3. Crystalline Deposits in Heat Exchangers

Transcription

1 Crystallization ouling in heat exchangers Crystalline Deposits in Heat Exchangers According to the mechanism responsible or the deposit generation, ouling o heat exchange suraces has been classiied into crystallization ouling, particulate ouling, chemical reaction ouling, corrosion ouling and biological ouling. Crystallization ouling is the deposition o a solid layer on a heat transer surace by a phase change process. In aqueous systems crystallization ouling mainly results rom the presence o dissolved inorganic salts in the lowing solution, which become supersaturated under the process conditions (precipitation ouling). Typical scaling problems in such cases are calcium carbonate, calcium sulphate and silica deposits. In oil pipelines and in dewaxing processes crystallization ouling can occur by solidiication o higher melting ractions o the organic mixture (reezing or solidiication ouling). Particulate ouling is the accumulation o particles suspended in a luid onto a heat exchange surace. Suspended particles can be crystals, ambient pollutants (sand, silt, clay), upstream corrosion products or products o chemical reactions occurring within the luid. Chemical reaction ouling involves deposits that are ormed as the results o chemical reactions at the heat transer surace. The heat exchanger surace material does not react itsel, although it may act as a catalyst. This kind o ouling is a common problem in chemical process industries, oil reineries and dairy plants. Corrosion ouling occurs when the heat exchanger material reacts with the luid to orm corrosion products on the heat transer surace. Biological ouling is the development and deposition o organic ilms consisting o micro-organisms and their products (microbial ouling) and the attachment and growth o macro organisms, such as barnacles or mussels (macrobial ouling). Generally, several ouling mechanisms occur at the same time, nearly always being mutually reinorcing. One notable exception is particle deposition occurring together with crystallization, which weakens an otherwise tenacious scale [Bot95]. Although the sedimentation o particles (thus particulate ouling) very oten is the origin o hard crystalline deposits orming in crystallizer vessels, the ocus in this work is given to crystallization ouling on a clean surace, which is the common ouling problem encountered in melt crystallization systems. Crystallization ouling is called reezing ouling in the literature. This might be because one o the most practical examples reezing o water pipes at winter is a typical case o this type o deposit ormation process. Other examples are blockage o chemical process lines or oil pipes and the reezing o liquid metals and glass, when poured through channels and nozzles [Wei97]. The progress o reezing ouling is presented in Fig An initiation period

2 26 Crystallization ouling in heat exchangers Maximum deposit thickness Deposit thickness A B C Figure 3.1 Development o a crystalline deposit in reezing ouling. A: induction period, B: surace nucleation period, C: growth o the deposit layer. Time (A) is needed or the surace nucleation (similar to crystallization processes described in Chapter 2.1). Ater the surace nucleation a period (B) can be observed, during which the surace is covered with the crystalline deposit. Depending on the temperature dierence on the surace and the properties o the crystallizing component, this period can be ininitely short, or total coverage may not be achievable at all. The growth o the deposit layer thickness (C) ollows. In ouling mitigation it is oten the aim to prolong the induction period as long as possible. The induction period can be extended according to Mersmann [Mer95] by: low heat lux and small temperature dierence at the wall optimal low velocity o crystal suspension smooth or coated pipes ultrasonic vibration on pipes presence o additives (ouling inhibitors) The resistance to heat transer due to the deposit layer is described by the ouling resistance, deined by R xd 1 1 = = (3.1) k h h d The nucleation on the heat exchanger surace can lead to surace roughening, which leads to increasing turbulence and higher heat transer coeicient on the surace. In this case the ouling resistance can get negative values. However, when the deposit thickness increases heat transer rom the surace is hindered. The driving orce or crystallization ouling is the

3 Crystallization ouling in heat exchangers 27 temperature dierence (undercooling) on the solid-liquid phase boundary. For a lat plate the temperature on the phase boundary can be calculated by assuming that the system is at pseudo-steady state by the equation: ( T T ) + + R = ( T T ) s a 1 ho x k w 1 [ h + N H ] b s i (3.2) For a pipe low this equation must be transormed to the ollowing orm: i i ( T T ) + + R = ( T T ) s a 1 ho D D o x k w D D log 1 [ h + N H ] b s i (3.3) It can be seen rom equations 3.2 and 3.3 that by increasing ouling resistance the temperature dierence between the deposit surace and the bulk solution becomes smaller. When the driving orce or the deposit ormation (crystallization) becomes smaller, the growth rate o the deposit gradually decreases. Finally, the limiting deposit thickness is reached, at which the heat transer is hindered by the deposit by such extent that no driving orce exists. Theoretically this limiting thickness is approached only asymptotically. In a practical case o a pipe low the whole pipe can o course be blocked beore the limiting thickness is achieved. For a given system this process can be inluenced by two main actors: The surace temperature, T s, and the low conditions inluencing the heat transer coeicient on the phase boundary, h o. For many technical applications the ormation o a rozen crust must be allowed, because o constructive limitations or additional costs needed or preventing the crystallization ouling. In such cases it is important to investigate the reeze shut o the system in order to prevent damage to process equipment by the increasing pressure drop [Wei97]. In addition to the pure heat exchange approach, it is important to understand that there are simultaneously a deposit growth phenomenon and a deposit removal phenomenon taking place in the system. How strong is the deposit removal eect depends on the one hand on the orces in the low, and on the other hand on the cohesion orces in the deposit layer and adhesion orces o the deposit to the heat exchange surace. The orces in the low can be aected by the low conditions, and in crystallization processes additionally by the suspension density o crystals in the lowing stream. The adhesion orces between the deposit and the heat exchange surace are inluenced by the surace temperature dierence and the surace properties. For the deposit removal rate has been presented an equation [För99] 1 3 µ g ρ m r = CR τ (3.4) ρ σ where C R is a constant, µ is the viscosity o the solution, g the acceleration by gravity and ρ and ρ the densities o the solution and the deposit, respectively. The shear strength o the deposit is given by σ and τ is the shear stress. This equation has been developed or the case

4 28 Crystallization ouling in heat exchangers o a removal rom the surace o a ully developed deposit. However, it can as well be applied to the induction period, where no ully developed deposit layer exists, or or the case where the whole deposit thickness is removed rom the surace. In these cases the shear strength o the deposit, σ, should be replaced by the adhesion orces between the deposit and the surace. 3.1 Eect o Flow Conditions It has been proved that when the low velocity increases in the laminar region the deposit thickness increases due to better heat exchange conditions on the surace. However, when the low velocity reaches the turbulent region and urther increases, the deposit thickness decreases due to higher shear orces o the luid stream. Thereby, the deposit thickness reaches a maximum value at a certain low velocity. In pipe low it very oten occurs that the deposit layer possesses a wavy proile, as presented in Fig [Hir85]. The reason or this are the changing low conditions along the pipe length. It can be seen that the deposit can at some low conditions melt and almost disappear. From Fig it can also be well observed how the low conditions (Re D ) and the surace temperature dierence (θ c ) are related to each other in the case o crystallization ouling. At higher Reynolds numbers (in this case higher low velocity) larger surace temperature dierences are necessary to produce a deposit layer o comparable thickness. The limiting deposit layer thickness and the conditions or ormation o wavy deposit layers have been well investigated (e.g. [Eps76]). However, the limiting low conditions have been paid less interest to. The reason may be that in the cases where reezing ouling usually occurs the low conditions may not be easily adjustable (long oil and water supply lines, low o metals). However, in chemical process equipment, such as crystallization processes, the lows can be chosen more reely, which oers a possibility or mitigation o crystallization ouling. It has been reported that shear dispersion causes crystals in a suspension to migrate toward the opposite direction o the shear rate gradient [Gra91, Mon94]. That means that a low has a tendency to hinder deposit ormation. When Fig. 3.1 and Fig are compared it can be seen that the curves presented have similar orm. However, Fig. 3.1 shows the deposit thickness vs. time and Fig vs. the pipe length. Anyway, it has been reported by Ribeiro et al. [Rib97] that, similar to the induction time in Fig. 3.1, an induction length is obtained in pipelines transporting crude oil at which no crystallization ouling occurs on the pipe surace. The authors have mentioned that up to this point the inner surace o the pipe has not yet reached the saturation temperature needed or crystallization to take place. However, this is not probable, because a sudden increase o the deposit thickness at the point o ormation is observed, which suggests that a certain undercooling already exists at this point.

5 Crystallization ouling in heat exchangers 29 Figure Steady-state wavy ice layers in a pipe low [Hir85]. Bohnet [Boh85] has developed an equation or the time needed to reach the asymptotic maximum ouling layer thickness by crystallization ouling. The equation can be written: R 2 ln = K P v t (3.1.1) * R where K P is constant including the physical parameters o the liquid and the crystalline deposit. From this ormula it can be concluded that by crystallization the layer thickness can be aected by the low velocity o the cooled stream. However, the low velocity appears also in the equations or mass transer and heat transer, and an absolute value cannot be calculated without knowing these eects. Anyway, the heat transer is known to play in melt crystallization processes much more important role than the mass transer. Thereby, the increment o heat transer by increasing low velocity inluences the process o ouling mitigation ar more positively than can be eected negatively by boosted mass transer rate. It was reported by Bott [Bot95] that the asymptotic ouling resistance due to wax deposits rom a kerosene stream was inversely proportional to the square o the Reynolds number. The way in which the low velocity aects the ouling process has been demonstrated in the work o Yang et al. [Yan02]. The authors investigated the induction time at dierent conditions or the case o CaCO 3. Higher low velocities at a ixed heat lux decreased the wall temperature through more eective heat exchange on the surace. Thereby, CaCO 3 being a salt with inverse solubility, was the driving orce or crystallization reduced and the induction time prolonged. The opposite was observed when the heat exchange surace temperature was kept constant increasing low velocity caused the induction time to get shorter. The reason is that, the driving orce being the same, the mass transer o ions was increased by the increasing turbulence, which aected the growth o the surace deposits positively. In addition to the heat and mass transer eects the removal phenomena was investigated. For this purpose the low was periodically switched between low rates o 1.39

6 30 Crystallization ouling in heat exchangers and 1.62 m/s every ive minutes. It was ound out that when the higher low rates were periodically used rom the beginning on, the induction time was longer and the deposit grew slower. However, when the same treatment was used to a grown deposit layer the deposit growth was aster than without the periodical change o the low velocity. The authors did not try to explain this phenomenon, but it should be clear that the adhesive and cohesive orces o the deposit play an important role on such a process. In this case the adhesive orces between the heat exchange surace and the deposit are weaker than the cohesive orces in a developed layer o the hard ionic crystals. When at lower low rates merely changes in heat and mass transer processes were observed, the removal was irst noticeable at the low rate o 1.62 m/s or the period o nucleation and spread over the surace. On the other hand, this low velocity again merely enhanced the heat and mass transer processes applied to a ully grown layer. From this can be concluded that every type o connecting orce (adhesive/cohesive) has its limiting low velocity (shear orce by the low), at which the removal orces grow in importance higher than the positive inluence to the deposit growth by the improved heat and mass transer. A reported way o ouling mitigation by adjusting the low conditions is pulsation o the process stream low, which increases the shear stress on the heat exchange surace [För99]. I the pulsation period is suiciently short (e.g. 10 s) the shear orces are suicient to remove ormed crystallites rom the heat exchange surace. When the length o the oscillation period is increased the amount o crystals and a longer residence time on the surace result in better attachment o the deposited material and necessitate a higher shear stress or deposit removal. 3.2 Eect o Crystalline Suspensions Söhnel et al. [Söh96] have investigated the inluence o the crystal suspension on the incrustation behaviour o sodium perborate. The authors showed that an incrustation orming on a clean surace took place in two steps: ormation o a complete layer and urther linear growth o the layer thickness. In a case where the surace was covered initially with a crystalline layer only the linear growth was observed rom beginning o the experiment. With the substance studied the authors concluded that the mechanism o initiation o incrustation was attachment o crystals to the surace, which was concluded rom the inluence o the suspension density. From the experimental results it could also be seen that despite that the initiation period or higher suspension densities through eased attachment is shorter, the urther growth is aster in the case o lower suspension density. This is due to the higher attrition orce o the higher suspension density. This was also partly proven by the observation that on the calm side o the bale the crystalline layer, even thinner than on the low side, always covered the whole surace, whereas on the low side the shear orces never allowed the whole surace to be covered. The mechanism in this case is not exactly crystallization ouling, but sedimentation with urther growth o the attached crystals. However, the results show that a suspension provides a higher shear orce or removal o the deposit layers.

7 Crystallization ouling in heat exchangers 31 The higher shear orce o the low at higher suspension densities is the result o increased apparent viscosity o the lowing luid by suspended solids. It has been ound that suspensions o high solids concentration show yield dilatant rheological behaviour [Wag77, Cha77, Ho74]. The same behaviour was observed by Haaiedh [Ha88], who has also shown the eect o increasing solids concentration presented in Fig Dilatant rheological behaviour means that the shear stress increases exponentially when the shear rate (low velocity) increases, by equation τ τ y K γ n = + n > 1 (3.2.1) Figure Eect o solids concentration on the apparent dynamic viscosity o a suspension [Ha88]. The result o a dilatant behaviour o a suspension is that an increment in the low velocity brings a larger increment in the shear stress than or a Newtonian luid (typical or a clear liquid), and that by higher solids content the shear stress also is higher. From Fig it can be seen that the apparent viscosity has a minimum accompanied with a slow rise in the case o yield dilatant rheology. In addition to shear stress, the viscosity also aects the processes o heat and mass transer. The higher viscosity aects these processes negatively, which increases the surace temperature dierence. It can be seen that running a suspension process at a low shear rate (low low velocity) can lead to problems with deposit ormation due to the high viscosity. Solids present in a system can also act as catalysts or nucleation, as shown by Vendel and Rasmuson [Ven97]. The authors ound out that the contact between a surace and a crystal can induce contact nuclei, which attach on the surace. However, secondary nuclei are always created in a supersaturated crystal suspension. Whether the nuclei attach and grow urther on

8 32 Crystallization ouling in heat exchangers the surace depends also on the surace characteristics. Vendel and Rasmuson also investigated the eect o surace characteristics and the results o their work are discussed in Chapter 3.3. Deposit removal by particle abrasion was studied by Mori et al. [Mor96]. It was proven that particle abrasion increases the heat transer coeicient on the surace. Ater the asymptotic ouling resistance was reached the heat transer coeicient could be returned to 95% o the initial value by addition o glass beads. However, when the value was corrected or the real removal process taking into account the improvement in the heat exchange coeicient, the ouling resistance o the deposit (ipso acto the thickness) was decreased by 43%. It was stated by the authors that using glass beads the heat transer coeicient approached a constant value when the suspension density o beads was over 50 kg/m 3. It was noticed that higher low rates did not result in more eective deposit removal. On the opposite, the deposit was even harder to remove. This was explained by saying that the luid velocity has an inluence on the ouling layer bond strength. 3.3 Eect o Surace Structure o Heat Exchanger The eect o the surace structure o the heat exchanger can be divided to two actors: the geometrical coniguration o the heat exchanger and the properties o the heat exchanger surace. While the geometrical construction brings its eect through its inluence on the low patters, the surace properties inluence the adhesive orces between the surace and the deposit. The adhesive orces can be divided into mechanical interactions and molecular interactions, such as Lishitz-van der Waals orces, Lewis acid base orces and double layer orces [Boh03]. The attachment orces between particles and a surace have been recently investigated by Sonnenberg and Schmidt [Son04], who modelled the attachment orces or various particle shapes based on the van der Waals orces between the particle and the surace. In her experiments Schuldei [Sch00] ound out that glass and stainless steel produced dierent results concerning the ouling o the heat exchange surace by crystallization. She ound out that in metal surace the crystallization ormed evenly on the surace. On the other hand, on the glass surace the incrustations started rom a joint position and grew rom that position on until the pipe was totally blocked. In a glass pipe the volume low was inluenced by an incrustation sooner at higher Reynolds numbers o the stream and also the pipe was sooner blocked when higher Reynolds numbers were applied. In a metal pipe the situation was dierent: As in a glass pipe the growth o an incrustation started earlier when the Reynolds number was higher, but the time needed or the blockage o the pipe was decreased. Also the time, which the equipment could operate without disruption, was shorter or the metal pipe, which is explained by the higher surace roughness o the metal pipe and higher heat conductivity o the wall material. The heat lux through metal and glass walls are dierent. From that ollows that the temperature at the inner surace o the metal wall is by cooling lower, which makes the growth easier and aster. Thereore, the pipe will be blocked

9 Crystallization ouling in heat exchangers 33 sooner. The way in which a pipe most eiciently could be blocked by ice was studied by Keary and Bowen [Kea98]. They state that the pipe diameter has inluence on the reezing behaviour through dierent turbulence conditions. The inluence o dierent surace materials on crystallization o organic chemicals has been investigated by Haasner [Haa02]. He showed that the eect o the surace material also depended on the compound system to be crystallized. The decisive actor or the ease o ormation o crystalline layers on a cooled surace was ound to be the interacial tension between the melt and the cooled surace. The orces inluencing the attachment o solid matter o a surace can be described with the help o the interacial tensions with the equation [Isr85]: Wcsl = γ + γ γ (3.3.1) cl sl cs W csl represents the work needed to remove a crystal rom the surace and to create the corresponding crystal-liquid and surace-liquid interaces. Vendel and Rasmuson [Ven97] have investigated the inluence o the surace material on the attachment o crystals. They ound that organic crystals tend to attach on telon suraces, but crystals ormed on a steel surace tend to detach rom the surace. It was also ound that the adhesion onto the surace was governed by the same physical properties as the catalytic activity o the crystal-surace interace on primary nucleation. These results are in good agreement with the results presented by Haasner [Haa02] or the inluence o telon suraces. As the conclusion was stated that increasing the ainity o the crystallizing substance to the solid surace promotes catalyzed primary nucleation, collision initiation and adhesion o the deposit on the surace. Nucleation can be avoided using special types o surace materials or by additives (like polyamides). In the case o special surace materials the negative aspects are high price and low heat transer coeicients. The additives may inluence the process negatively by inluencing the product quality or by accumulation in the process streams [Cor01]. Bansal et al. [Ban01] compared crystallization ouling properties in plate-and-rame and double-pipe heat exchangers. They stated that the ouling on the heat exchange surace depended strongly on the shear rates prevailing in the dierent apparatuses. High shear rates were beneicial in avoiding crystallization ouling. Fouling inhibition could in their work be more eiciently achieved in the plate heat exchanger, where the surace proile enhanced high local shear rates. 3.4 Summary o Existing Research on Crystallization Fouling In the theoretical part considering crystallization ouling on heat exchanger suraces the basics o ouling phenomena and their inluence on the heat transer are presented. The mechanisms by which crystalline incrustations are initiated and by which the deposit growth proceeds have been discussed. In Chapter 3.1 the ocus was given to the eect o low conditions on

10 34 Crystallization ouling in heat exchangers incrustation. In Chapter 3.2 has been presented the work on the inluence o crystal suspensions on the ouling o the heat exchange suraces. In Chapter 3.3 the structure o the heat exchange surace, especially the properties o the surace material, were discussed. The ouling phenomena are maniold and a large amount o research has been carried out in this ield. The research has, however, concentrated on the ouling mechanisms taking place in heat exchangers without phase change. The work on crystallization ouling has ocused mainly on incrustations caused by inorganic salts. The reezing ouling phenomena are usually discussed in context o solidiication o pipelines without any intentional heat exchange or phase change process. The lack o research on incrustations in melt crystallization processes probably has its reason in the severity o the problem. It is commonly thought that deposit ormation on cooled suraces in melt crystallization is inevitable, and cannot be mitigated simply by adjusting the process conditions. However, the same laws o physics are behind the deposit ormation in melt crystallization processes, as in any kind o crystallization ouling. This work serves to investigate the possibilities o incrustation mitigation in suspension melt crystallization processes making use o the process conditions. The inluence o the orces acting on the heat exchange surace are experimentally proved. Also the conditions, at which crystalline deposits orm on a cooled surace, and the inluence o these conditions on the layer structure, are investigated. With the research carried out in this work on the ouling phenomena the deposit ormation and the deposit removal processes, and how they are inluenced by the process conditions will be better understood. This gives the physical background or application o novel constructions or suspension melt crystallization processes.