(134a) Tubular Enhancement for Laminar Flow Reactors Using Static in-Line Devices

Introduction

Increasing efficiency on industrial plants is an ongoing endeavour, with efforts to reduce wastage, carbon emissions and production times. As a result, more and more plants are opting to use tubular reactors in place of traditional batch reactors, moving to more continuous processing.

The difference in flow mechanics results in a change in process requirements, and if not considered carefully can lead to poorly designed units either too large or underperforming. A difficulty in processing using tubular flow for reactions can be a result of inhomogeneity, where reactants are not well dispersed in their solvent. Particularly in high viscosity fluids, a high axial dispersion can be observed resulting in a non-ideal profile dissimilar to plug flow. Furthermore, poor heat transfer can result in non-uniform reactions, leading to poor product yields.

In-line static mixers and turbulators have been used to improve these issues, particularly where flow is viscous and prone to poor heat transfer and mixing. A range of devices are commercially available, predominantly static mixers, which enhance exchanger performance at the cost of pressure drop. These devices are often designed to cut the flow and incorporate sections of fluid into one another through forced channels or paths, and although these inserts can improve homogeneity of the fluid, a significant amount of energy is lost through friction and therefore requires more pumping power. Furthermore, improving mixedness does not necessarily result in better heat transfer or residence time.

We investigated the advantages of using two types of in-line turbulator in viscous applications, a wire matrix-based insert (hiTRAN), and a new plate type insert (hiVISC). These devices are designed for enhancing performance in tubular reactors, specifically improving residence time distributions and tube-side heat transfer.

Methodology

For this study, a combination of experimental and theoretical techniques were utilised. These methods measured the effect of tubular devices on heat transfer, pressure drop, residence time and gave an indication of mixing.

Heat transfer and pressure drop measurements

Heat transfer enhancement and pressure drop penalty was measured using the high viscosity flow rig (HVFR), consisting of a heating test section and a cooling test section which were operated simultaneously. Each test section consisted of a tube-in-tube configuration, with water acting as the service fluid on the shell side flowing counter currently. Glycerol was used as the process fluid due to its Newtonian behaviour and range of viscosities at the operational temperatures between 10 - 60 ^oC. Measurements of tube side and shell side temperature were taken on both inlets and outlets. Each test section was also fitted with a differential pressure sensor, which measured the in-tube pressure drop. The HVFR was also fitted with a Coriolis flowmeter and a bypass section to measure glycerol viscosity.

Experiments were conducted by inserting a tubular enhancement device into each test section and measuring the heat transfer over a range of flowrates and temperatures by adjusting the gear pump frequency using a frequency inverter. Temperatures were controlled using two separate flow loops, one for cooling coupled with a chiller unit and one for heating which used a controlled tank with an immersion heater.

Verified computational fluid dynamics (CFD) models were used to extend the scope of study for each of the devices. A computational model of the HVFR test sections were constructed in Ansys Fluent, and extensively calculated to minimise errors. The computational setup mimicked the conditions of HVFR tests, where heat transfer and pressure drop results accurately matched those observed in experiments. These results were then extended into deep laminar flow, by computationally increasing the fluid viscosity and thereby reducing the Reynolds number by an order of magnitude.

Mixing measurements

Mixedness was measured using planar laser induced fluorescence (PLIF), where a primary fluid is mixed with a secondary fluid doped with a fluorescent material and a plane is imaged using a laser sheet and camera. As the secondary fluid fluoresces under laser light, the resulting image consists of bright and dark areas representing each of the species which can then be computationally analysed to derive the mixing effectiveness of the enhancement device being tested.

The planar laser induced fluorescence rig (PLIFR) used glycerol as both the primary and secondary fluids, where the secondary fluid was doped with Rhodamine for the fluorescent effect. Flow was temperature controlled using a heating/cooling unit and a plate exchanger, giving a range of viscosities for glycerol and thereby a range of Reynolds numbers ranging from 0.03 - 16.5 (7.5 - 55 ^oC).

Pure glycerol flow was established first, followed by a central flow injection prior to the 23 mm internal diameter by 300 mm long test section. Imaging occurred after the testing section, where a laser sheet illuminated a plane through a clear Perspex section. Pressure drop across the tube internals was measured using a differential pressure sensor.

The resulting images had 13 μm spatial resolution, with 4096 levels of brightness. 50 images were captured for each test, combined, calibrated, and normalised to give a final result. Concentration-brightness linearity and temperature independence tests indicated pixel brightness was only affected by concentration of the secondary fluid.

Results and conclusions

Pressure drop

Baseline experiments were conducted using the HVFR with no tube internals, where isothermal results in pressure drop matched theoretical expectations. Experiments were conducted for Re = 0.1 - 100, by adjusting the pump frequency and changing the temperature of glycerol. Inserts increased the pressure drop observed in the test sections. For hiTRAN, increasing the packing density resulting in an increase in pressure drop, and the same trend was observed when reducing the pitch for hiVISC.

Heat transfer

As with pressure drop tests, initial baseline experiments for tube side heat transfer were conducted for the test sections without internals using the HVFR. The same tests were then conducted for each of the inserts. All inserts increased the heat transfer coefficient in comparison to the empty tube. Low density hiTRAN had the lowest heat transfer enhancement, which increased with packing density. Similarly, heat transfer coefficient increased with a reduction in pitch for hiVISC.

Heat transfer efficiency

An efficiency for each insert was calculated by computing a ratio between heat transfer enhancement over pressure loss in comparison to the empty tube results. Heat transfer coefficient was made dimensionless by calculating j-factor:

j = Nu Pr ^-n

Dimensionless pressure drop was calculated using Darcy friction factor, f:

ΔP / L = (f ρ v²) / (2 D)

Finally, insert efficiency, η, was calculated as follows:

η = [ (j _insert / j _empty) / (f _insert / f _empty) ^m ]

Using this definition, insert performance could be evaluated against the empty tube. η = 1 represented empty tube performance, whereas an increased value indicated increased heat transfer with less pressure penalty than an equivalent length of empty tube required for the same heat transfer.

hiTRAN inserts show an increasing η with increasing Re, indicating best performance where Re > 50, particularly for high packing density. hiVISC shows up to η = 2.3 at very low Re (< 1), which increases as Re increases. Trends show η may reduce as Re > 100. Both sets of inserts show advantages over commonly used helical type inserts often used for viscous heat transfer enhancement.

Mixing efficiency

As with heat transfer efficiency, an efficiency for mixing can be calculated to compare insert performance. By computing the coefficient of variance (CoV) for images obtained using the PLIFR, an indication of mixedness can be evaluated against friction factor f. Mixing efficiency, λ, can be defined as:

λ = CoV f ^-1/3

Tested inserts resulted in high values for λ, indicating an efficient mixing capability. This, however, corresponded to a high overall CoV, i.e. relatively poor mixing in comparison to other commercially available static mixers. Both hiTRAN and hiVISC act as low-intensity mixers, for which the intensity can be adjusted by varying packing density and pitch, respectively.

Conclusion

Using a variety of experimental methods, we characterised two types of tubular enhancement device (hiTRAN and hiVISC) in terms of pressure drop, heat transfer enhancement and mixing capability for high viscosity applications. Results showed both sets of inserts could deliver efficient enhancement, where pressure drop penalty was a fraction of the heat transfer enhancement. Applying these devices in flow reactors, viscous heat exchangers and other units can provide significant benefits in reducing costs, footprint, energy usage and carbon emissions.