CORROSION/RELIABILITY Failure analysis of an ethylene cracking heater finned tube Here's what caused the leak and the steps taken to prevent future leaks R. P. GUPTA and P. MOON, IPCL, Nagothane, India uring the hydrotest as per the D annual health check plan of this furnace, water leakage from the convection zone was noticed and further investigation, by opening che convection box cover at different locations, showed that a 4-in.@ finned tube ofthe first-pass ceil of the convection zone was leaking. ‘Appearance/morphology of dam- age. An elliptical shaped 12-mm X 7- mm puncture was noticed toward the north side of the convection aone atthe tube entry point in the convection 20 Erosion and erosion corrosion are char- acterized by a localized loss in thicknessin the form of pits, grooves, gullies, waves, rounded holes and valleys. This loss often tsa directional pattern (Fig. 1) Process details. Funetion of HG prebeater-1 furnace. Hydrocarbon preheater-1's function isto prcheat the HC (ce, ethane/propane) from 65°C to 125°C to 145°C through furnace flue gases. Preheater-1 contains HC with MDS catalyst in normal furnace crack ing operation. During the decoking cycle, the HC feed control valve remains closed whereas decoking steam and hot air are introduced through preheater-2. From pre~ hheater-2 onward, steam is ive, ie, circu ing through preheatcr2, the economizes, the superheater and furnace area and the dezok- ‘TABLE 4. Operating data ans Cracking mode parameters Ethane Propane Decoking (psatng prose, 64, 65,66, 54, 65,66 Stam was stagnant (not Ive) glen’ uring the decoking ye Operating cy 50 150 temperature, °C Savi TE DMS Hes DDS “Seam + yarne + condensate + hat air prof id handed 30.93 Production ce a0 dae Wade Decking ge he wor Divs Geb 1 (Big 2). The operating data are shown in ‘Table | Probable reasons of failure. Tube failure may have occurred duc to any one ora combination of the following prob- able reasons: + Faulty material of construction (moc) * Corrosion * Upset in process. Visual inspection, NDT and metal- lography of the failed fin cube were carried out to find the root cause of the fale. "Three small pieces were cut at different locations ofthis finned tube for study and visual inspection toward the north side of the convection zone atthe tube entry point in the convee- + Liquid stagnation marks were noticed between the 5 o'clock and 7 o'clock pos tions ofthe finned tube + Severe roughening and pitting were noticed at the 6 o'clock position on the internal surface of the tube near the pune- ture arca, which may be indicative of the presence of moisture from steam eonden- sate, Q,, sulfur (40-50 ppm) from DMDS, CO; from HC, and NH; fom hydrazine, * In the vicinity of the punctured area OD of the fins was found eaten away. + A black sticky process deposit was noticed between the 5 o’clock and 7 th ing network piping including the decoking At tube failed location: o'clock positions. pot. The only area where the steam is not _* As mentioned, an elliptical shaped At other locations (three samples of 6 live ie the steam is stagnant, is preheate- 12-mm X 7-mm puncture was noticed in, were cut) ‘TABLE 2. Chemical analysis results Elements Ge Sia tn Pax Seo Nine Cine Mma Cla You Beper 13 a 029 003 003 04 04 O15 04 008 ‘06 5 5 G8 1.06 Faedsomple 0205 0216 —~0582_——~=~=«B~C~*« co aCCORROSION/RELIABILITY + Liquid stagnation marks were noticed beeween the 5 o'clock and 7 o'clock posi tions ofthe finned cube + Roughening and pitting wer atthe 6 o'clock position on then surface, However, the extent of corro- sion is less in these samples compared the failed portion. ‘The failed sample was chemically ana lyzed and the results are shown in Table 2 Hardness of the cross-section of the failed samples was taken and found in che range of 82-85 HIRB, Hardness values are normal for this MOC. (roves |esonomize ee \ Supa \ Radiat box ‘The microstruc- ture of the failed finned cube cross ined and found to be normal, ie., ferrite and pearlite (Fig. 3). Approximate grain size was 7-8, meaning a fine-grained CS microstructure is normal for CSA 106 Gu B. Root-cause analysis. +The chemical, hardness and mecallo- graphic studies confirmed that the MOC of the ube is A 106 Gr. B, which isthe correct, MOC. No abnormality was noticed in the finned tube metallurgy. Tube failure due to faulty MOC is ruled out. + Seagnation of steam and condensate during the heater decoking cycle increases the corrosivity of the environment, reduces stability of the protective surface films and increases susceptibility to metal loss. Metal may be removed from the sur- face as dissolved iron or as solid corrosion -, products, which are mechanically swept oT an ae ne ae Re ee el eee od, from the finned tube surface. The size, shape, density and hardness of the impact ing medium also afect the meral oss rate from the finned tube. + Factors that contribute to an increase in cortosivty of the environment, such as temperature and pH, can increase suscep- tibility vo metal loss + Erosion can be caused by gas-borne catalyst particles or particles carried by a liquid such asa slurry (mixture of HC deposits, steam and condensate). This form of damage ocenrs as a result of eat lyst (DMDS) movement in the FCC reac- torlregenerator * Liquid level effects sion often occurs underneath solid sub- stance deposits that sometimes collect just above the liquid level on a metal part that is partly immersed in an electrolyte. The deposit usually remains moist or intermit- tently moist and dry. + CO, corrosion results when CO? dissolved in_ water forms carbonic acid (HyCOs). The acid may lower the pH and a sufficient quantity may promote general cotrosion and/or pitting corrosion of carbon ste + Sullus from DMDS catalyst—At a low enough temperature (about 138°C), the gas and water vapor will condense to form sulfurous and sulfuric acid, depend-era Opesirereccn eet cae ing upon the concentration of sulfur triox- ide, which ean lead to * Localized flow can cr cumferential groove called condensate grooving ifthe condensate contains a high level of NH, Op and ai * The eailier production cycle was quite short, ie, 25 days or approxi- rately half ofthe present situation. This means thatthe decoking cycle of 35-45 har repeated twice earlier over the pres- ent situation, These tubes are subjected to two different service conditions, i HC during nocmal operation and stag nate steam and condensate during decok- ing cycles. This stagnation condition of ie a deep cir- steam and condensate leads to cor and subsequent tube failure. * Corrosion is noticed throughout the tube internal surface but a che failed loca- tion ic is very high. This may be due to initial tube surface roughness or excessive accumulation of stagnate condensate at that location "The metal area where the liquid level fluctuates of the liquid agiates that is incezmittently weued is called the splash zone. Corrosion occurring in this area is called splash-2one corrosion + General metal los is defined as rela tively uniform chinning over a significant area of the equipment. Corrosion and ero- sion are never totaly uniforms a rule of thumb is that, ifthe metal loss rate among diferent points in an ates varies by a fac- tor of four or less, then damage is consid ced general. Stagnant steam and condensate during the heater decoking cycle increased the corrosvity ofthe environment and led to failure of the HC pecheat line finned tube in the convection 2one (Fig. 4) CORROSION/RELIABILITY TS pager s SootonacA males * eset tbe, Recommendations. * Scagnation of steam and condensate during che heater decoking cycle is avoided by providing a line from the HC control valve low drain point to the decoke pot so thae duting furnace decoking, positive flow through hydrocarbon preheater I can be ensured. A tube of the other heater similar 10 preheater-1 should be removed for the inspection to assess its health sipuiocRarHy API R571 (it lin, Dacre 2008), Sesion 3. Meat tie 3. ‘ASIAVEL 9, and of econ R. P. Gupta ic seer comoson and metal spect for PCL, agothane, India. He has over 16, ots of experience dealing with al c ‘pecs of mate electn, carson ‘monitoring and conto, aspecton, NOT metargiclalure analysand woubleskenting, ‘iting atria spect, waking spect, ee. | Mr Gupta hae a BF date metal fom Rajasthan Universtiy. na, and an Mech eae esta tngicering rr IT Dati, nda, He can beveahed at ‘mal andr Gopi coin P.R. Moon ica creson and met lrg specait for PC, ago thane Indi, He has cre 14 yeas of ference nthe Fld of nection, Roi] NOT, corrosion monitoringfcontro, etal, welding inspection and retalogresy. He ha dip. ngineaing eee me blag tom te Goverment Pec, Nagpur na, Ml pst] Scalable Why is CADWorx one of the fastest {growing and most productive pant design suites on the market? Because thas al he tots to produce Inttger plant designs, Inui + Ping + Sto! * cable vays/suctog * Colson detection + Equipment * lof material * fomoties + Flow schematics * irument op diagrams 1 Blrectonal ks antes roams + Wateteough and vualzaton CADWorx delivers! Contact us fin out tow you cen imorom your design effctenc. CHE Plant Focused. Industry Driven -+1 281-890-4566 + sales@coade.com RWWA erel [ol metal at) Erte onesies Sad con be aches at em: ra, Moonie coin Select 175 at vowsyarocarbonProcesing coms