Search in book...
Toggle Font Controls
Create new playlist

Name your new playlist

Playlist description (optional)
Sign In

Email address

Password

Forgot Password?

or

Continue with Facebook

Continue with Google
Sign Up

Full Name

Email address

Confirm Email Address

Password

or

Continue with Facebook

Continue with Google

Chapter Two

Basic Science of the Fouling Process

F. Coletti¹, B.D. Crittenden², and S. Macchietto³ ¹Chief Technology Officer, Hexxcell Ltd, UK ²Professor of Chemical Engineering, University of Bath, UK ³Professor of Process Systems Engineering, Imperial College London, UK

Abstract

Chapter 2 reviews the science of the fouling process. The basic fouling mechanisms—crystallization, particulate, chemical reaction, corrosion, and biological fouling—are described with particular focus on the mechanisms underlying crude oil fouling: chemical reaction and corrosion fouling. Possible routes of formation of crude oil fouling deposits are then discussed. These include autoxidation, corrosion, polymerization, thermal cracking, and asphaltene precipitation. Several events occurring in fouling—initiation, transport, attachment, removal, and ageing—are also discussed. Finally, the chapter reviews the effects that key variables such as crude oil composition, operating conditions, and surface conditions have on these events.

Keywords

Asphaltene precipitation; Autoxidation; Corrosion; Crude oil compatibility; Crude oil fouling; Events in fouling; Fouling mechanisms; Fouling science; Variables affecting fouling

The first attempt to rationalize and categorize fouling was made by Epstein (1983). He summarized the level of understanding in different aspects of the fouling process in what is now known as the 5 × 5 fouling matrix (Figure 2.1(a)). The five columns of the matrix list the different fouling mechanisms and the five rows indicate the subprocesses involved with it. The 5 × 5 fouling matrix has been updated over time to reflect the current level of understanding by Bohnet (1987) (Figure 2.1(b)) and, more recently, by experts in various aspects of the field (Figure 2.1(c)) participating in the “Heat Exchanger Fouling and Cleaning VIII” conference held in Austria in June 2009 (Müller-Steinhagen et al., 2009b). Over the years, many gaps in research seem to be filling. In particular transport, which was already identified as the most studied subprocess in the original Epstein matrix, has gained a remarkable level of attention across all fouling mechanisms types. At the other end of the scale, ageing was identified as the topic having received the least attention.

Figure 2.1 Evolution over time of the 5 × 5 fouling matrix summarizing the current level of understanding of fouling mechanisms and processes: (a) original 5 × 5 fouling matrix Epstein (1983); (b) interpretation by Bohnet (1987) and (c) by researchers attending the 2009 Heat Exchanger Fouling and Cleaning conference in Austria (Müller-Steinhagen et al. 2009b). Light to dark shading indicates increase in level of understanding (Coletti, 2010).

There are several reasons why the fouling problem has not been solved so far. There is disagreement between the researchers on the mechanisms involved in producing the deposits, on what the foulant precursors are, on the location of the reactions taking place (whether these are in the bulk of the fluid or at the tube surface) and, quite disappointingly, even the chemical structure of the species that constitute the fouling deposits is disputed. However, there is common ground on the fact that mitigation of fouling in industrial heat exchangers can only be achieved if a better understanding of the underlying science is achieved.

A number of review articles exist on fouling of heat transfer surfaces (De Deus, 1980; O'Callaghan, 1980; Collier, 1980; Epstein, 1981b; Somerscales, 1981; Bohnet, 1987; Bott, 1988b; Watkinson, 1988; Somerscales, 1990; Knudsen, 2002) and more specifically on crude oil fouling formation (Crittenden (1987a,b); Watkinson (1988, 1992); Crittenden et al. (1997); Panchal (2001); Watkinson and Wilson (1997).

This chapter aims at providing the reader with the basic concepts in fouling science, starting from the possible fouling mechanisms (Section 2.1)—with particular focus on chemical reaction fouling (Section 2.2)—and the subprocesses involved in fouling (Section 2.3). Then, a description of the variables affecting fouling and how these interact with each other is provided (Section 2.4) before a final conclusions (Section 2.5).

2.1. Fouling Mechanisms

The columns in the Epstein matrix (Figure 2.1) represent the five major categories into which fouling has been divided according to the mechanism that produces it:

1. Crystallization fouling — precipitation and deposition of dissolved salts that, under process conditions, become supersaturated (Hasson, 1981; Bott, 1988a; Yiantsios et al., 1995).

2. Particulate fouling — deposition of suspended particles on heat transfer surfaces (Gudmundsson, 1981; Epstein, 1988; Karabelas et al., 1997).

3. Chemical reaction fouling — deposition resulting from one or more chemical reactions between reactants contained in a flowing fluid creating accumulation on the heat transfer surface (Watkinson, 1988; Crittenden et al., 1997; Watkinson and Wilson, 1997).

4. Corrosion fouling — deposition produced by a chemical reaction that involves a reactant and the metal surface. The increased roughness of the surface may also promote fouling due to other mechanisms (Lister, 1981; Somerscales, 1997, 1999).

5. Biological fouling — formation of organic films consisting of microorganisms that promote attachment of macroorganisms such as mussels, algae, etc. (Kent, 1988; Melo, 1997).

Crude oil fouling in refinery preheat trains can happen via any of the above mechanisms, except biofouling. Crystallization may occur when salts are deposited from the aqueous fraction of crude oil upstream of the desalter. For example, Joshi (2001) showed the case of one heat exchanger processing crude at about 88 °C where salt deposition was the suspected cause of fouling. Particulate fouling may occur when dirt, clay, or rust suspended in the oil are transported to the heat transfer surface where they may become attached. Chemical reactions between the different species contained in the crude may be triggered and follow a variety of reaction pathways depending on crude composition and process conditions:

• Sulfur compounds in the oil can react with metal surfaces leading to corrosion fouling.

• Oxygen present in concentrations as low as a few ppm (Watkinson, 2005) in storage tanks might form insoluble gums primarily by the autoxidation mechanism.

• Polymerization of monomeric compounds can happen when cracked feedstock have been added to the crude.

• Components in the oil might undergo thermal decomposition (cracking) at high temperatures to ultimately form coke.

Another important mechanism in crude oil fouling is asphaltene precipitation. In this case, changes in composition, temperature, and pressure might cause asphaltenes or resins present in the oil to precipitate out. Although this is a physical phenomenon, it is often a series of chemical reactions that lead to the formation of the precursors responsible for asphaltene deposition.

Establishing which is the overarching mechanism is paramount in order to take remedial action at both design and operation stages. However, whereas for some fluids that undergo chemical reaction fouling (e.g., milk fouling, Beuf et al., 2003; Bansal and Chen, 2006) the species involved, reaction pathways, and underlying kinetics are well defined, crude oil has a complex and diverse composition that makes it extremely challenging to determine exact information for all these aspects. In fact, even if the species involved can be identified via chemical and physical characterization, it is extremely difficult to determine their exact role in the deposition process. Very often, analysis of the deposits only reveals the product of several reactions which prevents the possibility of establishing a link between precursors and deposits, thereby making it virtually impossible to identify the underlying fouling mechanisms. Moreover, to complicate even further the task of identifying the root causes of fouling, it is most likely a combination of mechanisms (mixed fouling) that is actually responsible for deposition (Crittenden et al., 1992). So far, it has only been possible to interpret experimental and field data to find indications of potential causes of fouling without finding conclusive theories on the actual deposition mechanisms. As a result, only conjectures on possible reaction mechanisms exist for crude oil and hydrocarbon fouling (Panchal and Watkinson, 1994; Watkinson and Wilson, 1997).

Industrial experience (see Section 3.1) shows that two mechanisms for organic fouling are more likely to occur in refinery preheat trains:

1. A combination of corrosion and chemical reaction fouling via the autoxidation mechanism.

2. Precipitation of asphaltenes that could adsorb or otherwise adhere to the heat transfer surface.

Although it is possible that operating conditions might be consistent with the latter mechanism, there is also evidence to suggest that the first mechanism is present. In this case, iron sulfide corrosion may have an important role to play in the induction period (see Section 2.3.1) and even in later steps of crude oil fouling. Hydrogen sulfide formed by the breakage of C–S bonds is believed to react with iron containing surfaces to form iron sulfide and this type of corrosion is known to promote fouling by increasing the surface area and decreasing the interfacial tension for stronger adhesion to insoluble asphaltenes (ESDU, 2000). Wang and Watkinson (2011) suggested that iron sulfide particles could be formed in suspension by the reaction of H₂S or organic sulfides with iron oxides formed elsewhere in the process leading to particulate fouling. Their alternate view was that iron sulfide first forms on the heat transfer surface itself. Iron sulfide is black and, without chemical analysis, could be mistaken for carbonaceous material. Whichever mechanism is operative, Joshi et al. (2009) reported that fouling deposits from crude oil exchangers often contain a substantial (35%+) amount of iron sulfide. They also reported that the rate of fouling can be reduced by changing the tube metallurgy. These observations prompt the suggestion that corrosion leads to a roughening of the heat transfer surface and that this roughening is then responsible for promoting the accumulation of fouling precursors. In confirmation, a number of studies have shown that significant amounts of both iron and sulfur are present in crude oil deposits (Wang and Watkinson, 2011; Young et al., 2011).

Details of the possible routes of crude oil fouling formation are presented below.

2.2. Routes to Crude Oil Fouling Formation

2.2.1. Corrosion

The general electrochemical mechanism of corrosion involves the removal of a metal ion from an anionic site leaving behind excess electrons on a metal. For example, with iron:

$F e_{metal} \to F e_{soln}^{2 +} + 2 e_{metal}$ $F e_{metal} \to F e_{soln}^{2 +} + 2 e_{metal}$

Electrons are simultaneously consumed in a balancing reaction at adjacent cathodic sites. For conventional oxidative corrosion the balancing reaction is usually the dissolution of dissolved oxygen and thence the formation of metal oxides or hydroxides:

$O_{2} + 2 H_{2} O + 4 e \to 4 O H^{-}$ $O_{2} + 2 H_{2} O + 4 e \to 4 O H^{-}$

For heat exchangers left outside in damp environments a certain amount of corrosion would occur (Figure 2.2) on metal surfaces that almost certainly would provide sites for other types of fouling to occur once installed in the process.

Wang and Watkinson (2011) studied the mixed organic/inorganic nature of deposits, noting that the iron and sulfur contents in crude oil fouling deposits from three studies ranged from 21% to 49% and 12–22%, respectively, with Fe/S ratios ranging from 0.5 to 1.4. Without doubt, iron sulfide corrosion was occurring. It is considered that the formation of iron sulfide could occur in parallel or in series with the formation of organic deposits. Sulfur is present in all crude oils at various percentage levels in the form of mercaptans, sulfides, disulfides, etc. Iron is present at ppm levels in crude oils but Wang and Watkinson (2011) suggested that corrosion upstream can provide additional dissolved iron sources such as napthenates. In the absence of oxygen, the formation of iron sulfide on a metallic heat transfer surface can proceed via (1) corrosion of iron by organic acids upstream, (2) thermal decomposition of acid salts, and (3) reaction of iron oxide with organic sulfur or hydrogen sulfide:

Figure 2.2 Refinery shell and tube heat exchanger tube sheet left outside after cleaning. (copyright B. D. Crittenden).

$2 RC H_{2} COOH + Fe \to {(RC H_{2} COO)}_{2} Fe + H_{2}$ $2 RC H_{2} COOH + Fe \to {(RC H_{2} COO)}_{2} Fe + H_{2}$

$2 RC H_{2} COOH + F e^{2 +} \to {(RC H_{2} COO)}_{2} Fe + 2 H^{+}$ $2 RC H_{2} COOH + F e^{2 +} \to {(RC H_{2} COO)}_{2} Fe + 2 H^{+}$

${(RC H_{2} COO)}_{2} Fe \to RC H_{2} COC H_{2} R + FeO + C O_{2}$ ${(RC H_{2} COO)}_{2} Fe \to RC H_{2} COC H_{2} R + FeO + C O_{2}$

$FeO + R^{'} –S– R^{″} \to FeS + R^{'} –O– R^{″}$ $FeO + R^{'} –S– R^{″} \to FeS + R^{'} –O– R^{″}$

$FeO + H_{2} S \to FeS + H_{2} O$ $FeO + H_{2} S \to FeS + H_{2} O$

The last four reactions (where R′–S–R″ represents an organic sulfide) can create FeS in the fluid bulk such that it can deposit elsewhere in a heat exchanger.

2.2.2. Autoxidation

Autoxidation is a free radical chain reaction mechanism that can operate in oils at moderately low temperatures (Crittenden, 1988a). Like all oxidation reactions, autoxidation is exothermic. Oxygen is required and is likely to be present dissolved at low concentrations in crude oils in storage. The first step is the abstraction of hydrogen from an organic molecule R–H by a free radical X

$R - H + X^{\cdot} \to R^{\cdot} + XH$ $R - H + X^{\cdot} \to R^{\cdot} + XH$

The chain reaction involves the reaction of molecular oxygen with the substrate radical R

and further hydrogen abstraction creating peroxy radicals and hydroperoxide molecules:

$R^{\cdot} + O_{2} \to RO O^{\cdot}$ $R^{\cdot} + O_{2} \to RO O^{\cdot}$

$RO O^{\cdot} + RH \to ROOH + R^{\cdot}$ $RO O^{\cdot} + RH \to ROOH + R^{\cdot}$

Once the autoxidation process has been initiated, homolysis of the weak O–O bond in the hydroperoxide leads to the formation of more free radicals that can take part in hydrogen abstraction. The reaction thereby breeds free radicals.

$ROOH \to R O^{\cdot} + O H^{\cdot}$ $ROOH \to R O^{\cdot} + O H^{\cdot}$

With a good supply of oxygen the concentration of substrate molecules would be low. Hence, of the following three possible free radical termination reactions, the last one is probably the most important:

$R^{\cdot} + R^{\cdot} \to R–R$ $R^{\cdot} + R^{\cdot} \to R–R$

$R^{\cdot} + RO O^{\cdot} \to ROOR$ $R^{\cdot} + RO O^{\cdot} \to ROOR$

$RO O^{\cdot} + RO O^{\cdot} \to ROOR + O_{2}$ $RO O^{\cdot} + RO O^{\cdot} \to ROOR + O_{2}$

Trace amounts of soluble metal salts, particularly of cobalt, manganese, iron, copper, chromium, lead, and nickel, can catalyze the rate of hydroperoxide homolysis and may also take part in initiation reactions:

$ROOH + C O^{2 +} \to R O^{\cdot} + O H^{-} + C O^{3 +}$ $ROOH + C O^{2 +} \to R O^{\cdot} + O H^{-} + C O^{3 +}$

$ROOH + C O^{3 +} \to RO O^{\cdot} + H^{+} + C O^{2 +}$ $ROOH + C O^{3 +} \to RO O^{\cdot} + H^{+} + C O^{2 +}$

$RH + C O^{3 +} \to R^{\cdot} + H^{+} + C O^{2 +}$ $RH + C O^{3 +} \to R^{\cdot} + H^{+} + C O^{2 +}$

Oxidation of aromatics and primary C–H bonds in alkanes is relatively difficult but becomes easier with secondary and tertiary C–H bonds. Thus, the presence of aromatics or naphthenic species can inhibit the deposition rate from straight chain alkanes whereas the rate is considerably higher if there is a C

C bond or an aromatic nucleus α to the group undergoing oxidation (Crittenden, 1988b). Hence the rate of autoxidation can be high for cracked feedstocks containing olefinic species.

As will be seen in Section 4.1, this an important route generally assumed to be responsible for chronic fouling affecting the long-term performance of refinery preheat trains.

2.2.3. Polymerization

Polymerization mechanisms involve the conversion of individual compounds (monomers) into a large multiple of themselves. Copolymerization can occur when two or more such monomers are involved in the creation of a high molar mass compound. Polymerization can be expected if there are olefinic compounds in the feedstock. Hence, polymerization as a fouling mechanism is probably limited to crude oils to which cracked feedstocks have been added.

Polymers can be created by a wide variety of processing techniques that include bulk, solution, suspension, emulsion, and precipitation methods (Crittenden, 1988a). The general mechanism of polymerization is broadly similar to that of autoxidation and involves radical formation and chain initiation, reaction chain propagation, and chain termination. The rate of each step and the length of the polymer chains depend on a number of factors including the concentration and activity of the reactants. Similar to autoxidation, species such as oxygen, halides, sulfides, nitrogen compounds, certain metals, and metallic compounds are all known to be capable of initiating polymerization. Under certain conditions, mercaptans and sulfides can take part in corrosion reactions with the heat exchanger surface to produce free radicals that could initiate polymerization reactions, for example,

$Fe + R–SH \to FeSH + R^{\cdot}$ $Fe + R–SH \to FeSH + R^{\cdot}$

If the polymer being formed becomes insoluble in its solution then it will precipitate out. This would occur wherever processing conditions were favorable, for example, in the bulk fluid, in the heat transfer film, or at the heat transfer surface.

2.2.4. Thermal Cracking

Thermal cracking of hydrocarbons tends to be restricted to temperatures in excess of 650 °K and is therefore more likely to arise in the crude distillation furnace rather than in the preheat exchangers. Even so, cracking and its associated mechanism of coking could occur in deposits that spend long durations on hot heat transfer surfaces in which temperatures increase as fouling proceeds.

In this mechanism, higher molar mass hydrocarbons are broken down into lighter alkanes and alkenes, the latter being able to take part in both autoxidation and polymerization mechanisms. Although a vast number of individual reactions can occur in the thermal cracking of complex hydrocarbon mixtures, the overall mechanism comprises the steps of initiation, hydrogen abstraction, radical decomposition, radical addition, and termination. Initiation involves a single molecule breaking into two free radicals at a C–C bond:

$R - R^{'} \to R^{\cdot} + R^{' \cdot}$ $R - R^{'} \to R^{\cdot} + R^{' \cdot}$

Only a small fraction of the feed might participate in the initiation step. Each free radical may then abstract hydrogen from a hydrocarbon molecule turning it into a free radical:

$R^{\cdot} + R^{″} R^{″'} \to RH + R^{″} R^{″' \cdot}$ $R^{\cdot} + R^{″} R^{″'} \to RH + R^{″} R^{″' \cdot}$

The more complex free radical can then decompose into a hydrogen free radical and an alkene:

$R^{″} R^{″' \cdot} \to R^{″} = R^{″'} + H^{\cdot}$ $R^{″} R^{″' \cdot} \to R^{″} = R^{″'} + H^{\cdot}$

Alkenes can react with free radicals to form larger free radicals, which can cause aromatic products to be formed when heavier feedstocks are involved:

$R^{″} R^{″' \cdot} + R^{″ ″} = R^{″ ″'} \to R^{″} R^{″'} R^{″ ″} R^{″ ″' \cdot}$ $R^{″} R^{″' \cdot} + R^{″ ″} = R^{″ ″'} \to R^{″} R^{″'} R^{″ ″} R^{″ ″' \cdot}$

Two common types of termination reaction are possible, namely recombination in which two free radicals combine to create a larger molecule and recombination in which one free radical transfers a hydrogen atom to the other one, yielding an alkene and an alkane. For example,

$R^{\cdot} + R^{″} R^{″' \cdot} \to R R^{″} R^{″'}$ $R^{\cdot} + R^{″} R^{″' \cdot} \to R R^{″} R^{″'}$

$R^{″} R^{″' \cdot} + R^{″} R^{″' \cdot} \to R^{″} = R^{″'} + R^{″} R^{″'}$ $R^{″} R^{″' \cdot} + R^{″} R^{″' \cdot} \to R^{″} = R^{″'} + R^{″} R^{″'}$

Thermal cracking is endothermic (i.e., ΔH is positive) but is favored thermodynamically (i.e., the free energy change ΔG is negative) because although the bond dissociation energy for a C–C bond is quite high the large and positive entropy change ΔS that results from the breakage of large molecules into several smaller ones means that the TΔS term in the Gibbs free energy equation is larger than the ΔH term:

$Δ G = Δ H - T Δ S$ $Δ G = Δ H - T Δ S$

(2.1)

This explains why thermal cracking occurs at higher temperatures. At high temperature, coke can be formed via the secondary or synthesis reactions of the products of the primary or cracking and dehydrogenation reactions. Figure 2.3 (Crittenden et al., 1988a, adapted from Fitzer et al. (1971)) summarizes the primary and secondary reactions. In the context of crude oil fouling, coke formation is noted again under ageing in Section 2.3.5.

2.2.5. Asphaltene Precipitation

Several authors (Wiehe, 2001a; Mason and Lin, 2003; Stark and Asomaning, 2003; Saleh et al., 2005a) indicated that asphaltene flocculation and deposition caused by asphaltene/oil incompatibility are responsible for crude oil fouling, especially at high temperatures (Eaton and Lux, 1984; Lambourn and Durrieu, 1986; Dickakian, 1989).

Asphaltenes are defined as the n-alkane-insoluble/toluene-soluble fraction of the crude (Watkinson, 2008) and form one of the four key classes of organic compounds that characterize the polarizability and polarity of crude oils according to the SARA (Saturate, Aromatic, Resin, & Asphaltene) analysis. Although asphaltenes have been studied in a number of works (Dickakian and Seay, 1988; Buenrostro-Gonzalez et al., 2004; Aguilera-Mercado et al., 2006), there is ongoing debate among the scientific community on their exact structure (Durand et al., 2010) and on whether an archipelago (Karimi et al., 2011; Murgich et al., 1999; Sheremata et al., 2004), continental or an island model (Groenzin and Mullins, 1999; Mullins, 2010, 2011; Sabbah et al., 2012) is more appropriate to describe them. More details on this debate and aggregation mechanisms are reported in Section 5.2.1.

Figure 2.3 General schematic of hydrocarbon pyrolysis (cracking and coking). Crittenden et al. (1988a), adapted from Fitzer et al. (1971).

Different ratios of SARA components produce different effects on the stability of asphaltenes in the crude. This plays a key role when two or more crudes are blended together. For example, both the resin and the asphaltene fractions of crude oils contain polar constituents with the former acting as a dispersant for the latter. If incompatible oils are mixed together, asphaltenes may flocculate and come out of solution resulting in severe fouling events. For this reason, major research efforts have been made to establish the effect of crude compatibility on asphaltene stability (Wiehe and Kennedy, 2000; Gabrienko et al., 2014).

Various ways of characterizing compatibility have been proposed. Asomaning and Watkinson (1997) quantified compatibility via the colloidal instability index (C.I.I.):

$C .I .I . = \frac{Saturates + Asphaltenes}{Aromatics + Resins}$ $C .I .I . = \frac{Saturates + Asphaltenes}{Aromatics + Resins}$

(2.2)

When C.I.I. is less than unity, the combined fractions of resins and aromatics are considered sufficiently high to keep the asphaltene fraction in solution or dispersion. On the other hand, when C.I.I. is greater than one, asphaltenes are expected to precipitate out. For such unstable oils, it has been found that precipitated asphaltenes have compositions that are similar to those of fouling deposits (Watkinson, 2005) and it has been assumed (Watkinson, 2005) that asphaltene fouling is simply a case of particulate deposition complicated by precipitation and agglomeration processes. Although Equation (2.2) provides a useful indication of stability, it has been argued that its use alone is not sufficient to predict asphaltene precipitation over a wide range of compositions. Thus Al-Atar and Watkinson (2001) included also the resin/asphaltene ratio as a parameter influencing fouling.

In view of the complexity and potential importance of asphaltenes in crude oil fouling, attempts have also been made to understand the underlying fundamental mechanisms involved in the phase boundaries of asphaltene–crude oil chemical systems (reviewed in Section 5.2).

Although oil blend compatibility plays an important role in fouling in PHTs, this is not the only cause of asphaltene deposition. Wiehe (2001b) showed that asphaltenes do not require being insoluble to produce fouling, pointing out that crudes exist that are self-incompatible (i.e., they do not need to be blended to produce fouling) and that even compatible oils can undergo asphaltene fouling. Complex chemical reactions may be triggered by the process conditions inside the heat exchangers leading to asphaltene deposition from crude oils. Eaton and Lux (1984) proposed the following series of degradation reactions:

Besides the limitations of analytical techniques (e.g., instrument resolution, limited solubility of the samples), the number and complexity of the reaction and physical processes involved make it very difficult to confirm the above mechanisms and even harder to measure the individual reactions rates and their respective kinetic parameters.

2.2.6. Generalised mechanism

Previous sections have highlighted that several uncertainties are involved with the identification of the mechanism overarching the fouling process. However, from all the experimental evidence reported in the literature, there seems to be agreement on a generalized reaction fouling mechanism which comprises the following three steps:

$Reactant A \overset{r_{1}}{\to} Precursor B \overset{r_{2}}{\to} Foulant C \overset{r_{3}}{\to} Aged deposit D$ $Reactant A \overset{r_{1}}{\to} Precursor B \overset{r_{2}}{\to} Foulant C \overset{r_{3}}{\to} Aged deposit D$

In the first step, soluble precursors are formed via a reaction (or series of reactions) among different species in the crude. In some cases, the formation of the precursors can happen before the crude enters the heat exchanger and Panchal and Watkinson (1994) found indications that these can have a more significant effect than the precursors generated in the exchanger itself. In the second step, the precursors react giving insoluble foulant species that will then deposit on the heat transfer surface. A third step involves chemical and physical changes of the deposits caused by the high wall temperatures. This last process goes under the name of ageing (see Section 2.3.5) and, as will be shown in later chapters, it plays an important role in the overall heat exchanger performance.

According to Panchal and Watkinson (1994), reactions r₁ and r₂ can take place in the bulk, in the boundary layer, or at the wall. Figure 2.4 summarizes the three possible scenarios:

• Case 1: Precursor generation in the bulk. In Case 1a, reaction r₂ occurs on the wall surface whereas in Case 1b, reaction r₂ occurs in the bulk itself.

• Case 2: Precursor generation in the boundary layer. In Case 2a, reaction r₂ occurs on the wall surface whereas in Case 2b, reaction r₂ occurs in the boundary layer itself.

• Case 3: Precursor generation on the wall surface.

Panchal and Watkinson (1994) developed a mathematical model for fouling in a single tube based on the reaction scheme shown in above. They compared model simulations in the different cases with experimental data finding a relatively good match to Case 1a and Case 2, suggesting that fouling may occur via reaction in the bulk or in the boundary layer. However, admittedly, they neglected reaction r₃ (i.e., ageing) which may affect the accuracy of their final conclusions. Indeed, the question of where the reactions take place is still open. Ebert and Panchal (1995) and Srinivasan and Watkinson (2005), for example, suggested that the fouling reaction takes place in the bulk film and not at the wall surface, whereas Crittenden et al. (1992) argued that most of the deposits is generated in the highest temperature region, i.e., at (or near) the surface.

A detailed description of the events in fouling formation is given in the following section.

Figure 2.4 Deposition mechanism where precursor generation is in the bulk (Case 1), in the boundary layer (Case 2), or at the wall surface (Case 3). Adapted from Panchal and Watkinson (1994).

2.3. Events in Crude Oil Fouling

Fouling is a dynamic process. Figure 2.5 shows idealized fouling curves where the fouling resistance is plotted against time, starting from a clean surface. A lag time, t_l, is often seen before the layer starts growing on the initially clean heat transfer surface. Depending on process conditions, fluid composition, and fouling mechanism, the fouling resistance evolves in different ways. In some cases, fouling just keeps growing at a linear rate, in some others, the thickness of the layer keeps increasing but the rate of deposition decreases over time (falling rate). If the conditions are such that the forces at which the foulants are deposited on the surface are balanced by the forces that either suppress deposition or remove deposited species, the observed behavior is defined as asymptotic.

Epstein (1983) considered that each of five generic types of fouling (crystallization, particulate, chemical reaction, corrosion, and biological) comprise five distinguishable steps (reported in the rows of his 5 × 5 matrix in Figure 2.1):

• Initiation (Yang et al., 2009a)

• Transport to surface (Panchal and Watkinson, 1994)

• Attachment (Visser, 1988a,b)

• Removal (Yiantsios and Karabelas, 1994; Bohnet et al., 1997)

• Ageing (Wilson et al., 2009; Coletti et al., 2010).

Although there must be some sequential element to any overall fouling process, in reality many of these steps could, and probably would, operate in parallel. When uncertain reaction mechanisms are taken into account, it is clear that crude oil fouling becomes a very complex physico-chemical process.

When material of increasing molar mass and structural complexity exceeds its local solubility in crude oil then it will form a solid. This could happen at the heat transfer surface or in a reaction zone within the fluid and close to the heat transfer surface where process conditions are favorable enough. Reactants (or fouling precursors) must be transported to the reaction zone by convection mechanisms and reaction products could be convected away perhaps to be deposited, or to take part in other reactions, elsewhere within the heat exchanger. The possibility also exists for the action of fluid shear to remove solid deposits in a wholesale fashion from the heat exchanger surface for them to be deposited elsewhere. Clearly, initiation must precede ageing but, once started, the steps of initiation, transport, attachment, removal, and ageing could in fact all proceed in parallel. For crude oil fouling, transport, attachment, and removal steps have been commonly deemed to occur in parallel. In contrast, the steps of initiation and ageing remain poorly understood. Although chemical reaction does not feature as a distinguishable step in Epstein's analysis, reactions are most likely to occur in the initiation, attachment, and ageing steps for crude oil fouling. It is unlikely that chemical reactions are involved in transport and removal.

Long before Epstein distinguished steps in fouling processes, Nelson (1934a) proposed a simple mechanism to account for fouling of oil refinery heat exchangers in which none of the steps were considered to be discrete and sequential. It was considered that the rate of fouling would be directly dependent on the volume of fluid in a heat transfer film which would be at a temperature higher than that of the bulk fluid. Hence, the volume of hydrocarbon likely to react would be reduced at higher velocity. In turn, this would mirror the common observation that fouling in oil refinery heat exchangers could be mitigated by using higher velocities. Furthermore, a higher velocity would not only reduce the residence time of the thinner film again leading to less fouling but also if there were a deposit removal mechanism taking place, then the rate of removal almost certainly would be increased by a higher velocity because the shear stress on the heat transfer surface would be higher. Nelson's idea clearly integrated the transport, attachment, and removal steps. Nevertheless, Nelson's simple concept dealt neither with initiation nor with ageing.

A brief discussion for each of these subprocesses is provided in the following sections.

2.3.1. Initiation

When a new or clean heat transfer surface is commissioned there might be little change in the overall heat transfer coefficient, U, for some time, creating the impression, almost certainly a false one, that nothing is happening. Lasting anything up to several days, the length of the initiation period is related to the nature of the surface (see Section 2.4.4) as well as to process conditions such as surface temperature and fluid velocity. In some cases initiation might actually increase rather than decrease the overall heat transfer coefficient, U due to the creation of roughness on the surface (Crittenden and Alderman, 1988). Such roughness creation might come from deposition of chemical elements within the crude oil and/or from corrosion-type reactions on the surface. In reality, very little is known about the initiation, or induction, of crude oil fouling and indeed there is often difficulty in even interpreting when the initiation step has started and ended. Saleh et al. (2003) arbitrarily used the time at which the change in reciprocal overall heat transfer coefficient (1/U) was 5% or more of the total change in 1/U to be the end of the induction period.

Yang et al. (2012) adopted a more generic and arguably a more scientific approach in which it was assumed that active foulant species stick to the surface and gradually cover it, the rate of change in surface coverage dθ/dt being proportional to the fractional free surface area (1 – θ). It was further assumed that the foulant already on the surface could act as a seed, thereby attracting more foulant such that the growth rate was first order in θ with a rate constant equal to k₁. In addition, by adopting a removal mechanism akin to that in adsorption phenomena, the removal rate of the coverage was assumed to be proportional to θ with a rate constant equal to k₂. A combination of these three assumptions led to the following relationship:

$\frac{d θ}{d t} = k_{1} θ (1 - θ) - k_{2} θ$ $\frac{d θ}{d t} = k_{1} θ (1 - θ) - k_{2} θ$

(2.3)

According to Equation 2.3, the fouling layer grows on this covered surface at a rate equal to θdR_f/dt, where R_f is the fouling resistance and dR_f/dt can be any established expression that accounts for the normal period of fouling of the heat exchanger. It was found that the surface coverage in the induction period θ was related to the rate constants and time t by the following equation:

$θ = \frac{k_{1} - k_{2}}{k_{1}} \frac{1}{1 + c e^{- (k_{1} - k_{2}) t}}$ $θ = \frac{k_{1} - k_{2}}{k_{1}} \frac{1}{1 + c e^{- (k_{1} - k_{2}) t}}$

(2.4)

Here, c is a constant of integration. Equation (2.4) produces a smooth sigmoidal curve ranging from θ = 0 at the beginning of the induction period to θ = 1 at its end. A time t_0.5 is defined to be the time when θ is equal to half its maximum value.

$t_{0.5} = \frac{\ln c}{k_{1} - k_{2}}$ $t_{0.5} = \frac{\ln c}{k_{1} - k_{2}}$

(2.5)

Yang et al. (2012) used data obtained during induction periods for quite different types of fouling including crude oil, whey protein, and calcium salt scaling. The induction period model supported experimental observations in which shorter induction periods are found with higher surface temperatures. The activation energy E for the induction period of the crude oil study was found to be 78.2 kJ mol⁻¹; this value is important in determining what effect the surface temperature might have on the length of the induction period. Yang et al. (2012) were able to show also that the length of the induction period was much reduced when the heat transfer surface was not clean initially. They claimed that fouling threshold phenomena could be captured by their interpretation of the induction period. Thus, if k₂ were to remain close to the value of k₁ then the surface coverage θ would remain low and the induction period would, accordingly, be very long. Hence, a boundary could be drawn between a fouling zone in which k₁ > k₂ and a nonfouling zone in which k₁ < k₂. What this theoretical interpretation of the induction period does not do is identify what physical and/or chemical processes are occurring at the heat transfer surface in the very early stages of crude oil fouling.

2.3.2. Transport

In all reaction mechanisms, the fouling precursors (foulants) must be transported by convective mechanisms from the bulk of the fluid toward the zone in which they are converted into deposits by attachment or adhesion to the heat transfer surface. In a similar manner, the foulant or deposit, if mobile enough, could be transported back toward the fluid bulk, perhaps to take part in further reactions elsewhere in the equipment. The transport of foulants and their precursors toward the heat transfer surface is traditionally based on diffusion and the film theory of convective mass transfer and the mechanisms would be different for suspended particulates and dissolved precursors. The transport of precursors is governed by the difference between their bulk concentration c_b and their concentration at the heat transfer surface c_s. The lumped parameter k_m is the mass transfer coefficient. Strictly, this equation should describe the mass (or molar) flux of foulant precursors:

$\frac{d m}{d t} = k_{m} (c_{b} - c_{s})$ $\frac{d m}{d t} = k_{m} (c_{b} - c_{s})$

(2.6)

The mass transfer coefficient k_m can be obtained from any appropriate correlation that relates the Sherwood number to the Reynolds and Schmidt numbers. If the deposit properties (density and thermal conductivity) were to remain independent of time then the rate of fouling would be proportional to the rate of mass deposition and hence the rate of fouling would be given by:

$\frac{d R_{f}}{d t} = k^{'} (c_{b} - c_{s})$ $\frac{d R_{f}}{d t} = k^{'} (c_{b} - c_{s})$

(2.7)

Here, k′ would be a coefficient that incorporates the mass transfer coefficient, k_m as well as the density and thermal conductivity of the deposit.

For particles already present or formed in the fluid bulk, fouling can be considered to comprise two distinct steps, namely transport to the wall region followed by adhesion or attachment to the wall (Watkinson, 2005). If the transport step controls the overall rate of deposition then the fouling rate should only be a weak function of surface and bulk temperature. The fouling rate should increase with increasing velocity, linearly in turbulent flow and more weakly in laminar flow (Watkinson, 2005). Epstein (1981) has summarized the mechanisms that could be involved in the transport of particles toward a surface (particulate fouling). Settling is unlikely to be relevant to crude oil fouling and so the principal reason as to whether a particle can get to a surface and remain there is related to the surface forces such as the London–van der Waals forces, which are always attractive. Electrical double-layer forces are attractive if the particle and surface have zeta potentials of opposite sign. A hydrostatic viscous interaction force may also be operative in which the fluid friction on the particle increases significantly as it moves normal to the plane of the surface. When interaction barriers are additive then there is a large energy barrier which must be overcome, which is equivalent to the activation energy of a chemical reaction. It is for this reason that the deposition flux is commonly assumed to comprise two resistive components:

${\dot{m}}_{d} = \frac{c_{b}}{(1 / k_{m}) + (1 / k_{r})}$ ${\dot{m}}_{d} = \frac{c_{b}}{(1 / k_{m}) + (1 / k_{r})}$

(2.8)

Here k_r is a rate constant applied to the energy barrier and commonly assumed to apply to a first-order process. The rate constant is expected to be a strong function of temperature; the relationship is commonly expressed in the form of an Arrhenius-type equation:

$k_{r} = k_{r \infty} e^{- E / R T_{w}}$ $k_{r} = k_{r \infty} e^{- E / R T_{w}}$

(2.9)

Where k_{r ∞} is the pre-exponential factor, T_w is the wall temperature and R the universal gas constant.

The mass transfer coefficient again can be obtained from correlations involving the Reynolds and Schmidt numbers. The particle diffusivity, which is required to calculate the Schmidt number, can be obtained from the Stokes–Einstein equation (Epstein, 1981). Clearly, the interactions change when deposits have already been established on the heat transfer surface. Although electrophoresis has been suggested as an important mechanism for particulate deposition (Epstein, 1981), it is unlikely to be operative in crude oil systems unless there is a significant aqueous fraction. Thermophoresis, a mechanism that can exert a force on a submicrometer particle due to a temperature change across it, is also unlikely to be effective in crude oil fouling. Because it is close to a hot surface when crude oil is being heated, the effect would be to repel the particle.

2.3.3. Attachment

Once transported, foulants either attach or adhere to the heat transfer surface or they leave the surface such that they may be deposited elsewhere. Factors affecting adhesion include surface energies, surface temperature, and shear forces acting at the surface, as well as the nature and compositions of previously deposited layers.

Little is known about the fundamental details of the adhesion of deposits from crude oils on heat transfer surfaces. Simple theories have been proposed to describe this phenomenon (Bennett, 2012) but they are difficult to validate given the challenges associated with collecting the relevant experimental data. While it is the complexity of crude oils which makes this an uncertain subject, some lessons can be drawn from fundamental studies of the other major type of reaction fouling, namely that from food processing. Here, the basic fundamentals of adhesion and the associated subject of removal have been provided by Visser (1988a,b).

Should the adhesion step control the overall fouling rate if it is a particulate-based mechanism then the overall fouling rate should increase strongly with surface temperature because the adhesion step would be sensitive to temperature. The overall fouling rate would not be strongly dependent on bulk temperature but should decrease with increase in velocity because less time would be available for adhesion to occur (Watkinson, 2005).

2.3.4. Removal

Removal mechanisms are poorly understood in comparison with deposition. Models of crude oil fouling commonly assume that the overall or net rate of deposition is the difference between a rate of deposition and a rate of removal. However, such a distinctive approach is almost certainly not the way crude oil fouling proceeds. Rather, it is almost certainly the case that many mechanisms including those responsible for deposition and those responsible for removal will operate together in a continuous manner. Epstein (1981) prefers to use the term “reentrainment” rather than “removal.” One reason for this is that on a clean heat transfer surface there cannot, in principle, be any removal since there would be no deposit to remove. Nonetheless, once deposits have started to form on the heat transfer surface the possibility then exists for them to be removed by the surface shear stress in turbulent flow or by diffusion should they be soluble in the chemical bulk. The latter mechanism becomes a possibility when deposition is the result of chemical reactions because the concentration of reaction products in the bulk fluid would be low.

Epstein (1981) summarized the reentrainment mechanisms that previous researchers had used in mathematical models of fouling processes. They include spalling, erosion, bond fracture, and dissolution. Each, in principle, could be considered applicable to crude oil fouling and may, of course, operate in parallel. Generally, models of removal have the rate of mass removal being directly proportional to the mass of deposit on the surface.

2.3.5. Ageing

The exposure of the fouling layer to high wall temperatures over extended periods can trigger chemical and physical transformations which alter the structure and properties of the deposits. Nelson (1934a,b) described initial deposition of crude oil fouling as involving a gel which changes its structure over time to a harder material similar to coke. These transformations not only alter the rheology of the deposit layer (Sileri et al., 2009) but also its thermal conductivity, and thus:

• The overall thermo-hydraulic behavior of the exchanger.

• The nature of the material recovered for analysis with analytical techniques. This makes it difficult to identify the material originally deposited (Wilson et al., 2009).

• The ease of deposit removal (Wilson, 2005). Ageing therefore has important implications for the development of cleaning strategies.

Within a fouling layer, the material is subject to a range of temperature histories so that ageing is non-uniform in space and highly time dependent. Timescales that enable appreciation of the effects of ageing are rather long and laboratory experiments rarely report deposit ageing (Wilson et al., 2009). Unlike the induction period, which happens at much shorter timescales, ageing plays a more important role in industrial equipment than in laboratory experiments. However, despite the importance of these effects, relatively little attention has been paid in the literature to these phenomena, particularly in the area of chemical reaction fouling (Figure 2.1).

Nelson (1934a) was perhaps the first to report on ageing through a semiquantitative model of crude oil furnaces. Atkins (1962) subsequently was the first to propose that a hydrocarbon fouling layer (for a fired heater) might comprise two layers: a porous, tarry layer and a hard crust layer. From this concept, Crittenden and Kolaczkowski (1979b) developed a mathematical model of fouling for two such layers whence the subject of ageing has been largely ignored, perhaps because it is so hard to study experimentally.

More recently, Ishiyama et al. (2010) developed a lumped parameter, first-order kinetics ageing model that is based on a series of layers laid down at fixed time steps to form a series of annuli with varying history. Once deposited, each layer is covered by the next layer with deposition occurring only at the deposit–oil interface. Although it was considered that the thickness of each layer would not change with time, it was assumed that a change in thermal conductivity of a layer would constitute the ageing process. Although the approach adopted is still idealistic, Ishiyama et al. were able to make exploratory studies of the outcomes from various scenarios including deposition with no ageing and deposition with ageing for operation with either constant wall temperature or constant heat flux. The model was then extended by Coletti et al. (2010) to account for the effect of ageing in distributed systems as summarized in Section 5.1. No experimental studies are known to validate these various ageing models.

Ageing in crude oil deposits almost certainly follows a coking mechanism (Section 2.2.4). According to Fitzer et al. (1971), the secondary reactions in a cracking and coking mechanism (Figure 2.3) involve cyclization of hydrocarbon chains, thereby forming aromatics and subsequently condensation of the aromatics to form polynuclear aromatic structures of high molar mass. Although the formation of aromatics tends to occur preferentially at temperatures above 950 °K, aromatic structures will already be present in a fouling deposit that has been created through an asphaltene deposition mechanism. Once present in a deposit, chemical condensation of aromatics can take place in the liquid phase at temperatures in the range 650–800 °K, the lower end of which will be experienced in deposits formed in the crude oil preheat exchanger train. At temperatures below 650 °K, polynuclear aromatics can cross-link to form higher molar mass compounds via both Diels–Alder and free radical pyrocondensation mechanisms (Fitzer et al., 1971).

2.4. Variables Affecting Fouling

Laboratory experiments have been performed in microbomb reactors (MBRs) (Section 3.1), stirred cells (Section 3.2), and recirculating closed loop apparatuses in laboratory equipment (Section 3.3) or in field units (Kuru et al., 1997) to assess the effects that different variables have on fouling. The key variables identified with these experiments are:

1. Crude oil composition and inorganic contaminants

2. Bulk and wall temperatures

3. Velocity (shear stress)

4. Surface conditions (e.g., roughness and roughness dynamics).

Notably, pressure is missing from the list above. Although it can have a significant effect in asphaltene precipitation during recovery of heavy crude oils (Asomaning et al., 2000), it is not known whether pressure could have an influence under typical refinery operating conditions (i.e., below 30–35 bar).

Through the experimental results available in literature, it can be established that mass transport in crude oil fouling increases weakly with both wall and bulk temperature and increases nearly linearly with velocity. Attachment, on the other hand, has been shown to increase strongly with wall temperature but only weakly with bulk temperature and decreases with increased velocity (Watkinson, 2008). The fact that the fouling rate decreases with increasing velocity suggests that crude fouling is controlled by adhesion or reaction.

Figure 2.6 Schematic of phenomena occurring in a tubular heat exchanger undergoing crude oil fouling. Boxes on the top of the figure show the phenomena whose increase reduces fouling rates and have a positive impact on the measured quantities: pressure drop and coil inlet temperature. Boxes at the bottom of the figure show phenomena that should decrease in order to mitigate fouling. Arrows indicate relations between phenomena (square boxes) and measured variables (circles). Reprinted with permission from Coletti and Macchietto. A dynamic, distributed model of shell-and-tube heat exchangers undergoing crude oil fouling, 2011. Ind. Eng. Chem. Res. 50(8): 4515-4533. Copyright (2011) American Chemical Society.

Figure 2.6 summarizes the interactions of the above variables involved in the tube-side crude oil fouling process and their combined effect on the two main measured quantities: pressure drop and outlet temperature.

The overall effect can be significant: deposition of foulants increases the resistance to heat transfer, but it also affects the fluid dynamics of the system by reducing the cross-sectional area which in turn increases the velocity, hence the convective heat transfer coefficient. The variables affecting fouling will be explored more in detail in the next sections.

2.4.1. Crude Oil Composition and Inorganic Contaminants

Crittenden (1988b) reviewed the extensive literature on how the presence of trace species can affect the deterioration of feedstocks when autoxidation is the dominant mechanism.

2.4.1.1. Oxygen

Deterioration and gum formation are not just dependent on dissolved oxygen concentration but also on the presence of chemically bound oxygen, nitrogen, and sulfur, as well as the presence of trace metal contaminants. Rigorous exclusion of dissolved oxygen (or air) can substantially reduce—or eliminate—deposition from hydrocarbons in the liquid phase, although the extent depends on the type and concentration of sulfur species present. As might be expected, the addition of organic peroxides to deoxygenated feedstocks has been found to be highly deleterious.

2.4.1.2. Sulfur

The importance and role of sulfur in crude oil fouling have been highlighted in Section 2.2.1 when discussing fouling mechanisms.

The effect of sulfur is complex (Crittenden, 1988b). Thiols, sulfides, disulfides, and some condensed thiophenes which can break down to form free radicals can be problematic but diphenyl sulfide and dibenzothiophene are not. Free sulfur, disulfides, polysulfides, and thiophenol promote sludge formation in storage tanks whereas thiophenes, aliphatic mercaptans, and aliphatic sulfides show little effect. The organo-sulfur structure is important as well with the thermal decomposition of alkanes being promoted by sulfur atoms in branched structured compounds. The presence of small amounts of nitrogen-containing compounds such as pyridines and pyrroles can be deleterious to hydrocarbon stability (Crittenden, 1988b).

2.4.1.3. Metals

Metal chlorides of iron, calcium, and magnesium hydrolyze to produce acids that promote hydrocarbon fouling substantially while the nature of metal surfaces themselves can strongly influence deposition rates with copper and vanadium being highly deleterious (Crittenden, 1988b). When present in the form of acetyl-acetonates, low levels of metals can dramatically increase deposition rates.

2.4.1.4. Asphaltenes

As reported in Section 2.2.5, Asphaltenes are defined as the n-alkane-insoluble/toluene-soluble fraction of the crude (Watkinson, 2008) and their role in crude oil fouling has been studied by several authors.

Using their patented batch stirred cell (Section 3.2), Eaton and Lux (1984) found that the addition of asphaltenes with molar mass in the range 1000 to 300,000 indeed promoted crude oil fouling whereas the addition of resins (molar mass <5000) had less serious effects. Both the concentration and the solubility of asphaltenes were found to be important factors. On the other hand, Venditti et al. (2009a) reported on preliminary experiments using a MBR, described in Section 3.1, that showed the role of asphaltenes to be somewhat ambiguous in deposit formation from the bulk at temperatures in the range 280–390 °C. There was evidence for the formation of larger molecules when the oil was heated even when no solids were observed at lower temperatures and, accordingly, there was evidence of chemical transformations. The formation of deposits in the MBR was favored by high temperatures and long residence times but, irrespective of the temperature, no deposits were found at low residence times (1 h). There was also evidence that carbonaceous deposits originated not only from the asphaltene fraction of crude oil but also from the crude after it had been deasphalted. Moreover, it was found that, under the same operating conditions, the deasphalted crude under testing could produce a larger amount of deposits than the original crude oil (Venditti et al., 2009a). Discussions with refiners reveals that even crudes with just traces of asphaltenes do produce fouling in the hot end of the PHT. This is confirmed by industrial observations and experimental work by Saleh et al. (2005b) who studied the fouling behavior of an Australian crude oil with low sulfur, ash, and asphaltene levels.

It should be noted that crude oil composition and its contaminants are not variables that can be generally manipulated whereas the process conditions of temperature and, in particular, velocity can be adjusted by the designer in order to mitigate fouling.

2.4.2. Temperature

As a general rule, high temperatures are usually associated with the promotion of chemical reaction fouling and corrosion and lead to shorter induction times. Furthermore, temperature has a direct effect on the ageing rate of the foulant layer. The deposit may be hardened and become difficult to remove or, alternatively, become weaker and tend to spall under the influence of temperature and time.

There has been some debate (Polley et al., 2002a) in the literature over whether the rate of the fouling reaction is a function of surface (wall) temperature, T_w, or rather the film temperature, T_f, between the bulk temperature, T_b, and the wall (Ebert and Panchal, 1995):

$T_{f} = 0.55 (T_{b} + T_{w})$ $T_{f} = 0.55 (T_{b} + T_{w})$

(2.10)

The temperature chosen to describe fouling rate dependence is associated with the zone where the reaction is assumed to take place (i.e., Case 1, 2, or 3 in Section 2.2.5).

2.4.3. Velocity and Shear Stress

Of all the variables, velocity is the one on which the designer has the greatest control (Bott, 1995b). An increase in velocity has a double-pronged effect on deposition of increasing the wall shear stress and increasing the convective heat transfer coefficient. The first is perhaps the more obvious effect on fouling: increased shear forces at the deposit/fluid interface impede fouling deposition and may even result in foulant removal. The role of shear stress in fouling behavior is highlighted by a plot by Joshi et al. (2009) that relates fouling rates to shear stress for several heat exchangers in six different PHTs, covering a range of tube-side velocities from 0.9–2.7 m s⁻¹. The majority of the exchangers considered had a design shear stress range of 5–8 Pa. The graph shows that the higher the shear stress, the lower the fouling, as expected, but it is surprising that temperature appears to have no effect on fouling rates, thereby apparently attributing the dominant role to shear stress alone.

The second is a less obvious—although important—effect of velocity on fouling. An increase in velocity, assuming that everything else remains constant, produces an increase in the convective heat transfer coefficient and thereby a reduction in total thermal resistance. As a result, the wall temperature is reduced, generating a beneficial effect on fouling rate.¹ This is the principle exploited by fouling mitigation devices based on a fixed wire matrix inserted on the tube-side of industrial units (Ritchie et al., 2009). On the other hand, if the deposition involves mass transfer of insoluble species (e.g., Case 1a and 2a in Figure 2.4), higher velocities will increase the diffusion toward the surface in the presence of a concentration gradient, thereby enhancing deposition.

Controversial effects of velocity on initial fouling rate have been reported by Crittenden et al. (1988a, 2009) in both model chemical and crude oil systems. At low surface temperatures the initial fouling rate decreased with increasing flow rate. At high surface temperatures the opposite dependency was observed. They concluded that a combination of high surface temperatures and low flow rates can lead to mass transfer of deposit precursors to the surface becoming the rate-determining step. This clashes with other observations, reported at the beginning of Section 2.4, that suggested adhesion or reaction is the rate-determining step. Industrial observations are that if the velocity is reduced, even for short periods, the result can be an increased deposition rate and, as a consequence, highly fouled surfaces. Once this has happened, even after the heat exchanger is brought back to normal operation, the deposits may have hardened their structure (aged) and not respond to increased velocities by detaching. This confirms the importance of maintaining the highest possible velocities allowed by other constraints given by an arbitrary earlier choice of allowable pressure drop (Butterworth, 2004) or by corrosion and vibration problems that may occur at large velocities (ESDU, 1989).

2.4.4. Surface Conditions and Roughness Dynamics

The importance of surface conditions has been addressed in several articles (Crittenden and Kolaczkowski, 1979a; Zhao and Müller-Steinhagen, 2001; Förster and Bohnet, 2001; Santos et al., 2004; Kukulka et al., 2010; Coletti et al., 2010). If the surface is corroded, it not only provides resistance to heat transfer but it also creates sites that encourage deposition. Additionally, corrosion products become particulate fouling for downstream sections of the heat exchanger.

Another important aspect of the surface is its roughness as it affects the convective heat transfer coefficient (Walker and Bott, 1973). The disruption of the viscous sublayer caused by a rough surface generates an increase in the turbulence level compared to that of smooth surfaces (Figure 2.7) which in turn produces higher heat transfer coefficients (Yaglom and Kader, 1974). A constant value of equivalent sand roughness is typically used, calculated from shorthand correlations. However, not only the initial status of the clean surface matters, but also its changes over time. The accumulation of foulant material on heat transfer surfaces often results in a progressive increase in roughness, with associated effects on heat transfer coefficient and pressure drop. Although in laminar flow the surface roughness has little effect on the friction factor and heat transfer coefficient (Shah and Sekulic, 2003), it plays an important role in the turbulent regime typical of industrial applications.

Figure 2.7 Effect of the growth of the deposits on a thermal surface. In clean conditions (a) the surface is relatively smooth and the boundary layer is well defined. Once deposition begins the deposits enhance surface roughness (b) and the laminar boundary layer is disrupted with related enhancement of the convective heat transfer coefficient.

The effect of roughness on the heat transfer coefficient is largely acknowledged in literature, but the effect of fouling on surface roughness dynamics has received relatively little attention (Ceylan and Kelbaliyev, 2003). Enhancement of heat transfer due to an initial increase in surface roughness generated in fouling tests has been reported and analyzed for particulate fouling by Crittenden and Alderman (1988), for crystallization fouling by Albert et al. (2009), and for corrosion fouling by Panchal (1988). For chemical reaction fouling, most of the reported cases are qualitative. Bott (1990) noted that, due to the complex interaction of variables, no systematic study on the effects of roughness on fouling had been undertaken. However, some quantitative results have been presented. Cousineau et al. (1988) reported the increase in heat transfer coefficient for a Bayer process in which sodium aluminosilicate scales deposit with kinetics of the second order. The apparent negative fouling resistance reported in the initial period (up to 40 h) of testing a crude oil by Knudsen et al. (1997) could be attributed to the progressive increase in surface roughness. Wilson and Watkinson (1996) reported both thermal and hydraulic roughness effects (from analysis of pressure drop measurements) in studies of autoxidation reaction fouling, while Asomaning et al. (2000) observed these in testing crude oils.

It should be pointed out that the dynamics of surface roughness are expected to apply mainly to laboratory testing, where more precise measurements of heat transfer are made, over short timescales. Industrial applications feature larger timescales, so that short-term variations in surface roughness are not noticed. Nonetheless, being able to capture this phenomenon is essential for the correct interpretation of accelerated fouling tests performed in pilot plants.

Quantitative modeling of roughness dynamics is in its infancy: for example, in their analysis of crude oil heat exchangers subject to fouling, Yeap et al. (2005) changed the surface roughness of clean tubes to that reported for bitumen by Kern (1988) to represent the foulant. An approach to modeling the increase in roughness due to fouling on tube surfaces was developed by Yiantsios and Karabelas (1994). They proposed a population balance model based on a population of roughness elements to describe deposition and removal in crystallization fouling. However, their focus was on the removal process and they did not discuss roughness in terms of hydrodynamics, pressure drop, and more importantly, impact on heat transfer. The first model for roughness dynamics in chemical reaction fouling was proposed by Coletti et al. (2010), which captured the effect of increased surface roughness due to fouling deposition on heat transfer coefficient and pressure drop. This model will be reviewed in Section 5.1.

2.5. Conclusions

In this Chapter, the complexity of the crude oil fouling problem has been described. In particular, it has been shown that despite the large research activity on the subject, much remains to be understood starting from the nature and composition of the crude oil itself and the chemical reactions that produce both precursor and fouling species. The chemical and physical phenomena that take place and interact with one another are extremely difficult to identify and isolate. Laboratory measurements and field experience have provided some useful indications on how key variables such as temperature and velocity affect fouling rates within the heat exchanger. However, further experimental work in obtaining fouling deposits under controlled conditions as well the development of analytical techniques that allow a better characterization of both crude oil and its deposits is needed. These will be described in Chapter 3 and Chapter 4, respectively.

..................Content has been hidden....................

You can't read the all page of ebook, please click here login for view all page.

Table of Contents for Chapter Two. Basic Science of the Fouling Process

Create new playlist

Sign In

Sign Up