Dec-2019

Flow maldistribution in shell and tube heat exchangers

Old designs of shell and tube heat exchangers need to be revamped to develop uniform flow distribution in the inlet header and mitigate tube-side fouling.

DEREK SUMSION and MIKE WATSON, Tube Tech International
TIM DORAU, RICHARD SCHAB, SIMON UNZ, MICHAEL BECKMANN and M REZA MALAYERI
Technische Universität Dresden

Article Summary

Heat exchangers are the workhorse of most chemical, petrochemical, food processing, and power generating processes. The global heat exchanger market is estimated to top $22.59 billion by 2023, with an increase of approximately 9% per annum.¹ Among many types of exchangers, approximately 65% of the heat exchanger market is still dominated by the shell and tube type heat exchanger (see Figure 1). It is largely favoured due to its long performance history, relative simplicity in design, operation and maintenance, as well as its wide temperature and pressure design ranges.

Manufacturers are presently under increasing pressure to produce heat exchangers that are more efficient in terms of heat recovery and use of material, while at the same time being faced with fluids that are increasingly difficult to process. One major problem directly related to these requirements is the deposition of unwanted materials on heat transfer surfaces, which occurs in most heat exchangers.² Conservative studies estimated that heat exchanger fouling leads to additional costs in the order of 0.25% of the GDP of industrialised countries, and that it is responsible for 2.5% of the total equivalent anthropogenic emissions of carbon dioxide.³

Fouling is generally a complicated process that involves a considerable number of independent variables.⁴ Among many dominant parameters, fluid velocity is known to have a potentially profound impact on fouling propensity. This perhaps is best established in general fouling modelling:

                  [1]
       
where subscripts d and r refer to deposit and removal, respectively. In this equation, m ̇r, removal rate, is a function of wall shear stress, and strength of deposit layer onto the surface, thus:

                 [2]

where τ is the wall shear stress, a function of fluid velocity; φ is the strength of deposit; and K is a proportionality constant. Field and experimental results show that, for instance in refineries, asphaltenic fouling at crude oil preheat train conditions can be mitigated when the wall shear stress exceeds approximately 10 Pa and is profoundly minimised when the wall shear stress is above 15 Pa. Therefore, increased velocity has long been sought as a possible means to minimise fouling. Nevertheless, in shell and tube heat exchangers, apart from the operating consequences of a higher pump requirement and increased pressure drop, this may not be a straightforward option due to the occurrence of flow maldistribution in the header.⁵ The latter implies that some exchanger tubes may experience lower velocities while others experience higher velocities. This would, in turn, mean that some tubes may also foul or even clog, causing unplanned shutdown of the exchanger. Accordingly, qualitative and quantitative prediction of flow maldistribution is of interest in heat exchanger engineering.

Flow maldistribution
In operation, a fluid flows through the tube bundle from the intake header to the rear header and, during its passage, it is heated or cooled by heat transfer through the walls of the tubes by a process or working fluid. In current manifestations of these devices, the product fluid flow rates through the individual tubes are not uniform but are greater in the tubes near or in line with the intake pipe, as these offer less flow resistance. It is common for the intake pipe to be connected to a side of the intake header and to extend at an angle to the axis of the tubes in the bundle. In such a configuration, there is still greater variation in the flow rates through the individual tubes.

The uniform distribution of flow in the tube bundle of shell and tube heat exchangers is an assumption in conventional heat exchanger design. Nevertheless, in practice flow maldistribution is an inevitable occurrence, which may have severe implications for the thermal and mechanical performance of heat exchangers.⁵ Flow maldistribution can be caused either by operating conditions (viscosity-induced or density-induced maldistribution, multiphase flow, fouling) and/or by exchanger geometry and mechanical design features (geometrical design, manufacturing deficiencies and tolerances). Geometry-induced flow maldistribution can further be categorised into passage-to-passage, manifold-induced, and gross flow maldistribution.

Computational fluid dynamics (CFD) results for the flow velocity distribution in the front header of a stereotype shell and tube heat exchanger (one-pass) with 464 tubes (see Figure 2) showed that there exists a noticeable flow maldistribution for a typical fluid velocity of 3.0 m/s. This can be explained by a flow circulation when flow enters the front head. This, in turn, results in the formation of dead zones in the tube sheet, and many of the tubes receive only a very low flow velocity.

Similar CFD studies have been conducted for a shell and tube exchanger with a two-pass configuration, which is commonly used in the preheat trains of refineries. Apart from directly plotting velocity on a plane (see Figure 2), another approach to visualising flow maldistribution across the tube bundle is a bar chart. The average velocity of each tube in the first pass is calculated at a position 1m downstream of the tube entrance. The resulting tube velocity is normalised using the average velocity in the whole tube bundle at 1m length, which is 0.497 m/s given an inlet velocity of 1.252 m/s. The results are summarised in Figure 3. A maximum deviation of ±35% from the average velocity can be observed. Only 339 of 585 tubes in the first pass are part of the ±10% range around the average value in the centre. Due to the non-linear relationship between the wall shear stress and fouling rate, the tubes on the lower end of the velocity distribution are especially prone to fouling. The assumption of uniform flow across the bundle, which is often used for heat exchanger calculations, is thus questionable for a two-pass heat exchanger with 1170 tubes.
One possibility to mitigate fouling is to use higher tube velocities that are directly related to an increase in wall shear stress. In general, a higher tube velocity can only be achieved at the expense of an increase in pressure drop, which in turn leads to the need to use stronger pumps, requiring more energy. Using 1.49 m/s average tube velocity (see Figure 4) slightly decreased the flow maldistribution in comparison to Figure 3, but the general velocity distribution is still far from being uniform. Three other average tube velocities between 0.497 m/s and 1.49 m/s were investigated, for which strong tube velocity maldistributions have been experienced.

To summarise the CFD results for one- and two-pass exchangers, it can be inferred that in each of the heat exchanger designs a significant proportion of tubes have a tube velocity below the desired uniform value. Accordingly, the optimisation of flow for better uniform flow distribution becomes imperative. There are several advantages to achieving an optimised, uniform flow rate through the tubes of a heat exchanger of this type, including improved heat transfer efficiency, reduced fouling rates, reduced energy consumption, and longer lifetime.