- Email: [email protected]
Effect of composition of coke deposited in delayed coker furnace tubes on on-line spalling
Fuel Processing Technology 172 (2018) 133–141
Contents lists available at ScienceDirect
Fuel Processing Technology journal homepage: www.elsevier.com/locate/fuproc
Research article
Effect of composition of coke deposited in delayed coker furnace tubes on on-line spalling
T
⁎
Harender Bishta, , Vijai Shankar Balachandranb, Mitra Patelb, Girish D. Sharmaa, Ashwani H. Yadava, Dwaipayan D. Biswasa, Sreenivas Pacharua, Sukumar Mandala, Asit K. Dasa a b
Reliance Industries Limited, Motikhavdi, Jamnagar 361140, India Reliance Industries Limited, Reliance Corporate Park, Ghansoli, Navi Mumbai 400701, India
A B S T R A C T Delayed coking is one of the most widely used residue upgradation process in crude oil refining where vacuum residue is thermally cracked and converted into distillates and petroleum coke. During heating of vacuum residue in coker furnace, coke is continuously deposited in furnace tubes which is removed at regular intervals by one of the three methods namely (i) pigging, (ii) steam air decoking or (iii) online spalling (OLS). OLS is one of the preferred methods of coke removal from coker furnace tubes. We have noticed very effective as well as “not so effective” OLS in a commercial delayed coker furnace. In present article, the changes in composition and properties of furnace coke during effective and ineffective OLS were systematically analyzed by various analytical tools such as Dilatometer, TGA, HRSEM, SEM-EDX and XRD. Based on these analyses, it was found that inorganic deposits such as iron sulphides and tube metal corrosion products play a major role in failure of OLS. Higher total acid number (TAN) of vacuum residue may be promoting the formation of iron sulfide deposits in furnace tube.
1. Introduction Delayed coking is one of the most widely used residue upgradation process where vacuum residue (VR) is thermally cracked and converted into distillates and petroleum coke. Typically VR feed enters in a delayed coker furnace at 300–350 °C and comes out at 485–505 °C. On reaching cracking temperature of ~ 400 °C, thermal cracking of alkyl side chains of asphaltenes gets initiated, consequently the core of asphaltene molecules become less soluble in VR [1,2]. The cracked asphaltenes core agglomerate and start to precipitate out of solution phase. These agglomerates on further propagation in furnace tubes experience still higher temperature ~ 500 °C and start depositing on the furnace tubes as coke [3,4]. The thermal conductivity of deposited coke [5,6] is typically between 0.3–3 W/m·K and that of P9 metal used in furnace tube is ~30 W/m·K [7]. Thus coke being ~30 times less conductive than heater tubes reduces the heat transfer coefficient of furnace tubes [8,9]. In order to maintain the desired furnace outlet temperature of feed, tube metal temperature is continuously increased up to the maximum allowable working temperature (MAWT). P9 metal used in delayed coker furnace tubes contains ~89 wt% Fe, ~9 wt% Cr and ~1 wt% Mo along with small amount of C (~ 0.1 wt%), Si (~ 0.22 wt%), Mn (0.73 wt%) and Ni (0.27 wt%) [10].
⁎
Deposition rate of coke and other associated inorganic materials in delayed coker depends on several factors such as feed velocity inside furnace tubes, [11,12] suspended solids in coker feed, [13] roughness of tube surface [14] and temperature of tube metal surface [8,9,12]. Once MAWT is attained, furnace tubes are required to be cleaned by one of the following three methods viz. a) Pigging [15] b) Steam air decoking [16] c) On-line spalling (OLS) [17]. During pigging the entire furnace is shut down and cooled to ambient temperature. A plastic pig, embedded with metal studs, is inserted in the fouled furnace tubes and pushed with help of high pressure water. Size of plastic pig is sequentially increased to a diameter slightly less than the inner diameter of furnace tube to remove maximum amount of coke from furnace tubes. In steam air decoking the fouled furnace is heated in presence of steam-air mixture to slowly burn the deposited coke. This method is less effective when significant amount of inorganic material is deposited in furnace tubes. OLS is one of the most economical methods of furnace decoking as it does not require complete furnace shutdown and external expert agencies to perform the spalling activities. OLS is done in one pass of furnace at a time whereas other passes of the same furnace are operated normally. Vacuum residue flow to the furnace pass where OLS is being carried out is reduced and steam flow is increased in tandem. Finally VR
Corresponding author. E-mail address: [email protected] (H. Bisht).
https://doi.org/10.1016/j.fuproc.2017.12.013 Received 17 August 2017; Received in revised form 19 December 2017; Accepted 19 December 2017 0378-3820/ © 2017 Elsevier B.V. All rights reserved.
Fuel Processing Technology 172 (2018) 133–141
H. Bisht et al.
flow is completely stopped and steam flow is maximized to get steam velocity in the range of 90 to 100 m/s. After attaining the desired steam velocity, temperature of furnace tubes is gradually raised to MAWT, typically in the range of 650–700 °C and maintained for 1–2 h to dislodge any loosely bound coke. Then temperature of furnace tubes is quickly dropped to 450–500 °C by manipulating fuel supply to furnace to give a mild thermal shock. This mild shock is given to ensure heater tubes are not plugged due to excessive coke removal. After mild thermal shock, temperature of furnace tubes is again increased to MAWT and hold for few hours. 2nd thermal shock is given by decreasing the tube temperature to 250–300 °C in < 1 h by manipulating fuel supply to furnace and introduction of boiler feed water in furnace tubes. 1 to 3 more shocks are given by repeating the sequence followed for 2nd deep shock to ensure maximum coke removal from tubes. After multiple thermal shocks, furnace tubes are heated again, vacuum residue is gradually introduces and steam flow is reduced to normal operation. Success of OLS is measured by reduction in tube metal temperature when the spalled pass is taken back in line for regular operation to maintain desired feed outlet temperature. OLS efficiency depends upon difference in coefficient of thermal expansion (CTE) between deposited coke and tube metal as well as binding strength of coke with tube metal surface. The former is primarily dependent on crystallinity of coke such as graphite content, bonding characteristics of C in constituent molecules [18] whereas the latter is dependent on the elemental composition of coke layer and its inorganic contents [14]. Elemental analysis of the deposited coke in a lab scale foulant testing unit showed C, Fe and S as major constituents [19,20]. In literature source of Fe and S has been attributed to feed and it was suggested that iron naphthenates in feed decomposes and reacts with sulfur to form iron sulphide, predominantly as Pyrrhotite [15]. In this article, we have compared the composition and properties of furnace coke removed during pigging from a ~9 Million Metric Ton per Annum commercial delayed coker furnace situated in Jamnagar, India. These coke samples were collected during pigging operation when OLS was very effective and when OLS was not so effective. This study attempts to understand the root cause of good and bad spalling of coke in a commercial delayed coker furnace. It was found that significant amount of iron sulfide was present in the coke deposited in furnace tubes during bad spalling period which may be a result of higher corrosion of tube metals and upstream equipment due to increased total acid number (TAN) and reduced sulfur in VR feed. CTE of bad spalling coke was higher and closure to that of P9 metal making is more prone to sticking during thermal shock of OLS.
Table 1 Average VR properties during good and bad spalling periods. Properties
Good spalling period
Bad spalling period
0.47 4.1 1.037 14.1 25.5 16 410 119 11.1
0.60 3.9 1.037 15.1 25.2 16 439 117 13.3
TAN, mg KOH/g Sulfur, wt% Density, g/cm3 Asphaltene, wt% CCR, wt% Recovery at 565 °C, wt% V, ppmw Ni, ppmw Fe, ppmw
coke and water mixture were collected from furnace tube outlet. This mixture was filtered through a mesh to collect coke flakes. The water along with finer coke powder was drained. The coke thus collected was kept in separate bags for each pass of the furnace. In this study, since the total thickness of coke layer was ~2 mm only, most of the coke flakes were intact having both tube side and feed side surfaces. In case, thickness of coke is higher there may be slicing of the coke layer into two or more layers hence care must be taken to identify the tube side and feed side surfaces. The screened coke samples were oven dried at 60 °C for 12 h to remove moisture. These coke samples have seen temperature of ~ 500 °C for more than 60 days in furnace tubes and were of different size and shape ranging from small granules to flakes of few centimeter sizes. 2.3. Hardgrove Grindability Index (HGI) HGI of coke samples was measured in HGI apparatus from Chemicals and Instruments Corporation, Kolkata as per ASTM D409. 2.4. Real density Real density of coke and ash samples was measures in Quantachrome Ultrapyc 1200e as per ASTM D2838. Coke samples were powdered to < 75 μm size before real density measurement. 2.5. X-ray diffraction XRD analysis was carried out by using Rigaku Miniflex600 power diffraction instrument. X-ray was generated from Cu X-ray tube operated at 60 kV and 1 mA. Graphite monochromator was used to get monochromatic Cu Kα radiation (1.54056 Ǻ) which was used for XRD measurement. The samples were powdered and uniformly spread over XRD sample holder. Then sample holder was placed on an auto sampler of the X-ray system. XRD patterns of the samples were collected in 5–50° (2θ) with scanning speed 1°/min and step size (0.01°). Phase identification in the sample was carried our using search match program of M/s Rigaku and using the JCPDS library.
2. Material and methods 2.1. Vacuum residue (VR) feed VR feed processed in delayed coker during good spalling and bad spalling period was produced from blends of 15–20 different crude oils. Although VR properties were maintained within specification window, there were some changes in feed properties during good spalling and bad spalling periods. Table 1 shows the average properties of VR feed processed during good spalling and bad spalling periods. Average TAN of VR was 0.47 mg KOH/g and 0.60 mg KOH/g during good spalling and bad spalling periods respectively. Average total sulfur in VR was 4.1 wt% during good spalling period whereas it has decreased to 3.9 wt % for bad spalling period. Other feed properties like CCR, metals, density were almost similar in both periods.
2.6. Thermogravimetric analysis TGA measurements were performed by using TA SDT-Q 600 instrument. Samples were placed in the analyzer and kept at 250 °C for 1 min. Subsequently, the temperature was increased from 250 °C to 900 °C at a heating rate of 100 °C/min. Decrease in weight of the sample w.r.t. the starting weight was recorded in percentage. Derivative of the curve was also determined in order to display peaks where the weight decrease was highest.
2.2. Furnace coke 2.7. CHNS analysis Coke collected after pigging of delayed coker furnace when OLS was effective was named as “good spalling” and that collected when OLS was ineffective was named “bad spalling”. During pigging the scrapped
CHNS measurement was performed using Elementar Vario micro cube instrument, where combustion tube and reduction tube were set at 134
Fuel Processing Technology 172 (2018) 133–141
H. Bisht et al.
temperature programmed desorption (TPD) trapping column for separation of N2, CO2, H2O and SO2 before detection by a thermal conductivity detector (TCD). Concentrations of carbon, sulphur, hydrogen and nitrogen were determined as per ASTM D5373 method using sulphanilamide as reference standard.
Table 2 Properties of good and bad spalling coke. Properties
Good spalling
Bad spalling
HGI Real density, g/cm3 Elemental analysis (wt%) Carbon Hydrogen Nitrogen Sulfur Others TGA % wt loss up to 150 °C in N2 (moisture) % wt loss up to 850 °C in N2 (moisture + VCM) Residue after 1000 °C in air
40 2.12
65 2.39
65.6 1.0 < 0.5 7.5 25.4
45.6 < 1.0 < 0.5 17.2 35.7
5.0 12.5 21.5
2.6 8.6 43.3
2.8. Ash generation 40 gm coke samples were crushed to make fine powder and heated up to 700 °C in a calcination furnace with 30SLPH air flow and maintained at this temperature for 12 h. The ash thus formed was analyzed for elemental analysis and real density. 2.9. Scanning electron microscopy (SEM) SEM imaging of coke samples was taken on NOVA NanoSEM650, FEI Co, USA with a field emission gun. The images were acquired with back scatter electron (BSE) mode using a concentric BSE detector. The samples were mounted on carbon tape and fractured to observe the cross section. The micrographs showed Z contrast based on atomic number to give bright domains for higher elements with varying grey scale. Coke samples were imaged on both sides. The side which was in contact with furnace tube was named as “Tube Side (T)” and the side which was in contact with vacuum residue was names “Feed Side (F)”. SEM-EDS was done with EDAX® SDD X-ray detector. Excitation was fixed at 10 keV and mapping was acquired using TEAM® software Version. 4.2.2.
Table 3 Ash of good and bad spalling coke. Sample
Spalling type
Sample wt, (gm)
Ash wt, (gm)
Ash, wt%
Real density, g/cm3
Sample 1
Bad spalling Good spalling
40.0129
10.0198
25.04
4.0335
40.0462
6.3219
15.79
3.8309
Sample 2
Table 4 Elemental composition of good and bad spalling cokea. Elements
C O Fe Ni Na Mg Al Si P Mo S Ca V a b
Good spalling
2.10. Coefficient of thermal expansion (CTE)
Bad spalling
Concentration (wt%)
Normalizedb (wt%)
Concentration (wt%)
Normalizedb (wt%)
0.3 37.5 4.7 1.5 4.5 0.4 2.2 6.0 2.4 29.5 3.7 3.1 3.9
0.0 39.2 4.9 1.6 4.7 0.4 2.3 6.3 2.5 30.9 0.0 3.2 4.1
1.9 40.3 19.8 1.3 2.0 0.6 1.0 5.0 0.9 17.3 3.6 2.8 3.6
0.0 42.7 21.0 1.4 2.1 0.7 1.0 5.3 0.9 18.3 0.0 2.9 3.8
Coefficient of thermal expansion of P9 metal, good spalling and bad spalling coke were measured on Unitherm Model 1161 Dilatometer from TA Instrument using Unitherm dilatometer Software. The instrument was calibrated with Crystallox standards. Argon gas was used as inert medium. 3. Results and discussion Hardgrove Grindability Index (HGI) of coke gives indication of its hardness/brittleness and hence to some extent the spalling character of the coke deposits. Higher HGI indicates softer coke whereas lower HGI indicate harder/brittle coke. Table 2 shows the HGI and real density of bad and good spalling coke. HGI value of bad spalling coke was much higher (softer coke) and was in the range from 59 to 90 whereas that of good spalling coke was ranging from 38 to 43 (more hard & brittle coke). This analysis suggests that the good spalling coke was harder and more brittle making them prone to crack during thermal shock of online spalling vis-à-vis bad spalling coke. Bad spalling coke is more prone to fine generation due to intercalated layer of iron sulfide thus showing higher HGI values. Real density of bad spalling coke was ranging from 2.2 to 2.6 gm/
Average of three measurements. Normalized after subtracting C & S values.
1150 °C & 850 °C respectively. 30–40 mg sample was weighed in tin containers and introduced into the combustion tube with help of an autosampler. The gases generated in combustion tube were swept out by helium into a copper filled tube, then passed through proprietary Table 5 Metal and oxygen content in good and bad spalling coke. Metal oxide
Mol. wt
Metal, wt%
O2, wt%
Metal good spalling, wt%
Cal. O2 good spalling, wt%
Metal bad spalling, wt%
Cal. O2 bad spalling, wt%
Fe2O3 NiO Na2O MgO Al2O3 SiO2 P4O6 MoO3 CaO V2O5 Total
159.7 74.7 62 40.3 102 60.1 219.9 143.9 56.1 181.9
69.9 78.6 74.2 60.3 52.9 46.8 56.3 66.6 71.5 56.0
30.1 21.4 25.8 39.7 47.1 53.2 43.7 33.4 28.5 44.0
0.77 0.26 0.74 0.06 0.36 1.00 0.39 4.88 0.50 0.65 9.60
0.33 0.07 0.26 0.04 0.32 1.14 0.30 2.44 0.20 0.51 5.61
5.25 0.34 0.52 0.17 0.25 1.32 0.23 4.58 0.73 0.96 14.36
2.26 0.09 0.18 0.11 0.22 1.51 0.18 2.29 0.29 0.75 7.89
135
Fuel Processing Technology 172 (2018) 133–141
H. Bisht et al.
Fig. 1. XRD of good spalling coke and bad spalling coke.
Intensity (cps)
Good Spalling Coke sample Good Spalling Coke sample Bad Spalling Coke sample
Intensity (cps)
Pyrrhotite 3T, syn Fe7S8, 00-025-0411
2θ θ
Fig. 2. BSE SEM images of bad spalling coke (BS) tube side (T), feed side (F) and cross section (X).
136
Fuel Processing Technology 172 (2018) 133–141
H. Bisht et al.
Fig. 3. BSE SEM images of good spalling coke (GS) tube side (T), feed side (F) and cross section (X).
cm3 whereas that of good spalling coke was in the range of 2.04 to 2.17 gm/cm3 (Table 2). In case of bad spalling coke the real density was found even higher than that of highly crystalline carbon e.g. graphite (2.09–2.23 gm/cm3). Presence of appreciable amount of pyrrhotite having real density of 4.61g/cm3 may be the reason for increase in real density of bad spalling coke.
3.2. Thermo gravimetric analysis (TGA) TGA of bad spalling coke in air showed ~55% weight loss with ~45% ash (inorganic residue) whereas good spalling coke showed ~79% weight loss and ~21% ash (Table 2). TGA results showed the combustion of coke started around 400 °C and completed below 600 °C. TGA performed in nitrogen atmosphere showed a weight loss of about 8–12 wt% out which 2.5–5 wt% was attributed to moisture which evaporates below 150 °C and remaining was attributed to volatile combustible matter (VCM). The inorganic residue obtained in presence of air after TGA analysis of good spalling coke was much lower (21–25 wt%) vis-à-vis bad spalling coke which contained 40–45 wt% residue. On the other hand, moisture and VCM content of bad spalling coke was lower than good spalling coke. These results corroborate that inorganic content of the deposited coke remains a determining factor for good and bad spalling.
3.1. CHNS analysis Elemental analysis of coke is tabulated in Table 2. Carbon % in good spalling coke was in the range of 63.2 to 67.9 wt% whereas Carbon % in bad spalling coke was significantly lower and varying from 40.6 to 56.5 wt%. The sulfur content in good spalling coke was ranging from 6.3 to 8.7 wt% whereas in bad spalling coke was much higher and ranging from 14.2 to 19.1 wt%. Carbon content of good spalling coke was much higher implying they were predominantly carbonaceous, whereas bad spalling coke was having high amount of inorganic matter. The inorganic matter was predominantly associated with sulfur as the sulfur content of bad spalling coke was significantly higher than that of good spalling coke.
3.3. Ash analysis As seen in Table 3, ash content obtained from good spalling and bad spalling coke was 15.79 wt% & 25.04 wt% respectively. 137
Fuel Processing Technology 172 (2018) 133–141
H. Bisht et al.
Fig. 4. Tube side SEM-EDS elemental mapping of bad spalling coke.
3.4. X-ray diffraction analysis (XRD)
Ash content of bad spalling coke was much higher than that of good spalling coke suggesting higher amount of inorganic material deposited along with coke during bad spalling period. Real density of ash generated from bad spalling coke was higher than that of good spalling coke. Among all oxides present in ash, NiO and Fe2O3 have highest densities (6.72 gm/cm3 & 5.25 gm/cm3 respectively) [21]. Considering the fact that quantity of NiO is same in both good and bad spalled coke ash, it follows logically to attribute the high real density of ash from bad spalling coke is due to higher level of Fe2O3. During calcination, metals present as sulfide get converted to oxide forms. The ash after calcination was milled and pellets were made from homogenized ash. The elemental composition of ash pellet was analyzed by SEM-EDS and reported in Table 4. Some residual C and sulfur was present in the ash. The metal and oxygen concentrations were normalized after subtracting the C and S from ash. Table 5 shows the inorganic oxides present in ash, their molecular weight and oxygen %. This data was used to calculate the oxygen content of ash which was compared with the oxygen data obtained by elemental analysis. The total oxygen content in ash by SEM-EDX was 37.5 wt% and 40.3 wt% for good and bad spalling coke respectively (Table 4). For good spalling ash the measured value is close to the calculated value of 36.9 wt% whereas the calculated oxygen content for ash of bad spalling coke was slightly lower than measured value (35.5 wt% vis-à-vis 40.3 wt%). Elemental analysis showed significant change in iron concentration of good and bad spalling ash. Other metals present in VR feed such as Ni, V, Al, Ca, Si, Na and Mg were present in similar concentration in both good and bad spalling coke. In addition to the above listed elements, the amount of Mo was significantly high in both good and bad spalled cokes. Based on metal analysis of VR feed (Table 1), such high concentration of Fe and Mo in deposited coke could not be attributed to feed metals and are expected to be coming from corrosion of tube metals or upstream equipment. This inference is also corroborated by higher TAN of the VR feed processes during bad spalling period.
XRD pattern of bad spalling coke matched with that of Pyrrhotite-3T having a formula of Fe7S8 (ICDD powder pattern: 00-025-0411) (Fig. 1). Carbon in bad spalling coke was present in more amorphous state and gave a broad shoulder like pattern whereas carbon in good spalling coke was less amorphous and the peaks were much sharp in comparison to bad spalling coke. The characteristic peak of graphitic carbon at 2θ 26.5° [002] was present in the XRD spectra of good spalling coke suggesting comparatively higher crystalline nature. 3.5. Scanning electron microscopy (SEM) In order to understand the microstructure and surface morphology, both coke samples were evaluated using SEM [22]. The objective of this analysis was a) to know the formation of various carbonaceous and inorganic phases during coking b) presence of different metals in coke c) identification and composition of inorganic phases intercalated in the coke layers. We chose to image coke samples in back scattering mode to obtain Z contrast which gave bright domains for higher atomic number elements with varying grey scale. This enabled to quickly demarcate carbonaceous and inorganic phases. Morphology and composition of deposited coke changed as we move inwards from tube side (T) to feed side (F). We have also fractured coke flakes to reveal the cross section and imaged for inclusions (X). 3.5.1. SEM of bad spalling coke Fig. 2 shows the back scattered electron (BSE) micrographs of bad spalling coke. Tube side images (BS1-T & BS2-T) show domains of inorganic phase (bright area) and carbonaceous phase (grey area). Careful examination of the micrographs shows that the bright inorganic phases were not appearing continuous due to peeling off from the coke surface. This may have resulted due to high mechanical shear during pigging operation which has exposed the underneath carbonaceous layer. These observations suggest that inorganic (non-carbonaceous) layers were continuously spread between tube metal and coke layer 138
Fuel Processing Technology 172 (2018) 133–141
H. Bisht et al.
Fig. 5. Feed side SEM-EDS elemental mapping of bad spalling coke.
which has resulted in firm binding of coke to tube metal surface. The composition of bright layer was predominantly iron sulfide. On the other hand, feed side images (BS1-F & BS2-F) showed coke surface was mostly composed of carbonaceous material with few isolated bright patches of inorganic layer. Careful observation of surface revealed that the inorganic phase was also intercalated between carbonaceous layers in bad spalling coke. Therefore we fractured the coke flakes to reveal the cross section (BS1-X & BS2-X) and it was observed that the bright inorganic layer is not only deposited between tube metal and coke but also intercalated in between different coke layers.
that of bad spalling coke. The morphology of tube side coke appeared mainly composed of carbon phase with very few isolated inorganic domains as seen in Fig. 3 (GS1-T & GS2-T). Unlike bad spalling coke, the inorganic domains are discrete suggesting that most of the tube metal surface was in contact with carbonaceous material which was easy to detach during OLS operation. Coke surface on the feed side (GS1-F & GS2-F) was also predominantly composed of carbonaceous material. Fractured cross sections of good spalling coke showed absence of any intercalated inorganic domains in the coke layer (GS1-X & GS2X).
3.5.2. SEM of good spalling coke BSE-SEM images of good spalling coke are markedly different from
3.5.3. Energy dispersive spectroscopy (SEM-EDS) To understand elemental composition of bright inorganic domains and grey carbonaceous domains, EDS mapping was done for both tube side (T) and feed side (F) surfaces of bad spalling cokes. This elemental mapping of whole coke surface showed that the bright domains on tube side (Fig. 4) and feed side (Fig. 5) were predominantly composed of iron, sulfur, manganese and chromium whereas the grey domains were predominantly composed of carbon and sulfur. Sulfur in VR mostly present in form of thiophenes, sulfides (both cyclic & acyclic) and sulfoxides [23]. Out of these, the sulfides are thermally reactive species and mostly contribute for formation of iron sulfides [3,24]. Metals present in feed e.g., Nickel and Vanadium were uniformly distributed in the grey domains along with coke.
Table 6 CTE of P9 metal, FeS, good spalling & bad spalling coke.
CTE, 10− 6/°C CTE Delta with P9 Metal a b
P9a
FeSb
Good spalling cokea
Bad spalling cokea
12.5 0
9.5–11.3 1.2–3.0
5.2 7.3
7.3 5.2
Average CTE measured in the range of 200–750 °C. Average CTE measured in the range of 27–427 °C [25–27].
139
Fuel Processing Technology 172 (2018) 133–141
H. Bisht et al.
Feed side (F) mapping showed other elements characteristic of VR feed such as Si, N, Na and Ca were distributed uniformly over the carbonaceous phase. Presence of all major metals of P9 tube e.g., Fe, Cr, Mn & Mo in bright domains suggests that majority of inorganic sulfide are coming from tube metals or upstream equipment having similar metal. Composition of bright domains (inorganic phase) and grey domains (carbonaceous phase) was also analyzed by EDS spot analysis. This revealed that inorganic phase (bright domains) of bad spalling coke contains very high concentration of iron sulfide. Furthermore, the composition of inorganic phase was same irrespective of its location whether it was present in-between tube surface and coke layer or sandwiched between coke layers. Previous lab scale fouling study done on SS-316 wire showed the formation of iron sulfide layer only at the wire surface [20], whereas in this study iron sulfide was found on tube side as well as intercalated in between coke layers. SEM images of bad spalling coke revealed that inorganic phases deposited in between tube surface and coke layer was more continuous in comparison to that between different coke layers. EDS analysis of grey domains showed that carbonaceous phase is predominantly composed of carbon and sulfur with uniform distribution of metals e.g., V & Ni present in VR feed.
showed that bad spalling coke had higher sulfur and lower carbon contents. XRD analysis of bad spalling coke revealed that Fe and S were present as Pyrrhotite-3T (Fe7S8) and the characteristic graphitic 002 peak at 26.5° was absent. SEM-BSE images of bad spalling coke showed inorganic domains were present on tube side as well as intercalated between coke layers. SEM-EDS analysis of these inorganic domains showed presence of Fe and S whereas carbonaceous domains showed presence of carbon and sulfur along with other elements of VR feed such as V, Ni, Si, P, N, Ca, Na etc. uniformly distributed. Presence of Mo along with Fe in inorganic domains suggests furnace tubes and upstream equipment as primary source of metals in inorganic deposits. Furthermore, similarity of composition of inorganic domains (i) between coke and tube surface and (ii) intercalated between different coke layers suggests they originate from same metal source. Higher amount of naphthenic acids in VR feed may be one of the root cause for iron sulfide deposited along with coke in furnace tubes. This study suggest that iron sulfide is one of the primary causes of ineffective OLS in commercial delayed cokers due to two main reasons (i) FeS helps in strong binding of coke with tube metal (ii) it increases the CTE of coke which prevents crack formation during thermal shocks of OLS.
3.6. CTE of good and bad spalling coke
Supplementary data to this article can be found online at https:// doi.org/10.1016/j.fuproc.2017.12.013.
Appendix A. Supplementary data
CTE values of P9 metal, good spalling and bad spalling coke were measured from 200 to 750 °C which is the typical temperature range for OLS. Average CTE value of good spalling coke was 5.2 × 10− 6/°C whereas that of bad spalling coke was 7.3 × 10− 6/°C. We have also measured the CTE value of P9 metal. CTE value of pure iron sulfide in the temperature range of 27–427 °C was taken from literature [25–27]. Table 6 shows average CTE values of P9 metal, coke, iron sulfide and their delta from P9 metal. The delta in CTE of coke and P9 metal has decreased form 7.3 × 10− 6/°C for good spalling coke to 5.2 × 10− 6/ °C for bad spalling coke. This decrease in delta CTE makes thermal shocks of OLS less effective in separating coke from tube surface.
References [1] I.A. Wiehe, Process Chemistry of Petroleum Macromolecules, 1st edition, CRC, June 2008, pp. 267–272 (292–293). [2] I.A. Wiehe, A phase-separation kinetic model for coke formation, Ind. Eng. Chem. Res. 32 (11) (1993) 2447–2454. [3] S. Rahmani, W. McCaffrey, M.R. Gray, Kinetics of solvent interactions with asphaltenes during coke formation, Energy Fuel 16 (1) (2002) 148–154. [4] A.H. Alshareef, A. Scherer, X. Tan, K. Azyat, J.M. Stryker, R.R. Tykwinski, et al., Formation of archipelago structures during thermal cracking implicates a chemical mechanism for the formation of petroleum asphaltenes, Energy Fuel 25 (5) (2011) 2130–2136. [5] F. Vogt, R. Tonti, M. Hunt, L. Edwards, A preview of anode coke quality in 2007, Light Metals (2004) 489–493. [6] K. Kasai, T. Murayama, Y. Ono, Measurement of effective thermal conductivity of coke, ISIJ Int. 33 (1993) 697–702. [7] H.Y. Lee, S.H. Lee, J.B. Kim, J.H. Lee, Creep–fatigue damage for a structure with dissimilar metal welds of modified 9Cr–1Mo steel and 316L stainless steel, Int. J. Fatigue 29 (2007) 1868–1879. [8] P.E. Eaton, J. Williams, Delayed coker furnace fouling control laboratory correlation to field experience, NACE International Conference Papers, NACE International, Houston, TX, 2008, pp. 1–7 (paper 8670). [9] W.C. Kuru, C.B. Panchal, C.F. Liao, J.W. Palen, W.A. Ebert, Development of high temperature, high pressure fouling units, Proc. Understanding Heat Exchanger Fouling and Its Mitigation, Engineering Foundation Conf, 1997. [10] P. Mathiazhagan, A.S. Khanna, Arab. J. Sci. Eng. 34 (2C) (2009) 159–176. [11] S. Asomaning, C.B. Panchal, C.F. Liao, Correlating field and laboratory data for crude oil fouling, Heat Transf. Eng. 21 (3) (2000) 17–23. [12] B.D. Crittenden, S.T. Kolaczkowski, T. Takemoto, D.Z. Phillips, Crude oil fouling in a pilot-scale parallel tube apparatus, ECI symposium series, volume RP5, Proceedings of 7th International Conference on Heat Exchanger Fouling and Cleaning - Challenges and Opportunities, Engineering Conferences International, Tomar, Portugal, July 1–6, 2007. [13] T. Kuppan, Heat Exchanger Design Handbook, 1st edition, CRC Press, Feb 2000, pp. 393–422. [14] B.A. Garrett-Price, S.A. Smith, R.L. Watts, J.G. Knudsen, W.J. Marner, J.W. Suitor, Fouling of Heat Exchangers, Characteristics, Costs, Prevention, Control and Removal, Noyes Publications, New Jersey, 1985, pp. 17–18. [15] R.J. Parker, R.A. McFarlane, Mitigation of fouling in bitumen furnaces by pigging, Energy Fuel 14 (2000) 11–13. [16] K.A. Catala, M.S. Karrs, G. Sieli, A.A. Faegh, Advances in delayed coking heat transfer equipment, Hydrocarb. Process. (Feb 2009) 45–54. [17] J. Adams, G.C. Hughes, Coker furnace on-line spalling – safe, clean, proven and profitable, Paper Presented at AFPM (American Fuel and Petrochemical Manufacturers) Paper # AM-12-2012, 72 2012, pp. 1–11. [18] Fuels & lubricants handbook: technology, properties, performance and testing, in: George E. Totten (Ed.), ASTM Manual Series MNI37WCD Chapter 29, Properties of Fuels, Petroleum Pitch, Petroleum Coke and Carbon Materials, 2003, pp. 776–777. [19] T. Stephenson, A. Kubis, M. Derakhshesh, M. Hazelton, C. Holt, P. Eaton, et al., Corrosion-fouling of 316 stainless steel and pure iron by hot oil, Energy Fuel 25 (2011) 4540–4551.
3.7. Effect of operating conditions and feed properties on deposited coke The operating conditions, such as coil outlet temperature, furnace inlet pressure and feed rate were similar during bad spalling and good spalling period. Therefore, to understand the reasons for higher amount of iron sulfide during bad spalling period, feed properties were compared. It was noticed that TAN of vacuum residue feed was higher during bad spalling period. Naphthenic acids can react with tube metals (Fe, Cr, Mn and Mo) to form metal salts as corrosion products below < 400 °C. At higher temperatures (> 400 °C) naphthenic acids decompose to produce non-corrosive products [28]. In coker fractionator bottom and convection section of coker furnace, which are maintained at < 400 °C these corrosion products can be formed and get converted to metal sulfides by reaction with hydrogen sulfide and deposit in the radiation section of furnace tube. 4. Conclusions Systematic analysis of coke deposited in commercial delayed coker furnace tubes during good and bad spalling periods was done to understand root causes of ineffective OLS. VR feed was having slightly higher TAN and lower Sulfur during bad spalling period whereas other feed properties e.g., Ni, V, Fe, Na, Ca, density and CCR were varying within same range during both periods. Coke deposited during the two periods showed significant difference in composition, crystallinity, hardness and density. Bad spalling coke was comparatively softer and amorphous than good spalling coke. Real density and ash content of bad spalling coke was higher indicating higher concentration of high density inorganic compounds e.g. Pyrrhotite. Elemental analysis 140
Fuel Processing Technology 172 (2018) 133–141
H. Bisht et al.
Fractions, Marcel Dekker Inc., New York, 1994, p. 384. [24] M. Srinivasan, A.P. Watkinson, Fouling of some Canadian crude oils, Heat Transfer Eng. 26 (1) (1997) 7–14. [25] M.W. Hussain, M.Y.H. Ansari, OSR-JAP 9 (5) (Sep.–Oct. 2017) 61–63 Ver. II. [26] S.S. Sharma, Proc. Ind. Acad. Sci. 341 (1951) 72. [27] R.S.B. Chrystal, Trans. Faraday Soc. 512 (1965) 1811. [28] W.A. Derungs, Naphthenic acid corrosion – an old enemy of the petroleum industry, Corrosion 12 (12) (1956) 41–46.
[20] M. Derakhshesh, P. Eaton, B. Newman, A. Hoff, D. Mitlin, M.R. Gray, Effect of asphaltene stability on fouling at delayed coking process furnace conditions, Energy Fuel 27 (2013) 1856–1864. [21] P. Patnaik (Ed.), Handbook of Inorganic Chemicals, Mc GrawHill Professional, 2002. [22] B. Weidong, W.C. McCaffrey, M.R. Gray, Agglomeration and deposition of coke during cracking of petroleum vacuum residue, Energy Fuel 21 (2007) 1205–1211. [23] K.H. Altgelt, M.M. Boduszynski, Composition and Analysis of Heavy Petroleum
141
Contents lists available at ScienceDirect
Fuel Processing Technology journal homepage: www.elsevier.com/locate/fuproc
Research article
Effect of composition of coke deposited in delayed coker furnace tubes on on-line spalling
T
⁎
Harender Bishta, , Vijai Shankar Balachandranb, Mitra Patelb, Girish D. Sharmaa, Ashwani H. Yadava, Dwaipayan D. Biswasa, Sreenivas Pacharua, Sukumar Mandala, Asit K. Dasa a b
Reliance Industries Limited, Motikhavdi, Jamnagar 361140, India Reliance Industries Limited, Reliance Corporate Park, Ghansoli, Navi Mumbai 400701, India
A B S T R A C T Delayed coking is one of the most widely used residue upgradation process in crude oil refining where vacuum residue is thermally cracked and converted into distillates and petroleum coke. During heating of vacuum residue in coker furnace, coke is continuously deposited in furnace tubes which is removed at regular intervals by one of the three methods namely (i) pigging, (ii) steam air decoking or (iii) online spalling (OLS). OLS is one of the preferred methods of coke removal from coker furnace tubes. We have noticed very effective as well as “not so effective” OLS in a commercial delayed coker furnace. In present article, the changes in composition and properties of furnace coke during effective and ineffective OLS were systematically analyzed by various analytical tools such as Dilatometer, TGA, HRSEM, SEM-EDX and XRD. Based on these analyses, it was found that inorganic deposits such as iron sulphides and tube metal corrosion products play a major role in failure of OLS. Higher total acid number (TAN) of vacuum residue may be promoting the formation of iron sulfide deposits in furnace tube.
1. Introduction Delayed coking is one of the most widely used residue upgradation process where vacuum residue (VR) is thermally cracked and converted into distillates and petroleum coke. Typically VR feed enters in a delayed coker furnace at 300–350 °C and comes out at 485–505 °C. On reaching cracking temperature of ~ 400 °C, thermal cracking of alkyl side chains of asphaltenes gets initiated, consequently the core of asphaltene molecules become less soluble in VR [1,2]. The cracked asphaltenes core agglomerate and start to precipitate out of solution phase. These agglomerates on further propagation in furnace tubes experience still higher temperature ~ 500 °C and start depositing on the furnace tubes as coke [3,4]. The thermal conductivity of deposited coke [5,6] is typically between 0.3–3 W/m·K and that of P9 metal used in furnace tube is ~30 W/m·K [7]. Thus coke being ~30 times less conductive than heater tubes reduces the heat transfer coefficient of furnace tubes [8,9]. In order to maintain the desired furnace outlet temperature of feed, tube metal temperature is continuously increased up to the maximum allowable working temperature (MAWT). P9 metal used in delayed coker furnace tubes contains ~89 wt% Fe, ~9 wt% Cr and ~1 wt% Mo along with small amount of C (~ 0.1 wt%), Si (~ 0.22 wt%), Mn (0.73 wt%) and Ni (0.27 wt%) [10].
⁎
Deposition rate of coke and other associated inorganic materials in delayed coker depends on several factors such as feed velocity inside furnace tubes, [11,12] suspended solids in coker feed, [13] roughness of tube surface [14] and temperature of tube metal surface [8,9,12]. Once MAWT is attained, furnace tubes are required to be cleaned by one of the following three methods viz. a) Pigging [15] b) Steam air decoking [16] c) On-line spalling (OLS) [17]. During pigging the entire furnace is shut down and cooled to ambient temperature. A plastic pig, embedded with metal studs, is inserted in the fouled furnace tubes and pushed with help of high pressure water. Size of plastic pig is sequentially increased to a diameter slightly less than the inner diameter of furnace tube to remove maximum amount of coke from furnace tubes. In steam air decoking the fouled furnace is heated in presence of steam-air mixture to slowly burn the deposited coke. This method is less effective when significant amount of inorganic material is deposited in furnace tubes. OLS is one of the most economical methods of furnace decoking as it does not require complete furnace shutdown and external expert agencies to perform the spalling activities. OLS is done in one pass of furnace at a time whereas other passes of the same furnace are operated normally. Vacuum residue flow to the furnace pass where OLS is being carried out is reduced and steam flow is increased in tandem. Finally VR
Corresponding author. E-mail address: [email protected] (H. Bisht).
https://doi.org/10.1016/j.fuproc.2017.12.013 Received 17 August 2017; Received in revised form 19 December 2017; Accepted 19 December 2017 0378-3820/ © 2017 Elsevier B.V. All rights reserved.
Fuel Processing Technology 172 (2018) 133–141
H. Bisht et al.
flow is completely stopped and steam flow is maximized to get steam velocity in the range of 90 to 100 m/s. After attaining the desired steam velocity, temperature of furnace tubes is gradually raised to MAWT, typically in the range of 650–700 °C and maintained for 1–2 h to dislodge any loosely bound coke. Then temperature of furnace tubes is quickly dropped to 450–500 °C by manipulating fuel supply to furnace to give a mild thermal shock. This mild shock is given to ensure heater tubes are not plugged due to excessive coke removal. After mild thermal shock, temperature of furnace tubes is again increased to MAWT and hold for few hours. 2nd thermal shock is given by decreasing the tube temperature to 250–300 °C in < 1 h by manipulating fuel supply to furnace and introduction of boiler feed water in furnace tubes. 1 to 3 more shocks are given by repeating the sequence followed for 2nd deep shock to ensure maximum coke removal from tubes. After multiple thermal shocks, furnace tubes are heated again, vacuum residue is gradually introduces and steam flow is reduced to normal operation. Success of OLS is measured by reduction in tube metal temperature when the spalled pass is taken back in line for regular operation to maintain desired feed outlet temperature. OLS efficiency depends upon difference in coefficient of thermal expansion (CTE) between deposited coke and tube metal as well as binding strength of coke with tube metal surface. The former is primarily dependent on crystallinity of coke such as graphite content, bonding characteristics of C in constituent molecules [18] whereas the latter is dependent on the elemental composition of coke layer and its inorganic contents [14]. Elemental analysis of the deposited coke in a lab scale foulant testing unit showed C, Fe and S as major constituents [19,20]. In literature source of Fe and S has been attributed to feed and it was suggested that iron naphthenates in feed decomposes and reacts with sulfur to form iron sulphide, predominantly as Pyrrhotite [15]. In this article, we have compared the composition and properties of furnace coke removed during pigging from a ~9 Million Metric Ton per Annum commercial delayed coker furnace situated in Jamnagar, India. These coke samples were collected during pigging operation when OLS was very effective and when OLS was not so effective. This study attempts to understand the root cause of good and bad spalling of coke in a commercial delayed coker furnace. It was found that significant amount of iron sulfide was present in the coke deposited in furnace tubes during bad spalling period which may be a result of higher corrosion of tube metals and upstream equipment due to increased total acid number (TAN) and reduced sulfur in VR feed. CTE of bad spalling coke was higher and closure to that of P9 metal making is more prone to sticking during thermal shock of OLS.
Table 1 Average VR properties during good and bad spalling periods. Properties
Good spalling period
Bad spalling period
0.47 4.1 1.037 14.1 25.5 16 410 119 11.1
0.60 3.9 1.037 15.1 25.2 16 439 117 13.3
TAN, mg KOH/g Sulfur, wt% Density, g/cm3 Asphaltene, wt% CCR, wt% Recovery at 565 °C, wt% V, ppmw Ni, ppmw Fe, ppmw
coke and water mixture were collected from furnace tube outlet. This mixture was filtered through a mesh to collect coke flakes. The water along with finer coke powder was drained. The coke thus collected was kept in separate bags for each pass of the furnace. In this study, since the total thickness of coke layer was ~2 mm only, most of the coke flakes were intact having both tube side and feed side surfaces. In case, thickness of coke is higher there may be slicing of the coke layer into two or more layers hence care must be taken to identify the tube side and feed side surfaces. The screened coke samples were oven dried at 60 °C for 12 h to remove moisture. These coke samples have seen temperature of ~ 500 °C for more than 60 days in furnace tubes and were of different size and shape ranging from small granules to flakes of few centimeter sizes. 2.3. Hardgrove Grindability Index (HGI) HGI of coke samples was measured in HGI apparatus from Chemicals and Instruments Corporation, Kolkata as per ASTM D409. 2.4. Real density Real density of coke and ash samples was measures in Quantachrome Ultrapyc 1200e as per ASTM D2838. Coke samples were powdered to < 75 μm size before real density measurement. 2.5. X-ray diffraction XRD analysis was carried out by using Rigaku Miniflex600 power diffraction instrument. X-ray was generated from Cu X-ray tube operated at 60 kV and 1 mA. Graphite monochromator was used to get monochromatic Cu Kα radiation (1.54056 Ǻ) which was used for XRD measurement. The samples were powdered and uniformly spread over XRD sample holder. Then sample holder was placed on an auto sampler of the X-ray system. XRD patterns of the samples were collected in 5–50° (2θ) with scanning speed 1°/min and step size (0.01°). Phase identification in the sample was carried our using search match program of M/s Rigaku and using the JCPDS library.
2. Material and methods 2.1. Vacuum residue (VR) feed VR feed processed in delayed coker during good spalling and bad spalling period was produced from blends of 15–20 different crude oils. Although VR properties were maintained within specification window, there were some changes in feed properties during good spalling and bad spalling periods. Table 1 shows the average properties of VR feed processed during good spalling and bad spalling periods. Average TAN of VR was 0.47 mg KOH/g and 0.60 mg KOH/g during good spalling and bad spalling periods respectively. Average total sulfur in VR was 4.1 wt% during good spalling period whereas it has decreased to 3.9 wt % for bad spalling period. Other feed properties like CCR, metals, density were almost similar in both periods.
2.6. Thermogravimetric analysis TGA measurements were performed by using TA SDT-Q 600 instrument. Samples were placed in the analyzer and kept at 250 °C for 1 min. Subsequently, the temperature was increased from 250 °C to 900 °C at a heating rate of 100 °C/min. Decrease in weight of the sample w.r.t. the starting weight was recorded in percentage. Derivative of the curve was also determined in order to display peaks where the weight decrease was highest.
2.2. Furnace coke 2.7. CHNS analysis Coke collected after pigging of delayed coker furnace when OLS was effective was named as “good spalling” and that collected when OLS was ineffective was named “bad spalling”. During pigging the scrapped
CHNS measurement was performed using Elementar Vario micro cube instrument, where combustion tube and reduction tube were set at 134
Fuel Processing Technology 172 (2018) 133–141
H. Bisht et al.
temperature programmed desorption (TPD) trapping column for separation of N2, CO2, H2O and SO2 before detection by a thermal conductivity detector (TCD). Concentrations of carbon, sulphur, hydrogen and nitrogen were determined as per ASTM D5373 method using sulphanilamide as reference standard.
Table 2 Properties of good and bad spalling coke. Properties
Good spalling
Bad spalling
HGI Real density, g/cm3 Elemental analysis (wt%) Carbon Hydrogen Nitrogen Sulfur Others TGA % wt loss up to 150 °C in N2 (moisture) % wt loss up to 850 °C in N2 (moisture + VCM) Residue after 1000 °C in air
40 2.12
65 2.39
65.6 1.0 < 0.5 7.5 25.4
45.6 < 1.0 < 0.5 17.2 35.7
5.0 12.5 21.5
2.6 8.6 43.3
2.8. Ash generation 40 gm coke samples were crushed to make fine powder and heated up to 700 °C in a calcination furnace with 30SLPH air flow and maintained at this temperature for 12 h. The ash thus formed was analyzed for elemental analysis and real density. 2.9. Scanning electron microscopy (SEM) SEM imaging of coke samples was taken on NOVA NanoSEM650, FEI Co, USA with a field emission gun. The images were acquired with back scatter electron (BSE) mode using a concentric BSE detector. The samples were mounted on carbon tape and fractured to observe the cross section. The micrographs showed Z contrast based on atomic number to give bright domains for higher elements with varying grey scale. Coke samples were imaged on both sides. The side which was in contact with furnace tube was named as “Tube Side (T)” and the side which was in contact with vacuum residue was names “Feed Side (F)”. SEM-EDS was done with EDAX® SDD X-ray detector. Excitation was fixed at 10 keV and mapping was acquired using TEAM® software Version. 4.2.2.
Table 3 Ash of good and bad spalling coke. Sample
Spalling type
Sample wt, (gm)
Ash wt, (gm)
Ash, wt%
Real density, g/cm3
Sample 1
Bad spalling Good spalling
40.0129
10.0198
25.04
4.0335
40.0462
6.3219
15.79
3.8309
Sample 2
Table 4 Elemental composition of good and bad spalling cokea. Elements
C O Fe Ni Na Mg Al Si P Mo S Ca V a b
Good spalling
2.10. Coefficient of thermal expansion (CTE)
Bad spalling
Concentration (wt%)
Normalizedb (wt%)
Concentration (wt%)
Normalizedb (wt%)
0.3 37.5 4.7 1.5 4.5 0.4 2.2 6.0 2.4 29.5 3.7 3.1 3.9
0.0 39.2 4.9 1.6 4.7 0.4 2.3 6.3 2.5 30.9 0.0 3.2 4.1
1.9 40.3 19.8 1.3 2.0 0.6 1.0 5.0 0.9 17.3 3.6 2.8 3.6
0.0 42.7 21.0 1.4 2.1 0.7 1.0 5.3 0.9 18.3 0.0 2.9 3.8
Coefficient of thermal expansion of P9 metal, good spalling and bad spalling coke were measured on Unitherm Model 1161 Dilatometer from TA Instrument using Unitherm dilatometer Software. The instrument was calibrated with Crystallox standards. Argon gas was used as inert medium. 3. Results and discussion Hardgrove Grindability Index (HGI) of coke gives indication of its hardness/brittleness and hence to some extent the spalling character of the coke deposits. Higher HGI indicates softer coke whereas lower HGI indicate harder/brittle coke. Table 2 shows the HGI and real density of bad and good spalling coke. HGI value of bad spalling coke was much higher (softer coke) and was in the range from 59 to 90 whereas that of good spalling coke was ranging from 38 to 43 (more hard & brittle coke). This analysis suggests that the good spalling coke was harder and more brittle making them prone to crack during thermal shock of online spalling vis-à-vis bad spalling coke. Bad spalling coke is more prone to fine generation due to intercalated layer of iron sulfide thus showing higher HGI values. Real density of bad spalling coke was ranging from 2.2 to 2.6 gm/
Average of three measurements. Normalized after subtracting C & S values.
1150 °C & 850 °C respectively. 30–40 mg sample was weighed in tin containers and introduced into the combustion tube with help of an autosampler. The gases generated in combustion tube were swept out by helium into a copper filled tube, then passed through proprietary Table 5 Metal and oxygen content in good and bad spalling coke. Metal oxide
Mol. wt
Metal, wt%
O2, wt%
Metal good spalling, wt%
Cal. O2 good spalling, wt%
Metal bad spalling, wt%
Cal. O2 bad spalling, wt%
Fe2O3 NiO Na2O MgO Al2O3 SiO2 P4O6 MoO3 CaO V2O5 Total
159.7 74.7 62 40.3 102 60.1 219.9 143.9 56.1 181.9
69.9 78.6 74.2 60.3 52.9 46.8 56.3 66.6 71.5 56.0
30.1 21.4 25.8 39.7 47.1 53.2 43.7 33.4 28.5 44.0
0.77 0.26 0.74 0.06 0.36 1.00 0.39 4.88 0.50 0.65 9.60
0.33 0.07 0.26 0.04 0.32 1.14 0.30 2.44 0.20 0.51 5.61
5.25 0.34 0.52 0.17 0.25 1.32 0.23 4.58 0.73 0.96 14.36
2.26 0.09 0.18 0.11 0.22 1.51 0.18 2.29 0.29 0.75 7.89
135
Fuel Processing Technology 172 (2018) 133–141
H. Bisht et al.
Fig. 1. XRD of good spalling coke and bad spalling coke.
Intensity (cps)
Good Spalling Coke sample Good Spalling Coke sample Bad Spalling Coke sample
Intensity (cps)
Pyrrhotite 3T, syn Fe7S8, 00-025-0411
2θ θ
Fig. 2. BSE SEM images of bad spalling coke (BS) tube side (T), feed side (F) and cross section (X).
136
Fuel Processing Technology 172 (2018) 133–141
H. Bisht et al.
Fig. 3. BSE SEM images of good spalling coke (GS) tube side (T), feed side (F) and cross section (X).
cm3 whereas that of good spalling coke was in the range of 2.04 to 2.17 gm/cm3 (Table 2). In case of bad spalling coke the real density was found even higher than that of highly crystalline carbon e.g. graphite (2.09–2.23 gm/cm3). Presence of appreciable amount of pyrrhotite having real density of 4.61g/cm3 may be the reason for increase in real density of bad spalling coke.
3.2. Thermo gravimetric analysis (TGA) TGA of bad spalling coke in air showed ~55% weight loss with ~45% ash (inorganic residue) whereas good spalling coke showed ~79% weight loss and ~21% ash (Table 2). TGA results showed the combustion of coke started around 400 °C and completed below 600 °C. TGA performed in nitrogen atmosphere showed a weight loss of about 8–12 wt% out which 2.5–5 wt% was attributed to moisture which evaporates below 150 °C and remaining was attributed to volatile combustible matter (VCM). The inorganic residue obtained in presence of air after TGA analysis of good spalling coke was much lower (21–25 wt%) vis-à-vis bad spalling coke which contained 40–45 wt% residue. On the other hand, moisture and VCM content of bad spalling coke was lower than good spalling coke. These results corroborate that inorganic content of the deposited coke remains a determining factor for good and bad spalling.
3.1. CHNS analysis Elemental analysis of coke is tabulated in Table 2. Carbon % in good spalling coke was in the range of 63.2 to 67.9 wt% whereas Carbon % in bad spalling coke was significantly lower and varying from 40.6 to 56.5 wt%. The sulfur content in good spalling coke was ranging from 6.3 to 8.7 wt% whereas in bad spalling coke was much higher and ranging from 14.2 to 19.1 wt%. Carbon content of good spalling coke was much higher implying they were predominantly carbonaceous, whereas bad spalling coke was having high amount of inorganic matter. The inorganic matter was predominantly associated with sulfur as the sulfur content of bad spalling coke was significantly higher than that of good spalling coke.
3.3. Ash analysis As seen in Table 3, ash content obtained from good spalling and bad spalling coke was 15.79 wt% & 25.04 wt% respectively. 137
Fuel Processing Technology 172 (2018) 133–141
H. Bisht et al.
Fig. 4. Tube side SEM-EDS elemental mapping of bad spalling coke.
3.4. X-ray diffraction analysis (XRD)
Ash content of bad spalling coke was much higher than that of good spalling coke suggesting higher amount of inorganic material deposited along with coke during bad spalling period. Real density of ash generated from bad spalling coke was higher than that of good spalling coke. Among all oxides present in ash, NiO and Fe2O3 have highest densities (6.72 gm/cm3 & 5.25 gm/cm3 respectively) [21]. Considering the fact that quantity of NiO is same in both good and bad spalled coke ash, it follows logically to attribute the high real density of ash from bad spalling coke is due to higher level of Fe2O3. During calcination, metals present as sulfide get converted to oxide forms. The ash after calcination was milled and pellets were made from homogenized ash. The elemental composition of ash pellet was analyzed by SEM-EDS and reported in Table 4. Some residual C and sulfur was present in the ash. The metal and oxygen concentrations were normalized after subtracting the C and S from ash. Table 5 shows the inorganic oxides present in ash, their molecular weight and oxygen %. This data was used to calculate the oxygen content of ash which was compared with the oxygen data obtained by elemental analysis. The total oxygen content in ash by SEM-EDX was 37.5 wt% and 40.3 wt% for good and bad spalling coke respectively (Table 4). For good spalling ash the measured value is close to the calculated value of 36.9 wt% whereas the calculated oxygen content for ash of bad spalling coke was slightly lower than measured value (35.5 wt% vis-à-vis 40.3 wt%). Elemental analysis showed significant change in iron concentration of good and bad spalling ash. Other metals present in VR feed such as Ni, V, Al, Ca, Si, Na and Mg were present in similar concentration in both good and bad spalling coke. In addition to the above listed elements, the amount of Mo was significantly high in both good and bad spalled cokes. Based on metal analysis of VR feed (Table 1), such high concentration of Fe and Mo in deposited coke could not be attributed to feed metals and are expected to be coming from corrosion of tube metals or upstream equipment. This inference is also corroborated by higher TAN of the VR feed processes during bad spalling period.
XRD pattern of bad spalling coke matched with that of Pyrrhotite-3T having a formula of Fe7S8 (ICDD powder pattern: 00-025-0411) (Fig. 1). Carbon in bad spalling coke was present in more amorphous state and gave a broad shoulder like pattern whereas carbon in good spalling coke was less amorphous and the peaks were much sharp in comparison to bad spalling coke. The characteristic peak of graphitic carbon at 2θ 26.5° [002] was present in the XRD spectra of good spalling coke suggesting comparatively higher crystalline nature. 3.5. Scanning electron microscopy (SEM) In order to understand the microstructure and surface morphology, both coke samples were evaluated using SEM [22]. The objective of this analysis was a) to know the formation of various carbonaceous and inorganic phases during coking b) presence of different metals in coke c) identification and composition of inorganic phases intercalated in the coke layers. We chose to image coke samples in back scattering mode to obtain Z contrast which gave bright domains for higher atomic number elements with varying grey scale. This enabled to quickly demarcate carbonaceous and inorganic phases. Morphology and composition of deposited coke changed as we move inwards from tube side (T) to feed side (F). We have also fractured coke flakes to reveal the cross section and imaged for inclusions (X). 3.5.1. SEM of bad spalling coke Fig. 2 shows the back scattered electron (BSE) micrographs of bad spalling coke. Tube side images (BS1-T & BS2-T) show domains of inorganic phase (bright area) and carbonaceous phase (grey area). Careful examination of the micrographs shows that the bright inorganic phases were not appearing continuous due to peeling off from the coke surface. This may have resulted due to high mechanical shear during pigging operation which has exposed the underneath carbonaceous layer. These observations suggest that inorganic (non-carbonaceous) layers were continuously spread between tube metal and coke layer 138
Fuel Processing Technology 172 (2018) 133–141
H. Bisht et al.
Fig. 5. Feed side SEM-EDS elemental mapping of bad spalling coke.
which has resulted in firm binding of coke to tube metal surface. The composition of bright layer was predominantly iron sulfide. On the other hand, feed side images (BS1-F & BS2-F) showed coke surface was mostly composed of carbonaceous material with few isolated bright patches of inorganic layer. Careful observation of surface revealed that the inorganic phase was also intercalated between carbonaceous layers in bad spalling coke. Therefore we fractured the coke flakes to reveal the cross section (BS1-X & BS2-X) and it was observed that the bright inorganic layer is not only deposited between tube metal and coke but also intercalated in between different coke layers.
that of bad spalling coke. The morphology of tube side coke appeared mainly composed of carbon phase with very few isolated inorganic domains as seen in Fig. 3 (GS1-T & GS2-T). Unlike bad spalling coke, the inorganic domains are discrete suggesting that most of the tube metal surface was in contact with carbonaceous material which was easy to detach during OLS operation. Coke surface on the feed side (GS1-F & GS2-F) was also predominantly composed of carbonaceous material. Fractured cross sections of good spalling coke showed absence of any intercalated inorganic domains in the coke layer (GS1-X & GS2X).
3.5.2. SEM of good spalling coke BSE-SEM images of good spalling coke are markedly different from
3.5.3. Energy dispersive spectroscopy (SEM-EDS) To understand elemental composition of bright inorganic domains and grey carbonaceous domains, EDS mapping was done for both tube side (T) and feed side (F) surfaces of bad spalling cokes. This elemental mapping of whole coke surface showed that the bright domains on tube side (Fig. 4) and feed side (Fig. 5) were predominantly composed of iron, sulfur, manganese and chromium whereas the grey domains were predominantly composed of carbon and sulfur. Sulfur in VR mostly present in form of thiophenes, sulfides (both cyclic & acyclic) and sulfoxides [23]. Out of these, the sulfides are thermally reactive species and mostly contribute for formation of iron sulfides [3,24]. Metals present in feed e.g., Nickel and Vanadium were uniformly distributed in the grey domains along with coke.
Table 6 CTE of P9 metal, FeS, good spalling & bad spalling coke.
CTE, 10− 6/°C CTE Delta with P9 Metal a b
P9a
FeSb
Good spalling cokea
Bad spalling cokea
12.5 0
9.5–11.3 1.2–3.0
5.2 7.3
7.3 5.2
Average CTE measured in the range of 200–750 °C. Average CTE measured in the range of 27–427 °C [25–27].
139
Fuel Processing Technology 172 (2018) 133–141
H. Bisht et al.
Feed side (F) mapping showed other elements characteristic of VR feed such as Si, N, Na and Ca were distributed uniformly over the carbonaceous phase. Presence of all major metals of P9 tube e.g., Fe, Cr, Mn & Mo in bright domains suggests that majority of inorganic sulfide are coming from tube metals or upstream equipment having similar metal. Composition of bright domains (inorganic phase) and grey domains (carbonaceous phase) was also analyzed by EDS spot analysis. This revealed that inorganic phase (bright domains) of bad spalling coke contains very high concentration of iron sulfide. Furthermore, the composition of inorganic phase was same irrespective of its location whether it was present in-between tube surface and coke layer or sandwiched between coke layers. Previous lab scale fouling study done on SS-316 wire showed the formation of iron sulfide layer only at the wire surface [20], whereas in this study iron sulfide was found on tube side as well as intercalated in between coke layers. SEM images of bad spalling coke revealed that inorganic phases deposited in between tube surface and coke layer was more continuous in comparison to that between different coke layers. EDS analysis of grey domains showed that carbonaceous phase is predominantly composed of carbon and sulfur with uniform distribution of metals e.g., V & Ni present in VR feed.
showed that bad spalling coke had higher sulfur and lower carbon contents. XRD analysis of bad spalling coke revealed that Fe and S were present as Pyrrhotite-3T (Fe7S8) and the characteristic graphitic 002 peak at 26.5° was absent. SEM-BSE images of bad spalling coke showed inorganic domains were present on tube side as well as intercalated between coke layers. SEM-EDS analysis of these inorganic domains showed presence of Fe and S whereas carbonaceous domains showed presence of carbon and sulfur along with other elements of VR feed such as V, Ni, Si, P, N, Ca, Na etc. uniformly distributed. Presence of Mo along with Fe in inorganic domains suggests furnace tubes and upstream equipment as primary source of metals in inorganic deposits. Furthermore, similarity of composition of inorganic domains (i) between coke and tube surface and (ii) intercalated between different coke layers suggests they originate from same metal source. Higher amount of naphthenic acids in VR feed may be one of the root cause for iron sulfide deposited along with coke in furnace tubes. This study suggest that iron sulfide is one of the primary causes of ineffective OLS in commercial delayed cokers due to two main reasons (i) FeS helps in strong binding of coke with tube metal (ii) it increases the CTE of coke which prevents crack formation during thermal shocks of OLS.
3.6. CTE of good and bad spalling coke
Supplementary data to this article can be found online at https:// doi.org/10.1016/j.fuproc.2017.12.013.
Appendix A. Supplementary data
CTE values of P9 metal, good spalling and bad spalling coke were measured from 200 to 750 °C which is the typical temperature range for OLS. Average CTE value of good spalling coke was 5.2 × 10− 6/°C whereas that of bad spalling coke was 7.3 × 10− 6/°C. We have also measured the CTE value of P9 metal. CTE value of pure iron sulfide in the temperature range of 27–427 °C was taken from literature [25–27]. Table 6 shows average CTE values of P9 metal, coke, iron sulfide and their delta from P9 metal. The delta in CTE of coke and P9 metal has decreased form 7.3 × 10− 6/°C for good spalling coke to 5.2 × 10− 6/ °C for bad spalling coke. This decrease in delta CTE makes thermal shocks of OLS less effective in separating coke from tube surface.
References [1] I.A. Wiehe, Process Chemistry of Petroleum Macromolecules, 1st edition, CRC, June 2008, pp. 267–272 (292–293). [2] I.A. Wiehe, A phase-separation kinetic model for coke formation, Ind. Eng. Chem. Res. 32 (11) (1993) 2447–2454. [3] S. Rahmani, W. McCaffrey, M.R. Gray, Kinetics of solvent interactions with asphaltenes during coke formation, Energy Fuel 16 (1) (2002) 148–154. [4] A.H. Alshareef, A. Scherer, X. Tan, K. Azyat, J.M. Stryker, R.R. Tykwinski, et al., Formation of archipelago structures during thermal cracking implicates a chemical mechanism for the formation of petroleum asphaltenes, Energy Fuel 25 (5) (2011) 2130–2136. [5] F. Vogt, R. Tonti, M. Hunt, L. Edwards, A preview of anode coke quality in 2007, Light Metals (2004) 489–493. [6] K. Kasai, T. Murayama, Y. Ono, Measurement of effective thermal conductivity of coke, ISIJ Int. 33 (1993) 697–702. [7] H.Y. Lee, S.H. Lee, J.B. Kim, J.H. Lee, Creep–fatigue damage for a structure with dissimilar metal welds of modified 9Cr–1Mo steel and 316L stainless steel, Int. J. Fatigue 29 (2007) 1868–1879. [8] P.E. Eaton, J. Williams, Delayed coker furnace fouling control laboratory correlation to field experience, NACE International Conference Papers, NACE International, Houston, TX, 2008, pp. 1–7 (paper 8670). [9] W.C. Kuru, C.B. Panchal, C.F. Liao, J.W. Palen, W.A. Ebert, Development of high temperature, high pressure fouling units, Proc. Understanding Heat Exchanger Fouling and Its Mitigation, Engineering Foundation Conf, 1997. [10] P. Mathiazhagan, A.S. Khanna, Arab. J. Sci. Eng. 34 (2C) (2009) 159–176. [11] S. Asomaning, C.B. Panchal, C.F. Liao, Correlating field and laboratory data for crude oil fouling, Heat Transf. Eng. 21 (3) (2000) 17–23. [12] B.D. Crittenden, S.T. Kolaczkowski, T. Takemoto, D.Z. Phillips, Crude oil fouling in a pilot-scale parallel tube apparatus, ECI symposium series, volume RP5, Proceedings of 7th International Conference on Heat Exchanger Fouling and Cleaning - Challenges and Opportunities, Engineering Conferences International, Tomar, Portugal, July 1–6, 2007. [13] T. Kuppan, Heat Exchanger Design Handbook, 1st edition, CRC Press, Feb 2000, pp. 393–422. [14] B.A. Garrett-Price, S.A. Smith, R.L. Watts, J.G. Knudsen, W.J. Marner, J.W. Suitor, Fouling of Heat Exchangers, Characteristics, Costs, Prevention, Control and Removal, Noyes Publications, New Jersey, 1985, pp. 17–18. [15] R.J. Parker, R.A. McFarlane, Mitigation of fouling in bitumen furnaces by pigging, Energy Fuel 14 (2000) 11–13. [16] K.A. Catala, M.S. Karrs, G. Sieli, A.A. Faegh, Advances in delayed coking heat transfer equipment, Hydrocarb. Process. (Feb 2009) 45–54. [17] J. Adams, G.C. Hughes, Coker furnace on-line spalling – safe, clean, proven and profitable, Paper Presented at AFPM (American Fuel and Petrochemical Manufacturers) Paper # AM-12-2012, 72 2012, pp. 1–11. [18] Fuels & lubricants handbook: technology, properties, performance and testing, in: George E. Totten (Ed.), ASTM Manual Series MNI37WCD Chapter 29, Properties of Fuels, Petroleum Pitch, Petroleum Coke and Carbon Materials, 2003, pp. 776–777. [19] T. Stephenson, A. Kubis, M. Derakhshesh, M. Hazelton, C. Holt, P. Eaton, et al., Corrosion-fouling of 316 stainless steel and pure iron by hot oil, Energy Fuel 25 (2011) 4540–4551.
3.7. Effect of operating conditions and feed properties on deposited coke The operating conditions, such as coil outlet temperature, furnace inlet pressure and feed rate were similar during bad spalling and good spalling period. Therefore, to understand the reasons for higher amount of iron sulfide during bad spalling period, feed properties were compared. It was noticed that TAN of vacuum residue feed was higher during bad spalling period. Naphthenic acids can react with tube metals (Fe, Cr, Mn and Mo) to form metal salts as corrosion products below < 400 °C. At higher temperatures (> 400 °C) naphthenic acids decompose to produce non-corrosive products [28]. In coker fractionator bottom and convection section of coker furnace, which are maintained at < 400 °C these corrosion products can be formed and get converted to metal sulfides by reaction with hydrogen sulfide and deposit in the radiation section of furnace tube. 4. Conclusions Systematic analysis of coke deposited in commercial delayed coker furnace tubes during good and bad spalling periods was done to understand root causes of ineffective OLS. VR feed was having slightly higher TAN and lower Sulfur during bad spalling period whereas other feed properties e.g., Ni, V, Fe, Na, Ca, density and CCR were varying within same range during both periods. Coke deposited during the two periods showed significant difference in composition, crystallinity, hardness and density. Bad spalling coke was comparatively softer and amorphous than good spalling coke. Real density and ash content of bad spalling coke was higher indicating higher concentration of high density inorganic compounds e.g. Pyrrhotite. Elemental analysis 140
Fuel Processing Technology 172 (2018) 133–141
H. Bisht et al.
Fractions, Marcel Dekker Inc., New York, 1994, p. 384. [24] M. Srinivasan, A.P. Watkinson, Fouling of some Canadian crude oils, Heat Transfer Eng. 26 (1) (1997) 7–14. [25] M.W. Hussain, M.Y.H. Ansari, OSR-JAP 9 (5) (Sep.–Oct. 2017) 61–63 Ver. II. [26] S.S. Sharma, Proc. Ind. Acad. Sci. 341 (1951) 72. [27] R.S.B. Chrystal, Trans. Faraday Soc. 512 (1965) 1811. [28] W.A. Derungs, Naphthenic acid corrosion – an old enemy of the petroleum industry, Corrosion 12 (12) (1956) 41–46.
[20] M. Derakhshesh, P. Eaton, B. Newman, A. Hoff, D. Mitlin, M.R. Gray, Effect of asphaltene stability on fouling at delayed coking process furnace conditions, Energy Fuel 27 (2013) 1856–1864. [21] P. Patnaik (Ed.), Handbook of Inorganic Chemicals, Mc GrawHill Professional, 2002. [22] B. Weidong, W.C. McCaffrey, M.R. Gray, Agglomeration and deposition of coke during cracking of petroleum vacuum residue, Energy Fuel 21 (2007) 1205–1211. [23] K.H. Altgelt, M.M. Boduszynski, Composition and Analysis of Heavy Petroleum
141