Page # 1 of 11

Fatigue Life Evaluation of Coke Drum Support Skirt / Shell Junction Using Transient Thermal Stress Analysis

HOT- BOX & WELD BUILD-UP

Note: The enclosed general information is for reference only. The details of project specific calculations will be provided “later”

FOSTER WHEELER USA CORP. 2020 DAIRY ASHFORD, HOUSTON, TEXAS 77077

May 22, 2007

Page # 2 of 11

1.0 INTRODUCTION

1.1 SCOPE

An analytical study is performed to estimate the fatigue life of support skirt to

shell junction for Coke drums based on transient thermal / mechanical stress

calculations and using finite element analysis method. The fatigue life is

estimated based on the actual operating skin temperature.

This engineering analysis also provides specific recommendations to optimize

specific process and operating parameters so that the life of skirt / shell junction

can be extended and fatigue damage reduced.

Although high stresses and cracking happen in nearly all coke drums due to

thermal cyclic fatigue, the damage can be minimized by controlling operating and

process parameters.

1.2 BACKGROUND

Coke drums undergo severe thermal and pressure cycling on a daily basis due to

alternative pre-heating, filling with coke, quenching and decoking operation. This

cyclic mode of operation results in significant thermal stresses at the support skirt

attachment to the drum.

Transient temperature gradients and stress reversals are developed in the skirt

during the cyclic cooling and heating process leading to thermal fatigue failure. If

the temperature ramps during quench and / or preheat / switch to coking cycles

are severe, it will create high thermal fatigue stresses resulting in premature skirt

junction cracking.

It should be noted that the ASME code fatigue data for welded vessel uses a

factor of safety of 2 on stress and 20 on cycles.

Page # 3 of 11

1.3 ASSUMPTIONS

The effect of the following parameters is not included in the analysis:

- Stress concentrations unintentionally introduced during fabrication

- Weld metal physical properties same as the base metal properties.

- Occasional operating “upsets” and wind loads are ignored.

- Only the effect of thermal, pressure and weight loads are included.

- Fatigue strength reduction factor (FSRF) of weld between the skirt and

drum at inner crotch is assumed to be equal to 1.0. FSRF depends on

the weld type and inspection level. For some coke drums, FSRF = 2.0

are required. This will lower the fatigue life.

Page # 4 of 11

2.0 RECOMMENDATIONS TO INCREASE COKE DRUM FATIGUE LIFE

Using transient thermal data of actual operating condition, the results of transient

thermal stress analysis based on finite element techniques provide an insight into

specific operating / process factors which have significant impact on fatigue life

of skirt / shell junction.

The analysis results provide the operator with optimum transients and control

parameters to improve the fatigue life of skirt / shell junction as noted below.

- Extended pre-heat and higher temperature after pre-heat has significant

positive effect on skirt junction fatigue life.

- Maximize the duration of switch to coke to minimize the temperature ramp.

- Faster quench rate has significant negative impact on skirt junction fatigue

life.

Note that there is a temperature gradient reversal in going from quench to

preheating / switch cycle due to transient cooling / heating sequence. This causes

a stress reversal and therefore higher stress range which leads to lower fatigue

life.

By combining transient thermal / stress analysis results and recommended process

parameters, the reliability of the existing coke drum operation can be improved.

Page # 5 of 11

3.0 Transient Thermal Data during Coking Cycle

The transient skin temperature data for coke drum is taken from FW data base of

operating unit and summarized below. Thermal ramps during quenching and

preheating / switch to coking cycle are noted in the following tabulation:

(The following data was used for transient thermal stress analysis)

Operating Event

Time

(min)

Operating temp

Temperature Rate of temp

change,

(0F / min)

# Description Duration 0C

0F

1

Quench # 1:

- Steam to

Fractionator

- Steam to

Blowdown

2

Quench # 2

Quenching and

Filling

3 Drain, Drill,

Reheat, Test

Does not govern thermal stress or fatigue life.

4 Preheating /

Switch to Coke

Page # 6 of 11

4.0 FATIGUE LIFE EVALUATION USING TRANSIENT THERMAL

ANALYSIS

Coke drums undergo severe thermal and pressure cycling on a daily basis when

subject to alternative pre-heating, filling up with coke, quenching and then

decoking operation. This cyclic mode of operation results in significant thermal

stresses at the support skirt attachment to the drum.

Temperature gradients are developed along the skirt during steady state and

transient thermal conditions inside the coke drum. Higher thermal gradient will

lead to higher thermal stresses and lower fatigue life. Transient temperature

gradients developed during the cyclic quenching and heating process are reverse

in nature and therefore cause reversal in bending stresses imposed on skirt.

In general, the magnitude of thermal stresses induced due to thermal condition of

coke drum during quenching and preheating / switch to coking are much higher

compared to steady state thermal condition.

Therefore, the fatigue evaluation of coke drum is governed by thermal transient

analysis. It is determined that the weld between the skirt and shell at the inner

crotch is subjected to highest bending stress reversal due to alternate cooling and

heating.

Stress reversal effect is considered to determine the stress range and to evaluate

fatigue life.

Finite element analysis is used to evaluate the thermal and stress profiles for the

transient conditions. Thermal cycle includes transient cooling condition and

transient heating condition.

A 2-D axi-symmetric FEA model is used for thermal and stress evaluation. The

FEA model also includes insulation, fireproofing and “hot box” area. Two models

are created for each design, namely, a thermal model and a stress model. In the

thermal model a determination is made of the transient temperature gradients in

the skirt during the heat up and quench cycle.

The temperature gradients from thermal model is then exported to the Stress

model to determine thermal stresses. Three cases are considered in the stress

model: 1) the thermal case alone, for heat up and quench conditions; 2) an internal

pressure case acting with dead and live loads in the drum and 3) the combined

thermal, internal pressure plus dead and weight loads during the heat up and

quench part of the coking cycle.

Page # 7 of 11

Upon determination of the maximum stresses for both transient conditions, the

stress reversal effect is considered to determine the full stress range and to

evaluate fatigue life.

The allowable criteria for the evaluation of fatigue due to peak stress is twice the

stress amplitude using fatigue curves from ASME Section VIII, Division 2,

Appendix V.

The results of this thermal / mechanical stress analysis are used to perform fatigue

evaluation at critical locations.

Load Cases and Load Combination

For thermal / mechanical stress analysis, following load cases are analyzed:

- Thermal gradient only

- (Pressure + Weight)

- (Thermal Gradient + Pressure + Weight)

CRITICAL LOCATIONS FOR FATIGUE LIFE EVALUATION

Finite element analysis results identify the following the critical location for

fatigue life evaluation:

- Weld between Skirt and Knuckle @ inner crotch (see FEA plot)

In order to comply with the requirements of Appendix – 5, principal stresses are

calculated using FEA stress results, followed by maximum stress intensity and

alternating stress amplitude, Salt, calculations. Salt is adjusted for fatigue strength

reduction factor and for modulus of elasticity at maximum metal temperature.

Page # 8 of 11

5.0 FEA MODEL AND BOUNDARY CONDITIONS

A 2-D axi-symmetric finite element model is constructed using ALGOR software.

The FEA model also includes insulation, fireproofing and “hot box” area (see

FEA plot Figure # 1 and 2).

The FEA model includes sufficient length of shell so that the fixed end boundary

condition at the skirt base plate does not effect the results at skirt / shell junction.

The geometry, Process and mechanical design conditions noted on design and

vendor drawing of Coke drum are used in FEA model.

The stress model consists of the same components, geometry and elements as the

thermal model. The nodal temperatures calculated in thermal model are

transferred to stress model to calculate thermal gradient stresses.

The skin temperature measured by TI’s on shell during current operation

immediately above the lower tangent line or estimated skin temperature for

“future” case is applied to the internal metal surfaces.

Forced heat transfer film co-efficient, h is applied to the outside surface and “hot-

box” enclosure. Hot-box is modeled to include the radiant heat effect using body-

to-body option in ALGOR. The thermal conductivity, K, specific heat, Cp, and

mass density of all components are input in the transient thermal model.

Two models are created for each design, namely, a thermal model and a stress

model. In the thermal model a determination is made of the transient temperature

gradients in the skirt during the heat up and quench cycle. The stress analysis

resulting in maximum stress intensity is determined using thermal profiles at

various time steps iteratively. The weld at inner crotch between skirt and shell is

determined to be the highest stressed location.

For stress model, the pressure loads are applied as surface loads on the inside of

the shell, knuckle and cone. Note that the pressure load also includes the

additional effect of static head due to content inside coke drum. End loads are

applied as equivalent pressure load to account for internal pressure effect and

weight loading acting at the end of shell.

Page # 9 of 11

Page # 10 of 11

Page # 11 of 11