Extraction of natural compounds using supercritical CO2: Going from the laboratory to the industrial application

• Experiments provide a mass transfer parameter for optimally pretreated substrates.
• Computer simulation (CS) helps optimizing industrial multi-vessel (MV) SFE plants.
• Pilot plant data will help refining simulation code to account for nonideal flow conditions.
• CS may be the single tool permitting the multivariable optimization required for MV SFEs.
• Technical restrictions to particle size, CO2 velocity, and plant cycle time await identification.

Despite industrial application for almost four decades, there is reluctance in some world regions to adopt supercritical (sc) CO2 extraction because of the wrong perception that it is not fully competitive. To refute this misconception, this manuscript analyzes economics of scCO2 extraction of vegetable oil from prepressed seeds. Selection of this application was due to the availability of a predictive mathematical model of the extraction process applicable for simulation purposes; inner microstructural changes of oilseeds during prepressing allow their extraction according to a shrinking core hypothesis. The predictive model has as its single parameter a particle-size and scCO2-condition-independent microstructural mass transfer factor that can be best-fitted to laboratory extractions, existing literature correlations to estimate other model parameters, such as the axial dispersion in packed beds operating with supercritical fluids, and the solubility of vegetable oils in scCO2. On the other hand, there is a need to correlate literature data for the film mass transfer coefficient to unveil the factors responsible for experimental data scattering. Because laboratory or pilot plant runs in single-extraction-vessel units cannot produce the simulated countercurrent contact in an industrial plant having ≥3 extraction vessels, mathematical simulation provides the relationship between oil yield and extraction time that can anchor precise estimations of extraction cost. Analysis of results unveiled differences in optimal extraction time (for minimal extraction cost) between production costs estimated in this work and the operational costs informed before. Because the operational cost does not include the capital cost of the industrial plant, the need appears to reduce its contribution to the total cost by increasing plant productivity. This is achieved reducing extraction time, which negatively influences oil yield.To make further progress in the optimization of industrial scCO2 extraction processes, this manuscript proposes refining the mathematical simulation approach, and studying those technical constraints whose manifestations become more prevalent on scale-up. Mathematical simulation can be adapted to alternative, sample-pretreatment dependent mass transfer mechanisms in the solid matrix. It can be refined also to account for the size distribution of the substrate, radial changes in superficial scCO2 velocity, axial changes in pressure, and radial/axial changes in temperature resulting from heterogeneous packing, pressure drop, and/or heat transfer from/to extraction vessel walls that may influence large-scale extractions. Large-scale experiments will allow studying these phenomena, as well as technical constraints to the decrease in particle size, increase in scCO2 velocity, and decrease in extraction time imposed by the agglomeration and decrease in packed bed permeability of the substrate, and thermal effects during reconditioning of extraction vessels. The latter effects should be included as restrictions in the optimization of the extraction process, which may limit the extraction rate and the size or number of extraction vessel that impact economics positively. Close collaboration with industry will facilitate tackling large-scale problems, as well as refining estimates of plant cost as a function of its size and/or configuration.

Figure optionsDownload as PowerPoint slide