Air Separation and Liquefaction

by Nils Tellier, PE, President, EPSIM Corporation (CSA CSM)
nils@epsim.us
All illustrations courtesy EPSIM Corporation

Background History of Air Separation and Liquefaction

This section builds on a rich history of methods to develop deep refrigeration and cryogenic liquefaction during the 19th Century. You are encouraged to read Cryo Central’s History of Cryogenics for more information.

Industrial refrigeration in the 19th Century was largely driven by butchers and brewers. Charles Tellier pushed the envelope to achieve temperatures of -79°C (-110°F) in 1866 by means of cascade refrigeration with NH₃ and SO₂. Carl Von Linde founded Gesellschaft für Lindes Eismaschinen Aktiengesellschaft (Linde AG) in 1879, liquefying CO₂ for the Guinness Brewery among other customers. He later developed and presented an industrial-scale 3-liter per hour air liquefier at the Paris World Fair in 1895. The liquefier process was entirely based on Joule-Thompson (adiabatic) expansion. Von Linde’s commercial liquefier compressed air to 200 Bars and expanded it, after pre-cooling against exiting vapor, to 30 Bars. It took about a week to cool down the liquefier and he was able to liquefy 30 percent of the process air right below the critical point.

In 1902 George Claude improved the refrigeration cycle by adding an expansion engine and a mechanical brake to reduce the enthalpy of the air vapor, and therefore its temperature, at nearly constant entropy. The idea had first been proposed by Carl Siemens in 1857 and attempted by Ernest Solvay in 1885. The isentropic expansion resulted in a dramatic reduction in cooling time and operating pressures, while increasing the throughput of liquefied air. George Claude founded Air Liquide that year.

At the dawn of the 20th century, cryogenically liquefied air had become an industrial commodity. The oxygen produced was primarily used for oxy-acetylene welding and cutting applications. Industrial oxygen was produced by distillation of air at cryogenic temperatures. The brewing and alcohol industries had a marked influence in the early days of cryogenics. The former shared their knowledge on heat exchangers and temperature control; the latter on fractional distillation. One caveat with conventional distillation is that the condenser at the top of the distillation column has to operate at a colder temperature than the most volatile component—nitrogen in our case. This was an expensive proposition and the first air separation plants ran at high pressure to raise the condensing temperature. Carl Von Linde modified the conventional distillation process in 1910 and invented the double column process as we know it today. He cut the distillation column in the middle and combined the condenser and reboiler into a single heat exchanger. The lower column operates at high pressure with a vapor feed and the upper column at low pressure with a liquid feed.

One more significant refinement leading to the modern cryogenic air separation and gas liquefaction industry was to replace the piston expansion engine by a more efficient turbo-expander technology. Credits go to Guido Zerkowitz (Linde AG) in 1935 and Pyotr Kapitsa in 1939.

Modern Air Separation

Air is a colorless and odorless gas at ambient temperature and pressure conditions. Moisture combines with atmospheric air in different amounts based on temperature, pressure and relative humidity. The following composition of dry air is approximate, due to the rapid rise in atmospheric CO₂ levels:

There are different techniques to extract—or concentrate—a component from the air. Hollow fiber membrane systems and pressure-swing adsorbers are two non-cryogenic means to extract nitrogen; vacuum-swing or pressure-vacuum swing adsorption, and ceramic membrane ionization are also used to produce oxygen. However, these processes are outside the scope of our presentation.

Cryogenic air separation remains the most efficient means of separating air into high purity gases and liquids at an industrial scale. The following flowchart outlines the basic process steps in cryogenic air separation:

Air Filtration and Compression

Atmospheric air is first filtered to remove airborne particles before it enters the Main Air Compressor. The compressor is the main energy user in an air separation plant. It is a significant determinant of electricity costs and costs per ton of the products oxygen, nitrogen or argon. The purpose of the compression step is to control the mass flow of air feeding the cryogenic distillation column at the condenser/reboiler operating pressure. When the air separation plant is in steady operation, it is the condenser/reboiler that pulls the air. The compressor then regulates the air flow and the balance act determines the operating pressure.

Compressors used in air separation plants are oil-free multistage intercooled centrifugal units. Depending on site location and plant production, the compressor may have three or four stages, although additional booster stages may be integrated in the compressor. The compressor comprises a motor-driven bull gear and small pinions located around the bull gear driving turbines at high speed for each stage of compression. The turbine wheels are housed in volutes so that the incoming air velocity is gradually reduced and converted to pressure and heat. Heat of compression and condensing moisture are removed through the water-cooled intercoolers. Compressors used for air separation are oil-free. It is of paramount importance that the air remains absolutely free of oil contamination. Any trace of oil entrained in the air would plug and possibly damage the cryogenic equipment, and present a risk of spontaneous explosion when in contact with oxygen.

Centrifugal compressors will surge when flow is reduced below a certain point at a given pressure. To prevent this, a surge controller measures in real time two independent variables, such as the discharge pressure and the flow or motor amps. When the controller sees that the compressor approaches its surge curve, it will open a recycle valve—a vent in the case of an air compressor—to increase flow. A second control, used to operate the entire air separation plant, is the air flow control from the compressor’s inlet guide vanes. The air flow variable is mass flow, where the volumetric air flow measurement is compensated in real time for temperature and pressure. Flow and pressure instrumentation on the compressor must be installed with wet saturated process air in mind, in a manner that moisture does not accumulate in low points of the instruments’ tubing.

Air Purification

The purpose of the purification step is to remove all traces of moisture and CO₂ from the high pressure oil-free air. There are two alternatives to accomplish this: front end purification by adsorption, and reversing heat exchangers (REVEX).

REVEX combines air purifying and cooling by means of switching passages in the cryogenic heat exchangers. Wet air is dried and cooled to near its liquefaction temperature against exiting oxygen and nitrogen. Moisture and CO₂ ice up and deposit in the warm section of a switching exchanger passage, gradually building up restriction. Cold air leaving the exchangers is near its liquefaction temperature, bone dry and free of CO₂. After several minutes, valves switch the incoming wet air to a clean set of exchanger passages and direct waste nitrogen to clean up the fouled passages. The REVEX technique has some drawbacks. It is noisy, does not remove such contaminants as propane and acetylene, and has a higher parasitic loss of air, translating into higher electric consumption from the compressor. REVEX is advantageous for its lower capital cost, simplicity of operations and fast start-up.

In contrast, front-end purification is considered state-of-the-art for air purification. It entails two or more adsorber vessels filled with activated alumina and molecular sieve. Alumina retains all the moisture from the air. The molecular sieve, typically 13X, removes the CO₂ and several hydrocarbon contaminants. One vessel is one line cleaning the air while the other vessel is being desorbed with dry nitrogen byproduct from the air separation plant. The vessels are sized to switch every four to eight hours. Although molecular sieve retains water, perhaps better than alumina, it only desorbs it at very high temperatures and requires a plant shutdown for an exceptional reactivation. High pressure air leaving the front-end purification is absolutely free of moisture and CO₂, and is called Clean Dry Air (CDA).

The “Cold Box” and Cryogenic Distillation

The high-pressure, oil-free air, cleaned of moisture and CO₂ traces, enters a “cold box” which houses the super-insulated cryogenic equipment. The cold box includes cryogenic heat exchangers, distillation columns, the turbo-expander and all the cold piping, carefully insulated with rockwool or perlite, and boxed in a carbon-steel outer jacket. Cryogenic cold boxes often appear as tall white structures, deceptively void of any outside feature.

Air entering the cold box is cooled against exiting product oxygen and nitrogen gases in multi-pass cryogenic exchangers called “main exchangers.” Modern cryogenic exchangers are made of brazed aluminum plates, with alternate layering of passages between the air, nitrogen and oxygen to ensure efficient heat exchange and to support high pressure differentials. As mentioned earlier, REVEX exchangers have switching passages for air and waste nitrogen whereas exchangers for front-end purification plants only have fixed passages. The main exchangers are separated between a “warm” exchanger and a “cold” one.

To further compensate for parasitic heat leaks and liquid production, additional refrigeration is provided by a turbo-expander following a Modified Claude refrigeration cycle. The expander inlet is located at the “midpoint” of the main exchangers, namely at the piping junction between the warm and the cold exchangers.

Plant efficiency can be measured by the exchangers’ approach temperatures at the warm end and at the cold end. A small temperature difference at the warm end shows good thermal efficiency, or low heat losses. Likewise, a small temperature difference at the cold end is a sign of good oxygen recovery. However since the air enthalpy does not add up to that of the product gases (oxygen and nitrogen), the cold end and warm end approach temperatures are mutually disruptive: reducing the cold end temperature difference increases the warm end one and vice-versa. A benefit of the Claude refrigeration cycle is that the expander increases the temperature difference at the midpoint and thus closes the approach temperatures simultaneously at the process warm and cold ends.

The air leaving the cold end of the main exchangers is close to its liquefaction point of -160°C (-260°F). It enters the lower distillation column, also called the high-pressure column or HP column—and nowadays referred to as Medium-Pressure column. Air vapor entering the HP column is pulled upward across distillation trays or packing by the condenser/reboiler located at the top of the column. Since the saturated air density increases by a factor of 35 when it turns from vapor to liquid, it is easy to visualize the force of the convection happening inside the HP column. The rising vapor is gradually stripped of oxygen by the falling liquid, by fractional distillation across the column trays or packing sections, such that the top of the HP column sees only pure nitrogen. Nitrogen production, gaseous or liquid, comes from the HP column head. The column sump holds a “rich liquid” composed of nitrogen, argon and approximately 35 percent oxygen. The rich liquid is expanded into the upper column as feed and a stream of liquid nitrogen from the top of the HP column is sent to the upper column as lean reflux.

The condenser/reboiler located at the top of the HP column consists of a heat exchanger immersed in low-pressure boiling liquid oxygen (LOX), with passages open to the HP column head. A variation introduced in the early 1990s, called falling film reboiler, comprises an exchanger open-ended on the low pressure side and fed by falling liquid and pumped LOX. Heat exchange with the condensing high-pressure nitrogen is improved by the turbulent LOX flow and increased liquid contact, resulting in better plant performance and stability of operation.

Gibb’s Law tells us that there is a direct relationship between pressure and temperature for a two-phase fluid of known composition. Hence by controlling the pressure at the top of the HP column we also control the condensing temperature of the HP nitrogen. Conversely, controlling the pressure at the sump of the LP column gives us control of the boiling temperature of the LOX bath. In summary, the temperature difference—and duty—at the condenser/reboiler is a function of the pressure ratio between the lower and upper columns.

This brings us to the upper column where further distillation takes place to purify the oxygen and, as the case may be, extract argon for further purification. The upper column operates at a low pressure, as close to atmospheric as possible. It is called the low-pressure or LP column. If we take into account the pressure drop across the main exchangers and some driving pressure to utilize the nitrogen byproduct for the regeneration of the air cleaning system, pressure in the LP column determines the HP column pressure required for the condenser/reboiler to work, which in turn determines the main air compressor discharge pressure. It is therefore more economical energy-wise to operate the LP column at the lowest possible pressure.

Similar to the HP column, the upper column separates components by fractional distillation across trays or packing. This is last similarity between the two columns. In the upper column, the feed is the rich liquid from the HP column. The liquid falls through the distillation trays and is gradually stripped of the more volatile nitrogen by the rising vapor. The rising vapor comes from boiling LOX around the condenser/reboiler. As the vapor climbs through the trays or packing, oxygen molecules condense and fall back to the sump. The vapor at the top of the LP column is low-purity nitrogen partially recondensed by the lean reflux sent from the top of the HP column. Through the distillation process, the falling liquid becomes gradually richer in oxygen and accumulates around the condenser/reboiler as high purity LOX. The rising vapor ends up as low-purity nitrogen at the top of the column, also the coldest point of the whole plant with a temperature around -195°C (-320°F). Nitrogen exiting the top of the LP column, often called waste nitrogen, is a byproduct of the air separation because it holds a small percentage of oxygen. The waste nitrogen is first used to sub-cool the lean reflux and then the rich liquid, to reduce their flashing before they are expanded into the LP column. After that the waste nitrogen enters the cold end of the main exchangers and is gradually warmed by cooling incoming HP air. Warm waste nitrogen is used to regenerate the CO₂ and moisture saturated adsorbant in the front end purification before being vented back to atmosphere. Alternatively, the waste nitrogen is routed through the fouled passages in the REVEX exchangers.

Oxygen is collected as a boiling liquid around the condenser/reboiler in the sump of the LP column. A small stream of LOX must be constantly drained from the sump to avoid hydrocarbon saturation and the risk of a spontaneous explosion. Traditionally, high-purity gaseous oxygen has been produced by letting the oxygen vapor exit the bottom of the LP column, enter the main exchanger’s cold end and gradually warm up against incoming air. The low pressure oxygen gas exiting the cold box is then compressed to meet client process requirements.

In the 1990s became increasingly common to pump the LOX at high pressure into the main exchangers, where the oxygen is vaporized and warmed against incoming air. This process is very efficient, because liquid pumping does not create heat of compression and a two-phase fluid cools the incoming air by leveraging the latent heat of vaporization in addition to its sensible heat. Figure 9 illustrates the cryogenic equipment inside a cold box for a gaseous oxygen plant with pumped LOX.

What about argon, which makes up nearly one percent of the air? Argon is particularly challenging because it is very close to oxygen in molecular weight and bubble point. The O₂-Ar phase diagram is very flat, which requires a very tall distillation column. Adding insult to injury, argon freezes and blocks process equipment at relatively “warm” temperatures. Squeezed between the boiling LOX and the cold reflux in the LP column, argon concentrates to about 30 percent near the center of the column where it is extracted and further purified in separate columns. Modern plants post 1992 are able to purify argon in a single column located next to the LP column. Argon is produced and marketed as a liquid.

In summary, the cryogenic separation of air to produce oxygen and nitrogen requires constant and simultaneous heat and mass balances. Cooling down an air separation plant from ambient conditions can take up to two days, usually following a week of cleaning or “deriming” the cold box internals. Loss of efficiencies in the process translate into higher power consumption by the main air compressor and thus higher production costs. Modern and retrofitted plants are highly automated in order to maintain efficient and robust operations with runtimes between scheduled shutdowns upwards of 35,000 hours.

A closing remark, perhaps the most important topic, is that of safety. Cryogenic temperatures, nitrogen and oxygen present safety hazards. Additionally, process air contamination from oil and hydrocarbons can combust spontaneously with oxygen. There are many safety features in the design and fabrication of an air separation plant, including hydrocarbon detection and control, oxygen cleanliness, oxygen velocity control and choice of materials of construction.