Correlating Feed Characteristics and Catalyst Properties with Fluid Catalytic Cracking Performance

Abstract

Feedstock quality has been proven to be the single variable that most affects fluid catalytic cracking (FCC) unit performance, but catalyst characteristics have also been reported in the literature to have a considerable effect on cracking process performance. How these two main variables of the FCC process complement each other in the search for ways to optimize the performance of the FCC unit is the subject of current research. Twenty-one feedstocks with K_W-characterizing factors ranging from 11.08 to 12.06, Conradson carbon contents ranging from 0.05 to 12.8 wt.%, and nitrogen contents ranging from 800 to 3590 ppm (wt/wt) (basic nitrogen from 172 to 1125 ppm (wt/wt)) were cracked on 21 catalysts with micro-activity between 67% and 76% (wt/wt) in a laboratory-based advanced catalytic evaluation (ACE) unit at a reaction temperature of 527 °C, catalyst–to-oil ratios between 3.5 and 12.0 wt/wt, and a catalyst time on stream of 30 s. Some of the feeds and catalysts tested in the laboratory FCC ACE unit were also examined in a commercial short-contact-time FCC unit resembling a UOP side-by-side design. It was found that conversion can be very well predicted in both the laboratory ACE and the commercial FCC units using multiple linear correlations developed in this work from information about the following feed properties: K_W-characterizing factor, nitrogen content, and micro-activity of the catalyst. The coke on the catalyst that controls the catalyst-to-oil ratio and the regenerator temperature in the commercial FCC unit could be calculated using the correlations developed in this work for the laboratory ACE and commercial FCC units, based on feed characteristics and catalyst micro-activity. Due to the greater slope of the Δ coke/Δ micro-activity dependence observed in the ACE FCC unit, the more active catalysts show weaker results compared to the less active catalysts at a constant coke yield. In contrast, catalysts with higher activity are preferable for operation in the commercial FCC plant because they provide higher conversion at the same coke yield due to the lower slope of the Δ coke/Δ micro-activity relationship.

1. Introduction

Fluid catalytic cracking (FCC), first introduced in 1942 in response to the demand for high-octane aviation gasoline during World War II, has become the heart of petroleum refining [1,2,3]. Over the eight decades of its existence, it has evolved from a producer of gasoline derived from gas oil to a producer of diesel fuel and propylene derived from feedstocks containing residues and bio-oils [4,5,6,7]. Over the years, FCC technology has proven its high flexibility in terms of processing feedstocks of different qualities [8,9,10,11], which may explain its widespread application in petroleum refining. About 600 FCC plants from 825 petroleum refineries are operating worldwide [12]. Optimal FCC plant performance is vital to the competitiveness of any refinery that uses an FCC process in its processing scheme. The right balance between feed characteristics and catalyst properties, consistent with the design constraints of the industrial catalytic cracking plant, is key to improving the performance of the cracking process. It has been shown that feedstock quality is the single variable that most influences the performance of a catalytic cracking plant [12]. Stratiev [12] systematically reviewed 213 publications appearing in the literature for the period of 1961 to 2025 that investigated the influence of feedstock characteristics on FCC performance and identified the best characteristic descriptors that allow prediction of the crackability of feedstock for FCC. It has been shown that, the higher the hydrogen content and the lower the aromatic carbon content, the higher the FCC conversion, reaching a maximum of 85% (wt/wt) for feeds with a maximum hydrogen content of 13.7% (wt/wt) and a minimum aromatic carbon content of 4.4% (wt/wt) [12]. Lappas et al. [13] have shown in their study that, the higher the saturate content of the FCC feed (the higher the hydrogen content [14]), the lower the aromatic content and the higher the olefin content of the FCC gasoline. More aromatic FCC feeds give lower gasoline yields, but this gasoline has a higher octane number than the gasoline produced from paraffinic feeds [15]. The activity of the equilibrium catalyst used in commercial FCC plants all over the world has shown a continuous increasing trend over the years [16,17]. It increased from 64 in 1997 [16] to 74 (wt/wt) in 2016 [17]. Along with enhancement of catalyst activity, a trend towards magnification of processing heavier, more contaminated residue feeds is reported by Pope et al. [17]. The percentage of FCC units that crack residual feedstocks has enlarged from about 40% in the early 2000s to about 50% in the second half of the second decade of the 21st century [17]. One may conclude that the heavier, more difficult to crack feeds require higher activity catalyst. Definitely, a higher-activity catalyst provides a higher conversion for more refractory feeds, as reported by Ng et al. [9], but coke selectivity is the other factor that controls the conversion level in an industrial FCC unit. The benefit of a highly active catalyst, which is characterized by high coke selectivity, may not be observed in a commercial FCC plant, as it may cause excessively high temperatures in the regenerator, which will lead to a decrease in the catalyst-to-oil ratio and ultimately to a decrease in the FCC conversion level, as well as a decrease in throughput. Thus, a proper balance between catalyst activity and coke selectivity is crucial to achieve the maximum conversion level for a given feedstock in a specific commercial FCC plant. To achieve this optimum performance, one needs to know how the conversion level and coke amount laid down on the catalyst depend on feed characteristics and catalyst activity. To establish the relationship between feedstock characteristics and catalyst activity, conversion, and the amount of coke deposited on the catalyst, 21 different feedstocks were cracked over 21 commercial catalysts with activities (micro-activity) between 67 and 76% (wt/wt) in a laboratory ACE unit. Correlations were developed to calculate conversion and coke yield (coke percent of the catalyst), and they were used to define the optimum performance of a short-contact-time commercial FCC unit. To the best of our knowledge, such correlations have not been reported in the literature yet.

The aims of this study are to discuss the dependence of the conversion and the amount of coke deposited on the catalyst during the cracking process on the properties of the feedstock and the activity of the catalyst, as well as to develop quantitative correlations allowing calculation and prediction of both the conversion and the level of coke on the catalyst.

2. Materials and Methods

Twenty-one FCC feeds whose properties are presented in

Table 1. Physical and chemical properties of FCC feeds cracked in this study.

FCC Feed Property	Method	Feed 1	Feed 2	Feed 3	Feed 4	Feed 5	Feed 6	Feed 7	Feed 8	Feed 9	Feed 10	Feed 11
D15°C, g/cm3	ASTM D4052 [18]	0.9355	0.9139	0.9174	0.9285	0.934	0.926	0.9181	0.973	0.901	0.920	0.914
Refractive index at 20 °C	ASTM D1747 [19]	1.5291	1.5143	1.5165	1.5267	1.5318	1.5282	1.5181	1.5613	1.5000	1.5106	1.5065
Molecular Weight, g/mol	[20]	366	345	378	358	407	325	316	583	367	388	342
Sulfur, % (wt/wt)	ASTM D4294 [21]	0.6089	0.26	0.37	0.5	0.683	0.612	0.471	0.985	0.176	1.695	1.608
Nitrogen, % (wt/wt)	ASTM D3228 [22]	0.27	0.20	0.19	0.27	0.35	0.36	0.33	0.25	0.08	0.15	0.14
Basic N, ppm (wt/wt)	ASTM UOP269-10 [23]	757	561	533	786	1099	1125	1009	701	224	421	398
Viscosity (98.89 °C), cSt	ASTM D445 [24]	9.6	7.1	8.9	8.5	12.8	6.8	6.2	152.4	7.4	9.7	7.0
Concarbon, %	ASTM D189 [25]	0.30	0.20	0.20	0.20	2.42	0.25	0.11	12.8	0.1	0.4	0.05
Initial Boiling Point (IBP), °C	ASTM D7169 [26]	299	310	310	308	338	255	277	224	282	311	303
5% Evaporated (wt/wt), °C	ASTM D7169 [26]	345	341	343	341	383	310	336	463	335	356	343
10% Evaporated (wt/wt), °C	ASTM D7169 [26]	365	356	360	357	402	331	357	534	357	377	361
50% Evaporated (wt/wt), °C	ASTM D7169 [26]	450	429	453	442	476	417	408	573	441	460	427
90% Evaporated (wt/wt), °C	ASTM D7169 [26]	525	510	518	516	540	524	475	614	528	541	482
95% Evaporated (wt/wt), °C	ASTM D7169 [26]	541	527	529	529	555	544	499	681	550	562	496
Final Boiling Point (FBP), °C	ASTM D7169 [26]	588	649	561	568	607	615	576	711	613	628	551
Kw-factor	[27]	11.67	11.83	11.92	11.72	11.83	11.61	11.66	11.83	12.06	11.92	11.82
Saturates, % (wt/wt)	In-house [28]	47	58	56.6	49.0	44.3	47.3	48.8	30	55.7	50.6	53.1
Aromatics, % (wt/wt)	In-house [28]	47.1	34.3	41.3	46.4	53.1	50.2	49	63.3	43.5	46.8	45.3
Light Aromatics, % (wt/wt)	In-house [28]	13	13.8	14.5	14.0	15.5	12.7	19.5	N.D.	20.9	13.8	14.2
Medium Aromatics, % (wt/wt)	In-house [28]	10	13.8	6.7	8.4	11.3	8	7.4	N.D.	9.91	16.9	14.3
Heavy Aromatics, % (wt/wt)	In-house [28]	24.1	6.7	20.1	24.0	26.3	29.5	22.1	N.D.	12.6	16.1	16.8
Resins, % (wt/wt)	In-house [28]	5.9	20.1	2.1	4.6	2.6	2.5	2.2	6.7	0.9	2.6	1.6
CA (n-d-M), % (wt/wt)	ASTM D3238 [29]	32.0	26.0	25.9	32.7	34.6	36.4	29.3	47.0	15.3	22.9	21.6
P, % (wt/wt)	[30]	61.1	62.9	64.4	61.6	65.3	60.1	60.8	88.7	65.1	65.0	63.4
Naph, % (wt/wt)	[30]	21.9	23.3	21.7	22.1	18.4	23.4	24.3	-4.4	23.7	22.0	24.0
A, % (wt/wt)	[30]	17.0	13.9	14.0	16.3	16.3	16.5	14.8	15.7	11.2	13.0	12.6
CA (Total), % (wt/wt)	[31]	25.1	20.4	20.2	25.3	25.5	27.6	22.8	24.4	13.0	15.1	15.1
H (Total), % (wt/wt)	[31]	11.8	12.4	12.3	12.0	12.0	11.9	12.2	12.1	12.8	12.3	12.4
D15°C, g/cm3	ASTM D4052 [18]		0.907	0.910	0.916	0.909	0.964	0.939	0.978	0.919	0.922	0.921
Refractive Index at 20 °C	ASTM D1747 [19]		1.5062	1.5111	1.5170	1.5111	1.5577	1.5393	1.5689	1.5132	1.5196	1.5190
Molecular Weight, g/mol	[20]		362	356	365	354	295	268	401	366	348	385
Sulfur, % (wt/wt)	ASTM D4294 [21]		0.164	0.294	0.269	0.381	0.480	0.369	0.666	0.386	0.321	0.283
Nitrogen, % (wt/wt)	ASTM D3228 [22]		0.12	0.16	0.15	0.14	0.35	0.25	0.28	0.18	0.1657	0.940
Basic N, ppm (wt/wt)	ASTM UOP269-10 [23]		313.59	414	393	367	1070	686	819	505	439	172
Viscosity (98.89 °C), cSt	ASTM D445 [24]		7.4	7.4	8.1	7.2	7.1	5.4	21.1	8.3	7.6	9.6
Concarbon, %	ASTM D189 [25]		0.1859	0.084	0.239	0.3911	0.34	0.15	2.67	0.2	0.07	0.51
Initial Boiling Point (IBP), °C	ASTM D7169 [26]		303	291	287	282	254	234	213	285	N.A.	N.A.
5% Evaporated (wt/wt), °C	ASTM D7169 [26]		353	350	353	328	316	300	395	340	349	368
10% Evaporated (wt/wt), °C	ASTM D7169 [26]		371	369	371	350	336	322	416	361	365	387
50% Evaporated (wt/wt), °C	ASTM D7169 [26]		439	436	444	434	403	373	484	445	433	459
90% Evaporated (wt/wt), °C	ASTM D7169 [26]		517	514	529	518	479	433	541	526	513	564
95% Evaporated (wt/wt), °C	ASTM D7169 [26]		536	533	556	540	503	455	556	544	533	596
Final Boiling Point (FBP), °C	ASTM D7169 [26]		584	588		607	555	517	595	598	576	688
Kw-factor	[27]		11.98	11.92	11.89	11.92	11.08	11.19	11.33	11.86	11.75	11.90
Saturates, % (wt/wt)	In-house [28]		N.D.	N.D.	N.D.	56.4	35	45	31	53.9	N.D.	N.D.
Aromatics, % (wt/wt)	In-house [28]		N.D.	N.D.	N.D.	42.5	62.2	52.7	66.5	43.7	N.D.	N.D.
Light Aromatics, % (wt/wt)	In-house [28]		N.D.	N.D.	N.D.	18.0	10.3	17.1	13.1	17.6	N.D.	N.D.
Medium Aromatics, % (wt/wt)	In-house [28]		N.D.	N.D.	N.D.	7.1	7.5	6.9	10.8	8.9	N.D.	N.D.
Heavy Aromatics, % (wt/wt)	In-house [28]		N.D.	N.D.	N.D.	17.4	36.1	28.7	32.9	17.1	N.D.	N.D.
Resins, % (wt/wt)	In-house [28]		N.D.	N.D.	N.D.	1.0	2.8	2.3	2.5	2.4	N.D.	N.D.
CA (n-d-M), % (wt/wt)	ASTM D3238 [29]		25.4	31.4	37.2	32.5	75.9	62.7	82.0	29.0	38.2	36.7
P, % (wt/wt)	[30]		64.5	64.0	63.8	64.3	47.9	52.8	59.7	63.1	61.7	64.3
Naph, % (wt/wt)	[30]		23.2	22.9	22.1	22.8	27.3	27.3	16.0	23.0	23.2	21.2
A, % (wt/wt)	[30]		12.3	13.1	14.1	12.9	24.8	19.9	24.3	13.9	15.1	14.5
CA (Total), % (wt/wt)	[31]		16.2	18.8	21.2	19.1	40.8	33.8	39.9	18.3	22.3	20.9
H (Total), % (wt/wt)	[31]		12.6	12.5	12.3	12.5	10.8	11.4	10.9	12.3	12.2	12.3

Note: N.D. = not determined; N.A. = not available.

were cracked in a laboratory ACE FCC unit. The standard methods and analytical procedures used to measure the properties of the FCC feeds are also listed in Table 1.

Table 2. Characteristics of the FCC catalysts used in this study.

Sample	Micro-Activity, % (wt/wt)	Al2O3, % (wt/wt) d.b.	RE2O3, % (wt/wt) d.b.	MgO, % (wt/wt) d.b.	P2O5, % (wt/wt) d.b.	Na2O, % (wt/wt) d.b.	SO4, % (wt/wt)	Carbon, % (wt/wt)	Ni, ppm (wt/wt)	V, ppm (wt/wt)	SA, m2/g	ZSA, m2/g	MSA, m2/g	UCS, Å
Catalyst 1	73.0	45.2	1.0	0.5033	0.26	0.19	N.A.	0.13	99	208	138	N.A.	N.A.	24.28
Catalyst 2	73.2	45.3	1.08	0.5033	0.23	0.19	N.A.	0.15	101	210	134	N.A.	N.A.	24.28
Catalyst 3	73.8	45.5	1.36	0.4920	0.11	0.2	N.A.	0.18	101	210	139	N.A.	N.A.	24.28
Catalyst 4	73.3	42.3	1.7	0.0486	0.07	0.26	0.06	N.A.	N.A.	N.A.	160	122	38	24.28
Catalyst 5	73.2	44.3	1.8	0.0461	0.09	0.28	0.07	N.A.	N.A.	N.A.	161	124	37	24.282
Catalyst 6	71.0	43.0	1.6	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	161	115	46	24.27
Catalyst 7	75.2	40.5	2.7	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	187	158	29	24.309
Catalyst 8	67.0	40.3	1.4	N.A.	N.A.	0.3	N.A.	N.A.	N.A.	N.A.	206	171	35	24.256
Catalyst 9	71.0	40.6	1.9	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	192	155	37	24.276
Catalyst 10	75.0	43.0	2.6	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	171	143	28	24.314
Catalyst 11	73.5	42.2	2.5	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	191	151	40	24.282
Catalyst 12	75.5	42.2	2.9	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	152	122	30	24.301
Catalyst 13	76.3	N.A.	2.31	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	153	114	39	24.31
Catalyst 14	72.8	N.A.	1.75	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	193	142	51	24.27
Catalyst 15	75.2	N.A.	1.59	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	195	143	51	24.3
Catalyst 16	74.2	N.A.	1.54	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	221	149	73	24.27
Catalyst 17	70.3	N.A.	2.22	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	153	86	66	24.27
Catalyst 18	73.5	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	24.28
Catalyst 19	72.5	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	24.28
Catalyst 20	73.3	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.
Catalyst 21	76.0	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.

Note: N.A. = not available.

summarizes some characteristics of the 21 catalysts used in this study. All catalysts investigated in this study are commercial-grade, and for this reason not all characteristics can be revealed.

Cracking experiments with the 21 FCC feeds from Table 1 on the catalysts shown in Table 2 were carried out in an ACE laboratory FCC unit [32], a diagram of which is shown in Figure 1. The experiments were performed at reaction temperature of 527 °C; a catalyst time on stream of 30 s; and a catalyst-to-oil ratio variation between 3.5 and 12.0 wt/wt The product fractionation was based on the gas chromatographic high temperature simulated distillation (ASTM D 7169) with the following ranges: cracked naphtha = C₅-221 °C; LCO (light cycle oil) = 221–343 °C; HCO (heavy cycle oil) = >343 °C. The cracking experiment was considered correct if the mass balance was within the range of 98% to 102% (wt/wt). The yields were normalized to 100% (wt/wt).

Feed conversion (X, % (wt/wt)) was calculated as 100 − LCO yield − HCO yield. The weight hourly space velocity (WHSV) was calculated using Equation (1).

$$WHSV={\displaystyle \frac{3600}{CTO\ast TOS},{\mathrm{h}}^{-1}}$$

(1)

where

CTO = catalyst-to-oil ratio, wt/wt

TOS = time on stream (30 s)

Assuming that the ACE fluidized bed reactor provides a perfectly mixed flow of the catalyst with a “plug” gas flow [33], a “plug” reactor model, whose integrated form is shown in Equation (2), can be used to calculate the reaction order and apparent kinetic constant.

$$Conversion=\left(1{-\left(\right(n-1)\times k\times \tau +1)}^{{\displaystyle \frac{1}{1-n}}}\right)\times 100$$

(2)

where

k = apparent kinetic constant; h⁻¹⁻ⁿ$\times$ frac.⁻¹;

τ = reaction time = ${\displaystyle \frac{1}{WHSV}}$, hour.

Some of the feeds from Table 1 and the catalysts from Table 2 were examined in a commercial FCC unit, whose process diagram is depicted in Figure 2. More details about this commercial FCC unit and how it operates are given in [34].

Intercriteria analysis (ICrA), whose essence is well described in our recent study [35], was employed to search for the presence of statistically meaningful relations among the feed properties, the catalyst characteristics and conversion, and the amount of coke deposited on the catalyst during the catalytic cracking process. The theory and application of ICrA are detailed in [36]. An intuitionistic fuzzy pair (IFP) $〈\mu ,\vartheta 〉$ $\in$ [0, 1] was used, where μ defines the degree of membership, validity, proximity, etc., while υ determines the degree of non-membership, non-validity, and non-proximity. For μ in the range 0.75–1.00 and υ between 0 and 0.25, an area of statistically meaningful positive consonance is identified, while at μ = 0–0.25 and υ = 0.75–1.00, a space of statistically meaningful negative consonance is derived. All other cases are considered dissonance. The results from the ICrA evaluation are presented in the form of an intuitionistic fuzzy interpretation triangle. When the IFP is close to the vertex with coordinates $〈\mathrm{1,0}〉,$ it is considered that the investigated variables are in positive consonance. When the IFP is close to the vertex with coordinates $〈\mathrm{0,1}〉$, this means that the two variables which concur to this point are in negative consonance. For the remaining IFPs, it is deemed that they are in dissonance. The statistically significant relationships between feedstock properties and catalyst characteristics related to FCC conversion and the amount of coke on the catalyst, as determined by ICrA, were further processed by multiple linear regression to establish correlations for calculating conversion and the quantity of coke on the catalyst.

3. Results and Discussion

3.1. Laboratory FCC ACE Experiments

It is generally accepted that the kinetics of catalytic cracking of vacuum gas oils obey a reaction order close to 2 [37,38,39,40]. However, our experimental data show that, if a reaction order of 2 is applied in all cases studied, the error in predicting the conversion can be relatively high for some regions of reaction times. As an example, cracking a feedstock with a K_W of 11.93 and a nitrogen content of 0.147 wt.% on a catalyst with a micro-activity of 69.8% (wt/wt), a surface area (SA) of 167 g/m², a rare earth oxide (RE₂O₃) content of 1.68% (wt/wt), and an average particle size (APS) of 101 μ at three reaction temperatures (527, 539, and 560 °C) showed that a reaction order of 2 underestimates conversion at low reaction times and overestimates it at high reaction times for the reaction temperatures of 539 °C and 560 °C, as illustrated in Figure 3a.

The data in Figure 3b indicate that catalytic cracking of the examined feed on the catalyst mentioned above at reaction temperatures of 539 °C and 560 °C is best described by a reaction order of 3.1, whereas at a reaction temperature of 527 °C, the best reaction order is indeed close to 2. A possible explanation for this finding could be that, under harsher conditions, the more reactive components are cracked very easily, and the unconverted feedstock is enriched in refractory species that show a lower dependence of the reaction rate on the reaction time, which in turn leads to the observation of a higher reaction order.

The data from 219 cracking experiments of 41 feed/catalyst pairs involving 21 FCC feeds and 21 catalysts were processed using Equation (2) to examine the generally accepted second order [37,38,39,40] and which order can provide the best fit. The best-fit order for the 41 feed/catalyst cases was found to vary between 1.2 and 3.05, and the average reaction order for these cases was 2.28. Figure S1 shows the kinetic plots of the 41 feed/catalyst cases with the best-fit order values and the apparent kinetic constants.

Figure 4 indicates that the second order gives a standard error of 1.23% (wt/wt) for the 219 cracking experiments, while the best-fit order results in a two-fold lower standard error (0.52% (wt/wt)). It is clear that, if the values of the second-order apparent kinetic constant are used to correlate with the properties of the feedstock and the characteristics of the catalyst, the resulting correlation will also be characterized by an error, which would ultimately increase the overall error in the conversion prediction. For that reason, in this study it was decided to correlate the feed and catalyst features with conversion at a catalyst-to-oil ratio (CTO) of 7.5 wt/wt This CTO was, to a certain extent, chosen arbitrarily, but also considering that 7.5 wt/wt is the typical CTO for most commercial FCC units working nowadays [17].

In order to investigate the relationships between conversion at a CTO of 7.5 wt/wt and the characteristics of feeds and catalysts, ICrA was performed. The data from the ICrA evaluation in terms of μ and υ are summarized in Figure 5.

Figure 5 presents a graphical intuitionistic fuzzy interpretation triangle for ICrA. The green dots imply the existence of positive consonance, while the blue dots denote the presence of negative consonance between the feed and catalyst characteristics and conversion at a CTO of 7.5 wt/wt As can be seen from the data in Figure 5, most of the feed and catalyst features and conversion are in dissonance (purple dots). A summary of selected properties that exhibited positive and negative consonance, quantified by μ and υ values, are given in

Table 3. μ values from ICrA evaluation of the properties of the feeds and catalysts shown in Table 1 and Table 2.

μ	Conv	MA	Δ Coke	D15	RI 20 °C	MW	N	Basic N	CCR	Kw	CA (n-d-M)	P	Naph	A	CA (Total)	H (Total)
Conv	1.00	0.49	0.32	0.16	0.12	0.57	0.14	0.16	0.35	0.82	0.20	0.77	0.39	0.10	0.09	0.82
MA	0.49	1.00	0.63	0.49	0.47	0.40	0.49	0.50	0.39	0.38	0.48	0.36	0.46	0.49	0.50	0.40
Δ Coke	0.32	0.63	1.00	0.74	0.70	0.50	0.68	0.69	0.54	0.25	0.61	0.30	0.37	0.69	0.68	0.21
D15	0.16	0.49	0.74	1.00	0.92	0.51	0.81	0.81	0.58	0.16	0.82	0.21	0.46	0.92	0.88	0.08
RI 20 °C	0.12	0.47	0.70	0.92	1.00	0.48	0.77	0.76	0.57	0.18	0.83	0.22	0.47	0.91	0.90	0.10
MW	0.57	0.40	0.50	0.51	0.48	1.00	0.43	0.43	0.62	0.63	0.44	0.68	0.08	0.45	0.42	0.55
N	0.14	0.49	0.68	0.81	0.77	0.43	1.00	0.98	0.49	0.20	0.70	0.23	0.54	0.82	0.82	0.21
Basic N	0.16	0.50	0.69	0.81	0.76	0.43	0.98	1.00	0.49	0.20	0.70	0.24	0.54	0.83	0.82	0.21
CCR	0.35	0.39	0.54	0.58	0.57	0.62	0.49	0.49	1.00	0.44	0.54	0.47	0.24	0.54	0.54	0.37
Kw	0.82	0.38	0.25	0.16	0.18	0.63	0.20	0.20	0.44	1.00	0.23	0.88	0.38	0.13	0.14	0.89
CA (n-d-M)	0.20	0.48	0.61	0.82	0.83	0.44	0.70	0.70	0.54	0.23	1.00	0.24	0.49	0.84	0.88	0.20
P	0.77	0.36	0.30	0.21	0.22	0.68	0.23	0.24	0.47	0.88	0.24	1.00	0.32	0.17	0.17	0.81
Naph	0.39	0.46	0.37	0.46	0.47	0.08	0.54	0.54	0.24	0.38	0.49	0.32	1.00	0.49	0.52	0.48
A	0.10	0.49	0.69	0.92	0.91	0.45	0.82	0.83	0.54	0.13	0.84	0.17	0.49	1.00	0.95	0.06
CA (Total)	0.09	0.50	0.68	0.88	0.90	0.42	0.82	0.82	0.54	0.14	0.88	0.17	0.52	0.95	1.00	0.11
H (Total)	0.82	0.40	0.21	0.08	0.10	0.55	0.21	0.21	0.37	0.89	0.20	0.81	0.48	0.06	0.11	1.00

and

Table 4. υ values from ICrA evaluation of the properties of the feeds and catalysts shown in Table 1 and Table 2.

ν	Conv	MA	Δ Coke	D15	RI 20 °C	MW	N	Basic N	CCR	Kw	CA (n-d-M)	P	Naph	A	CA (Total)	H (Total)
Conv	0.00	0.44	0.64	0.79	0.79	0.36	0.79	0.79	0.47	0.11	0.74	0.14	0.50	0.84	0.85	0.12
MA	0.44	0.00	0.28	0.40	0.39	0.47	0.40	0.40	0.39	0.50	0.41	0.49	0.40	0.40	0.40	0.49
Δ Coke	0.64	0.28	0.00	0.18	0.19	0.40	0.24	0.24	0.27	0.66	0.31	0.59	0.50	0.23	0.24	0.71
D15	0.79	0.40	0.18	0.00	0.04	0.47	0.18	0.18	0.30	0.82	0.17	0.72	0.47	0.06	0.11	0.90
RI 20 °C	0.79	0.39	0.19	0.04	0.00	0.47	0.18	0.20	0.27	0.78	0.11	0.71	0.46	0.04	0.06	0.85
MW	0.36	0.47	0.40	0.47	0.47	0.00	0.54	0.54	0.24	0.33	0.52	0.24	0.83	0.51	0.55	0.42
N	0.79	0.40	0.24	0.18	0.18	0.54	0.00	0.00	0.37	0.77	0.28	0.70	0.39	0.15	0.16	0.77
Basic N	0.79	0.40	0.24	0.18	0.20	0.54	0.00	0.00	0.37	0.78	0.29	0.71	0.40	0.17	0.18	0.79
CCR	0.47	0.39	0.27	0.30	0.27	0.24	0.37	0.37	0.00	0.42	0.32	0.35	0.57	0.33	0.33	0.49
Kw	0.11	0.50	0.66	0.82	0.78	0.33	0.77	0.78	0.42	0.00	0.73	0.06	0.54	0.84	0.84	0.08
CA (n-d-M)	0.74	0.41	0.31	0.17	0.11	0.52	0.28	0.29	0.32	0.73	0.00	0.70	0.44	0.14	0.11	0.78
P	0.14	0.49	0.59	0.72	0.71	0.24	0.70	0.71	0.35	0.06	0.70	0.00	0.59	0.77	0.78	0.14
Naph	0.50	0.40	0.50	0.47	0.46	0.83	0.39	0.40	0.57	0.54	0.44	0.59	0.00	0.44	0.43	0.45
A	0.84	0.4	0.23	0.06	0.04	0.51	0.15	0.17	0.33	0.84	0.14	0.8	0.4	0	0.04	0.93
CA (Total)	0.85	0.40	0.24	0.11	0.06	0.55	0.16	0.18	0.33	0.84	0.11	0.78	0.43	0.04	0.00	0.89
H (Total)	0.12	0.49	0.71	0.9	0.85	0.42	0.77	0.79	0.49	0.08	0.78	0.1	0.5	0.9	0.89	0

.

One can see from the data in Table 3 and Table 4 that conversion is in positive consonance with feed characteristics: K_W (μ = 0.82; υ = 0.11); P (μ = 0.77; υ = 0.14); and H (μ = 0.82; υ = 0.12). Conversion is in negative consonance with feed characteristics: D₁₅ (μ = 0.16; υ = 0.79); RI (μ = 0.12; υ = 0.79); N (μ = 0.14; υ = 0.79); basic N (μ = 0.16; υ = 0.79); and C_A (total) (μ = 009; υ = 0.85). ICrA, like other statistical tools, evaluates the presence or absence of statistically meaningful relations between pairs of variables. However, it cannot assess the combined effect of several variables on the target parameter. This can be done using multiple regression analysis. By employing this approach, we found that variables such as feed K_W, nitrogen content (N), and catalyst micro-activity (MA) have a statistically significant effect on conversion level. The developed multiple correlation linking conversion at a CTO of 7.5 wt/wt with the characteristics mentioned above, using 41 feed/catalyst cases, is presented as Equation (3).

$$\begin{gathered} {Conversion}_{ACE FCCU}=-198.338+17.337\times K_W-35.203\times N+0.946\times MA \\ \mathrm{R}=0.994; \mathrm{standard}\mathrm{error}=0.93\% (\mathrm{wt}/\mathrm{wt}) \end{gathered}$$

(3)

It is confirmed that all variables in Equation (3) have a significant influence on conversion, as judged by the very low p-values (the highest p-value was on the order of 10⁻¹⁶). This finding suggests that ICrA, or any correlation analysis, is good to apply together with multiple regression in order to identify the single and combined effects of several variables on the target parameter. It is worth mentioning here that the combination of the FCC feed characteristics K_W and N and the catalyst feature MA provided the correlation with the highest accuracy among other correlations, which used other parameters shown in Table 1 and Table 2. The other correlations developed in this work are not discussed here, because Equation (3) presents the best correlation.

Table S1 summarizes that feed K_W and N and catalyst micro-activity MA data, conversion at a CTO of 7.5 wt/wt, and the data calculated using Equation (3), and Figure 6 shows a parity graph of calculated and measured conversion at a CTO of 7.5 wt/wt.

In addition to conversion, another parameter affecting the performance of a commercial FCC plant is the coke deposited on the catalyst, as it affects the regenerator temperature and the associated catalyst-to-oil ratio [41,42]. The higher the amount of coke deposited on the catalyst, the higher the regenerator temperature and the lower the catalyst-to-oil ratio. In this research, the coke laid down on the catalyst surface was denoted as Δ coke. Although Δ coke, as evident from the data in Table 3 and Table 4, did not show a statistically meaningful relation with any feed or catalyst property, as quantified by the ICrA evaluation, multiple regression analysis indicated that the feed characteristics of K_W, Conradson carbon content (CCR), and molecular weight (MW), together with catalyst micro-activity, have a significant impact on Δ coke. The developed multiple correlation in this study that allows calculation of Δ coke is presented as Equation (4).

$$\begin{gathered} {∆Coke}_{ACE FCCU}=3.3502+0.0662\times CCR+0.00312\times MW-0.5519\times K_W+0.03508\times MA \\ \mathrm{R}=0.966; \mathrm{standard}\mathrm{error}\mathrm{of}0.084\% (\mathrm{wt}/\mathrm{wt}) \end{gathered}$$

(4)

The agreement between the measured Δ coke and the Δ coke calculated using Equation (4) is shown in Figure 7.

Equation (4) shows that only feed K_W has a suppressing effect on the amount of coke deposited on the catalyst surface. Feed Conradson carbon content and molecular weight, along with catalyst micro-activity, promote catalyst coking, which is in line with reports in the literature [43,44,45].

Equations (3) and (4) indicate that feeds with a high K_W are more reactive and less coke-forming. Each 0.1 increase in K_W is associated with an increase in conversion of 1.7% (wt/wt) and a reduction in coke on the catalyst by 0.055% (wt/wt). While nitrogen affects the reactivity of the feedstock, as apparent from Equation (3), it appears to have no influence on the coke on the catalyst, as evident from Equation (4). The activity of the catalyst, expressed by the value of its micro-activity, affects both conversion (see Equation (3)) and coking on the catalyst (see Equation (4)), as it increases both. While the former has a positive effect on conversion at the same catalyst-to-oil ratio, the latter has a negative effect on conversion, because more coke on the catalyst leads to lowering of the catalyst-to-oil ratio, and consequently decreases conversion [46]. In order to evaluate how catalyst activity influences conversion = at the same coke yield, considering that commercial FCC units are typically operated in a constant coke yield mode, the performance of two catalysts with different micro-activities (MAs of 71 and 76% (wt/wt)) was compared at constant coke yield during cracking of two feeds with different crackability (Feed 14, with a K_W of 11.886 and an N content of 0.150% (wt/wt); Feed 12, with a K_W of 11.98 and an N content of 0.120% (wt/wt)). Figure 8 presents graphs of the dependence of coke yields on conversion for the catalysts and feeds mentioned above.

The interpolated conversion at constant coke of 2.3% (wt/wt) for the more refractive Feed 14 for the lower-activity Catalyst 6 (MA of 71% (wt/wt)) is 68.6% (wt/wt), while for the more active Catalyst 21 (MA of 76% (wt/wt)) it is 64.0% (wt/wt). These results suggest that the more active Catalyst 21 would exhibit worse performance in a commercial heat-balanced FCC unit when more refractory feeds are cracked.

The interpolated conversion at constant coke of 2.3% (wt/wt) for the more reactive Feed 12 for the lower-activity Catalyst 9 (MA of 71% (wt/wt)) is 69.9% (wt/wt), while for the more active Catalyst 12 (MA of 75.5% (wt/wt)) it is 68.7% (wt/wt). These results show again that the more active catalyst performs worse when the comparison is made at constant coke yield.

These comparative results suggest that more active catalysts, although resulting in higher conversion at a constant catalyst-to-oil ratio, exhibit poorer performance at constant coke yield due to the overwhelming effect of the reduced CTO, when ACE cracking data are used. The observed trend toward increased micro-activity of catalysts used in the commercial FCC units worldwide, however, supposes that more active catalysts may perform better in commercial FCC units [16,17]. That was the reason why we investigated the performance of catalysts of various activity when feeds with diverse quality were cracked in a commercial FCC unit, whose process diagram is shown in Figure 2. The commercial cracking experiment results are discussed in the next section (Section 3.2).

3.2. Commercial FCC Experiments

The commercial cracking experiments were performed at a reaction temperature of 540 °C and catalyst-to-oil ratios in the range of 7 to10 wt/wt The micro-activity of catalysts employed in the commercial FCC unit varied between 69 and 76% (wt/wt). The feed characteristic variations were as follows: 11.77 ≤ K_W ≤ 11.98; 0.123 ≤ N ≤ 0.166% (wt/wt). From the operation of the commercial FCC unit for a period of 9 years, 12 cases were selected, comparing conversion at the same catalyst-to-oil ratio of 7.5 wt/wt Similar to the experiments conducted in the laboratory ACE unit, the less active catalysts formed less coke on their surface in the commercial FCC unit as well, making comparison at the same CTO a challenging task. For this reason, periods were selected when the catalyst stripping zone in the FCC reactor was operating at lower efficiency, which increased the typical Δ coke value of the less active catalysts. This allowed collection of conversion data at a CTO of 7.5 wt/wt for catalysts with different micro-activities. Using these 12 cases, the data from which are summarized in Table S2, allowed the development of a multiple correlation, shown as Equation (5).

$$\begin{gathered} {Conversion}_{commercial FCCU}=-191.922+18.286\times K_W-63.025\times N+0.809\times MA \\ \mathrm{R}=0.998; \mathrm{standard}\mathrm{error}\mathrm{of}0.27\% (\mathrm{wt}/\mathrm{wt}) \end{gathered}$$

(5)

Figure 9 shows a parity plot of the measured conversion and conversion calculated usgin Equation (5) in a commercial FCC unit at a CTO of 7.5 wt/wt.

One can see, comparing Equation (3) with Equation (5), that feed K_W has almost the same value of the regression coefficient, and the micro-activity coefficients are also very close, but the coefficient of nitrogen content in the commercial FCCU regression is twice as high as that in the laboratory ACE FCC unit. It appears that the much shorter catalyst contact time in the commercial FCC unit (between 2 and 3 s) favors the inhibitory effect of nitrogen to a greater extent compared to that in the laboratory ACE FCC unit with a longer contact time (30 s). The results of vacuum gas oil cracking reported by Caeiro et al. [47] and Stratiev [48] have shown that, during laboratory experiments conducted in ACE FCC [47] and micro-activity test (MAT) [48] FCC units, the inhibitory effect of nitrogen is more pronounced at lower catalyst-to-oil ratios, i.e., at shorter reaction times (${\displaystyle \frac{1}{WHSV}}$) (see Equation (1)). If the reaction time in the commercial FCC unit with a 3 s catalyst contact time and a catalyst-to-oil ratio of 7.5 wt/wt is calculated using Equation (1), a reaction time of 0.00625 h is obtained. The reaction time in the laboratory ACE FCC unit at a catalyst-to-oil ratio of 7.5 wt/wt and a catalyst-to-oil contact time of 30 s is equal to 0.0625 h, which is 10 times longer than that in the commercial FCC unit. Therefore, the observed stronger inhibitory effect of nitrogen in the commercial unit with a short contact time compared to that in the laboratory ACE FCC unit with a longer contact time may be attributed to the tenfold shorter reaction time in the commercial FCC unit compared to that in the laboratory ACE FCC unit. Nevertheless, the physicochemical mechanism behind this observation is still unclear and requires further study.

Similar to the Δ coke from the laboratory ACE FCC unit, the Δ coke in the commercial FCC unit was also found to depend on the feed K_W and the micro-activity of the catalyst, as shown in the developed multiple correlation (Equation (6)).

$$\begin{gathered} {∆Coke}_{commercial FCCU}=4.2402-0.4016\times K_W+0.01463\times MA \\ \mathrm{R}=0.941; \mathrm{standard}\mathrm{error}0.026\% (\mathrm{wt}/\mathrm{wt}) \end{gathered}$$

(6)

The agreement between observed Δ coke and Δ coke computed using Equation (6) from the studied commercial FCC unit is presented in Figure 10.

In contrast to the ACE FCC Δ coke correlation, the commercial Δ coke correlation does not include the feed properties of Conradson carbon content and molecular weight. This can be explain by the very narrow range of variation in these properties (348 g/mol ≤ MW ≤ 365 g/mol; 0.07% (wt/wt) ≤ CCR ≤ 0.39% (wt/wt)), which prevented them from having a notable influence on the Δ coke in the commercial FCC unit. The data in Figure 7 and Figure 10 show that the correlations developed to calculate Δ coke in both the laboratory ACE FCC unit and the commercial FCC unit are less accurate than the correlations developed to calculate the conversion at a CTO of 7.5 wt/wt (Figure 6 and Figure 9). The reason for this observation may lie in the different coke-forming properties of the catalysts tested in this study, which can depend on various structural and compositional characteristics of the catalysts [49,50]. Comparing Equation (6) (commercial FCC plant) with Equation (4) (laboratory ACE FCC plant), it can be seen that the effect of catalyst activity on Δ coke is higher by a factor of for the ACE plant compared to the commercial FCC plant. The shorter contact time of the commercial FCC unit results in a lower influence of catalyst activity on Δ coke. As discussed by Wear [46], Δ coke affects both the catalyst-to-oil ratio and the regenerator dense bed temperature. The data in Figure 11 for the commercial FCC unit confirm this statement. They show that a higher Δ coke value, which is typical for more active catalysts, is associated with a decrease in CTO and an increase in regenerator temperature. For the commercial FCC unit, the conversion increment at each unit of CTO magnification is approximately 2% (wt/wt) in the investigated CTO range between 7 and 10 wt/wt The data in Figure 11a show that, for each 0.1% (wt/wt) increment in Δ coke, the CTO drops by 1.6 wt/wt, which is a 3.2% (wt/wt) lower conversion.

Using Equations (5) and (6) along with the regression embedded in Figure 11 allows assessment of the effect of catalyst activity on the heat-balanced FCC unit conversion level. For example, a micro-activity difference of 4 wt.% would result in 3.2 wt.% higher conversion at a CTO of 7.5 wt/wt and a 0.06% (wt/wt) higher Δ coke, which is a 0.9 wt/wt lower CTO and lower conversion, as a consequence of the CTP being diminished by 1.8% (wt/wt). The total effect, however, would be equal to 3.2 − 1.8 = 1.4 wt.% higher conversion, implying that the more active catalyst would result in higher conversion of the feed in the commercial FCC unit, in contrast to the conclusion made on the base of the ACE tests that was discussed in the previous section. To confirm this deduction, tests in the commercial FCC unit were performed with two catalysts with micro-activities of 76 and 72% (wt/wt), which were used to crack two different FCC feeds (K_W of 11.98 and N of 0.12% (wt/wt); and K_W of 11.77 and N of 0.17% (wt/wt)), the results of which are summarized in

Table 5. FCC commercial test results with two catalysts with different activities and two feeds with different reactivities.

	High-Activity Catalyst	Low-Activity Catalyst
Parameters	Refractory Feed (KW of 11.98 and N of 0.12% (wt/wt))
Micro-activity,% (wt/wt)	76	72
Δ Coke, % (wt/wt)	0.63	0.55
Catalyst-to-oil ratio, wt/wt	7.71	8.84
Conversion, % (wt/wt)	74.7	73.0
Coke yield, % (wt/wt)	4.83	4.90
	Reactive feed (KW of 11.77 and N of 0.17% (wt/wt))
Micro-activity, % (wt/wt)	75.5	72
Δ Coke, % (wt/wt)	0.55	0.48
Catalyst-to -oil ratio, wt/wt	8.91	9.99
Conversion, % (wt/wt)	81.6	80.4
Coke yield, % (wt/wt)	4.90	4.80

.

The data in Table 5 confirm that the more active catalysts perform better in the commercial FCC unit, providing conversion enhancement between 1.7 wt.% for the refractory feed and 1.2% (wt/wt) for the more reactive feed. The coke yield was between 4.8% and 4.9% (wt/wt), reaffirming that the heat-balanced commercial FCC unit operates at a constant coke yield.

The more active catalysts make more coke on their surface, which increases the regenerator temperature, which can be an issue if the maximum temperature limit in the regenerator is reached, as discussed in our earlier study [51]. In this case, a lower active catalyst is recommended to cope with the higher regenerator temperature. Nevertheless, the catalyst activity should be reduced very carefully, as the results from this study show that more active catalysts perform better in a commercial heat-balanced FCC unit. In the laboratory ACE FCC unit, the more active catalysts perform worse at constant coke yield due to the steeper slope of the Δ coke/Δ micro-activity relationship compared to that observed in the commercial FCC unit.

4. Conclusions

Twenty-one different feeds were cracked over 21 distinct catalysts in a laboratory ACE FCC unit at various catalyst-to-oil ratios. It was found that the conversion of the investigated feeds did not obey the generally accepted second order. Instead, the best-fit order was determined to vary between 1.2 and 3.05.

Conversion at a catalyst-to-oil ratio of 7.5 wt/wt and a reaction temperature of 527 °C was examined to determine the relationship with feed properties and catalyst features, using ICrA. It was determined that the feed properties of K_W-characterizing factor, paraffin portion (P), hydrogen content (H), density (D¹⁵), refractive index (RI), nitrogen content (N), basic nitrogen content (basic N), and aromatic carbon content (C_A (total)) have statistically meaningful consonance with the conversion level at a catalyst-to-oil ratio of 7.5 wt/wt Multiple linear regression analysis showed that the best correlation for calculating conversion is that which uses the K_W-characterizing factor and the nitrogen content of the feedstock, as well as the micro-activity of the catalyst.

Another correlation was developed to calculate the coke content on the catalyst, showing that Δ coke depends on the feed properties of K_W-characterizing factor, Conradson carbon content (CCR), and molecular weight (MW), as well as the micro-activity of the catalyst.

The same correlations for calculating conversion at a catalyst-to-oil ratio of 7.5 wt/wt and Δ coke were developed based on data from a commercial FCC unit operating at a reaction temperature of 540 °C. While the regression coefficients of the K_W factor characterizing the feedstock and the micro-activity of the catalyst were close to those of the multiple correlation developed based on data from the ACE FCC laboratory experiment, the nitrogen content in the commercial FCC feed showed an inhibitory effect that was twice as strong as that seen with the ACE FCC laboratory process. The stronger inhibitory effect of nitrogen on catalytic cracking of gas oil in the commercial FCC unit may be attributed to the tenfold shorter reaction time compared to that seen in the laboratory ACE FCC unit, an observation that is in line with other studies examining the poisoning effect of nitrogen on catalytic cracking of gas oil. However, additional research is needed to determine out the physicochemical mechanism underlying this observation.

The commercial FCC Δ coke correlation shows a 2.4 times lower regression coefficient value for catalyst micro-activity compared to that of the ACE FCC Δ coke correlation. This means that the ACE FCC cracking experiments overestimate the effect of catalyst micro-activity on Δ coke compared to the commercial FCC experiments. This may explain why more active catalysts show worse performance in the ACE FCC plant when the performance is compared at a constant coke yield. More active catalysts in the commercial heat-balanced FCC plant demonstrate higher conversion at the same coke yield due to the steeper slope of the Δ coke/Δ micro-activity relationship. Catalysts with higher activity are preferable for operation in the commercial FCC plant, because they result in higher conversion at the same coke yield. This statement is valid unless the maximum limit of the regenerator temperature is reached. In such cases, catalysts with lower activity that produce less Δ coke are required.