Publications Repository - Helmholtz-Zentrum Dresden-Rossendorf

Distillation is the leading thermal separation technology that is carried out in many industrial tray columns worldwide. Although distillation columns are expensive in terms of cost and energy, they will remain in service due to unavailability of any equivalent industrially-viable alternative. However, rising energy costs and urgent needs to reduce greenhouse gas emissions demand improvements in the energy efficiency of separation processes, globally. This can be achieved by tuning the dynamics of the evolving two-phase dispersion on column trays via design modification and revamping. Thus, it becomes necessary to understand how the two phases evolve over the tray and how they link to tray efficiency for given tray designs, systems and operating conditions. Only then, the cost and energy reduction can be achieved by strategically iterating the tray design and revamps with respect to the resulting tray efficiency. To pursue this strategy, accurate prediction of the separation efficiency based on flow and mixing patterns on the trays is an important prerequisite.

In this thesis, the mathematical models relying on flow and mixing patterns for predicting the tray efficiencies were reviewed. These models were developed based on the analyses of two-phase flow, crossflow hydraulics and mass transfer over the trays. Several limitations in the existing models were identified that could lead to inaccurate tray efficiency predictions. First, the conventional models do not account for any variation in the local two-phase flow in their formulation. These models rather consider a homogeneous flow scenario based on flow monitoring at the tray boundaries only, which indicates a black box efficiency estimation. Second, the existing models do not consider any vapor flow maldistribution, which can be detrimental to the tray efficiency. In response to these limitations, a new model based on refinement of the conventional residence time distribution (RTD) model (referred to as the ‘Refined RRTD model’) was proposed. The new model involves geometric partitioning of the tray into compartments along the flow path length, which permits computing the tray efficiency through quantification of the efficiency of the individual compartments. The proposed model ensures that the fluid dynamics of each compartment contribute towards the overall tray efficiency, which specifically targets the black box prediction of the tray efficiency by the conventional models. The tray discretization further aids in analyzing the impact of vapor flow maldistribution on the tray efficiency. In the initial assessment, the new model capabilities were demonstrated in appropriate case studies after theoretical validation of the model for the limiting cases of the two-phase flows. For the experimental validation of the new model, a full hydrodynamic and mass transfer description of the two-phase dispersion specific to the tray operation is indispensable. Because of the inherently complex dispersion characteristics, significant advancements in the imaging and efficiency modeling methods were required.

In this thesis, a DN800 column simulator equipped with two sieve trays (each with 13.55% fractional free area) was used with air and tap water as the working fluids. Deionized water was used as a tracer. The gas loadings in the column in terms of 𝐹-factor were 1.77 Pa0.5 and 2.05 Pa0.5, whereas the weir loadings were 2.15 m3m-1h-1, 4.30 m3m-1h-1 and 6.45 m3m-1h-1. An advanced multiplex flow profiler comprising 776 dual-tip conductivity probes for simultaneous conductivity measurements was introduced for hydrodynamic characterization. The spatial resolution of the profiler based on the inter-probe distance was 21 mm × 24 mm, whereas the temporal resolution was 5000 Hz. The design characteristics of the new profiler, electronic scheme, measurement principle, reference framework, and data processing schemes are explained in detail. By analyzing the two-phase dispersion data gathered by the profiler at multiple elevations above the tray, the effective froth height distributions were obtained for the first time based on a newly proposed approach. Uniform froth heights were seen over the majority of the tray deck, whereas both minimum and maximum froth heights were detected immediately after the tray inlet. Based on threshold-based calculation (accompanied by γ-ray CT scans), 3D time-averaged liquid holdup distributions were visualized for the first time, too. Homogeneous liquid holdup distributions were observed at multiple elevations above the deck with the highest holdups occurring near the average effective froth heights. The detailed flow and mixing patterns of the liquid in the two-phase dispersion were retrieved via tracer monitoring. With respect to tray centerline, axisymmetric liquid flow and mixing patterns were detected with parabolic velocity distributions near the tray inlet. The liquid velocities over the remaining tray deck were nearly uniform for the prescribed loadings. Eventually, the RRTD model was applied by discretizing the tray geometrically, and accordingly employing the available hydrodynamic data. The conventional models often applied in the literature were also evaluated with the new model.

For evaluating the model predictions, a new system add-on for the existing air-water column facility was proposed for direct efficiency measurements. The air-led stripping of isobutyl acetate from the aqueous solution is a safe and viable approach that overcomes numerous limitations posed by the existing chemical systems. Based on liquid sampling at different tray locations, the liquid concentration distributions were obtained at each operating condition via UV spectroscopy. The tray and point efficiencies as well as stripping factors were calculated from those distributions. Because of the low liquid diffusivity and high liquid backmixing, low efficiencies were observed at the given loadings. The model predictions were consistent with the experimental counterparts (even for the extrapolated values of the involved parameters), because of the uniform liquid flow and mixing in the compartments. For the given predictions, those corresponding to the new RRTD model were the most accurate. Additional hydrodynamic and efficiency data are needed for more conclusive evidence regarding the promise of the RRTD model.