PTQ Q4 2023 Issue

900

Competitor A Competitor B Dynocel 680

1-Hexene Q-value Water Q-value

800

700

600

500

400

300

200

100

Dynocel 680 Competitor A

100

150

Pure alumina

Pure 13X MS

200

250

300

Temperature (˚C)

Figure 1: Left Propylene reactivity tests for Evonik and Competitor co-formed adsorbents; Right: Heat release measurement for different adsorbent classes

detected by a drop in pressure during the temperature hold step, liquid formation due to oligomerisation of the olefin, and/or adsorbent colouration. Reactivity tests have been conducted on different co-formed products and with several olefinic streams. Reactivity tests were performed in harsh conditions (high temperature, high pressure, and sealed vessel). They, therefore, do not represent ‘normal’ operating conditions but rather show the difference in reactivity of the different products and what can happen in case of upset. Results obtained for propylene are shown in the left section of Figure 1 . While Evonik and Competitor A products showed low/no reactivity, pressure recorded during Competitor B product tests drastically dropped at around 200°C (pressure: 600 psig – 41 barg). This indicates that Competitor B product had reacted with propylene during the test. The yellow colour of the unloaded product confirmed the presence of oligomers at the adsorbent surface. As a reminder, oligomerisation is an exothermic reaction that can, on an industrial scale, generate an uncontrollable chain reaction with large heat release.  Heat release: Heat release (exothermic response) is measured with the Q value, which is recorded as the amount of heat generated by the adsorption of a compound at the surface of a solid. The lower the value, the safer the process. Q values have been measured for the adsorption of water and hexene at the surface of several classes of adsorbent (see Figure 1, right ). Alumina-based adsorbent showed little heat release dur- ing the test, whereas pure large pore diameter molecular sieves (13X type) showed a massive heat release, explaining the need for a pre-load step when used in olefinic processes. As expected, co-formed products showed Q values between pure alumina and pure molecular sieves data. However, Competitor A co-formed product still exhibited a fairly high Q value, justifying the need for a pre-load step for the latter. Conversely, Dynocel 680 hybrid adsorbent showed a lower heat release, indicating that a pre-load step is not needed. These pilot tests confirmed that Evonik’s hybrid adsorbent exhibits both low reactivity towards unsaturated streams and low exothermic response. These important

features have been further verified at an industrial level, as demonstrated in the following case study with PCK Refinery in Schwedt/Oder, Germany. Adsorbent application in PCK Refinery With 240 000 barrels per day production, PCK Refinery sup - plies 90% of the Berlin and Brandenburg areas with gaso- line, jet fuel, diesel, and fuel oil. PCK Refinery operates a C₃ dehydration unit in its FCC complex. The purpose of the unit is to dry the highly olefinic C₃ cut (propylene/propane: 80/20) to below 1 ppmw water and to remove traces of ammonia to below 15 ppbw. The C₃ dehydration unit consists of two adsorbers, with one in adsorption service (liquid phase) and the other in regeneration mode (gas phase). Unlike typically seen in this type of unit, the regeneration media for the dryer is not an inert gas but the effluent from the other adsorber in adsorp - tion service. Therefore, the regeneration media contains 80 wt% propylene, which makes safety and reliability focal points in the operation. Historical concern Always looking for the most efficient and safe process, PCK’s selection of adsorbent and internals has evolved with time, as shown in Figures 2 and 3 . Figure 2 schematises the dif- ferent loading plans from unit installation in 1994 until today, while Figure 3 shows the temperature profile recorded dur - ing the first and following regenerations for the different adsorbent used. Before 2007 , t he dehydration unit was initially filled with a Type 3A molecular sieve only. As Figure 3a shows, using 3A-type molecular sieves only, the regeneration outlet tem- perature always remained lower than the inlet temperature, and the curve shape did not differ between the first and later cycles. This confirmed that low pore diameter molecular sieves have no intrinsic reactivity towards propylene. 2007-2013 : In 2007, with the need for NH₃ removal, a layer of hybrid adsorbent (Hybrid 1 from Competitor B) was chosen to be added to the dryers. During the first regener - ation, a temperature peak was observed in the treater. This

PTQ Q4 2023