Figure 4. (A) CAD model of the portion of the MCD system contained within the vacuum can. (B) CAD model of the custom-built condenser. (C) Image of the MCD system with the vacuum can lid lying on a flat surface, when running the orientation is inverted with the column hanging down from the lid.
The incoming air is dried, and CO2 removed using adsorption cartridges containing 13X molecular sieve material. Mass flow controllers are used to control the flowrate of the incoming air and the bottoms stream exiting the reboiler. The condenser—shown in Figure 4(B)—contains a heat transfer surface custom-machined from a copper block contained in a stainless-steel housing. It was situated around and attached to the cold head of the cryocooler, a Sunpower CryoTel® GT 16 W unit. During testing, the cryocooler cold head operated around 77 K (‑196°C), at which temperature it provided 16 W of lift (per manufacturer specifications). The feed entered the bottom of the condenser and the nitrogen rich gas exited to top of the condenser.
To provide thermal insulation, the heat exchanger, column, and condenser were all enclosed within the vacuum can. The pressure inside the can was maintained below 10‑4 torr, and the pieces of equipment inside the vacuum can were each wrapped in 5‑10 layers of 500 DM cryogenic laminate (multi-layer insulation). Swagelok and flange fittings were used for all connections. Type K thermocouples were placed on the inlet and outlet of the various components to monitor temperatures throughout the system. Opto 22 software was used to monitor the temperatures, flowrates, and pressures inside the vacuum can, as well as the pressure of the process fluid in the distillation system. The power inputs to the cryocooler and the reboiler were also modulated using Opto 22. A vacuum pump was connected to the reboiler for continuous removal of liquid. The flow rate was measured by a mass flow controller and verified with a Mesa Labs Definer 220‑L DryCal. A small sample of the reboiler vapor was periodically withdrawn and sent to the Dycor residual gas analyzer for compositional analysis. The residual gas analyzer pulls samples at vacuum pressures, so the gas was assumed to be ideal. Consequently, the total pressure \(P\) was assumed to equal the sum of the individual species partial pressures \(P_{i}\), and the species mole fractions \(y_{i}\) were assumed to be directly proportional to the species partial pressures (i.e.,\(y_{i}=\frac{P_{i}}{P}\)).

Process Flow

Atmospheric air was first processed through the adsorption cartridges. The CO2-free dry air is then pre-cooled in the recuperator inside the vacuum can. The cooled dry air then flows into the condenser and was partially condensed. The uncondensed vapor—mostly nitrogen—flows back through the heat exchanger. The condensed liquid collects in the bottom of the condenser and then flows into the column, passes down through the column, and is eventually collected in the reboiler. Along the way, it interacts continuously with the upward moving vapor produced in the reboiler. The less volatile components—primarily oxygen and argon but also other heavy components such as xenon—became concentrated in the liquid, while the more volatile nitrogen becomes concentrated in the vapor. A portion of the liquid in the reboiler is periodically withdrawn and sampled.

Operation

During startup, the reboiler is turned off while waiting for the system to cool down to operating temperatures. This cooling process took a couple of hours. All three columns were tested under similar conditions with maximum reflux to compare separation efficiency. For these tests, the feed air flow rate was fixed at 1 SLM (standard volumetric flow) and the flow rate of the bottoms liquid product was kept below 0.01 SLM (also standard volumetric flow). The AMPH device was also tested to determine the maximum flow of oxygen-rich product (minimum mole fraction 0.90) that could be produced by the equipment. For these tests the feed air flow rate was fixed at 5 SLM and the bottoms liquid product flow was adjusted from 0.140 to 0.400 SLM in increments of 0.054 SLM.

Modeling

The MCD system was modeled using process simulation software, the flowsheet for which is shown in Figure 5. This is a very simple model, consisting of only a heat exchanger to model the recuperator and a distillation column to model the MCD column.