May-2024

Flexible downscaling of MAPD removal from C3/C4 olefin streams

This study simulates selective hydrogenation of a combined C3/C4 cut with 1,3-butadiene, aiming to retain the target products under relevant industrial conditions.

Edgar Jordan, Charlotte Fritsch and Joachim Haertlé
hte GmbH

Article Summary

Raw olefinic C3/C4 streams in integrated petrochemical refineries, mainly the product of naphtha steam crackers or FCC units, usually contain low-weight percentages of acetylene derivatives, known as MAPD compounds. MAPD stands for methylacetylene (propyne, MA) and propadiene (PD), which are the most common impurities; vinylacetylene (but-1-en-3-yne, VA) and ethylacetylene (1-butyne, EA) can also be formed. These side products must be removed to protect downstream processes from severe catalyst poisoning and unwanted side reactions of those highly reactive substances. For instance, polymer-grade propylene must not exceed an MAPD impurity level of 1 ppmw.

The predominant method for removing these compounds is selective hydrogenation of the triple bonds while preserving the olefinic species. Whereas selective hydrogenation of acetylene in the C2 fraction is performed in the gaseous phase, C3/C4 selective hydrogenation is predominantly carried out in a liquid state. This is for reasons of energy efficiency and the smaller reactor volume.

Another benefit of a liquid state reaction is that greenoil formed as a byproduct during the hydrogenation reaction is washed out of the catalyst bed. Greenoil is an unwanted side product in C3-C4 selective hydrogenation, formed by the oligomerisation of olefins to various heavier unsaturated hydrocarbons. This greenoil can lead to catalyst deactivation as well as contamination in the unit.

Commercial selective hydrogenation processes can be differentiated into front-end and tail-end configurations. Front-end hydrogenation processes are characterised by an excess of hydrogen (H2) present in the raw product stream from the steam cracking or FCC unit. In contrast, tail-end selective hydrogenation, situated behind the de-methaniser and de-ethaniser columns, is controlled by feeding a stoichiometric amount of hydrogen to the hydrocarbon stream. Whereas for front-end hydrogenation, NiCo catalysts are common, tail-end hydrogenation catalysts are precious metals, mainly based on promoted palladium (Pd). Commercial process licensors and catalyst manufacturers for selective hydrogenation include Axens, UOP, Evonik, Lummus, Linde, KBR, and BASF.

Tail-end liquid state selective hydrogenation often incorporates multistage operation. In the first stage especially, a recycle stream of the product can be integrated to control the acetylene concentration and, in turn, the exothermicity of the reaction.^1-4

For operators of tail-end selective hydrogenation processes, many aspects of the catalyst and process performance are important to investigate, such as the long-term stability of the catalyst, the acetylene derivatives conversion level as a function of feed composition, the optimal operating window of the process, and the control of greenoil formation.

Experimental set-up
The experimental unit was equipped with a flexible up-and-downflow configuration. Liquid feed was pumped by a two-piston syringe pump, whereas the hydrogen flow was controlled by a mass flow controller (see Figure 1). The pressure was controlled using a backpressure regulator situated after (liquid phase operation) or before (gas phase operation) the reactor. The resulting product mixture was diluted with N2 downstream, whereas any greenoil formed was kept liquid and thus separated from the product stream. The temperature of the catalyst bed was controlled by a heating/cooling mantle connected to a thermostat equipped with a parallel electrical heating circuit. The accessible temperatures ranged from sub-ambient to 600°C.

With this flexible set-up, it was possible to run the reaction at various pressures and temperatures in both liquid and gas phase. Although most commercial selective hydrogenations of these C3/C4 species are performed in the liquid phase, the experiment additionally compared liquid and gas phase performance to demonstrate the unit’s flexibility to accommodate different light olefin selective hydrogenation reactions. In liquid phase operation, the impact of upflow vs downflow configuration was investigated.

The hydrocarbon feedstock simulates a C3-C4 olefin stream. It should be noted that the feedstock contains 7.25 wt% 4-vinylcyclohexane as a dimerisation product of 1,3-butadiene, extending the educt spectrum up to C8.

The effluent stream was analysed with an Agilent gas chromatograph (GC) equipped with a flame ionisation detector (FID) and an RTX-Alumina BOND column. Conversion was calculated based on the weight and molar educt, and product flow rates, with the educt flow rates defined by repeated bypass measurements between the experiments.

Results and discussion
Operation under standard conditions
A commercial Lindlar catalyst was used (5 wt% Pd, selectively poisoned by lead (Pb) on a calcium carbonate carrier, Merck).5 The reactor with an inner diameter of 8 mm was filled with a 2:3 (by volume) mixture of the Lindlar catalyst and silicon carbide as inert material. The total catalyst zone length was 77 mm. Reference conditions for testing were a reactor temperature of 35°C, a pressure of 24 barg, a molar hydrogen/VA ratio of 2, a hydrocarbon feed-based liquid hourly space velocity (LHSV) of 22 h-1, and an upflow configuration of the reactor system.

This approximates a typical operating window of commercial units with a hydrogen dosage just below the stoichiometric amount with respect to all acetylenes and PD. These parameters were systematically varied during the testing program, as will be further discussed.

Under these bespoke standard conditions, the following conversions were observed: 78.6 wt% for VA, 49.4 wt% for EA, and 54.1 wt% for MA. The higher conversion of VA compared to MA and EA is characteristic of Pd-based catalysts. Hydrogen was quantitatively consumed, which is also typical for a tail-end selective hydrogenation process. However, containing 5 wt% Pd, the commercially available Lindlar catalyst provided a fivefold higher active metal load than a typical industrially available catalyst for this reaction and, hence, was significantly more active. This explains the high BD loss of 9.53 wt%.

LHSV variation
Starting from the standard conditions, the hydrocarbon-based LHSV varied in a broad range between 7.2 h-1 and 82.4 h-1. Due to the high activity of the Lindlar catalyst used, the hydrogen conversion remained at nearly 100 mol% over the whole LHSV range. At higher temperatures, a temperature increase of the catalyst zone could be detected, with the zone temperature ranging from 35.5°C at LHSV = 7.2 h-1 to 36.7°C at LHSV = 82.4 h-1. The heater/cooler system was able to keep the catalyst zone temperature constant within a narrow temperature range also at high rates of heat production by the hydrogenation reaction.