In our pilot work, the reactor bed's axial temperature delta has ranged from 8 degrees C to 31 degrees C, depending on feed rate, inlet temperature, and catalyst age. This is consequential: at the lower end, C10 to C14 selectivity is in the range we want for jet-cut fractions. At the higher end, products shift toward lighter olefins and heavier wax-precursor chains, reducing jet yield and increasing downstream fractionation load. We measured this distribution systematically across 18 pilot runs over roughly 14 months at our Zurich facility.
We are writing this because academic literature well documents the link between axial temperature profiles and selectivity in general olefin oligomerization, but discusses it less for methanol-to-jet. The MtJ feedstream, especially the olefin distribution leaving the MTO stage, interacts with thermal gradients in ways that models built for traditional naphtha or Fischer-Tropsch feeds do not fully capture.
Why Temperature Non-Uniformity Matters in an MtJ Reactor Bed
Olefin oligomerization over acid zeolite catalysts releases heat. Shorter olefins join into longer chains, but the heat release varies along the bed because reactant concentration is highest at the inlet. In an adiabatic fixed-bed reactor, the catalyst bed top therefore runs hotter than the bottom, sometimes substantially so, depending on inlet olefin concentration and space velocity.
Selectivity is temperature-sensitive because reaction pathways compete. At moderate temperatures, roughly 220 to 280 degrees C for our catalyst formulations, sequential olefin addition dominates, producing C8 to C16 chains with branching patterns favorable for kerosene. As temperature rises, cracking competes more strongly and shifts products toward lighter fractions. Above approximately 320 degrees C, coke forms significantly faster and catalyst deactivation accelerates.
These temperature thresholds are not fixed cutoffs. They vary with catalyst age, feed composition, and zeolite formulation. In practice, selectivity drifts gradually as the axial profile changes during a run, whether upper-bed deactivation is responsible or inlet olefin concentration changes in the upstream MTO stage.
How We Measure Bed Temperature Distribution
Our pilot reactor has a multi-point thermocouple assembly with fixed positions at seven axial locations through the catalyst bed, plus two sensors in the inter-bed quench zone. Radial measurement is limited at this scale, so we assume reasonable radial uniformity. That assumption holds reasonably well for a well-packed bed at pilot diameter, but needs verification at larger diameter.
Alongside the fixed thermocouples, we logged inlet and outlet gas composition with online GC at approximately 15-minute intervals throughout all 18 pilot runs. Continuous temperature records combined with periodic composition data let us track how temperature profile and product selectivity co-evolve during each run.
Our process control platform ingests temperature data in real time. Historical run data then enters the optimization layer's dataset for recommending conditions in subsequent runs.
What 18 Pilot Runs Showed
Across 18 pilot runs, we saw three distinct temperature profiles: flat, front-loaded, and progressive.
Flat profiles appear with moderate feed rates, fresh catalyst, and mid-range inlet olefin concentration within our operating window. The axial delta is typically 8 to 14 degrees C. Jet-range selectivity, the C8 to C16 product fraction, consistently reaches our target of 68 to 74 percent of hydrocarbon product by mass in internal pilot campaign data. Downstream fractionation is straightforward in these runs, and yield per unit feedstock is highest.
Front-loaded profiles arise when inlet olefin concentration reaches the upper end of our operating window or space velocity is raised for throughput. The top 20 to 30 percent of the bed runs substantially hotter, producing axial deltas of 22 to 31 degrees C. Under these conditions, internal benchmarks show jet-range selectivity falling to approximately 58 to 63 percent by mass, with more lighter fractions routed to naphtha or off-gas instead of jet.
Progressive profiles indicate partially deactivated catalyst and typically appear in the back half of catalyst life. As upper-bed activity declines, the reaction front moves lower and the temperature peak shifts downstream. Real-time diagnosis is harder because outlet temperature may look acceptable while the internal distribution turns unfavorable.
The Hotspot Problem and Carbon Chain Length Distribution
The front-loaded case presents the clearest controllable yield loss. If the upper bed zone stays at 300 degrees C or above for extended periods, longer-chain intermediates crack faster, increasing the lighter fraction at the expense of jet-range material. Coke also deposits faster on acid sites in that zone, shortening effective catalyst life for the whole bed.
An early-run surprise was that elevated temperature does not affect both sides of the jet cut equally. We lose more from C12 to C16 than from C8 to C10 because heavier jet-range components preferentially crack back to C6 to C8. The product therefore moves toward the lighter end of the ASTM D7566 Annex A7 distillation envelope, which can complicate blending with conventional Jet-A when the blend partner has already displaced much of the lighter fraction.
We are not saying high-temperature operation is never acceptable; we are saying it requires an explicit yield tradeoff. Throughput needs can justify a front-loaded profile, but ignoring its selectivity effect creates losses that look like catalyst problems even when operating conditions are responsible.
Turning Profile Data Into Operating Recommendations
Tobias Maurer and the platform team have been converting temperature-profile patterns into operating recommendations. Thermocouple data arrives in real time, but the link between the current profile and expected end-of-run yield depends on catalyst lifetime, upstream MTO output, and planned run duration.
Across 18 runs, we gathered enough data for a working model of the flat-to-front-loaded transition. It uses inlet olefin concentration, upper-bed temperature, and catalyst age to estimate whether the profile will stay stable or drift front-loaded within the next several hours. Above a defined probability threshold, the system recommends lowering feed rate or adjusting inter-bed quench flow, depending on the more direct root-cause response.
This is incremental, not a fully closed-loop thermal control system. Still, internal pilot benchmarks show a consistent, measurable improvement of roughly 4 to 6 percentage points in run-average jet selectivity when profile recommendations were applied versus earlier runs without them, on a mass basis. That gain adds directly to yield per campaign.
What This Means for Scale-Up
The pilot reactor used for this dataset has a bed diameter small enough that radial temperature gradients are not significant. At commercial scale, that assumption fails. In a 1.5 to 2 meter diameter bed, center-to-wall temperature differences will interact with the axial gradient, creating two-dimensional non-uniformity. Its selectivity effects will be more complex than those observed in our pilot data.
We cannot model commercial-scale thermal behavior from our current dataset. That is an honest limitation. The pilot data does provide a well-characterized view of axial profile dynamics and their selectivity effects, which is needed for scale-up. Our measurement architecture, profile classification, and profile-drift prediction model are designed to extend to larger reactors as geometry and thermocouple arrangement are adapted.
Scale does not alter the link between temperature non-uniformity and selectivity. For a facility running the MtJ process, pilot-scale understanding makes scale-up an engineering exercise, not a discovery exercise.