In a fixed-bed tubular reactor for orthoxylene oxidation to phthalic anhydride, the temperature maximum located in the first third of the catalyst bed determines catalyst lifetime, product distribution, and safe operating latitude more than any other measured plant variable. The process gas is formed by preheating filtered air to 170–200 °C and mixing it with vaporised orthoxylene at a controlled loading typically reported between 60 g Nm−3 and 80 g Nm−3 at normal gas flow. The resulting mixture enters a tube bundle containing 10,000–30,000 vertical tubes, each with an inner diameter in the range of 21–26 mm and a tube length of 3,000–4,000 mm, packed with a vanadium pentoxide–titania anatase catalyst shaped as rings or lobed extrudates. Selective oxidation to phthalic anhydride releases approximately −1.11 MJ mol−1; total combustion to carbon oxides and water releases roughly −4.38 MJ mol−1. Because both reactions occur on the same catalytic surface, any local temperature increase accelerates the more exothermic combustion pathways more strongly than the desired partial oxidation, producing a self-reinforcing hot spot unless heat removal matches the local generation rate. The outer wall of each tube is cooled by a molten salt mixture composed of potassium nitrate, sodium nitrite, and sodium nitrate, maintained at a bath temperature commonly specified between 340 °C and 390 °C. Radial heat transfer inside the packed bed is the main thermal bottleneck: heat generated on the centreline must cross a stagnant gas film, the pellet surface, and a low-conductivity packed bed before reaching the tube wall. A temperature difference of 40–80 K between the hot spot and the coolant is common, and commercial catalyst suppliers usually set a maximum internal temperature limit of 420–450 °C because higher peaks trigger anatase-to-rutile sintering, potassium migration, and irreversible selectivity loss. The reactor pressure envelope is constructed to ASME BPVC Section VIII Division 1, and the tube bundle is fabricated from seamless heat exchanger tubing conforming to ASTM A213/A213M.

What Thermodynamic and Kinetic Constraints Convert a Benign Temperature Rise into a Runaway Hot Spot?

At the kinetic level, phthalic anhydride formation and total oxidation exhibit a measurable difference in apparent activation energy, with many supported vanadate catalysts giving selective oxidation apparent activation energies in the 80–130 kJ mol−1 band while the over-oxidation pathway consumes additional intermediate species. The immediate consequence is that selectivity to phthalic anhydride degrades as the bed temperature passes 420 °C; over-oxidation consumes the product itself via ring cleavage and decarbonylation, yielding maleic anhydride, citraconic anhydride, carbon monoxide, and carbon dioxide. Because the complete combustion of orthoxylene liberates approximately 3.3 MJ mol−1 more enthalpy than the selective oxidation, a small loss of selectivity produces a disproportionate increase in local heat release. The selective pathway depletes the adsorbed orthoxylene surface species, while total oxidation continues on sites that have become oxygen-rich and thermally activated. A hot spot therefore does not merely shift the selectivity curve; it changes the steady-state oxygen coverage on the vanadium oxide surface and can create a secondary heat-release zone where phthalic anhydride itself is oxidised. The temperature maximum is not stationary, because inhibition from adsorbed maleic anhydride or fouling in the inlet zone causes the hot spot to migrate downstream over campaign life. Operators monitor this migration and adjust bath temperature upward, but the kinetic asymmetry means that the peak temperature is not linearly proportional to the bath setpoint; above a threshold usually near 430–450 °C, the coupling between temperature and over-oxidation becomes self-amplifying and can exceed the cooling capacity of the coolant film.

The practical failure mode at production scale is not catastrophic decomposition of orthoxylene but progressive catalytic damage that shortens campaign life and shifts the selectivity–temperature operating curve. Fixed-bed units built with 25 mm inner diameter tubes and 3,000 mm length may run continuously for 24–36 months before catalyst replacement, but only if the peak bed temperature is held below the catalyst-specific ceiling for the entire campaign. Temperature control is typically achieved by cascading orthoxylene feed flow, air flow, and salt bath circulation rate; the salt bath temperature is not the only manipulated variable because gas-side temperature rise can be reduced by lowering o-xylene loading, increasing air flow, or adjusting the coolant inlet temperature. Batch-to-batch variance in catalyst ring density, anatase crystallite size, and promoter distribution is a known contributor to hot spot variability; sieving and lot-audit procedures are required because a 0.1 g cm−3 change in apparent pellet density can shift local pressure drop and gas bypass. Tube-to-tube flow maldistribution caused by differences in packed-bed void fraction of ±2% is sufficient to create individual tube temperatures that deviate by more than 10 K from the bundle average. Feed-to-bundle distribution is therefore checked with hot-wire anemometry or differential pressure surveys during shutdown; a tube with restricted gas flow can starve the catalyst and simultaneously raise the local temperature, producing a high-temperature channel in the tube sheet region. The hot spot response is lagged and non-minimum-phase: a step increase in o-xylene loading initially raises conversion and heat release near the inlet before the coolant system can remove the additional duty, so operators typically impose loading changes at rates no greater than 5–10 g Nm−3 min−1.

Molten Salt Bath Hydrodynamics and Radial Heat Flux Limitations

On the shell side of the fixed bed reactor, the molten salt bath functions as both heat transfer medium and thermal flywheel. A common ternary salt is the Hitec-type mixture of 53 mass% KNO3, 40 mass% NaNO2, and 7 mass% NaNO3, with a melting point near 142 °C and a maximum film temperature limit of about 450–480 °C. Salt is circulated through the tube bundle by axial flow pumps or gas-lift circulators; the objective is to provide sufficient cross-flow velocity across each tube to produce a coolant-side heat transfer coefficient of the order of 1,000–2,000 W m−2 K−1, which exceeds the gas-side wall film coefficient by a factor of 10–30. The resistance to heat transfer therefore resides almost entirely in the process gas film and the packed bed itself. At typical Reynolds numbers for gas inside a 25 mm tube of 5,000–20,000, the wall film heat transfer coefficient may be only 100–250 W m−2 K−1. The effective radial conductivity of the packed bed, which includes static conduction, gas-phase conduction, and radiation, commonly falls in the range 0.4–1.0 W m−1 K−1 for ceramic catalyst rings at 350–420 °C. The radial Peclet number for heat in packed beds is low enough that a hot spot at the tube centreline remains separated from the wall by a radial temperature difference of 15–40 K even under stable operation.

Salt bath degradation introduces additional constraints on hot spot control. Prolonged exposure to temperatures above 460 °C accelerates nitrite oxidation to nitrate and water, changing the melting point and reducing heat transfer density. Moisture ingress into the salt from heat tracing leaks or from combustion water is a known operational hazard; steam can cause localised boiling and salt carryover into vent headers without forming sodium metal, but the resulting solidification in stagnant zones is sufficient to block circulation channels. To prevent solidification in stagnant zones, the salt bath is maintained above 180 °C during shutdown, and air lance sparging is used before restarting circulation. Salt bath temperature is controlled by a cascade loop that monitors the highest of six to twelve axial thermocouples embedded in selected reactor tubes; the setpoint of the bath heater is modulated to keep the maximum internal temperature below the catalyst ceiling. The measurement elements are Type K thermocouples conforming to IEC 60584-1, and the thermowells are spaced along the centreline and near-wall positions to capture radial temperature asymmetry. In some plants, the salt bath is cooled by submerged U-tube exchangers that generate medium-pressure steam; the total heat recovery duty across a world-scale reactor is substantial, and the bath circulation system is configured with redundant pumps because loss of circulation can allow the hot spot to grow at a rate of several kelvin per minute in the first metre of the bed.

Feed composition control directly modifies the heat generation profile and is therefore treated as part of the hot spot control strategy. The pre-reactor mixing section must ensure that orthoxylene vapour is completely homogenised with air before the tube sheet; stratified orthoxylene pockets produce a distribution of inlet concentrations that can push individual tubes into high-temperature excursions even when the average concentration is stable. Static mixers, evaporation chambers, and heated transfer lines are used to avoid condensation. The o-xylene dew point at 60–80 g Nm−3 and near-atmospheric pressure is below 80 °C, but transfer lines are operated at 100–150 °C to provide margin. Air is filtered and dried below 0.1 g water kg−1 because moisture competes with o-xylene on the vanadium oxide surface and influences acid-site density. The concentration controller often receives a feed-forward signal from reactor pressure drop and off-gas oxygen measurement rather than relying solely on mass flow ratio. When off-gas oxygen falls below 2 vol%, the risk of local oxygen depletion and runaway reduction of vanadium oxide increases; the standard countermeasure is to reduce o-xylene flow while maintaining air flow. Impurities in the orthoxylene feed, particularly sulfur compounds and organonitrogen species, are controlled below 10 mg kg−1 because they deactivate the active sites and shift the hot spot profile downstream. Avoidance of amine-based additives in the feed system is mandatory because basic nitrogen compounds alter vanadium oxide surface acidity and promote local overheating by changing the ratio of selective to total oxidation.

When Catalyst Beds Are Axially Diluted with Inert Alumina Rings

Axial dilution is a standard industrial strategy for matching local heat generation to local heat removal without changing the shell-side coolant temperature. The first 20–50 cm of the bed is often loaded with inert alpha-alumina rings or rings with very low catalytic activity to preheat the feed and flatten the inlet reaction spike. In a common configuration, the bed is divided into three zones: an inert preheat section, a 1:1 volumetric dilution zone, and a fully active rear zone. The dilution ratio is expressed as the volume fraction of active catalyst to total ring volume; reducing this ratio from 1.0 to 0.5 in the hot spot region can reduce the peak temperature by 15–30 K and shift the peak about 10–30 cm downstream. The penalty is that the residence time required for the same conversion increases or the o-xylene loading must be reduced, which lowers the mass of phthalic anhydride produced per tube per day. Dilution with low-surface-area alpha-alumina also increases the bed thermal conductivity slightly because the inert rings are denser and have a higher solid thermal conductivity than porous titania-supported catalyst rings. Pressure drop across the bed is lowered by the use of larger inert rings, but gas bypass at the wall may increase if the tube-to-pellet diameter ratio is outside the range 8:1–12:1. Catalyst pellet shape contributes to hot spot management: Raschig rings, wagon-wheel shapes, and simple hollow cylinders offer lower pressure drop and higher geometric surface area than spheres; the internal hole improves intraparticle heat transfer and reduces the effective reaction rate per unit bed volume by lowering the active mass fraction. The maximum hot spot temperature is therefore a function of both dilution profile and pellet geometry.

During the first 500–1,500 hours of operation, the hot spot position shifts as a consequence of selective vanadium oxidation state redistribution and blockage of the most active sites in the inlet zone by coke or adsorbed heavy aromatic residues. Reactor operators compensate by raising the molten salt bath temperature in small increments of 0.5–2.0 °C per month while continuously comparing off-gas oxygen and phthalic anhydride product acidity. When the bath temperature reaches the catalyst end-of-run ceiling, usually 390–410 °C for an extended campaign, the tube bundle is isolated, cooled, and unloaded. A tube with inadequate hot spot control can show a radial temperature difference greater than 60 K during early operation, and such tubes are frequently identified by infrared thermography during shutdown before they fail a pressure test. The thermowell data from these tubes is used to revise the dilution profile for the next loading; this is a site-specific iterative procedure rather than a single design fix. Published data for specific commercial dilution schedules is limited because loading patterns are proprietary, but the underlying trend is consistent: hot spot peak temperature falls with decreasing active fraction in the first metre, while the axial conversion profile becomes broader and the salt bath setpoint must be raised to maintain end-of-run conversion.