Heat exchange fouling.
Heat exchangers are omnipresent in the chemical process industries. Even a moderate size oil refinery, for example, will typically contain dozens if not hundreds of exchangers. The design of such exchangers requires specification of a heat transfer surface area, A, which is given by an equation of the form
A = q[Sigma]R/[Delta][T.sub.eff] (1)
while the rating of an already functioning exchanger is given by
q = A [multiplied by] [T.sub.eff]/[Sigma]R (2)
where q is the heat transfer rate and [Delta][T.sub.eff] the effective temperature difference driving force between the hot and the cold sides of the exchanger. Assuming, for simplicity, a plate-type exchanger in which A is the same on both the hot and the cold sides, the term [Sigma]R is given by
[Sigma]R = [R.sub.o] + [R.sub.fh] + [R.sub.fc] (3)
where [R.sub.o] is the resistance (for unit area) to heat transfer of the exchanger under clean conditions, [R.sub.fh] the additional resistance acquired by the exchanger due to fouling of its hot surface, and [R.sub.fc] the corresponding fouling resistance on the cold surface. [R.sub.o] can commonly be predicted with confidence to two significant figures from the existing state of knowledge in the field of heat transfer, but prior estimation of [R.sub.f], the values of which may eventually approach in magnitude that of [R.sub.o], is rarely reliable even to one significant figure.
The unit area resistance to heat transfer caused by a fouling deposit, [R.sub.f], has been dubbed the "fouling factor" by the Tubular Exchangers Manufacturers Association (TEMA), whose listings of [R.sub.f] values are used by most heat exchanger designer. Although TEMA continues to revise the values of [R.sub.f] for various process streams listed in earlier editions of its widely used Standards, and to increase the number of process streams for which it specifies values of [R.sub.f], the specification are mainly single-valued, ignoring in most cases the effect on [R.sub.f] of such key process variables as fluid velocity, bulk temperature and composition, and surface temperature. More importantly, the use of single invariant values of [R.sub.f] in conjunction with Equations (1) and (3) to specify A ignores the transient nature of fouling and treats the heat exchange process as if it were instantaneously at a steady state with fixed values of [R.sub.fh] and [R.sub.fc]. Since initially the heat exchanger surfaces are clean, this means that the exchanger is overdesigned at start-up, which may dictate the use of lower initial fluid velocities and give rise to higher initial surface temperatures than prescribed by the design, and thus result in more and faster fouling than would otherwise be the case.
The variation of [R.sub.f] with time, i.e., the kinetics of fouling, has therefore been the subject of considerable study during the past three decades. Aside from the intrinsic curiosity-driven interest in this subject, which involves application of almost every technical fundamental of chemical engineering, the investigations have also been prompted by the very large costs incurred in industry as a result of heat transfer surface fouling. The economic penalties include higher capital expenditures for oversized plants, costs of cleaning heat exchangers and of antifoulants, loss of production due to shutdown for cleaning or due to reduced overall plant efficiency and, last but not least, energy losses due to increased thermal inefficiencies (hence greater fuel consumption) and increased pressure drops (hence greater pumping power requirements). Sessions on fouling are now held at most heat transfer conferences, and conferences entirely devoted to fouling, with published proceedings, are also now a regular occurrence.
It is convenient to classify fouling from liquids according to the key physical/chemical process essential to the particular fouling phenomenon. The five primary categories thus identified are particulate fouling from suspension, crystallization fouling from solution, chemical reaction fouling by deposit formation at the heat transfer surface via chemical reactions in which the surface itself is not a reactant, corrosion fouling due to accumulation of indigenous corrosion products on the heat transfer surface, and biofouling due to attachment of macro-organisms or microorganisms to the surface. In the Chemical Engineering Department at UBC, we have limited our efforts to studies of the first three categories, though in some cases the results have been perturbed by unplanned corrosion. The principal investigators have been Paul Watkinson, FCIC, in all three categories, with recent emphasis on chemical reaction fouling from hydrocarbon streams; Bruce Bowen in particle deposition under isothermal conditions; and the present author, in both particulate and chemical reaction fouling. Though in industrial practice two or more of these fouling modes commonly occur together, our studies have thus far been limited to a single mode in isolation and, where possible, with a single species for any mode, e.g, one size, shape and density of particle, one crystallizing salt or one chemical reactant. In this way it is hoped to build up some fundamental understanding of the individual fouling modes before studying their frequently synergistic interactions.
The principal method used in our Department for monitoring thermal fouling involves the application of a known constant and uniform heat flux (e.g., by electrical heating) at the single heat transfer surface under study and the measurement by means of a thermocouple of the consequent wall temperature rise, [T.sub.w](t) - [T.sub.wo], due to buildup of the fouling deposit from the fluid whose constant bulk temperature is [T.sub.b]. Since for the initial clean surface,
q/A = [T.sub.wo] - [T.sub.b]/[R.sub.o] (4)
and for the fouled surface after some time t,
q/A = [T.sub.w](t) - [T.sub.b]/[R.sub.o] + [R.sub.f](t) (5)
combinations of Equations (4) and (5) results in
[R.sub.f](t) = [T.sub.w] (t) - [T.sub.wo] / q/A (6)
Characteristic fouling curves are shown in Figure 1. A delay time [t.sub.D] is commonly associated with all fouling modes other than particulate fouling. The asymptotic curve is distinguishable from the falling rate curve by the fact that as t approaches infinity, [R.sub.f] approaches a finite value in the former case while [R.sub.f] approaches infinity in the latter. Linear fouling is commonly associated with crystallization from pure particle-free supersaturated solutions under constant heat flux conditions; non-asymptotic falling rate fouling with crystallization under conditions in which the temperature of the deposit-fluid interface falls with time (which it does not normally do under constant heat flux conditions); asymptotic fouling with any fouling mode in which either deposit attachment to the surface is eventually inhibited entirely by various possible autoretardation mechanisms, or deposit detachment increases with deposit thickness to the point where it equals deposit attachment; and sawtooth fouling with periodic shedding of deposit clumps due, for example, to velocity surges or to deposit aging. The absolute location of any fouling curve depends on such process variables as fluid composition, fluid velocity and surface temperature.
The deposition process for all three categories of fouling under study at UBC (and also for biofouling) involves a transport step in series with deposit attachment to the surface, sometimes followed by piecemeal detachment. The transport step has been well modelled from mass transfer theory, but models for the deposit attachment step are under development at UBC and elsewhere. Also under study, mainly under the direction of Paul Watkinson, is the fouling of organic streams under both oxygenated conditions (autoxidation), for which considerable progress has been reported in recent years, and non-oxidative conditions (e.g., thermal decomposition, polymerization), which has received less attention. In the latter category, special attention is now being given to the role of asphaltenes in the fouling of crude oils.
The increased understanding of fouling which is achieved by controlled laboratory research and analysis, as well as by more casual observations and measurements in industry, bears its most immediate fruit in giving some indictions on how to operate existing equipment in such a way as to minimize or at least reduce fouling, but is very slow in influencing design procedures for heat exchangers. Nevertheless there is an influence. If, for example, one considers the effect of fluid velocity on fouling of heat transfer surfaces, one finds that whereas increasing the velocity increases mass transfer, it also decreases deposit attachment and enhances detachment. More often than not, especially in turbulent flow, it is one or both of the last two factors which over-rides the first, in which case raising the velocity will tend to inhibit fouling. It is notable that, whereas some four decades ago cooling water in heat exchanger tubes was commonly assigned a velocity of 1 metre/sec for design purposes, that velocity has since been increased to 2 metres/sec.
Norman Epstein, FCIC is a nominally reared professor in the Department of Chemical and Bio-Resource Engineering at the University of British Columbia (UBC), Vancouver, BC. He is also an Editor Emeritus of The Canadian Journal of Chemical Engineering.
|Printer friendly Cite/link Email Feedback|
|Publication:||Canadian Chemical News|
|Date:||Nov 1, 1996|
|Previous Article:||Restructuring in the chemical industry: fiscal fat trimming or corporate anorexia.|
|Next Article:||Power of the sun.|