8.37

Separation Controls, Air M. H. MACCONNELL

(2005)

INTRODUCTION This section will focus on the different processes and associated control strategies that are used in the air separation processes. The nominal composition of air is provided in 1 Table 8.37a. Air separation processes primarily produce oxygen, nitrogen, or argon in either the gas or the liquid phase. This section includes discussions of process controls that are associated with three air separation technology areas: membrane, adsorption, and cryogenic distillation in the manufacture of oxygen, nitrogen, and argon. The typical product purities and production rates of the three highlighted technologies are provided in Table 8.37b. Production volume, product mix, and product purity determine the design of the individual air separation facilities. From a production volume point of view, oxygen is the most significant product. Oxygen is the third-most-produced industrial chemical in the world today; the majority of the production plants use 2 cryogenic air separation facilities. The consumption of oxy3 gen by market is illustrated in Table 8.37c.

TABLE 8.37b Production Attributes of Air Separation Technologies

Technology

Purity Range

Production Range Tons per Day (TPD)

Nitrogen membrane Nitrogen PSA

5%–1000 ppm O2 in N2 5%–500 ppm O2 in N2 200—5 ppm O2 in N2

8–10,000 SCFH 4–116 MSCFH @ 95% N2 0.4–12 MSCFH @ 5 ppm

2-Bed oxygenVSA

90–93% O2

15–110 TPD

Single-bed oxygen VSA

90–93% O2

4–5 TPD

Cryogenic distillation

99.9% O2 < 0.5 ppm O2 in N2

60–4000 TPD

(PSA) process. In either case, the principle of operation involves creating physical conditions in a vessel that permit an adsorbent material that is packed in the vessel to temporarily remove or adsorb the impurities of interest. Once the adsorbent bed is full, the physical conditions in the vessel

ADSORPTION TECHNOLOGY The Process The adsorption technologies used are either the vacuum swing adsorption (VSA) or the pressure swing adsorption

TABLE 8.37c Consumption of Oxygen by Market United States Market

TABLE 8.37a Nominal Composition of Dry Air

Component

Percent by Volume

Component

Parts per Million by Volume (ppmV)

Users %

Western Europe Market

Users %

Primary metals production

49%

Primary metals production

40%

Chemicals and gasification

25%

Chemicals and gasification

27%

Petroleum refineries

6%

Fabricated metal products

6%

Nitrogen

78.084

Carbon dioxide

Oxygen

20.946

Neon

18.2

Welding and cutting

6%

Health services

6%

Helium

5.2

6%

1.1

Petroleum refineries

5%

Krypton

Clay, glass, and concrete products

Xenon

0.09

Health services

4%

Pulp and paper

4%

Methane

1–15

Pulp and paper

2%

Water treatment

3%

0–0.5

Water treatment

1%

Other

4%

Other

1%

Argon

0.934

Acetylene Other hydrocarbons

350–400

0–5

2123 © 2006 by Béla Lipták

2124

Control and Optimization of Unit Operations

are switched to a different state to permit the initiation of desorption of the impurities from the adsorbent bed. Adsorption processes depend on the affinity of certain natural and synthetic materials to preferentially adsorb the nitrogen or the oxygen molecules. Zeolites, for example, are aluminosilicates that have nonuniform electric fields in their void spaces that cause preferential adsorption of polar molecules relative to nonpolar ones. For this reason, when air is passed through the zeolite, nitrogen is more strongly adsorbed than are oxygen or argon molecules. As air passes through the bed, nitrogen is retained and an oxygen-rich stream is produced in the vessel. Carbon molecular sieves can also be used as the air separation media. In carbon molecular sieves, the pore sizes are approximately the same size as the air molecules. Because oxygen molecules are smaller than nitrogen molecules, the oxygen molecules diffuse more quickly into the pores than do the nitrogen molecules. This is a kinetic adsorption process. In short, carbon molecular sieves are selective for oxygen, and zeolites are selective for nitrogen. Control of the Adsorption Process Typically, programmable logic controllers (PLCs) are used to control the VSA or PSA units. Analog, on/off, and timer functions are all utilized to control product purity, to operate the compressor and regeneration valves, and ultimately to obtain maximum productivity and efficiency. Locating the Adsorbate Front If the position of the adsorbate front and the associated mass-transfer zone along the axial length of the adsorber is known, it can be very useful, because it determines the composition and ultimate product purity as well as the required timing of the regeneration step. Because of the temperature and pressure variations in this process, the commercially available oxygen sensors are not useable for direct on-line measurement. Therefore, the operating parameters are established mathematically from pilot studies that infer the composition on the basis of energy balances and axially distributed temperature and pressure 4 measurements. After the product leaves the unit, its quality is analyzed to ensure that product specifications are met. This analysis is also used to determine regeneration valve timing based on the pilot scale-up process and operating experience. Vacuum Swing Adsorption Vacuum swing adsorption is principally used in the generation of oxygen gas in the purity range of 90–93%. Manufacturers of VSA units provide adsorbent materials that adsorb nitrogen at atmospheric and slightly elevated pressures. This permits generation of an enriched oxygen stream that can be accumulated in a receiver. Depending on the amount of adsorbent, the bed will be allowed to continue adsorption of nitrogen for a fixed period of time until it is nearly saturated. After

© 2006 by Béla Lipták

Desorption phase

Purge check valve 10 psig spring ASL FI

PI

Adsorption phase

PI

AAL % O2 90–93% Oxygen

O2 Buffer tank

Adsorber tank

KY Timer controls Adsorption phase

Desorption phase Silencer Inlet Vent

Blower

FIG. 8.37d Components and controls of a small-scale vacuum swing adsorption unit. (Courtesy of Air Products and Chemicals, Inc.)

this, the beds are switched to vacuum, and the desorption of nitrogen occurs and the nitrogen is vented. In a small-scale VSA oxygen generator, a timer (KY in Figure 8.37d) controls the blower that, during the desorption phase, generates the vacuum in the adsorber bed, causing the nitrogen, moisture, and other contaminants to be desorbed and vented. When the vacuum level in the adsorber bed is low enough, the adsorber vessel is purged with oxygen to sweep the remaining nitrogen from the vessel. This purge continues for a timed period, after which the automatic fourway valve returns the system into the adsorption phase, and the blower pushes fresh air into the adsorber bed. When the pressure in the adsorber bed exceeds the pressure in the oxygen storage tank, oxygen will start flowing into the storage tank and will continue to do so until the feed timer times out and the cycle starts over again. This process is depicted in Figure 8.37d. Paramagnetic oxygen analyzers (single or redundant) are typically used to continuously monitor the product quality and to alarm (AAL on Figure 8.37d) and initiate automatic interlocks to trip the equipment (which protects the customer by switching the oxygen supply to a backup), if purity falls below the 93–95% range. Larger-scale VSA units operate in a similar manner. Pressure Swing Adsorption The PSA technology utilizes co-adsorption, a phenomenon where both nitrogen and oxygen are adsorbed, but they are

8.37 Separation Controls, Air

2125

HIV PT Adsorber A

Vent

O2 AIC

Adsorber B

Nitrogen buffer FO

I

N2 to Customer

FC FT

Vent

PT

PT

TE

TE

Air in Compressor

FIG. 8.37e Process and instrumentation diagram of a pressure swing adsorption-type nitrogen generator. (Courtesy of Air Products and Chemicals, Inc.)

adsorbed at different rates. The process utilizes a two-tank (or tower) arrangement (see Figure 8.37e) in which two primary processing steps, adsorption and transfer, are performed. Unlike a desiccant-type adsorber that removes essentially all of the target molecules, the PSA carbon molecular sieve (CMS) takes advantage of the difference in adsorption rates (kinetic adsorption) to achieve the separation objectives. The production rate of the PSA process can be limited by the adsorptive capacity of the adsorber, or when the bed is very cold, by the rate of desorption of oxygen from the carbon molecular sieve. The PSA units are often installed out of doors and need to be retuned for winter and summer operations, if ambient temperature conditions are substantially different. A simplistic explanation of this temperature sensitivity is that hot molecules vibrate faster than cold molecules and, therefore, do not adsorb as readily into the carbon pores when hot. When the beds are cold, the opposite occurs, and therefore, it takes longer for the molecules to desorb from the carbon pores. Design Considerations The carbon in the CMS is typically an extruded pellet approximately 2 mm in diameter and 1 cm in length. The carbon in the CMS bed is soft and can easily be pulverized into dust by agitation. Therefore, the bed must not be agitated. If the carbon pellets turn to dust and the dust is vented, this is uneconomical because the CMS is relatively expensive and because the carbon dust shortens the life of the valve seats. Therefore, PSA vessel sizing and gas flow controls must guarantee that the gas velocity through the PSA unit is low, on the order of 1–5 ft/s. The product purity is a function of the gas velocity, the timing cycle of the tower, and the

© 2006 by Béla Lipták

CMS selectivity grade. The sizing of the PSA vessel should also consider the required production rate and the CMS efficiency rating. In a PSA-type nitrogen generator, it is also important to control the air-to-nitrogen ratio. If the nitrogen product has to be 0.5% pure, an air-to-nitrogen ratio of 3:1 is acceptable, whereas if high-purity (5–10 parts per million [ppm] O2) nitrogen product is required, the ratio of air to nitrogen should be around 5:1. The PSA vent dump volume largely impacts the air-tonitrogen ratio. It is sometimes possible to increase efficiency by using a little more air or cycling the beds faster. Process Measurements The product purity is generally measured with fuel cell-based electrochemical oxygen analyzers (for a detailed description of the different oxygen analyzers and their features, see Section 8.42 in Chapter 4 in the first volume of this handbook). For critical applications, it is often appropriate to install two oxygen analyzers in parallel to provide redundancy and improved reliability. Given that the PSA units are subject to wide ambient temperature variation, it is important that the oxygen analyzer be provided with good temperature compensation. The PSA unit should always be provided with the capability for remote monitoring of the process conditions and of the product quality. Because many of the PSA units are located in remote locations, the remote monitoring of PSA installations is also used to dispatch maintenance crews from a central location as needed. As shown in Figure 8.37e, the analog measurements typically include the measurement of product composition, product flow rate, compressor discharge (feed) pressure and temperature, N2 supply pressure to the customer, product

2126

Control and Optimization of Unit Operations

Water removal

Air inlet

I

Fine aerosol removal PDI

PDI

Compressor

I

I

PSL TSL TIC PT

Carbon filter

TE

TE

LS Superheater

I

Drain

QIT Vent

QIC

PCV FI

FT

I

% or ppm O2 Analyzer

FC I

N2 Product to customer

BPR

Membrane modules

PT TE

PT FO

O2 rich permeate

N2 Receiver

Vent

FIG. 8.37f Process and instrumentation diagram of a pressure membrane-type nitrogen generator. (Courtesy of Air Products and Chemicals, Inc.)

buffer tank pressure, and ambient temperature. A number of on/off status measurements are also made to monitor valve positions and compressor status. Programmable logic controllers are used to control the adsorption–desorption cycle of the plant, which essentially involves opening and closing the eight or so on/off valves on a time cycle sequence. These valves are either ball or butterfly valves, depending on the line size. The valve actuators are typically double-acting air cylinders, except for the fail closed (FC) nitrogen product valve and the fail open (FO) nitrogen product vent valve, which are spring-loaded to provide the required safe failure positions in the event of a product quality excursion or unit shutdown.

MEMBRANE AIR SEPARATION Membrane systems are used for the continuous separation of components in mixed gas streams. The production volume and the achievable purity are generally lower than from adsorption or cryogenic processes. The most commonly used membrane systems are in the production of nitrogen, although membrane separation is also used in gas drying applications where water molecules permeate very quickly through the membrane material. The predominance of nitrogen generators as membrane separators is because the smaller size of the oxygen molecules guarantees the superior permeability of oxygen through the membranes. As a result of this difference in permeability, membrane systems can only be used to produce oxygen in a concentration range of 25–50%, whereas 99.9% nitrogen purity is achievable.

© 2006 by Béla Lipták

Process Description The membrane separation process starts with an air compressor (typically, an oil-flooded screw design) that compresses air to 100–200 psig (Figure 8.37f). The moisture and aerosols are removed from the compressed air by a carbon filter, and then it is superheated to prevent any condensation in the membrane modules. The superheated gas passes through a bank of parallel membranes, where oxygen permeates through to the low-pressure side of the membrane and is either vented or recycled. One or two electrochemical gas oxygen analyzers (QIT in Figure 8.37f) continuously sample the high-pressure nitrogen product as it leaves the membrane modules. The product quality is controlled by the manipulation of the product flow control valve that is feeding the accumulation vessel. If the oxygen concentration in the product exceeds the allowable limits, the FC on/off valve on the receiver inlet trips closed and the FO on/off valve vents the product, until purity is reestablished. A back-pressure regulator (BPR) on the membrane system outlet prevents drawing more product than is available at the specified purity. Flux and Selectivity The membrane separation process is based on the difference in permeation or diffusion rates of nitrogen and oxygen molecules through a polymeric membrane that separates the high- and the low-pressure streams. Flux and selectivity are the two properties that bear directly on system design. Flux determines the membrane surface area, and for a given gas, flux is a function of the partial pressure difference and inversely of the membrane thickness. In addition, each

8.37 Separation Controls, Air

type of membrane material has a characteristic proportionality constant called permeability. Selectivity is the ratio of the permeability of the gases that are to be separated. In other words, permeability is the ratio of the speeds at which the respective gases diffuse through the membrane. Membrane System Sizing Sizing of the membrane system is based primarily on two parameters: the length of the membranes in each module and the number of modules installed in parallel. For a given amount of flow, the length of the membrane modules tends to determine the purity of the product. In general terms, the product purity is a function of the flow rate through the modules, while the number of parallel modules determines the volumetric production capacity. The efficiency of a particular membrane module is a measure of the amount of nitrogen produced from a unit of air consumed. The efficiency and production capacity depend on the overall permeability and selectivity of the membranes, because there is always a certain amount of nitrogen that is lost with the oxygen. Therefore, a given membrane combination will produce a known amount of nitrogen product of a specific purity, if the unit is operating normally. Physical Description The membranes are hollow polymeric fibers. The diameter of the fibers is on the order of some hundreds of microns. The fibers are composed of any of a number of polymers, including polysulfones, polycarbonates, and polyimides. The fibers are bundled together in a fashion similar to that of a shell and tube heat exchanger The bundle of fibers of specified length is effectively sealed on both ends to tube sheets. During operation, the higher-pressure air passes through the inside of the fibers, and oxygen is permeated to the outside (the shell side). The entire membrane adsorption system is packaged in a cabinet with operator controls conveniently located on the front panel. Cabinets are normally installed indoors at the customer location.

CRYOGENIC AIR SEPARATION The term cryogenic refers to any material or process that operates at very low temperatures, typically below 120°K (–243°F, or –153°C). The use of distillation to separate air into its component is a mature technology. The use of cryogenic engineering in the liquefaction of air was initially used at the beginning of the 20th century. Cryogenic processes produce oxygen, nitrogen, argon, and other rare gases from atmospheric air. The selection from among these processes and cycles depends on the product volume, phase, and purity requirements as well as on the capital cost and energy consumption constraints. An advantage of cryogenic over adsorption processes is their ability to coproduce oxygen, nitrogen, and argon as well

© 2006 by Béla Lipták

2127

as their ability to liquefy a portion of the total output. Storage of liquefied product in vacuum-insulated tanks serves as backup for gaseous product systems during outages or for the liquid filling of truck trailers for merchant sales. This section will describe the low-pressure (LP) cycle and the associated process controls in a typical operation. Cryogenic processes utilize the same distillation principles as do higher temperature processes, except that the distillation equipment, heat exchangers, and associated piping are located inside an insulated enclosure referred to as a cold box or can, depending on whether the enclosure is rectangular or cylindrical in shape, respectively. Front-End Air Purification The cryogenic air separation process consists of two main operations. The first is the front-end purification of the air feed before it is sent to distillation. This step is described in Figure 8.37g. Once the air is purified, the air stream is split in two and both streams are sent to the cryogenic distillation (cold box) operation, which is shown in Figure 8.37h. The main heat exchanger is shown on both figures, completely on Figure 8.37h and partially on Figure 8.37g. Feed Air Preparation The most commonly used process for the production of oxygen and/or nitrogen is the low pressure or double column cycle. In this process, the main feed air is compressed to a range of 60–200 psia, the heat of compression is removed by a cooler, and the high boiling impurities (H2O, CO2, N2O, and certain hydrocarbons) are removed by an adsorption system (Figure 8.37g). The purified air is divided into two streams, one of which is compressed with a booster compressor (shown on Figure 8.37h, which is a continuation of Figure 8.37g) to a range of 100–1200 psia. Both feed air streams are cooled in the main heat exchanger (schematically shown in both Figures 8.37g and 8.37h) by indirect heat exchange with cryogenic return streams. Reversing Heat Exchangers Prior to 1980, most air separation plants used reversing main heat exchangers to remove the high boiling impurities such as moisture and carbon dioxide (CO2) from the feed air. In these early plants, the moisture and CO2 in the feed air to the cold box were allowed to freeze in the main heat exchanger that was cooled by the counterflow of cryogenic gas. This accumulation was allowed to continue for a set time period or until the exchanger began to increase its pressure drop or began to lose some of its heat-transfer capability. At that point, the flow direction through the exchanger was reversed, and dry, CO2-free waste gas was sent through it to purge the accumulated impurities to an atmospheric vent. The reversing heat exchangers were very large and tended to require substantial maintenance due to leaks, which were caused by frequent and severe temperature cycling. Moreover, unlike the type of front-end adsorber systems shown in Figure 8.37g, the reversing exchanger also suffered from an

2128

Control and Optimization of Unit Operations

H2O CO2 FIC QIT QIT

Air to booster (Continued on Fig. 8.37h compressor as stream #1)

Molecular sieve controls PT To timed actuated valves

Air inlet

PT TSA

TSA vent

or PSA Vent TS

or PSA TS

TIC QIT H2O Steam

FIC If TSA

>

PY

FIC

Chilled water PIC

= Silencer LIC

Cooler Multi-stage centrifugal compressor with intercoolers

CWR If TSA

Main heat exchanger Air to HP column (#2 on Fig. 8.37h)

Regeneration gas from LP column (#3 on Fig. 8.37h)

HP - High pressure LP - Low pressure PSA - Pressure swing adsorption TSA - Temperature swing adsorption

FIG. 8.37g The equipment and main instrumentation used in the feed section of the cryogenic distillation process. (Courtesy of Air Products and Chemicals, Inc.)

inability to remove acetylene and other particular hydrocarbons, which could then concentrate in the liquid oxygen and, thereby, cause an unsafe condition, having the potential for energy release. To address this safety concern, plants that use reversing heat exchangers continuously process all their liquid oxygen in the distillation system through adsorption beds to remove the hydrocarbons. The adsorbers are then periodically regenerated with nitrogen. Feed Air Adsorption Systems Today, most cryogenic air separation plants are built with an adsorption system front end that features either temperature swing adsorption (TSA) or pressure swing adsorption (PSA), similar to the process illustrated in Figure 8.37g. In either case, two or more vessels packed with alumina beads or with a molecular sieve are used, which packing has an affinity for high boiling impurities such as moisture, carbon dioxide, nitrous oxide, acetylene, and a number of other impurities. Removal of these impurities is critical, because if they freeze in the cryogenic section of the plant, they obstruct the flow passages. Both the TSA and the PSA processes consist of two or more vessels packed with adsorbant beds. During operation, one vessel is on-line and is removing the impurities from the compressed feed air, while the other vessel is being regenerated by being purged with dry and CO2-free waste gas, which

© 2006 by Béla Lipták

is then vented to atmosphere. There are a number of operational differences between the two technologies that will be described in the paragraphs that follow. Programmable control systems provide the required valve switching logic for the operation. This logic is based on a variety of process measurements and timing functions. Because the front-end system does not completely remove all impurities, the distillation system must be periodically shut down and defrosted. This defrosting operation is either scheduled on a preventive maintenance basis (typically 3–5 years) or scheduled sooner if dictated by deteriorating operating conditions such as increasing pressure drops across heat exchangers. Temperature Swing Adsorption The TSA process consists of two or more vessels that are packed with a bed of adsorbant material. These vessels are switched from on-line to regeneration modes every 5 to 12 hr, depending on the design of the plant. While one vessel is on-line and is removing impurities from the compressed air feed, the other is being regenerated by a countercurrent purge flow of high-temperature dry and CO2-free waste gas (stream #3 on Figure 8.37g). The TSA waste gas heater uses whatever heat source is most economical. This could be low-pressure steam, electrical, or direct-fired natural gas heat. The TSA beds have greater affinity for adsorption when operated at lower temperatures. For this reason and to enhance removal efficiency, the compressed feed air is cooled with a direct contact after

8.37 Separation Controls, Air

2129

FIC FIC

GOX to pipeline

ppm O2

PIC

QIT Booster compressor

Air from adsorption front end (#1)

LIN

GAN

PIC

< 1 ppm O2

% O2 QIT PIC

FIC FIC

LP GAN to reactivation (#3)

Pure N2

After cooler

% O2 LIC QIT

FIC

Waste to reactivation (#3)

> FY

LIN % O2 QIT

PIC

LP GAN to compressor

PIC

Vent

PIC

LPC FIC

Air from adsorbers (#2)

FIC FIC

5%Ar, 95%O2, Trace N2

HP GAN

Main heat exchanger

99+% O2

QIT

~ 95% Ar Crude Argon To Pure Argon Distillation

LIC

Reboiler/ condenser

ppm THC ppm O2 QIT

QIT FIC THC

LIC

QIT

% O2

%O2 QIT

Sidearm Column

PIC

Pure N2

TIC

Liquefied air

Expander

Air

LOX Pump

HPC

30–40% O2

LIC

TIC

FT

LOX to Storage

GAN Crude LOX

LOX to Disposal

Subcooler GAN - Gaseous nitrogen GO - Gaseous oxygen HP - High pressure HPC - High pressure column LIN - Liquid nitrogen LOX - Liquid oxygen LP - Low pressure LPC - Low pressure column

FIG. 8.37h The equipment and main instrumentation used in the cryogenic distillation process in which the pumped liquid oxygen (LOX) cycle produces N2, O2, and crude argon. (Courtesy of Air Products and Chemicals, Inc.)

cooler (DCAC) that uses chilled water that is generated by the cooling effect of the excess cryogenic nitrogen waste gas that is not used to regenerate the beds. By operating at lower temperature, the impurity removal capacity is increased. This is because the required size of the TSA unit increases with water load, which in turn increases with air temperature. Pressure Swing Adsorption The PSA process also consists of two or more vessels that are packed with a bed of adsorbant material. These vessels are also switched from on-line to regeneration modes but they are switched several times per hour, much more frequently than with TSA. While one vessel

© 2006 by Béla Lipták

is on-line removing impurities from the feed air, the other is being regenerated by being purged with dry CO2-free waste gas, which is then vented. Comparison of TSA vs. PSA TSA adsorber beds are most commonly used in the newer air separation plants while PSA is considered a niche application that is most often used in smaller plants. The PSA process is sometimes favored if there is no economical heat source available for regenerating the TSA, such as low-pressure steam or natural gas. The PSA does utilize a simpler, less capital-intensive compressor after cooler, as compared to the more expensive

2130

Control and Optimization of Unit Operations

DCAC, which is most often associated with TSA. The PSA process depends on pressure difference to separate impurities from the air and requires more regeneration gas than does TSA, which uses heat to regenerate. Because the PSA process is often oversized to obtain the required capacity, PSA cycling can result in a variation in the feed airflow to distillation. Finally, TSA provides a secondary benefit if there are airborne contaminants that need to be scrubbed, because its DCAC tower does also provide scrubbing. Cryogenic Distillation (Cold Box) Process The equipment, piping, and basic instrumentation used in the cryogenic distillation (cold box) process is shown in Figure 8.37h. Here, the higher-pressure air from the booster compressor on Figure 8.37h is liquefied and fed through a Joule Thompson (JT)-type expander valve. In Figure 8.37h, the liquefied air stream is shown as a feed to the high-pressure column (HPC), but it could also be fed to either or both the HPC and LPC columns. Stream #2, the lower-pressure air stream, is shown as a feed gas to a point lower in the HPC.

Argon Product Argon is produced at the top of the sidearm column where impurities include oxygen at a range of 1 ppm to 4.0% and nitrogen at the ppm levels. If pure argon is required and the product from the sidearm column contains unacceptable oxygen levels (e.g., up to 4%), then oxygen is removed through additional purification steps. Pure argon is produced either directly from distillation or through a process that uses catalytic deoxidation of a crude argon stream. Waste Gas The last remaining major process stream is the waste stream that is extracted from the upper portion of the LPC and contains principally nitrogen but also some small fractions of oxygen and argon, depending on the recovery characteristics of the plant. This purge stream provides refrigeration for the main heat exchanger and dry, CO2-free regeneration gas for the reactivation of the front-end purifiers (stream #3 in Figures 8.37g and 8.37h). CRYOGENIC INSTRUMENTATION Severe-Service Control Valves

Oxygen Production Figure 8.37h describes the design of the pumped liquid oxygen cycle, the “Pumped LOX” cycle. As the name implies, in a pumped LOX plant, the liquid oxygen (LOX) is pumped from the low-pressure column (LPC) sump to serve other process needs; this will be described in more detail later. Oxygen is purified at the bottom of the LPC and is removed to storage either as gaseous oxygen (GOX) or LOX, depending on the customer requirements. The boiling point and relative volatility of argon is between that of nitrogen and oxygen. In the mid-section of the LPC column, the argon concentration is about 10%, and it is drawn off as feed to a sidearm column that produces crude argon. All the oxygen, which enters with the gaseous air (stream #2 in Figure 8.37h), leaves in the bottoms from the highpressure column as impure or crude LOX, in which the oxygen concentration is in the range of 30–40%. The crude LOX is flashed down to a lower pressure and is either fed to an intermediate stage in the low-pressure column, or if argon is to be recovered, it is used to operate the sidearm condenser and then fed to the low-pressure column as a more vapor-rich stream. Nitrogen Production Nitrogen, having a greater relative volatility than oxygen, concentrates as it rises through the HPC. The vapor from the top of the column is condensed against boiling liquid oxygen in the reboiler/condenser. The condensed overhead is divided into a reflux stream that is returned to the HPC and a reflux stream that is sent to the top of the LPC. Nitrogen is purified at the top of both the HPC and LPC, and this product can be extracted from either column overhead, depending on the process cycle, but is often withdrawn from the LPC overhead to prevent limiting the HPC reboiler/ condenser of reflux.

© 2006 by Béla Lipták

The LOX control valves or air JT valves often operate at pressure drop of 1000 psig or more, where flashing or cavitation normally occurs (Figure 6.1y in Chapter 6 describes some anticavitation valve designs). The valve seats in such severe services must be hardened, using such material as stellited trim. It is also desirable to keep the flow velocities as low as possible to reduce erosion of the valve trim and to minimize noise (see Section 6.14 in Chapter 6 for details). Flashing and cavitation can cause severe erosion of the control valve trim, which can destroy the valve. It is for this reason that special valve designs are required for such services. These valves always require positioners. The outlet area of severe-service valves should be maximized to better handle the expanding gas and to limit or prevent cavitation as much as possible. Design is complicated in multicomponent streams when flashing and cavitation process conditions exist. Still, one can properly specify and size control valves for cavitating service, while flashing is a consequence of process conditions and cannot be eliminated through valve selection or design. Impulse and Sample Lines All impulse or sample lines must include blow-down connections for periodic defrost operations. The impulse lines serving pressure transmitters, differential flow transmitters, and differential pressure level transmitters should be provided with liquid sealed legs inside the cold box. This design prevents large frost accumulations on the impulse lines and on the valves that are located outside the cold box, caused by the boiling of cryogenic liquid and freezing of ambient moisture. On the other hand, continuously flowing cryogenic liquid samples to analyzers should not have seal loops, because these loops can cause the boiling of the sample inside the cold

8.37 Separation Controls, Air

box and, thereby, cause the distilling of the sample, which makes it nonrepresentative. The design of impulse and sample tubing/piping must also consider the stress caused by thermal expansion and contraction of cryogenic equipment, as well as the weight loads imposed by the insulation (typically perlite) that is packed in the cold box. Flow Elements Venturi and nozzle-type flowmeters (Section 2.29 in Chapter 2 in Volume 1) work well and are frequently used for the measurement of cryogenic liquid and gas flows. In some cases, cryogenic flows are approximated on the basis of the opening of control valves and of the pressure drop across them. Outside the cold box, elbow meters (Section 2.29 in Chapter 2 in Volume 1) are occasionally used where line sizes are very large and where it is desired to minimize the pressure drop across the flow sensor, such as in the case of measuring TSA or PSA regeneration gas flows. The nominal 4:1 turndown capability of the differential flow devices is usually acceptable, because this rangeability exceeds the turndown requirement of the ASU plant itself. The turndown capability of cryogenic liquid flow measurement applications is often limited by the flashing that occurs at low flows, as the control valve is throttled to the point where flashing starts. Nonetheless, smart transmitters can provide wider turndown than is possible with conventional differential pressure transmitters in flow measurement applications. In order to increase plant efficiency, it is important that the instrumentation and control valves used will provide the required accuracy and rangeability. Inversely, if the control loop components are improperly specified or sized, they will not be able to accurately meet the controller set points, and this can cause sustained upsets. Temperature Measurement Temperature measurements are made by thermocouples and resistance temperature devices (RTDs). They are described in detail in Sections 4.10 and 4.13 of the first volume. They are provided with bar stock thermowells that are welded into the process lines and vessels as needed. Thermocouple extension wire or RTD cables are run from the temperature element head to junction boxes on the outer face of the cold box. Process Analysis in ASU The three major purposes of using analyzers in an air separation plant are the monitoring and control of product quality, the performance of process control, and the continuous maintenance of process safety. In a typical sampling system, the pressure of the sample drawn from the process is reduced, and this sample gas is sent to a remote analyzer panel. In order to lower the dead

© 2006 by Béla Lipták

2131

time of the measurement, bypass purge flow is provided around the analyzer. This also helps to minimize the impact of leaks in the sample line. The sample system response time is computed from the sum of the sample and bypass flows, taking into account the transport system volume and pressure. The primary sample is taken from a tap that is situated in an ideal location, and sample stream-switching capability is provided at the analyzer panel, to use alternative samples taken from auxiliary sample taps, when necessary. Product Quality Analysis Product quality analysis is required for the monitoring and control of the oxygen, nitrogen, and argon products of the plant. The primary analytical method used for measuring the purity of the oxygen product is paramagnetic (Section 8.42 in Chapter 8 in Volume 1). The sensor in the paramagnetic analyzer is pressure-compensated and is either the dumbbell or the thermal wind variety. The specification for the oxygen product calls for 99% + purity, and the main impurity in this stream is argon. The specification for the nitrogen product requires that its oxygen impurity be at the parts per million level. This is measured electrochemically by fuel cell, zirconium oxide, or Coulometric sensors (Section 8.42 in Chapter 8 in Volume 1). The product specifications for the argon product limit the impurities of oxygen nitrogen and total hydrocarbon (THC) to trace amounts. The trace oxygen analysis is made by the same methods that are used for the measurement of trace oxygen in the nitrogen product. The trace nitrogen in argon can be measured by either gas chromatography (Section 8.12 in Volume 1) or by spectrographic quartz plasma-type analyzers. THC is measured with a flame ionization detector (FID), and the results are reported as methane-equivalent THC. Normally, all three products are analyzed for moisture (dew point or ppm), and this is typically done by analyzers having aluminum oxide or quartz crystal microbalance sensors (Section 8.33 in Chapter 8 in Volume 1). Trace levels of carbon monoxide, hydrogen, helium, and neon are also present in atmospheric air. They are likely to pass through the HPC and concentrate in the nitrogen product and reflux. It is sometimes necessary to install a noncondensable purge on the condensing side of the LPC reboiler to purge these gases. Similarly, small traces of krypton and xenon, which are present in ambient air, will also process through the system and concentrate in the LOX product in the HPC sump. These trace components can also be recovered and concentrated by specialized processes. The processing of krypton and xenon must also include the removal of hydrocarbons, because if they are allowed to concentrate in the oxygen, unsafe conditions can evolve. Safety Analyzers For the purpose of safety, carbon dioxide and total hydrocarbons are measured in the ppm range or lower.

2132

Control and Optimization of Unit Operations

Samples are typically taken from the feed air supply to the plant and from the liquid oxygen streams. The trace carbon dioxide is typically measured with non-dispersive infrared (NDIR) methods that utilize the Luft principle or by gas filter correlation (GFC) methods (Section 8.27 in Volume 1). Analysis of total hydrocarbon is typically performed using flame ionization detector analyzers, and the results are reported as methane equivalent hydrocarbon in ppm. If thermal deoxidation units (Deoxo) are used to remove oxygen from crude argon, a safety analyzer is required to monitor the oxygen level in the crude argon feed to the Deoxo process to prevent high oxygen levels from causing unsafe thermal runaway of the catalytic bed. A paramagnetic or electrochemical method is used to monitor the percentage of oxygen in the feed to the Deoxo process. Occasionally, safety analyzers are used to monitor for the unlikely situation where high levels of oxygen (> 21%) contaminate the waste gas stream used to purge the front-end adsorbers. Such monitoring can become necessary because adsorbers that are constructed of incompatible materials may present a risk of fire when exposed to enriched oxygen. Finally, area monitors are used to analyze the air quality in occupied rooms to ensure that ventilation is adequate and to provide alarms if either an oxygen-depletion or an oxygenenrichment condition is evolving, with the corresponding risk of asphyxiation or oxygen fire. Process Control Analyzers The product purity analyzers discussed earlier are also used in controlling the air separation process. The product purity or safety analyzers also trigger alarms and safety interlocks. Other process control-related analytical loops include the control of excess hydrogen in the crude argon from the Deoxo process. The concentration of hydrogen is measured by thermal conductivity type analyzers (Section 8.57 in Chapter 8 in Volume 1). Carbon dioxide in the feed air is measured after the frontend adsorption system and is monitored by NDIR-type analyzers (Section 8.9 in Chapter 8 in Volume 1). Dew point is detected by aluminum oxide sensors (Section 8.33 in Chapter 8 in Volume 1) on the outlet of the TSA regeneration gas steam heat exchangers and compressor aftercoolers. In case a leak causes the dew point to rise, an alarm is actuated. The hydrogen supply to the Deoxo systems is monitored for trace methane, because this methane is not removed by the Deoxo process or by argon distillation, and it would become an impurity in the pure argon product. Nitrogen in the sidearm column feed is either inferred on the basis of LPC operating conditions or is directly measured by gas chromatography or by ion mobility spectrometers. The gas chromatograph first removes the oxygen and then chromatographically separates the nitrogen and argon for analysis. The ion mobility analyzers provide an effective method of nitrogen analysis in this mixed gas stream but have

© 2006 by Béla Lipták

the disadvantage of using a radioactive source and therefore requiring special permits and are rarely used today.

REGULATORY AND FEEDFORWARD CONTROLS The controls described in the following paragraphs refer to the subsystems of the process shown in Figures 8.37g and 8.37h. The control loops are not shown in detail, and the feedforward and logic/safety controls are not shown at all. Therefore, the reader is asked that for in-depth, detailed discussions of the related algorithms, dead times, time constants, and tuning, refer to Chapters 2 and 8, where they are discussed. Both regulatory and advanced process controls are used to optimize the production of a specified product mix and to maintain product purity at minimum operating cost. In controlling the air separation processes, these control strategies have to be adapted to optimize production against purity and energy constraints, while also correcting disturbances that would upset the steady-state operation. Regulatory controls serve to respond to upsets caused by feed airflow disturbances, which can be caused by upsets resulting from TSA or PSA regeneration or from diurnal variations in cooling water temperature, which can affect heat transfer. Advanced control technology is also used to provide load-following optimization strategies so as to ramp plant production in response to variations in customer gas demand. Process control systems used to implement these objectives include large distributed control systems (DCSs), which are often linked to supervisory control computers for implementation of advanced controls by means of various forms of model predictive control (MPC). In the following paragraphs, some specific control problems and solutions will be discussed. Main Air Compressor Flow As illustrated in Figure 8.37i, the compressed air is vented to provide the added throughput required to keep the main air compressor (MAC) out of surge. (See Section 8.15 for details on compressor control and optimization.) The surge flow measurement is typically inferred from differential pressure measurement across the compressor stages. The surge line for the compressor is confirmed by surge tests. Some compressor suppliers do not use surge control, but only open the vent when a maximum operating pressure is reached. The level of control sophistication applied is a function of the relative importance to save energy by reducing the surge margin. Stabilization of Pressure Surge During regeneration of the front-end adsorbers, it is important to minimize the pressure surges, because these upsets have the potential of causing a disruption of the downstream distillation process, which in turn can disrupt the argon product

8.37 Separation Controls, Air

Users Compressor

Pi

Pd

∆PT

∆P

∆PY

Ratio relay

m(∆P) + b FT

FCV

SP

h Surge measurement

FIC

FO

PI DA

To vent or recycle Surge valve opens if h < m(∆P) + b

Pressure increase Pd – Pi, PSI

Surge line

Compressor speeds 100% 90% 80%

70% 60% 50%

Differential pressure h, PSI

FIG. 8.37i The flow across the compressor (h) is kept above the surge limit by venting, if the user demand is insufficient to keep it out of surge.

quality profile. Unlike the reversing exchangers that cycled multiple times per hour, the TSA adsorbers are regenerated only once in several hours. If one has properly determined the process dead times and time constants, feedforward control can be used to attenuate these disturbances. The feedforward loop might include a ramping function to increase the main air compressor flow (the multistage centrifugal compressor in Figure 8.37g), in order to compensate for the increased airflow required when the off-line adsorber vessel is being pressurized after regeneration. After the regenerated unit is on-line and the beds are switched to regenerate the other adsorber vessel, the main air compressor flow is ramped back to its normal operating flow. Sidearm Nitrogen Control The feed flow disruption caused by adsorber regeneration is particularly important when argon is being produced, because the argon composition profile in the LPC is sensitive to feed flow, and feed flow upsets introduce a risk of excess nitrogen entering the sidearm column. Nitrogen impurity is normally

© 2006 by Béla Lipták

2133

in the low ppm range. Nitrogen excursions in excess of fractions of a percent in the sidearm column could vapor-lock the condenser and cause loss of vapor flow and reflux, resulting in the draining of the contents of the sidearm column into the LPC. After a dumping episode, the reestablishing of oxygen purity in the LPC and also of the proper operation of the sidearm column can take many hours to complete. The challenge is in managing the sidearm column feed to maximize argon concentration while preventing too much nitrogen, which will cause the sidearm column to dump. On the other hand, if the plant is operated conservatively with little or no nitrogen in the sidearm, argon will ultimately be lost in the waste and recovery will be lower than otherwise possible. Control of nitrogen in the sidearm is implemented by determination of composition in the sidearm or at a location several stages above the sidearm feed point by direct analytical measurement (see QIT in Figure 8.37h). Operating personnel will tend to run the distillation column more conservatively to limit the nitrogen concentration in the sidearm to very low levels, but this also reduces recovery because incrementally more argon will be lost in the waste stream. Therefore, to maintain high argon recovery, the plant requires precise control of sidearm composition, and advanced control systems are capable of providing it. Product Purity Control In general, for a plant with a liquid product and a fixed air flow, the product purity is inversely related to the flow rate of product that is being removed from the plant. Therefore, product flow control is often used to control product purity. This control strategy is complicated by the fact that control loops often interact, necessitating feedforward control algorithms and model predictive control strategies for optimal load-following. The performance of a load-following optimization system is dependent upon the maximum ramping rate of the air separation plant, the severity of demand changes, and the capacitance of the system. The objective is to minimize the venting of overproduction, while avoiding (minimizing) supplemental liquid vaporization, which makes up for production deficiencies. Oxygen Recovery Controller In steady-state operation, where the customer has specified a constant GOX production rate, the oxygen purity is a function of the oxygen recovery. The fixed GOX customer flow being the independent variable, the supplied airflow to the plant is the primary manipulated variable that impacts oxygen recovery. Therefore, the oxygen purity controller (QIT at the bottom of LPC in Figure 8.37h) can act as a cascade master to manipulate the ratio of feed air to product flow. This manipulation affects the power consumption of the plant, and for a fixed number of distillation stages, increasing the feed

2134

Control and Optimization of Unit Operations

airflow will improve the product purity. The control loop is configured as a triple cascade where the O2 recovery controller (QIT, %O2) is the cascade master of the purity controller (QIT, ppm THC), which is the cascade master of the feed flow controller (FIC on the compressor in Figure 8.37g).

Today, taking a portion of the air after it has been purified (stream #1 in Figure 8.37g), boosting the pressure with a booster compressor, cooling the stream, and then expanding it through a mechanical expander provides the refrigeration required for liquefaction.

Front-End Purity and Hydrocarbon Accumulation

Expander Configuration

Methane, ethane, propane, and to a certain extent ethylene are not adsorbed in the TSA or PSA-type front-end air purification processes. Acetylene, which is dangerous due to low solubility, is captured in the TSA or PSA molecular sieve. In addition, there are impurities that can accumulate and plug equipment, such as traces of CO2 that slip through the frontend adsorbers. Plugging increases the potential for accumulating local concentrations of hydrocarbons by blocking passages in the LOX sump reboiler/condenser section (bottom of HPC column in Figure 8.37h), which normally are continuously flushed by siphon action. Trapped oxygen may create a localized concentration of hydrocarbons as a result of dry boiling in the blocked passages and may eventually leave a residue of pure hydrocarbon in a pure oxygen atmosphere that could potentially autoignite, with potentially severe consequences. Normally, in a pumped LOX plant, it is the front-end CO2 removal characteristics that determine the minimum amount of LOX that must be removed as a product or in a purge stream.

The refrigeration required in oxygen plants to make up for the heat leak from the environment is provided by machinery that expands a portion of the feed airflow to the low-pressure column. The work done by the expander can be recovered as electrical power. The recovery can be achieved by using a generator-loaded expander, or by taking that portion of the feed that is to be expanded and first compressing it to a higher pressure by a compressor that is mounted on the same shaft as the expander. This machine is commonly referred to as a compander. In a plant making GOX only, gaseous air would be taken from the TSA, cooled in the main heat exchanger, and expanded directly into the LP column. More refrigeration is needed if either some amount of liquid oxygen product is also needed or if the expander flow is starting to impact the oxygen recovery. In such a case, a compander can be used (instead of recovering the work as electricity), to boost that fraction of the air that is to be expanded. This way, a higher pressure ratio can be obtained across the expander.

Hydrocarbon Concentration Factor

Refrigeration to the plant is typically controlled by manipulation of expander flow. Normally, an expander flow is so selected as to match the cooling needed to produce a little more than the minimum LOX production required. At a constant expander flow, the liquid oxygen sump level controller (LIC at bottom of HPC in Figure 8.37h) will provide steadystate operation.

A hydrocarbon concentration factor can be determined on a material-balance basis by determining how much LOX is being removed from the LPC reboiler compared to the total oxygen in the air that is entering the system. In a pumped LOX plant, the concentration factor is approximately 5 because the entire oxygen product is leaving the column as a liquid. In a plant that produces gaseous oxygen, where most of the oxygen product is generated as GOX, with only a small amount as LOX being produced, the concentration factor might be in the range of 100–500. Total hydrocarbon is continuously measured at several points in the plant. These include either the LPC sump (pure oxygen) or the crude liquid oxygen from the sidearm column condenser sump.

REFRIGERATION CONTROLS In the early days of cryogenic air separation, the feed air was liquefied using a so-called “split cycle” that involved compressing the air to approximately 2000 psig and then cooling the compressed gas. After that, the cold gas was sent through an expander valve to cool it by the Joule Thompson effect. This expander valve is referred to as a Joule Thompson valve or JT valve.

© 2006 by Béla Lipták

Refrigeration Balance Controller

Main Heat Exchanger Control On a pumped LOX plant, high-pressure air from the booster compressor is condensed in the main heat exchanger, and the JT valve essentially controls the liquid level in the exchanger, while the booster compressor discharge pressure effectively controls the vapor inventory in the exchanger. The oxygen in the feed to the plant is an independent variable based on customer oxygen flow requirements, and the booster compressor discharge pressure is held constant. By maintaining a constant discharge pressure, the molecules that condense in the exchanger are effectively offset by the oxygen molecules that are vaporized. The liquid level measurement in the exchanger is difficult, particularly, because above the critical pressure, the vapor/liquid interface disappears. Knowledge of the liquid level is important, as this determines the available surface area for condensation as well as the magnitude of the temperature approach. The temperature approach translates into

8.37 Separation Controls, Air

a pressure ratio between the high-pressure air and the oxygen that is boiling in the exchanger. There are multiple complex positioning strategies for the temperature element of the temperature controller (TIC) that throttles the JT control valve. Proper tuning of this controller is also important to provide the required refrigeration.

Load Valve gain (Gv)

m1

Load (u)

m +

+

Gc

Load

Load

e +

Set point (r)

For stable control: Gc × Gv × Gp × Gs = 0.5

Gp

Process gain (Gp)

There is no doubt that advanced controls can improve the plant efficiency, as well as argon recovery, over what is possible through regulatory controls alone. Advanced control systems can interface with the distributed control system using supervisory computers that are used for regulatory process control to assist in control and optimization. Advanced control systems, using model predictive control, can predict evolving conditions and, as such, can optimize the operation through model-based anticipation. Without advanced control, the plant will still run reliably and will deliver good product, but operators tend to pick safe controller set points, which can be further away from operating or safety constraints than necessary. The safety margins used by the operator are usually a function of both their motivation and their experience/education. Therefore, while the plant will run reliably under manual operator control, its operation will not be optimized. Such unoptimized operation is likely to result in significant losses in argon recovery and is a significant increase in air compressor power consumption. The advanced controls monitor the operating constraints while meeting the production target at minimum waste of either product or energy. In the air separation process, advanced controls can improve the production and purity of GOX, the sidearm feed composition, the flow and purity of argon, and the total cryogenic liquid production.

Gv

Controller gain (Gc)

ADVANCED CONTROL

2135

–b

c Sensor gain (Gs) Gs

Load

FIG. 8.37j If the process gain varies with load, it is necessary to compensate for that nonlinearity. This can be done by characterizing the control valve (shown above) or can be done by characterizing the measurement.

the nonlinearity from the measurement (of nitrogen purity, in this specific case) by a compensator. Such characterizer can take the logarithm of the product quality signal, for example, to arrive at a constant loop gain.

Characterizers to Compensate Process Nonlinearity

Operating Close to Constraints

Nitrogen purity can be controlled by using the nitrogen purity controller as the cascade master of the airflow controller that sets the gaseous air feed flow to the HPC as its slave, if its set point is not already controlled by GOX purity. While linear response is assumed between the composition of the overhead product (GAN) and changes in boil-up and reflux ratio, these relationships are not linear over the whole range of the operation. As the column approaches a state of operation that corresponds to minimum reflux and maximum recovery, it becomes difficult to recover any more nitrogen, and the bottom liquid (LOX) will approach equilibrium with the air feed. If, under these conditions, more nitrogen is taken off, the process gain drastically changes and product purity drops (oxygen impurity in the nitrogen rises). In general terms, if the process gain varies with load, stability of the loop can only be maintained if the variation in the process gain is compensated, because stable control requires a stable gain product of the loop components of about 0.5 (Figure 8.37j). One way to achieve this is to remove

The most efficient plant operation is obtained if the process is run close to the limitations that are set by the equipment used, the available raw material and utility supplies, market conditions, product quality requirements, power consumption, and so on. If one focuses on the objective of minimizing energy consumption, then airflow should be reduced until the limitations set by the purity requirements of the product are approached. This is because an increase in airflow also increases the boil-up and reflux rates, which in turn results in higher product purity, but at a higher investment of energy. The ratio of the power consumption and the production rate of a plant expresses its energy efficiency. Air separation plants can have six or more constraints that are alternately approached during operation. When a constraint is approached, the control system should load or unload the plant, as needed to maintain a margin from the constraint. The more constraints are approached at the same time, the more difficult the control challenge becomes, but selective control can handle that task.

© 2006 by Béla Lipták

2136

Control and Optimization of Unit Operations

Therefore, good plant design should eliminate the possibility of the plant operation being able to approach multiple constraints that require conflicting control responses. For example, no extra stages of distillation should be added in order to obtain better purity than the purity needed at near ideal recovery.

CONCLUSIONS In this section, the process controls for air separation processes using adsorption, membrane, and cryogenic technologies have been discussed. Cryogenic plants are the economic choice for most medium to large-scale oxygen and nitrogen production applications, because of their lower unit power consumption and economies of scale. They represent the only choice for argon production. LOX, LIN, and LAR facilities are available for backup and peak shaving to gaseous supply systems and are readily available on-site, using the same equipment that produces the gaseous products. This eliminates the need for addon liquefiers or the use of supply contracts. Adsorption systems for the production of oxygen can supply 85–95% oxygen and are best suited for applications requiring less than 100 tons per day production. Nitrogenproducing PSA and nitrogen membrane systems are economical for smaller scale production volumes and where the high purity products of cryogenic systems are not required.

ABBREVIATIONS ASU CMS DCAC Deoxo GFC GOX HPC JT LAR LIN

Air separation unit Carbon molecular sieve Direct contact after cooler Deoxidation unit Gas filter correlation Gaseous oxygen High pressure column Joule Thompson Liquid argon Liquid nitrogen

© 2006 by Béla Lipták

LOX LPC MAC MPC NDIR ppmV PSA THC TPD TSA VSA

Liquid oxygen Low pressure column Main air compressor Model predictive control Non-dispersive infrared Volumetric parts per million Pressure swing adsorption Total hydrocarbon Tons per day Temperature swing adsorption Vacuum swing adsorption

References 1. 2. 3. 4.

Parker, S. P., Ed., Encyclopedia of Science & Technology, Volume 3, 8th edition, New York: McGraw-Hill, 1997. www.airproducts.com. SRI International, CEH, November 1999. Beh, C. C. K., and Webley, P. A., “A Method for the Determination of Composition Profiles in Industrial Air Separation, Pressure Swing Adsorption Systems,” Department of Chemical Engineering, Monash University, October 2002.

Bibliography Butricia, A. J., Out of Thin Air, Praeger, 1990. Castle, W. F., “Air Separation and Liquefaction: Recent Developments and Prospects for the Beginning of the New Millennium,” International Journal of Refrigeration, 25, pp. 158–172, 2002. Murphy, K., Odorski, A., Smith, A., and Ward, T., “Oxygen Production Technologies for Non-Ferrous Smelting Applications,” Air Products and Chemicals, Inc., 1987. “Safe Operation of Reboilers/Condensers in Air Separation Units,” Document 65/99, Brussels, Belgium: European Industrial Gas Association, Industrial Gas Committee. “Safe Practices Guide for Cryogenic Air Separation Plants,” CGA P-8, Chantilly, VA: Compressed Gas Association. Schmidt, W., Kovak, K., Licht, W., and Feldman, S., “Managing Trace Contaminants in Cryogenic Air Separation,” AIChE Spring Meeting, Atlanta, GA, March 5–9, 2000. Schmidt, W., Winegardner, K., Dennehy, M., and Castle-Smith, H., “Safe Design and Operation of a Cryogenic Air Separation Unit,” Process Safety Progress, Vol. 20, No.4, pp. 269–279, December 2001. Standen, A., Ed., Encyclopedia of Chemical Technology, 2nd Edition, New York: Interscience, Volume 14, 1967.