Optimization of a Concentric-Ring Rotating Packed Bed for Enhanced Offshore Natural Gas Dehydration

Abstract

Facing the harsh offshore environment characterized by severe space constraints and continuous platform motion, this study develops an optimized rotating packed bed (RPB) for compact and robust triethylene glycol dehydration. Through integrated experimental and computational investigation, the concentric-ring rotor was identified as superior among four configurations, consistently achieving dehydration equilibrium above 80% under lean TEG conditions. CFD analysis revealed its fundamental mechanism: synergistic matching between the centrifugal force field and annular flow paths yields the most uniform liquid distribution. This enabled the establishment of a strong predictive correlation (R² = 0.935) between simulated liquid uniformity and experimental dehydration performance. Guided by flow field diagnostics, targeted structural optimizations increased dehydration equilibrium from 86.1% to 92.25% while reducing system pressure drop by 73%. Parametric studies defined an optimal operating envelope at a gas-to-liquid ratio of 60:1 and system pressure of 2 MPa, achieving peak efficiency of 96.42% with robust performance across 50–150% load variations. This work demonstrates a simulation-guided pathway for intensifying separation processes, providing a validated framework for designing marine-adapted dehydration technology.

1. Introduction

The offshore oil and gas industry faces a dual challenge in natural gas dehydration. First, the high cost of platform space and strict deck-load limits render traditional bulky equipment such as gravity separators and packed towers increasingly impractical [1,2,3,4,5]. Second, the constant swaying and tilting of the platform caused by wind, waves, and currents disrupts separation processes that rely on a stable gravitational field, leading to performance degradation and operational risks [6,7,8,9]. These constraints create an urgent need for compact, motion-insensitive dehydration solutions.

Although existing studies have widely demonstrated the potential of rotating packed beds (RPBs) in intensifying gas–liquid mass transfer, most of these efforts were directed at land-based conditions, model systems, or batch operations, with optimization objectives often focusing on a single efficiency or mass transfer coefficient. Zhao et al. [10] found that an RPB increased the ozone decomposition rate to about four times that of a bubble column reactor. Feng et al. [11] utilized liquid detention to enhance CO₂ absorption efficiency and the volumetric mass transfer coefficient by 55.6% and 113.1%, respectively, and showed that polyethyleneimine retained approximately 85% desorption efficiency after five cycles. He et al. [12] significantly improved the production rate of bisphenol S by enhancing micro-mixing and mass transfer in an RPB. Nie et al. [13] achieved rapid methane hydrate formation and proposed an anti-clogging packing structure. Moreover, Zhu et al. [14] confirmed the significant enhancement of organic liquid dispersion by an oleophobic surface, although their work was limited to static liquid–liquid initial dispersion. Subsequent diversification of RPB configurations (packings, baffles, blades, stator-rotor designs, structured foams, etc.) and extensions to nanoparticle synthesis have all reflected ongoing optimization of hydrodynamics and mass transfer [15,16,17,18,19,20]. Nevertheless, the systematic, mechanically driven structural optimization of an RPB specifically designed for offshore natural gas dehydration with triethylene glycol (TEG)—one that explicitly coordinates maximum efficiency per unit volume, minimum energy consumption (pressure drop), and inherent robustness against marine motion conditions—has not yet been adequately addressed.

To fill this gap, the present study is guided by the quantitative correlation between liquid distribution uniformity and the degree of dehydration equilibrium. Using a combination of computational fluid dynamics simulations and experiments, we carry out a mechanism-driven structural optimization of a concentric-ring RPB. By investigating the effects of key structural parameters such as annular gap width, opening ratio, and rotor configuration on liquid distribution, mass transfer, and pressure drop, we aim to reveal the intrinsic structure-performance coupling mechanism. The ultimate goal is to simultaneously achieve a significant increase in dehydration efficiency and a substantial reduction in pressure drop, thereby providing an efficient and compact technical solution for offshore natural gas dehydration.

2. Materials and Methods

2.1. Experimental Setup and Procedure

2.1.1. Experimental System

The hypergravity dehydration system employed in this study comprises three primary components (Figure 1): (1) the rotating packed bed (RPB) module, (2) the stationary base, and (3) the electric drive motor. As shown schematically in Figure 2, the rotating bed is the core component of the high-gravity system, functioning to secure and drive the rotor structure and create a high-gravity field for enhanced gas–liquid mass transfer.

This study focuses on the concentric-circle rotating bed. It adopts a simplified baffle-modified structure consisting of concentric perforated rings; the interior features two-layer concentric perforated dynamic rings arranged radially from the outer to inner positions, with staggered hole distribution between the two dynamic rings. The 4-mm-diameter holes are uniformly distributed in 20 columns along the circumference. Gas–liquid phases achieve countercurrent contact through the channels between dynamic rings via these perforations. Figure 3 shows the structure and dimensions of different rotating beds.

The flow diagram of the supergravity dehydration simulation experiment is shown in Figure 4.

The instruments and meters used in this experiment are shown in

Table 1. Instruments and meters used in the experiment.

Name	Manufacturer	Model	Specifications
Compressor	Taizhou Aotusi Industry and Trade Co., Ltd. Taizhou, China	OTS550	2400 L/h
Pump	Zhongshan Gaoshuo Electronics Co., Ltd. Zhongshan, China	AB52	100 L/h
Dew point meter	Taiwan Hengxin Technology Co., Ltd. Taiwan, China	AZ8726	−78.7~50 °C
Pressure gauge	Detu Instrument Shanghai Co., Ltd. Shanghai, China	TESTO510	1 Pa~5 MPa
Filter separator	Jiangsu Zhongsheng High tech Industry Co., Ltd. Nanjing, China	ZS-GLS-05	500~5000 Nm3/d
Digital constant temperature water bath	Changzhou Huaao Instrument Manufacturing Co., Ltd. Changzhou, China	HH-2	0~100 °C
Thermostat water bath	Shanghai Yinze Instrument Equipment Co., Ltd.	HH-601	-
Higee dehydration unit	Self-built. Shanghai, China	-	-
Micro moisture analyzer	Shanghai Changji Geological Instrument Co., Ltd. Shanghai, China	SYD-2122B	3 μg~100 mg H2O

.

The designed and assembled experimental setup for supergravity-enhanced triethylene glycol (TEG) dehydration of natural gas achieved a measurement accuracy of 0.1 °C for water dew point, 1 Pa for pressure, and 0.5 °C for temperature.

2.1.2. Experimental Procedure and Parameters

Based on a self-constructed high-gravity triethylene glycol dehydration system, the experimental procedure was conducted as follows: methane gas is passed through an air compressor and then through a container containing a certain amount of water. The humidified air is introduced into a filter separator for preliminary removal of liquid water, after which it enters the RPB dehydration device while lean TEG, maintained at a set temperature via a constant-temperature water bath, is delivered by a peristaltic pump and sprayed onto the inner periphery of the rotor. Under high-gravity conditions, the liquid spread uniformly across the rotating packing, enabling efficient gas–liquid contact and effective dehydration. The dehydrated dry gas exited from the top of the unit, and the rich TEG was discharged from the bottom for regeneration and reuse. Dew points at the inlet and outlet of the reactor were measured using a probe-type dew point meter, and the pressure drop across the unit was monitored with a differential pressure gauge.

Upon assembly of each rotating bed configuration, the experimental setup was purged for 2–3 min using compressed air prior to formal testing. The rotational speed was precisely controlled by the rotating bed. The rotational speed was then incrementally increased from 100 rpm to 1500 rpm in steps of 200 rpm, maintaining each level for 30 s, while verifying consistent dew point readings at both the inlet and outlet to ensure system stability. Under baseline conditions including a lean TEG flow rate of 100 L/h, purity of 98.55%, and a gas feed rate of 2400 L/h, the effects of rotational speed (200, 400, 600, 800, 1000, and 1200 rpm, adjusted via the rotating bed) on dehydration performance were systematically investigated. Data for pressure drop and outlet dew point were recorded after maintaining each operational condition for 1 min to ensure steady-state measurement.

In the natural gas dehydration system of triethylene glycol, it represents the best dehydration effect that the system can theoretically achieve. It is usually used to evaluate the dehydration performance, and the calculation formula is as follows:

$$UI={\displaystyle \frac{P_v}{P_s}}\times 100\%$$

(1)

In the equation, UI represents the dehydration equilibrium degree; $P_v$ denotes the saturated vapor pressure of water above the lean TEG solution, and $P_S$ signifies the saturated vapor pressure of water in the outlet air.

The vapor pressures of water were determined as follows. The saturated vapor pressure of water above the lean TEG solution was calculated using Raoult’s law, which is the product of the mole fraction of water in the lean TEG and the saturated vapor pressure of pure water at the corresponding temperature. The mole fraction of water in the lean TEG was measured using a SYD-2122B fully automatic micro-moisture titrator based on the Karl Fischer coulometric method. Meanwhile, the saturated vapor pressure of water in the outlet air was obtained by multiplying the mole fraction of water in the outlet air by the standard atmospheric pressure. The mole fraction of water in the outlet air was derived from the measured dew point of the effluent air stream.

2.2. Numerical Modeling

2.2.1. Governing Equations and Physical Models

The absorption of water vapor from natural gas by triethylene glycol is a purely physical process without any accompanying chemical reactions, and the same applies to the absorption of water vapor from air [21,22]. However, there is currently no accurate description of the mass transfer coefficient or correlation for the triethylene glycol absorption process under a high-gravity field, which makes it impossible for simulation results to directly reflect the absorption extent. As a result, the influence of temperature on the simulation outcomes is negligible, and the thermal effects can be disregarded during the simulation. The energy equation is thus excluded, while the viscosity of triethylene glycol is considered due to its certain impact on the flow dynamics [23,24]. Sensitivity analysis conducted on the viscosity variation corresponding to typical ±10 °C fluctuations indicates that the relative change in the UI is less than 2.2%, and the fluctuation in the dehydration equilibrium degree is less than 0.8%.

The physical properties are simplified and characterized by the density and viscosity at the experimentally determined optimal temperature of 40 °C, ensuring better alignment with practical conditions. Therefore, this section only considers the continuity equation and the momentum equation.

Continuity equation:

$${\displaystyle \frac{\partial \left(\rho u_x\right)}{\partial x}}+{\displaystyle \frac{\partial \left(\rho u_y\right)}{\partial y}}+{\displaystyle \frac{\partial \left(\rho u_z\right)}{\partial z}}=S_m$$

(2)

Here x, y, z represent the three spatial directions of the flow, while u_x, u_y, u_z denote the corresponding velocity components in each direction. The term S_m stands for the source term in the continuity equation, which is set to zero in the present simulation.

Momentum equation:

$${\displaystyle \frac{\partial \left(\rho u_x\right)}{\partial t}}+\nabla \cdot \left(\rho u_x\overrightarrow{u}\right)=-{\displaystyle \frac{\partial p}{\partial x}}+{\displaystyle \frac{\partial {\tau }_{xx}}{\partial x}}+{\displaystyle \frac{\partial {\tau }_{yx}}{\partial y}}+{\displaystyle \frac{\partial {\tau }_{zx}}{\partial z}}+\rho f_x$$

(3)

$${\displaystyle \frac{\partial \left(\rho u_y\right)}{\partial t}}+\nabla \cdot \left(\rho u_y\overrightarrow{u}\right)=-{\displaystyle \frac{\partial p}{\partial y}}+{\displaystyle \frac{\partial {\tau }_{xy}}{\partial x}}+{\displaystyle \frac{\partial {\tau }_{yy}}{\partial y}}+{\displaystyle \frac{\partial {\tau }_{zy}}{\partial z}}+\rho f_y$$

(4)

$${\displaystyle \frac{\partial \left(\rho u_z\right)}{\partial t}}+\nabla \cdot \left(\rho u_z\overrightarrow{u}\right)=-{\displaystyle \frac{\partial p}{\partial z}}+{\displaystyle \frac{\partial {\tau }_{xz}}{\partial x}}+{\displaystyle \frac{\partial {\tau }_{yz}}{\partial y}}+{\displaystyle \frac{\partial {\tau }_{zz}}{\partial z}}+\rho f_z$$

(5)

Here ${\tau }_{xx}$, ${\tau }_{xy}$, ${\tau }_{xz}$ denote the components of the stress tensor τ, while $f_x$, $f_y$, $f_z$ represent the unit body forces acting in the three coordinate directions.

The simulation focused on analyzing gas–liquid flow dynamics and mass transfer mechanisms within the rotating packed bed, addressing both mass and momentum conservation under intense swirling conditions. Given the flow complexity involving rotational shear, jet mixing, and boundary layer separation, turbulence modeling required careful consideration. Based on prior validation studies [25,26], the RNG k-ε model was selected over standard k-ε for its demonstrated accuracy in rotating shear flows, free jets, and separation scenarios—key features of RPB hydrodynamics.

Turbulent kinetic energy (k) equation:

$${\displaystyle \frac{\partial (\rho k)}{\partial t}}+{\displaystyle \frac{\partial (\rho k{u}_i)}{\partial x_i}}={\displaystyle \frac{\partial }{\partial x_j}}\left({\alpha }_k{\mu }_{eff}{\displaystyle \frac{\partial k}{\partial x_j}}\right)+G_k-\rho ϵ-Y_M+S_k$$

(6)

Dissipation rate (ε) equation:

$${\displaystyle \frac{\partial (\rho ϵ)}{\partial t}}+{\displaystyle \frac{\partial (\rho ϵu_i)}{\partial x_i}}={\displaystyle \frac{\partial }{\partial x_j}}\left({\alpha }_{ϵ}{\mu }_{eff}{\displaystyle \frac{\partial ϵ}{\partial x_j}}\right)+C_{1ϵ}{\displaystyle \frac{ϵ}{k}}{G}_k-C_{2ϵ}\rho {\displaystyle \frac{{ϵ}^2}{k}}-R_{ϵ}+S_{ϵ}$$

(7)

Here, $t$ denotes time; $x_i$ and $x_j$ represent spatial coordinate directions; $u_i$ is the mean velocity component; and $\rho$ stands for fluid density. The turbulent kinetic energy and its dissipation rate are expressed as $k$ and $\epsilon$, respectively. $G_k$ refers to the generation term of turbulent kinetic energy resulting from mean velocity gradients, while ${\mu }_{eff}$ indicates the effective dynamic viscosity, defined as ${\mu }_{eff}=\mu +{\mu }_t$, where $\mu$ is the molecular viscosity and ${\mu }_t$ denotes the turbulent viscosity. The parameters ${\alpha }_k$ and ${\alpha }_{\epsilon }$ represent the inverse effective Prandtl numbers for $k$ and $\epsilon$, respectively. In the RNG model, these parameters vary with the flow field. $C_{1\epsilon }$ and $C_{2\epsilon }$ are model constants, and $Y_M$ describes the contribution of fluctuating dilatation to the overall dissipation rate in compressible turbulence. $S_k$ and $S_{\epsilon }$ denote user-defined source terms. $R_{\epsilon }$ is a distinctive additional term in the RNG model, serving as a key differentiator from the standard k–ε model. This term is designed to account for the effects of small-scale turbulence and improve predictions in shear flows. It is typically formulated as follows:

$$R_{\epsilon }={\displaystyle \frac{C_{\mu }\rho {\eta }^3(1-\eta /{\eta }_0)}{1+\beta {\eta }^3}}{\displaystyle \frac{{\epsilon }^2}{k}}$$

(8)

where $\eta$ represents the ratio of the turbulent to mean strain time scales, defined as

$$\eta ={(2E_{ij}·E_{ij})}^{{\displaystyle \frac{1}{2}}}{\displaystyle \frac{k}{\epsilon }}$$

(9)

with $E_{ij}$ being the modulus of the mean strain rate tensor. The constants ${\eta }_0$ and $\beta$ are specific to the RNG model.

As the TEG-water vapor absorption involves no chemical reactions, a multiphase approach treating gas and liquid as interpenetrating continua was adopted. Comparative evaluation confirmed the Mixture model’s superiority for this system, efficiently resolving phase interactions while maintaining computational stability. The presence of both stationary and rotating components necessitated a sliding mesh technique for transient simulation, capturing critical rotor-stator interaction effects absent in steady-state approaches.

2.2.2. Physical Model

In simulating the flow of gas and triethylene glycol within the rotating bed, chemical reactions were not considered, and a multiphase flow model was employed. The gas and liquid phases were treated as interpenetrating continua, making the Mixture model more suitable for the configuration of this study. Since the rotating bed under investigation contains both stationary and rotating components, and the interaction forces between them cannot be neglected, a sliding mesh technique was adopted for the transient solution. The main body establishes a 1:1 scale model based on the experimental setup, with basic parameters as shown in

Table 2. Basic parameters of rotating bed model.

Project	Numerical Value
Liquid flow rate, L/h	100
Gas flow rate, L/h	2400
Liquid phase density, g/cm3	1.1
Gas phase density, kg/m3	1.206
Size of the upper and lower end caps of the swivel (diameter × thickness), m	Φ0.09 × 0.004
Size of the gas inlet and outlet pipes, m	Φ0.015 × 0.002
Size of the liquid inlet and outlet pipes, m	Φ0.005 × 0.001
Jet hole size (diameter × quantity), m	Φ0.001 × 16
Shell size (diameter × height × wall thickness), m	Φ0.1 × 0.09 × 0.005

and Figure 4.

2.2.3. Boundary Conditions and Solution Strategy

The gas phase (treated as incompressible) and the liquid TEG phase both employed velocity-inlet and pressure-outlet (atmospheric) boundaries. The gas inlet velocity is 4 m/s, and the liquid inlet velocity is 2 m/s. Turbulence parameters are defined by turbulence intensity and hydraulic diameter: gas phase: 5.2% and 15 mm; liquid phase: 6.8% and 1 mm. The initial gas volume fraction is 1, and liquid is gradually injected from the nozzles until a periodic steady state is reached. A sliding mesh is used between the rotor and the casing; the computational domain is divided into rotating and stationary zones, with flux interpolation at the interface [27,28]. Taking into account both computational accuracy and resource consumption, a medium grid is selected for subsequent calculations. The near-wall area is processed using standard wall functions, and after post-processing statistics, the y + values on the wall are mainly distributed in the range of 30–150, which meets the requirements of logarithmic law layers. Both the rotor and the stationary wall are equipped with 5 layers of prismatic mesh (with a first layer height of 0.02 mm and a growth rate of 1.2) to analyze the velocity gradient within the boundary layer.

A robust solution strategy was first implemented. The FLUENT pressure-based solver was employed for transient analysis to accurately capture rotor-stator interactions. A timestep of 0.001 s was selected to ensure Courant number stability, and a total physical time of 10 s (10,000 iterations) was simulated to achieve period-independent solutions. Convergence criteria mandated residual reductions below 10⁻⁴ for all parameters per timestep.

To verify grid independence, this study used three sets of schemes: coarse grid (approximately 850,000), medium grid (approximately 1.65 million), and fine grid (approximately 2.8 million) for independence testing. Based on the medium grid, the pressure drop and uniformity index changes in the fine grid are both less than 1%.

2.2.4. Liquid Distribution Uniformity and Its Correlation with Dehydration Performance

Liquid distribution characteristics serve as a fundamental factor influencing mass transfer efficiency in rotating beds. Uniform liquid distribution ensures full wetting of the packing surface, effectively increasing the gas–liquid contact area and thereby significantly enhancing the mass transfer process. In contrast, non-uniform distribution leads to insufficient wetting, reduces the mass transfer area, and diminishes gas–liquid contact efficiency, ultimately impairing overall mass transfer performance [29,30]. Research by Zhang [31] clearly pointed out that “uneven liquid distribution in RPB has a serious impact on mass transfer performance” and employed CFD methods to quantify the maldistribution index. Liao [32] improved the overall performance of RPB by optimizing the liquid distributor structure. Similarly, Han’s study on the hydrodynamics and liquid-solid mass transfer in micro-packed bed reactors with copper foam packing highlighted the significant influence of liquid distribution on liquid-solid mass transfer. Liu’s work further revealed the impact of initial liquid dispersion on liquid distribution and mass transfer performance in rotating packed beds. Numerous studies support the view that liquid distribution performance can serve as an effective indicator for evaluating mass transfer effectiveness [33,34]. Therefore, liquid distribution characteristics can serve as effective simulation outputs for evaluating mass transfer effectiveness. It should be noted that the validity of UI as a surrogate indicator for dehydration performance is based on the premise that physical absorption dominates and the mass transfer resistance is mainly controlled by the gas–liquid interfacial area.

This study established a three-dimensional model that matches the experimental conditions to accurately represent the true flow field characteristics and capture the liquid phase distribution. We employed FLUENT’s built-in uniformity index (UI) calculation to assess distribution quality, which quantifies liquid volume fraction variations across specified surfaces to reflect distribution homogeneity. The UI ranges from 0 to 1 and is calculated using the following equation:

$$UI=1-{\displaystyle \frac{{\displaystyle \sum _{i=1}^n\left|{\phi }_i-{\overline{\phi }}_{\alpha }\right|A_i}}{2\left|{\overline{\phi }}_{\alpha }\right|{\displaystyle \sum _{i=1}^n{A}_i}}}$$

(10)

where $i$ represents the face index for a surface discretized into n computational faces, and $\overline{{\varphi }_a}$ denotes the area-averaged liquid volume fraction across the entire surface.

Where $A_i$ represents the area of the i-th computational face in the surface mesh, and ${\varphi }_i$ denotes the liquid volume fraction at the i-th face. The index i ranges from 1 to n, where n is the total number of discretized faces comprising the surface domain.

Through experimental and simulation-based multiple correlation analysis of the uniformity index and dehydration equilibrium under different operating conditions, the scatter plot of the fitted curve is shown in Figure 5. The goodness-of-fit reaches 0.935, indicating an excellent fitting effect. Therefore, based on the relationship established between the distribution uniformity index and the dehydration equilibrium, Equation (11) is used in subsequent calculations to determine the dehydration equilibrium. The limitation of UI as a mass transfer performance indicator lies in that it only characterizes the geometric uniformity of liquid distribution, rather than the actual mass transfer rate of water vapor. Therefore, the UI-performance correlation established in this work is applicable only to dehydration processes dominated by physical absorption; caution is required when applying engineering extrapolation to real natural gas systems involving chemical reactions or complex components.

The correlation between the uniformity index and the dehydration equilibrium degree was fitted as follows:

$$\alpha =1.114UI+0.0092$$

(11)

where $\alpha$ denotes the dehydration equilibrium degree (%); $UI$ represents the uniformity index.

2.2.5. Model Validation

To ensure the accuracy of the numerical simulations, a comprehensive experimental validation of the established CFD model was conducted. The validation focused on two key performance parameters: system pressure drop and dehydration equilibrium degree.

Figure 6 compares the simulated and experimental pressure drops at different rotational speeds, showing consistent trends with an average deviation of 6.5% and a maximum deviation of 9.8%. More critically, Figure 7 presents the comparison between the simulated and actual dehydration equilibrium efficiency values, revealing excellent agreement with an average deviation of merely 0.3% and a maximum deviation of 0.7%. These comparative results robustly confirm the high reliability of the present simulation methodology. The model not only accurately predicts the hydrodynamic characteristics (e.g., pressure drop) within the rotating bed but also, through the established correlation between the uniformity index and the dehydration equilibrium degree, successfully replicates the system’s final dehydration performance. Therefore, this validated model is deemed suitable for the subsequent in-depth analysis of flow field characteristics and dehydration performance within the rotating packed bed.

3. Results

3.1. Comparison of Dehydration Equilibrium Performance of Rotating Packed Beds with Different Structures

In the triethylene glycol dehydration process, dehydration equilibrium is utilized to evaluate its dehydration performance. Experimental results were obtained using a self-built apparatus through Equation (1).

Table 3. Dehydration equilibrium degree of four rotating beds at different rotating speeds (%).

Type	Rotational Speed (rpm)
	200	400	600	800	1000	1200
Blade	72.34	72.99	74.41	75.85	75.68	75.89
Spiral	68.12	70.54	71.08	72.29	73.42	74.89
Concentric-Ring	80.21	81.17	83.44	86.54	86.7	87.11
Stator-Rotor	66.71	68.39	70.97	72.75	73.12	73.02

presents the dehydration equilibrium data for four types of rotating beds at different rotational speeds under lean triethylene glycol conditions at 40 °C.

As can be seen from Table 3, among the four types of rotating beds studied, the concentric-circle rotating bed exhibits the best dehydration performance, with a dehydration balance consistently above 80%.

3.2. Flow Field Characterization

To investigate the distribution characteristics of gas–liquid two-phase flow in different rotating beds, this paper simulates the liquid volume fraction contours inside various rotating beds under the condition of a rotational speed of 800 rpm and a TEG liquid temperature of 40 °C (as shown in Figure 8A–D and Figure 9A–D). By comparing the diffusion range of the gas phase in each figure, the impact of different rotating beds on the uniformity of gas distribution and the gas–liquid contact interface can be intuitively analyzed.

The concentric-ring rotating bed in Figure 8C enables the most uniform liquid distribution, with its core advantage lying in the high degree of matching among structural design, flow characteristics, and pressure distribution. Its design features annular layered arrangements around the center, with circumferentially symmetric flow channels free from unidirectional obstructions. The multi-layered flow channels form multiple sets of parallel dispersion paths, covering a wider cross-sectional area. Moreover, the annular structure in Figure 9C allows the centrifugal force generated by rotation to act uniformly on the liquid, establishing a stable radial pressure gradient that progressively drives the liquid outward through successive ring layers. This design aligns with the radial dispersion tendency of centrifugal force, thereby preventing local accumulation. Additionally, the continuous annular flow channel cross-sections and consistent turning angles reduce localized flow resistance, avoiding issues such as asymmetric resistance or excessive local pressure drop. The resulting uniform pressure distribution further ensures symmetric radial flow velocities and balanced circumferential forces while preventing the formation of vortices or dead zones in the flow field. This effectively eliminates both local liquid accumulation and dispersion segregation. Compared to other structural types prone to sudden local pressure changes, the concentric-ring rotating bed achieves more stable and uniform circumferential liquid dispersion. The uniformity index of the simulation results is presented in

Table 4. Uniformity index of the simulation.

Vane Type	Spiral Type	Concentric-Ring Type	Stator-Rotor
67.02	65.40	77.26	63.17

below.

Since the simulation result uniformity index and the experimental result dehydration balance degree exhibit a linear relationship according to Formula (9), it indicates that the concentric circular rotating bed has the best dehydration performance.

3.3. Investigation of Flow Fields, Structural Optimization, and Performance Enhancement in Concentric Rotating Beds

3.3.1. Structural Optimization and Design

Based on the findings presented in Section 3.1 and Section 3.2, the concentric-ring rotating bed was identified as the most promising configuration. To further explore its performance potential, this section focuses on structural optimization and systematic evaluation, aiming to enhance its overall performance through targeted improvements. However, based on flow field analysis, several issues have been identified in the existing concentric-ring rotating bed structure:

• Liquid Distribution Issues: The synchronous rotation of the spray pipe and rotor structure, combined with lateral liquid discharge, leads to insufficient liquid dispersion in the upper rotor space, weakening liquid phase mixing.
• Pressure Drop Concerns: Gas ejection through peripheral orifices contributes disproportionately to system pressure drop.
• Low-efficiency Zones: The region between the outer rotating ring and cavity exhibits inadequate liquid holdup, reducing effective interfacial area.

To address these challenges, the following structural modifications were implemented:

• The spray pipe was redesigned as a stationary vertical pipe with perforated walls, thereby decoupling it from the rotor assembly and ensuring axial liquid distribution.
• Gas Outlet Redesign: The orifice-based gas exhaust was replaced with direct tubular venting to minimize flow resistance.
• Rotating Assembly Modification: The outer rotating ring was integrated with the rotor base plate edge, eliminating the low-efficiency interfacial zone, while an additional rotating ring was incorporated to augment interfacial contact.

3.3.2. Performance Evaluation of the Optimized Bed

Building upon the original concentric-ring rotating bed design, we developed an improved structural model following the same modeling methodology detailed in Section 2.2. As shown in Figure 10, the geometric model achieves satisfactory computational accuracy and demonstrates good convergence characteristics with 253,234 nodes and 1,266,146 mesh elements.

Inside the rotor, three concentric perforated movable rings are arranged radially, with diameters of Φ16 mm, Φ49 mm, and Φ70 mm, respectively, and the inner diameter of the housing is Φ90 mm. The annular gaps between adjacent movable rings and between the outermost ring and the housing are 4.5 mm, 4 mm, and 5 mm, respectively. Each movable ring is provided with circular perforations of Φ1 mm, uniformly distributed in 20 columns along the circumferential direction. The perforations in adjacent layers are staggered to enhance the disturbance and contact between the gas and liquid phases as they pass through the openings.

Figure 11 presents the liquid phase volume fraction contours on the radial cross-section under operating conditions of 2400 L/h gas flow, 100 L/h liquid flow, and 800 rpm rotational speed, with all parameters maintained identical to the pre-modification configuration. The liquid distribution demonstrates significantly improved uniformity compared to the original design, with dehydration equilibrium increasing from 86.1% to 92.25%, indicating enhanced dehydration efficiency. Figure 12 displays the pressure distribution contours on the radial cross-section under identical operating conditions (2400 L/h gas flow, 100 L/h liquid flow, 800 rpm). The modified configuration shows a substantial reduction in pressure drop from 5850 Pa to 1570 Pa compared to the baseline design.

3.3.3. Comprehensive Assessment of the Operational Performance of the Optimized Rotating Bed

Following the confirmation of enhanced baseline performance, the operational envelope of the optimized rotating bed was further explored by investigating the effects of key process parameters, namely gas–liquid ratio, system pressure, and load fluctuation. Under a constant gas flow rate of 2400 L/h, the liquid flow rate was varied to investigate the influence of the gas–liquid ratio on dehydration performance, using an experimental device with a driving motor rated at 120 W that runs smoothly at a maximum speed of 1500 rpm.

Analysis of

Table 5. Uniformity index of different gas–liquid ratios and corresponding dehydration equilibrium degree.

Liquid Volume (L/h)	Gas–Liquid Ratio	Uniformity Index	Dehydration Equilibrium Degree (%)
240	10	0.8121	91.42
120	20	0.8187	92.16
80	30	0.8234	92.67
60	40	0.8268	93.05
48	50	0.8307	93.49
40	60	0.8289	93.31
34.3	70	0.8273	93.14
30	80	0.8248	92.83
26.7	90	0.8231	92.64
24	100	0.8223	92.55
21.81	110	0.8217	92.45
20	120	0.8210	92.34
18.5	130	0.8198	92.22
17.1	140	0.8174	92.07
16	150	0.8157	91.65

data reveals that dehydration efficiency deteriorates at both lower and higher gas–liquid ratios within the studied range. Under constant gas flow conditions, lower gas–liquid ratios correspond to excessive liquid flow which reduces gas-induced turbulence, thereby impairing liquid distribution uniformity, while higher ratios result in insufficient liquid volume that limits dispersion coverage. However, the overall variation in dehydration equilibrium efficiency remains relatively small, indicating a wide operable range for the gas–liquid ratio. Under the present simulation conditions, considering both minimal liquid consumption and optimal dehydration performance, the ideal gas–liquid ratio is approximately 60:1, achieving a dehydration equilibrium efficiency of 93.31%.

Having identified the optimal gas–liquid ratio, the influence of system pressure was subsequently investigated under this condition (60:1). As demonstrated in Figure 13, increasing pressure generally enhances dehydration equilibrium efficiency. This improvement occurs because elevated pressure strengthens the overcoming of liquid film resistance, promoting more thorough gas–liquid contact and consequently more uniform liquid distribution. However, beyond 2 MPa, the rate of efficiency improvement diminishes as excessive pressure impedes countercurrent gas–liquid contact and may induce liquid re-entrainment or even flooding, adversely affecting mass transfer. Consequently, the optimal rotational bed pressure for this system is approximately 2 MPa, achieving a dehydration equilibrium efficiency of 96.42%.

In the simulations, the system was configured for a nominal gas capacity of 2400 L/h. With the liquid flow rate fixed at 40 L/h, corresponding to an optimal gas–liquid ratio of 60:1, the gas flow rate was varied to assess the operational adaptability range. The resulting uniformity indices and the corresponding dehydration equilibrium efficiencies are presented in

Table 6. Uniformity index of different gas volumes and corresponding dehydration equilibrium degree.

Gas Volume (L/h)	Device Load Rate (%)	Uniformity Index	Dehydration Balance (%)
4080	170	0.8202	92.32
3840	160	0.8231	92.64
3600	150	0.8255	92.91
3360	140	0.8262	92.99
3120	130	0.8268	93.06
2880	120	0.8275	93.14
2640	110	0.8284	93.24
2400	100	0.8289	93.31
2160	90	0.8281	93.20
1920	80	0.8273	93.11
1680	70	0.8265	93.02
1440	60	0.8252	92.88
1200	50	0.8241	92.75

.

Analysis of Table 6 data indicates that gas flow variations have minimal impact on dehydration performance, with the calculated dew-point difference between maximum and minimum dehydration equilibrium efficiencies remaining below 1 °C across the 50–150% load capacity range, demonstrating robust operational adaptability within this operating window.

4. Conclusions

This integrated experimental and computational study demonstrates that the concentric-ring rotating packed bed (RPB) represents a superior configuration for intensifying the TEG dehydration process. By establishing a direct, quantitative link between microscale flow uniformity and macroscale process efficiency, this work provides a validated, simulation-guided framework for the rational design and optimization of high-gravity separation equipment.

The principal findings are delineated as follows:

• Performance Superiority of Concentric-Ring Design: Comparative experimental evaluation of four RPB structures identified the concentric-ring geometry as optimal, consistently achieving the highest dehydration equilibrium (>80%). Its inherent advantage stems from a structural synergy that aligns centrifugal force distribution with annular flow paths, promoting uniform radial dispersion and minimizing maldistribution.
• Mechanistic Insight and Predictive Correlation: Computational Fluid Dynamics (CFD) analysis, validated against experimental data, elucidated the fundamental flow mechanisms. A robust linear correlation (R² = 0.935) was established between the simulated liquid distribution uniformity index (UI) and the measured dehydration performance, effectively bridging detailed flow field characterization with ultimate process outcome and establishing UI as a reliable predictive metric.
• Targeted Optimization Yields Significant Gains: Guided by flow field diagnostics, specific structural modifications—including a stationary distributor and streamlined gas outlet—were implemented. The optimized design delivered a decisive improvement: a 7.1% relative increase in dehydration equilibrium (to 92.25%) coupled with a 73% reduction in system pressure drop, demonstrating simultaneous enhancement of separation efficiency and energy economy.
• Defined Operational Envelope and Robustness: Parametric studies established an optimal operating point at a gas–liquid ratio of 60:1 and a system pressure of ~2 MPa, achieving a peak dehydration equilibrium of 96.42%. Furthermore, the system exhibited excellent operational flexibility, maintaining stable performance with a dew-point variation of less than 1 °C across a 50–150% load range.

In summary, this study goes beyond mere performance comparisons to delve into the relationship between structure, flow, and performance within rotating packed beds (RPBs). It successfully demonstrates a simulation-based optimization approach, enabling the design of equipment with both higher efficiency and lower energy consumption. This provides a specific blueprint for the development of process intensification technology and offers valuable insights for subsequent dehydration on offshore oil and gas platforms.