Journal of Pressure Equipment and Systems 4 (2006) 29-32

The failure probability analysis of ethylene cracking furnace tube

based on risk
Wenhe Wang a,*, Shiming Shen a,b
a
College of Mechanical and Power Engineering of Nanjing University of Technology, Nanjing 210009, China
b
Center of Engineering Risk Analysis & Research, SINOPEC, Nanjing 210009, China

Abstract

The ethylene cracking furnace is the primary equipment for the petrochemical industry. The furnace tube is the key component of
cracking in the furnace. Failure can cause serious consequences such as fire and explosion. The Monte Carlo method is used to apply
the risk analysis theory of API 581. The qualitative analysis leads to failure probability of the ethylene cracking furnace tubes the fail-
ure probability of furnace tube. The results supply a reliable foundation for prolonging risk-based inspection period and risk manage-
ment of furnace tube.

Keywords: Ethylene cracking furnace; Furnace tube; Failure probability; Risk.

1. Introduction Carlo method and risk based inspection of API581 are

applied to compute the failure probability of a ethylene
The ethylene cracking furnace is important equip- cracking furnace tube in a petrochemical plant, and the
ment in the enterprises of petrochemical industry; the results of calculation are compared each other for sup-
furnace tube is its important component. The influence plying the reliable basis for risk analysis of the furnace
factors on failure of tube are various under high tem- tube.
perature of tube. The damage of furnace tube includes
creep, carburization and corrosion and so on. Damage 2.1. Failure probability analysis of furnace tube
may result in cracking, break and melting, and bring on based on Monte Carlo method
leakage of high temperature media, and lead to conse- The Monte Carlo method is also called random si-
quently fire and explosion. So failure probability of mulation method, its basic thought is sample with
furnace tube is more concerned. probability density function of the basic random varia-
With the rapid development of risk assessment in ble, the failure probability is defined by comparing
petrochemical industry, the risk analysis and assess- with a sample number of failure and the total sample
ment research for oil refining devices are attracted number [7].The Monte Carlo method is widely applied
more and more people’s attention. Failure likelihood to the reliability computation of the chemical equip-
(failure probability) is a mainly part in risk analysis of ment and analysis of failure probability has no limit in
devices [1, 2]. In the paper, Monte Carlo method and the application scope, so long as the limiting condition
risk based inspection of API 581 are applied to com- equation is known, results obtained by Monte Carlo
pute the failure probability of a ethylene cracking fur- method approach to the fact [8, 9].
nace tube in a petrochemical plant, it provided reliable
data for risk assessment, and insured a long period op- 2.1.1. Parameters and failure causes of the cracking
eration of the furnace tube. furnace tube
The material of the furnace tube is the HP series
2. Failure probability analysis methods of the casting condition, its microstructure is the eutectic car-
cracking furnace tube bide and the austenite [11], internal radius Ri=46.95mm,
external radius Ro=56.95 mm. The operation parame-
The computation methods of failure probability are ters of the ethylene cracking furnace tube: operation
various; there are mainly the first order second mo- pressure Pi=0.106MPa, inner wall temperature
ment methods, the response surface method, Monte Ti=838℃, outer wall temperature To=1050℃.
Carlo method and so on [4-7]. In the paper, Monte By failure mode and effect analysis (FMEA), risk
of the furnace tube is the highest in the entire equip-
* ment, meanwhile, creep and carburization are consi-
Corresponding Author.
E-mail Address: [email protected] (Wenhe Wang). dered as the mainly causes for rupture of the ethylene
30 Wenhe Wang et al. / The failure probability analysis of ethylene cracking furnace tube based on risk

cracking furnace tube [3]. meters etc.). In assessment, a very important point is to
calculate future failure probability according to the
2.1.2. The failure limiting condition equation of the present failure mechanism of the equipment. In
cracking furnace tube API581, there are three assessment methods, qualita-
Cumulated life fraction value is known as evaluation tive, half quantitative and quantitative analysis me-
criteria of the cracking furnace tube, according to Lar- thods [15]. In the paper, the quantitative assessment
son-Miller method and life theory [11], when cumu- method of API581 is applied to predict the failure
lated life fraction value reaches critical life-span frac- probability of furnace tube according to the failure
tion value, the furnace tube is considered as rupture by probability of the apparatus and statistics database of
creep. In the paper, the critical life-span fraction value the incident reason, and appropriate factors.
is used as a criterion and the failure probability of fur-
nace tube is computed [12-14]. 2.2.1 Generic failure probability for furnace tube
By the stress strength interference model, the failure API581 provides the generic failure probability of the
limiting condition equation of furnace tube can be furnace tube about three kinds of size. The failure
written with the formulas (1) and (2) [9, 10]: probability of the furnace tube can be calculated di-
rectly by the similar failure probability of long-term
F (t )  Fcr  0 (1) creep in the technological module of furnace tube. If
the short-term creep is taken place, the failure proba-
bility of furnace tube can be obtained by multiplying
M
L
F (t )   ( ) (2) similar failure probability with coefficient 100.Results
i 1 Li are shown in Table 2 [15].

In these parameters, F(t) is life-span fraction for

Table 1. Failure probability of tube at certain life (under creep &
furnace tubes at time t (hour), Li is rupture life-span carburization).
under stress σt and temperature t, ∆L is operation pe- Operation Times
Failure Probability
riod, Fcr is the critical life-span fraction value. (hours)
33141.94 0.000E+00
2.1.3. The failure probability of the furnace tube 38048.97 4.999E-04
under creep and carburization 43538.02 2.900E-03
The furnace tube failure probability under interfe- 55914.96 5.041E-02
rence of creep and carburization, which is shown as the 64005.95 8.079E-02
non-coupled superposition for circumference stress, is 72343.38 0.1611E+00
91637.16 0.4135E+00
only concerned in this paper. Cumulated life-span frac- 102716.7 0.5498E+00
tion value is as judgment method by Monte Carlo me- 128060.0 0.7893E+00
thod. the failure probability of furnace tube can be ob- 142454.6 0.8669E+00
tained under different operation time [3] (Where opera- 158088.7 0.9209E+00
tion pressure Pi is normally distributed, the variation 193353.7 0.9759E+00
coefficient is 0.1; Ti and T0 are the temperature of in- 281842.3 0.9990E+00
ternal and external walls respectively, are the normally
distributed, the variation coefficient is 0.01; Number of
the sample is 10000 times).
Failure probability at a certain life-span value is
shown in table 1.
Fig. 1 shows that the relation between failure proba-
bility and operation times of the furnace tube. From
Fig. 1, it is seen that the failure probability of the fur-
nace tube is increased constantly with the increasing of
operation times. If regarding 5% of failure probability
as the ultimate life-span, then the service life of fur-
nace tube is 6.4 years.

2.2. Failure probability computation of furnace

tube based on API 581
The assessment of failure probability based on
Risk-based Inspection(RBI) is carried out according to
the failure mechanism of the devices (fluid medium in Fig. 1. Relationship between failure probability and operation
devices, the material of the devices and process para- times of furnace tube.
 Journal of Pressure Equipment and Systems 4 (2006) 29-32 31

Table 2. Generic failure probability for furnace tube.

Hole Size Failure Probability(long-term) Failure Probability(short-term)
6.35mm 0 0
25.4mm 4.62*10-6 4.62*10-4
101.6mm 1.32*10-6 1.32*10-4
Rupture 6.60*10-7 6.60*10-5

Table 3. Failure probabilities of furnace tube for creep.

Hole Size/mm 25.4 101.6 Rupture
FTMSFST 5.5
FTMSFLT 0.55
Failure Probability(long-term)/a 4.62*10-6 1.32*10-6 6.60*10-7
Failure Probability(short-term)/a 4.62*10-4 1.32*10-4 6.60*10-5

2.2.2 Computation of the secondary factor of ure causes and the secondary factor of technological
long-term creep in the technological module module, the heating furnace tube long-term annual
The material of cracking furnace tube is HP-Nb, the failure probability of creep can be obtained, shown as
tube is made by the centrifugal foundry. The mean Table 3 [15].
Larson Miller parameter of material, Lmavg, can be de-
scribed as [11]: 3. Conclusions

Lmavg  65.05956-2.17933ln S (3) (1) In the situation that failure probability is 5%, and
considering simultaneously the interaction of creep and
carburization, the furnace tube operation life is 6.4
where, S is stress of furnace tube (MPa).The Larson
years. The results indicated that the creep and carburi-
Miller parameter at the current operation condition, lm,
zation are main reasons of the cracking furnace tube
is calculated using the following equation:
rupture.
(2) Because of having not considered the influence
TMT  460 (4)
lm  ( )  ( Logti  C ) of carburization, the result of failure probability com-
1000 putation based on API581 is lower than that using
Monte Carlo method.
where TMT is operating tube metal temperature in de- (3) The failure probability of cracking furnace tube
gree; ti is total time in service in hours and C is the calculated by Monte Carlo method is compared with
material coefficient which is 57 in here. The failure that using API581. The results provide failure possibil-
coefficient can be expressed as: ity data for risk assessment of the cracking furnace
3
tube.
 0.2854* X 2  0.5905* X  0.3001
FLT = min(1,100.6723* X ) (5)
References
X  ( Lmavg  l m ) / LM delta (6)
[1] Li Q. Risk assessment method based upon RBI. Safety,
Health and Environment 2004; 02: 34-38.
FTMSFLT  0.55e13*FLT (7) [2] Ma QL. Discussion on risk evaluation mode for petro-
chemical plants. Contemporary Chemical Industry2001;
30(3): 157-159.
where LMdelta is the average difference between the [3] Zhao T. Study on the Risk Based Maintenance Plan for
mean and minimum Larson Miller curves in API Ethylene Cracking Furnace. Nanjing: Nanjing University
RP530; X can be calculated by formula (6), and then of Technology, 2005.
X=0.3 in here. The failure factor FLT can be calculated [4] Yu L, Zheng YL. Response surface method by stepwise
regression. Journal of Dalian University of Technology
by formula (5), and then FLT = 0, the technical module
1999; 06: 56-72.
subfactor of long-term creep FTMSFLT can be computed [5] Wu QE, Zhuo JS. A sequential response surface method
using the formula (7), and then FTMSFLT=0.55. with various f and its application to structural reliability
analysis. Journal HoHai University 2001; 29: 45-52.
2.2.3. Computation result of failure probability in [6] Zhao SG, Lan Y. Monte Carlo method and its application to
API581 structural reliability analysis of terrene edge. Yellow River,
2006; 05: 65-66.
According to the similar failure probability of the
[7] Xu ZJ. Monte Carlo method. Beijing: The Press of Science,
heating furnace in API581, and considering creep fail- 1985.
32 Wenhe Wang et al. / The failure probability analysis of ethylene cracking furnace tube based on risk

[8] Dai SH, Wang ME. Reliability engineering and its applica- [12] Garofalo, Whitmore F, Domis RW. Creep and creep-rupture
tion in chemical equipment. Beijing: Chemical Industry relationships in an austenitic stainless steel.
Press, 1987. Trans.Metall.Soc 1961; 221: 310-319.
[9] Dai S H，Wang M O. Reliability analysis in engineering [13] Wailles, K F A. Random and systematic Factors in the
applications. New York, Van Norstrand Reinhold Press, scatter of creep Data.C.P.No.935. London：Aeronautical
1992. Research Council，1967.
[10] Zhou CY. Probabilistic life prediction, reliability assess- [14] Sprung I, Zilberstein VA. In understanding variability in
ment and preventive maintenance technology. Nanjing: creep and rupture behavior. New York: ASME, 1988.
Nanjing University of Technology, 2000. [15] American Petroleum Institute, Risk- Based Inspection,
[11] Soudabeh, John W H Price. Case study of the use of Base Resource Document, API 581.
API581 on HK and HP material furnace tubes. Journal of
Pressure Vessel Technology 2005; 127: 49-53.