Available online 9 May 2012

Keywords:Carburization

of the formed matrix in carburized zones, the carburization sensors were successfully

g fur

enhancement especially at decoking process, intensify the carburization phenomenon [7]. Carburization leads to the forma-tion of metal carbides in the grains and grain boundaries that consequently reduce the mechanical properties, creep resis-tance, service life time and weldability of the tubes. Moreover, the micro-cracks propagation and tube embrittlement areascribed to the occurrence of carburization process [3,7].

1350-6307/$ - see front matter 2012 Elsevier Ltd. All rights reserved.

Corresponding author.E-mail addresses: hkhodamorad@gmail.com, h_khodamorad@yahoo.com (S.H. Khodamorad).

Engineering Failure Analysis 25 (2012) 8188

Contents lists available at SciVerse ScienceDirect

Engineering Failure Analysishttp://dx.doi.org/10.1016/j.engfailanal.2012.04.006of heat resistant stainless steel; these parts signicantly inuence the economical/technical features of these plants. Eachgroup of heat resistant steel tubes is called a coil and a furnace includes several coils. Fig. 1 shows a schematic view ofthe coil utilized in Jam petrochemical complex (Asalooyeh, Iran).

Typically, these coils work at temperatures 8001100 C [1,2]. After some preheating stages, the uid (gas or liquid) feedenters through the inlet of coil, heats up to the thermal cracking temperature and then ethylene and other products exit fromthe outlet. Products of several coils are gathered and sent to renery stage. Any failure in these tubes results in shut-down ofthe cracking furnaces and wasting both cost and time. Hence, the technical inspection and monitoring of the cracking tubesis mandatory.

Carburization and creep are the main phenomena that cause failure in these tubes and decrease their life time [36].Thermal cracking of hydrocarbon feed in the tubes results in formation of free carbon and its diffusion into the tube. Thedepletion of free carbon and formation of an adherent coke layer on the inner surface of the tubes, as well as temperatureOvalnessCarburization sensorSemi-robot

1. Introduction

The tubes used in thermal crackinemployed for examination of the carburized layers. The comparison between data acquiredvia this sensor-based method and those obtained from chemical etching implied a highenough compatibility and appropriate accuracy of the sensor-based measurements. Thisfeasible analyzing method was effectively utilized in a semi-robot that moves on surfaceof the tubes and establishes a carburization prole. Finally, these carburization prolestogether with ovalness graphs were employed for assessment of the points susceptibleto embrittlement and failure in petrochemical industries.

2012 Elsevier Ltd. All rights reserved.

naces in an ethylene manufacturing process are usually made of HP and HK gradesInspection of carburization and ovalness in ethylene cracking tubesby using a semi-robot

S. Heidar Khodamorad a,, Davoud F. Haghshenas ba Technical Inspection Department of Jam Petrochemical Complex (JPC), Asalooyeh, IranbDepartment of Mining and Metallurgical Engineering, Amirkabir University of Technology, Tehran, Iran

a r t i c l e i n f o

Article history:Received 26 February 2012Received in revised form 21 April 2012Accepted 30 April 2012

a b s t r a c t

Carburization and creep are the main causes of failure in the heat resistant tubes of ethyl-ene cracking furnaces leading to a decrease in ductility and embrittlement of the crackingtubes. The aim of present study was to develop a semi-robot for inspection and classica-tion of the carburized zones in heat resistant tubes. Regarding the magnetic characteristics

journal homepage: www.elsevier .com/locate /engfai lanal

82 S.H. Khodamorad, D.F. Haghshenas / Engineering Failure Analysis 25 (2012) 8188A conventional method for the evaluation of carburization is the measurement of fractional volumetric ratio of chromium

Fig. 1. Schematic of heat resistant ethylene cracking coil.carbides in the micro-structure of tube walls [8]. However, this method requires cutting off the tube (destructive method)and taking specimens from several zones of the tubes (time-consuming). It is worthy of note that during carburization, metalcarbides such as M7C6 and M23C6 are formed and chromium accumulates at carburized zones [57]. Also, in carburized zonesthe eld structure is rich of Ni and FeNi and thus, these zones become ferromagnetic (Ni and NiFe are paramagnetic).Therefore the other method, which is a very effective and useful technique under industrial conditions, is based on ferromag-netic property of the tubes; i.e. the non-magnetic (austenitic) microstructure of these tubes becomes ferromagnetic due tocarburization reaction and when the carburized section of a tube is subjected to a magnetic eld by applying a magnetic incorporated probe, the lines of magnetic force deect toward the carburized section.

The magnetic ux measurement can be applicably used to correlate the magnetic response of carburized austenitic steeltubes (ferromagnetic behavior) with the volumetric fraction of chromium carbides (Cr23C6 and Cr7C3) in the tube wall [9]. Forinstance, Stevens et al. [10] evaluated the changes in magnetic properties of ethylene pyrolysis tubes (due to carburization)by the use of vibrating sample magnetometer (VSM) measurements [10]. Takahashi et al. [11] also studied the chromiumdepletion in nickel based alloys (Inconel 600) employing magnetic force microscopy (MFM) [11].

A literature survey shows that there are some attempts to develop an appropriate technique for the detection of magneticcarburized layers inside thermal cracking tubes. Kasai et al. [12] have employed a C core probe and a magnetizer for analyz-ing such magnetic carburized layers [12]. Additionally, da Silva et al. [13] could successfully characterize the carburization ofHP steels by non destructive magnetic testing. Moreover, Silva et al. [8] studied the carburization of ethylene pyrolysis tubesby magnetic measurements. These works indicate that magnetic ux detection is a suitable and feasible way for analyzingthe carburization phenomenon in the cracking tubes [8,13].

Another sign that shows the risk of rupture in thermal cracking tubes is the ovalness of tube. Carbides formation in inter-nal layers of the tubes leads to volume expansion of these areas. Generally, there is a proximate relation between the car-burization ratio and the creep remaining life with outer diameter (OD) of the tubes [14,15]. Uneven carburization around thetube circumference results in ovalness and the developing high stress and brittle points. Additionally, tube ovalness is sig-nicantly intensied when carburization is very severe and non-uniform.

In present study, the magnet tests together with ovalness measurements have been conducted for detection and evalu-ation the high risk places in thermal cracking tubes in Jam petrochemical complex (Asalooyeh, Iran). From the industrialpoint of view, the maximum allowance of carburization in tube wall is 50% (according to manufacturer recommendations)to keep the process conditions suitable and safe and to decrease risk of carburization. Accordingly, a group of carburizationand ovalness sensors have been utilized to develop a semi-robot that can move on the cracking tubes and make the prolesof carburization and ovalness along the tubes.

Table 1Chemical composition of HP40 + Nb tube.

Comp. C Si Mn Cr Ni Nb Ti

wt% 0.45 1.5 1.0 25 35.0 1.5 Addition

S.H. Khodamorad, D.F. Haghshenas / Engineering Failure Analysis 25 (2012) 8188 832. Materials and methods

Table 1 lists the chemical composition of HP40 + Nb heat resistant stainless steel utilized as thermal cracking tube.The carburization sensor consisted of two main parts; a small suspended permanent magnet (for producing a stable mag-

netic eld) and a x linear analogue electronic magnetic eld sensor (for monitoring and recording the magnetic eld vari-ations). Fig. 2a illustrates a schematic view of this sensor. When a permanent magnet is placed in front of a carburized zoneof the tube, the suspended permanent magnet will pull toward the tube and the magnetic eld around the sensor will vary.These variation in magnetic eld is an indicative of the carburized zones and carburization depth. The ovalness sensors werealso employed for recording the variation of tube outer diameter; i.e. when two ovalness sensors are placed at the x posi-tions opposite to each other (along the diameter of a circle), the variation of diameter can be detectible. Six carburizationsensors together with six ovalness sensors were applied to develop a semi-robot; this semi-robot was able to move overthe tube surface and simultaneously, record the data corresponding to carburization and ovalness of the tube. Fig. 2b depicts

Fig. 2. Schematic of (a) carburization sensor and (b) arrangement of sensors around the tube.the arrangement of these sensors on the developed semi-robot. The results acquired by the sensors were recorded by the useof an electronic data logger and then, converted to the carburization and ovalness proles.

All the tests were carried out with four replicates and conducted in a randomized order to avoid systematic bias. In orderto evaluate the accuracy of data obtained from the sensors, they were compared and veried by the data obtained fromchemical etching of 72 samples using NH3/HF/demineralized water as the etchant solution (NACE TM0498, 2006). The widthof carburized layer was visually checked, measured by digital caliper and compared with that obtained by the sensors.

Fig. 3. A view of developed semi-robot.

84 S.H. Khodamorad, D.F. Haghshenas / Engineering Failure Analysis 25 (2012) 81883. Results and discussion

3.1. Semi-robot development

As a matter of experience, detrimental physical/chemical processes occurred in some zones of heat resistant stainlesssteel tubes, are unpredictable. Furthermore, thermal distribution in these furnaces is non-uniform; i.e. carburization ofthe tubes is an unexpected process in the inner side of tubes. Thus, random inspection of the tubes is not a reliable approachfor monitoring and controlling these tubes.

A semi-robot was established that is able to vertically move over the external surface of a thermal cracking tube (5 in.diameter) and to bear around 3 kg weight; this system consists of a four-electrical-motor-gearbox complex with eightwheels for moving two carrier rings. Six carburization sensors were positioned between these two carrier rings and eachof these two rings included six ovalness sensors. A 16-channel electronical data logger was designed to save the output volt-ages of sensors in a memory card. A photo interrupter commanded the data logger to save voltages of sensors. Particularsoftware was developed to read the recorded data and plot it as carburization and ovalness proles. Fig. 3 shows a viewof the developed semi-robot. Each carburization sensor was equipped by two stainless steel wheels for easy movementand keeping a constant distance between the internal suspended permanent magnet and tube.

3.2. Reproducibility of carburization sensors

For evaluating the reproducibility of the data obtained by a certain sensor at different runs, two replicates that obtainedfrom a given carburized tube by each sensor were compared with each other. The results of this comparison are presented in

Fig. 4. Comparison of results obtained from sensor 1 (a) and 5 (b) in two different runs.

S.H. Khodamorad, D.F. Haghshenas / Engineering Failure Analysis 25 (2012) 8188 85Fig. 4a and b for sensors No. 1 and No. 5, respectively. As can be seen in Fig. 4a, the data obtained by the rst sensor are verysimilar in two runs on a specic carburized tube. However, the negligible differences in voltage response of the sensor areattributed to the errors coming from the magnetic eld sensors and data recording system. Fig. 4b depicts voltage valuesobtained in run 1 versus those obtained in run 2. The slope of correlated line between these two sets of data, as well asthe value of R2, indicates high reproducibility of the sensor.

3.3. Comparability of carburization sensors

In order to examine that whether the different carburization sensors result in a unique response or not, the data corre-sponding to a given sensor (second and third) were compared to that related to other sensors (fth and sixth) under sameconditions (Fig. 5). As can be observed in Fig. 5a, slope of the linear correlation between two sets of data (1), as well as thevalue of R2 (0.95) indicates the suitable comparability of the sensors. Also, the voltages recorded by sensors 3 and 6 (Fig. 5b)are in the range of 3.84 V and contrary to Fig. 4a, there is no pitch point in this prole (Fig. 5b). This indicates that these twosensors were placed on a non-carburized zone. The proximity of the data obtained by the third and sixth sensors reveals highthe enough comparability of these sensors (Fig. 5b).

3.4. Accuracy (tolerance) of ovalness sensors

The measurements via ovalness sensors were conducted for the samples prepared from carburized tubes (5 cm length);the diameter of these samples was also measured along the ovalness sensor position. A comparison between the results

Fig. 5. Comparison of results obtained from (a) 2nd and 5th sensor and (b) 3rd and 6th sensor.

86 S.H. Khodamorad, D.F. Haghshenas / Engineering Failure Analysis 25 (2012) 8188obtained via ovalness sensors and those acquired by direct measurements (by digital caliper) are presented in Fig. 6. Accord-ing to the standard error value (

S.H. Khodamorad, D.F. Haghshenas / Engineering Failure Analysis 25 (2012) 8188 87In this study the carburized layers inside thermal cracking tubes was effectively identied and quantied by the use of amagnetic sensor-based semi-robot under industrial conditions. Besides, the ovalness intensied during carburization pro-cess was evaluated. The reproducibility and comparability of the carburization sensors showed the appropriate applicabilityof the developed semi-robot. Moreover, the accuracy (tolerance) of ovalness sensors revealed that these sensors could esti-Also, Table 2 lists the values of carburized percent (resulted from etching test) and its corresponding sensor voltage atdifferent tube lengths. Based on the obtained results, a linear correlation can be established between the sensor voltageand the carburization percent (Fig. 8). Regarding the result of line tting procedure presented in Fig. 8, a linear correlationthat tted 94% of the variation in the data is proposed.

4. Conclusions

Fig. 8. Correlation of practical carburization measurements vs. sensor voltage.mate the value of diameter with an error less than 1%. A linear correlation could be also established between the sensor volt-age and the carburization percent that tted 94% of the variation in the data.

Acknowledgment

The authors would like to acknowledge the Jam Petrochemical Complex Inspection Department and turning unit for theircooperation.

References

[1] Dini G, Vaghe MM, Lotani M, Jafari M, Safaei-Rad M, Navabi M. Computational and experimental failure analysis of continuous-annealing furnaceradiant tubes exposed to excessive temperature. Eng Fail Anal 2008;15:44557.

[2] Tari V, Najazadeh A, Aghaei MH, Mazloumi MA. Failure analysis of ethylene cracking tube. J Fail Anal Prev 2009;9:31622.[3] Khodamorad SH, Haghshenas Fatmehsari D, Rezaei H, Sadeghipour A. Analysis of ethylene cracking furnace tubes. Eng Fail Anal 2012;21:18.[4] Guan K, Xu H, Wang Z. Analysis of failed ethylene cracking tubes. Eng Fail Anal 2005;12:42031.[5] Tawancy HM. Degradation of mechanical strength of pyrolysis furnace tubes by high temperature carburization in a petrochemical plant. Eng Fail Anal

2009;16:217985.[6] Chauhan A, Anwar M, Montero K, White H, Si W. Internal carburization and carbide precipitation in FeNiCr alloy tubing retired from ethylene

pyrolysis service. J Phase Eqilib Diffus 2006;27:68490.[7] Ul-Hamid A, Tawancy HM, Mohammed AI, Al-Jaroudi SS, Abbas NM. Failure of ethylene furnace outlet transfer line due to overheating. J Fail Anal Prev

2005;4:5461.[8] Silva IC, Silva LL, Silva RS, Rebello JMA, Bruno AC. Carburization of ethylene pyrolysis tubes determined by magnetic measurements and genetic

algorithm. J Scripta Mater 2007;56:31720.[9] Silva IC, Rebello JMA, Bruno AC, Jacques PJ, Nysten B, Dille J. Structural and magnetic characterization of a carburized cast austenitic steel. J Scripta

Mater 2008;59:10103.[10] Stevens KJ, Tack AJ, Thomas CW, Stewart D. Through-wall carburization detection in ethylene pyrolysis tubes. J Phys D: Appl Phys 2001;34:81422.[11] Takahashi S, Sato Y, Kamada Y, Abe T. Study of chromium depletion by magnetic method in Ni-based alloys. J Magn Magn Mater 2004;269:13949.[12] Kasai N, Owaga S, Oikawa T, Sekine K, Hasegawa K. Detection of carburization in ethylene pyrolysis furnace tubes by a C core probe with

Magnetization. J Nondestruct Eval 2010;29:17580.

[13] da Silva IC, da Silva RS, Rebello JMA, Bruno AC, Silveira TF. Characterization of carburization of HP steels by non destructive magnetic testing. NDT&EInt 2006;39:56977.

[14] Maharaj C, Imbert C, Dear j. Failure analysis and creep remaining life of hydrogen reformer outlet pigtail tubes. Eng Fail Anal 2008;15:107687.[15] Swaminathan J, Guguloth K, Gunjan M, Roy P, Ghosh R. Failure analysis and remaining life assessment of service exposed primary reformer heater

tubes. Eng Fail Anal 2008;15:31131.

88 S.H. Khodamorad, D.F. Haghshenas / Engineering Failure Analysis 25 (2012) 8188

Inspection of carburization and ovalness in ethylene cracking tubes by using a semi-robot1 Introduction2 Materials and methods3 Results and discussion3.1 Semi-robot development3.2 Reproducibility of carburization sensors3.3 Comparability of carburization sensors3.4 Accuracy (tolerance) of ovalness sensors3.5 Etching test

4 ConclusionsAcknowledgmentReferences