GAS CONDITIONING AND PROCESSING Volume 2: The Equipment

GAS CONDITIONING AND PROCESSINGVolume 2: The Equipment Modules

9th Edition

This book is based upon, and is the successor to, the classic work of the same title authored by Dr. John M. Campbell, Sr. from 1966-2013.

© 2017 PetroSkills, all rights reserved PetroSkills, LLC.

All rights reserved. No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the publisher. Requests for permission to make copies of any part of this publication should be mailed to:

PermissionsPetroSkills, LLC.2930 South Yale AvenueTulsa, Oklahoma U.S.A. 74114

3rd Printing, June 2017Printed in the United States of AmericaISBN 978-0-9703449-5-3COPYRIG

HT

Petr

oSkill

s

Disclaimer

The author John M. Campbell | PetroSkills takes no position as to whether any method, apparatus, or product mentioned herein, is or will be covered by a patent or other intellectual property. Furthermore, the information contained herein does not grant the right, by implication or otherwise, to manufacture, sell, offer for sale and/or use any method, apparatus or product covered by a patent or other intellectual property right; nor does it insure anyone against liability for infringement of same.

Neither John M. Campbell | PetroSkills nor any co-author or other party involved with the writing, preparation, publishing or distribution of these materials shall be responsible or liable in any way for any loss, damage or claim with respect to the use of the information, apparatus, equipment, methods or processes disclosed or described herein. There is no warranty or representation, express or implied, with respect to the accuracy, completeness, or usefulness of the information contained herein. All express or implied warranties, including any warranty of fi tness for any particular purpose are expressly disclaimed.

Published by: PetroSkills, LLC.2930 South Yale AvenueTulsa, Oklahoma U.S.A. 74114Phone: (918) 828-2500Fax: (918) 828-2580Website: www.petroskills.comE-mail: [email protected]

Contributors: Author: John M. CampbellEditors: Robert A. Hubbard,

Kindra Snow-McGregor

Special Thanks: Illustrators: Amy Ashby – Graphic Specialist, Chris Snyder – Production Artist, Jamie Browne – Production Artist, Chris Wong – Technical Illustrator

File Administrator: Sandie Womble, DBALayout: Susan Pittman – Publication Specialist

Kevin Cheyne – Printing & Shipping ManagerT.J. Gerlach – Content Specialist

Printed by: Taylor Publishing Company, Dallas Texas U.S.A.

COPYRIGHT

Petr

oSkill

s

In Memory of Dr. John M. Campbell

Dr. John Campbell passed away on 12 September, 2013. He founded John M. Campbell & Company in 1968 and authored the fi rst edition of Gas Conditioning and Processing which was published in the same year. Dr. Campbell enjoyed a distinguished career in the oil and gas industry beginning in the late 1940s and was recognized with numerous awards including the John Franklin Carll Distinguished Professional Award from the Society of Petroleum Engineers in 1978 and the Hanlon Award from the Gas Processor’s Association in 1987.

Dr. Campbell was not only widely acknowledged as a technical leader in the industry, but also as an outstanding educator. He attended the University of Oklahoma from 1946-1951 receiving Masters and Doctoral degrees in Chemical Engineering. During this time he was fi rst introduced to teaching when he was given lecture responsibilities in several undergraduate classes.

After 3 years in industry, Dr. Campbell returned to the University of Oklahoma as a professor in the Department of Petroleum Engineering where he remained until 1968, including 12 years as head of the department.

During this time Dr. Campbell developed and taught several short courses for companies in the oil and gas business worldwide. These experiences paved the way for the success he enjoyed over the following 25 years and as an instructor and consultant.

"I met Dr. Campbell when I joined the company in 1980. He was a remarkable person and a natural leader. He had the unique ability to distill complex topics into understandable pieces. He was technically brilliant, but understood the human side of our business. I am proud to have known him and enjoyed his friendship. I will miss him."

—Bob Hubbard

COPYRIGHT

Petr

oSkill

s

ABOUT JOHN M. CAMPBELL | PETROSKILLS

John M. Campbell & Co. was established in 1968 to provide professional services to the oil and gas industry in the areas of production facilities, gas processing and economic analysis. Through a merger with PetroSkills in 2012 the combined company has the capability to deliver these services across the full span of the industry from the reservoir to the marketplace.

PetroSkills training programs combine a strong theoretical foundation with relevant practical experience. The outcome is a focused, real-world, results-oriented event for course participants. Our instructors average 30 years of hands-on experience in the industry, and the depth and breadth of their technical expertise is unmatched by other training providers.

PetroSkills provides technical training courses world-wide and has trained over 375,000 petroleum professionals. Our Surface Facilities course topics include: Gas Processing, Process Facilities, Mechani-cal, Instrumentation, Controls & Electrical, Offshore, Pipeline, LNG and Operations and Maintenance. For more details on Surface Facilities training, visit petroskills.com.

PetroSkills Alliance

The PetroSkills Alliance is a unique consortium of over 30 companies in the petroleum industry who focus on creating, managing and evolving industry-leading programs for developing and assuring individual and organizational competency. By leveraging their collective expertise and best practices in important but not unique subject matter, Alliance members benefi t from a more effi cient investment in human capital development while assuring themselves that they are able to provide the best possible devel-opment resources to their people and organizations. Since 2001 the PetroSkills Alliance has grown steadily, with worldwide members representing every segment of the petroleum industry.

Consulting & Solutions

PetroSkills Consulting Services are built on extensive real-world experience across the entire oil and gas value chain. We combine the insights from our alliance member companies with the knowledge of seasoned industry experts to help with your most complex challenges. From talent development and organizational performance, to engineering and operational needs, we are driven by one goal: to improve performance. Visit petroskills.com/consulting for further information.

PetroSkills Compass is the industry’s most comprehensive competency management and assurance solution showing progression and growth development of technical professionals’ career paths for upstream, midstream and downstream energy companies. Through PetroSkills Compass, our comprehensive features and capabilities for assessing and documenting employee skills enables companies to focus on using accurate, accessible information about employees to plan for the future.

COPYRIGHT

Petr

oSkill

s

CHAPTER 18 409

18

ADSORPTION DEHYDRATION AND HYDROCARBON REMOVAL

Glycol dehydration or refrigeration with hydrate inhibition are used for most applications where dehydration of natural gas to sales gas specifi cation is required. Solid bed dehydration (also called dry desiccant or adsorption dehydration) is often the superior alternative in applications where water removal to a dew point below –40 to –50 °C [–40 to –58 °F] is required. The chart below shows a general comparison of these three dehydration technologies.

Glycol Dehydration (TEG) Chilling with MEG Injection Solid Bed AdsorptionPreferred for dewpoint depressions up to 40-50°C [72-90°F]

Water dewpoint depressions of 60-70°C [108-126°F] achievable

Preferred for water dewpoint depres-sions greater than 80°C [144°F]

Dewpoints below –40°C [–40°F] require special designs but are achievable

Dewpoints of –40°C [–40°F] possible Dewpoints below –100°C [–152°F] possible

High pressure contactor required, regen system is small, BTEX and acid gas emissions

Refrigeration system required, regeneration system is small, BTEX and acid gas emissions

High pressure adsorbers, switch-ing valves and regeneration heater required

Meets water dewpoint specifi cation only

Typically used when meeting an HCDP specifi cation where refrigera-tion is required

Can be multifunctional (removal of hydrocarbons or sulfur compounds)

The following list shows applications where solid bed adsorbents are typically used:1. Dehydration to water dewpoints less than –40 to –50°C [–40 to –58°F], such as those required

upstream of NGL extraction plants utilizing expanders and LNG plants.2. Hydrocarbon dewpoint control units where simultaneous extraction of water and hydrocarbon

is required to meet both of the respective sales specifi cations—well suited for hydrocarbon dewpoint control on lean, high pressure gas streams. These applications generally use silica gel.

3. Simultaneous dehydration and sweetening of natural gas where concentration of sulfur com-pounds is less than about 100 ppm.

4. Dehydration of gases containing H2S where H2S solubility in glycol can cause emission prob-lems at the regenerator.

5. Dehydration and trace sulfur compound (H2S, COS, mercaptans) removal for LPG and natural gas streams.

6. Dehydration of NGL streams.

COPYRIGHT

Petr

oSkill

s


410 VOLUME 2: THE EQUIPMENT MODULES

Adsorption describes any process wherein molecules from the gas or liquid are held on the surface of a solid by physical attraction. Commercial adsorbents have interconnecting cavities with enormous surface area available to attract polar compounds such as water as well as non-polar compounds such as hydrocarbons. Adsorbents may be divided into two classes—those which owe their “activity” to surface adsorption and capillary condensation (physical adsorption), and those which react chemically (chemical adsorption). Chemisorbents are used for trace mercury removal and trace H2S removal from natural gas. They are usually, but not always, non-regenerable. The physical adsorbents, which are the subject of this chapter, require use of a material which has the following characteristics:

1. Large surface area and pore volume for high capacity2. Possesses “activity” for the components to be removed3. Mass transfer rate is high, which means the mass transfer zone (Figure 18.5) is small4. Easily and economically regenerated5. Maintains an acceptable percentage of its initial adsorption capacity with time6. Acceptable pressure drop during the life of the adsorption unit7. High mechanical strength to resist crushing, dust formation, and damage due to thermal cycling8. Inexpensive, non-corrosive, non-toxic, chemically inert; possesses a high bulk density and high

capacity9. No appreciable change in volume during adsorption and desorption, and should retain strength

when “wet”10. Commercially available and proven in the fi eldMost commercial adsorbents will have a total surface area of 500 to 800 m2/g [2 400 000 to

3 900 000 ft2/lbm]. A spoonful would have a surface area equivalent to a football fi eld! This exceptionally large area is only achieved by producing material with a large interior surface resulting from capillaries or a crystalline-type lattice. The exterior surface of the particles is almost negligible.

Impurities such as water, carbon dioxide, hydrogen sulfi de, and other sulfur components enter the pore spaces and are held onto the surface inside the voids. In the natural gas industry, a Thermal Swing Adsorption cycle (TSA) is used. Impurities are adsorbed at low temperature and high pressure. The adsorbent is regenerated at high temperature. The regeneration pressure can be high or low.

The materials which meet the previous requirements may be divided into several general categories, such as: Alumina: a purer, manufactured version of bauxite

1. Silica Gel: composed largely of SiO2; manufactured by chemical reaction2. Molecular Sieves: an alkali metal alumino-silicate (zeolite)3. Activated Carbon: a carbon product treated and activated to have adsorptive capacity

All but carbon are used for dehydration. Carbon has desirable properties for hydrocarbon removal and adsorption of certain impurities but possesses negligible water capacity.

The basic nomenclature used in the industry is as follows:1. Adsorbent: the material (molecular sieves, silica gel, activated alumina, activated carbon—often

referred to as desiccant) that is removing the impurity2. Adsorbate: the impurity being removed (H2O, H2S, RSH, etc.)3. Adsorption: the adhesion of molecules onto a solid surface

COPYRIGHT

Petr

oSkill

s

CHAPTER 18 411


Desiccant PropertiesTable 18.1 presents properties of common desiccants.

The available pore volume determines the potential capacity for adsorption. In essence, monolayer adsorption occurs. Molecular sieves have precise pore openings, but silica gels and activated aluminas are amorphous and have a range of pore openings up to 100 Angstroms (Å). Types 3A, 4A, and 5A molecular sieves contain cavities 11.4 Å in diameter with respective nominal pore openings of 3, 4, and 5 Å (opening size for adsorption), where 108 angstroms (Å) = 1 cm.

As the pore diameter approaches the molecular diameter of the adsorbate, capillary condensation takes place for those molecules that exist in the liquid phase at adsorber pressure and temperature. Silica gel has a higher equilibrium capacity than molecular sieves because of its larger pore volume.

The pore opening at the surface of the desiccant must be large enough to admit the molecules being adsorbed to the interior of the particle where most of the surface area exists.

Table 18.1 Typical Desiccant Properties

Desiccant ShapeBulk Density,

kg/m3 [lbm/ft3] Particle Size

Heat Capacity, kJ/kg·°C

[Btu/(lbm-°F)]

Alumina Alcoa F200 Beads769[48]

714 Tyler mesh 1/8" / 3/16" / 1/4"

1.00[0.24]

Activated Alumina UOP A-201 Beads 737[46]

3-6 mesh or5-8 mesh

0.92[0.22]

Mole Sieve Grace – Davidson 4A Beads 673-721[42-45]

4-8 mesh or8-12 mesh

0.96[0.23]

UOP MOLSIV™ 4A-DG Extrudate 641-705[40-44]

1/8" or1/16" pellets

1.00[0.24]

Mole Sieve Zeochem 4A Beads 705-737[44-46]

4-8 mesh or8-12 mesh

1.00[0.24]

Silica Gel Sorbead® - R Beads 785[49]

5 x 8 mesh 1.05[0.25]

Silica Gel Sorbead® - H Beads 721[45]

5 x 8 mesh 1.05[0.25]

Silica Gel Sorbead® WS Beads 721[45]

5 x 8 mesh 1.05[0.25]

Table 18.2 shows the nominal diameter of common molecules determined from the Lennard-Jones relationship. This is called the nominal diameter because the molecules are not spheres and their ability to enter a given size opening depends on their direction of approach. Also, they are fl exible and can “squeeze” through an opening to some degree.

The various commercial desiccants can be divided into three broad categories: alumina, gel, and molecular sieves. Within each are a series of trade names.

Alumina is a hydrated form of aluminum oxide (Al2O3). When manufactured it is essentially iron free. In its natural state (bauxite) it contains varying amounts of iron. It is activated by driving-off part of the hydrated water adsorbed on the surface.

COPYRIGHT

Petr

oSkill

s



Gel is a granular, amorphous solid. Silica gel is the generic name for a gel manufactured from sulfuric acid and sodium silicate. It is essentially 100% silicon dioxide (SiO2). Other gels like alumina gel may be largely a form of Al2O3. Some gel-type desiccants are some combination of these two.

Molecular sieves are alkali metal crystalline aluminosilicates, very similar to natural clays. Type 4A molecular sieves are composed of Na2O3, Al2O3 and SiO2. Types 3A and 5A are produced by ion exchange of about 75% of the Na ions by potassium and calcium ions, respectively. Type 13X is populated by Na ions. All types have a pH of about 10 when in a slurry form with water and are stable in the pH range of 5-12.

Molecular sieves have electric charges on the inner surfaces of the crystal cavities, which are attracted to similar charges on polar molecules. The polarity of the water molecule plays an important part in this attraction. Such molecules, including hydrogen sulfi de, ammonia, carbon dioxide, and the alcohols, are adsorbed in prefer-ence to non-polar molecules.

In Figure 18.1, there are truncated octahedron groups that are connected to each other by either a four-membered structure (Type A) or a six-membered structure (Type X). A network of cavities is created. The adsorption takes place in these voids. The effective pore diameter is determined by the choice of cation (K+, Na+ or Ca++) and its position in the structure.

The nominal diameters of molecules that can enter the crystalline structure and be adsorbed are shown in Table 18.3. The X type can adsorb all molecules adsorbed by the A type with somewhat

higher capacity. 13X can adsorb large molecules such as mercaptans.

The selective capacity of molecular sieves for different sizes of molecules is important. To a degree, you can exclude those molecular sizes too large to enter the crystal. This is why a Type 3A or Type 4A molecular sieve might be used for drying. Molecular sieves are likewise used at higher absorption tempera-tures because their capacity above 38°C [100°F] declines less than gel or alumina. Table 18.3 summarizes the characteristics and applications of common molecular sieves.

Molecular sieves are supplied in either pellets (extrudates) or beads. The pellets will have precise particle diameters, typically 3.2 mm (1/8 inch) or 1.6 mm (1/16 inch). The particle diameter of beads is found from the mesh size used: 4x8 mesh beads are in a range of sized (2.4-4.7 mm [0.09-0.19 in]), with a nominal diameter of 3.2 mm [0.125 in]: 8x12 mesh beads range from 2.4-1.4 mm [0.06-0.09 in], and have a nominal diameter of 1.6 mm [0.0625 in].

Table 18.2 Nominal Diameters of Various

Molecules Using the Lennard-Jones Relation (18.1)

Compounds AngstromsH2 2.89O2 3.46N2 3.64

CO2 3.30H2O 2.65NH3 2.60CH4 3.80C3H8 4.30

CH3OH 3.76nC4H10 4.30iC4H10 5.28

H2S 3.60

Structure of Zeolite A Structure of Zeolite X

Figure 18.1 Molecular Sieve Structures (Courtesy: Zeochem)COPYRIG

HT

Petr

oSkill

s

CHAPTER 18 413


Table 18.3 Basic Characteristics of Molecular Sieves

Basic Type

Nominal Pore Diameter

(Angstroms) Available Form

Equilibrium H2O Capacity

(% wt) Molecules Adsorbed Molecules Excluded Applications

3A 3 Powder1/16 in Pellets 1/8 in Pellets 8-12 Beads4-8 Beads

2621212121

Molecules with an effective diameter <3 angstroms, including H2O and NH3

Molecules with an effective diameter >3 angstroms, e.g., ethane, CO2, H2S, methanol

Selected to minimize co-adsorption of unwanted impurities, such as metha-nol or light olefi ns, and to reduce the formation of COS.Dry olefi ns, methanol, ethanol, and natural gas.

4A 4 Powder1/16 in Pellets 1/8 in Pellets 8-12 Beads 4-8 Beads

14 30 Mesh

27.52222222222

Molecules with an effective diameter <4 angstroms, including ethanol, H2S, CO2, SO2, C2H4, C2H6, and C3H6

Molecules with an effective diameter >4 angstroms, e.g., propane

Most commonly used for natural gas dehydration.Dry natural gas, remove H2S.4A sieve can also be manufactured to remove trace amounts of CO2 for LNG applications.

5A 5 Powder 1/16 in Pellets 1/8 in Pellets 8-12 Beads4-8 Beads

2621.521.521.521.5

Molecules with an effective diameter <5 angstroms, including n-C4H9OH, n-C4H10, C3H8 to C22 H46

Molecules with an effective diameter >5 angstroms, e.g., iso compounds and all 4 carbon rings

Separates normal paraf-fi ns from branched-chain and cyclic hydrocarbons through a selective adsorption process, remove H2S, and normal mercaptans.

13X 10 Powder1/16 in Pellets 1/8 in Pellets 8-12 Beads 4-8 Beads

3026262626

Molecules with an effective diameter < 10 angstroms

Molecules with an effective diameter > 10 angstroms

Remove mercaptans and H2S from hydrocarbon liquids, remove H2O and CO2 from air plant feed, remove iso-mercaptans from natural gas.

Note: 8-12 and 4-8 refers to the Tyler screen size.4-8 beads is equivalent to a nominal diameter of 3.2 mm [1/8 in].8-12 beads is equivalent to a nominal diameter of 1.6 mm [1/16 in].Chart shows typical molecular sieve types only. It is common for vendors to customize these basic forms for specifi c use.Equilibrium H2O Capacity taken at 2.333 kPa and 25°C (0.34 psia and 77°F).Acid resistant sieves are available for dehydration of natural gas containing high concentrations of acid gas (H2S + CO2).Each type adsorbs the molecules listed plus those in the preceding row.

Desiccant SelectionThe selection of a desiccant for a particular application depends on several factors: water dewpoint

specifi cation, presence of contaminants (especially sulfur compounds), co-adsorption of heavy hydrocar-bons, and cost.

All commercial desiccants are capable of producing water dewpoints below –60°C [–76°F]. In a well-designed and properly operated unit, the following water dewpoints are achievable:

Desiccant Outlet DewpointAlumina –73°C [–100°F]Silica Gel –60°C [–76°F]

Molecular Sieves < 0.1 ppmv

COPYRIGHT

Petr

oSkill

s



Molecular sieves are routinely used to dehydrate gas upstream of deep NGL extraction plants, LNG liquefaction, or NRU (nitrogen rejection unit) facilities which operate at temperatures well below –100°C [−150°F]. In these applications water content measurement is diffi cult. The lowest reliable measurement of water in the fi eld is 0.1 ppmv.(18.21)

For most gas drying applications requiring effl uent water concentrations below 1 ppmv molecular sieve will be the fi rst choice due to the very low outlet water dewpoints and higher useful capacity. Molecular sieves are also used in applications requiring removal of trace levels of sulfur compounds. Molecular sieves require higher regeneration heat loads compared to gels or alumina.

In the presence of a gas stream saturated with water, aluminas have a higher equilibrium capacity for water than molecular sieves, but the water loading declines rapidly as the relative saturation of the gas stream decreases. Aluminas also have a lower heat of regeneration than sieves. However, the limited outlet water dewpoints achievable with alumina preclude their use in very low temperature gas processing appli-cations.

Silica gel is sometimes used when both a moderate water and hydrocarbon dewpoint must be met. Some silica gels have an appreciable capacity for C5+ hydrocarbons as well as for water. This allows both water and hydrocarbon removal to be achieved in a single unit. The equilibrium capacity of gel for hydro-carbons is lower than for water; consequently, the bed saturates with hydrocarbons much more quickly than for water. This results in short adsorption cycle times, sometimes less than 1 hour, so the name Short Cycle Units (SCU) is often applied to these installations.

THE BASIC SYSTEM

Figure 18.2 shows a simple dry desiccant system. It consists of two towers, each containing desiccant. One is drying while the other is regenerating. During regeneration, all adsorbed materials are desorbed by heat to prepare the tower for its next adsorption cycle. In essence, an adsorption system concen-trates the impurity, such as water, from the feed stream into the regeneration stream.

At the time shown in Figure 18.2, Adsorber A is drying and Adsorber B is in regeneration. The main gas stream fl ows into the top of tower A and out the bottom. As later discussions will detail, the regeneration cycle consists of two main parts: heating and cooling. During the heating portion, the regeneration gas is heated to 204-316°C [400-600°F]. The temperature depends on the desiccant being used and the character of the material to be desorbed. Figure 18.2 indicates an outside source of clean, dry gas is being used for regeneration.

Once the desiccant bed has been heated to the desired temperature, the regeneration gas by-passes the heater to permit cooling of the bed. This cooling normally ceases when the bed is 10-15°C [18-27°F] warmer than the inlet gas temperature.

The hot regeneration gas leaving the tower is cooled to condense the desorbed water or hydrocar-bons. After separation, the cooled regeneration gas can be returned downstream of the regeneration gas source or routed to sales, depending on whether or not outlet gas specifi cation will be met. Sometimes the spent regeneration stream is sent to a small amine unit to remove sulfur compounds from the regeneration gas and/or a small glycol unit to remove the water prior to blending. In dehydration applications, desiccant vessels are typically externally insulated and the regeneration gas rate will normally be 5-15% of the total throughput to the dehydration process, with 10% being a good average.

In gas dehydration, fl ow during adsorption is typically downward because of the higher allowable superfi cial velocity compared to upfl ow. During regeneration, heating is typically upfl ow even though it

COPYRIGHT

Petr

oSkill

s

CHAPTER 18 415


requires more valves and piping. Most bed contamination occurs at the top. By heating upfl ow, the sieve at the bottom of the bed is not exposed to the contaminants desorbed during the regeneration cycle. Since the natural gas leaving the tower during the adsorption cycle will be in equilibrium with the adsorbent in the bottom of the tower, upward regeneration with clean, dry gas is preferred when dehydrating to water dewpoints below –100°C [–152°F].

The fl ow direction for cooling gas is optional. Upfl ow cooling saves two switching valves per tower (since unheated regeneration gas may be used) but requires dry gas. Downfl ow cooling (same direction as adsorption) is preferred if the cooling gas contains water.

There are three basic sources of regeneration gas in gas dehydration:1. Inlet gas2. Dry product gas from the adsorption unit3. Dry effl uent (tail) gas from a downstream cryogenic unit, such as the demethanizer overhead

The fi rst source, 1, involves some degree of re-saturation of the bed during cooling, which limits the useful bed capacity. The second source, 2, uses the dry gas exiting the adsorbent. The last source, 3, is the most effi cient and is the norm in low temperature NGL extraction plants. Both 1 and 2 will also increase the amount of adsorbent required compared to 3.

One design factor is the number of adsorber vessels. Many large dry desiccant units for natural gas drying contain more than two towers to reduce the overall cost of dehydration and to increase fl exibility.

There are several ways to use multiple adsorbers. As shown in Figure 18.3, two adsorbers are operating in adsorption in a staggered, parallel cycle with the third being heated and then cooled.

= Adsorption Step Time= Heating= Cooling= Depressurization= Repressurization

Time Shown

Adsorber A

Adsorber B

A

AH

H

C

C

A HC

TC

LCLC

RegenerationGas

Fuel Gas

Dry Process Gas

Water to Disposal

Spent Regeneration Gas

AdsorberA

AdsorberB

RegenerationGas Cooler

RegenerationGas Separator

Dry Gas Filter

Wet Feed Gas

To Liquid Disposal

Inlet Coalescing Filter

RegenerationGas Heater

Figure 18.2 Flow Sheet of a Basic Two-Tower Dry Desiccant Unit

COPYRIGHT

Petr

oSkill

s



In Figure 18.3, the shaded area inside adsorbers A and B shows the progress of water adsorption in the bed. This is the portion of the bed that is saturated with water. Below this shaded area, the desiccant is capable of adsorbing more water.

The adsorption front in bed A has travelled further through the vessel than in bed B because bed A

has been on-stream longer. Before the leading edge of this front reaches the outlet, bed A will be switched to regeneration (R), and bed B and the newly regenerated bed C will now be on adsorption. Thus, at any one time, the two dehydrating towers possess different degrees of water saturation. By the time bed A is ready for regeneration, bed C must be ready to go on adsorption.

The operating sequence of the adsorbers in Figure 18.4 is shown in Table 18.4.

The time required to complete the operation in each column is referred to as the step time. The total time that any one tower is on adsorption is equal to A1 + A2. The total time available for regeneration (heat,

Figure 18.3 Dry Desiccant Adsorbers

CA B

Table 18.4 Operating Sequence of a Three Adsorber System

Bed A: A2 R A1Bed B: A1 A2 RBed C: R A1 A2Time:

TC

RegenerationGas

RegenerationGas Heater

To Liquid Disposal

RegenerationGas Cooler

RegenerationGas Separator

Dry Gas FilterFuel Gas

AdsorberA

Inlet Coalescing

Filter

AdsorberB

AdsorberC

Dry Process Gas

Water to Disposal

SpentRegenerated GasWet Feed Gas

LC

= Adsorption Step Time= Heating= Cooling= Depressurization= Repressurization

Time Shown

Adsorber A

Adsorber B

Adsorber C

A1

A1

A1

A2

A2

A2

H

H

H

C

C

C

A1,2 HC

LC

Figure 18.4 Three-Tower System Flow Schematic

COPYRIGHT

Petr

oSkill

s

CHAPTER 18 417


cool, de-pressurization, re-pressurization, and stand-by) must equal one step time. Note that for systems with more than two adsorbers, the step time (time for regeneration) will be a shorter duration relative to the adsorp-tion cycle.

The time required to fi nish a complete regeneration cycle in each column is referred to as the step time, and is provided in Equation (18.1).

(18.1)

Shorter step times increase the required regeneration gas rate and the size of the regeneration equipment. Equation (18.1) can be rewritten as total cycle time, tcycle = ta + tstep = ta + ta/(n – 1), where tcycle is the total cycle time of each tower.

Figure 18.4 shows a fl ow schematic for a three-tower system. Adsorbers A and B are operating in parallel adsorption and Adsorber C is shown in the heating cycle.

THE NATURE OF ADSORPTION

Figure 18.5(a) illustrates the basic behavior of an adsorbent bed in gas dehydration service. During normal operation in the adsorption cycle, three separate zones exist in the bed:

1. Equilibrium zone2. Mass transfer zone (MTZ)3. Active zone

In the equilibrium zone, the desiccant is fully saturated with water. It has reached its equilibrium water capacity based on inlet gas conditions and has no further capacity to adsorb water.

The second zone is the mass transfer zone (MTZ). Virtually all of the mass transfer takes place in the MTZ. A concentration gradient exists across the MTZ. This is illustrated in Figure 18.5(b) for various times throughout the adsorption cycle. Curves 1-3 show the formation of the MTZ over successive time intervals; curve 4 refl ects the concentration gradient for the MTZ position in Figure 18.5(a). Curve 6 shows the concentration gradient at breakthrough. Notice the adsorbate (water) bed saturation is 0% at the leading edge of the MTZ and 100% at the trailing edge.

The third zone is the active zone. In the active zone, the desiccant has its full capacity for water and contains only that amount of residual water left from the previous regeneration cycle.

When the leading edge of the MTZ reaches the end of the bed, breakthrough occurs. If the adsorp-tion process is allowed to continue, the water content of the outlet gas will increase following the traditional “S” curve. Breakthrough curves are illustrated in Figure 18.5(c) for three MTZ lengths. Imagine a water analyzer probe recording water breakthrough readings at the dryer outlet of the vessel on adsorption. Curve 1 in Figure 18.5(c) shows a breakthrough curve if the adsorbent reached instantaneous equilibrium and there was no mass transfer zone. Curves 2 and 3 show the character of the water breakthrough curves for moderate and larger mass transfer zones, respectively. A large MTZ means less adsorbent will be fully saturated with water at the end of the adsorption cycle. In other words, a large MTZ results in more under-utilized adsorbent compared to a small MTZ.

t n – 1t

stepa=

Where: tstep = time available for regenerationta = adsorption timen = number of adsorbers

COPYRIGHT

Petr

oSkill

s



Adsorption is a transient process. During the adsorption cycle, weakly held components can be desorbed by the more strongly held ones. Figure 18.5(d) shows the location of MTZs in multicomponent adsorption typical of hydrocarbon and water adsorption on silica gel. As the gas enters the bed, components are adsorbed at different rates. After the process has proceeded for a very short period of time, a series of adsorption zones will appear. These zones represent that portion of the tower involved in the adsorp-tion of any given component. Behind a particular zone, all of a component entering has been adsorbed on the bed. Ahead of the zone, the adsorbent still has its full equilibrium capacity to adsorb that particular component. These zones form and move through the desiccant bed. Water would be the last zone formed. On all adsorbents except carbon, water will displace the hydrocarbons if enough time is allowed to do so. Multicomponent adsorption on silica gels will be discussed in more detail in the Hydrocarbon Recovery section.

Figure 18.5(e) shows the effect of desiccant size on the length of the MTZ. A smaller desiccant size results in a shorter MTZ which in turn increases the useful capacity of the desiccant. However, a smaller desiccant size increases the pressure drop across the bed. Selection of desiccant size is an important consid-eration in the design and specifi cation of an adsorption unit.

Desiccant CapacityFigure 18.6a and Figure 18.6b shows a set of isotherms for Type 4A molecular sieve.(18.22) The

y-axis shows the equilibrium loading, Xe, of the sieve as a function of water partial pressure (plotted on the x-axis) for several system temperatures. The temperature curves are referred to as isotherms. The isotherms show that the equilibrium water capacity of the sieve is dependent on temperature.

d) Schematic View of Bed Saturations

0

H2O C6+ C5 C4

Bed Length

Saturation1 2 3 4

Time

1) 14-20 Mesh2) 8-10 Mesh3) 5-6 Mesh4) 3-8 Mesh

e)

0

1.0

a) Product

EquilibriumZone

Mass Transfer

Zone

ActiveZone

Feed

32

Time

1

00

1.0

=0.95

c) Character of the Breakthrough Curve

CCo

CCo

Feed

B

E

12

34 5

Bed Length

Saturation

00

vZFeed

6

b) Variation of Adsorption Zone Front with Time

Ads

orba

teBe

d Sa

tura

tion

Ads

orba

teBe

d Sa

tura

tion

CCo

Co=0.05

Figure 18.5 Schematic Portrayal of Adsorption Process

COPYRIGHT

Petr

oSkill

s

CHAPTER 18 419


Fig

ure

18.

6a

Isot

herm

s fo

r U

OP

MO

LSIV

4A

– S

I Uni

ts (C

ourt

esy:

UO

P L

LC, A

Hon

eyw

ell C

ompa

ny)

Xe, kg of Water Adsorbed / 100 kg Activated Adsorbent

Pa

rtia

l P

ress

ure

of

Wa

ter,

kP

a

1.0

E-07

0.0

5.0

10.0

15.0

20.0

25.0

1.0

E-06

1.0

E-05

1.0

E-04

1.0

E-03

1.0

E-02

1.0

E-01

1.0

E+00

1.0

E+01

1.0

E+02

1.0

E+03

1.0

E+04

Activ

atio

n Co

nditi

ons:

350°

C an

d <1

0 M

icro

ns H

g

0°C

25°C

40°C

65°C

200°

C26

0°C

315°

C

©2003 UOP LLC All Rights Reserved. Used with Permission.

COPYRIGHT

Petr

oSkill

s



Fig

ure

18.

6b

Isot

herm

s fo

r U

OP

MO

LSIV

4A

– F

PS

Uni

ts (

Cou

rtes

y: U

OP

LLC

, A H

oney

wel

l Com

pany

)

Xe, lbm of Water Adsorbed / 100 lbm Activated Adsorbent

Pa

rtia

l P

ress

ure

of

Wa

ter,

psi

a

1.0

E-08

0.0

5.0

10.0

15.0

20.0

25.0

1.0

E-07

1.0

E-06

1.0

E-05

1.0

E-04

1.0

E-03

1.0

E-02

1.0

E-01

1.0

E+00

1.0

E+01

1.0

E+02

1.0

E+03

1.0

E+04

32°F

77°F

100°

F15

0°F

400°

F50

0°F

600°

F

Activ

atio

n Co

nditi

ons:

662°

F an

d <1

0 M

icro

ns H

g

©2003 UOP LLC All Rights Reserved. Used with Permission.

COPYRIGHT

Petr

oSkill

s

CHAPTER 18 421


The water partial pressure can be found from Equation (18.2):

(18.2)

The mol fraction of water can be calculated from the water content of the gas using Equations (6.2) (SI), and (6.3) (FPS) provided in chapter 6.

Figure 18.6a and Figure 18.6b can be used to estimate the equilibrium loading, Xe, of the molecular sieve at adsorber feed conditions. It can also be used to estimate the equilibrium loading at regeneration conditions. The equilibrium loading at regeneration conditions sets the minimum achievable water content of the dry gas. There will always be some water left on the molecular sieve following regeneration. This equilibrium condition can also be found from Figure 18.6a and Figure 18.6b. These calculations are demon-strated in Example 18.1 and Example 18.2.

Adsorbents such as molecular sieves can be regenerated by reducing the partial pressure of the adsorbate or increasing the system temperature. The former is referred to as Pressure Swing Adsorption (PSA) and the latter is referred to as Thermal Swing Adsorption (TSA). TSA is used in gas dehydration.

pp y Pw w=

SI FPS

Where: ppw = water partial pressure kPa psia

yw = water mole fraction of the natural gasP = pressure of the natural gas kPa psia

Example 18.1: Water-saturated natural gas is fed to a three-tower system containing Type 4A molecular sieves. The feed rate is 11.3 106 std m3/d [400 MMscfd] of 20.3 MW gas. The inlet pressure is 6205 kPa [900 psia], and the inlet temperature is 30°C [86°F]. What is the equilibrium loading of the sieve, Xe, at feed conditions?

SI Solution:

Step 1. Determine the water content of natural gas from Figure 6.1a at feed conditions. Water content = 690 kg/106 std m3

Step 2. Convert the water content of the gas to ppmv using Equation (6.2). Determine the water partial pressure, ppw, from Equation (18.2).

y 761 400690 kg H O/10 std m

9.06 10 906 ppmvw2

6 3–4#== =c m

ppw = (9.06 10–4)(6205) = 5.6 kPa

Step 3. From Figure 18.6a, interpolating between the 25°C and 40°C isotherms to fi nd 30°C at a water partial pressure of 5.6 kPa. The theoretical equilibrium loading, Xe, is 23.5 kg of water adsorbed/100 kg of adsorbent. This is often written as a percent, for planning purposes use 23 %.

FPS Solution:

Step 1. Determine the water content of natural gas from Figure 6.1b at feed conditions. Water content = 44 lbm/MMscf

COPYRIGHT

Petr

oSkill

s



Example 18.2: Using the data in Example 18.1, determine the equilibrium loading, Xe, at regeneration conditions, and the lowest product gas specifi cation in ppmv water achievable if the bed is regenerated using a water saturated regeneration gas having a water partial pres-sure of 5.6 kPa [0.8 psia] at 260°C [500°F].

SI Solution:

Step 1. Referring to Figure 18.6a (isotherms), at 260°C [500°F] and a water partial pressure of 5.6 kPa fi nd Xe = 3.8 kg of water per 100 kg of adsorbent. This means a Type 4A molecular sieve at the specifi ed regeneration conditions holds 3.8 kg water/100 kg adsorbent.

Step 2. Determine the lowest product gas specifi cation achievable for the regeneration equilibrium loading conditions.

From Figure 18.6a, locate Xe of 3.8 % on the 30°C [86°F] isotherm. Read a water partial pressure, ppw, of 4.8 10–4 kPa.

Step 3. Calculate the achievable product gas water specifi cation from Equation (18.2) using the water partial pressure from Step 2.

yw = ppw/P = (4.8 10–4 kPa)/(6205 kPa) = 8.0 10–8 = 0.08 ppmv

This is the lowest theoretical water content that can be achieved if regenerating with water-saturated gas. This value will be smaller if dry regeneration gas is used.

FPS Solution:

Step 1. Referring to Figure 18.6b, at 500°F and a water partial pressure of 0.8 psia fi nd Xe of 3.8%. This means a Type 4A molecular sieve at the specifi ed regeneration conditions holds 3.8 lbm of water/100 lbm of adsorbent.

Step 2. Determine the lowest product gas specifi cation achievable for the regeneration equilib-rium loading conditions. From Figure 18.6b (isotherms), locate Xe of 3.8 wt % on the 86°F isotherm. Read a water partial pressure of 7.0 10–5 psia.

Example 18.1 (Cont'd):

Step 2. Convert the water content of the gas to ppmv using Equation (6.2). Determine the water partial pressure, ppw, from Equation (18.2).

y 47 43044 lbm/MMscf 9.28 10 928 ppmvw

–4#= = =

ppw = (9.28 10–4)(900) = 0.8 psia

Step 3. From Figure 18.6b, interpolate between the 77°F and 100°F isotherms to fi nd the 86°F isotherm at a water partial pressure of 0.8 psia. The theoretical equilibrium loading, Xe, is 23.5 lbm of water adsorbed/100 lbm of adsorbent. For planning purposes, use 23 lbm of water adsorbed/100 lbm of adsorbent. This is often written as 23 wt. %.

COPYRIGHT

Petr

oSkill

s

CHAPTER 18 423



Step 3. Calculate the product specifi cation from Equation (18.2) using the water partial pressure from Step 2.

yw = ppw/P = (7.0 10–5 psia)/(900 psia) = 8.0 10–8 = 0.08 ppmv.

This is the lowest theoretical product specifi cation that can be made if regenerating with water-saturated feed gas. This value will be smaller if dry regeneration gas is used.

As demonstrated in Examples (18.1) and (18.2), Figure 18.6a and Figure 18.6b can be used to estimate the equilibrium water loading on the sieve at adsorption conditions, regeneration conditions, and to estimate the fi nal dry gas water content.

Figure 18.7 shows the static water equilib-rium capacity for a number of different adsorbents at 25°C. Notice that the shape of the molecular sieve isotherm is steep in the region between a relative humidity of 0 and 10%. The isotherm then fl attens out at roughly 15% relative humidity, with a maximum percent of water adsorbed of 22% (water adsorption capacity). This is favorable because high equilibrium loadings are achieved at low water concentrations (low relative humidity). In compari-son, the other adsorbent isotherms do not achieve a comparable capacity until the relative humidity is at least 40%. Molecular sieve’s high adsorption capacity at low water concentrations makes it the adsorbent of choice to achieve low water dewpoints.

The capacity of the desiccant is dependent upon the regeneration conditions. The water content (dewpoint) of the dry gas exiting the bed during the adsorption cycle will depend on the equilibrium loading at the bottom of the bed after the regeneration cycle is complete. The working capacity of the bed is dependent upon the average residual loading of water remaining in the entire bed after the regeneration cycle is complete. This residual water loading depends upon the temperature, pressure, and water content of the regeneration gas stream (both heating and cooling) as well as the direction of the regeneration gas fl ow.

The average residual loading in the bed will be greater than the equilibrium residual loading at the bottom of the bed (assuming upfl ow regeneration). This is because the regeneration cycle time is insuffi -cient to reach equilibrium conditions throughout the entire bed. For this reason an average residual loading should be used for estimating desiccant capacity. For molecular sieve the average residual loading can be assumed to be between 2-4 %.(18.2) In most cases, a value of 4% is appropriate.

The difference between the equilibrium loading during the adsorption cycle and the average residual loading is the net equilibrium loading. Xnew shown in Equation (18.3).

Relative Humidity, %

Wa

ter

Ad

sorb

ed

, %

1008060402000

4

8

12

16

20

24

28

Sorb

ead

HSo

rbea

d H

H-156 Alumina

Molecular SieveMolecular SieveSorb

ead

R

Sorb

ead

R

Figure 18.7 Static Equilibrium Curves for Various Commercial DesiccantsCOPYRIG

HT

Petr

oSkill

s



Example 18.3: From Example 18.1, determine the new net equilibrium loading using Equation (18.3)

Xe = 23% Xr = 4% Xnew = (23 – 4) = 19%

ΔXnew is for fresh activated molecular sieve. Over time, the equilibrium capacity of the desiccant will decline.

(18.3)

Where: Xnew = net equilibrium loading, new %Xe = inlet equilibrium loading %Xr = assumed average residual loading (4%)

X X – Xnew e r=

Figure 18.8 shows typical capacity decline curves for molecular sieve. The decline in capacity occurs for several reasons. One is a gradual loss of crystalline structure and/or pore closure.(18.3) A more troublesome cause of capacity decline is contamination of the sieve due to liquid carryover from upstream separation equipment. Liquid water, heavy hydrocarbons, glycol, amines, etc. can damage the desiccant and cause rapid loss of capacity.(18.4) Degradation due to contamination occurs primarily through blockage of the small capillary or lattice openings, which control access to the interior surface area. If these contami-nants are present upstream of the dehydration system, provision for removing them should be provided. In addition to a normal impingement separator, a coalescing fi lter separator or other high-effi ciency separation device is required.

A third source of capacity loss is due to “refl uxing” where water which has been desorbed during the heating cycle re-condenses in the upper part of the bed due to a large temperature gradient across the bed early in the heating cycle. This will be discussed in more detail in a later section.

The y-axis in Figure 18.8 is referred to as the life factor, FL and is plotted as a function of completed cycles (adsorption plus regeneration is one cycle) for three different operating conditions: Good, Average and Poor. Common to each curve is a rapid loss of capacity early in the sieve life with a slower capacity loss as more cycles are completed.(18.5)

Molecular sieve units should be designed based on a realistic life factor, FL. The value of the life factor is determined from experience and system design. The application of a life factor results in an aged equilibrium loading, ΔXa which includes the effect of the decline of equilibrium capacity as shown in Equation (18.4).

(18.4)

When an optimistic value of FL, such as the “Good” curve in Figure 18.8 is selected for predicting the performance of an adsorption unit, problems can occur if the actual conditions match an “Average” or “Poor” life curve. The more rapid decline of the aged net equilibrium loading, Xa, will often require that the adsorption cycle time be shortened. This, in turn, decreases the time available to regenerate the bed since a bed must be completely regenerated before the adsorption bed(s) experiences breakthrough.

FL=Xa Xnew

Where: ΔXa = net equilibrium loading, aged %ΔXnew = net equilibrium loading, new %

FL = life factor

COPYRIGHT

Petr

oSkill

s

CHAPTER 18 425


What this means in practice is that higher regeneration gas fl ows and heater duties are required. It is good practice to provide a margin of additional regeneration heater capacity to accommodate these potential shorter regeneration times.

Conversely, during early sieve life the net equilibrium loading of the desiccant is greater than the “aged” equilibrium loading. This allows longer adsorption times. For example, if the design adsorption time is 12 hours including the life factor, it is likely that early in the life of the unit it may be possible to operate using a 16-20 hour adsorption cycle which reduces the number of cycles and increases the desiccant life (increase FL ).

FL

, Lif

e F

ac

tor

00 500 1000 1500 2000 2500 3000 3500

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

1.00

Number of Cycles

Poor

Good

Average

Poor

Good

Average

Figure 18.8 Generic Molecular Sieve Capacity Decline Curves

Example 18.4: Using the data from Example 18.1 and the results from Example 18.3, determine the aged delta-loading. Assume a three bed system using 8-hour step times and a 3-year life based on “average” performance.

Step 1. Determine the number of cycles per tower.

(1 step/8 hrs)(1 cycle/3 steps)(24 hrs/day)(365 days/yr)(3 yrs/life) = 1095 cycles per tower

Step 2. From Example 18.3, ΔXnew = 19 %.

Step 3. Estimate ΔXa from Equation (18.4).

From Figure 18.8, FL = 0.60

ΔXa = (0.60)(19) = 11.4 wt %. This is the aged net equilibrium loading at the end of the three-year life.

COPYRIGHT

Petr

oSkill

s



The mass of the sieve in the net equivalent equilibrium zone, me, is calculated from Equation (18.5).

(18.5)

Equation (18.5) assumes that the average water saturation in the MTZ is 50%. This is shown in Figure 18.9 where a vertical line is drawn through the infl ection point, which creates two equal areas, I and II. Area I has no water loaded on the adsorbent, while area II has some water loaded on it. Because the two areas are assumed to be equal, the adsorbent in the rectangle “abcd” can be considered to be fully loaded with water and is the mass of sieve in the net equivalent equilibrium zone.(18.23) This allows us to take credit for half the weight of the mass transfer zone as shown in Equation (18.9).

The useful capacity (sometimes referred to as breakthrough loading) of the desiccant will be less than the adsorption cycle loading because, at breakthrough, the MTZ remains in the bed and the MTZ is not fully saturated with water. Thus, a longer MTZ results in a decrease in the useful capacity.

The length of the MTZ depends on gas composition, gas velocity, size and bulk density of desiccant, co-adsorption of other components, relative water saturation, and bed contamination. It can vary from 0.1 to 0.2 meters (0.3 to 0.7 feet) up to 1.5 to 2.0 meters (4.9 to 6.6 feet). The precise determination of MTZ is beyond the scope of this book. The MTZ length of an aged sieve may be estimated from Equation (18.6).(18.6)

(18.6)

The MTZ can also be expressed in terms of mass of desiccant:

(18.7)

m Xm t

100H 0/hre

a

a2=^ ê ^h ho h

SI FPSWhere: me = mass of sieve in net equivalent equilibrium zone kg lbm

ΔXa = net equilibrium loading, aged % % %mH20/hr = amount of water removed per hour see Eqn. (18.11)

ta = adsorption cycle time h hr

SI FPSWhere: MTZL = aged mass transfer zone length m ft

vg = superfi cial gas velocity m/min ft/minA = constant 10.7 35

KMS = particle size factor = 0.52 m for 3.2 mm pellets or 4x8 mesh beads

= 1.7 ft for 1/8" pellets or 4x8 mesh beads

= 0.26 m for 1.6 mm pellets or 8x12 mesh beads

= 0.85 ft for 1/16" pellets or 8x12 mesh beads

FL = life factor From Figure 18.8 From Figure 18.8

MTZ v /A K /FL g0.3

MS L= _ î h

SI FPSWhere: D = bed diameter m ft

tMS = bulk density of molecular sieve Table 18.1 Table 18.1

m v /A K 4D /FMTZ g

0.3MS

2

MS L= rt_ ^ î h h; E

COPYRIGHT

Petr

oSkill

s

CHAPTER 18 427


The useful capacity at breakthrough is shown by Equation (18.8).

(18.8)

The total amount of desiccant required in the bed is equal to the mass in the net equivalent equilib-rium zone, me, and half the mass in the MTZ, mMTZ, because half of the mass in the MTZ is already accounted for in the net equivalent equilibrium zone.

(18.9) The time to water breakthrough can be estimated from Equation (18.10).

(18.10)

The amount of water removed per hour is given in Equation (18.11)

(18.11)

In order to maximize the useful capacity of the adsorbent, the length of MTZ relative to the total bed height must be decreased. FL should be maximized by properly designing the inlet separation equipment. Note that the two are interrelated—a low life factor will result in a longer MTZ. As shown in Figure 18.5 (e), one way to decrease the MTZ is by using smaller size desiccant. This, however, will result in increased

Figure 18.9 Actual and Net Equivalent Equilibrium Zones

Flow

Xa

Net Equivalent Equilibrium ZoneHypotheticalUnused Bed

Xe

Xr

Actual Equilibrium Zone Actual MTZ

a

d c

b SI FPSWhere: Xuseful = useful capacity of

molecular sieve at break through

wt % wt %

ΔXa = net equilibrium load-ing, aged

wt % wt %

me = mass of sieve in the net equivalent equilibrium zone

kg lbm

mMTZ = mass of sieve in MTZ

kg lbm

mT = total amount of sieve in bed

kg lbm

X mX m

usefulT

a e= c m

m mT e 0.5 mMTZ= +

m 100X m

btH O/hr

useful T

2

=H ^ ^^ ^h hh h

SI FPS

Where: Hbt = time to water breakthrough h hrXuseful = useful capacity of molecular sieve wt% wt%

mT = total amount of molecular sieve in bed kg lbmmH2O/hr = amount of water removed per hour kg/h lbm/hr

Table 18.5 Example Useful Capacity of Desiccants

x, wt% Activated Alumina 5−12Silica Gel 5−8Molecular Sieve (4A) 7−14

m 24W q per bed

H O/hr2= ^^ ^ hh h

SI FPSWhere: mH2O/hr = amount of water removed per hour kg/h lbm/hr

W = water content of the gas kg/106 std m3 lbm/MMscfq per bed = gas fl ow per bed at standard conditions 106 std m3/d MMscfd

COPYRIGHT

Petr

oSkill

s



pressure drop so the choice of desiccant size must be carefully evaluated. The length of the MTZ is also a function of the gas velocity. (18.24)

An example of desiccant useful capacity values are shown in Table 18.5.

ADSORPTION SYSTEM DESIGN

The design of fi xed bed adsorption systems is complex. The desiccant manufacturers will typically perform these calculations with their own proprietary software. The output from the desiccant manufacturer will be a design summary sheet. Unlike many other processes, these calculations cannot be done with a process simulator. Most molecular sieve dehydration design summary sheets will contain the information that is provided in the adsorption system analysis method outlined in Table 18.6.

Table 18.6 Adsorption System Analysis Method

Adsorption System Analysis Step Information to defi ne / Calculations to be completed

1 Establish the design basis: feed gas fl owrate, composition, temperature, pressure, and product gas water content or dewpoint specifi cation.

2 Defi ne the total number of towers, size and type of adsorbent(s), fl ow direction; cycle times for adsorption and regeneration.

3 Estimate adsorbent bed diameter and mass or volume of adsorbent required per vessel.4 Estimate the height of the bed, the expected initial pressure drop through the bed, and

height of the adsorber vessel.5 Estimate the regeneration gas fl owrate for external or internal insulation. Defi ne regenera-

tion gas composition, heating and cooling temperatures, pressures and fl ow directions.6 Size regeneration gas heater and cooler.

Items 1 and 2 are typically provided by the end user. The remaining items are provided by the adsorption process vendor. The following discussion will provide relatively simple manual calculations which are often useful for preliminary designs, scoping studies and troubleshooting. These calculations also provide insight into molecular sieve design and operation.

The procedure as outlined in Table 18.6 follows:Step 1 Establish the design basis: feed gas fl owrate, composition, temperature, pressure, and

product gas water content or dewpoint specifi cation.From Example 18.1:The feed rate is 11.3 106 std m3/d [400 MMscfd] of 20.3 MW gas. The inlet pressure is 6205 kPa [900 psia], and the inlet temperature is 30°C [86°F]. The feed gas compressibility factor is 0.83.For the purposes of these calculations, it is assumed that all of the water is removed.

Step 2 Defi ne the total number of towers, size and type of adsorbents, fl ow direction and cycle times per tower.A three adsorber unit containing 4A mol sieve 4x8 mesh beads with two towers in adsorption and one tower in regeneration will be assumed. The adsorption cycle is assumed to be 16 hours (8 hour step time). The bulk density of the mol sieve beads is 705 kg/m3 [44 lbm/ft3].

Step 3 Estimate the adsorber diameter. The diameter of the adsorbers is set by the superfi cial gas velocity and allowable pressure drop. Several methods of determining the bed diameter are presented in the literature. Equation (18.12) is based on the Ergun equation.(18.7)

COPYRIGHT

Petr

oSkill

s

CHAPTER 18 429


(18.12)

Constants B and C for Equation (18.12) are:

Desiccant Shape and SizeSI FPS

B C B C4x8 bead 4.16 0.001 35 0.056 0 0.000 088 93.2 mm (1/8") extrudate 5.36 0.001 89 0.072 2 0.000 1248x12 bead 11.3 0.002 07 0.152 0.000 1361.6 mm (1/16") extrudate 17.7 0.003 19 0.238 0.000 210

Many designs are based on a fresh molecular sieve bed pressure drop of 41 kPa (6 psi). However, pressure drop increases over the bed life, and most molecular sieves can tolerate a pressure drop in excess of 69 kPa (10 psi).(18.8)

When Equation (18.12) is written in terms of velocity, it becomes:

(18.13)

Equation (18.14) gives a useful approximation of the maximum allowable superfi cial gas velocity during adsorption:

(18.14)

The actual fl owrate, qg, can be calculated from Equation (18.15).

(18.15)

SI FPSWhere: ΔP/L = pressure drop/length kPa/m psi/ft

n = gas viscosity cp cptg

= gas density kg/m3 lbm/ft3vg = superfi cial gas velocity m/min ft/min

LP B v C vg g g

2= + t

v 2C–B B 4C L

P

gmaxg

2g

0.5

=+ +

t

t^ ah k: D

v Agmax

g=

t

SI FPSWhere: vgmax = maximum superfi cial gas velocity m/min ft/min

ρg = gas density kg/m3 lbm/ft3A = constant: 3.2 mm [1/8"] 74 61

1.6 mm [1/16"] 53 43

q 1440q

PP

TT zg

s

s= c cm m

SI FPSWhere: qg = actual gas fl ow actual m3/min acfm

q = standard gas fl ow std m3/d scf/dayPs = standard pressure kPa psiaP = actual pressure kPa psiaT = actual temperature K °RTs = standard temperature K °Rz = gas compressibility factor at T and P — —

COPYRIGHT

Petr

oSkill

s



Calculation of the minimum tower diameter follows:

(18.16)

It is recommended to assume the actual vessel diameter is the next larger standard vessel size.

The velocity of the gas through the vessel based upon the actual vessel diameter can be calculated from Equation (18.17), and should be used for the remainder of the sizing calculations.

(18.17)

D v4q

mg

g=r

SI FPSWhere: Dm = minimum vessel diameter m ft

qg = actual gas fl owrate m3/min ft3/minvg = maximum allowable superfi cial gas velocity m/min ft/min

v /4 Dq

g 2g=

r hSI FPS

Where: vg = actual vapor superfi cial velocity m/min ft/minD = selected diameter of vessel m ft

Example 18.5: Estimate the adsorber diameter and mass of desiccant required with the data provided in Steps 1-3 listed above.

SI Solution:

A. Calculate fl owrate per tower:

q d11.3 10 std m

2 adsorbers1 5.65 10 std m /d

6 36 3#

#== c am kB. Calculate the actual volumetric fl owrate per tower from Equation (18.15):

q 14405.65 10

6205101

288303 0.83 56 m /ming

63# == c a a ^m k k h

C. Calculate the maximum superfi cial velocity from Equation (18.14):

0.83 8.314 303 60 mkg

g 3= =t ^ ^ ^h h h6205 20.3 Eqn. (3.3)

v60

74 9.5 m/mingmax = =

D. Estimate the minimum bed diameter from Equation (18.16): D 3.14 9.5

4 56 2.74mm = =^ ^^ h hh h

use the next larger standard vessel size: D = 2.90 mE. Determine actual velocity through the bed at selected vessel diameter from

Equation (18.17):

v3.14/4 2.90

56 8.5 m/ming 2= =^ ^h h

COPYRIGHT

Petr

oSkill

s

CHAPTER 18 431



F. Estimate the mass of desiccant in the net equivalent equilibrium section at end of life (aged condition), me from Equation (18.5):

From Example 18.1, feed gas water content = 690 kg/106 std m3 Adsorption cycle time = 16 hr From Example 18.4, ΔXa = 11.4% From Equation (18.11) the water removed per hour =

m 11.4162 kg/hr 16 hr cycle time 100

22 700 kge ==^ ^ ^h h h

G. Estimate mMTZ at end of life conditions from Equation (18.7):

From Example 18.4, FL = 0.60

From problem statement in Step 2, tMS = 705 kg/m3

m 10.78.5 0.52 4

3.14 2.90MTZ

0.3 2

= a ^ ^ ^ck h h h m 7050.60 3760 kg=a k

H. Estimate total mass of sieve, mT from Equation (18.9), and Xuseful from Equation (18.8) at end of life:

m 22 700 (0.5)(3760)T 24 580 kg=

X 24 580 10.5%useful = =c 11.4 22 700^ hmI. As a check on the cycle time assumption and bed sizing, estimate the water

breakthrough time from Equation (18.10):

And should match the specifi ed 16 hour adsorption cycle time.

Note that the total mass of desiccant in the bed is set by:• The useful capacity of the desiccant, Xuseful, at the end of sieve life and• A 16 hr. adsorption cycle time, ta, at the end of sieve life

When unit is fi rst brought online Xuseful will be signifi cantly higher than at the end of sieve life and the adsorption cycle time can be much longer.

J. Estimate Xuseful for new desiccant from Equation (18.8):

From Example 18.3, ΔXnew = 19%

X 24 580 17.6%useful = =c 19 22 700 m

m 24690 5.65 162 kg/hH O/hr2

= =^ ^h h

162 10010.5 24 580 16hbt = =H ^ ^h hh h

COPYRIGHT

Petr

oSkill

s




K. Estimate new desiccant breakthrough time after initial startup from Equation (18.10).

162 10017.6 24 580 26.7 hbt = =H ^ ^h h

FPS Solution:

A. Calculate fl owrate per tower:

q d400 MMscf

2 adsorbers1 200 MMscfd== a ak k

B. Calculate the actual volumetric fl owrate per tower from Equation (18.15):

q 1440200 10

90014.7

520546 0.83g

61977 ft /min3# == c ^m h

C. Calculate the maximum superfi cial velocity from Equation (18.14):

3.8 ftlbm Equation 3.3g 3t ^0.83 10.73 546^ h900 20.3 h Eqn. (3.3)

v3.861 31.3 ft/mingmax = =

D. Estimate the minimum bed diameter from Equation (18.16):

D 3.14 31.34 1976 9.0 ftm = =^ ^^ h hh h

use the next larger standard vessel size: D = 9.5 ftE. Determine actual velocity through the bed at selected vessel diameter from

Equation (18.17):

v3.14/4 9.5

1976 27.9 ft/ming 2= =^ ^h hF. Estimate the mass of desiccant in the net equivalent equilibrium section at end of life

(aged condition), me from Equation (18.5):

From Example 18.1, feed gas water content = 44 lbm/MMscf Adsorption cycle time = 16 hr From Example 18.4, ΔXa = 11.4%

From Equation (18.11) the water removed per hour =

m 11.4367 lbm/hr 16 h cycle time 100

51 500 lbme ==^ ^ ^h h h

G. Estimate mMTZ at end of life conditions from Equation (18.7):

From Example (18.4), FL = 0.60 From Table 18.1, tMS = 44 lbm/ft3

m 2444 200 367 lbm/hrH O/hr2

= =^ ^h h

COPYRIGHT

Petr

oSkill

s

CHAPTER 18 433



m 3527.9 1.7 4

3.14 9.5 8250 lbmMTZ

0.3 2

== a ^ ^ ^ck h h h m0.6044a k

H. Calculate total mass of sieve, mT from Equation (18.9), and Xuseful from Equation (18.8) at end of life:

m 51500 (0.5)(8250)T = + 55 625 lbm=

X 55 625 10.6%useful = =c 51^11.4 500hm

I. As a check on the cycle time assumption and bed sizing, estimate the water break-through time from Equation (18.10).

367 10010.6 55 625 16 hbt = =H ^ ^h h

This should match the specifi ed 16 hour adsorption cycle time.

Note that the total mass of desiccant in the bed is set by:• The useful capacity of the desiccant, Xuseful, at the end of sieve life and• A 16 hr. adsorption cycle time, ta, at the end of sieve life

When unit is fi rst brought online Xuseful will be signifi cantly higher than at the end of sieve life and the adsorption cycle time can be much longer.

J. Calculate Xuseful for new desiccant from Equation (18.8).

From Example 18.3, ΔXnew = 19%

X 55 625 17.6%useful = =c 19 51500^ hm

K. Estimate new desiccant breakthrough time after initial startup from Equation (18.10).

367 10017.6 55 625 26.7 hbt = =H ^ ^h h

Step 4 Estimate the height of the bed, the expected initial pressure drop through the bed, and height of the adsorber vessel.

Once the bed diameter and required mass of adsorbent are determined, the bed height, hB, can be calculated from Equation (18.18).

(18.18) h4 D

mB

MS2

T=t

r^ ` ^h j hSI FPS

Where: hB = required height of bed m ftmT = total amount of molecular sieve in bed kg lbmD = bed diameter m ft

tMS = density of molecular sieve Table 18.1 Table 18.1

COPYRIGHT

Petr

oSkill

s



Example 18.6: From Example 18.5, estimate the bed height required, initial pressure drop through the bed, and the estimated adsorber height.

SI Solution:

A. Estimate bed height, hB from Equation (18.18):

h705 3.14/4 2.9

24 580 5.3 mB 2^ ^ ^h h hB. Estimate actual bed pressure drop from Equation (18.12) and compare to the recom-

mended maximum pressure drop of 41 kPa. The viscosity of the gas, n, is 0.013 cp.

From Example 18.5, tg = 60 kg/m3 and vg = 8.5 m/min.

LP 4.16 0.013 8.5 0.00135 60 8.5 6.3 kPa/m2= + =^ ^ ^ ^ ^ ^h h h h h h

P 6.3 5.3 33 kPa. This is acceptable.^ ^C. Estimate adsorber height, hv from Equation (18.19):

h 5.3 m 1.5 m 6.8 mv =

An additional 3700 kg of desiccant could be added to the bed and remain within the pressure drop constraint of 41 kPa. This would increase the bed height by 1.2 m, and the adsorber vessel height to 8 m. This would allow for an increased adsorption cycle time of 3-4 hours, which would reduce the number of cycles per year, and extend the desiccant life.

The actual tower height (seam to seam) will be the bed height plus the height of bed supports including support balls, and suffi cient space to ensure good fl ow distribution at the top of the bed. This additional height is typically 1-1.5 m [3.3-5 ft]. An estimate for the vessel height is provided in Equation (18.19).

(18.19)

An example adsorption tower is shown in Figure 18.10 Figure 18.10 Schematic of Adsorber Loading and Internals

150 mm [6 in]Ceramic Balls

Floating Screen

Ceramic Support Balls

Fixed Screen

Ceramic Balls

Adsorbent

75 mm [3 in]

75 mm [3 in]

SI FPSWhere: hv = seam to seam height of

adsorber vesselm ft

hb = required adsorbent bed height

m ft

A = additional height required 1.5 m 5 ft

h h Av b= +

COPYRIGHT

Petr

oSkill

s

CHAPTER 18 435



FPS Solution:

A. Determine bed height, hB from Equation (18.18):

h44 3.14/4 9.5

55 625 17.8 ftB 2= =^ ^ ^h h hB. Calculate actual bed pressure drop from Equation (18.12) and compare to the recom-

mended maximum pressure drop of 6 psi. The viscosity of the gas, μ, is 0.013 cp. From Example 18.5, tg = 3.8 lbm/ft3 and vg = 31.3 ft/min.

LP 0.0560 0.013 27.9 0.000 088 9 3.8 27.9 0.28 psi/ft2= + =^ ^ ^ ^ ^ ^h h h h h h

P 0.28 17.8 5.0 psia. This is acceptable.^ ^C. Estimate adsorber height, hv from Eqn 18.19:

h 17.8 ft 5 ft 22.8 ftv =

An additional 11 200 lbm of desiccant could be added to the bed and remain within the pressure drop constraint of 6 psi. This would increase the bed height 3.6 ft, and the adsorber vessel height to 26.4 ft. This would allow for an increased adsorption cycle time of 3-4 hours, which would reduce the number of cycles per year, and extend the desiccant life.

If the selected bed diameter results in the pressure drop through the bed exceeding the recom-mended limit there are two options that may be considered:

1. Increase the assumed bed diameter to the next larger standard vessel size2. Decrease the adsorption cycle time

Note that if the adsorption cycle time is decreased, there will be less time available for the regeneration cycle. This reduced regeneration cycle time will increase regeneration gas fl owrate, and size of the regen-eration equipment.

The procedure to estimate the size of molecular sieve beds outlined in this text should not be a substitute for good judgment. Each situation should be evaluated on a case-by-case basis.

In addition to the P across the bed, the overall pressure drop through the adsorption system includes P through the piping manifold and switching valves and can have a signifi cant effect on CAPEX and OPEX of the facility.

A three or four-adsorber system offers the advantages of better bed geometry and more effective utilization of the adsorbent because of a smaller MTZ and increased operating fl exibility compared to a two-tower system.

The drawbacks of using multiple adsorbers during adsorption are higher capital costs and reduced time for regeneration.

Sensitivity studies comparing various combinations of number of adsorbers, adsorber diameters, pressure drops, step times, and regeneration gas rates would highlight the impact these variables have on the economic viability of a particular project.

COPYRIGHT

Petr

oSkill

s



Step 5 Determine the regeneration gas fl owrate, based on insulation type (external or internal), gas composition, heating and cooling temperatures, pressures and fl ow directions.

The required regeneration gas fl owrate required is a function of step times, the source of regen-eration gas, and the type of adsorbent that is used. The disposition of the spent regeneration gas must also be decided, as additional processing may be required to remove water and/or other concentration peaks, such as H2S.

The heating cycle is most critical because if the bed is not adequately regenerated it will not be able to meet the required water content (dewpoint). Heating must accomplish all of the following:1. Heat the desiccant to at least 204-316°C [400-600°F] (depends on desiccant)2. Heat and vaporize the adsorbed water3. Heat and vaporize any hydrocarbons on the bed4. Heat the vessel shell and steel internals (if not insulated internally)5. Heat the valves and piping between the regeneration heater and the adsorbers6. Supply heat lost through the insulation

Example 18.7: For comparison purposes, Table 18.7 provides the results of a three adsorber design with two alternative four adsorber designs. The sizing basis is for the conditions provided in Example 18.5 and Example 18.6, with the exception that the maximum allowable pressure drop of ~ 41 kPa [6 psi] was used for determining the maximum desiccant bed and vessel height.

Table 18.7 Comparison of Three Adsorber Unit with Four Adsorber Unit Designs

Parameter Three Adsorber UnitFour Adsorber Unit

(Three in Adsorption)

Four Adsorber Unit (Three in Adsorption)

Reduced Cycle TimeAdsorption Cycle Time 19.6 hr 24 hr 14.8 hrNumber of Cycles 894 821 1332FL 0.60 0.61 0.56ΔXa 11.4% 11.6% 10.7 %q per Adsorber 5.65 106 std/m3

[200 MMscfd]3.77 106 std/m3

[133.3 MMscfd]3.77 106 std/m3

[133.3 MMscfd]d 2.90 m [9.5 ft] 2.44 m [8 ft] 2.29 m [7.5 ft]vg 8.5 m/min [27.9 ft/min] 8.0 m/min [26.2 ft/min] 9.0 m/min [29.8 ft/min]me 27 900 kg [63 000 lbm] 22 400 kg [50 600 lbm] 15 000 kg [33 900 lbm]mMTZ 3760 kg [8250 lbm] 2570 kg [5650 lbm] 2600 kg [5600 lbm]mT per Adsorber 29 780 kg [67 100 lbm] 23 700 kg [53 400] 16 300 kg [36 700 lbm]Xuseful (aged) 10.7% 11.0% 9.9%Θbt (aged) 19.6 hr 24 hr 14.8 hrXuseful (new) 17.8% 18.0% 19.0%Θbt (new) 33 hr 39.4 hr 28.5 hrhb 6.4 m [21.5 ft] 7.2 m [24.2 ft] 5.6 m [18.9 ft]hv 7.9 m [26.5 ft] 8.7 m [29.2 ft] 7.1 m [23.9 ft]ΔP 40 kPa [6.0 psi] 40.2 kPa [6.0 psi] 40.1 kPa [6.0 psi]

COPYRIGHT

Petr

oSkill

s

CHAPTER 18 437


If the desiccant is a Type 4A or Type 5A molecular sieve, the bed temperature at the end of the heating cycle will be 260-288°C [500-550°F]. With a Type 3A molecular sieve, silica gel, and alumina, the maximum regeneration temperature typically ranges from 180-232°C [350-450°F]. The actual regeneration temperature should be the minimum required to adequately regenerate the bed.

Figure 18.11 shows a regeneration temperature profi le for a Type 4A molecular sieve. The tempera-ture, TH, is the hot regeneration gas into the bed. The temperature profi le T1 to T4 is the outlet gas temperature leaving the bed during the heating cycle. In this case, when the bed outlet temperature (T4) reaches approxi-mately 260°C [500°F], the heating cycle is fi nished and the cooling cycle begins. The temperature profi le T4 to T5 shows the bed outlet temperature during the cooling cycle.

The regeneration cycle can be divided into four (4) specifi c time intervals. Interval A (QA) is virtually all sensible heat. It represents the time required to heat the bed, steel, and adsorbed water from T1 to T2. At T2, desorption of the water begins.

Interval B (QB) is where the most of the water is driven off of the adsorbent. This requires suffi cient heat to revaporize the water and break the attractive forces that bind the water to the surface of the adsorbent. This is often called the heat of desorption. This value is approximately 4190 kJ/kg [1800 Btu/lbm] for sieves and 3260 kJ/kg [1400 Btu/lbm] for alumina and gels. The temperature profi le for interval B is T2 to T3. For commercial desiccants, T2 is about 110°C [230°F] and T3 is about 140°C [284°F].(18.9, 18.10)

Once the bulk of the water has been driven from the bed, interval C (QC) represents the time required to remove heavy contaminants and residual water. This is sometimes referred to as driving the “boot” off the bed. The temperature profi le for QC is from T3 to T4.

Interval D (QD ) is where the bed is cooled from T4 to T5 which is the end of the regeneration cycle. The regeneration gas fl owrate is established by an energy balance. The regeneration gas rate must

Tem

pe

ratu

re, °

C

Tem

pe

ratu

re, °

F

Typical Temperature Record of Dry Molecular Sieve

50

0

100

0

-25

200

300

400

500

600

0 1 2 3 4 5 6 7 8

T1

T2TB

T3

TC

TD

T5

T4TH

Start ofCycle

End ofCycle

100

250

300

150

200

A B C D

Regeneration Gas Outlet Temperature

Regeneration Gas Inlet Temperature

Ambient Air Temperature

Hours of Cycle

TA

Figure 18.11 Typical Regeneration Temperature Curves for a 4A Sieve

COPYRIGHT

Petr

oSkill

s



be adequate to deliver the required heat input in the time available. Likewise, it must also be suffi cient to deliver the cooling in the time available.

In equation form, this energy balance is summarized as follows:

(18.20a)

(18.20b)

(18.20c)

(18.20d)

Equations (18.20a)-(18.20d) can be solved for the mass fl ow of regeneration gas (m) subject to the constraint:

(18.21)

Equations (18.20a)-(18.20d) and (18.21) can only be solved by trial and error and are too tedious for manual calculations. One problem is the diffi culty to determine the effective temperature differences ΔTA, ΔTB, etc.

The temperature difference at any time is represented by the difference between curve 1 and curve 2 in Figure 18.11.

An alternative is to calculate the total heating load, QH, (QH = QA + QB + QC) and cooling load, QD. This approach raises the problem of determining the overall temperature difference (T) for the heating and cooling cycles. The temperature difference at any time is represented by the difference between curve 1 and curve 2 in Figure 18.11.

One method of estimating the average temperature difference is to use a log-mean temperature difference (Equation (18.22) and Equation (18.23)).

(18.22)

(18.23)

The calculation of the regeneration rate then follows:

(18.24)

Q mC TA p A A= D H

Q mC TB p B B= D H

Q mC TC p C C= D H

Q mC TD p D D= D H

SI FPSWhere: Q = heat load in a time interval kJ Btu

m = mass fl ow of regeneration gas kg/h lbm/hrCp = heat capacity of regeneration gas kJ/kg•°C Btu/lbm-°FΔT = temperatures at different time intervals show on Figure 18.11 °C °FΘ = length of time interval h hr

time available for regenerationA B C D1+ + +H H H H

Tln T – T

T – TT – T – T – T

heatingH

H 4

H 1

H 1 H 4= d^ ^

nh h

Tln T – T

T – TT – T – T – T

coolingD

5 1

4 1

4 1 5 1= d^ ^

nh h

m C TQ

C TQ

p H H

H

p D D

D= =

COPYRIGHT

Petr

oSkill

s

CHAPTER 18 439


Where HH and HD represent the total heating and cooling times, respectively.

To determine the actual temperature difference, numerical integration would be required.

Many references suggest that TH – T1 be substituted in place of ΔTH and QH be multiplied by 2.5 to simplify the calculation. The 2.5 factor is equivalent to assuming the effi ciency of the heating process is 40% (1/2.5). This eliminates the calculation of a mean temperature difference and gives results surprisingly close to the more complex methods. This is summarized in Equation (18.25).

(18.25)

The calculation of the regeneration gas rate is still iterative since the total regeneration time HH+ HD must be less than the time available. The regeneration requirements that must be met are:

a. To ensure good fl ow distribution and water removal, the minimum heating gas superfi -cial velocity should be greater than a velocity which results in a pressure drop equal to 0.23 kPa/m [0.01 psi/ft]. This velocity can be estimated from Equation (18.13).

b. The maximum heating gas superfi cial velocity should not result in a pressure drop greater than 5.4 kPa/m [0.24 psi/ft] during upfl ow regeneration to avoid lifting the bed.

c. De-pressurization and re-pressurization steps should not exceed 345 kPa/min [50 psi/min] to avoid bed lifting.

For the fi rst approximation, the following assumptions can be used:

a. Assume the regeneration gas fl owrate, m, is 10% of the process gas rate.b. Allocate 60-70% of the total regeneration time to heating, and 30-40% to cooling.

Stand-by time may be added to provide an extra design margin.

The heat loads QH and QO can be calculated by an energy balance. QH is the sum of sensible heat required to heat the steel, desiccant, support balls, valves, piping, plus the heat of desorption.

Calculation of QH (Heating Duty)From Figure 18.11, QH includes the heating intervals A, B, and C as shown in Equation (18.26):

(18.26)

Where QA is mainly sensible heat, QB is the desorption of the water, and QC is driving off the “boot” of the bed.

(18.27)

(18.28)

(18.29)

m C T – T2.5 Q

p H 1 H

H= ^ h

Q Q Q QH A B C= + +

Q 1.2 m C T – T (m ) C T – T m C T – TA steel p 2 1 T p 2 1 H O p 2 12+^ ^ ^h h hS D L

h hQ 1.2 m C T – T m C T – T m hB steel p 3 2 T p 3 2 H O desorp2+ D^ ^ ^ ^ ^ ^ _h h h h iS D

Q 1.2 m C T – T m C T – TC steel p 4 3 T p 4 3^ ^ ^ ^ ^h h hS D

COPYRIGHT

Petr

oSkill

s



Summing all of the heat load equations and including heat loss to the environment, QH reduces to:

(18.30) = +^ ^ ^ ^Q 1.2 m C T – T m – T m C T – T m hsteel p 4 1 T 1 H O p 3 1 H O desorpH 22+D^ ^ ^ ^ ^ _h h h h h is LD Qmisc_ i

SI FPSWhere: msteel = mass of steel kg lbm

mH2O = mass of water to be removed from bed kg lbmΔhdesorp = Heat of desorption kJ/kg Btu/lbm

Qmisc = heat loss and misc. heat loads kJ MMBtu/hr

Calculation of QD (Cooling Duty)

(18.31)

The mass of steel in Equation (18.30) is increased by 20% to account for the steel bed supports, piping, nozzles, manways, etc. The heat capacities, CP , in Equation (18.30) and (18.31) are those values applicable to the material shown in each term and are provided in Table 18.8. The last term in Equation (18.30) (Q misc.) accounts for heat losses and miscellaneous heat load (e.g., support balls, etc.). Q misc. is often estimated to be 15% of the calculated value.

Adsorber Vessel The heat required for the adsorber

vessel will depend on whether internal or external insulation is used. Where it is applied, internal insulation is of two types: (1) a steel “can” inside the shell that provides a stagnant gas space between the bed and shell or (2) cast or sprayed internal. If internal insulation is used, the castable type is typically selected for molecular sieve service. When internal insulation is used the adsorber vessel sensible heat load is only about 25% that of an exter-nally insulated tower. This reduces the overall heating load and is particularly benefi cial when the time available for heating is limited, e.g., short-cycle units.

Small cracks in internal insulation can contribute to bypassing gas around the molecular sieve bed. Bypassing just 0.1% of the gas can prevent successful expander or LNG plant operation. “Can type” internal insulation can develop cracks in the seal welds, and the can may warp over time due to

Table 18.8 Physical Properties for Heat Balance Calculations

Heat Capacity, Cp

CpS = Steel 0.50 kJ/kg•°C [0.12 Btu/lbm-°F]

CpL = Liquid Water 4.19 kJ/kg•°C [1.00 Btu/lbm-°F]

CpD = Desiccant see Table 18.1

Heat of DesorptionH2O on sieves 4187 kJ/kg [1800 Btu/lbm]H2O on gel or alumina 3256 kJ/kg [1400 Btu/lbm]Hydrocarbons 465 kJ/kg [200 Btu/lbm]

DensityCarbon Steel 7800 kg/m3 [490 lbm/ft3]Support Balls 1500 kg/m3 [94 lbm/ft3]

Q 1.2 m C T – T m – TD steel p 4 5 T p 4 5^ ^ ^ ^ ^h h hS D

DeadGas Space

Can

Vent

RefractoryInsulation

Can Type Cast Type

Figure 18.12 Types of Internal Insulation

COPYRIGHT

Petr

oSkill

s

CHAPTER 18 441


the temperature cycling. Both insulation types are illustrated in Figure 18.12. With internal insulation, the actual bed diameter usually is about 150 mm [6 in] less than the shell I.D.

External insulation is used in the majority of molecular sieve dehydrator installations. The approximate wall thickness of an adsorber tower is shown in Figure 18.13. This is based on

ASME Section VIII Division 1 and A-516 Gr 70 plate steel. For Section VIII Division 2 and BS 5500, use a wall thickness approximately 70% of that in Figure 18.13.

The mass of the steel may be estimated from Equation (11.29) in Chapter 11, page 47.

Regeneration Gas HeaterRegeneration gas heaters come in various types and confi gurations. Direct-fi red heaters are the

most common, salt-bath heaters are sometimes used in smaller units. Waste heat recovery from turbine exhaust may be feasible in some units. Regardless of the type of heater used, Equation (18.32) is used to estimate the duty.

(18.32)

De

sig

n P

ress

ure

, MP

aD

esi

gn

Pre

ssu

re, p

sia

Vessel I.D., mm

Vessel I.D., in

Wall Thickness, mm

Wall Thickness, in.

100806050403020

10

10

5

00 500 1000 1500 2000 2500 3000

2000

1500

1000

500

00 12 24 36 48 60 72 84 96 108 120

14

1.0 2.51.5

0.5

3.0 4.03.52.0

Figure 18.13 Approximate Pressure Vessel Wall Thickness

SI FPSWhere: QRH = regeneration heater duty kJ/h Btu/hr

m = regeneration gas mass fl ow kg/h lbm/hrCp = average regeneration gas heat capacity kJ/kg•°C Btu/lbm–°FTH = heater outlet temperature °C °FT1 = capacity heater outlet temperature °C °F

Q mC T – TRH p H 1= ^ h

COPYRIGHT

Petr

oSkill

s



Example 18.8: Using the data in Example 18.5 estimate:1. The regeneration gas fl owrate2. The heating time3. The cooling time4. The regeneration gas heater duty

The regeneration gas is “dry,” with the following properties (refer to Figure 18.11 for the defi nitions of T1, T2, T3, T4, T5, TH and the intervals A, B, C and D). External insulation will be used.Regeneration gas data:

z = 1.0, P= 2068 kPa [300 psia], γ= 0.59, average Cp = 2.68 kJ/kg•°C [0.64 Btu/lbm-°F], μ = 0.02 cp

Other data requirements for the calculation:

TH T1 T2 T3 T4 T5

288°C [550°F]

30°C[86°F]

110°C [230°F]

140°C [284°F]

260°C [500°F]

55°C[131°F]

Heat of desorption = 4187 kJ/kg water [1800 Btu/lbm] mT desiccant = 24 580 kg [55 625 lbm], Cp desiccant = 1.0 kJ/kg °C [0.24 Btu/lbm-°F] mH2O removed = 2600 kg [5870 lbm] (from Example 18.5) Vessel Design Pressure = 6830 kPa [990 psia] Vessel Dimensions: diameter: 2.9 m [9.5 ft], height: 6.8 m [22.8 ft]

SI Solution:

A. Estimate the mass of steel in each adsorber.

Mass of steel: From Figure 18.13, fi nd the wall thickness, t, is approximately 85 mm. From Equation (11.29), fi nd the mass of steel:

m 0.032 6.8 m 2900 5 mm 53 640 kgsteel == m

B. Estimate heat load, QH, from Equation (18.30).

Use the average plateau temperature during the plateau = (T2 + T3)/2 = (110 + 140)/2 = 125°C to calculate the sensible heat of water.

Steel: (1.2)(53 640 kg)(0.5 kJ/kg•°C)(260 – 30°C) = 7400 MJ Desiccant: (24 580 kg)(1.0 kJ/kg•°C)(260 – 30°C) = 5650 MJ

Regeneration Gas CoolerCalculation of the cooler load for all three heating intervals (A, B, and C) is needed to fi nd the

highest cooling load. It will normally occur in interval B where the majority of the water is desorbed from the bed and must be condensed in the regeneration gas cooler. The normal temperature approach will be 16-20°C [29-38°F] for air cooling and 8-10°C [15-18°F] for water cooling.

COPYRIGHT

Petr

oSkill

s

CHAPTER 18 443



Water (sensible): (2600 kg)(4.19 kJ/kg•°C)(125 – 30°C) = 1000 MJ Water (desorption): (2600 kg)(4187 kJ/kg) = 10 900 MJ Sub-total QH = 7400 + 5650 + 1000 + 10 900 = 24 950 MJ Add 15% to cover heat losses, support balls, etc. = 3740 MJ QH = 24 950 + 3740 = 28 690 MJ. C. Determine regeneration time.

From Equation (18.1), tstep = 16 h /(3 – 1) = 8 h

Depressurization: 345 kPa/min6205 – 2068 kPa 12 min=^^ h h

Repressurization: 345 kPa/min6205 – 2068 kPa 12 min=^^ h h

Total estimated pressure equalization time = 12 + 12 = 24 min. Assume 0.5 h Time for heat and cool = 8 – 0.5 = 7.5 hours Assume 65% for heat, 35% for cool: HH = (0.65)(7.5 hrs) = 4.88 hrs.

HC = (0.35)(7.5 hrs) = 2.62 hrs.

In this calculation, stand-by time has been ignored. Any addition of stand-by time would have to be deducted from the heating/cooling cycle.

D. Calculate the fl owrate of regeneration gas from Equation (18.25) using 4.88 hours for heating:

C Cm 2.68 kJ/kg 288 – 30 4.88 h2.5 28 690 000 kJ 21 300 kg/h:

== % %^ ^ ^^ ^ h h h

Convert the regen gas fl owrate to a standard volumetric fl owrate to determine percent-age of feed gas:

q h21 300 kg

0.59 29 kg1 kmol

kmol23.64 std m

d24 h 0.71 10 std ms

36 3#== a ^ ^c c ak h h m m k

Regen gas fl owrate percent of feed:

This is an acceptable percentage of the feed gas.

E. Check cooling load, QD, from Equation (18.31):

Steel: (1.2)(53 640 kg)(0.5 kJ/kg•°C)(260 – 55°C) = 6600 MJ Desiccant: (24 580 kg)(1.0 kJ/kg•°C)(260 – 55°C) = 5040 MJ QD = 6600 + 5040 = 11 640 MJ

11.3 10 std m /d0.71 10 std m /d 100 6.3%6 3

6 3

## == c ^m h

COPYRIGHT

Petr

oSkill

s




Calculate the time required for cooling from Equation (18.20c) for the regen gas fl owrate:

Estimate ΔTD as the log-mean temperature difference:

ΔTD = (230 – 25)/ln 230/25 = 92.4°CFor a regeneration gas fl owrate of 21 300 kg/hr:

t = 2.2 hr (note this is less than the cooling time allotted of 2.62 hr) OK F. Check to see if ΔP/L ≥ 0.23 kPa/m during heating using Equation (18.12).

Find vg from Equation (18.17):

qa = m/t

m = (21 300 kg/hr)(1 h/60 min) = 355 kg/min

t = (P)(MW)/zRT

t = (2068 kPa) [(0.59)(29 kg/kmol)] / (1)(8.314)(561 K) = 7.59 kg/m3

qa = (355 kg/min) / (7.59 kg/m3) = 46.8 m3/min

vg = qa/(π/4D2) = 46.8 m3/min/((π/4)(2.9)2 m2) = 7.1 m/min

Determine ΔP/L for beads from Equation (18.12):

LP 4.16 0.02 7.1 0.00135 7.59 7.1 1.1 kPa/m2 =^ ^ ^ ^h h

G. Estimate the regeneration gas heater size from Equation (18.32).

= = C CQ mC T – T21 300 kg/h

2.68 kJ/kg 288 – 30 4100 kW3600RH p H 1 : =% %^ ^^ ^ ^h h h h h

FPS Solution:

A. Estimate the mass of steel in each tower.

Mass of steel:

From Figure 18.13, fi nd the wall thickness, t, is approximately 3.4 in.

From Equation (11.29), fi nd the mass of steel:

m 13.8 22.8 ft 3.4 in 121 950 lbmsteel == ^ h114 in

B. Estimate heat load, QH, from Equation (18.30).

Use the average plateau temperature during the plateau = (T2 + T3)/2 = (230 + 284)/2 = 257°F to calculate the sensible heat of water.

Steel: (1.2)(121 950 lbm)(0.12 Btu/lbm-°F)(500 – 86°F) = 7.3 MMBtu

Desiccant: (55 625 lbm)(0.24 Btu/lbm-°F)(500 – 86°F) = 5.5 MMBtu

Water (sensible): (5870 lbm)(1.0 Btu/lbm-°F)(257 – 86°F) = 1.0 MMBtu

COPYRIGHT

Petr

oSkill

s

CHAPTER 18 445



Water (desorption): (5870 lbm)(1800 Btu/lbm) = 10.6 MMBtu Sub-total QH = 7.3 + 5.5 + 1.0 + 10.6 = 24.4 MMBtu Add 15% to cover piping, valves, losses, etc. = 3.6 MMBtu QH = 24.4 + 3.6 = 28.0 MMBtu.

C. Determine regeneration time.

From Equation (18.1), tstep = 16 h /(3-1) = 8 h

Depressurization: 50 psi/min900 – 300 psia

12 min=^^ h h

Repressurization: 50 psi/min900 – 300 psia

12 min=^^ h h Total estimated pressure equalization time = 12 + 12 = 24 min. Assume 0.5 h Time for heat and cool = 8 – 0.5 = 7.5 hours Assume 65% for heat, 35% for cool: ϴH = (0.65)(7.5 hrs) = 4.88 hrs.

ϴc = (0.35)(7.5 hrs) = 2.62 hrs.

In this calculation, stand-by time has been ignored. Any addition of stand-by time would have to be deducted from the heating/cooling cycle.

D. Calculate the fl owrate of regeneration gas from Equation (18.25) using 4.88 hours for heating:

m 4.88 hr^ ^ ^h h hF F0.64 Btu/lbm- 550 – 86 48 300 lbm/hr== 2.5 28 000 000 Btu^ hConvert the regen gas fl owrate to a standard volumetric fl owrate to determine percent-age of feed gas:

q hr48 300 lbm

0.59 29 lbm1 lbmol

lbmol379.5 ft

d24 hr

1 10 scfMMscf 25.7 MMscfds

3

6#== a ^ ^c c a ck h h m m k m

Regen gas fl owrate percent of feed:

400 MMscfd25.7 MMscfd 100 6.4%= =a ^k h

This is an acceptable percentage of the feed gas.E. Check cooling load, QD, from Equation (18.31):

Steel: (1.2)(121 950 lbm)(0.12 Btu/lbm °F)(500 – 131°F) = 6.5 MMBtu Desiccant: (55 625 lbm)(0.24 Btu/lbm °F)(500 – 131°F) = 5.0 MMBtu

QD = 6.5 +5.0 = 11.5 MMBtu

Calculate the time required for cooling from Equation (18.20c) for the regen gas fl owrate:

COPYRIGHT

Petr

oSkill

s




Estimate ΔTD as the log-mean temperature difference: ΔTD = (414 – 45)/ln 414/45 = 166°F

For a regeneration gas fl owrate of 48 300 lbm/hr:

QD = (48 300 lbm/hr)(0.64 Btu/lbm-°F)(166°F)(t) = 11.5 MMBtu

t = 2.2 hr which is less than the cooling time allotted of 2.62 hr. OK.

F. Check to see if ΔP/L ≥ 0.01 psi/ft during heating using Equation (18.12)

Find vg from Equation (18.17):

qa = m/ρ m = (48 300 lbm/hr)(1 h/60 min) = 805 lbm/min ρ = (P)(MW)/zRT ρ = (300 psia) [(0.59)(29 lbm/lb-mol)] / ((1)(10.73)(1010 °R)) = 0.47 lbm/ft3 qa = (805 lbm/min)/(0.47 lbm/ft3) = 1710 ft3/min vg = qa/(π/4D2) = 1710 ft3/min/((π/4)(9.5)2 ft2) = 24 ft/min Determine ΔP/L for beads from Equation (18.12):

LP 0.0560 0.02 24 0.000 088 9 0.47 24 0.05 psi/ft2 =^ ^ ^ ^h h h h

G. Estimate the size regeneration gas heater from Equation (18.32).

QRH = mCP(TH – T1) = (48 300 lbm/hr)( 0.64 Btu/lbm-°F)(550 – 86 °F) = 14.3 MMBtu/hr.

Regeneration Gas ConsiderationsDesign of the regeneration gas system depends on many factors including source and disposition of

the regeneration gas, sales or process specifi cations for the dehydrated gas and type of process downstream of the mol sieve unit, etc. For low temperature NGL extraction plants the regeneration gas will typically be taken from the process gas stream downstream of the demethanizer. This gas will be dry and will contain no heavy hydrocarbons. The regeneration pressure can either be high pressure (typically sales gas pressure) or low pressure (typically the expander booster compressor discharge). In the case of low pressure regenera-tion gas, the attainment of an acceptable residual water loading on the molecular sieve is usually limited by the rate of heat addition to the system. In low-pressure regeneration, the heating cycle can be switched to cooling when T4 reaches about 90% of TH. In some cases the switch from heating to cooling is done sooner than this and in a way that allows the cooling gas to push the hot heating gas out of the bed. This is called a “thermal pulse” cycle. At high pressure the limitation may be the water holding capacity of the regeneration gas. In high-pressure regeneration, the outlet temperature, T4, shown on Figure 18.12, may be held for a period of time to ensure that all of the desorbed water is being carried away. The time for such a “heat soak” will depend on the system operating conditions.

For LNG applications, the regeneration gas is generally taken from the fuel gas system and returned to the fuel gas system after use. This allows for low pressure regeneration. In some cases the source of the regeneration gas is immediately downstream of the mol sieve unit. The wet regeneration gas leaving the regeneration gas scrubber will often be recycled to the feed of the mol sieve unit using a small high speed

COPYRIGHT

Petr

oSkill

s

CHAPTER 18 447


centrifugal compressor. This design means the regeneration will take place at high pressure and the mol sieve unit will have to be designed to handle a larger inlet rate (feed gas + regeneration gas).

The disposition of the regeneration gas exiting the regeneration gas cooler and scrubber is an additional unit design decision. This gas will contain water and if it is blended directly into the sales gas the sales gas may not meet the water dew point specifi cation. For example, assume the regeneration gas fl ow is 8% of the sales gas and this gas leaves the regeneration gas scrubber water-saturated with a water content of 950 kg/106 std m3 (60 lbm/MMscf). The water content of the blended gas will be (0.92)(0) + (0.08)(950) = 76 kg/106 std m3 (4.8 lbm/MMscf). If this exceeds the sale gas specifi cation the regeneration gas will require dehydration which is typically done with a small glycol unit. The regeneration gas will be water saturated only during that portion of the regeneration cycle where water condenses in the regeneration scrubber (typically interval B in Figure 18.12).

Table 18.9 provides a summary of regeneration pressure advantages and disadvantages.

Table 18.9 Low Pressure vs. High Pressure Regeneration

Low Pressure High PressureAdvantages • Generally requires lower regeneration gas rates

because it has a higher water carrying capacity and more easily meets the minimum velocity requirement

• Can reduce refl uxing during regeneration (refl uxing is discussed in the following section)

• Eliminates or dramatically decreases the pressure equalization times

• Can be taken from export compressor discharge before the discharge cooler which decreases the regeneration heater duty

• More likely to be blended into sales gas stream without dehydration and if dehydration is required the unit will be smaller

Disadvantages • Effl uent regen gas holds more water and is less likely to meet sales gas specifi cations when blended or requires a larger glycol dehydration unit if it needs to be dried

• Longer pressure equalization times required when switching from adsorption to regeneration and vice versa which shortens the time available for heating and cooling

• Requires higher regeneration rates• May be contaminated with lube oil if taken

from export compressor discharge (more of a problem with reciprocating and screw compressors than centrifugals)

• More likely to lead to refl uxing during regeneration

Another regeneration issue can arise when the gas being dehydrated contains trace quantities of H2S and other sulfur compounds. The concentration of these components in the feed gas may meet the sales gas specifi cation but depending on the mol sieve type they can be adsorbed on the bed. As the feed gas enters the bed, both water and sulfur compounds will be adsorbed. The water is preferentially adsorbed relative to H2S and displaces the H2S in the equilibrium zone. As the adsorption cycle continues, so does this displacement. This can lead to a situation where the H2S concentration exiting the bed is higher than the inlet concentration for a period of time. More importantly, sulfur compounds will be retained on the on the bed at the end of the adsorption cycle. This will result in an H2S peak during regeneration. The spent regeneration gas will have to be handled in a way to mitigate the effects of this peak.

Regeneration RefluxOne cause of poor adsorber performance, such as high pressure drop, dusting, shortened adsorbing

time, or poor outlet dewpoint, can be traced to a condition often referred to as “refl uxing.” This situation occurs when water driven from the lower part of the bed during regeneration condenses on the cooler vessel walls and desiccant near the top of the bed and subsequently “rains” back down into the bed.

When refl uxing occurs, the condensed water drains down the bed until it contacts the heat zone moving up the bed. At this point, the water will boil. This rolling boil can wash away the binder and grind

COPYRIGHT

Petr

oSkill

s



the molecular sieve into a powder. Over time this powder is baked into a hard, donut-shaped cake. Even in less severe cases, this refl uxing phenome-non will reduce the molecular sieve capacity. Regeneration refl ux is especially troublesome if there is an amine carryover into the absorber from an upstream amine unit.(18.11)

There are options that can be adopted or prevent to minimize refl ux. A few of the more common methods are listed below. Reference 18.12 provides an excellent discussion of regeneration refl ux and prevention options.1. Utilize low pressure regeneration gas2. Minimize heat loss by using internal insulation3. Reduce the rate of heat input by ramping up the

temperature of the regeneration gas during the heating cycle; this will increase the time required for heating so may not be feasible in some systems

4. Work with adsorbent suppliers to see if they have adsorbent formulations modifi ed to be more re-sistant to the effects of refl uxing(18.11)

MECHANICAL DESIGN CONSIDERATIONS

Once a unit confi guration has been selected the mechanical design of the unit can proceed. This involves design of the vessels, including distribution devices, associated valves, regeneration gas heater, and piping layout.

Vessel Design As discussed in an earlier section, the vessel size is set by superfi cial gas velocity and mass of

desiccant required. Bed weights are substantial and, coupled with the ΔP forces incurred during adsorption, require that adequate support be provided. Because of the large temperature differences between the cycles, provisions must be made for thermal expansion of the bed support system.

Traditionally, wire mesh and grating assem-blies have been used to support desiccant beds. Johnson screens are also used, particularly for smaller sieve sizes. A Johnson screen is shown in Figure 18.15.

The annular space between the vessel wall and the edge of the bed support screen must be sealed to prevent loss of sieve particles. Heat resistant ceramic or fi berglass rope is packed between the grating and the vessel wall to prevent leakage. Inert ceramic balls are typically installed above this mechanical assembly.

Figure 18.14 Regeneration Reflux of High Pressure Gas

Top of Bed

Refluxing Section

Condensing Zone

AdsorbentHeat Zone

Figure 18.15 Johnson Screens (Courtesy of Johnson Screens)

COPYRIGHT

Petr

oSkill

s

CHAPTER 18 449


Uniform gas fl ow distribution through the bed is critical for effective dehydration. Poor distribu-tion, i.e., channeling, may result from several causes: poor distribution at the inlet and/or outlet nozzles. This may also contribute to impingement of high velocity gas on the bed surface. Baffl e plates and distribu-tion headers have proven successful in improving distribution.

Another type of gas channeling results from liquid carryover. After striking the inlet baffl e plate, liquids travel to the vessel wall and run down the perimeter of the bed. The liquid coats the adsorbent in contact with the vessel wall. During regeneration, heavy hydrocarbons can crack and form “coke like’ deposit on the desiccant which impedes mass transfer. The result is adsorption channeling, which shows up as a steady water leakage into the product gas.

A moisture sample probe should be located in the dehydrator desiccant bed about 0.5 m [1.6 ft] from the outlet end of the bed. This probe, when used in conjunction with the outlet gas moisture probe, offers a source of valuable information for troubleshooting dehydration problems, particularly if a possible channel exists down the wall of the vessel. It also permits capacity tests for optimizing time cycles.

All instrumentation openings, particularly in liquid treaters, should be equipped with screens to prevent molecular sieve entry.

Desiccant LoadingThe loading of adsorption dehydration and treating vessels is an important step in assuring that the

unit will perform as designed. The following procedures are recommended:1. Inert ceramic balls (75-150 mm [3-6 in] in depth) are usually placed on top of the bottom sup-

port screen to minimize adsorbent nesting in the screen opening and to aid in gas distribution. The diameter of the inert balls should be two to four times the size of the molecular sieve par-ticles. Usually two sizes of inert balls are used. The larger inert balls near the screen must be retained by the screen.

2. After the support balls have been loaded, the moisture probe should be inserted. Installation of probes is diffi cult after the desiccant has been loaded.

3. The desiccant should be loaded in a way that minimizes breakage and abrasion. Common methods include a plastic or cloth sock that reaches to the support balls from the top loading nozzle, a dump bucket or specially designed pneumatic loading systems which suck the desic-cant from bulk containers. Regardless of the method used it is important that the desiccant be uniformly distributed across the bed to prevent particle segregation based on size.

4. A layer of inert balls (19 mm to 25.4 mm [3/4" to 1" in] diameter) should be placed on top of the adsorption bed. This layer of balls will prevent movement of desiccant on the surface caused by high local velocities. A fl oating 20x20 mesh stainless steel screen should be placed between the inert balls and the molecular sieve to prevent migration of the denser balls down into the bed.

Feed Gas ConditioningWithout exception, the most frequently encountered operating problem in adsorption systems is

poor feed gas conditioning. The gas entering the bed should be free of all entrained hydrocarbons, treating chemicals (such as glycol and amine), free water, and solids. Although most desiccants are designed to withstand some carryover, signifi cant or persistent carryover of entrained material will cause premature reduction of bed capacity and/or mechanical damage to the desiccant material.

COPYRIGHT

Petr

oSkill

s



A properly sized impingement separator followed by a coalescing fi lter separator should be installed upstream of any solid bed dehydration system.

If the feed gas pressure is above cricondentherm pressure retrograde condensation may occur across the switching valves, piping manifold, and the bed itself. In these cases, it is advisable to heat the feed gas slightly (5-10°C [9-18°F]) to avoid condensation of hydrocarbons.

Some of the most common bed contaminants are:

Hydrocarbons: Heavy hydrocarbons such as crude oil, condensates, or lube oil are adsorbed by the binder in macropores which are much larger than the active sites for water. These high boiling-point fractions can undergo a number of reactions during regeneration, including cracking, polymerization, or coking. These high MW, non-volatile hydrocarbon components block or slow the diffusion of water molecules to the active sites. This results in a loss of useful capacity and premature breakthrough.

Light hydrocarbons, such as NGLs, can also be adsorbed in the macropores. Again this slows diffusion, hence the adsorption rate, and results in premature breakthrough. The difference is that light hydrocarbons are driven off the bed during regeneration and do not leave a non-volatile residue. Light hydrocarbon contamination can occur due to entrainment or retro-grade condensation.

Glycols: Glycols behave similarly to heavy oils in adsorption bed contamination. They are adsorbed in the macropores and decompose during regeneration forming a complex sludge that slows adsorption and “cements” molecular sieve particles together to form chunks. This encourages gas channeling which in turn leads to premature breakthrough.

Amines: Amines can also contribute to coking. In addition, ammonia is formed during regen-eration. Ammonia can attack the binder and weaken the physical structure of the molecular sieve, especially during high-pressure regeneration when regeneration refuxing is likely to occur. “Water wash trays” at the top of the amine contactor are recommended to minimize the carryover of amine compounds.

Salt: Salt usually enters a desiccant bed dissolved in entrained water. When the water is vaporized the salt accumulates and blocks macropores. In extreme case it can bind the desiccant particles together to form “chunks.” At some point it is necessary to replace the adsorbent. This can be a signifi cant problem in liquid dehydrators handling LPG, propylene, etc. from “salt dome” storage caverns.

Oxygen: If oxygen is present in the feed gas or regeneration gas, it will react with H2S and other sulfur compounds resulting in the deposition of elemental sulfur. In extreme cases, this can lead to the particle binding discussed previously for other contaminants.

Complications resulting from oxygen are not limited to the production of sulfur. Reaction with the hydrocarbons can form water and encourage deposition of “coke-like” deposits. Since oxygen can enter a system by a number of routes, it’s a good idea to request an oxygen analysis prior to unit start up.

H2S/Sulfur Compounds: H2S is adsorbed on Types 4A, 5A, and 13X molecular sieves. When H2S and CO2 are present in the feed gas, special molecular sieves, which minimize the formation of COS, should be considered. Type 3A molecular sieve falls into this category. H2S peaks during regeneration is another problem and were discussed previously.

If mercaptans are in the feed, they may decompose during regeneration and form a second H2S peak. The decomposition of the mercaptans can cause a capacity reduction similar to heavy hydrocarbons.

Methanol: Methanol is frequently used for hydrate inhibition in the production and gathering systems. Methanol has a vapor pressure higher than water, so signifi cant quantities of methanol can be

COPYRIGHT

Petr

oSkill

s

CHAPTER 18 451


present in the vapor phase in the feed gas. Methanol is adsorbed on Type 4A molecular sieves and increases the length of the mass transfer zone, thereby reducing the useful capacity of desiccant to adsorb water. It can also decompose during regeneration or form formaldehyde if oxygen is present. Reducing the regeneration temperature to 232°C [450°F] when Type 4A molecular sieves are in service will help minimize decompo-sition. If methanol is known to be present in the feed gas, the molecular sieve supplier should be notifi ed so that additional adsorption capacity can be included in the design. Some installations have successfully used Type 3A molecular sieves to prevent the adsorption of methanol. Type 3A molecular sieves are not as rugged as Type 4A. They are more sensitive to hydrothermal damage, have longer mass transfer zones compared to Type 4A, and require a lower regeneration temperature, typically 232°C [450°F].

Reference 18.17 provides a good summary of common molecular sieve contaminants.

SPECIAL CONSIDERATIONS

Adsorbents have been used to dehydrate, sweeten, and remove NGLs from natural gas streams for over 60 years. Early units employed silica gel and/or alumina. Molecular sieves were introduced to the gas processing industry in the 1960s. There have been many innovative applications over the years. A few of these are:

Compound Beds: Compound beds are desiccant beds that use more than one desiccant type or desiccant size. The purpose is to increase the useful capacity of the bed by increas-ing the equilibrium capacity or shortening the MTZ, or both. Assuming downfl ow adsorption, the most common example of a compound bed is the use of 8x12 mesh or a 1.6 mm diameter pellet (1/16 inch) molecular sieve at the bottom of the bed and 4x8 mesh or 3.2-mm diameter pellets (1/8 inch) of the same type of molecular sieve at the top. The equilibrium capacity of the two layers is the same, but the rate of water adsorption in the smaller desiccant is faster, hence a smaller MTZ.

Another compound bed application involves the use of two types of molecular sieves in one vessel. Type 4A molecular sieves could be installed on top of Type 13X molecular sieves if both water and iso-mercaptan removal is needed.

Compound beds using activated alumina or silica gel on top of molecular sieves have also been used. The alumina or silica gel has a higher static equilibrium capacity for water; and may also protect the molecular sieves from liquid carryover. These approaches have shown mixed success. It has been reported that activated alumina will lose most of its adsorption capacity if subjected to liquid water.(18.13) Even if no liquid water enters the vessel, the useful capacity of activated alumina at actual operating conditions might be lower than that of the molecular sieves. The protective layer of silica gel has a very limited capacity for liquids and might be saturated very quickly.(18.11)

H2S Removal: This application is generally limited to those cases where the concentration of H2S in the feed gas is less than 200-300 ppmv. The advantage of using molecular sieves is the simultaneous dehydration and sweetening of natural gas. A Type 5A molecular sieve is commonly used for this applica-tion; however, a Type 4A molecular sieve also has some useful capacity for H2S.

There are several important differences between H2S and H2O adsorption. First, the equilibrium capacity for H2S is much lower than for water. In addition, the MTZ is longer for H2S. These two factors mean much lower useful capacities for H2S relative to water. The result is signifi cantly shorter cycle times and/or longer beds. Water will be adsorbed along with the H2S, but the amount of adsorbed water will be small when the application is for H2S removal. This is the case because the breakthrough occurs with H2S not H2O.

COPYRIGHT

Petr

oSkill

s



A more detailed discussion of gas sweetening using adsorption processes can be found in Volume 4 of the Campbell “Gas Conditioning and Processing” series, Gas Treating and Sulfur Recovery.

Adsorption vs. Glycol: In general, when the outlet water dewpoint requirement is above –40°C [–40°F], glycol (TEG) dehydration or glycol (MEG) injection is generally preferred due to lower capital and operating costs. If H2S is present in the gas, a fi xed-bed system may be chosen over glycol because of the co-absorption of H2S in the glycol and subsequent environmental impact at the regenerator.

Another factor may affect the dehydration decision is the presence of aromatic hydrocarbons in the feed gas.

Aromatic hydrocarbons are not adsorbed on molecular sieves and are recovered as a liquid product in silica gel units; hence, no environmental problem. Some companies have used dry desiccant units (in lieu of glycol) for this reason.

Another related topic that is often raised in the design of dry desiccant systems is whether or not to dehydrate the gas with glycol upstream of the adsorption. This may be diffi cult to justify economically, but there are a number of positive benefi ts to this confi guration. First, the water content of the feed gas is signifi cantly lower, meaning longer adsorption cycle times and fewer regeneration cycles. This can increase the life of the desiccant and possibly save several bed change outs over the facility life.

HYDROCARBON RECOVERY

The basic mechanism for hydrocarbon recovery is similar but more complex than dehydration. This involves multiple mass transfer zone behavior.

Hydrocarbon recovery, condenser operation has a critical effect on recovery. The adsorption bed simply serves to concentrate the recoverable components so that condensation is more effi cient. The temperature and pressure of condensation is a critical parameter governing plant performance. Depending on the length of the adsorption cycle, a silica gel plant with ambient condensation is limited to minimal butane recovery and 75-90% of the pentanes and heavier.

The regeneration gas fl ow is typically 10-15% of the feed gas. The net effect is to make the regen-eration gas 6-10 times richer in condensable hydrocarbons than the feed gas.

Silica gel plants have been used primarily on lean gas streams (< 0.5% C5+) for hydrocarbon and water dewpoint control where other methods of processing are not economically attractive. The most popular application has been the processing of gas produced from gas storage reservoirs. The untapped potential for adsorption technology as an alternative to refrigeration appears large particularly at higher pressures (above the cricondenbar) or in situations where low pressure drop and faster startup are desired.

Figure 18.16 shows a variation on the 3-tower system used for Short Cycle Units (SCU) or Hydro-carbon Recovery Units when simultaneous adsorption of C5+ hydrocarbons and water is required. In this system, one bed is in adsorption, one in heating, and one in cooling. In other words, only one tower is undergoing adsorption while two towers are in the regeneration cycle.

The adsorption cycle in these facilities may sometimes be less than one hour. This does not leave suffi cient time to completely heat and cool one bed during the regeneration cycle, hence the use of heating and cooling respectively. In this scheme, feed gas is used as the regeneration gas source. Heating counter-current to adsorption will remove the light hydrocarbon components from the bottom of the bed, which helps strip off the heavy hydrocarbons that are adsorbed higher in the bed.

COPYRIGHT

Petr

oSkill

s

CHAPTER 18 453


In Figure 18.16, the regeneration gas fl ow is created by holding a small amount of back pressure on the inlet side of the unit and using the control valve ΔP to force the regeneration gas through the loop. Alter-natively, a small, in-line, high-speed centrifugal compressor can be installed in the regeneration loop. The latter requires higher maintenance but is more energy effi cient.

FRC

Abs

orbe

r

Hea

ting

Cool

ing

Inlet Gas

Outlet Gas

Heater

Sepa

rato

r

Figure 18.16 Schematic View of a Typical Three-Tower Plant Using Cooling and Heating in that Order

Tower 1

Tower 2

Tower 3

Heating G

as In

Heating G

as Out

Cooling G

as Out

Cooling G

as In

Proce

ss Gas I

n

Proce

ss Gas O

ut

Figure 18.17 Pipe Manifold for Three Tower Adsorption Plant (Tower 1 is Adsorbing, Tower 2 is Heating, Tower 3 is Cooling)

COPYRIGHT

Petr

oSkill

s



Figure 18.17 shows the piping manifold for a three-tower plant using a general confi guration like that shown in Figure 18.16. Two-way switching valves are shown. The most common switching valves employed are Orbit® ball valves, although others are used.

Process CharacteristicsThe capacity of most silica gel is about the same for hydrocarbons as for water. Activated carbon,

of course, has no effective capacity for water.

Figure 18.18 shows the equilibrium capacity of silica gel for various hydrocarbons in a two-compo-nent gas where the second component is methane. The fi gure shows both static equilibrium (from cell tests) and dynamic equilibrium or fresh breakthrough loading (from fl ow tests).

Notice that all curves are approaching the monolayer capacity of the gel. The monolayer capacity is found by assuming that only one layer of molecules is held to the solid surface. Knowing both surface area and molecule size, the capacity of the gel can be estimated. This means that ultimate capacity for any component adsorbed on silica gel is fi xed by surface area, provided that the component is small enough to enter the interior of the adsorbing particle.

Figure 18.18 Hydrocarbon Equilibrium on Silica Gel(18.14)

COPYRIGHT

Petr

oSkill

s

CHAPTER 18 455


Actual capacity for any component is fi xed by the zone movement previously described, bed geometry, equilibrium capacity, and gas fl owrate. The surface of the adsorbent is always occupied by some molecule. As the zone of a given component progresses down the bed, it must displace the molecules already there. The rate of displacement depends on their relative equilibrium loading.

Theoretically, the zone for any component cannot move any faster than it can completely displace the materials ahead of it. In actual practice, at commercial fl owrates, the zones tend to “over-run” each other. Therefore, true chromatographic separation does not occur. This is illustrated in Figure 18.19.

Once the front of the mass transfer zone (MTZ) for a particular component reaches the outlet of the bed, the ratio of outlet to inlet concen-tration (C/C0) starts to increase. When this ratio reaches one, all primary adsorption ceases for that component (in this case, component 1). Desorption now begins because the MTZ for the less volatile component (component 2) behind it is displacing the component 1 previously adsorbed. The concen-tration ratio rises above unity but again approaches one when this next zone starts breaking out the end of the bed. This process continues until the cycle is terminated and regeneration of the bed begins.

In Figure 18.19, C0 = component inlet concentration and C = component outlet concentration.

Area A is representative of the amount of component 1 adsorbed, and Area B is representative of the amount desorbed by the zone following behind. The latter is smaller than the former. At time θB for component 2, (Area A—Area B)/Area A is generally about 0.35-0.40. Thus, immediate displacement has not occurred.

This is shown in Table 18.10 by the test data on a fi eld unit.

Even recognizing a 6-10% error in sampling and analysis, no sharp separation has occurred. The propane has broken through in less than two minutes. This data is not characteristic of an effi cient process because even good pentanes recovery is not obtained early in this example cycle. Beyond this point, the exit stream is being enriched.

If liquid recovery is the goal, some net amount of component is available even after its zone passes from the tower. The recovery will simply be less. If hydrocarbon dewpoint control is the goal, such enrich-ment is probably intolerable for components heavier than C6. In such a case, C6 breakthrough approximates the maximum cycle time. Liquid recovery plants tend to use shorter cycles than dewpoint control plants, since the recovery of C5 is often a process objective. In dewpoint control plants, the objective is to meet a hydrocarbon dewpoint specifi cation, which may require effi cient recovery of C7 or C8+ components only.

Figure 18.19 Illustration of the Adsorption/Desorption Process

B

A

1.0

Time

B

1 2

CC0

Table 18.10 Silica Gel Bed Adsorption Performance

Time, min C3 iC5 nC5

Inlet mol% (C0)0 1.020 0.120 0.085

Outlet mol% (C)2 0.804 0.031 0.01212 0.976 0.074 0.03122 0.938 0.075 0.04332 0.962 0.077 0.06542 0.970 0.155 0.13052 0.952 0.109 0.102

COPYRIGHT

Petr

oSkill

s



Figure 18.20 shows the breakthrough curves for several natural gas components on a silica gel bed.(18.15) No time values are shown on the x-axis because breakthrough times depend on gas velocity, bed length, and inlet composition of C5+ components. Component breakthrough typically occurs in order of increasing boiling point, with the exception of aromatic hydrocarbons, which are retained for a longer time due to their slight polarity.

With activated carbon, the zones tend to move slower. One reason is that water does not promote displacement. Basically though, the zone speed is lower because of the greater affi nity of carbon for the lighter hydrocarbons.

Regeneration and RecoveryFigure 18.21 summarizes the regeneration behavior of a short-cycle plant. Notice that the materials

do not desorb at a constant rate. The pentanes and lighter components start desorbing almost immediately. The hexanes and heavier concentration in the exit gas peak after a fi nite time.

Figure 18.21 (a) & (b) show the composition of various hydrocarbon components in the regenera-tion gas vs. time. Obviously, the composition of the stream to the regeneration gas cooler (condenser) varies continually with time. For this reason, a series of fl ash calculations must be made with time to accurately represent the liquid recovery to be expected. The minimum amount of regeneration gas necessary to increase the concentration of recoverable components in the condenser should be used. The use of refrigeration in the condenser will likewise enhance recovery. Even at temperatures as high as 16-20°C [60-68°F], a marked improvement in recovery effi ciency may be realized.

Elapsed Time (min)

Re

lati

ve

Co

nce

ntr

ati

on

(O

utl

et/

Inle

t)

0.0

0.2

0.4

0.6

0.8

1.0

1.2

C7 Benzene

Toluene

C8

C9

C6

C5C5C6C7C8C9

Benzene

Toluene

Figure 18.20 Breakthrough Profiles for Hydrocarbons on Silica Gel

COPYRIGHT

Petr

oSkill

s

CHAPTER 18 457


General SpecificationProper design of silica gel and activated carbon hydrocarbon recovery units is complex and requires

special simulation software. A detailed design method is beyond the scope of this text, although the princi-ples governing it have been presented here. References 18.15 and 18.18 summarize developments in the use of silica gel units for hydrocarbon dewpoint control.

Most obvious problems stem from improper specifi cation. The following minimum specifi cations are suggested:1. Gas fl ow limited to the rate previously discussed for dehydration (to promote desiccant life)2. Cycle length should be fi xed by the following considerations:

a. Not less than 15 minutes for gas containing pentanes and heavierb. Adsorption time set by breakthrough time for the component for which recovery is

desired or which must be removed for dewpoint control

Tem

pera

ture

, °F

Tem

pera

ture

, °F

Bed Outlet Temperature, °F

Out

let B

ed Te

mpe

ratu

re, °

F

Time, % of Cycle Length

Peta

ne, m

ol %

d) Characteristic Pentane Content of Regeneration

Gas vs. Bed Outlet Temperature

c) Outlet Bed Temperature Produces

Characteristic Curve

a) With Butane, Pentane, Hexane b) With Butane, Hexane, Heptane

Rege

nera

tion

Gas

Com

posi

tion,

mol

%

Rege

nera

tion

Gas

Com

posi

tion,

mol

%

Time, minTime, min

500

400

300

200

100

500

400

300

200

100

500

400

300

200

100

00

00 20 40 60 80 100 0 40 80 120 160 200 240 280 320 360

5 10 15 20 25 30 350 5 10 15 20 25 30 35

1

2

3

4

5

0

0.2

0.4

0.6

0.8

1.0

0

0.2

0.4

0.6

0.8

1.0

1.2

n-Butanen-Pentane

n-Hexane

Pentane

Hep

taneH

exan

e

Note: 1. Concentrations for n-butane and n-hexane in Figure a) are actually one-tenth of the values shown on the y-axis. 2. Concentrations for n-pentane and n-hexane in Figure b) are actually one-tenth of the value shown on the y-axis. 3. Composition measurements were taken at 3 minute intervals.

Note: Points shown representspread of data obtained.

Figure 18.21 Typical Regeneration for Hydrocarbon Adsorbers(18.16)

COPYRIGHT

Petr

oSkill

s



3. Bed length should be at least 5 m [16.5 ft]4. Inlet regeneration gas temperature should be at least 230°C [446°F] and preferably 260°C

[500°F] when processing gases containing pentanes and heavier5. The alternative of using refrigeration instead of ambient cooling in the condenser should be

consideredItems 1-4 are not independent; each affects the other. If the adsorption cycle length is less than

15 minutes, it is almost impossible to properly regenerate the bed. If these 15 minutes are greater than the breakthrough time for a key component, some compromise is needed. Breakthrough time depends on gas velocity and bed length (for a given gas composition and adsorbent). Economics and/or process needs will govern the compromise. It is recommended that the vendor furnish the adsorption effi ciency as well as condenser recovery. Adsorption effi ciency is simply that fraction of the component entering during the proposed cycle length that is retained on the adsorbent. This provides the basis to determine the relative merit of the competitive bids and provides the information required to request changes prior to purchase.

LIQUID DEHYDRATION

Silica gel, alumina, and molecular sieves may be used to dry hydrocarbon liquids. The fl ow scheme is similar to that for gas. Some larger plants are designed so that the fl ow may be reversed to “loosen” the bed if it has been compacted or to free the retaining screens of sediment. This provision is seldom needed for fractionated liquids. If there is any possibility of free liquid water being present, a coalescer should be provided upstream of the desiccant beds.

One difference between gas and liquid dehydration is in the regeneration cycle. The two systems commonly used to provide regeneration are:

1. Gas.2. Closed Vapor (either gas or vaporized product).

Table 18.11 summarizes these processes. Natural gas (sales or residue) is typically used for regen-eration in liquid desiccant systems. The gas is heated to approximately 160°C [320°F] for alumina; higher temperatures are required for molecular sieves. Liquid dehydrators are often operated upfl ow for adsorption and downfl ow for regeneration.

Table 18.11 Comparison of Regeneration Practices

Method Advantage Disadvantages Common UsageNatural Gas • Uses readily available

material• Low operating cost• Simple construction• Readily adaptable to auto-

matic control

• Introduces some additional safety hazards

• Requires compressor if high pressure gas not available

• Field locations, Product Pipelines, Non-Volatile Liquids

Closed Vapor(vaporized `product)

• Simple operation• Low cost of operation• Minimizes loss of valuable

volatile liquids• No contamination of

product

• Control of system more critical

• Requires pumping equip-ment

• Requires effi cient conden-sation of exit regeneration gas

• With volatile liquids such as propane, butane, etc.

• Where composition of feed is substantially constant

COPYRIGHT

Petr

oSkill

s

CHAPTER 18 459


Design ConsiderationsThe solubility of water in sweet hydrocarbons is shown in Figure 18.22. Notice that it is far more

soluble in many unsaturated and aromatic hydrocarbons than in the normal paraffi ns. Knowledge of compo-sition is thus very important. The presence of sulfur compounds enhances water solubility.

If no data are available for the specifi c liquid, a weight fraction relationship may be used to estimate liquid mixture water content.

Most contracts specify that the dried liquid show a negative result to the Cobalt Bromide test, which is equivalent to a water content of 15-30 ppmw. This is virtually bone dry, for ppmw is weight percent times 10 000. This requires the assumption all incoming water is removed in the bed.

Liquid velocity should be 1-2 m/min. This will fi x tower diameter. Tower length will usually be shorter for gas. An (L/D) ratio of 2-3:1 is common. As little as three seconds contact time is commonly provided. Some operators require a minimum bed length of 1.5 meters.

Activated alumina has been used quite widely for liquid drying since it is relatively inexpensive and tower costs are minimized at low pressure. An effective capacity of 5-12 kg of water per 100 kg alumina is common; this is equivalent to that in gas service. The capacity of gels and sieves in liquid dehydration service is likewise similar to their gas drying capacity.

Figure 18.22 Solubility of Water In HydrocarbonsCOPYRIGHT

Petr

oSkill

s



A Type 3A molecular sieve is useful for drying liquids to minimize co-adsorption of NGL compo-nents and heavier hydrocarbon molecules. The pore opening size is too small to admit these contaminants to the interior particle surface. Olefi ns, for example, may “tie-up” the available surface and reduce water capacity when using alumina.

Molecular sieves may be used also to both dry and sweeten liquids as discussed in Volume 4.

REFERENCES18.1 Malino, H.M., “Fundamentals of Adsorptive Dehydration”, 62nd Annual LRGCC, (Feb. 2012). 18.2 Meyer, P., “Easy and Sophisticated Debottlenecking of Molecular Sieve Plants”, Hydrocarbon World, Vol. 5 Issue 1, (2010).18.3 Meyer, Peter, “Hydrothermal Damaging of Molecular Sieve and How to Prevent It”. GPA Europe Conference, Paris, (Feb.

21, 2003).18.4 Elizondo, T., R. Bombardieri, “Molecular Sieve Dehydration Operational Problems and Solutions”, 87th GPA Annual Con-

vention, Grapevine, (March, 2008).18.5 Carlsson, A.F., Rajani, J.B., Kodde, A.J., “Finding the Fountain of Youth for a Mol Siv Dehydration Unit”, 83rd GPA Annual

Convention, New Orleans, (March 2004).18.6 Gas Processors Association Engineering Data Book, 13th Edition, 2012.18.7 Ergun, S. “Fluid Flow Through Packed Columns,” Chem. Eng. Prog., Vol. 48, No. 2 (Feb. 1952), p. 89.18.8 Private communications with Hank Rastelli, UOP, (March 2012).18.9 Meyer, P., “Molecular Sieve Troubleshooting”, GPA Europe Annual Conference, Lisboa, (Sept. 2010). 18.10 Rastelli, H., Stiltner, J. “Modeling to Optimize Natural Gas Dehydration”, GAS (2009).18.11 Rastelli, H., and Stiltner, J., “Extending Molecular Sieve Life in Natural Gas Dehydration Units”, 86th GPA Annual Conven-

tion, San Antonio, (March, 2007).18.12 Trent, R. E., “Regeneration Refl ux,” Zeochem Product Bulletin.18.13 LRGCC Round Table discussions, Hank Rastelli, UOP.18.14 Ashford, F., PhD. Thesis, Univ. of Oklahoma (1970).18.15 Harris, D. M. and D. W. Ingram, “Adsorption: The Flexible Solution for Gas Processing for NAM’s Underground Storage

System,” European GPA Continental Meeting, Budapest (1999).18.16 Campbell, J. M., Oil and Gas J., (Feb. 21, 1966), p. 93.18.17 Dewhirst, Mike, “Effect of Contaminants on Molecular Sieves,” 14th Annual GPA European Chapter Meeting (Sept. 1997),

Antwerp, Belgium.18.18 Schultz, T., “Gas Conditioning for Underground Storage Applications,” European GPA Continental Meeting, Zurich (Sept.

1998).18.19 Turnock, P. H. and K. J. Gustafson, Gas Cond. Conf., Norman, Oklahoma (1972).18.20 Kunkel, L. V. and J. W. Chobotuk, NGPA Meeting, Dallas, Texas (1973).18.21 SpectraSensors, “SS2100 Water Measurements in Desiccant Dryer Outlets”, Product Code 33701 Bulletin.18.22 An Introduction to Zeolite Molecular Sieves, UOP white paper.18.23 Lukchis, G.M., “Adsorption Systems Parts I, II and III”, Chem Eng., (June 11 1973, July 9, 1973, Aug 6 1973).18.24 Leavitt, F.W., “Non-Isothermal Adsorption in Large Fixed Beds”, Chem. Eng. Prog. 58, 54-59, (1962).

COPYRIGHT

Petr

oSkill

s

Documents

GAS CONDITIONING AND PROCESSING Volume 2: The Equipment