Matching Operating Performance in Membrane Separations
Membrane separation has become widely established in the process industry. For example, CO2 removal from natural gas streams with membranes has been applied since the 80s. Vendors of traditional CO2 removal technologies have also incorporated membrane-based processes.
In the example below, a gas plant processing a rather acid stream with 11% of CO2 and 85 % of methane and other heavier hydrocarbons, is equipped with membrane banks supplied by a major membrane technology vendor. The membrane banks are used to separate a CO2-rich permeate stream, taking advantage of the greater selectivity of the membrane elements for CO2 over the hydrocarbon components. The CO2-rich is then discarded, whereas the sweetened hydrocarbon stream is sold as natural gas within specification (< 8.5 % in CO2).
For the simulation, the engineer has obtained some typical permeabilities for one of the main materials of the membrane, silicone rubber. Upon a brief literature review, the permeabilities for the key components are as follows:
These permeabilities are then input onto the Permeabilities tab of the membrane unit operation in the Symmetry process software platform. The permeabilities for ethane, and propane are not available but for the first pass of the simulation, the methane permeabilities will be used for these components. As shown later, these permeabilities will be adjusted based on actual plant performance.
Figure 1. Specification of a membrane unit operation for CO2 separation
The thickness, operating area of the membrane banks and permeate pressure (20 psia or 138 kPa), are input such that the unit operation can solve.
Once the membrane has solved, the sweetened gas stream flows are inspected and compared to the calculated plant components flowrates. The need for and adjustment is evident as there’s a mismatch between the model and the actual plant performance as shown below:
Figure 2. Solved membrane and comparison to operating data
The actual permeabilities can be estimated using the Model Regression tool. The membrane permeabilities should be added as Regressed Variables and the sweetened gas component flowrates calculated from the actual plant measurements are specified as the Experimental Values as shown below.
Figure 3. Configuration of a regression model for a membrane
Once the regression has finished the newly calculated permeabilities calculate new sweetened gas flowrates. Since the component error for propane is still high (13%), an alternative scale for permeabilities will be tried out by using the log10 of the permeabilities, a new Symmetry process platform 2020 feature, as depicted below.
Figure 4. Configuration of a regression model for a membrane using Log10 of permeabilities
The error results for this regression are within the expected measurement range (< 0.31 %) and are more acceptable and will be used by setting the regressed values as specifications of the membrane unit operation.
Upon inspection of the sweetened hydrocarbon stream in the Symmetry process platform model, it can be observed that the membrane model matches the actual plant performance.
Figure 5. Final model of the membrane with the regressed permeabilities
The Symmetry process platform offers a rigorous membrane unit operation to allow for a comprehensive modeling of a gas processing facility that uses this type of separation. Combining these models with simulation tools such as the Model Regression enables the engineer to quickly customize and adjust the model to match actual plant performance.
Please contact us if you have any questions or feedback on how to best utilize the membrane unit operation within the Symmetry process platform.
Roberto Guillén, Engineering Software Developer, VMG Calgary