in Construction 61 (2016) 98–111

: 23 m
: 26 m
Hook
60 m
Equalizer

10 m

30 m
((-2.5,2.5) (2.5,2.5)

10 m
(0,0)

(-2.5,-2.5) (2.5,-2.5)

Load 300 ton

(10 x 10 x 5)

Fig. 13. Modeling of the test case and connections between the equalizer and the load.

Fig. 14. Simulation result: Motion of the equalizer and the load.
 S.-H. Ham et al. / Automation in Construction 61 (2016) 98–111 105

Fig. 15. Simulation result: Motion of the sheaves in the equalizer.

matrix and the position vector of the center of gravity. Thus, the con- 3.3. Modeling of pulleys
straint equation of the wire rope can be rewritten as:
A constraint-based wire rope can be applied to make fixed and mov-
g ðqÞ ¼ rG1 þ R1 rA=G1 −rG2 −R2 rB=G2 −ltotal ¼ 0 ð17Þ ing pulleys. Fig. 8 shows an example of fixed and moving pulleys using
the constraint-based wire rope. If the total length is denoted as ltotal, the
where, rG1 and rG2 are the position vectors and R1 and R2 are the rota- total length is equal to the sum of the segments between the fixed and
tion matrixes of Body1 and Body2, respectively. rA/G1 means the vector moving pulleys, and the circumference of the pulleys.
from G1 to A and rB/G2 means the vector from G2 to B. If we let l ¼ rG1 þ Thus, the constraint equation of the wire rope connecting the fixed
RG1 rA=G1 −rG2 −R G2 rB=G2 , and n ¼ jljl , we obtain the derivation of the con- and moving pulleys in the above example is given by:
straint equation as follows:  
πD1 πD2
g ðqÞ ¼ la þ lb þ lc þ þ −ltotal
 2 2 
     πD1 πD2
¼ jrA −rB j þ jrC −rD j þ jrE −r F j þ þ −ltotal ¼ 0: ð19Þ
G ¼ n − n  rA=G1 −n n  rB=G2 : ð18Þ 2 2

Wire rope tension

[ton]

Time
[sec]

Fig. 16. Graphs of wire rope tensions between the load and the equalizer.
106 S.-H. Ham et al. / Automation in Construction 61 (2016) 98–111

Lower Goliath Crane Lower Trolley

Upper
Trolley Trolley
Upper
Trolley

Equalizer Equalizer

Load
Dock

Load (43 x 23 x 11, 350 ton)

Fig. 17. Modeling of load lifting using the Goliath crane.

4. Test cases
Equalizer1
Equalizer3 This section presents two test cases to validate the mathematical
model explained in the previous section. One involves several fixed
and moving pulleys, and the other is an equalizer.
Equalizer2
4.1. Pulleys

This test case uses pulleys with a constraint-based wire rope. The
purpose of this case is to show that the constraint-based wire rope
can model fixed and moving pulleys, as explained in Section 3.3.

4.1.1. Modeling
Fig. 10 shows two fixed pulleys and two moving pulleys. One end of
a wire rope is attached to the fixed body, and the other is attached to a
Fig. 18. Wire ropes connected between the load and three equalizers.
box of 11 kg weight.

4.1.2. Formulation
3.4. Modeling of the equalizer From Eqs. (17) and (18), we obtain the constraint equation g and its
derivative G of the wire rope constraint as follows:
The equalizer can be realized by a combination of several fixed and
moving pulleys, as shown in Fig. 9.
gk ¼ jrA −rB j þ jrC −rD j þ jrE −r F j þ jrG −rH j þ r J −rK −ltotal ¼ 0 ð21Þ
If there are a number of n moving pulleys with a same diameter
D, the constraint equation of the equalizing rope can be written as a
given equation: 2 3
 −nAB þ nCD 
6 nAB  rB=G1 − nCD  rC=G1 7
6 7
6 −nEF þ nGH 7
G ¼6
T  7 ð22Þ
! 6 nE F  r F=G2 − nGH  rG=G2 7
X
m
πD 6 7
g ðqÞ ¼ 2li þ m −ltotal ¼ 0: ð20Þ 4 −n JK 5
i¼1
2 n JK  rK=G3

1 Equalizer1 2
Equalizer3

Equalizer2

Fig. 19. Height change of pulleys of the equalizer during the simulation.
 S.-H. Ham et al. / Automation in Construction 61 (2016) 98–111 107

1 2 3

4 5 6

Fig. 20. Simulation procedure of load lifting using the Goliath crane.

Wire rope tension (Equalizer 1)

[ton]

Time [sec]

Wire rope tension (Equalizer 2)

[ton]

Time [sec]

Wire rope tension (Equalizer 3)

[ton]

Time [sec]

Fig. 21. Graphs of wire rope tensions between the load and each equalizer.
108 S.-H. Ham et al. / Automation in Construction 61 (2016) 98–111

where, rB/G1 is the vector from the center of gravity of Body1 to point B
r −r
in the inertial frame, and ni j ¼ jrii −r jj j is the normal vector from the
point i to the point j.

4.1.3. Simulation result

Because the weight of each pulley is set to 20 kg, the box starts 3,600 ton capacity
to move downward, whereas the moving pulleys move upward, as floating crane
shown in Fig. 11.

4 Equalizers
4.2. Equalizer 7.5 [m]

60 [deg]
This test case shows the modeling of the equalizer and the wire rope
tension adjusted equally by the real mechanism, as explained in
Section 2.

4.2.1. Modeling Load

(34m x 27m x 11m, 1,000 ton)
By adding more fixed and moving pulleys, we can construct an
equalizer. For example, an equalizer with five moving pulleys is
modeled, as shown in Fig. 12. The hook and the equalizer are connected
Fig. 22. Modeling of load lifting using a floating crane.
by hinge joints, where the rotating axis is aligned to the vertical vector.
The sheave diameter of the moving pulleys is 0.8 m. The main dimen-
sions of the equalizer are indicated in Fig. 12. 5.1.2. Simulation results
To test the mechanism of the equalizer, a simple load was lifted with Fig. 19 shows the pulley height change of the equalizer during the
one equalizer. Fig. 13 shows the modeling of the equalizer test case and simulation. At the first time point, the heights of the equalizer moving
connections between the equalizer and the load. The load is assumed to pulleys are at the same level. However, to control the wire rope tension
be a 10 × 10 × 5 m box, with a weight of 300 ton. To check the equalizing equally, the pulleys are moved up and down.
process, the initial length of each wire rope is set at 23 m or 26 m. Simulation procedure of load lifting using the equalizer is summa-
rized in Fig. 20. From Fig. 20 (1) to (3), the three equalizers move to
their equilibrium positions and orientations. From Fig. 20 (4) to (6),
4.2.2. Simulation result
the load lifts off the ground.
The motion of the equalizer and the load are shown in Fig. 14. Be-
cause ‘Line1’ and ‘Line5’ are shorter than the others, wire rope tensions
are acting on the sheaves connected by ‘Line1’ and ‘Line5’ at the first
time point. Due to these tensions, the sheaves start to move up and
down. The motions of the sheaves are enlarged in Fig. 15. In addition, Equalizer3
we can also see that the equalizer is rotated due to the hinge joint,
which is used to connect the equalizer and the hook.
Fig. 16 shows the wire rope tensions of ‘Line1’, ‘Line2’, ‘Line3’, ‘Line4’, Equalizer1
and ‘Line5’. If one of the tensions is larger or smaller than the others, the
length is shortened or extended to adjust the tension equally. Finally, all
of the wire rope tensions become equal. Equalizer4
5. Load lifting simulation

This section presents the two load lifting simulations using a Goliath
crane and a floating crane, which are commonly used in shipbuilding.
Equalizer2

5.1. Load lifting simulation using a Goliath crane

The first application is load lifting simulation using a Goliath crane.

The Goliath crane is gantry-type crane used to erect loads from a pre-
erection area to the dock [16]. It is widely used in shipyards.

5.1.1. Modeling
Fig. 17 shows the modeling of the load, the dock, and the Goliath
crane. At the top of the crane, there are lower and upper trolleys.
Three equalizers are suspended by the trolleys. The load model was
exported from ship CAD data, including hull plates and outfitting.
Its size is 43 × 23 × 11 m (length × breadth × height) and its weight
is 350 ton.
We assumed that the equalizer has five connection points, like the
equalizer in Section 4.2. In total, 15 connection points on the load are
connected to three equalizers, as shown in Fig. 18. Fig. 23. Wire ropes connected between the load and four equalizers.
 S.-H. Ham et al. / Automation in Construction 61 (2016) 98–111 109

1 2 3

Trim by stern
Even keel

4 5 6

Fig. 24. Simulation procedure of load lifting using a floating crane.

The wire rope tensions between the load and each equalizer are
shown in Fig. 21. The five lines of the wire ropes in the graphs almost
overlap. If one of the tensions goes high, the others are also increased
in a moment. Thus, the wire rope tensions become similar even though
they show oscillating results during lifting the load.
5.2. Load lifting simulation using a floating crane
The second application is a load lifting simulation using a floating
crane. As the weight and size of erected loads and modules of offshore
project have increased, floating cranes have recently been used to lift
heavy loads, exceeding 1000 tons.
5.2.2. Simulation result
The simulation procedure of load lifting by the floating crane
is shown in Fig. 24. At the beginning of the simulation, the floating
crane is trimmed by the stern to be on even keel after lifting the load
(Fig. 24 (1)). While the wire ropes are hoisted up, the attitude of the
floating crane is set to be almost on an even keel (Fig. 24 (2), (3)). Final-
ly, the load is lifted off the ground (Fig. 24 (4), (5), and (6)).
Fig. 25 shows the equalizers before and after equalizing. Before
equalizing, the heights of the moving pulleys of the equalizers are at
the same level. However, after equalizing, the pulleys move up and
down to find equilibrium positions.
The wire rope tensions between the load and each equalizer are
shown in Fig. 26. The five lines of the wire ropes in the graphs almost
overlap as expected.

5.2.1. Modeling
Fig. 22 shows the modeling of the load and the floating crane. In this
simulation, a 3600 ton capacity floating crane with four hooks and
equalizers, possessed by many shipbuilding companies, was used. The
load model was also exported from ship CAD data including hull plates
and outfitting. Its size is 34 × 27 × 11 m (length × breadth × height), and
its weight is 1000 ton.
We used the same equalizers as in the previous example. In total,
20 connection points of the load are connected to four equalizers, as
shown in Fig. 23.
6. Conclusions and future works
In this study, we presented the real mechanism of an equalizer,
which is operated by fixed and moving pulleys sequentially connected
by a single equalizing wire rope. For the physics-based simulation of
load lifting, the DEL equation with constraint equations was briefly
derived. Using the stretching constraint of the wire rope, a constraint-
based wire rope was adopted. The constraint equation of the constraint-
based wire rope is such that the summation of the segments is equal to

Before equalizing After equalizing

Fig. 25. Position change of the equalizers before and after equalizing.
110 S.-H. Ham et al. / Automation in Construction 61 (2016) 98–111

Wire rope tension

of Equalizer 1
[ton]

Time [sec]

Wire rope tension

of Equalizer 2
[ton]

Time [sec]

Wire rope tension

of Equalizer 3
[ton]

Time [sec]

Wire rope tension

of Equalizer 4
[ton]

Time [sec]

Fig. 26. Graphs of wire rope tensions between the load and each equalizer.
 S.-H. Ham et al. / Automation in Construction 61 (2016) 98–111
the total length. We extended its usage to model the pulleys and eventu-
ally an equalizer.
To verify the mathematical model of the pulleys and the equalizer
based on the DEL equation with constraint-based wire ropes, pulleys
and an equalizer were tested. The results showed the expected motions
of the sheaves. Finally, we applied it to load lifting simulations using a
Goliath crane and a floating crane.
In future, the DEL equation and constraint-based wire rope that we
have presented need to be extended, to be fully useful in real-world ap-
plications, including subsea simulation [17]. To simulate cases that have
not been tried before due to the absence of mathematical modeling, we
will integrate more functionalities, such as contact between the load
and the wire rope, bending or twisting of the wire rope, and hydraulic
systems. Additionally, the simulation results will be compared with
real operations to validate the program in further research.
Acknowledgements
This work was partially supported by
a) BK21 Plus, Education & Research Center for Offshore Plant Engineers
(COPE) of Seoul National University, Republic of Korea
b) Engineering Research Institute of Seoul National University, Repub-
lic of Korea,
c) Research Institute of Marine Systems Engineering of Seoul National
University, Republic of Korea, and
d) Engineering Development Research Center (EDRC) funded by the
Ministry of Trade, Industry & Energy (MOTIE), Republic of Korea.
References
[1] J.H. Cha, Nonlinear dynamic response and discrete event/discrete time simulation of
a heavy cargo suspended by a floating crane(Ph.D. Thesis) Seoul National University,
Seoul, Korea, 2008.
111
[2] J.H. Cha, K.Y. Lee, M.I. Roh, Dynamic response simulation of a heavy cargo suspended
by a floating crane based on multibody system dynamics, Ocean Eng. 37 (14-15)
(2010) 1273–1291.
[3] J.H. Cha, S.H. Ham, K.Y. Lee, M.I. Roh, Application of a topological modelling ap-
proach of multi-body system dynamics to simulation of multi-floating cranes in
shipyards, Proceedings of the Institution of Mechanical Engineers, Part K Journal
of Multi-body Dynamics, 224(4) 2010, pp. 365–373.
[4] J.H. Cha, M.I. Roh, K.Y. Lee, Integrated simulation framework for the process plan-
ning of ships and offshore structures, Robot. Comput. Integr. Manuf. 26 (5) (2010)
430–453.
[5] E. Imanishi, T. Nanjo, T. Kobayashi, Dynamic simulation of wire rope with contact, J.
Mech. Sci. Technol. 23 (2009) 1083–1088.
[6] U. Lugrís, J.L. Escalona, D. Dopico, J. Cuadrado, Efficient and accurate simulation of
the rope–sheave interaction in weight-lifting machines, Proceedings of the Institu-
tion of Mechanical Engineers, Part K Journal of Multi-body Dynamics, 225 2011,
pp. 331–343.
[7] I. García-Fernández, M. Pla-Castells, R.J. Martínez-Durá, Elevation cable modeling for
interactive simulation of cranes, Proc. ACM SIGGRAPH/Eurographics Symposium on
Computer Animation 2008, pp. 173–181.
[8] M. Servin, C. Lacoursière, Massless cable for real-time simulation, Comput. Graphics
Forum 26 (2007) 172–184.
[9] J.M. Wendlandt, J.E. Marsden, Mechanical integrators derived from a discrete varia-
tional principle, Phys. D 106 (1997) 223–246.
[10] J.E. Marsden, M. West, Discrete mechanics and variational integrators, Acta Numer.
10 (2001) 357–514.
[11] A.J. Lew, Variational time integrators in computational solid mechanics(PhD thesis)
California Institute of Technology, 2003.
[12] G.R. Fowles, G.L. Cassiday, Analytical Mechanics, Thomson Learning, 2005.
[13] F. Crivelli, The Störmer–Verlet Method, Swiss Federal Institute of Technology,
Zurich, 2008.
[14] J. Baumgarte, Stabilization of constraints and integrals of motions in dynamical sys-
tems, Comput. Methods Appl. Mech. Eng. 1 (1972) 1–16.
[15] E. Eich, M. Hanke, Regularization methods for constrained mechanical multibody
systems, ZAMM-J. Appl. Math. Mech. 75 (10) (1995) 761–773.
[16] S. Ha, N.K. Ku, M.I. Roh, H.J. Hwang, Multibody system dynamics simulator for pro-
cess simulation of ships and offshore plants in shipyards, Adv. Eng. Softw. 85 (2015)
12–25.
[17] J.H. Woo, J.H. Nam, K.H. Ko, Development of a simulation method for the subsea pro-
duction system, J. Comput. Des. Eng. 1 (3) (2014) 173–186.