Internal Gas–Liquid Separation in Industrial Ebullated Bed Hydroprocessors

Abstract

Ebullated bed reactors used in heavy oil upgrading internally recycle up to 90% of net liquid flow to maintain fluidized conditions. Internal phase separators result in significant gas re-entrainment within the recycled liquid, leading to high gas holdup (40 to 60%), reduced processing capacity, and over-cracking of vapor products. Phase separation in two commercial designs is assessed using computational fluid dynamics and compared to pilot-scale results. Liquid short-circuiting within first-generation recycle pan designs was identified as a key factor, leading to a reduced separation efficiency of 1 to 2 mm bubbles compared to second-generation flow-through designs. Negligible performance improvements were observed for <1 mm bubbles, highlighting the need to exercise caution when scaling these designs to hydroprocessing conditions where submillimeter bubble sizes are predominant. A practical correlation for gas separation efficiency based on the liquid residence time is proposed and compared to numerical and experimental data.

Fluid Dynamics Modeling of a Commercial Ebullated Bed Hydroprocessor

“…Since bubbles do not readily coalesce in the bed, 9 the original bubble size distribution generated at the grid likely impacts gas−liquid separation at the recycle pan and gas entrained in the liquid recycle increases bed gas holdup at the expense of liquid holdup and product yield. Previous studies on ebullated bed hydroprocessors focused on the scaling of bed fluid dynamics, 10−13 simulation of gas− liquid separation at the recycle pan, 14,15 modeling of overall reactor fluid dynamics, 16−19 and the impact of gas−liquid distribution system on bubble dynamics and properties. 20−22 Building on the works of Safoniuk et al 10 and Macchi et al, 11,12 Pjontek et al 13 proposed a scaling approach for bed fluid dynamics.…”

“…Lane et al 15 later simulated the performance of the second generation recycle pan (flow through pan) by leveraging previously developed model. 14 An improved separation performance relative to the two stage recycle cups was reported for the flow through pan.…”

“…Previous studies on ebullated bed hydroprocessors focused on the scaling of bed fluid dynamics, − simulation of gas–liquid separation at the recycle pan, , modeling of overall reactor fluid dynamics, − and the impact of gas–liquid distribution system on bubble dynamics and properties. − Building on the works of Safoniuk et al and Macchi et al, , Pjontek et al proposed a scaling approach for bed fluid dynamics. The proposed scaling required matching of five dimensionless groups ( U g / U l , ρ g / ρ l , ρ l / ρ s , Ar L – S , and Re L – S ) and bubble coalescence behavior.…”

“…Lane et al . then incorporated their previously developed separation models , into an overall fluid dynamics model of the LC-Fining unit. As the bubble size information was not available at the time, an initial guess of the mean size was made to predict gas–liquid separation efficiency and subsequent iterations were necessary to determine the bubble size that provided the best fit to commercial operating data for the freeboard gas holdup.…”

“…Bubble size distribution and drag models proposed in the study of Mach et al will be used to predict phase holdups in the unit. Further, the size distribution model will be coupled with the gas–liquid separation model proposed by Lane et al to better predict recycled gas and liquid flow rates and, consequently, phase holdups. Two new numerical frameworks using mass of the catalyst inventory or the recycle pump curve as inputs to determine the bed height (criterion for convergence) are described in the next section.…”