1. Introduction
It is typical for petroleum refineries to process blends of crude oils instead of an individual crude oil [
1]. This practice aims at enhancing refinery margins by including in the crude diet cheaper so-called “opportunity” crudes, as crude-oil costs can account for around 80% of a refinery’s turnover [
2,
3]. Refineries usually blend premium crudes with low-quality crudes to gain from the higher profit margins of low-quality crudes. The crude oils coming from different fields have diverse quality properties described by the true boiling point (TBP) curves, and, generally, these crude oils, which have higher content of distillate fractions (naphtha, kerosene, diesel, and vacuum gas oil), are more valuable and have higher prices [
4]. The crude selection process usually involves the use of a linear programming tool that models the performance of the refinery searching for the economical optimum [
5,
6,
7,
8]. The complexity of refinery operations makes it difficult to formulate the suitable planning optimization models [
9]. One of the challenges in the linear programming models is how to properly represent the operation of crude distillation units (CDUs) [
8,
10]. They are among the most important process units in the refinery because they separate the crude oil in fractions (naphtha, kerosene, diesel oil, atmospheric residue, vacuum gas oils, etc.), which are processed in the downstream units (upgrading and conversion units). An inaccurate CDU operation representation will give inaccurate yields of the intermediate streams, and the entire downstream processing modelling will be erroneous. Thus, the overall economic performance evaluation of the refinery is directly influenced by the proper simulation of CDU operation [
8,
10]. Different models of CDU operation simulation have been reported in the literature [
7,
11,
12,
13]. However, our search of the literature has shown that the CDU operation simulation has not considered the deviation of distillation yields from crude oil blends from the additive blending rule, as reported in the research of Li et al. [
14,
15], when crude oil blends are processed. Unfortunately, the reports of Li et al. [
14,
15] employed equilibrium/flash vaporization equipment, and no data about the accuracy of this equipment have been announced. Moreover, Wang [
16] mentioned that the error of Chinese domestic distillation experimental results is large, and the reliability of the results is questionable. All of this was a reason for us to perform TBP distillation of different individual crude oils and their blends in a standardized equipment operating under requirements of the standards ASTM D 2892 [
17] and ASTM D 5236 [
18]. Before the performance of the TBP distillation experiments with the individual crude oils and their blends, several runs with four crude oils were carried out to demonstrate the repeatability of the equipment used.
The aim of this research was to test the validity of the additive blending rule for twelve different crude oil blends from seven individual crude oils and an imported atmospheric residue and search for a possible explanation in cases of observed deviations from the additive blending rule.
3. Results
Table 5,
Table 6,
Table 7,
Table 8 and
Table 9 present the results of the TBP measured and estimated yields of the crude oil blends from applying the additive blending rule shown in Equation (2).
where
BlendTBPyield is the yield of the TBP fraction of the crude oil blend, wt.%;
wt.frac.Crude1 is the weight fraction of crude oil 1 in the crude oil blend;
Crude1TBPyield is the yield of the TBP fraction in the crude oil 1, wt.%;
wt.frac.Crude2 is the weight fraction of crude oil 2 in the crude oil blend; and
Crude2TBPyield is the yield of the TBP fraction in the crude oil 2, wt.%.
Table 5,
Table 6,
Table 7,
Table 8 and
Table 9 also present the results of the measured and estimated density of the crude oil blends, assuming regular solution behavior of the crude oil blend, that is, no excess volume of mixing. The estimated density was calculated using Equation (3).
where
ρregular(estimated) is the estimated density of the crude oil blend, assuming regular solution behavior, g/cm
3;
ρCrude1 is the measured density of the first crude oil participating in the crude oil blend, g/cm
3; and
ρCrude2 is the measured density of the second crude oil participating in the crude oil blend, g/cm
3.
The data in
Table 5 show that the differences between measured and estimated TBP yields of the fractions of light naphtha (IBP-100 °C), heavy naphtha (100–180 °C), VGO (360–540 °C), and VR (>540 °C) are much bigger than the repeatability limits of the maximum 0.9 wt.% for the TBP analysis. The same is valid for the differences between measured and estimated crude-oil-blend densities, reporting differences much bigger than the repeatability limit of 0.00001 g/cm
3.
The data in
Figure 1 indicate the presence of clear relation of the difference between measured and estimated density, and the TBP yields of light naphtha, VGO, and VR to the content of Iranian-1 (IHCO-1) in the crude oil blends Sirtica–Iranian-1. This suggests that the existence of excess volume of mixing a result from the difference between the molecular interactions between the crude oils Sirtica and Iranian-1 may be the reason for the observed difference between measured and estimated TBP yields. The higher the density difference, the bigger the ∆TBP yields of naphtha, VGO, and VR are. It is interesting to note here that lower yields of naphtha are at the expense of higher yields of the heavy oils VGO and VR. It is also evident from the data in
Figure 1 that the magnitude of the differences between densities and TBP yields decreases with the reduction of the share of Sirtica crude oil in the blend (increasing the share of Iranian-1).
The data in
Table 6 indicate that, for all the studied Urals–Arab-Medium crude oil blends, the differences between measured and estimated TBP yields of light naphtha (IBP-110 °C) are much bigger than the repeatability limit of the TBP analysis. If one compares the data for the TBP light naphtha yields of both individual Urals and Arab Medium crude oils shown in
Table 1 (8.2 and 9.1 wt.%, respectively) with the measured yields of the blends of these crudes shown in
Table 6 (5.5, 6.3, and 6.5 wt.%), one can understand that some kind of suppression to extract the light naphtha from the blends Urals and Arab Medium occurred.
The data in
Figure 2 display that the difference in density correlates with the difference (∆) in the light naphtha yield (
Figure 2a) and that the ∆ density increases with the enhancement of the share of Urals in the crude oil blend Urals–Arab Medium.
The data in
Table 7 indicate that only the blend 33%Urals/67%Iranian-2 demonstrates that the ∆TBP yields of light naphtha and diesel are bigger than the repeatability limits of the TBP analysis. All other ∆TBP yields are within or very close to the repeatability limits.
The data in
Table 8 show that the ∆TBP yields of light and heavy naphtha and VGO are bigger than the repeatability limits of the TBP analysis of the blend 50%(80%Urals/20%Iranian-2)/50%AR. For the blend 75%(80%Urals/20%Iranian-2)/25%AR, only the ∆TBP yield of VGO is bigger than the repeatability limit of the TBP analysis.
The data in
Table 9 indicate that all ∆TBP yields are within the repeatability limits of the TBP analysis.
4. Discussion
For all twelve of the investigated crude oil blends from the seven individual crude oils, the blend densities deviate from the value of the density of the regular solution. This implies that the excess molar volume of the studied crude oils is not zero. In order to calculate the values of the excess molar volume (
VE) for the 12 crude oil blends, Equation (4) was employed, as reported in [
20].
where
M1 and
M2 are the molar masses of the crude oils participating in the blend,
ρ1 and
ρ2 are the densities of the individual crude oils, and
ρ denotes the density of the mixture.
Besides the excess molar volume, the relative changes in volume, ∆V, were also calculated by the use of Equation (5), as reported in [
21].
where
xi,
mi, and
are the mole fraction, molar mass, and molar volume of the individual component i, respectively; and
ρm is the density of the mixture. The molecular weights of the studied crude oils and their blends were estimated by the new empirical correlation reported in our recent research [
22].
Table 10 and
Table 11 summarize the values of the excess molar volume and ∆V of the 12 investigated crude oil blends, as estimated using Equations (3) and (4).
One can see from the data in
Table 10 and
Table 11 that there are negative and positive values for the molar excess volume and the relative changes in volume, ∆V. The positive excess molar volume indicates volume expansion upon mixing and, thus, a repulsive interaction of mixing crude oils or weaker interactions than the interactions of the individual crude oils. The negative excess volume shows stronger interactions of mixed molecules than individual molecules before mixing.
In order to assess the relations between the ∆TBP yields and the molar excess volume and the relative changes in volume, ∆V, an intercriteria analysis (ICrA) evaluation was performed. As input data, the ICrA method requires an m × n table with the measurements or evaluations of m objects against n criteria. As a result, it returns an n × n table with intuitionistic fuzzy pairs, defining the degrees of relation between each pair of criteria—hence the name “intercriteria”—and allows us to make informed decisions that take into account the inherent uncertainty that complex real-life problems exhibit. For the sake of terminological precision, in ICrA, the use of the term “correlation” when discussing the relationship between the criteria is avoided; the terms “positive consonance”, “negative consonance”, and “dissonance” are used instead. For laboratory experiments where the conditions are controlled to a higher extent than those in the industrial experiments, the meaning of μ = 0.75 ÷ 1.00; υ = 0÷0.25 denotes a statistically meaningful significant positive relation, where the strong positive consonance exhibits values of μ = 0.95 ÷ 1.00; υ = 0 ÷ 0.05, and the weak positive consonance exhibits values of μ = 0.75 ÷ 0.85; υ = 0.15 ÷ 0.25. Respectively, the values of negative consonance with μ = 0 ÷ 0.25; υ = 0.75 ÷ 1.00 mean a statistically meaningful negative relation, where the strong negative consonance exhibits values of μ = 0 ÷ 0.05; υ = 0.95 ÷ 1.00, and the weak negative consonance exhibits values of μ = 0.15 ÷ 0.25; υ = 0.75 ÷ 0.85. All other cases are considered to be dissonance. For a more detailed explanation of the essence of ICrA, the reader can refer to our previous study [
23].
Table 12 and
Table 13 summarize the values of μ and υ obtained from the ICrA evaluation for the 12 studied crude oil blends.
It is evident from the data in
Table 12 and
Table 13 that statistically meaningful negative relations exist between the ∆TBP yields light naphtha, kerosene, and VGO and between the ∆TBP yields of heavy naphtha and vacuum residue. This means that when the ∆TBP yields of light naphtha and kerosene go up, the ∆TBP yield of VGO decreases. When the ∆TBP yield of heavy naphtha increases, that of the vacuum residue diminishes. From these data, one may conclude that a deviation from the additive blending rule will be mainly a result from changes in the yields of the light crude oil fractions light and heavy naphtha and kerosene; and in the heavy oil fractions VGO, and VR.
It is also evident from the data in
Table 12 and
Table 13 that the molar excess volume and the relative changes in volume, ∆V, have a positive statistically meaningful consonance with the ∆TBP yields of light and heavy naphtha and a negative statistically meaningful consonance with the ∆TBP VR yield. This implies that an increase in the values of the
VE and ∆V will be associated with an increase in the values of ∆TBP yields of light and heavy naphtha, whereas an enhancement of the values of the
VE and ∆V will be accompanied by a reduction in the ∆TBP yield of the VR.
Figure 3 presents an illustration of the relation of the molar excess volume (
Figure 3a) and the relative changes in volume, ∆V (
Figure 3b), to the ∆TBP yields of heavy naphtha (100–180 °C) and vacuum residue (>540 °C).
It is evident from the data in
Figure 3a,b that at zero molar excess volume and with the relative changes in volume, ∆V, the ∆TBP yields of heavy naphtha and VR are also zero; that is, the yields of these crude oil fractions coincide with those estimated by the additive blending rule. Therefore, the lack of different intermolecular interactions between the individual crude oils to blend is a prerequisite for fulfillment of the additive blending rule. The presence of different intermolecular interactions between the individual crude oils to blend suggests the appearance of deviation from the additive blending rule. The data in
Figure 3 suggest that a reduction in the yield of heavy naphtha relative to that estimated by the additive blending rule is associated with an enhancement of the ∆TBP VR yield. Obviously, a redistribution between crude oil components occurs, resulting in a reduction in heavy naphtha recovery at the expense of a higher VR recovery rate. It is clear that this case is not favorable from an economic point of view because a lower production of high-value heavy naphtha and higher production of the lower-value vacuum residue will be registered when the molar excess volume of the blended crude oils is negative. The higher the magnitude of the negative value of the excess molar volume, the lower the yield of the high-value heavy naphtha and the higher the yield of the low-value vacuum residue are.
Out of the twelve studied crude oil blends, eight exhibited a deviation from the additive blending rule that was bigger than the repeatability limits of the TBP analysis, and four displayed deviations within the repeatability limits, thus confirming the validity of the additive blending rule for these four crude oil blends. A good correlation between the ∆TBP yields and the ∆ density for the cases where the ∆TBP yields were bigger than the TBP repeatability limits was observed (
Figure 1 and
Figure 2). This suggests that the deviation from the regular solution behavior may be the reason for the deviation from the additive blending rule. The observed relations of the ∆TBP yields of heavy naphtha and VR to the molar excess volume,
VE, and the relative changes in the molar volume, ∆V (
Figure 3), could be considered as evidence for the relation of the deviation from the regular solution behavior to the deviation from the additive blending rule. Thus, calculating
VE and ∆V by using Equations (3) and (4) may be deemed an indicator for expected observations of appearance or lack of deviations from the additive blending rule.