Sei sulla pagina 1di 12

Poster PO-08

LEARNING CURVE MANAGEMENT ON CONSTRUCTION COST


REDUCTION IN LNG STORAGE TANKS

GESTION DE LA COURBE DAPPRENTISSAGE POUR


LEXECUTION DE LA CONSTRUCTION DE RESERVOIRS
DE STOCKAGE DE GNL
Gyu-Hyuk Lim
General Manager of Natural Gas Supply
Korea Gas Corporation
Joseph H. Cho, Ph. D.
LNG Process/Onshore Operation
KBR
Sung-Woo Kang
Manager of International Projects Group
Korea Gas Corporation

ABSTRACT
Unlike in other areas, construction performance and learning factor analysis in LNG
tank construction have been limited, compared to their important roles in LNG facilities.
The LNG tank construction cost is the largest single cost in a receiving terminal. The
specific construction cost of LNG storage tanks have decreased with an increase in
storage capacity. However, quantitatively utilizing the past experiences in cost estimation
and profitability analysis was a long term unsettled task in LNG projects.
Learning curve resources were accumulated by Kogas and KBR from three different
types of storage tanks: membrane type aboveground tanks, 9% Nickel full containment
storage tanks, and membrane type in-ground tanks (total 28 tanks, each type built at the
same site). Case based construction analysis was performed in view of tank type,
construction method, construction duration, accumulated experience, storage capacity,
and other parameters that affect construction performance. This paper presents the effects
of the Learning Curve Management (LCM) on construction performance and cost
reduction based on past experiences. A re-learning factor is incorporated in the learning
function and the ultimate level of performance has been implanted in the new model of
the learning curve. This paper demonstrates that the LCM, derived from the case based
construction analysis, provides a reasonable basis of cost analysis, cost estimation, and
profitability studies, all of which are related to the evaluation of future costs and
confidence level in an analysis.

RESUME
Contrairement dautres secteurs, lexcution de performance et lanalyse dtude
factorielle dans la construction de rservoirs de GNL ont t inactivement limits
jusquici, compar leurs rles importants dans les facilits de GNL. Le cot de
construction de rservoirs de GNL est le cot singulier le plus important dans un terminal
mthanier. Le cot spcifique de construction de rservoirs de stockage de GNL a
diminu avec laugmentation des capacits de stockage. Cependant, utiliser

PO-08.1
SESSIONS CONTENTS
Poster PO-08

quantitativement les expriences antrieures dans les analyses dvaluation du cot et de


rentabilit fut une tache perturbe long terme dans les projets de GNL.
Les ressources de la courbe dapprentissage ont t accumules pour trois types
diffrents de rservoirs de stockage : rservoirs au sol de type membrane, rservoirs en
bton prcontraint et 9% Nickel et rservoirs enterrs de type membrane (total de 28
rservoirs, chaque type construit sur un mme site). Une analyse de construction par cas
fut excute base sur le type de rservoir, la mthode de construction, la dure de la
construction, lexprience accumule, la capacit de stockage et dautres paramtres qui
affectent lexcution de la construction. Cet article prsente les effets que la gestion de la
courbe dapprentissage (LCM) a sur lexcution de la construction et la rduction du cot
par rapport aux expriences antrieures. Un facteur de re-apprentissage est incorpor dans
la fonction dapprentissage et le niveau ultime dexcution fut implant dans le nouveau
modle de la courbe dapprentissage. Cet article dmontre que LCM, driv dune
analyse par cas, fournit une base raisonnable pour lanalyse du cot, lestimation du cot
et les tudes de rentabilit proches lvaluation de cots futurs et au niveau de confiance
dune analyse.

INTRODUCTION
The construction of an LNG storage tank requires the longest construction period in
LNG plants. Its cost is also the largest single cost in LNG terminal construction.
Technical innovations and cost reductions in LNG facilities, such as in LNG storage
tanks, have strengthened LNG relative to other energy sources. Ever since the worlds
first LNG storage facility was built at Arzew, Algeria, in 1962, over 275 units of LNG
storage tanks, of which capacity range is from 30,000 m3 to 200,000 m3, have been built
and operated thus far.
Over the last three decades there have been no articles or papers focused on LNG tank
construction performance analysis. Most companies do not measure learning curve
performance in construction activities. In order to investigate the effect of past experience
on construction performance (cost and construction period) with the same type of the
LNG storage tanks, learning curve theory is introduced for a logical evaluation of
construction performance. Construction histories of three different types of LNG storage
tanks (totaling 28 tanks) were selected and analyzed by Kogas and KBR in view of cost
(escalated as of the end of 2001) and time (construction duration excluding foundation).
Korea Gas Corporation (Kogas) constructed three (3) above-ground membrane tanks
at Pyeong-Taek terminal in 1986 and added one more membrane tank in 1987. Kogas
performed an expansion project which added six (6) more storage tanks at the same site.
All of them are above-ground membrane tank and their capacities are 100,000 m3. Ten
(10) units of pre-stressed concrete (PC) 9% Nickel full containment (FC) type tanks have
been constructed over a time interval at the Incheon terminal. The terminal has also
constructed eight (8) units of membrane type underground storage tanks of which
capacity ranges from 140,000 to 200,000 m3. These three types of storage tanks are the
most widely used storage tanks in LNG terminals as well as in LNG plants.
The objectives of this study are to develop the construction analysis process and to
investigate the effects of learning based on the construction performance from the past
experience. The investigation results were used to develop a new learning curve model

PO-08.2
SESSIONS CONTENTS
Poster PO-08

which allows estimating construction cost and duration, and for profitability analysis with
a high confidence level.

LEARNING FACTOR ANALYSIS


Although the learning curve technique possesses an important management tool,
which aids in predicting cost and productivity for massive production, the traditional
model can not be applied directly to the construction of an LNG storage tank. This
inability to use the traditional model directly is due in part to the intrinsic difference that
exists between LNG tank construction and mass production. The LNG tank project has
some intermittent basis and time delays between projects. Typically, after the completion
of a project, the well trained employees may leave or may be assigned to another project.
The intermittent construction pattern, resulting in re-learning, must be considered due to
the consequential project time delays (the forget effect).
This paper proposes using the newly developed learning curve management (LCM)
concept for the LNG tank construction. Figure 1 illustrates the basic learning curve for
LNG tank construction. Tank construction processes exhibit (in many cases) a learning
curve behavior that can be expressed by Eq. 1.

n
y = C1 exp + C 3 (1)
2
C
where,
y : construction period for n unit of tank constructed
n : number of tanks constructed
C1 : constant reflecting storage volume and type
C2 : constant reflecting the construction performance
C3 : constant reflecting the ideal minimum construction period for the type of tank
50

48
Construction Period, [mon]

46

44

C1 C2 Reflects Rate of Learning


42

40
C3
38

36
1 3 5 7 9 11

Cumulative Tank Construction Units


Fig. 1 Typical LNG tank Construction Learning Curve

PO-08.3
SESSIONS CONTENTS
Poster PO-08

The learning-curve function is a relationship of variables C1, C2, and C3. C1 implies a
readiness, ability to be prepared, and a difficulty in the area or in technology. C2 relates to
how fast the organization is learning. C3 is the level of performance for any company or
contractor. This level reflects the companys technical and operational capacity and
management policies as it pertains to storage tank construction.
One common way of assessing tank construction performance is the specific LNG
tank construction cost, defined as a cost for unit storage volume ($/m3). However,
performance should also relate to the rate of learning (C2), the improvement of tank
construction in a series (C1), and the capability to maintain an achieved performance level
(C3). The capability of a company to construct LNG tanks directly relates to the
management of these variables, C1, C2, and C3. These specifically involve the successful
learning through experience, labor skills available, regional experience, employee
training, keeping well-trained employees, understanding policies and procedures by the
company, and the availability of tools and information accessible to the company.
Greater learning leads to better possibilities of long-lasting improvement [1]. The
higher the level of response accuracy to the construction results, the better the chances of
increasing construction performance and capacity. Construction personnel experience is
critical to both the reduction of inefficient work and to the improvement of construction
capabilities. Construction experience is a composite of local type experience - tanks
constructed with the same type design. It also requires general training for each
construction process. Likewise, a captain needs skills and training in order to navigate an
LNG carrier.
Overall construction performance has been greatly influenced by a companys
construction capacity and tank construction personnel. The less the organizational
construction capacity, the more the organization is at the mercy of the available tank
construction personnels experience in interpreting the specific work procedure and
organizing the low-level workforce. That is why a company that depends solely on its
construction personnel for performance becomes unproductive when personnel leave the
organization [2].

LEARNING CURVE RESOURCES


In order to obtain a good analysis in construction performance and to investigate the
effects of LCM, selecting high quality learning curve resources is quite important. Some
construction conditions, such as tank type, location, construction technology applied,
regulations, degree of infrastructure, etc, greatly affect construction performance.
Normally, at LNG receiving terminals, a small number of LNG storage tanks are built at
the initial stage of construction and then more storage tanks are added depending on the
gas market maturity.
LNG tank construction for terminal expansions may be affected by operations. It is
imperative that companies involved in expansion establish a safety construction plan to
help minimize the negative effects of simultaneous operation and construction at the same
site. Restricted access to the site and limited availability of the construction area
significantly reduces the construction improvement capabilities in some cases. The site
location has an impact on the overall cost and schedule: camp cost, labor, and import of
construction material and equipment. Since learning curve resources selected for this

PO-08.4
SESSIONS CONTENTS
Poster PO-08

study have well-established plans that help minimize the number of constraints, the
effects of expansion on construction capabilities are considered insignificant.
The following tables summarize the learning curve resources used in the construction
analysis process and in the LCM. The construction duration excludes the period for site
improvement and piling because the required period varies from location to location,
even within the same site. Table 1 shows the construction history of the membrane type
above-ground tank. All of these tanks have a capacity of 100,000 m3 and were
constructed by the same contractor. A long time interval (about 6 years) exists between
the first tanks (# 4) in operation and the first expansion tank (# 5) in operation.
Table 2 presents the history of the PC 9% Nickel FC type aboveground tanks. All of
these tanks at the Incheon terminal have the same as 100,000 m3 and were consecutively
constructed by one contractor.
The construction history of membrane type inground tanks are shown in Table 3. The
first four of each type of tank (#1 - 2, #3 - 4) have been constructed by two different
companies (A, B). The remaining two tanks have just been completed and two are still
under construction by a different contractor (C).

Table 1 Membrane Type Aboveground Tanks [3]


Tank No. Capacity (m3) Operation start year Construction period (mon)*1
#13 100,000 Oct. 1986 36 - 38
#4 100,000 Oct. 1987 34
#5 100,000 Nov. 1993 37
#6 100,000 Nov. 1994 34.5
#7 100,000 Nov. 1995 33.5
# 8 9 100,000 Dec. 1997 32
# 10 100,000 Dec. 1998 31.5

Table 2 PC 9% Ni Full Containment Tanks [4]


Tank No. Capacity (m3) Operation start year Construction period (mon)*1
#13 100,000 Oct. 1996 35 - 36
#4 100,000 Sep. 1997 33
#5-6 100,000 Sep. 1998 34
#7-8 100,000 Aug. 1999 32.5 - 33
# 8 9 100,000 Nov. 1999 32 - 32.5
# 10 100,000 Mar. 2000 32

Table 3 Membrane Type Inground Tanks [5]


Tank No. Capacity (m3) Operation start year Construction period (mon)*2
#12 140,000 Oct. 2001 44 - 48 (A)
#3-4 200,000 Dec. 2002 53 - 55 (B)
#5-6 200,000 Sep. 2003 52 - 54 (C)
#7-8 200,000 Apr. 2004 51 (C) *3
Note 1: Constructed by one contractor at the same site
Note 2: Blanks illustrate different contractors
Note 3: Tanks under construction

PO-08.5
SESSIONS CONTENTS
Poster PO-08

Parameters Affecting Construction Performance


Parameters that affect construction performance and cost drivers have been identified
for an LNG storage unit. These involve the following:
Types of tanks (Single, Double, Full containment, and Membrane type)
Sizes of tanks
Construction schedules (contractual obligation)
Locations (construction access constraints)
Seismic designs
Operational pressures and heat leaks (Boil-off gas handling)
Tank spacing
Level of employee training
Contractors construction capabilities

LNG Tanks Constructed


The tank specifications used in this study are summarized in Table 4.

Table 4 LNG Tank Specifications


Description Membrane A/G *1 9% Ni A/G Membrane U/G
*2
Capacity (m3) 100,000 100,000 140,000 200,000
Primary Barrier membrane (1.2mm) 9% Ni membrane (2.0
BOG Rate (%) 0.1 0.075 mm)
Side Structure PC wall PC wall 0.1
Roof Concrete dome Concrete concrete wall
Inner Dia. (m) 57.5 dome Concrete dome
Outer Dia. (m) 59.5 68.0 72
Height (m) 41.4 70.0 78.6
Max. Liquid Level 39.9 36 53
(m) 29.9 49.2
*1: Above ground
*2: Under ground

LEARNING CURVE MANAGEMENT (LCM)


Figure 2 illustrates the LCM for above-ground membrane tanks. The original LC
excluding the re-learning effect can be expressed by:

n
y = 10.19 exp + 30.9 (2)
7.34

The required period of the first tank construction was about 38 months. Site
improvements and piling are excluded in this period in order to accurately evaluate the
accurate effects of LCM on construction performance. The duration for the foundation
(including piling) varies depending on the source rock distribution, even with tanks (same
design and capacity) constructed at the same site. The required period for the 2nd and 3rd
were almost the same because the tanks were constructed simultaneously.

PO-08.6
SESSIONS CONTENTS
Poster PO-08

However, the construction of LNG tank #5 began 6 years later after the construction
of tank #4. The LC shows the re-learning effect. The requirement of re-learning can be
depicted through the company that lost experienced personnel, causing low learning
levels within the company. The re-learning process rapidly recovers the companys
construction capabilities. The normal expected construction period would be 31 months.
However, the re-learning adds 6.0 more months to the schedule. If the company
effectively maintains its construction capabilities, it could decrease the re-learning period.
After re-learning the construction process, construction performance dramatically
improves. The re-learning effect (C2) is almost twice as faster than in the case where the
company lacked previous experience.
A tank was constructed almost every year throughout the expansion stage. Therefore,
a deduction can be made which suggests that the trained engineers and skilled technicians
were well maintained throughout the remainder of the expansion stage. The limit on the
construction period (C3) with the above-ground membrane storage tanks (100,000 m3)
under the given conditions is about 31 months. These conditions involve a companys
construction performance capacity, its learning capabilities, its construction
characteristics of the given tank type, and its general construction conditions at the site.

40

Re-learned C1
Construction Period , [months]

38

C2
36
Re-learned C2

C1
34

32
C3

30

28
1 2 3 4 5 6 7 8 9 10

Cumulative Tank Construction Units

Fig. 2 LCM for Above-ground Membrane Tank

The company involved in 9% Nickel FC tank has previous experience, which enables
it to build a similar type of cryogenic tank for ethylene. The LC also shows the smooth
learning effect based on past experiences (Fig. 3). The LC can be expressed as follows:

n
y = 5.43 exp + 31.6 (3)
4.55

The learning rate is lower than that of the membrane tank. Several reasons for this are
explained as follows: the storage tanks are constructed within a relatively short period of

PO-08.7
SESSIONS CONTENTS
Poster PO-08

time, the company did not have enough time to accumulate the learning effect on the
construction performance, and a limitation existed on the available skilled employees and
well trained engineers to build the tanks within the short time period. A significant
increase in construction performance appeared in tank #4. The large block prefabrication
method was applied to the roof structure [6].

38
Construction Period, [months]

36
C2

C1
34

C3
32

30

28
1 2 3 4 5 6 7 8 9 10

Cumulative Tank Construction Units


Fig. 3 LCM for 9% Nickel Above-ground PC Tank

Figure 4 illustrates the LCM for in-ground membrane tanks. The storage capacities of
the first two tanks constructed by a company (A) are 140,000 m3. Tanks #3 and #4 were
constructed by a company (B). The storage capacities of tanks #3 8 are 200,000 m3.
Tanks #5 and #6 were constructed by a company (C). The rest of the tanks (#7 8) are
also under construction by company (C). Figure 4 highlights an interesting detail in the
LCM for the different capacity tanks and for the different contractors. The construction
experience can be transferred indirectly to the company who constructed the same types
of tanks later on. The latter company has a steeper LC in construction performance. The
specific construction performance, which is defined as the construction monthly period
divided by LNG storage volume (1000 m3), of companies A and B result in values of
0.343 and 0.275 for 140,000 m3 and 200,000 m3, respectively. The reduced specific
performance can be interpreted to indicate that as the LCM for the large tank increases, a
reduction in the construction costs follows. The ultimate values of C3 in each storage
capacity are projected at 42 and 51 months for 140,000 m3 and 200,000 m3, respectively.
C3 values highly depend on the depth of the slurry walls.

PO-08.8
SESSIONS CONTENTS
Poster PO-08

59

200, 000 Kl
constructed by
Construction Period, [months]

55
a company C

51

Capacity
Factor
47 200, 000 Kl
constructed by
a company B

43

140, 000 Kl
39 constructed by
a company A

35
1 2 3 4 5 6 7 8 9
Cumulative Tank Construction Units
Fig. 4 LCM for In-ground Membrane Tank

Quantitatively utilizing past experiences in cost savings and achieving an increase in


profitability were long term unsettled issues in the LNG industry. Since the construction
of a tank requires a large number of materials and different kinds of labor, the exact
escalation in cost is also difficult to determine before construction. In-ground tanks are
excluded in the investigation on the effect of past experiences on the storage tank cost
because their storage capacities were different and a diverse group of contractors were
involved. The 9% Nickel above-ground FC tanks also turned out to be an improper model
since the contract policy was changed: the additional construction was on a renewed
reimbursable contract basis. The contract was renewed twice, once for three tanks (# 4
6) and again for four tanks (# 7 10). The original construction was based on lump sum
contract.
On the other hand, the above ground membrane tank shows a good model. These
tanks were contracted almost on a yearly basis during the expansion stage at the Pyeong-
taek terminal. Figure 5 presents cost effectiveness of past experiences on the storage
tanks. The tank costs escalated towards the end of 2001 with the inflation rate of industry
manufactured goods issued by the Korean National Bank. The tank costs were then
divided by the last tank cost, which provides the basis for specific tank construction cost.
The contract for the first three tanks was awarded to SN Technigaz, France, who served
as the primary contractor and a local company was nominated as the sub-contractor. The
fourth tank, built about one year later, was constructed by the local company who served
as the primary contractor, while SN Technigaz functioned as the sub-contractor. The
overhead costs, related to the design, were not applied to the tank cost because the same
design was applied. Six years later, the fifth tank was constructed. The cost of this tank
was estimated based on escalation of the previous tank. The main materials for the tank
construction, such as membrane and insulation panels, were locally manufactured for the
sixth through tenth tank. This localization seems to be a reason why costs decreased in
the construction of tanks # 6 7.

PO-08.9
SESSIONS CONTENTS
Poster PO-08

The estimation of future tank costs or contractors profitability is essential in LNG


projects. Tank cost effectiveness exhibited (in many cases) an LC behavior (Fig. 5) that
can be expressed as:

n
(TIC ) f = C P exp + Cm (4)
Cv
where,
(TIC)f : future installed cost
n : number of tanks in the field
Cp : constant reflecting storage volume
Cv : constant reflecting construction performance
Cm : constant reflecting the ideal minimum cost for an area

120

115
Specific Tank Construction Cost, [%]

110

105

100

95

90

85

80
1 2 3 4 5 6 7 8 9 10 11

Cumulative Tank Construction Units

Fig. 5 Cost Effectiveness of Past Experiences for Above-ground Membrane Tank

Segmented sub-activities for the PC 9% Nickel FC tank is illustrated in Fig. 6. Since


some of the sub-activities carried out overlapped on the time scale (i.e., roof fabrication is
performed during fabrication of PC side-walls), each segment shows the net period. The
actual amount of manpower input was not available. However, if this information were
available, the specific work rate could then be compared and the reasons for
improvements in C1, C2, and C3 could also be obtained from past experiences.
This type of data manipulation allows the analyst to quickly focus on various
activities with a specific productivity rate (daily work production divided by manpower
input). The construction analyst should always be investigating for value added
improvements in construction performance. The quicker the analyst can focus on high
impact activities, the more time is available for the next level of construction analysis.
Based on past experience, C2 learning parameters increase during roof fabrication and
inner shell work mainly due to large block steel roof fabrication. All of the steel blocks

PO-08.10
SESSIONS CONTENTS
Poster PO-08

were prefabricated in a nearby workshop and then moved to a temporary shop where the
roof insulation was installed. The completed large blocks were transported to the
construction site [6]. The new construction method demonstrably shortened the field
construction period. A new welding jig had also developed and applied on the inner shell
welding job based on the site welding experience. The new welding jig reduced the repair
job, resulting in an increase in welding constructability. However, the shortened roof
fabrication period is not directly affected by the shortening of the whole tank construction
because it is mainly performed inside the tank during the PC wall erection. The harder
work that has a lower LC effect is side-wall work.

1000
Working Period for Each Activity, [days]

Others

800
Cool-
down

600 Insulation

Inner-
shell
400
Roof fab

200 Sidewall

0
1 2 3 4 5 6 7 8 9 10

Cumulative Tank Construction Units

Fig. 6 Segmented Activities Distribution for 9% Nickel Tank (100,000 m3)

CONCLUSIONS
1. The developed LC model for LNG tank construction well represents the
improvements of construction performance based on past construction experience,
and improves estimations of future costs for three types of LNG tanks.
2. LCM can be a useful tool for project developers to estimate future tank costs, and
for contractors to confidently predict the profitability of the companies
construction activities.
3. The construction analysis strengthens the level of learning based on past
experiences, resulting in an increase in construction performance and capability.
4. In order to minimize the re-learning factor (forget effect), a company needs to
organize the developed construction processes, manage learning resources, hold
experienced employees, and pay careful attention to the indirect activity of macro
LCM.

PO-08.11
SESSIONS CONTENTS
Poster PO-08

ACKNOWLEDGEMENT
The authors wish to thank Korea Gas Corporation (KOGAS) for their permission to
reference their construction and operation data of LNG terminals. Special thanks are
extended particularly to J.S. Jung at Kogas Maintenance and Engineering Co. and S. B.
Hong at Daewoo Engineering & Construction Co. for their kind assistance and useful
discussion.

REFERENCES CITED
1. Hosin, Y.A.: Learning Curve for Interrupted Production, Proc., 1989 Int. Industrial
Engineering Conference, Institute of Industrial Engineers, June 23, 1989.
2. Millheim, K., Maidla, E., and Kravis, S.: An Example of the Drilling Analysis
Process for Extended Reach Wells, SPE Paper 49111, the Annual Technical
Conference and Exhibition, New Orleans, USA, 27-30 September 1998.
3. Construction Report of Pyeongtaek Expansion Project, Korea Gas Corporation, 1988.
4. Construction Report of Incheon Terminal, Korea Gas Corporation, 2000.
5. Construction Report of LNG Storage Tanks #11 14, Korea Gas Corporation, 2002.
6. Kwon, Y.J.: Commissioning of Korea Gass New Receiving Terminal at Incheon,
Proc., LNG 12, Perth, West Australia, May 4- 7 1998.

PO-08.12
SESSIONS CONTENTS

Potrebbero piacerti anche