Академический Документы
Профессиональный Документы
Культура Документы
Engineering HSE Group, HSE Systems Department, Engineering Division, JGC Corporation, 2-3-1, Minato Mirai, Nishi-ku, Yokohama
220-6001, Japan
b Laboratory for Safety Engineering and Risk Management, Yokohama National University, Hodogaya-ku, Yokohama 240-8501, Japan
a b s t r a c t
Many base load onshore LNG plants use large number of Air-Fin-Coolers normally mounted on the center pipe rack
of the LNG process train. Further, the LNG plant modularized approach requires large, complex structures (modules)
for supporting the LNG process equipment and for allowing sea and land transportation. This results in additional
congestion of the plant and large voids under module-deck, which are conned by large girders. Thus, in case of
leaks, the proper ventilation to reduce the accumulation of gas is critical for the safety of the plant.
This paper evaluates the Air-Fin-Cooler induced air ow in modularized LNG plants using Computational Fluid
Dynamics (CFD) analysis.
The results of this evaluation show that the ventilation of the Air-Fin-Cooler induced air ow is inuenced by the
process train orientation. Further, a moderate increase is observed in specic design conditions or areas, such as
shorter separation distances between modules. Based on the results of this evaluation, four design measures are
proposed to optimize the use of Air-Fin-Cooler, such as train orientation against prevailing wind direction and use
of the grating deck material.
2012 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
Keywords: Air-Fin-Cooler; Forced ventilation; Separation distance; LNG
1.
Introduction
Corresponding author. Tel.: +81 45 682 8505; fax: +81 45 682 8850.
E-mail addresses: tanabe.masayuki@jgc.co.jp (M. Tanabe), atsumi@ynu.ac.jp (A. Miyake).
Received 23 September 2011; Received in revised form 11 July 2012; Accepted 4 September 2012
0957-5820/$ see front matter 2012 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.psep.2012.09.001
352
353
Purpose
Modeling
Advantage
Disadvantage
Remarks
To evaluate
detailed air ow
for each case
Actual geometry
for all equipment
Air ow based on
AFC fan
performance
curve
Can
provide/quantify
detailed air
ow/ventilation for
each module
To determine
trends of in/out
ows (target
volumea )
considering
high/low packing
density
Actual geometry
only for large
equipment
Porosity used for
congested area
(where
calculated
porosity is less
than 0.9)
Constant AFC air
ow
Used in this
study
Target volumes in this study are above deck area, below deck area and gap.
AFC species the required duty for the process uid cooling
and AFC is designed to provide the required air ow rate for
the duty accordingly. However, the Air Change per Hour (ACH),
which is an indicator of ventilation and the function of area
volume and air ow rate, is not normally calculated. Thus, in
the current standard design practice, the AFC forced air ow
is not effectively used for enhancing the ventilation (i.e., to
reduce possibility of ammable gas accumulation), in particular with modular design, during emergencies.
The study is planned in two steps (1) to quantify ventilation by ACH as general indication for ventilation effect and
(2) to check gas dispersion trend (e.g., buoyancy, release direction). This paper covers the rst step and estimates the forced
ventilation effect of AFC (i.e., the increase of ACH due to the
AFC induced air ow over natural ventilation) inside modularized LNG process trains which have higher congestion than
traditional onshore stick built LNG plant. Computational Fluid
Dynamics (CFD) analysis has been used for the estimates and
for evaluating the design measures for increasing ACH without
modifying AFC process design, such as
1. The increase of ventilation in the modules and gaps due to
wind conditions and orientation of the trains
2. The impact on AFC induced air ventilation of
modules separation distances
deck material
Based on the ndings from this paper, the second step will
cover gas dispersion study using CFD for the following model
geometries and leak parameters
1. Geometries
wind conditions and orientation of the trains
modules separation distances
deck material
2. Leak parameters
leak release direction (downward and horizontal)
buoyancy (methane gas, LNG, propane).
2.
Methodology
2.1.
Strategy
2.2.
Study basis
2.3.
354
R
3600
(1)
where
Qa : Air Flow Rate (m3 /s)
Vmod : Free Module Volume (m3 )
R : Air Change Rate per Hour
Since the ACH calculation is simply related to free volume
in the area and air ow rate passing through the area, it is
important to correctly identify the detailed air ow inside the
area (Horan and Finn, 2008; Matsuura et al., 2010; Deru and
Burns, 2003). Therefore, CFD analysis has been used.
2.4.
The detailed air ow inside the process train, which is a component of the ACH calculation, is affected by such factors as
3.
3.1.
General
3.2.
355
Fig. 5 Side view of deck concept (solid plate vs. grating decks).
3.3.
Simulated scenarios
The CFD analysis has been done for the cases shown in
Table 2. The 2 wind speed cases are selected for the study,
which are 5 m/s and 10 m/s. Since AFC forced ventilation
Conditions
a ,b
AFC-on/off
Wind speed
Train orientation (wind
direction)
Sensitivity analysis
Separation distances
between modules
Deck oor material
a
b
Run cases
AFC-on
AFC-off
5 m/s
10 m/s
Perpendicular wind direction to
process train in the longitudinal
direction (from North)
Parallel wind direction to process
train in the longitudinal direction
(from West)
8m
15 m
25 m
Grating
Plating
356
Local
Instrument
Room
Local Electrical
Room
Refrigeration
Compressor
Package
Refrigeration
Compressor
Package
Symmetry planes
Fig. 7 N-Wind model area.
3.4.
Simulation model
357
domain outlets with no constraint on temperature. The outlets from the AFCs are dened as domain inlets, with the air
entering the domain constrained to have the temperature
dened by the actual design information. The temperature
is applied uniformly across the boundary.
Mesh
CFD model mesh is established as shown in Figs. 11 and 12.
Finer mesh size is applied to the modules and AFC and the
mesh size is gradually enlarged toward the model boundary. The minimum and maximum mesh sizes are 0.5 m and
5 m. The total number of mesh elements is summarized in
Table 3.
Equations
The equations used in the CFD are summarized in Table 4.
The large equipment is modeled explicitly in the CFD model.
The loss term due to small pieces of equipment is established as a friction factor only (Figs. 1315).
Module congestion is represented by applying porosity
less than unity and a ow resistance source term for the
momentum equation based on the Modied Porosity Distributed Resistance (MPDR) model (Vianna, 2009).
In CFX there is a signicant computational cost associated
with using porosity not equal to unity. Hence the full porous
model is only applied to congested regions with a volume
porosity below 0.90. It is judged that regions with porosity greater than this would not affect the ow signicantly
enough to justify the computational expense of using porosity
less than unity, since the terms in equations for porosity less
than unitiy would be almost identical to those in equations
for porosity equal to unity in Table 4.
It would be impractical to calculate the volume porosity
for each computational cell in the domain, so each identied congested region is assumed to have uniform volume
Fig. 10 CFD model boundary for W-Wind case (parallel to the process train axis).
Fig. 11 Mesh cut plane for N Wind (Perpendicular to the Process Train Axis) simulations.
358
Fig. 12 Mesh cut plane for W Wind (parallel to the process train axis) simulations.
porosity throughout the region. The volume porosity for each
congested region is calculated independently from the actual
model data, with the region boundaries identied through
manual inspection of the model. The locations of these regions
Module separation
8m
15 m
25 m
8 m (plated deck)
3,794,883
5,169,127
4,641,226
5,517,911
4,699,063
6,252,662
3,782,113
5,169,160
Equations
Remarks
Turbulence
Commonly used
Based on eddy-viscosity concept
Energy
t + ( U h) = ( T) + :
U + SE
Steady/Transient
Continuity
t + ( K U) = 0
Porosity equal to unity:
(U)
t + ( U U) =
p + + SM Porosity less than
unity:
t ( U) + ((K U) U)
Momentum
(h)
Flow resistance
Friction factor
e K U + (U) 23 U
SM p
SM,i = Kloss,i (/2)|U|Ui
Ri = SM,i
|u |
Kloss,i = 1 |U|i Aw fi
fi,pl = 0.0048Re0.2
i
fipr =
0.23 +
0.11
1.08 Re0.15
i
0.5
P
D
Rei =
|ui |Dn
|ui |D
359
4.
The results of the CFD analysis are shown in Tables 79. The
results are summarized based on the effect of AFC-on, separation distance, and deck oor material in Sections 4.1, 4.2 and
4.3, respectively.
4.1.
360
Piperack-1
Piperack-2
Piperack-3
Piperack-4
Piperack-5
Piperack-6
Piperack-7
Equipment Module-1
Equipment Module-2
0.823
0.805
0.839
0.840
0.831
0.853
0.862
0.732
0.741
Pitch between
obstacles, P (m)
Characteristic
dimension, D (m)
Piperack-1
Piperack-2
Piperack-3
Piperack-4
Piperack-5
Piperack-6
Piperack-7
Equipment Module-1
Equipment Module-2
N/A
N/A
4.05
4.05
0.67
0.42
N/A
5.07
4.48
2.5
2.5
3
3
1
0.8
8
5
4.5
0.3
0.3
0.25
0.25
0.6
0.6
0.64
1.3
1.25
Area
ACH
AFC-on
Above deck
Below deck
Whole module
Above deck
Below deck
Whole module
Above deck
Below deck
N-wind 5 m/s
NE
NW
SE
SW
N-GAP
S-GAP
660
667
603
590
684
624
690
694
607
590
722
634
856
787
732
728
1048
828
491
490
485
483
501
496
507
507
500
495
518
509
342
348
402
392
375
424
34
36
24
22
37
26
36
37
21
19
39
25
150
127
82
86
179
95
N-wind 10 m/s
NE
NW
SE
SW
N-GAP
S-GAP
1070
1076
1029
1020
1108
1083
1107
1109
1043
1029
1142
1110
1143
1057
1092
1082
1082
1183
976
975
956
951
951
978
1009
1009
985
975
975
1004
678
695
790
772
772
837
10
10
8
7
11
11
10
10
6
6
11
11
69
52
38
40
78
41
W-wind 5 m/s
M1
M2
M3
M4
GAP12
GAP23
GAP34
528
640
719
752
2534
2759
2909
538
654
735
767
2568
2809
2968
531
631
651
650
2329
2409
2360
489
470
437
422
2369
2211
2069
505
482
450
435
2438
2269
2125
361
356
365
348
1717
1657
1627
8
36
64
78
7
25
41
6
36
63
77
5
24
40
47
78
78
87
36
45
45
W-wind 10 m/s
M1
M2
M3
M4
GAP12
GAP23
GAP34
921
883
934
1018
4174
4160
4299
946
898
944
1035
4218
4200
4362
814
849
851
988
3866
3851
3706
978
938
873
847
4735
4405
4134
1010
962
898
871
4868
4514
4239
728
716
732
697
3466
3352
3284
6
6
7
20
12
6
4
6
7
5
19
13
7
3
12
19
16
42
12
15
13
Whole module
AFC-off
361
362
Area
ACH
Whole module
8 m SD
15 m SD
Above deck
25 m SD
8 m SD
Below deck
15 m SD
25 m SD
8 m SD
15 m SD
25 m SD
N-wind 5 m/s
NE
NW
SE
SW
N-GAP
S-GAP
660
667
603
590
684
624
669
668
617
607
674
638
665
672
605
593
671
610
690
694
607
590
722
634
700
694
620
608
716
646
692
693
607
593
702
614
856
787
732
728
1048
828
811
765
701
680
920
713
779
770
664
664
831
682
N-wind 10 m/s
NE
NW
SE
SW
N-GAP
S-GAP
1070
1076
1029
1020
1108
1083
1083
1080
1033
1019
1084
1052
1075
1087
1017
1025
1085
1033
1107
1109
1043
1029
1142
1110
1112
1107
1050
1032
1120
1069
1106
1113
1027
1037
1115
1047
1143
1057
1092
1082
1333
1183
1063
1027
1053
997
1171
1017
1060
1061
982
1012
1124
1024
W-wind 5 m/s
M1
M2
M3
M4
GAP12
GAP23
GAP34
528
640
719
752
2534
2759
2909
549
681
737
763
1418
1582
1695
568
714
773
766
946
1073
1131
538
654
735
767
2568
2809
2968
557
694
750
776
1436
1608
1731
577
726
786
780
960
1092
1150
531
631
651
650
2329
2409
2360
535
621
625
636
1319
1361
1403
564
634
682
673
850
957
939
W-wind 10 m/s
M1
M2
M3
M4
GAP12
GAP23
GAP34
921
883
934
1018
4174
4160
4295
925
903
980
1079
2248
2316
2523
946
924
1038
1125
1384
1483
1644
946
898
944
1035
4218
4200
4362
941
916
995
1095
2264
2335
2569
966
935
1052
1137
1396
1507
1661
814
849
851
988
3866
3851
3706
829
905
921
986
2195
2185
2167
835
859
1020
1018
1294
1354
1510
Area ACH
Below deck
AFC-on
Above deck
AFC-off
AFC-on
Whole module
AFC-off
AFC-on
AFC-off
Grated Plated Grated Plated Grated Plated Grated Plated Grated Plated Grated Plated
N-wind 5 m/s
NE
NW
SE
SW
856
787
732
728
643
645
641
650
342
348
402
392
341
343
367
370
690
694
607
590
667
675
594
583
507
507
500
495
506
507
494
492
660
667
603
590
658
664
597
587
491
490
485
483
491
491
482
481
N-wind 10 m/s
NE
NW
SE
SW
1143
1057
1092
1082
951
958
979
981
678
695
790
772
679
683
725
730
1107
1109
1043
1029
1082
1091
1028
1018
1009
1009
985
975
1007
1009
972
969
1070
1076
1029
1020
1064
1074
1022
1013
976
975
956
951
976
978
948
946
W-wind 5 m/s
M1
M2
M3
M4
531
631
651
650
482
574
598
617
361
356
365
348
343
337
324
310
538
654
735
767
533
646
723
761
505
482
450
435
502
474
442
427
528
640
719
752
529
639
712
747
489
470
437
422
487
461
431
416
W-wind 10 m/s
M1
M2
M3
M4
814
849
851
988
742
824
875
944
728
716
731
697
693
687
659
625
946
898
944
1035
961
903
941
1014
1010
962
898
871
1004
954
888
857
921
883
934
1018
939
891
935
1007
978
938
873
847
974
929
867
835
363
4.2.
364
4.3.
4.4.
Recommended design option enhancing
ventilation
Based on the above results (Sections 4.14.3) in the Step-1
ventilation study, the following design approaches are identied as possible optimization for the use of AFC-on ventilation
for reducing possibility of ammable gas accumulation in
onshore modularized LNG plant.
The AFC fans should be kept running even in emergency conditions to reduce the amount of ammable gas
365
The above design measures shall be carefully evaluated also from other aspects, such as adverse effect by
reducing the separation distance, hot air circulation, operation/maintenance aspects. Especially, hot air circulation
phenomenon and mitigation measures shall be evaluated in
order to minimize the reduction of the AFC performance and
production rate when the train orientation is perpendicular
to the wind direction. However, since modularized approach
increases congestion of the plant and creates large voids under
the module deck, which are conned by very large girders, we
believe that the proposed approach shall be considered.
The future paper (Step-2) will further evaluate gas dispersion effect considering other parameters, such as buoyancy of
leaked gas and release direction.
5.
Conclusion
366
References
van den Berg, A.C., Versloot, N.H.A., 2003. The multi-energy
critical separation distance. J. Loss Prevent. Process. 16,
111120.
Huser, A., Foyn, T., Skottene, M., 2009. A CFD based approach to
the correlation of maximum explosion overpressure to
process plant parameters. J. Loss Prevent. Process. 22, 324331.
Horan, J.M., Finn, D.P., 2008. Sensitivity of air change rates in a
naturally ventilated atrium space subject to variations in
external wind speed and direction. Energ. Buildings 40,
15771585.
Matsuura, K., Nakano, M., Ishimoto, J., 2010. Forced ventilation
for sensing-based risk mitigation of leaking hydrogen in a
partially open space. Int. J. Hydrogen Energy 35, 47764786.
Paterson, K., Tam, V.H.Y., Moros, T., Ward-Gittos, D., 2000. The
design of BP ETAP platform against gas explosions. J. Loss
Prevent. Process. 13, 7379.
Deru, M., Burns, P., 2003. Inltration and Natural Ventilation
Model for Whole-building Energy Simulation of Residential
Buildings. National Renewable Energy Laboratory.
Tanabe, M., Miyake, A., 2010. Safety design approach for onshore
modularized LNG liquefaction plant. J. Loss Prevent. Process.
23, 507514.
Pitblado, R., Baik, J., Raghunathan, V., 2006. LNG decision making
approaches compared. J. Hazard. Mater. 130, 148154.
Tam, V., Mores, T., Webb, S., Allinson, J., Lee, R., Bilimoria, E., 1996.
Application of ALARP to the design of the BP Andrew platform
against smoke and gas ingress and gas explosion. J. Loss
Prevent. Process. 9, 317322.
Vianna, S., 2009. Numerical Simulation of Accidental Explosions
in Offshore Production Plant. University of Cambridge.