Determination of Coke Drum
Fitness for Service
Richard Boswell, P.E.
Stress Engineering Services, Inc.
CITGO Refinery Lake Charles, LA
CITGO F-201 COKE DRUMS
n Built 1968 n ASME VIII/ Div 1
n Diameter = 21 feet n 7 Plate Rings
n Cyl Length = 70 feet n Top : 1” Clad plate
n Top : 60 psi at 836 oF n Bottom : 1.64” Clad
n Bottom : 60 + Hydro n Mat’l : SA-515 Gr. 55
at 899 oF n Clad : 7/64” 405 SS
n 18 hr Coking Cycle n Replaced in 1996 after
How Do Coke Drums Fail?
n Primary failure mechanisms are:
– Bulging and distortion of shell plates typically
20 - 40 ft above the skirt attachment weld.
– Circumferential cracking adjacent to welds and
bulges both (OD and ID initiated).
– Cracking and bulging in area of skirt to shell
Why do Coke Drums Fail?
n Typically coke drums are designed as
pressure vessels utilizing the ASME
Pressure Vessel Code.
n The ASME code assumes that internal
pressure is the primary (largest) stress to
which a pressure vessel is subjected.
Actual Measured Coke Drum Shell Stress
Maximum Hoop Stress
due to Pressure
Erroneous Design Assumptions
n Erroneous Assumption # 1
– Primary stress is a coke drum is caused by internal
n Erroneous Assumption # 2
– The 3 to 1 safety factor built into the ASME code is
sufficient to accommodate “secondary” stresses
caused by thermal gradients, fatigue and bending.
How Can We Identify the Source of
the Primary Stress in Coke Drums?
n Dynamic modeling of coke drum operation using
the finite element method (FEM)was utilized to
determine source of the high measured stresses.
n Analysis was based on actual shell stress/strain
data obtained by monitoring 300 + coke drum
cycles 1.5 (Years)
Results of FEM Analysis
n FE analysis indicated 3 sources of high
stresses measured on coke drums:
– Localized hot spot (thermal gradient stress)
– Coke/shell differential shrinkage stress
– Skirt/shell attachment (thermal gradient stress)
FEM Results (bulge analysis)
n Corrugations influence membrane and surface
stress across bulge.
n Higher stress due to larger diameter.
n Higher stress due to ring bending moments.
n Larger membrane stress at “valley” of bulge.
n Circ. welds are typically at “valley”.
Life Limiting Factors
n Low Cycle Fatigue Life.
n Reduction of wall thickness in bulge
n “Squatting” and leaning of drum due to
Low Cycle Fatigue Life
n Hoop and axial stresses are cyclic and
average 40-60 ksi.
n Yeild strength of the base metal was found
to be 32ksi.
n Low cycle fatigue was determined to be a
life limiting factor.
Stages of Fatigue Life in Steel
Length of Crack in.
1st Stage 2nd Stage Stage
0 10 20 30 40 50 60 70 80 90 98
Fatigue Life Expended (%)
Fatigue Life Determination
n Drums operated at 24hr cycle for 20yrs and 19
hour cycle for 8 years.
n Total cycles on drums at present is 1845 + 3652 =
n Using 50 ksi as average cyclic stress fatigue life
from ASME VIII Div 2 curves = 5000 Cycles.
Fatigue Limits for Carbon and
Low Alloy Steels
Applied Stress (psi)
10 100 1,000 10,000 100,000 1,000,000
# of Cycles
Laser Profile Map of Coke Drum
Bulge Growth Rate
Shell ID 126”
Wall Thickness in Valley Between Bulges
1.44” nom .
n Thermal gradients during quench are not
n Cylindrical flexibility accommodates
shrinkage without high stress
n Thermal gradients through wall are reduced
by inertia and insulation
FITNESS FOR SERVICE ISSUES
n How do bulges affect structural integrity?
n How predictable is crack/bulge growth?
n Can End of life be predicted?
n Can we continue to operate these vessels in a
reliable, and predictably safe manner?
n Over 300 cycles monitored with Strain
Gages and Thermocouples
n Hoop and Axial stress range +/- Yield
n When range is higher, defect damage can be
n Limited monitoring of a few cycles could
capture minimum loading : false security
n F-201 B Nominal inside radius = 126”
n November ‘94 survey : 8.4% bulge
n May ‘95 survey : 9.5 % bulge
n Other profiles did not increase this much
n Vertical and Circumferential surveys
n At circ. weld, wall thinned by 15% for 6”
both above and below
n No significant thinning at bulges
n Drums are shortening
n A series of finite element models made for
n Drum is a corrugated cylinder
n Manually digitized at 1 foot intervals
n Axisymmetric shell model in ABAQUS
n Loaded by internal pressure with
hydrostatic gradient to bottom
n Corrugated Cylinders Have Higher Stress
than Straight Cylinders
n Consider Ring Bending Moment Stress as a
n Compare Pressure + Ring Bending to
Primary Bending Stress Allowable
n New Pressure Rating Should be Considered
n Awareness of operation and response is
essential for FFS evaluation
n Drums experience very high short duration
stress during quench
n Occasionally this stress is very large
n Inspection is justified after large stress
Stress Engineering Services 11
n Long term results defined load conditions
for new drums
n Designed beyond the scope of a pressure
n Improved economics and safety
n Greater reliability
n Less maintenance, longer life
TYPICAL COKING CYCLE
Typical Stresses During Quench Cycle
OPERATING HOOP STRESS
OPERATING AXIAL STRESS