When the heat is on

Reactors with high heat flux capacity have been shown to yield cost and productivity benefits for certain classes of chemical batch reactions. Jose P Arencibia, II, of Kelvin Cryosystems, outlines a novel design technology

A new generation of reactors with high heat flux capacity can control exothermic batch reactions over a wide range of temperatures and sizes. By novel design of the heat transfer surfaces, high cooling loads and rates can be matched to any operating point with low radial temperature gradients and high space-time yields. Significant improvements to operating costs and productivity have been demonstrated. Furthermore, the ability to operate at low absolute temperatures reduces the evolution of VOCs and toxic vapours.

There are two broad classes of chemical batch reactions able to benefit from these techniques: highly exothermic reactions and those requiring operating temperatures lower than conventional refrigeration temperatures.

Unlike, say, heat exchangers, people rarely actually design a reactor. They frequently specify what is required of the vessel and trust that the standard configuration will deliver the required outcome. Conventional designs of low temperature reactors (LTR) are based on a modified standard stirred tank reactor with external half pipe heat transfer surface and possibly an internal coil.

One of the key design elements for a reactor is the radial temperature gradient, measured between the centreline and the wall of the vessel. A 10°C radial temperature gradient equates approximately to a rate of reaction at the centreline double that at the wall. In most applications this temperature gradient does not matter. However, for highly exothermic reactions and ones with critical side reactions this could be the difference between success and failure of the process.

effective reactions

By addressing the fundamentals of reactor/heat transfer design in the Kelvin LTR, significant benefits can be gained over conventional reactors and first generation LTRs. This results in cooling rates five to ten times faster than a first generation LTR, higher space-time yields (a measure of productivity/yield/quality) and radial temperature gradients as low as 1°C.

The space-time yield (Y) can be considered as the ultimate measure of effectiveness of a given chemical reaction. Y, representing the amount of the desired chemical species produced for a given reaction volume per unit time, is a measure of the efficiency of conversion and hence an indicator of profitability.

For most batch reactions an initial amount of a solvent is charged to the

vessel. The temperature of the initial charge is adjusted to the start temperature before commencing the controlled addition of the reactive ingredient. The rate of addition is limited by the cooling capacity of the vessel in order to maintain the desired reaction temperature. However, a Kelvin LTR can maintain the reaction mass at any desired temperature between -195°C and +120°C, chosen to optimise the chemistry, not constrained by the vessel heat transfer capacity.

conventional cooling

A conventional reactor system (Figure 1) is cooled by circulating an appropriate fluid through the external jacket and possibly an internal coil. Typically the heat transfer fluid is a proprietary material such as Syltherm, relying on sensible heat transfer. These fluids have been shown to be effective at high temperatures. However, low temperature applications are difficult if not impossible.

Substitution of the sensible heat transfer fluid by a cryogenic one is the basis of the first generation LTR design. The fluid most commonly chosen is liquid nitrogen (LN), having a boiling range of -195°C to -178°C by modifying the pressure.

To understand why a conventional reactor design fails when using cryogenic services it is necessary to understand how the geometry affects the heat removed from the reaction mass. LN passing through the external half pipe jacket (and equally any internal coil) boils. As soon as boiling starts, the nitrogen expands volumetrically with the production of gaseous nitrogen. With the onset of boiling, the flow regime in the heat transfer surfaces passes through a range of transient states, single phase, two phase, slug flow and mixed slug/annular flow.

These transient flow regimes are unpredictable and uncontrollable, giving rise to variable heat transfer coefficients from very high to very low. The result of this unstable flow regime is local hot and cold (freezing) spots and unpredictable reaction mass temperature gradients. This flow regime results in two-phase flow from the cooling circuit vent with alternating gas and liquid emissions. The consequences of this are reduced product yield with impaired quality, high operating costs and end user dissatisfaction.

Consider a real system designed to meet the customer requirements set out in table 1. To meet the customer requirement with a conventional (first generation LTR) design, there is insufficient heat transfer surface available in an external jacket, therefore internal coils will be needed. Even with the internal coil the cooling heat load constraint of 138 kW cannot be met.

For structural integrity the minimum pipe size for the coil is 50mm, resulting in a large number of spacers and supports. This affects the vessel cleaning, conflicting with the pharmaceutical application constraint. In addition the coil adversely affects the radial temperature gradient.

It is clear from this assessment that a conventional first generation LTR design will not meet the customer requirement.

The second generation LTR reactor addresses and solves all of the above issues. This is achieved by separating the phase change of the liquid nitrogen from the sensible heating of the gas phase. To ensure the LN will boil efficiently, the evolved gas must be at sufficiently low velocity to avoid any remaining liquid being carried over as droplets.

In the Kelvin Cryosystems second generation LTR reactor (Figure 2), the LN boils inside specially designed mixing baffles immersed in the bulk of the reaction mix. The sensible heat load is transferred to the external jacket coil.

For the customer requirement above, four mixing baffles are used in parallel to provide cooling to the batch. For this particular application each baffle is 150mm schedule10 pipe with a total cross sectional area equivalent to 0.082m². The vaporised LN from the baffles is combined together and piped directly to orthogonal cross section external jacket. This geometry provides 27% additional area for flow compared with a standard configuration.

In the design of the Kelvin LTR system heat transfer in the reactor is structured so that phase change without LN carryover occurs inside the mixing baffles. The mechanical design of the mixing baffles is critical in the effectiveness of LN/vapour separation. The supply of cooling to the reaction mass can be predicted and hence controlled. This gives high heat transfer coefficients in the mixing baffles of the order of 3000 to 6000W/m² K. The equivalent HTC in the jacket is 300-600W/m² K.

The 8,000l reactor supplied to meet the customer requirement above has proved to provide more than 138kW of cooling capacity with cooling rates in excess of 1°C/min and a radial temperature gradient of less than 2°C. The comparative performance of the Kelvin LTR and the first generation LTR it replaced, with the Kelvin LTR as the base, is shown in table 2.

reliable and reproducible

The design of a Kelvin LTR allows reliable and reproducible scaling of reactor size from the laboratory unit at 1.0l to 12,000l capacity. The availability of small-scale equipment is critical in the take-up of new techniques. Standard laboratory/pilot equipment cannot reproduce the operating envelope described above, preventing the development chemist from exploring new and exciting areas.

It has been shown that a slightly simpler geometry for the laboratory/pilot size units produces data that can be used to reliably scale the desired process to the required production scale.

There are working full-size units in the US, Ireland and the UK, some of which have been operating successfully for more than four years. Clients have often replaced first generation LTRs with Kelvin LTR, as is the case with two of the three units installed in Europe.