MechChem Africa February 2019

⎪ SAIChE IChemE news ⎪

“The critical number for a fuel blending product, however, is its research octane number (RON), which governs the fuel’s suit- ability for use in different modern engines,” says Prinsloo, adding that Sasol’s commercial CTLprocesses strive toproduce fuel products with specific target RONs. “The catalysts are critical for achieving this. They are critical for speeding up the re- action and making it more efficient. But they operate at high temperatures and they coke up, so they need to be routinely regenerated to keep them working effectively. Typically, the catalyst will deactivate within a week under the process conditions in a laboratory reactor,” he says. “The naphtha reforming technology has evolved from a simple fixed bed reactor sys- tem, where the catalyst has to be removed to be regenerated externally, to a highly sophisticated system where the catalyst is continuouslymoved through the reactor into the regenerator and back,” Prinsloo explains. In the regenerator, the catalyst is purged to remove hydrocarbons and hydrogen, the coke is burned off the catalyst before it is re- chlorinated and reduced. “Thewhole process is known as continuous catalytic regenera- tion (CCR) and it is currently used in Sasol’s modern plants in the manufacture of our high-octane fuel blending products,” he says. He adds that the paraffinic feed used in the Sasol process is particularly lean, which requires a much higher temperature to pro- duce the specific octane number required. This lowers product yield and causes catalyst to degenerate far more quickly. “Pushing catalyst andprocessingboundar- ies has always been imperative at Sasol. This has become especially true when targeting reformatewith a RONabove 95while retain- ing an acceptable yield and keeping within commercial fuel specifications, particularly with regard to benzene. The bench-scale challenge Akey role of Sasol’s Research andTechnology team is to seek out new and better catalysts for use in its CTL plants. “We routinely per- form bench-scale tests of catalysts for use in onour FT refinery processes using conditions that are as close as possible to those in our production units,” Prinsloo continues. But doing this manually presents chal- lenges. “First, we are looking to see the effect of the catalyst on the yieldof the targeted fuel anditsRONnumber.Ideally,thewholesample collected needs to have the same RON, but this can be difficult to achieve because the catalyst is slowly deactivating as the test proceeds, sometimes faster than the sample can be analysed,” he says. “We also need to collect a large enough sample of the reformate to do engine analy-

Sasol’s Research and Technology team used a laboratory-scale Zeton catalyst testing system as the basis for its automated bench-scale control system for evaluating reforming catalysts.

ses, which can take a week for each sample,” he adds. The process involves several endothermic and exothermic reactions and temperature swings introduce significant variation in the productcomposition.“Thetemperatureofthe catalyticbeddirectlyaffects theRONnumber and, as well as keeping the temperature as uniform as possible, a gradual increase in bed temperature can be used to counteract the deactivation process,” Prinsloo tells MechChem Africa . Sasol’s new bench scale solution “The control system we developed can compare very small differences in the refor- mate yield when using different commercial catalysts,” he says. “By usingmultidisciplinary analytical and advanced process control techniques, we have achieved substantial im- provementinbench-scalenaphthareforming.” Summarising Sasol’s new bench process, he says that the RON of the product is con- tinuously measured while the temperature and the heat flow over the catalyst bed is accurately controlled. “We slowly ramp up the catalyst bed temperature based on real- time RON measurements of the reformate. In addition, the control loop maintains an exceptionally uniform bed temperature with far less heat flux drift across the reactor zone,” he says. The core innovation involved in achieving this was developing a way of measuring the RON in real time so that it couldbeused in the control loop to automate the whole process. “Octane numbers of reformate samples are measured in a number of ways. The most accurate of these is to use a calibrated CFR engine to determine a motor octane number (MON). But 500 to 1 000 ml samples are required before the test can begin, which takes at least aweek to reformat bench scale, so using this test routinely is impractical for research. “An alternative is to do detailed composi- tional analyses and spectroscopy, but special chromatographic systems have to be used

and, althoughused inmany large laboratories, the equipment requires specialised knowl- edge,isveryexpensiveandtheteststaketime. “Wehavedevelopeda cost-effective alter- native that canapproximate theRONnumber using simple gas chromatography (GC) for component separation, combined with more sophisticated calculations and modelling to arrive at an estimated RON and MON. These values are then occasionally calibrated against more accurate CFR engine tests. “None of these methods were suitable for real time use in a feedback-based control system, however,” Prinsloo says. “We therefore pursued the use of near- infrared (NIR) spectrometry. We used our historical GC RON estimates to calibrate a real time on-line NIR analyser and set it up to continuously monitor the reformate stream from the bench-test system. By adopting error-based feedback control between the output reformate octane number and the preset required RON, we are able to slowly adjust the bed temperature upwards to keep the activity level of the catalyst at the point where it produces a reformatewith the exact RON required,” Prinsloo explains. “This system has enabled us to generate 1 900 equivalent octane number analyses that were corroborated via only 80 CFR engine tests, which translates into a more than 20-fold reduction in the testing load for laboratory octane number determination. “And this has already benefited our pro- duction units. By identifying catalysts better suited to our processes, we are achieving significantly better yields while lowering catalytic reforming demands and associated costs,” he concludes. A project worthy of winning any innova- tion award. q Acknowledgements: This work has been shared with the international community via a keynote lecture byNico Prinsloo at the 2016CATSA con- ference and a paper published in the American Chemical Society’s Industrial Engineering and Chemistry Research journal of May 2017.

February 2019 • MechChem Africa ¦ 7