Professional Documents
Culture Documents
net/publication/340597808
CITATIONS READS
4 3,292
4 authors, including:
Dong Xu
Tianjin University
94 PUBLICATIONS 864 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
Study on meso-processes and mechanisms of local scouring around submarine pipeline and sand depositing near artificial seaweed under complicated conditions View
project
All content following this page was uploaded by Haiming Zhu on 01 June 2021.
Abstract
This paper focuses on the optimization of self-propelled modular transporter (SPMT)
load-outs. In common practice, a series of independent sub-systems are usually es-
tablished based on the idea of multi-tasking. However, the conventional strategy
features weak capability of finding globally optimal ballast solution thus has ineffi-
cient performance in actual applications. In this study, we developed a new ballast
strategy with a unified ballast system. The strategy dynamically allocates ballast
resources and compensates cargo loads as well as tide variation together by taking
globally optimized ballast actions. On the basis of the proposed strategy, we es-
tablished the approaches of ballast trials simulation, operation duration prediction
and feasibility assessment for load-out plans. Then a simulation-based optimiza-
tion method was developed to lower the time cost and prevent loading failures. In
demonstration cases, the proposed ballast strategy was tested and proved to be able
to reduce the time consumption by up to three quarters. The optimization method
was applied in finding appropriate start time, barge and pumping capacity for partic-
ular projects. The above-mentioned optimization solutions can enhance the existing
analysis workflow and help engineers make informed decisions.
Keywords: self-propelled modular transporter (SPMT), load-out, ballast strategy,
simulation-based optimization, ballast simulation
1 1. Introduction
2 In ocean engineering industry, gigantic offshore structures, including jackets,
3 modules, topsides, etc., are built in shipyards and then transported to offshore in-
∗
Corresponding author.
Email address: dzf@tju.edu.cn (Zunfeng Du)
19 SPMT load-out is an emerging technique in recent years (Zheng and Wang, 2016).
20 The general process can be described as follows. Firstly, the cargo will be jacked up
21 by SPMT, secured with lashings, and then transported to load-out quay through
22 planned route. As shown in Fig. 3, the barge is usually moored with a perpendicular
23 angle to the quay. A set of ramps or link beams are used to bridge the gap. During
24 load-out operation, the SPMT is moved forward slowly until the cargo reaches the
25 specified position. Ballast actions are taken continuously through the process to keep
26 the barge at a constant level and equilibrium state. Then the operators will lower
27 the SPMT platform, lay the cargo onto prepared supporting structures, and perform
28 sea-fastening. Finally, the SPMT will be driven off the barge.
29 The enormous weight and dimension of cargo increases the difficulty and time
2
Fig. 2. Unit combination and suspension grouping.
3
34 2005; Kaup et al., 2016; Sinha et al., 2018; Guan et al., 2018). An allowable barge
35 deflection envelope can be obtained to show the maximum misalignment between
36 deck and quay that the carried structure can tolerate without overstressing (Piter
37 et al., 1989; Sircar et al., 1990; Kenney et al., 1994).
38 As for the ballast plan, the idea of multi-tasking is usually applied which implies
39 to allocate tanks and pumps to establish a series of independent ballast sub-systems.
40 E.g., Ferguson et al. (1983), Kenney et al. (1994), and Yang et al. (2005) divided the
41 tanks and pumps into four sub-systems to compensate for cargo weight, trimming
42 moment, heeling moment and tide variation, respectively. Fig. 4 shows an example
43 of tanks arrangement. In addition to those sub-systems, Dearing et al. (1985) also
44 designated a few tanks for hog correction and Piter et al. (1989) suggested using four
45 large compartments near the bow and stern to adjust the variation in behavior that
46 was different from prediction. However, there are some obvious disadvantages:
47 1) A tide-compensating sub-system is deployed to make the rest of ballast operation
48 tide-independent. Due to the low precision in estimating the overall operation
49 time, the tide-compensating sub-system is usually required to be capable for a
50 full tide cycle compensation, which leads to the necessity of using large ships.
51 2) Although the most efficient ballast actions for each sub-system will be taken,
52 the combined solution may not be optimal. E.g., if the operation takes place in
53 a rising tide condition, the tide-compensating sub-system will ballast the tanks
54 while the weight-compensating sub-system will suggest a de-ballast action. These
55 contradictory behaviors will result in a waste of time and ballast resources.
56 3) The pumps and tanks allocation is made statically based on worst-case conditions.
57 The capacity of certain sub-systems may not be fully utilized in normal scenarios.
58 To the authors’ knowledge, there are a limited number of studies concerning the
59 optimization of load-out plan. Kurniawan and Ma (2009) studied the method of
4
60 minimizing deflection and curvature of barge as well as maximizing ballast efficiency
61 by optimizing ballast plan. However, the proposed method considered neither the
62 pumping speed nor the variation of tide, making it hard to be applied in actual
63 projects. Silvianita et al. (2018) conducted a time and cost analysis of jacket struc-
64 ture load-out. However, the time analysis was conducted from a macro perspective
65 without detailed investigation on the ballast process. Besides, the loading window
66 of tide (as Fig. 5 shows) is another factor that can be optimized. If it is chosen
67 unwisely, the tide variation may overload the ballast system or even lead to contin-
68 gencies or failures. However, there are hardly any optimization approaches except
69 for those based on rough calculation and experiences.
81 2. Methodology
82 2.1. Mechanical models
83 This paper cannot cover every aspect of the analysis. A great deal of work should
84 be done beforehand to make sure the following requirements are met:
5
85 1) The strength of all involved components is analyzed in various load cases using
86 FEM.
87 2) The configuration of SPMT is studied and verified. Some important norms are:
88 a) The load of cargo should be distributed evenly among suspension groups. b)
89 The axle loads cannot exceed the bearing limit of the structure or the ground.
90 c) The SPMT has sufficient stability. d) The PPUs will provide enough driving
91 force (Zheng and Wang, 2016).
92 3) The platform deflection due to cargo load should be insignificant. Apart from
93 making appropriate SPMT configuration and cargo arrangement, cargo weight
94 management (Wong et al., 2011) and usage of out-riggers (Naqvi et al., 2014) are
95 also considerable approaches to avoid serious deflection.
96 4) Sufficient securing means are applied. The lashing forces and friction forces can
97 balance the wind load and inertial forces. Relative movements between cargo and
98 SPMT are prevented.
99 5) The barge is well restrained in horizontal directions. The moorings and fend-
100 ers can eliminate horizontal movements and balance wind load, wave load, and
101 friction forces of the tires.
102 The model of the hydraulic system of SPMT is shown in Fig. 6. According to
103 Pascal’s Principle, the axles in a suspension group experience the same axle load.
104 Thus, the SPMT and cargo are equivalently supported at the geometric center of each
105 suspension group. Normally, the platform is set level as the initial state. When the
106 trailer drives through an uneven surface, the length of the hydraulic cylinders will be
107 adjusted automatically to adapt to the bumps and hollows. Due to the incompressible
108 nature and constant volume of the hydraulic fluid, variation in suspension length of
109 each group sums zero. Based on these constraint equations, the inclination and
110 height change of the platform can be calculated.
6
Fig. 7. Model of cargo and SPMT.
111 Since there are no relative movements, the cargo and SPMT can be modeled as
112 a combined structure (Fig. 7). The stiffness of the structure is relatively high. For
113 three points supporting system, the structure can be considered rigid. The axle loads
114 of each group can be calculated by solving force and moment balance equations. For
115 four points supporting system, the stiffness characteristic of the structure is required
116 for calculating skew loads (DNV GL, 2018). That information can be obtained by
117 analyzing the reaction forces when a unitary displacement is compulsively applied
118 on each suspension group (Sinha et al., 2018). Horizontal forces, including wind load
119 and inertia force, will induce a dynamic moment on the structure. The influence
120 can be represented by shifting the original center of gravity (COG) to an equivalent
121 position while the value of gravity remains the same. However, ballast plans are
122 made based on static analysis. Those dynamic factors are out of scope for this study
123 and thus ignored in the mechanical models.
124 Fig. 8 presents a rigid model of barge for ballast calculation. The axle loads
125 and ramp loads are replaced by concentrated loads, which will be changed along
126 with the moving of SPMT. Deadweight, ballast loads, and buoyancy are modeled
127 as distributed loads. It is assumed that horizontal loads are balanced by reaction
128 forces of the moorings and fenders. Ship hydrostatic theory is applied to obtain the
129 floatation characteristics of the barge.
130 The ballast system of a specific vessel has its unique characteristics. Its perfor-
131 mance depends on the arrangement of ballast pipelines, quantity and performance
132 curve of pumps, the volume and shape of tanks, etc. To be more general, this study
7
Fig. 8. Model of barge.
8
Fig. 9. The established coordinate system.
156 barge requires a downward force F and anti-trim moment Mx to reach the desired
157 status, the follows equations will be established.
(P
n
δi ρg = F
Pi=1
n (1)
i=1 δi ρgxi = Mx
158 where ρ is the density of water and g is the gravity acceleration. Every combination
159 of [δ1 , δ2 , . . . , δn ] that satisfies the above conditions represents a reasonable ballast
160 action. There are infinite solutions for this linear system if n > 2.
161 A ballast action includes two ways of water transfer. “External transfer” refers
162 to pumping water into or out of the barge. “Internal transfer” denotes transferring
163 water between internal tanks. For a given ballast action [δ1 , δ2 , . . . , δn ], δext and δint
164 represent the volumes of external and internal transfer, respectively. They can be
165 calculated by the following equations:
n
X |F |
δext = δi =
ρg
i=1
n ! (2)
1 X
n X Xn
|F |
1
δint = |δi | − δi = |δi | −
2 i=1 2 2ρg
i=1 i=1
166 Regardless of the execution sequence of pumping works, the general progress of
167 ballast will not be influenced as long as all pumps are fully utilized. Therefore, the
168 total time cost t of ballast action [δ1 , δ2 , . . . , δn ] can be calculated by
1 X
n
δext + δint |F |
t= = |δi | + (3)
s 2s i=1 2ρgs
9
170 If F ̸= 0, F and Mx can be replaced by an equivalent force with the same
171 magnitude acting at x = Mx /F . On condition that Mx /F = xa ∈ {x1 , x2 , . . . , xn },
172 we can find the simplest solution of equation (1) where δa = F/ρg and t = F/ρgs.
173 In this case, only the ath row will be used in the ballast action (Fig. 10a).
174 Other than that case, at least two rows will be involved. Here we denote the
175 index of involved rows by a and b where a, b ∈ {1, 2, . . . , n}. Equation (1) will be
176 simplified as (
δa + δb = F/ρg
(4)
δa xa + δb xb = Mx /ρg
177 with the solution of
F xb − Mx
δ a =
ρg(xb − xa ) (5)
F xa − Mx
δ b =
ρg(xa − xb )
178 Then
1 |F | |F xa − Mx | + |F xb − Mx | |F |
t= (|δa | + |δb |) + = + (6)
2s 2ρgs 2ρgs|xb − xa | 2ρgs
179 If F = 0, then
|Mx | |Mx |
t= ≥ (7)
ρgs|xa − xb | ρgs|xn − x1 |
180 This implies that selecting the farthest two tank rows for ballasting is the least
181 time-consuming solution in this case (Fig. 10b).
182 If F ̸= 0 and Mx /F ̸∈ [x1 , xn ], then (xa − Mx /F )(xb − Mx /F ) > 0. We can
183 assume Mx /F < xa < xb . Let λi = xi − Mx /F > 0 which measures the distance
184 from the ith row to the acting point of the equivalent force in x direction, then
|F | 2
+1
t= 1+ (8)
2ρgs λb /λa − 1
185 where
xn − Mx /F
1 < λb /λa ≤ . (9)
x1 − Mx /F
186 Thus, we will obtain a minimum t if the first and last tank rows are selected (Fig.
187 10c).
188 If F ̸= 0, Mx /F ̸∈ {x1 , x2 , . . . , xn } and Mx /F ∈ (x1 , xn ), then
|(F xa − Mx ) + (Mx − F xb )| |F | |F |
t≥ + = (10)
2ρgs|xb − xa | 2ρgs ρgs
10
189 The left and right sides will be equal if and only if (xa − Mx /F )(xb − Mx /F ) < 0.
190 Since rows at stern and bow would always be selected in other cases, they can be
191 excluded in this case to avoid duplicated use. Therefore, the two rows that are closest
192 to x = Mx /F will be used (Fig. 10d).
193 Fig. 10 depicts all the possibilities of load cases and the corresponding basic
194 solutions.
(a) F ̸= 0 ∩ Mx /F = xa (b) F = 0
11
Fig. 11. Method of heel adjustment.
12
Fig. 12. Modification to the ballast solution for tank volume limit.
226 According to equation (3), the actual time cost ta can be calculated by
1 X ′
n
|F − Fb (te )|
ta = |δi | + (12)
2s i=1 2ρgs
227 where ta can be seen as a function of te , i.e., ta = f (te ).
13
230 estimated time. For the case ta < te , the solution can be executed if the ballast
231 action is conducted slowly, but it is not efficient. The optimal solution appears at
232 ta = te . Obviously, in some special cases, ta cannot be found within a reasonable
233 time span, which means the ballast system is unable to fulfill the task in the given
234 situation. Then, it is necessary to reconsider the load-out plan and make some
235 modifications.
236 In order to obtain the optimal solution, the accuracy of the estimated time con-
237 sumption needs to be improved. Iteration methods are employed in this study. The
238 most typical iteration algorithm, which will set ta of current iteration as te for next
239 iteration, is proved unstable. The calculation may be divergent and result in infi-
240 nite loops. Thus, we use an iteration algorithm inspired by the half-interval search
241 method. The general process is divided into two stages. In the first stage, we locate
242 a small interval where the desired te lies:
243 1) Set the value of te . For the first iteration, te = 0.
244 2) Compute ta and error = ta − te . For the first iteration, error > 0.
245 3) Increase te by ξ. Repeat the previous step until error < 0
246 4) The internal (te − ξ, te ) will be the wanted interval.
247 In the second stage, we find te using half-interval search method:
248 1) Set te to be the middle value of the interval.
249 2) Calculate ta and error.
250 3) Eliminate the half interval where error on both ends have the same sign.
251 4) The search continues on the remaining half. Repeat previous steps. Stop search-
252 ing when |error| on both ends are small enough.
253 5) Output te which is the accurate time estimation.
254 With the estimated time consumption, the ballast action generated based on equation
255 (11) will be a feasible solution.
14
264 in a certain step due to the constraints of pumping capacity, tank volume, or ballast
265 water reservation.
266 “LoadoutSimulator”, an in-house program, was developed to implement the sim-
267 ulation. The algorithms of both conventional and proposed ballast strategies were
268 integrated in the program. Then, we established a simulation-based optimization
269 method which could evaluate the settings of a load-out plan based on the results of
270 virtual ballast trails. E.g., the loading process with different start times can be sim-
271 ulated. Fig. 14 reveals the relationship between the overall time consumption and
272 the start times. It can be seen that if the timing is wise, the failure of loading can be
273 avoided and the time cost will be low. Other load-out settings can also be optimized.
274 E.g., the ballast trials can be conducted with different barges. The performance of
275 different barges can be evaluated by comparing the time consumption curves. As
276 shown in Fig. 15, using Barge B will lead to the longest operation duration while
277 using Barge C may end up with failures. Therefore, Barge A is the right choice for
278 this task. More description can be found in the following section.
15
Fig. 15. Assessing the performance of barges.
287 Based on the dimensions and mass of the cargo, an SPMT model is developed
288 with twelve 6-axle-line units, eight 4-axle-line units, and four PPUs. The axles are
289 divided into 3 groups to form 3 points support, as shown in Fig. 16.
290 We conducted a static analysis to verify the SPMT configuration. Some critical
291 results are listed in Table 3. The axle loads and ground pressure of the 3 groups are
292 close to each other and do not exceed limit values. Thus the SPMT configuration is
293 qualified for further load-out operation.
Table 1
Cargo information.
294 The information of available vessels is provided in Table 4 and the tide chart at
295 the quay is shown in Fig. 17. The tide has 2 tidal periods in a day. The curve rises
296 and reaches higher high tide at around 6:00 in the morning, then drops to lower low
297 tide in noon. The lower high tide and higher low tide are reached at about 18:00
298 and 24:00 in the night, respectively. The operation will take place under weather
299 restricted condition thus wind load and hydrodynamic loads were ignored.
16
Table 2
Specifications of the SPMT units.
303 process (Fig. 18). The static load and trimming moment are caused by the weight
304 transfer of SPMT and cargo. The dynamic load and trimming moment are induced
305 by the variation of static load and moment. The axle loads are applied onto the
306 barge discretely, which results in the sudden changes of static loads (Fig. 18a, Fig.
307 18b) as well as a series of spikes of dynamic loads (Fig. 18c, Fig. 18d).
308 Tank volume and pumping speed are the bottlenecks for static and dynamic
309 loads respectively. In order to compensate for a full tide cycle, a pump with the
310 speed of 3480 m3 /h and tanks with the volume of 6589 m3 are needed to eliminate
311 the influence of the tide completely. When making the tank arrangement, only a
312 half of the tank volume is available for allocation because the other half is reserved
17
Table 3
Analysis results of the SPMT configuration.
Table 4
Specifications of available barges.
313 for the initial ballast. The tanks at stern and bow are selected for trim adjustment
314 and thus are used for moment calculation. The final allocation is demonstrated in
315 Table 5 and Fig. 19. As shown, some tanks are assigned for multiple tasks. Heel
316 adjustment cannot be neglected because it is a common scenario that the forces are
317 not balanced in the lateral direction. Notice that this is an approximate calculation
318 and the allocation may not be optimal.
319 Based on the allocation plan, we can simulate the loading process using conven-
320 tional ballast strategy. Two start times were chosen for demonstration. Fig. 20
321 presents the produced ballast sequence for the loading starting from 9:00. Fig. 21
322 shows the sequence for 15:00. “Distance” in the figures measures how far SPMT goes
323 in load-out direction. Distance of 0.0 m corresponds to the position where the first
324 axle line reaches the quay-side edge of the ramps. “Time” represents the estimated
325 time consumption. The number in each cell stands for the volume percentage of
18
Fig. 17. Tide chart of the test quay.
Table 5
Final allocation of tanks and pumps.
326 water in the tank row. As can be seen, the tide-compensating sub-system makes
327 the whole operation tide-independent. However, a large number of ballast resources
328 are used for tide compensation, and thus the total duration of about 307 minutes is
329 required in these two trials.
19
(a) Static load (b) Static trimming moment
335 minutes. As for the operation starting at 15:00, the time cost will be 73 minutes.
20
Fig. 20. Ballast sequence using conventional strategy with start time of 9:00.
21
Fig. 21. Ballast sequence using conventional strategy with start time of 15:00.
22
Fig. 22. Ballast sequence using proposed strategy with start time of 9:00.
23
Fig. 23. Ballast sequence using proposed strategy with start time of 15:00.
24
336 3.4. Optimization of load-out plan
337 The time consumption curves of both ballast strategies are depicted in Fig. 24.
338 The results of the previous 4 trials are marked in the diagram. The tide level curve
339 is also presented to reveal the relationship between tide level variation and loading
340 duration. As can be seen, the operation can be accomplished at any time of the day
341 using either ballast strategy. If the conventional ballast strategy is employed, the
342 time cost will be relatively high (307.3 minutes) and irrelevant to tidal conditions.
343 Using the proposed ballast strategy can reduce the time consumption significantly
344 (71.8-174.2 minutes) but the results are decided by the start time. To be more
345 specific, a rising tide is helpful for compensating the weight of cargo, reducing the
346 burden of the ballast system. Thus the operation duration will be short. On the other
347 hand, if the loading process takes place with a dropping tide, the time consumption
348 will increase significantly.
Fig. 24. Time consumption curves for conventional and proposed ballast strategies.
349 The same analysis was conducted on all the available barges. The comparison
350 results can be seen in Fig. 25. Barge A offers stable performance, regardless of the
351 tidal condition. Barge B has low pumping capacity and, therefore, costs more time
352 than barge 1 in general. The trial will fail if the operation experiences a low tide
353 because the ballast water reservation will be exhausted. For Barge C, the proper-
354 time window is relatively narrow. In conclusion, Barge A is the best option for this
355 task. We can pick the start time in the span from 0:00 to 6:00 or from 10:00 to 18:00
356 to acquire a short operation duration of about 80 minutes.
357 The influence of pumping capacity was studied. Time consumption curves of the
358 three example barges with different pumping capacities are depicted in Fig. 26. As
359 shown, the overall time consumption is reduced when extra pumps are integrated.
25
Fig. 25. Comparison of barge performance.
360 The reduction will be obvious if the operation takes place under tide dropping peri-
361 ods. Besides, the range of improper start time shifts towards the extremely low tide
362 and shrinks a little in width (Fig. 26b). The reason can be explained as follows. If
363 the extremely low tide occurs in the later period of load-out when cargo weight is
364 completely transferred onto the deck, the limited volume of remained ballast water
365 will be inadequate for tide compensation, and thus the possibility of loading failure
366 will increase. Therefore, the range of improper start time usually appears ahead
367 of the extremely low tide. If the loading schedule is tighter, that range will shrink
368 and shift towards extremely low tide. This analysis suggests that pumping capacity
369 affects not only the time cost but also the failure odds of a load-out operation. There-
370 fore, increasing or decreasing the capacity based on demands could be a solution of
371 preventing loading failures for the operations start at improper timings.
372 4. Conclusions
373 In this study, we investigated the engineering background of SPMT load-out and
374 summed up the conventional ballast strategy based on the literature. We found out
375 that the multi-tasking strategy which was widely utilized features weak capability
376 of finding globally optimal ballast solution thus has inefficient performance in actual
377 applications. We also noticed that there was a lack of research in the aspect of
378 load-out plan optimization.
379 Then, we proposed a new ballast strategy with a single centralized ballast system
380 instead of a series of sub-systems. The core idea is to compensate the cargo weight,
381 trimming moment, heeling moment, and tide variation at the same time by taking
382 globally optimized ballast actions. The ballast resources will be allocated dynami-
26
(a) Cases for Barge A.
27
383 cally in accordance with the actual load conditions. The mathematical model proves
384 that ballast solutions rendered by this study require less time cost than conventional
385 solutions.
386 On the basis of the proposed strategy, ballast trials can be conducted to predict
387 the time consumption and feasibility of load-out plans. We developed a simulation-
388 based optimization method which enables the engineers to evaluate load-out settings
389 based on virtual ballast trials to reduce time consumption and avoid loading failures.
390 In case studies, the proposed ballast strategy was tested and proved able to reduce
391 the time consumption by up to three quarters. The simulation-based optimization
392 method was applied in selecting appropriate start time, barge and pumping capacity
393 for particular load-out projects.
394 The proposed strategy and methods in this study can enhance the existing anal-
395 ysis workflow and help engineers make informed decisions. However, this study can
396 be improved in several aspects in future studies. Firstly, some other processes with
397 regard to SPMT load-outs can be included, such as detaching the cargo from SPMT,
398 driving the SPMT off the barge, etc. Secondly, dynamic loads and elastic character-
399 istics can be integrated to gain more accurate results.
400 Acknowledgements
401 This research was supported by the National Natural Science Foundation of
402 China (No.51109158, No.51621092), the Tianjin Key Research Program of Applica-
403 tion Foundation and Advanced Technology (19JCYBJC21900), and China Offshore
404 Oil Engineering (Qingdao) Co. Ltd.
405 References
406 Ali, I., 2014. Structural modelling of offshore module for loadout, transportation and
407 installation. Master’s thesis. University of Stavanger, Norway.
408 Dearing, B., Lucas, M., Snell, C., et al., 1985. The design and development of loadout
409 procedures for the lena guyed and tower, in: Offshore Technology Conference,
410 Offshore Technology Conference. doi:10.4043/5046-MS.
412 Ferguson, N., Zarate, H., Kitani, T., Inokoshi, O., Masuda, S., et al., 1983. An
413 analytical study and systematic monitoring procedure developed for the load-out
414 operation of the North Rankin Jacket‘A’, in: Offshore Technology Conference,
415 Offshore Technology Conference. doi:10.4043/4573-MS.
28
416 GL Noble Denton, 2016. Guidelines for load-outs.
417 Guan, G., Yang, X., Yang, Q., Peng, C., 2018. Study on simulation experiment
418 of the force of floating dock during off-loading based on ansys, in: 2018 Interna-
419 tional Conference on Mathematics, Modelling, Simulation and Algorithms (MMSA
420 2018), Atlantis Press. doi:10.2991/mmsa-18.2018.16.
421 Ham, S.H., Roh, M.I., 2018. Dynamic analysis of block offloading using self-propelled
422 modular transporters. Automation in Construction 96, 411–432. doi:10.1016/j.
423 autcon.2018.10.002.
424 Kaup, M., Jurczak, W., Kaup, J., 2016. Design methodology of strength verification
425 of platform during load out of the Arkutun Dagi SE-Topside 43.800 MT. Polish
426 Maritime Research 23, 117–128. doi:10.1515/pomr-2016-0078.
427 Kenney, J., Knoll, D., Brooks, A., Crimmins, D., Wilkinson, G., Danos, J., et al.,
428 1994. Loadout of auger deck and hull-deck mating operations, in: Offshore Tech-
429 nology Conference, Offshore Technology Conference. doi:10.4043/7625-MS.
430 Kurniawan, A., Ma, G., 2009. Optimization of ballast plan in launch jacket load-
431 out. Structural and Multidisciplinary Optimization 38, 267–288. doi:10.1007/
432 s00158-008-0287-7.
433 Lu, L., Wang, X., Wang, H., 2014. Leveling system controlled by electro-hydraulic
434 proportional valves in self-propelled modular transporter (SPMT). International
435 Journal of Scientific & Engineering Research 5, 1250–1254. URL: https://pdfs.
436 semanticscholar.org/fd2d/67f2132ae22c6ccbf2188a4b13b8ed6f7254.pdf.
437 Naqvi, D., Wey, E., Morgan, J., Miller, M., Nguyen, T., 2014. Transportation
438 considerations in module design, in: Structures Congress 2014, pp. 1771–1781.
439 doi:10.1061/9780784413357.156.
440 Piter, E., Digre, K., Tabone, A., et al., 1989. Bullwinkle loadout analysis, in: Offshore
441 Technology Conference, Offshore Technology Conference. doi:10.4043/6095-MS.
442 Silvianita, Pradana, R., Chamelia, D., Dhanistha, W., 2018. Time and cost analysis
443 of jacket structure load out using skidding, in: IOP Conference Series: Earth
444 and Environmental Science, IOP Publishing. pp. 012–024. doi:10.2991/10.1088/
445 1755-1315/202/1/012024.
29
446 Sinha, S., Luquiau, E., Reddi, S., et al., 2018. Engineering simulation for load-
447 out of a tension leg platform, in: Offshore Technology Conference Asia, Offshore
448 Technology Conference. doi:10.4043/28451-MS.
449 Sircar, S., Chandra, T., Mills, T., Roberson, W., Schultz, A., et al., 1990. Analytical
450 predictions and field observations of the loadout operations of the ’Kilauea Jacket,
451 in: Offshore Technology Conference, Offshore Technology Conference. doi:10.
452 4043/6263-MS.
453 Wong, S.S., Shive, A., Warren, H., 2011. Weight management considerations for
454 onshore modularized petrochemical facilities, in: Structures Congress 2011, pp.
455 2458–2472. doi:10.1061/41171(401)213.
456 Yang, Y., Park, B., Ha, S., 2005. Development of load-out design methodology and
457 numerical strength evaluation for on-ground-build floating storage and offloading
458 system. Ocean engineering 32, 986–1014. doi:10.1016/j.oceaneng.2004.10.014.
459 Zheng, M., Wang, J., 2016. Load-out design of jacket with spmt, in: 2016 Inter-
460 national Conference on Civil, Transportation and Environment, Atlantis Press.
461 doi:10.2991/iccte-16.2016.37.
30