See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/318852821

PVP2017-65870 ASSESSMENT OF THE CRITICAL TILTING ANGLE OF A COKE

DRUM VESSEL SUBJECT TO SEISMIC LOADING

Conference Paper · July 2017

DOI: 10.1115/PVP2017-65870

CITATIONS READS

0 395

4 authors, including:

Julian Bedoya
Stress Engineering Services, Inc.
17 PUBLICATIONS   145 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Julian Bedoya on 02 August 2017.

The user has requested enhancement of the downloaded file.

 Proceedings of the ASME 2017 Pressure Vessels & Piping Conference
PVP2017
July 16-20, 2017, Waikoloa, Hawaii, USA

PVP2017-65870

ASSESSMENT OF THE CRITICAL TILTING ANGLE OF A COKE DRUM VESSEL

SUBJECT TO SEISMIC LOADING

Jorge PENSO Julian J. BEDOYA

Shell Global Solutions (US) Inc. Stress Engineering Services Inc.
Houston, Texas, USA Houston, Texas, USA

Scott BOUSE Sathish K. RAMAMOORTHY

Stress Engineering Services Inc. Stress Engineering Services Inc.
Houston, Texas, USA Houston, Texas, USA

ABSTRACT RSFa – Allowable Remaining Strength Factor

Assessment of a coke drum for seismic stability is EP – Elastic-Plastic Material Model
generally a well-documented load case, with ample examples EPP – Elastic-Perfectly Plastic Material Model
and historical design experience. However, the long-term LRFD – Load Resistance Factor Design
effects of non-uniform temperature distributions have in many PEEQ – Total Equivalent Plastic Strain
cases led to a phenomenon in coke drums commonly referred to FEA – Finite Element Analysis(es)
as the “banana effect”. The banana effect leads to out-of-plumb, FEM – Finite Element Method
and local weakening of the shell due to bulging, and creates a mph – Miles per Hour
more vulnerable structure where the contributions of tilting I – Importance Factor for Wind or Seismic per ASCE
angle and effective seismic loads vary with each stage of the Z – Seismic Acceleration per ASCE
coking cycle. This paper outlines one practical approach to this S – Spectral Response per ASCE
challenging problem, along with the derivations and analytical ID – Inside Diameter or Internal Surface
techniques necessary to support the resulting conclusions. The OD – Outside Diameter or External Surface
framework developed was able to compare the influences of OAL – Overall Length
tilting and seismic accelerations independently, even when psi – Pounds per Square Inch
acting simultaneously. Using this technique, the analysis was lbf – Pound Force
able to efficiently predict a relationship between an expected kip – 1,000 lbf
seismic load and the critical drum tilt angle, with only minor P – Internal Pressure
adjustments needed to account for non-linear vertical D – Dead Weight
distribution of the shear force. PS – Static Head Pressure
W – Wind Loading
NOMENCLATURE E – Earthquake Loading
API – American Petroleum Association LC – Load Case
ASME – American Society of Mechanical Engineers β – Load Factor Coefficient found in API 579
ASCE – American Society of Civil Engineers ƟRealTilt – Real Tilt Angle – an iterated quantity
BPVC – Boiler and Pressure Vessel Code ƟEqTilt – Equivalent angle of tilt considering the real tilt
VIII-1 – ASME BPVC Section VIII Division 1 angle as well as the gravity and seismic load resultant force
VIII-2 – ASME BPVC Section VIII Division 2 h – Elevation where the static head is zero pressure
ASCE – American Society of Civil Engineers o – Elevation of the origin the where the coordinate
UBC – Universal Building Code transformation (rotation) is taking place (i.e., pivot point)
FFS – Fitness For Service F – Resultant from Gravity and Seismic Load
RSF – Remaining Strength Factor

1 Copyright © 2017 by ASME and

Shell Global Solutions (US) Inc.
 FShear – Shear Load per ASCE 7-05, without higher mode water during the quench, and finally empty after decoking,
effects could have an effect on the structural stability of the coke drum.
MSEIS,Tilt – Seismic moment in tilted configuration However, in regions of high seismic activity, large masses can
MCalcASCE7-05 – Moment per ASCE 7-05 whereby shear experience significant accelerations. In these regions a robust
force is integrated over the height of the vessel. skirt, anchor bolts, and base ring and compression ring are
MMakeupMoment – Additional moment required to account for needed to prevent overturning of the coke drum.
dynamic amplification per Section 12.8 of ASCE 7-05.
CM – Center of Mass This paper discusses an assessment performed on 2 coke drums
y’(CM) – Elevation of Center of Mass in a refinery in a high seismic zone that was concerned about
X – Global Horizontal Coordinate the reduction in the original design margin of the seismic load
Y – Global Vertical Coordinate with the existing tilt angle, and needed to determine the critical
x’ – Local or Model Horizontal Coordinate tilt angle when the design margin was no longer acceptable1. To
y’ – Local or Model Vertical Coordinate address this problem, a finite element model of the actual
geometry and the existing bulges was generated from an
INTRODUCTION internal laser scan.
Coke drums are large cylindrical pressure vessels that are
used to upgrade residuum at refineries by thermally cracking The assessment per API 579-1 was performed in three different
the feed (residuum) and recovering light hydrocarbons that are configurations:
sent back for fractionation and further economic gain. Coke
drums typically operate in pairs in a process where the vessels a) “Hot” with contents to the fill height at a design
are operating out of phase with one another. Therefore, a pair of temperature of 875°F.
coke drums provides the refinery a continuous “upgrading” b) “Cold” and filled to the top of the overhead
process, but the actual pressure vessels operate in a batch mode nozzle, prior to unheading at a design temperature
(1). of 300°F.
c) Empty at a temperature of 90°F.
Seismic and wind calculations are significant loadings for these
vessels due to their large size, with 30-foot diameter and 100- The significance of the three cases described above is that both
foot tall pressure vessels being commonplace. In addition, these the magnitude and the distribution of the seismic loads are
pressure vessels are filled with the residuum coming into the different for each case2. Case (a) will have the material in the
vessel at temperatures of approximately 900°F, and a modest weakest condition due to the highest design temperature; case
internal pressure. In these units the actual formation of coke is (b) will have the highest mass, and therefore the highest vertical
delayed by the flowing fluid until it reaches the wide coke drum load, but also the highest accelerated mass, and case (c) will
and begins to solidify, hence the more appropriate name have the least mass, and the “easiest” mass to overturn. Figure
Delayed Coker (1). As the feed continues to enter the drum 1 shows an example simplified plot of total pressure
until a predetermined fill height, thermal gradients generated by (hydrostatic plus internal gas pressure) at the bottom flange vs.
non-uniform cooling of the coke bed are imparted on the vessel time where the cyclic nature of the mass variation vs. time can
itself. Once the upgrading process has completed, the resulting be appreciated3.
solid – petroleum coke – must be cooled and removed from the
drum, before the next cycle begins. The cooling – termed
quenching – also leads to significant thermal gradients, which
in turn cause localized stresses from hot/cold spots. The
cumulative effect of these hot/cold spots after many cycles
often times results in permanent deformations known as bulges,
eventually leading to a myriad of failure mechanisms (3 – 10).
In addition, a preferential and permanent deformation of the
drum shell, also known as “the banana effect” is not an unusual
consequence. Furthermore, the banana effect has been
associated with single side inlet entry of the feed in the coke
drum, although the effects of single side inlet entry will not be
specifically addressed in this paper.

This preferential deformation, aside from generating high

stresses in piping, and potentially overloading the skirt, the skirt 1
At the time of analysis, the actual tilt angle had not been characterized.
to shell weldment, and even anchor bolts, also alters the way 2
The center of gravity in each condition changes, as the shear force
seismic and wind loads act on the drum. The constant change in distribution, the magnitude and the accelerated mass.
3
This is an aggressive cycle due to the short cycle time, and resulting
liquid level, first from coke alone during the fill, then coke + severe thermal transients.

2 Copyright © 2017 by ASME and

Shell Global Solutions (US) Inc.
 Table 1: Summary of Coke Drum Design Data.
Coke Drum Design Data
Design Code 1992
Year of Construction 1994
Design Temperature 875°F(Coking)/300°F(Quench)
Internal Pressure 50 psi
Fluid Specific Gravity 0.88(Coke);1.48(Coke +Water)
Wind Design Data ASCE 7-88, Exposure C, 70
mph, I=1.0
Earthquake Design Data UBC-1991, Zone 4, I=1.0,
Z=0.4g, S=1.0
Shell/Top and Bottom SA-387 Grade 11, Class 2; SA-
Heads/Skirt Material 240 410S Cladding on Wetted
Surfaces
Base and Compression SA-516-70
Figure 1: Example Total Pressure at the Bottom of the Blind Flange vs. Rings/ Chair Gussets
Cycle Time Anchor Bolts 40 x 3-inch Diameter, SA-193-
B7
Three Level 3 analyses utilizing API 579-1 / ASME FFS-1 Shell ID 26 Feet (including Cladding)
(2007) (2), (herein referred to as API 579) were performed for Shell Nominal Thickness 1.082 inches – 0.905 inches,
each coke drum – two total, for each case (a), (b) and (c) above. plus 0.110 inches of Cladding
As previously mentioned, the aim of the assessment was to OAL 103 Feet 6 Inches
determine the critical tilting angle at which the seismic/wind
Empty Weight (Erection) 480,000 lbf
loads no longer had an acceptable design margin. The coke
Quench Weight(Coke + H2O) 3,878,000 lbf
drums were designed and fabricated to the ASME Boiler and
Operating Coke Weight 1,925,000 lbf
Pressure Vessel Code, herein referred to as “the Code” Section
VIII, Division 1 (VIII-1) 1992 Edition including the 1992
Addenda (11). As such, the original design margin on plastic
collapse was 4.04. Using the API 579 rules, an “allowable” The procedure for the assessment was per the description
Remaining Strength Factor (RSFa) of 0.9 was used, and in API 579 Annex B1 in Section B1.2.4. The material properties
therefore a design margin for existing equipment was accepted used for the shell, skirt, base ring, compression ring and
by the refinery as 3.6 (2). adjoining gussets were Elastic-Plastic (EP) determined using
the procedure described in Annex 3-D of VIII-2 (12) for the
temperatures described above in cases (a), (b) and (c). The
material model of the SA-193 B7 bolts included elastic-
METHODS
perfectly plastic (EPP) response for the sake of conservatism; in
Finite element analyses (FEA) were performed on two
case the bolts became overloaded, the failure of each individual
coke drums at a refinery located in a zone of high seismic
bolt could be accounted for in the analysis. In the assessment of
activity to determine the angle of tilt of the drum relative to the
the vessel and skirt system, it was the intention of the authors to
concrete tabletop at which the pressure vessels no longer met
also consider a laser scan of the skirt; however, the refinery
the design margin of 3.6 against plastic collapse. The design
could not wait the required time to obtain this data in a cost-
margin was determined by applying the guidelines in API 579
effective manner. Therefore, to account for “code acceptable”
2007 edition5, as well as the refinery operating the coke drums
imperfections in the skirt, a 1% ovality was included in the skirt
accepting the use of an RSFa of 0.9. The input data for the
geometry. It was acknowledged that if the skirt contained any
geometry was obtained from an internal laser scan of each of
buckling, the results of an “unbuckled” but oval skirt would be
the two coke drums. The laser scan data was collected in
unconservative, and the analysis would have to be repeated.
February of 2016; therefore, the data was a fairly accurate
Therefore, the concrete tabletop was assumed to be level, and
representation of the current geometry of the drums. Table 1
the skirt was assumed not to have buckled. Further verification
below shows the main design details of the coke drums.
validated these assumptions. Figure 2 shows the EP material
model used for the shell and skirt, while Figure 3 shows the
material properties used for the SA-193 B7 bolts.

4
For equipment designed after 1999, the original design margin is
reduced to 3.5 instead of 4.0.
5
The analyses were performed in April and May of 2016, prior to the
2016 version of API 579-1 / ASME FFS-1.

3 Copyright © 2017 by ASME and

Shell Global Solutions (US) Inc.
 Figure 4: Geometry Details of the Finite Element Model Used in the
Assessments.

Figure 2: Elastic-Plastic Constitutive Model for Shell and Skirt Materials The loading for the assessments was in accordance with
at Analysis Temperatures.
Table B1.4 of API 579. In discussions with the refinery, live
loading from piping was not considered as its effect on the
plastic collapse due to tilting from seismic loading was
determined to be small. Similarly, snow loads were also not
considered. Thermal expansion effects were included by
considering the height of the vessel at its corresponding
temperature. All material properties were analyzed at the
corresponding temperature for each case. The finite element
mesh had approximately 86,000 elements. Figure 5 shows an
example of the finite element mesh. Figure 6 shows the mesh
viewed from the top, as well as the earthquake and wind
loading direction relative to the ovality in the skirt. In addition,
for conservatism, the weak axis of the skirt ovality was aligned
with the bulge having the greatest deformation for both drums.
Figure 7 shows the radial coordinates in inches for each drum
analyzed. Note that the nominal radius for each drum is 156
inches.
Figure 3: Elastic-Perfectly Plastic Constitutive Model Used for the Bolting.

The analyses were performed using the ABAQUS general

finite element method (FEM) software, version 6.14. The
element type used for the shell, skirt, compression ring, base
ring and gusset plates in the chair region were shell elements,
type S4, with four integration points in the plane, and 5 section
points through the thickness. The element-type used for the
bolts were beam elements in space, type B31. Figure 4 shows
the different sections of the finite element model.

Figure 5: Finite Element Mesh Example Used for Both Drums

4 Copyright © 2017 by ASME and

Shell Global Solutions (US) Inc.
 constrained to a node on the compression ring plate to simulate
the presence and effect of the nuts.

Table 2 shows the load cases analyzed for each coke drum,
where the number specifically refers to the Global Criteria per
Table B1.4 in API 579. In addition, the allowable remaining
strength factor (RSFa) was 0.9 as agreed by the refinery, and the
original design margin for Code vessels built prior to the 1999
Edition of the ASME VIII-1 is 4.0 (2). Therefore, the combined
load factor coefficient β was equal to 3.6.
Figure 6: Finite Element Mesh Viewed from the Top; Note the Wind and
Earthquake Loading Directions Relative to the Permitted Ovality in the Table 2: Load cases Analyzed for Each Drum (2).
ASME Code. Operational Load Case Required Factored Loads
Condition
Coke LC 4 – Coke 0.86β(P+Ps+D)+1.1βW
LC 5 – Coke 0.86β(P+Ps+D)+0.71βE
Quench LC 4 – Quench 0.86β(P+Ps+D)+1.1βW
LC 5 – Quench 0.86β(P+Ps+D)+0.71βE
Empty LC 4 – Empty 0.86β(P+Ps+D)+1.1βW
LC 5 – Empty 0.86β(P+Ps+D)+0.71βE

The weight of the contents of the drums was modeled using a

hydrostatic pressure load with the same magnitude as defined in
the design conditions in the drawings. Please note that the
weight of the bottom unheading device was neglected in this
analysis – the impact of this omission is expected to be
negligible since the weight of the bottom unheading device is
on the order of 1% of the weight of the operating drum. The
factored seismic and gravity loads were applied as body forces
at the center of mass of the coke drums. The “unfactored”
gravity load is “1.0g”, where g is the acceleration due to
gravity. The load factor acting on the dead loads is equal to
0.86. The “unfactored” seismic load “E” is equal to 0.36g, and
was calculated using UBC-1991. The load factor acting on the
seismic load is 0.71 – See Table 2 above. Therefore the angle of
the factored resultant between gravity and seismic loads is
equal to the inverse tangent of the ratio of the factored seismic
load to the factored gravity load. This angle – termed the
equivalent tilt angle ƟEqTilt of the resultant body force between
the factored gravity and the factored seismic horizontal
acceleration was calculated as 16.55°. The finite element model
was therefore rotated by the Equivalent Tilt Angle ƟEqTilt6. Any
additional tilt angle (i.e., ƟRealTilt) is added on this resultant
angle. Ultimately, a series of finite element analyses with
increasing Equivalent Tilt Angle were run until convergence of
Figure 7: Radial Coordinates in inches for Both Drums Analyzed Based on
the load cases in Table 2 was such that the API 579 Load Factor
3D Scan. Nominal Radius is 156 inches. Coefficient β was reasonably close to 3.6. Note that for load
cases without the seismic load (i.e., wind), the ƟRealTilt and
The boundary conditions were defined as having contact ƟEqTilt were equal. Figure 8 shows an image and free body
between the concrete – modeled as a rigid surface, and the base diagram of the tilted coke drum with local and global
ring. This contact pair definition was allowed to come into coordinate systems, and the resulting equations to calculate the
contact as well as out of contact to accommodate lift-off of the angles (equations 1–3). The analyses were carried out in the
vessel. Furthermore, the bottom node of each bolt element was global coordinate system with the model rotated to the ƟEqTilt
fixed in all directions to simulate the restraint in the concrete
for each anchor bolt. Finally, although the nuts were not
explicitly included in the model, the top node of each bolt was 6
For a real tilt angle of 0°, the equivalent tilt angle was 16.55°.

5 Copyright © 2017 by ASME and

Shell Global Solutions (US) Inc.
angle, and the resultant (seismic + gravity) load directed in the (ELF) procedure given in section 12.8 of ASCE 7-05 matched
local negative vertical direction. with the original design loads. The ELF procedure is a
simplified static analysis procedure that accounts the effects of
the inelastic dynamic effects due to the seismic loads.

The vertical distribution of seismic force was calculated per

Section 12.8.3 in ASCE 7-05 (13).

Equation 12.8-11 (ASCE 7-05)

Equation 12.8-12 (ASCE 7-05)

∑

Equation 12.8-12 is based on the simplified first mode shape of

a cantilevered column. The deformed shape of the cantilevered
structure is a function of the exponent , which is related to the
fundamental period of the structure . The exponent k is
intended to approximate the effect of higher modes, which are
generally more dominant in structures with a longer
Figure 8: Free Body Diagram and Relationship Between Coordinate
Systems of Coke Drum Subjected to Seismic Loads.
fundamental period of vibration. The horizontal forces
computed using Eq. 12.8-11 do not reflect the actual inertial
The coordinate transformation above also required some slight forces imparted on a structure at any particular point in time.
adjustments to the static pressure definition, as defined in Instead, they are intended to provide lateral seismic forces at
Equation 4 (see nomenclature section). Please note that the individual levels that are consistent with enveloped results from
elevation of the origin “o” is simply an origin, and not the more accurate analyses.
elevation of maximum pressure; the elevation of maximum
pressure is at the bottom flange. Since the vertical distribution of the shear force is nonlinear, a
small make-up moment will be needed to ensure the base
This simple derivation is convenient because as fluids always moment calculated using the simple derivations match with the
behave according to their local gravity environment, rotating ASCE 7 calculations.
the model about the origin {o} – elevation at which the rotation
is applied, ƟRealTilt – provides a robust accounting of all internal Equations (5), (6) and (7) show the relationship between the
fluid loads and all loads that are subject to any generic applied moment from seismic loading due to the tilt; the
acceleration vector. applied moment calculation per ASCE 7 using a linear vertical
distribution, and the required “Make-up Moment” when using
The objective of this assessment was to estimate the reduction ASCE 7 Section 12.8 to include the dynamic amplification
in the design margin from the as designed condition. Therefore, factor in the finite element model.
the seismic and wind design loading of the coke drums was
based on the original design values and design codes rather (5)
than the latest design standards. This approach is justified since
there is no significant change in the stiffness and loading from
the original design condition. Based on the design data shown (6)
in Table 1, the design seismic loads were obtained from the
UBC-1991 code – Universal Building Code, and the wind
design loading was based on the ASCE 7-88 Standard – (7)
“Minimum Design Loads for Buildings and Other Structures”
published by the American Society of Civil Engineers (ASCE). Lastly, as a clarification for the acceptance criteria, the design
The design wind and seismic loads (base shear and bending margin is calculated as shown in Equation (8):
moment) were confirmed by taking the coke drum design data,
and independently verifying those using existing pressure
vessel design software packages.
(8)
Original design calculations were not available to verify the
analysis procedure used to obtain the base shear force and The remaining Strength Factor, which the allowable value is
bending moment at the base plate. The base shear force and defined as 0.9, and accepted by the refinery is determined as
bending moment calculated using the Equivalent Lateral Force shown in Equation (9):

6 Copyright © 2017 by ASME and

Shell Global Solutions (US) Inc.
 Table 3: Summary of Wind and Seismic Loads.
(9) Shear Force at Base Ring Bending Moment at Base
(lbf) Ring (kip-ft)
Wind Seismic Wind Seismic8
Figure 9 shows an exaggerated depiction of the hydrostatic
ASCE 7-88 UBC-1991 ASCE 7-88 UBC-1991
pressure distribution in the equivalent tilt configuration. Figure Coke 22,471 881,715 961.9 38,329
10 shows a depiction of the wind pressure for the load cases Condition
with wind loading. Quench 22,471 1,597,122 961.9 86,088
Condition

RESULTS
The results of the assessments are presented next. First and
foremost, validation of the inputs was sought by checking the
reaction forces obtained in the FEA vs. the expected reaction
force from a hand calculation. Figure 11 shows the excellent
correspondence between the hand calculation and the finite
element model analyzed. Note that the magnitude to the loads
are inclusive of the factors per Table B1.2 in API 579. Table 4
shows a summary of the results for each load case analyzed for
one drum; however, the results for the second drum were nearly
the same – as expected. Additionally, as can be seen in Table 4,
the seismic case during the quench condition is the governing
condition, and actually did not pass the original design margin
of 4.0, obtaining an RSF of 0.93. An RSF of 0.93 corresponds
to an original design margin of 3.72 using the ASCE 7-05,
paragraph 12.8. Therefore, the detailed results will be focused
Figure 9: Depiction of Static Pressure in Coke Drum at Equivalent Tilt only on this condition9 (LC5-Q). Additional iterations of this
Angle. Note that the tilt angle is exaggerated simply to show the effects it
has on the static pressure.
load case were run until the convergence reached a value as
close to 3.6 as reasonable – the last real tilt angle analyzed per
paragraph 12.8 of ASCE 7-05 was 1.3°, which gave a plastic
collapse margin of 3.57, or an RSFa of 0.893.

Figure 11: Overturning Moment Check at the Base Ring: Seismic

Figure 10: Illustration of Wind Loading in Load Cases with Wind Loading. Condition.

Finally, Table 3 shows the unfactored7 Seismic and Wind loads

per their applicable codes. (LRFD) methodologies to account for lower probabilities for loads to act at
their maximum magnitude simultaneously.
8
Equivalent Lateral Force Method (ASCE 7-05 Paragraph 12.8)
7 9
This is the actual design load. The applied load in the FEA is per API This was expected; however, for Code compliance and completeness, all
579, which includes the load factors using Load Resistance Factor Design other applicable cases were assessed.

7 Copyright © 2017 by ASME and

Shell Global Solutions (US) Inc.
 Table 4: Summary of Results at a Tilt Angle of 0°. model was predicting a shear-like failure at these bulges from
Oper. Load Required Load Factors RSF RSFa the significant seismic acceleration of the top-heavy vessel.
Condition Case
Coke LC4-C 0.86β(P+Ps+D)+1.1βW > 1.0 0.9
LC5-C 0.86β(P+Ps+D)+0.71βE > 1.0 0.9
Quench LC4-Q 0.86β(P+Ps+D)+1.1βW > 1.0 0.9
LC5-Q 0.86β(P+Ps+D)+0.71βE 0.9310 0.9
Empty LC4-E 0.86β(P+Ps+D)+1.1βW > 1.0 0.9
LC4-E 0.86β(P+Ps+D)+0.71βE > 1.0 0.9

Figure 12 shows a plot of the load factor β vs. Total Equivalent

Plastic Strain (PEEQ) for the coke drums at a Real Tilt angle of
1.3° and 0.0°. In addition load factors of 3.6, 1.0 are also
shown.

Figure 13: Contour plot of PEEQ of Both Analyzed Coke Drums at the
Critical Tilting Angle using Section 12.8 of ASCE 7-05.

CONCLUSIONS
In conclusion, this study has outlined a practical approach
to assessing bulged and permanently deformed leaning coke
drums subjected to seismic loads, while simplifying the
treatment and application of gravity and seismic accelerations.
The equivalent lateral force procedure provides a simple way to
account for effects of the inelastic dynamic response in a static
analysis. The methodology used in this paper is limited in that
Figure 12: Load Factor vs. PEEQ for Analyzed Coke Drums for Real Tilt
Angles for Seismic Load Case 5 in Quench Condition.
the moment applied must be supplemented to account for the
non-linear distribution of the shear force in the upper
Figure 13 shows a contour plot of PEEQ and a graphical elevations, making the calculation of a simple, but necessary
depiction to scale of what a tilt angle of 1.3° looks like. It is make-up moment in order to obtain the correct loading. Finally,
important to remember that this result was obtained using from a practical perspective, the results of this study show that
section 12.8 in ASCE 7-05. Although modest, it is important to the required angle of tilt required to jeopardize the seismic
note that with a vessel of over 100 feet tall, a tilt angle of 1.3° is stability of the structure even in high seismic regions, although
quite significant. In fact, the top nozzle has a diameter of 36 modest, is still large enough where operational problems with
inches, and the corresponding misalignment at the top nozzle drilling and cutting will be early symptoms that leaning is
from a 1.3° tilting angle is approximately 24.375 inches. These occurring.
results show that this refinery would be unable to insert the drill However, there are additional considerations with these
stem inside the coke drum to initiate the cutting process and conclusions, namely:
still not exceed the critical angle, assuming the skirt is not
buckled, the anchor bolts are sound, and the skirt to cone/shell  The skirt was conservatively modeled in a
junction is not compromised. In addition, a 24.375-inch configuration where the maximum ovality of the
horizontal misalignment at the top plane may also prove to be vessel was also present on the skirt, aligned with
problematic for the piping – this was not addressed specifically the worst bulge, and the loading occurring in the
in this paper, since the refinery did not have this severity of the weakest direction. A buckled or even partially
tilt angle, but this assessment could be performed and buckled skirt would make the critical tilting angle
recommendations made to avoid compromising the piping smaller than predicted. However, the refinery
based on maximum deformations/loads. Regardless, it is also performed follow-up inspection, and found the
important to highlight that the actual failure mode for both skirt geometry to be in good condition with no
drums, was the largest deformed bulge, which for both drums deformations; therefore, these results show that
happened to be at the bottom shell course. The finite element the continued operations of the coke drum on this
basis was justified.
 Existing inspection results of the skirt to shell
10
Lowest RSF – worst case scenario. weldment did not show evidence of cracking.

8 Copyright © 2017 by ASME and

Shell Global Solutions (US) Inc.
  The anchor bolts, concrete, base ring and (11) ASME, “Section VIII Division 1 Rules for
compression ring structures did not show Construction of Pressure Vessels”, 2015 ASME, New
evidence of degradation. York, N.Y.
(12) ASME, “Section VIII Division 1 Rules for
Therefore, the refinery’s proactive, and swift action enabled Construction of Pressure Vessels”, 1992 ASME, New
them to make an informed decision about the continued York, N.Y.
operation of these coke drums. (13) ASME, “Section VIII Division 2 Rules for
Construction of Pressure Vessels”, 2015 ASME, New
York, N.Y.
ACKNOWLEDGMENTS (14) ASCE Standard ASCE/SEI 7-05, “Minimum Design
The authors wish to recognize the contributions of Mr. Loads for Buildings and Other Structures”, 2005,
Luiz C. Largura Jr. P.E., for having performed finite element Reston, VA.
work, and to Shell Global Solutions (US) Inc. for having
funded this work.

REFERENCES

(1) Leffler, W.L., Petroleum Refining in Nontechnical

Language, 4th Edition, 2008, PennWell Corporation,
Tulsa, OK.
(2) API 579-1 / ASME FFS-1 Standard for Fitness For
Service (2007) American Society of Mechanical
Engineers, New York
(3) N.A. Weil and F. S. Rapasky, “Experience with Vessels
of Delayed-Coking Units”, Proceedings American
Petroleum Institute, 1958, pp. 214-232, vol 38 [III],
American Petroleum Institute
(4) 1996 API Coke Drum Survey (2003) American
Petroleum Institute, Washington, DC.
(5) Farraro, T. and Boswell, R.S. (1996) “Determination
of Coke Drum Fitness-For-Service and Remaining
Life”, API Spring Meeting, Operating Practices
Symposium, American Petroleum Institute,
Washington, DC
(6) Ramos, A., Rios. C.C., Vargas, J., Johnsen, E.,
Gonzalez, M., “Delayed Coking Assessment Using
Field Measurements & FEA”, PVP v 368 1998: 231-
237 ASME, New York, NY
(7) Penso, J.A., Lattarulo, Y.M., Seijas, A.J., Torres, J.,
Howden, D., and Tsai, C.L., “Understanding Failure
Mechanisms to Improve Reliability of Coke Drums”,
PVP v 395 1999, ASME, New York, NY
(8) Antalffy, L.P., Malek, D.W., Pfeifer, J.A., Stewart,
C.W., Grimsley, B., Shockley, R., “Innovations in
Delayed Coking Coke Drum Design”, PVP v 388
1999, ASME, New York, NY
(9) Boswell, R.S., Wright, B., “State-of-the-Art
Improvements in Coke Drum Design and Life
Extension Practices”, Proceedings of CREEP8, 2007,
ASME, New York, N.Y.
(10) Pieper, C.J., Shockley, L.R., Stewart, C.W., “Coke
Drum Design – Longer Life Through Innovation”,
AIChE Spring National Meeting, 2000, AIChE, New
York, N.Y.

9 Copyright © 2017 by ASME and

Shell Global Solutions (US) Inc.

View publication stats