REACTION ENGINEERING OF COAL LIQUEFACTIONmegacarbon.com/synfuels/literature/coal liquefactionold.pdf · REACTION ENGINEERING OF COAL LIQUEFACTION Ken K. Robinson Mega-Carbon Company

REACTION ENGINEERING OF COAL LIQUEFACTION

Ken K. Robinson Mega-Carbon Company St. Charles, IL

Table of Contents

INTRODUCTION

A Brief History of Coal Liquefaction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Process Overview: What Goes In and What Goes Out . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Need to Motivate Remaining Sections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

COALS AND CONVERSION CHEMISTRY

Coals: Classification, Nomenclature and Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Coal Reaction Chemistry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Coal Expts ---> Solubility Lumps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Model Compound Expts ---> Elementary Steps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

THE ROLE OF CATALYSTS Catalysis

Coal Liquefaction Catalysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Coal Liquids Upgrading via Hydroprocessing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Mass Transport Aspects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Catalyst Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

REACTION MODELING

Global Models (Lumped Models) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Molecular Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Mechanistic Models

REACTOR TYPES/MODELS AND PROCESS DESIGN

General: What are Reactors? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Laboratory Reactors

Coal Liquefaction Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Model Compounds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Pilot Plant ProcessesConjecture About Commercial Scale . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

INTRODUCTION

A Brief History of Coal Liquefaction

Interest in coal liquefaction has had several high and low periods over the last 60 years. It starts just

prior to World War II with the Bergius Process used by Germany to fuel their war effort, continues

through the 50s and 60s with the research by the U.S. Bureau of Mines, and then hit a extensive pace

following the Arab oil embargo of 1973. Since the early 1980s, interest has waned due to the decrease

in crude oil prices, but the recent invasion of Kuwait by Iraq has caused the world to reevaluate their

position and think about a national energy policy which will provide incentives to develop U.S. natural

resources. The largest source of liquid hydrocarbons is crude oil and the United States has found and

used up the cheap domestic crude that it once had. When demand exceeds supply, or an errant ruler

decides to invade a neighboring country, the price of crude oil will increase. This makes a wide variety

of options available to produce liquid hydrocarbons from alternative energy resources. Coal liquefaction

would be an attractive option for the United States due to the vast coal reserves.

In a 1988 article, Lumpkin (Lumpkin 1988) said "the economics derived early in the 1980s established

the price of transportation fuels from coal liquefaction at $80 per barrel or higher. However there have

been dramatic improvements in the technology since 1983 that have not been widely appreciated.

Recent designs and cost estimates show that a 60% decrease in the cost of liquid fuels from coal relates

to about $35/barrel for crude oil".

Direct liquefaction involves the addition of hydrogen to coal in a solvent slurry at elevated temperature

and pressure. The solvent provides a convenient transportation medium for the coal and enhances heat

and mass transfer during chemical reaction. In many processes, the solvent shuttles hydrogen from the

gas phase to the coal and is called a donor solvent. The elevated temperature cracks the coal molecules

by thermally rupturing carbon-carbon linkages and increases the rate of reaction. High pressure keeps

the solvent and products in the liquid phase, prevents coke build-up on the reactor walls and catalyst

surface, and promotes hydrogenation by maintaining a high partial pressure of hydrogen. Catalysts are

normally used to increase the rates of the desirable reactions which include cracking, hydrogenation ,

and removal of oxygen, nitrogen, and sulfur.

Coal consists of complex macromolecules without repeating monomer units that are built primarily of

carbon and hydrogen but also consist of significant amounts of oxygen, nitrogen, and sulfur. The

constituent units tend to be mostly substituted aromatics or hydroaromatics and the degree of

condensation increases as the coal matures. Coal has a hydrogen-to-carbon ratio which is significantly

less that petroleum ; converting coal into liquids requires either the addition of substantial amounts of

hydrogen or the removal of excess carbon.

The coal time line for the 20th century (Schlesinger 1980) helps to trace the major events that have

occurred in our recent history. Referring to Figure kkr-1, direct coal liquefaction technology developed

from work on hydrogenating coal tar by Bergius in Germany in the 1920s (Fischer 1925). A major

improvement was the development of sulfur-resistant hydrogenation catalyst by Pier in BASF's

laboratories in Ludwigshaven, Germany in 1924. The first German brown coal was hydrogenated in

1929. In the 1930s, several plants were built in Germany, and one in Billingham, England. The original

Bergius plants consisted of liquid-phase hydrogenation of coal/recycle slurry followed by vapor-phase

hydrogenation of middle oils (180 - 325°C). Both hydrogenation reactors operated with catalyst and

pressures of 5000-10,000 psig.

German capacity increased to the point that during World War II, eight additional plants were built in

Germany. In 1943, the installed capacity was over 100,000 BSD in 15 plants processing about 50,000

tons of dry coal per day. The US Bureau of Mines tested the German technology after World War II in a

200 barrel per day pilot plant. All of these efforts were technically successful but could not compete

economically with inexpensive petroleum from the Middle East in the early 1950s.

Second generation technology came along in the late 1970s and early 1980s. The pilot plants to develop

these processes included:

1. SRC-II(solvent refined coal) in Tacoma, Washington

2. EDS (Exxon Donor Solvent) in Baytown, Texas and

3. H-Coal in Catlettsburg, Kentucky.

Both the SRC-II and EDS process depended on a donor solvent for hydrogen during the liquefaction and

used the minerals in the coal for a catalyst. EDS did use a catalytic stage to hydrotreat part of the recycle

solvent in a second reactor but no catalyst was used in the coal liquefaction unit. The H-Coal process

was developed by Hydrocarbon Research Inc.(HRI) and was derived from their H-Oil process for

petroleum resid upgrading. The basis of the process was a novel catalytic reactor in which the catalyst

was ebullated in the liquid phase, similar to the more familiar gas-phase fluidized bed processes used in

the petroleum industry. The advantage of this type reactor is that the reactor contents are well mixed

helping to alleviate the excessive heat release associated with coal liquefaction plus the ability to add

and withdraw catalyst from the reactor while it is operating so that the catalyst activity is maintained at a

relatively high and constant level. The U.S. Department of Energy (DOE) and a consortium of industrial

sponsors built a 250 ton per day pilot plant at Catlettsburg, Kentucky which ran from June 1980 to

January 1983. The plant was a technical success and confirmed yields by smaller scale equipment and a

wide variety of mechanical equipment was tested including ebbulation pumps, pressure letdown and

block valves, and hydroclones to concentrate the solids-containing product. Although the process was

not economic, two recent developments promised further cost reductions. The first was a new catalyst

developed by Amoco researchers with a unique bimodal pore size distribution resulting in less catalyst

deactivation, and most importantly higher liquid yields. The second improvement came from Chevron

Research, a two stage liquefaction scheme which lowered hydrogen consumption by tailoring reaction

conditions in each stage to favor the reactions of coal dissolution, hydrogenation of the donor solvent.

thermal/catalytic conversion of the high molecular weight species (residuum) and removal of sulfur,

nitrogen, and oxygen.

Process Overview of Coal Liquefaction

Direct coal liquefaction refers to the direct reaction of coal with hydrogen to form liquids. The hydrogen

reacts with oxygen, sulfur, and nitrogen in coal to remove them as water, hydrogen sulfide, and

ammonia. The hydrogen is also required to substantially increase the H/C ratio before it becomes a

liquid. One of the key differences between coal and petroleum is the much lower H/C atomic ratio of

coal (-.7 vs 1.2 for petroleum). Consequently the conversion of coal to petroleum-like products requires

direct hydrogen addition and this adds considerable expense to the product.

The ideal technology for direct coal liquefaction for gasoline production would:

- produce only C5-400°F products which contain no nitrogen, oxygen, or sulfur

- produce no light hydrocarbon gases, since these gases consume large amounts of

hydrogen

- produce aromatic liquid for octane requirements to reduce hydrogen consumption

- remove oxygen from coal by carbon rejection to carbon oxides rather than

hydrogenation

None of the current coal liquefaction processes accomplish this. Second-generation liquefaction

processes now under development are descendants of the classic Bergius technology. Most involve,

more or less, catalyzed interactions between molecular hydrogen and coal-oil slurries of elevated

pressure and temperature. The new processes (2500 psi, 800-850°F) are different; 1. They operate at

less severe conditions, 2. They place a great deal of emphasis on H transfer from the slurry oil to the

coal/ 3. Careful attention is paid to producing process-derived solvent to recycle to the front end of the

process for slurrying the coal.

The following table lists these "second-generation" processes and how they compare:

Table 1Coal Liquefaction Process Summary

Name Reactor Type Demonstration Unit Capacity, TPD

H-Coal ebulated bed w/recycle Catlettsburg, KY 200

Exxon Donor Solvent plug-flow dissolver +catalytic solventhydrotreater

Baytown, TX 250

SRC I & II plug-flowpreheater/dissolver

Tacoma, WA andWilsonville, AL

506

Two stageliquefaction

plug-flowpreheater/dissolver &ebullated bedhydrotreater*

Wilsonville, AL 6

* Interstage deashing (moderate coupling) or Post-2nd stage deashing (tight coupling)

The Need for Reaction Engineering in Coal Liquefaction

Commercial coal liquefaction plants represent huge investments in capital. For example, the

Breckenridge H-Coal plant was sized to produce 80,000 BSD of liquid products and cost estimates were

in excess of 2.5 billion dollars in the mid 80's. Small errors in equipment sizing or yield translate to

millions of dollars in unnecessary expense. Thus it is extremely important to do the best job possible on

sizing, modeling, and specifying reaction conditions for the coal liquefaction reactor(s).

We need to have a clear understanding of reactors at three stages of development. At the earliest stage,

the laboratory reactors explore new reaction conditions, catalyst formulations, or coal types. The

reaction time should be well defined (avoid long heat-up and cool-down periods) and sufficient mixing to

minimize mass transfer limitations. Reaction kinetics may also be studied at this stage so the rate

models should not have some artifact of the experimental test embedded in one of the parameters.

The second stage of development is typically aimed at mimicking commercial operations by employing

recycle streams to achieve realistic simulations of the integrated process. Isothermal conditions are

usually maintained in the reactor. This is one of the more bothersome aspects of coal liquefaction, since

the coal liquids produced must be fractionated and then a solvent oil sent back to the slurry tank so that

the "process runs on its own juice". The challenge is significant. For example the H-Coal process

demonstration unit (PDU) at Trenton, NJ was a vertical 4" tube with a single bubble cap at the bottom to

introduce liquid feed and hydrogen gas. This was scaled to a 5 ft. diameter H-Coal reactor at

Catlettsburg with over 250 bubble caps and now the added complexity of localized flow variations in the

ebullated catalyst bed.

Finally we move to commercial scale and adiabatic operation. High hydrogen consumption translates to

substantial heat liberated in the liquefaction reaction. Reactor control and stability are extremely

important. Higher-than-desired temperatures lead to cracking reactions and subsequent lowering of

liquid yields. Furthermore coke will accumulate on the inner reactor walls and eventually alter the flow

regime and plug the reactor. We need to design the commercial reactor so that it is sized properly, can

be started-up and shut-down safely, and operated stably at steady-state conditions. It is a formidable

problem for the reaction engineer but if he/she is careful and rigorous, the end product will be success.

COALS AND CONVERSION CHEMISTRY

Coals: Classification, Nomenclature and Structure

The following is an excerpt from Peter Given's (Given 1986) course on coal science at Pennsylvania

State University. His excellent and vibrant description of coal cannot really be improved upon by the

author so it has been left relatively untouched.

"Coal can reasonably be described as an organic rock. As such, it is, like other rocks, a heterogeneous

assembly of more specifically definable components, and the plant debris from which it originated has

undergone a progressive series of alterations as a result of geological or geochemical processes.

Whenever materials are used, it is desirable to be able to relate properties as determined by simple

measurements in the laboratory to behavior in an industrial process; if such a relationship can be

established, one has a basis for selecting the best material for a given process, or for modifying

operating conditions to utilize most effectively the material available. Therefore, as with any other

material, it is important to be able to classify coals and coal components in some manner helpful to

utilization, on the basis of standard laboratory tests.

There are at least seven characteristics of coals that are or may be relevant to their behavior in

conversion processes: rank, geological history, mineral content, trace element distribution, petrographic

composition, chemical structural parameters and pore structure. The coal "series" allows coals to be

classified by the carbon content as:

peat ---6 lignite ---6 sub-bituminous ---6 bituminous ---6anthracite

When we study coals, what we see in nature is an apparently continuous series of alterations from living

plant material through dead material, peat, lignite, sub-bituminous and bituminous coals and anthracite to

graphite. The process of alteration from the peat stage onward is called metamorphism, and the term

"rank" is used in the scientific literature to signify the degree of alteration or metamorphism. Rank is of

such importance in defining the properties of coals that we must discuss it at some length.

Table 2

ASTM System for Classifying Coals By Rank

Class Group Fixed

Carbon,maf

Volatile

Matter,maf

Heating

Value,BTU/lb

Anthracite metaanthracite

anthracite

semianthracite

>98

92-98

86-92

<2

2-8

8-14

Bitumenous low-volatile

med volatile

high volatile A

high volatile B

high volatile C

78-86

69-78

<69

14-22

22-31

>31 >14,000

Subbitumenous subbitumenous A

subbitumenous B

subbitumenous C

10,500-11,500

9,500-10,500

Lignitic lignite A

lignite B

6,300-8,300

<6300

For simplicity, Table 2 makes it appear that coals are distributed along a single development line. In fact

any plot of a coal property against a measure of the degree of metamorphism is a band, rather than a

line. That is, there have evidently been a number of related pathways by which plant debris has been

progressively altered and passed along the coal series. Nevertheless, the degree of alteration is of

overriding importance in determining what set of properties a coal will have. If you are going to use a

coal, its rank is always the first thing, and sometimes even the only thing, that you want to know about it.

Table 3.Approximate Values of Some Coal Properties in Different Rank Ranges

High Vol. Bit. Bituminous Lignite Subbit. C B A Med Vol Low Vol Anthracite

% C (ash free) 65-72 72-76 76-78 78-80 80-87 89 90 93

% H 4.5 5 5.5 5.5 5.5 4.5 3.5 2.5

% O 30 18 13 10 10-4 3-4 3 2

% O as COOH 13-10 5-2 0 0 0 0 0 0

% O as OH 15-10 12-10 9 ? 7-3 1-2 0-1 0

Aromatic C atoms% of total C

50 65 ? ? 75 80-85 85-90 90-95

Av. no. benz.rings/layer

1-2 ? 2-3 2-3 2-3 2-3 5? >25?

Vol. Matter, % 40-50 35-50 35-45 ? 31-40 31-20 20-10 <10

Reflectance, %,Vitrinite

.2-.3 .3-.4 0.5 0.6 0.6-1.0 1.4 1.8 4

Density

Total surface area

Plasticity andcoke formation

Calorific value,maf free, BTU/lb.

7000 10,000 12,000 13,500 14,500 15,000 15,800 15,200

"Rank" is as a qualitative concept. Obviously we need some sort of quantitative parameter to characterize the

stage of alteration reached by a coal in its progression from plant material to anthracite. Since metamorphism of

coal involves systematic changes in so many of its properties and structural characteristics, we should expect that

several parameters would be needed simultaneously to specify adequately the degree of metamorphism. But

there are obvious practical difficulties in following this course...

The A.S.T.M. system used for commercial purposes in the U.S., and the International Systems used in Europe,

depend upon volatile matter to classify the higher rank coals, and calorific values for the lower, which is fair

enough as far as it goes. These properties are used to classify coals into about ten categories, as shown in Figure

KKR-2. People sometimes speak of a coal as being, for example, of high volatile A bituminous rank, as if the

statement uniquely defined the rank. It would be more accurate and less liable to lead to misconceptions if one

spoke of a coal as being in a certain rank range. Each rank category in fact covers a portion of the total rank

range, and the high volatile A category in particular contains coals of considerably different degrees of

metamorphism (e.g. the carbon content may be between about 80 and 87%, and the oxygen content between 11

and 4%).

A further characteristic of coals that may be of practical importance is their geological history. Coals are known in

almost every state of this nation, though in some states the deposits are too small to be of much commercial

significance. There were two principal coal-forming periods, the coals of the central and western states being

much younger than those of the eastern and southeastern states. This means that coals on different sides of the

continent were derived from plants of quite different type, since a great deal of evolution occurred in the 150

million years between the Carboniferous and the Cretaceous. Moreover, for any one geological age, coals were

deposited in a number of distinct basins in different parts of the country, and experienced distinctly different

geological histories.

For these various reasons, coals from the different major deposits may have properties that differ in ways of

considerable importance to the coal user. Therefore, it is of practical significance to use a geological

classification for the nation's coal deposits, as shown in Table 4 below. In some of the Provinces, notably

Provinces 2 and 4, there are several distinct basins of somewhat different histories. With this reservation, one

might expect to find some degree of homogeneity among coals of any one province, but to find differences

between Provinces.

Ranks of Coal

Schobert's (Schobert 1987) description of the various coal ranks follows:

BROWN COAL

Compared to lignite, Brown coal has a very high moisture content, sometimes more than 60% as it is mined, and

it is a younger coal geologically. Deposits of brown coal show distinct layering; the layers are distinguishable by

color. When brown coal dries on exposure to air, large pieces tend to disintegrate, a process called slacking.

When stored in stockpiles or bins, brown coal occasionally catches fire by spontaneous combustion. Brown coal

is easy to ignite but has a low heating value, about 3000 Btu/lb as mined. An enormous deposit of brown coal

occurs in the state of Victoria, Australia. Brown coal also occurs in central and eastern Europe.

LIGNITE

Among the coals of the United States, lignite is the lowest in rank, containing less than 75% carbon on a moisture-

and ash-free basis. The moisture content of lignite is high, but not as high as that of brown coals. The highest

moisture content of lignite mined in the United States is 40-42% (Gascoyne mine in Bowman County, North

Dakota). Lignite is relatively soft and ranges from brown to black. Because lignite has not progressed very far in

maturation, many lignite deposits contain easily recognizable plant remains up to the size of branches or stumps.

The reproductive parts of plants - spores and pollen - are well-preserved in lignite.

Two major deposits of lignite occur in the United States. The Fort Union region spreads over North Dakota,

Montana, Wyoming, and parts of Saskatchewan and Manitoba. This lignite deposit is the largest in the world.

The Gulf Coast lignites stretch from southern Texas, across Louisiana and Arkansas, and into Mississippi and

Alabama.

SUBBITUMINOUS COAL

Subbituminous coal is intermediate in rank between lignite and bituminous coal. Subbituminous coals have

matured to a point at which the woody texture often seen in brown coals or lignites is no longer apparent.

Subbituminous coals are black, having none of the brownish color seen in some lignites. Years ago

subbituminous coal was sometimes called black lignite. Subbituminous coals have the same tendency toward

slacking and spontaneous combustion as lignites and brown coals have. Large deposits of subbituminous coals

occur in the United States in the Rocky Mountain states.

BITUMINOUS COAL

Bituminous coal supplies most of the energy that comes from coal. Until the 1970s the low-rank coals provided

only about 2% of the national coal consumption. Bituminous coal and anthracite made up the rest; bituminous

coal had about a 6:1 margin over anthracite.

Bituminous coal is black but frequently appears to be banded with alternating layers of glossy black and dull black.

It breaks into prismatic blocks. The moisture and volatile matter contents are lower and the heating value is

higher than those of subbituminous coal. Bituminous coal shows little tendency to decrepitate on weathering or

experience spontaneous combustion.

When some varieties of bituminous coal are heated in the absence of air, they soften. Heating also removes

volatile matter as gases, which bubble through the softened mass of coal. The product resolidifies as a shiny,

porous, hard, black solid known as coke. Coke is the fuel used in blast furnaces to make iron.

Bituminous coal occurs in the Appalachian Mountain chain, running from Alabama up into Pennsylvania. A

second major deposit lies in the center of the United States, particularly in Illinois. Bituminous coal is widespread

throughout the world. The principal countries having bituminous coal include Great Britain, Germany, the Soviet

Union, China, and Australia.

ANTHRACITE

Anthracite ranks highest among coals, a position merited for several reasons. Having a very low volatile matter

content, anthracite burns with a hot, clean flame with no smoke or soot. This feature makes it an ideal domestic

fuel. Compared with bituminous coal, anthracite burns more slowly and gives off heat more uniformly. Anthracite

is very low in moisture, about 3%, and is low in sulfur. It is stable in storage and, unlike other coals, can be

handled without dust forming. Its heating value comes close to that of the best bituminous coals.

Anthracite is jet black and usually has a high luster. It is the hardest and most dense coal. In the days when coal

meant only bituminous coal and anthracite, the terms soft coal and hard coal were often used to distinguish the

two. Anthracite was occasionally called stone coal. It was also known as black diamond because of its hardness,

luster, and commercial value.

The many good qualities of anthracite make it a premium fuel. Unfortunately, its geographic distribution limits its

use. In the United States, significant deposits of anthracite occur only in a small region of northeastern

Pennsylvania. Therefore, it is not as readily available on the market as are other coals and thus commands a

premium price.

Coal Analysis

PROXIMATE ANALYSIS

The first approach to examining coal is based on the behavior of coal when it is heated. Heating is one of the

most easily available ways of bringing about changes in substances. All processes that make use of coal involve

heating it. For these reasons, an examination of the effects of heat on coal is a likely starting point for

characterizing the various kinds of coal.

If coal is heated gently at 105°C, it loses weight. The material that is removed at this temperature is water. This

test is the basis for determining the amount of water in coal. Usually, water is referred to as moisture in coal

analysis.

After the moisture has been removed, the temperature can be increased. Because coal burns if heated in air, this

test is done in an inert atmosphere. Heating coal at several hundred degrees Celsius causes another weight loss.

Some of the material that escapes in this experiment is a mixture of gases including carbon dioxide and methane.

The remainder condenses as an oily organic liquid and a tar. Ordinarily, the gases, oil, and tar are not analyzed

further to determine the amounts of pure compounds present in each. Instead, the total amount of gas, oil, and

tar evolution is lumped together and called volatile matter.

The solid material remaining after the volatile matter has been removed is a black char resembling charcoal. If

air is admitted to the apparatus, the black char burns and leaves behind an inorganic solid called ash. The

amount of coal material in the char is called fixed carbon and is calculated by subtracting the percentages of

moisture, volatile matter, and ash from 100.

The set of procedures for determining moisture, volatile matter, and ash and for calculating fixed carbon is known

as the proximate analysis. Use of this term is misleading because proximate is often thought to be short for

approximate. In fact, nothing is approximate about proximate analysis. The procedures for performing a

proximate analysis are very carefully defined and rigorously followed to ensure repeatability from one laboratory

to another. The specific details are prescribed by standards-setting organizations in various countries. In the

United States, the proximate and other analyses are governed by standards established by the American Society

for Testing and Materials (ASTM). The term proximate refers to the practice of lumping together a group of

constituents and reporting those constituents as one generic item, for example, volatile matter.

ULTIMATE ANALYSIS

Although the proximate analysis is helpful in practical applications, it doesn't provide information about the actual

composition of coal. The elemental composition is determined by a second set of analytical procedures, which

collectively are known as the ultimate analysis.

The elements that are normally determined in the ultimate analysis are carbon, hydrogen, sulfur, and nitrogen.

Carbon and hydrogen are measured by burning the coal and collecting the resulting carbon dioxide and water in

chemicals that will absorb them. By weighing the absorbents before and after the analysis, the amounts of carbon

dioxide and water can be determined. The carbon in carbon dioxide and hydrogen in water are known with great

accuracy.

In one of the common methods used to determine sulfur, the sulfur can be converted to sodium sulfate, which in

turn can be converted to barium sulfate. Barium sulfate is highly insoluble in water and therefore can be

collected, dried, and weighed.

Nitrogen is usually determined by converting it to ammonia and absorbing the ammonia in a known amount of

acid. The amount of acid remaining after the ammonia has been absorbed is subtracted from the amount used

initially. The difference represents the acid neutralized by the ammonia and thus tells the amount of ammonia

made in the experiment. Then, the amount of nitrogen present in the coal sample can be calculated.

Relatively simple analytical methods to determine oxygen in coal are not available. The most commonly used

method today involves bombarding the coal with neutrons to convert the oxygen to a very unstable isotop 22 of

nitrogen (nitrogen-16). The nitrogen-16 decays rapidly by emitting beta (ß) and gamma (() rays, which can be

detected and counted. A comparison of the results from a coal sample with those from a pure compound in which

the oxygen content is well-known provides a measure of the oxygen in the coal. Very few laboratories have

access to the appropriate neutron source, that is, a nuclear reactor. Many others do not want the expense, time

delay, or bother involved in sending samples to a reactor facility. Consequently, the amount of oxygen is

determined by adding together the percentages of the other four elements and subtracting this amount from 100.

The disadvantage of this procedure is that all of the experimental errors in the carbon, hydrogen, nitrogen, and

sulfur measurements are propagated into the calculated value for oxygen.

SULFUR IN COAL

When coal is analyzed carefully for both major and trace constituents, almost every element will be found. About

the only exceptions are the highly radioactive elements that are made artificially and some of the noble gases. Of

the roughly 80 elements found in coal, the one besides carbon having the most significant effect on coal use is

sulfur. When coal or coal liquids are are burned, the sulfur oxides formed cause air pollution if allowed to escape

into the environment. The sulfur in coal has caused the spending of millions of dollars on equipment and

processes to capture the sulfur oxides or to reduce the sulfur content of coal before it is burned. Sulfur occurs in

coal in three forms:

1.organic sulfur, which is a part of the molecular composition of the coal itself;

2.pyritic sulfur, which occurs in the mineral pyrite and some related minerals; and

3.sulfate sulfur, which is mostly iron sulfates.

In most coals, sulfate sulfur is a small fraction of the total sulfur content. In low-rank coals, organic sulfur may

contribute half or more of the total sulfur, but in most bituminous coals, the majority of the sulfur is pyritic sulfur.

Inorganic Constituents in Coal

Coal contains a rich variety of inorganic constituents. Mineral grains in coal can sometimes be seen with the

unaided eye and are easily seen through a microscope. When coal is burned it leaves behind the inconsumable

inorganic residue known as ash. The ash is the product of reactions or transformations of the inorganic

components of coal caused by the high temperatures of the combustion process. Several questions concern

these inorganic species: Where did they come from? What are their components? How do they behave?

SOURCES

Several sources contribute inorganic material to coal. The first is the coal-forming plants themselves. Most

plants contain at least small amounts of inorganic substances; that is why burning logs in a fireplace always leave

some ash. However, some plants concentrate significant quantities of inorganic material. For example, the

scouring rushes concentrate enough silica in their cell walls to give them a gritty texture that pioneers found useful

for cleaning cooking utensils or scouring floors. The scouring rushes are examples of the last surviving genus

(the horsetails) of a class of plants that were prevalent in the coal swamps of the Carboniferous period. Some of

the inorganic components of coal were in the original plant material.

During the diagenetic stage of coal formation, as the organic sediments were accumulating in the swamps, ample

opportunities occurred for inorganic debris to be deposited by either water currents or wind. This debris would

settle into the organic sediments and become part of the coal that eventually formed. Clay particles are deposited

into coal in this way. Minerals transported into the coal-forming environment by wind or water are called detrital

minerals.

Although all types of inorganic material in coal are important, either because of their deleterious effects in coal

processing or because of their adverse environmental impact and potential health hazards [76; see also Section

15.1], the relative abundance of different inorganic constituents makes it appropriate to distinguish between major

and minor constituents. The major constituents are

1.clay minerals (aluminosilicates), which occur mostly as illite, kaolinite, montmorillonite, and mixed illite-

montmorillonite and commonly make up as much as 50% of the total mineral matter contents;

2. carbonate minerals, principally calcite (CaCO 3), siderite (FeCO 3), dolomite (FeCO 3@MgCO3), and

ankerite (2CaCO 3@MgCO 3@FeCO 3), but frequently also present as variously composed mixed

carbonates of Ca, Mg, Mn, and Fe;

3. sulfides, which are mainly present as pyrite and marcasite (i.e., dimorphs of FeS x, which only differ in

crystal habit), but which have also been reported in the form of galena (PbS) and sphalerite (ZnS); and

4. silica, which is ubiquitous as quartz and usually accounts for up to 20% of all mineral matter.

Coal Petrology

The following is an excerpt from W. Spachman's (Spachman 1986) course taught at Pennsylvania State

University.

Coal petrology is the science concerned with the nature, origin, and evolution and significance of coal as a rock

material, and coal petrography is the subscience concerned with the description of coal materials and the practical

use of compositional descriptions. Widespread acceptance of "coal" as a class of rock materials has come about

only recently, hence, most "rock" petrologists are not well informed with respect to the composition of coal, and

coal petrology, accordingly, is a young science.

Coal may be usefully characterized at several different organizational levels, including:

1. the elemental level,

2. the molecular or sub-molecular level,

3. the phyterallevel,

4. the maceral-mineral level,

5. the lithotype level, and

6. the lithobody level.

From the standpoint of the coal petrologist, coal consists of two basic classes of materials, i.e. inorganic,

crystalline, minerals and organic, phytogenic, non-crystalline macerals. The macerals form the carbonaceous,

combustible fraction of the coal and by definition they must comprise more than half of the rock mass. Both

macerals and minerals occur in the coal as grains, particles, or fragments, commonly ranging in size from 2 cubic

microns to several cubic centimeters or larger.

The minerals commonly encountered in coal seams are:

Table KKR-6

I. Silicates

2 2 3 2Montmorillonite 4SiO @Al O (1+x)H O4 y 8-y 4 4 4 6 20Illite (OH) K (Si @Al @Fe @Mg @Mg )O

2 2 3 2Kaolinite 2H O@Al O @2SiO2 2 6 4 2 4Muscovite [K ] [Al Si ] [Al ] O (OH)

XII IV VI

8 5 2 3 8Chlorite H (Mg, Fe) Al Si O

II. Oxides

2Quartz SiO2 3Hematite Fe O

III. Sulfides

2Pyrite FeS2Marcasite FeS

IV. Sulfates

4 2Gypsum CaSO @2H O2 6 12 4 4Jarosite K Fe OOH) (SO )

V. Carbonates

3 3 3Ankerite 2CaCO @MgCO @FeCO3Calcite CaCO3Siderite FeCO

THE ROLE OF CATALYSTS

Coal Liquefaction Catalysis

Catalysis plays an important role in coal liquefaction and has led to important technical advances and improved

economics for producing liquid synfuels. A catalyst is defined as a substance which accelerates or changes the

course of a chemical reaction without being consumed itself. The influence of the catalyst on liquefaction rate is

actually not very noticeable since many of the liquefaction reactions take place in the absence of the catalyst. It is

generally believed that some of the minerals (i.e. pyrite) present in coal provide a catalytic function, particularly

for bituminous coals. However, once the coal has been initially liquefied, the catalyst will help to hydrocrack the

coal liquid constituents, remove sulfur, nitrogen, and oxygen as hydrogen sulfide, ammonia, and water and

rehydrogenate the hydrogen donor solvent.

Possible improvements to coal liquefaction catalysts can be summarized below:

Table 8Catalyst Improvements for Coal Liquefaction

Improvement Benefit

Higher activity Smaller reactor

More selectiveLess hydrogen usage

More effective heteroatom removal

More stable Lower catalyst usage

The way in which one makes catalyst improvements is not straight-forward. A heterogeneous catalyst is generally

preferred in coal liquefaction since separation of products from the catalyst is simplified. However, mixing of a

soluble catalyst with the coal slurry provides intimate contacting at the molecular scale. The diagram below

illustrates the catalyst parameters that have to be considered in an effective application:

(INSERT FIGURE KKR-3)

Table 9

Active Portion Nonactive Portion

Type of active components Type of support

Level of active components Support geometry

Stabilizer - Promoters Surface area

Bifunctionality Pore size distribution

Impurities

Reaction Steps in Catalytic Liquefaction

In catalytic coal liquefaction, with a heterogeneous catalyst, we will first list the various steps that occur in the

reaction path. In a slurry reactor, hydrogen-rich gas is bubbled through coal liquid containing suspended catalyst

particles. For simplicity, lets consider the reaction, hydrogenation of the donor solvent to replenish atomic

hydrogen lost (donated to reactive coal fragments to stabilize them). The catalytic hydrogenation reaction is:

According to film theory, the various steps in the overall process are the following:

1.Gas-liquid interface - hydrogen gas dissolves into liquid phase according to Henry's law

2.

Liquid film - dissolved hydrogen is transported to bulk liquid phase through the liquid film next to bubble.

3.

Catalyst film - dissolved hydrogen in bulk phase diffuses from the main body of the liquid to the catalyst surface.

4.

Catalytic reaction with pore diffusion - dissolved hydrogen diffuses into pores of the catalyst and reacts.

4'.Product transport out of pore - hydroaromatic diffuses out of catalyst particle to outside surface.

3'.Catalyst film - hydroaromatic transfers from the catalyst surface to bulk fluid through catalyst film.

2'.

Liquid film - bulk phase hydroaromatic is transported to gas phase through liquid film to gas-liquid interface of

bubble.

1'.

Gas liquid interface - a portion of the hydroaromatic vaporizes into the gas phase according to Henry's law.

Most of the above resistances occur in series but a few are in parallel, such as pore diffusion. The electrical

analog of a catalytic reaction helps to illustrate this resistance concept as shown below:

If the resistances were all in series, we could combine them by straight-forward algebra. However, pore diffusion,

a parallel resistance, complicates the mathematics and prevents a simple combining of terms. If the slowest step

in the process (i.e. rate controlling) is a series step, we can derive a simple rate expression for use in the reactor

performance equation. We will develop these in more detail later.

Nomenclature

Aip - partial pressure of A at gas-liquid interface

AiC - liquid concentration of A at gas-liquid interface

AH - Henry's law distribution constant for A

S - interfacial bubble area

AdN - molar flux of A, moles A/time dt

Alk - liquid film transfer coefficient

AlC - concentration of A in bulk liquid phase

exS - catalyst external area

Afk - catalyst film transfer coefficient for A

AsC - concentration of A at catalyst surface

cdN - molar flux of C, moles C/time df

cfk - catalyst film coefficient for C

csC - concentration of C at catalyst surface

clC - concentration of C in bulk liquid phase

ciC - concentration of C at gas-liquid interface

clk - liquid film transfer coefficient for C

cH - Henry's law distribution constant for C

cip - partial pressure of C at gas-liquid interface

The Porous Structure of Catalysts

Coal liquefaction catalysts are highly porous materials, and typically show some aspects of pore diffusion control.

Molecular transport in pores depends on the size of the pore. We can categorize pores as follows:

1. Bulk flow - Large pores

2. Ordinary diffusion - Large pores

3. Knudsen diffusion - Small pores

The relative balance of the surface reaction rate to the diffusion rate establishes the concentration profile in the

catalyst particle. If diffusion is rapid or surface reactions slow, then the concentration of reactant will not drop

very sharply as we travel down the pore. On the other hand, a slow diffusion rate or fast surface reaction will

produce a sharp concentration gradient in the catalyst particle; then the reaction on the external surface is much

higher than in the particle interior.

Ideal Catalyst

For the moment, consider a single cylindrical pore in a catalyst. The

differential mass balance for a first order reaction down the pore (Levenspiel 1972) is:

(INSERT FIGURE KKR-6)

Solution of the above differential equation leads to effectiveness factor plots like the following

sThe Thiele modulus, N, contains the intrinsic kinetic constant, k , that is frequently unknown. To get around this

problem, a new dimensionless modulus, M, is sometimes defined as:

M = R observed rate2

eff s D C volume

This produces the following modified effectiveness factor plot:

Real Catalysts

The mathematical development for the effectiveness factor, 0, as a function of the Thiele modulus is based on a

single size catalyst pore. We need to remind ourselves that real catalysts have a distribution of pores of many

different sizes. When catalyst extrudates are formed, the powder is compressed into a single large pellet and can

be visualized as:

The table below lists the properties of some catalysts developed by Amoco and W. R. Grace under an EPRI

contract.(Kim 1979) The "bimodal" pore size distribution is characterized by the inclusion of macropores greater

than 1000 Å in diameter.

This inclusion of these larger transport pores is important for activity maintenance and a higher catalyst

effectiveness factor. The mean diameter of the smaller pores (APD) should be around 120 Å to accommodate

the large coal-derived molecules. When the larger pores are "squeezed" out of the catalyst particle during

extrusion, or simply not built in, then the pore size distribution is called unimodal with a single hump in the pore

size frequency plot.

Table 10Catalyst Inspection Data

Composition, Wt% APD SA PV PV of 10 -10 Å ABD3 5

CoO MoOCatalyst

3

Å m /g cc/g % of total g/cc2

Bimodal Catalysts

HDS-1442A 3.1 13.2 58 323 .64 26.1 .567

CoMo-G120B

3.2 16.2 122 154 .60 11.1 .665

Mo-G120B --- 14.9 123 167 .65 9.6 .656

Unimodal Catalysts

CoMo-G120U

3.0 15.5 115 162 .62 4.7 .733

0.5 CoMo-G120U

0.5 16.0 121 160 .64 3.2 .723

Two of the bimodal catalysts are plotted in the frequency plot below along with one unimodal Amocat . AlthoughTM

the bimodal American Cyanamid HDS-1442A has a significant amount of macropore volume, the APD of the

smaller mesopores is too small for hydroliquefaction of coal. When exposed to a coal liquefaction environment,

the smaller pores rapidly fill with coke causing the pore mouth to neck down and eventually plug.

Not only does the catalyst effectiveness factor drop but ultimately the interior of the catalyst particle is isolated

from the exterior by the coke-filled pores. This will be discussed more in a later section on catalyst deactivation.

Which Rate Equations to Use

The rate equations for catalytic hydroliquefaction can take many forms depending which step or steps are slowest

and rate controlling. Referring back to the mass transport steps (Figure 2) the rate equations for these various

situations are listed below:

External Mass Transport

- Hydrogen Transfer Controls at Bubble Interface

When the transport of hydrogen gas from bubbles to the bulk liquid is the slowest step the rate

equation is:

iThe interfacial area of the gas bubbles, a , is a critical parameter in this case. If the gas bubbles are large, we can

increase reaction rate by better dispersion of the gas into the liquid phase, to produce smaller bubbles (increasing

ia ). Obviously there is a practical limit, because an unstable foam will eventually develop as bubble size is

decreased.

- Film Resistance Effects at Catalyst Surface

When the catalyst is functioning well and the catalytic rate fast, then external mass transfer may be the rate

limiting step. As soon as the reactants reach the external catalyst surface, they react. This causes the film

around the catalyst particle to limit transport of fresh reactant to the surface and, in turn, to control the overall

reaction rate.

Catalytic Reaction Controls

Pore Diffusion Limitations

We defined the effectiveness factor earlier as the ratio of the diffusion affected rate to the non-diffusion

rate or

Thus to account for pore diffusion limitations, the effectiveness factor is included in the rate equation as a

multiplier of the catalytic rate constant. If we begin to include other transport effects, the effectiveness factor will

still be associated with the catalytic rate constant. Referring back to the case where film resistance effects were

present at the catalyst surface we would use, the effectiveness factor as follows:

Estimation of the effectiveness factor was described earlier in the section on porous structure of catalysts. We

restate the key aspects.

Research Data for Reactor Design

Fulton (Fulton 1986) has pointed out that reactor design for a heterogeneous catalyst involves two major phases.

First, the researcher collects experimental reaction-rate data that are free of heat- and mass- transfer effects (i.e.,

intrinsic kinetic data). Second, the design engineer combines the intrinsic rate with the specific-heat and mass-

transfer effects appropriate to the plant-sized reactor to be designed.

In the second phase, the design of a commercial liquefaction reactor, the design engineer must take into account

the variety of heat-transfer, mass-transfer and reaction phenomena that occur.

In the first phase, the generation of intrinsic reaction-rate data, the researcher must:

1. Build and operate the reactor to minimize axial and radial transport (intrareactor) effects; e.g., isothermal

operation is highly desirable. We should define time "zero" (no long heat-up) in batch reactor test by

using a charging bomb, to inject the reactants into the reactor after heat-up.

2. Operate the reactor at a mixing rate large enough to remove heat- and mass-transfer gradients between

the catalyst particle and the bulk fluid (particle-fluid or interphase).

3. Use catalyst particles small enough to eliminate the influence of intraparticle heat and mass gradients.

With these precautions taken, the observed reaction rate will most likely be the intrinsic rate -- i.e., the true

catalytic reaction rate. As a result, the researcher needs methods for determining whether transport effects are

influencing the laboratory data. The "limiting criteria" allow the researcher to calculate or experimentally

determine whether these transport effects are important. They, in turn, guide the researcher to reactor operating

conditions that will provide data free from transport effects.

@ Mixing Intensity to Eliminate Mass Transfer Effects

@ Reduce catalyst particle size to reduce pore diffusion effects

Key rate parameters are modified by the influence of different transport regimes. For example, if diffusion is

strongly influencing the overall reaction rate, the energy of activation determined experimentally will be one-half

the intrinsic reaction energy of activation. As a result, the Arrhenius correlation for reaction rate would be grossly

in error.

The kinetic parameters (order and activation energy) are altered as the data-collection basis moves from the

chemical-reaction, and through the intrapellet and interphase, controlling domains. The influence of pellet size

and fluid velocity on rate becomes greater as bulk mass transfer becomes more important. Displayed below is

the characteristic behavior of kinetic parameters appropriate to each controlling regime.

Table 11

A A A-r = kC = A exp[-)E/RT}Cn n

DomainActivationEnergy Order

Pellet Size

FluidVelocity

ChemicalReaction )E n Independent

Independent

Diffusionreaction )E/2 (n+1)/2 1/L

Independent

Bulk masstransfer 5 kcal/mole 1st (1/L)3/2

v1/2

Catalyst Deactivation

In general, catalyst deactivation occurs by three mechanisms, 1.) sintering of the supported metals on the surface,

2.) poisoning by compounds which strongly and irreversibly adsorb on to the active sites and 3.) fouling, where

coke forms on the surface of the catalyst which limits access to the active metals and also reduces the diameter

of the pores, eventually leading to pore mouth plugging.

Coking, sintering and ash deposition all contribute to catalyst deactivation in coal liquefaction, the degree

depending upon the coal type and processing conditions. Used catalyst characterizations are described below for

work in pilot plant studies at Amoco. (Kim 1979)

Effect of Coking

Table 12 presents the loss on ignition (LOI) for various used catalysts. Irrespective of catalyst type and on-stream

hours, the loss on ignition (LOI) is almost always 20 wt% based on spent catalyst weight. Elemental analyses

show that a majority of this deposition is undoubtedly "coke" with the atomic H/C ratio ranging between 0.78 and

1.0. The second largest constituent is sulfur which amounts to about 5 wt%. The H-Coal catalyst in the table was

obtained from the bench scale H-Coal bench unit of HRI after 600 hours on stream operating in the syncrude

mode followed by testing in the Amoco continuous pilot plant for 50 additional hours; a somewhat higher LOI

(26%) is observed. The two unpromoted molybdenum catalysts (at the bottom) are no different from the rest of

the catalysts as far as LOI.

Table 12IGNITION LOSSES OF USED CATALYSTS

Amoco Pilot Plant Tests (Kim 1979)

Type Run No. CatalystHours on Slurry LOI, Wt% S, Wt%

5140 CoMo-G200B 136 21 4.5

Cobalt-molybdena 5149 CoMo-G120B 166 20 -

5138 CoMo-CT100 200 21 -

5135 HDS-1442A 190 21 -

5147 H-Coal (H-CoalUnit-Syncrude

650 26 -

5142 NiW-G120B 48 21 5.5

Nickel-molybdena orNickel-tungsten

5141 NiMo-G120B 137 24 5.1

5146 NiMo-G120B/Si 147 20 4.1

5145 NiMoRe-G120B 160 22 5.1

Miscellaneous 5144 SnMo-G120B 112 20 4.9

5143 Mo-CT100 95 22 4.5

Unpromoted molybdena 5148 Mo-G120B 140 20 3.7

The electron microprobe analysis of carbon profile across the catalyst pellets is shown in Figure KKR-13.

Both HDS-1442A and the Amoco CoMo catalyst show very high carbon concentrations at the outer

perimeter with some lower level of carbon across the entire cross-section, indicating pore mouth

plugging. An aged catalyst from the pilot plant was ground and retested in a batch unit; performance was

almost equivalent to the fresh catalyst. Other observations from the microprobe analyses are that HDS-

1442A shows stronger and more sparse carbon intensities in the catalyst interior, compared to the more

uniform distribution for CoMo on Grace 120 D catalyst.

Contaminants on Used and Regenerated Catalysts

The used as well as regenerated (calcination in air) catalysts were analyzed for metals contaminants by

emission spectroscopy. The results on calcium, iron and titanium are presented in Table KKR-13 on a

coke free basis. Titanium is clearly the major contaminant followed by, to a less extent, iron, silicon,

sodium, magnesium and calcium. The trace metals deposited on the catalyst are primarily from the coal.

The slurry oil typically contained less than one ppm titanium, 24 ppm Fe, 9 ppm Na, 5 ppm Mg, 2 ppm

Si, and 1 ppm Ca.

It appears that the titanium on spent catalyst is present in the carbonaceous deposit as well as the

catalyst support. When compared to the regenerated catalyst, about 60-70% titanium measured on the

spent catalyst is retained on the regenerated catalyst. It is not clear how the titanium leaves during the

regeneration. Calcium on the H-Coal syncrude catalyst is unusually high, perhaps due to the difference

in H-Coal slurry oil and a longer time on stream.

Table KKR-13TRACE METALS ON USED AND REGENERATED CATALYSTS

Amoco Pilot Plant Tests (Kim 1979)

Catalyst Hours

on StreamWt%, Coke-free BasisCalcium Iron Titanium

H-Coal (Syncrude)UsedRegen

650-

0.940.29

0.80.7

2.61.8

CoMo-G 120BUsedRegen

130-

0.140.12

0.60.4

2.31.4

NiMo-G120B/Si

Used 147 0.49 1.0 3.7

Regen - 0.24 0.9 2.4

Mo-G 120B

Used 140 0.13 0.8 3.0

Regen - 0.15 0.4 2.3

Mo-CT 100

Used 95 0.26 0.6 3.8

Regen - 0.15 0.4 2.3

Pore Size Distribution of Used Catalyst

Pore size distributions of fresh, used and regenerated catalysts were compared to examine patterns of

pore plugging. Figure KKR-14 shows pore size distributions obtained by mercury porosimetry for fresh

HDS-1442A and Amocat-1A, CoMo on 120 D alumina, tested by HRI. Pore volume is plotted versus

pore diameter and both catalysts clearly show bimodal pore distributions. HDS-1442A, which has

smaller pores than the Amoco CoMo catalyst, has a macropore (>1000 D diameter) volume about 27%

of the total, whereas the macropore volume of the Amoco catalyst is somewhat lower (18% of the total

pore volume). Based on performance data, the Amocat-1A catalyst is believed to have a more favorable

pore distribution, due to the larger mesopores around 125 D diameter.

The pore size distribution of the used HDS-1442A catalyst which had 190 hours is compared with the

fresh catalyst in Figure 3-17. The pore volume of the used catalyst is shown on a coke free basis in

order to make a one-to-one comparison with the fresh catalyst. The macropores on the right are virtually

unaltered after use. The macropores probably serve as feeder pores which transport materials into the

catalyst interior. On the other hand, the small pores in the range of 40-100 Å diameter, which contains

most of the surface area, have been greatly reduced. The reduction in pore volume and correspondingly

surface area is caused by coking and/or sintering. When regenerated, these pores are only moderately

restored. Small pores are usually more susceptible to sintering.

The pore size distribution of fresh and used Amocat-1A (CoMo on Grace 120 alumina) are shown in

Figure 3-18. This particular sample has been aged for 170 hours in the continuous pilot plant. Like the

HDS-1442A catalyst, the macropores of Amocat-1A are unaltered. The pore size distribution for the

used catalyst implies a whole new range of small pores had been created. However, this is due to partial

blockage of the pores by the coke causing a downward shift in the size distribution. Regeneration

restored most of these active pores in the 40-200 D diameter range. The net effect was that only about

20% of the fresh pores are destroyed during the course of liquefaction. The used catalyst retains more

than one half of its micropore volume, based on the regenerated pore volume.

Deactivation Models

In order to translate the differences in benzene soluble conversion to catalyst activity, a simple kinetic

scheme was proposed, which consists of two first order parallel reactions; one path represents coal going

to asphaltols, defined as THF soluble-benzene insoluble material, and then proceeds to benzene soluble

product; the other path represents coal going directly to benzene soluble material.

The above reaction scheme is not intended as a process model but serves as an estimation of catalyst

activity consistent with observations. Two assumptions were made; the first assumption deals with which

one of the two reaction paths is likely to be affected the most by a catalyst. Step 3 is assumed to be the

catalyzed reaction step. Since THF soluble conversions are the same for both thermal and catalytic

2liquefaction of Illinois No. 6 coal, Step 2 should be primarily thermal allowing k to be estimated from

thermal data. Thermal reaction also results in some benzene soluble product directly from coal as

shown in Step 1. We assume that a major function of catalyst is to convert asphaltols to benzene

soluble material. Of course, the mineral matter of coal has some small catalytic action on preasphaltene

conversion, but benzene solubles thus produced may be regarded as if derived directly from the thermal

reaction of coal. The second assumption is that the stirred autoclave reactor approximates perfect

mixing. Then, the following simple algebraic expressions for the rate constants can be derived.

Using the data obtained with HDS-1442A catalyst at 427°C and 48 minutes residence time, the values of

1 2 3 3k , k and k are calculated as 10.4, 6.3 and 7.0 hour , respectively. The rate constant k is a measure of-1

catalyst activity for liquefaction conversion. Since the space velocity remained constant in the aging

3test, the values of k /LHSV are plotted in Figure 3-11 for the unpromoted molybdenum and two other

CoMo catalysts.

The plot suggests first order deactivation since activity declines with time in a logarithmic fashion. The

unpromoted molybdenum catalyst is no more active than CoMo version initially, but it clearly maintains

activity better. For instance, at 100 hours on stream, the molybdenum catalyst shows a liquefaction

activity 40% higher than CoMo on the same alumina support and 160% higher than HDS-1442A.

Studies on spent catalysts from the Wilsonville pilot plant (Stohl 1985) give even more insight on catalyst

deactivation. It should be noted that the catalyst was in the second of two stages and not directly

exposed to the initial coal liquefaction environment, in the first stage dissolver.

The catalyst used in all three runs was Shell 324M with 12.4 wt % Mo, 2.8 wt % Ni and 2.7 wt % P on an

alumina support. Catalyst samples were withdrawn periodically throughout each run with only minor

(5%) catalyst additions to the hydrotreater. All samples were Soxhlet extracted with tetrahydrofuran and

dried under vacuum at 100°C prior to characterization and activity testing. A portion of each extracted

sample was regenerated by slowly heating to 500°C in a muffle furnace in air. Regeneration removed

the carbonaceous deposits but did not affect the contaminant metals. Studies of both aged and

regenerated catalysts enable separation of the effects of both carbonaceous deposits and contaminant

metals. The types of processing options practiced at Wilsonville include:

Run 242 ITSL integrated two stage liquefaction (interstage deashing)

Run 246 DITSL doubly integrated two stage liquefaction (two recycle streams)

Run 247 RITSL reconfigured integrated two stage liquefaction (post 2nd stage deashing)

Characterization of the catalysts included studies of the buildup of carbonaceous deposits and

contaminant metals as well as changes in physical properties. Quantitative carbon analyses show that

the carbon buildup is very rapid in samples from all three runs. Catalysts from Wilsonville run 246 attain

a carbon content of 7.3 wt % during DITSL processing which then increases to 9.0 wt % during ITSL

processing. Catalysts from Willsonville run 242 (ITSL processing) attain a carbon content of 9.3 wt %

after about 133 lb resid/lb catalyst which then remains constant throughout the remainder of he run.

Samples from Wilsonville run 247 with RITSL processing attain the highest carbon content of 10.8 wt %,

which decreases twice during the run to values of -10 wt % after 100 lb resid/lb catalyst and to 8.9 wt %

after 300 lb resid/lb catalyst. These two decreases in carbon are correlated with temperature increases

in the reactor from 680 to 700°F and from 700 to 710°F, respectively. The carbon contents in catalyst

samples from these three runs are correlated with the quantity of heavy resid and unconverted coal in

the hydrotreater feed. DITSL processing has the lightest feed whereas RITSL processing has the

heaviest feed because deashing, (which normally removes some unconverted coal and heavy resid) was

performed after the 2nd stage catalytic hydrotreater.

Many contaminant metals are deposited on the catalyst during coal liquefaction processing. However,

Fe and Ti are the most abundant so only data obtained on these two metals is shown. Iron contents,

determined by atomic absorption, as a function of catalyst age are given in Figure 2. It is interesting that

the catalyst exposed to high-ash feed has the lowest Fe concentration. This suggests that the high Fe

content in run 242 catalyst is not derived from the Illinois No. 6 coal, which was also used in run 247, but

from upstream equipment perhaps related to the deasher.

The lowest Ti content was observed on the run 246 catalysts which processed subbituminous coal. This

result is in agreement with previous studies on catalysts from single stage direct coal liquefaction

processes which showed that Wyodak coal resulted in lower Ti buildup than Illinois No. 6 coal. Since the

Wyodak coal and Illinois No. 6 coal have similar Ti contents, the difference must be due to the presence

of Ti phases in the subbituminous coal that are less reactive with the catalyst than those of the

bituminous coal. Run 247 catalysts with an age of 546 lb resid/lb catalyst show about a 30% higher Ti

content than run 242 catalysts with a similar age. This suggests that deashing is only partially effective

in removing the Ti phase which reacts with the catalyst.

Electron microprobe analyses of samples from each run with ages between 500 and 600 lb resid/lb

catalyst were used to determine the distributions of the contaminant metals. These distributions vary

widely in regard to maximum contaminant metal content but show less than 25% variation in penetration

depth. Figures 4 and 5 show typical Fe and Ti distributions in cross sections of the catalyst extrudates

that were about halfway down the length. The run 242 sample with the highest Fe content shows a

penetration depth of -50% of the pellet radius. The lowest Fe penetration is observed on the run 247

sample and the lowest Ti penetration is on the run 246 sample. Both run 242 and run 247 samples show

Ti penetration depths on the order of 50% of the pellet radius.

An equation has been derived which relates extrudate activity to the physical properties of the catalysts(2)

and the fraction of intrinsic activity lost due to deactivation by carbonaceous deposits and contaminant

metals.

At the start of coal processing, the largest activity lost is due to carbonaceous deposits. However, losses

due to contaminant metals also become significant and cause a continuous decline in activity after the

effects of the carbonaceous deposits have levelled off. The most significant difference between runs

242 and 246 is due to the lower deactivation caused by contaminant metals in run 246. The overall

deactivation rates, however, are similar in these two runs. Run 247 has a deactivation rate due to

contaminant metals between those observed for runs 246 and 242. However, the loss of extrudate

activity early in run 247 which is due to carbonaceous deposits is significantly greater than observed in

either run 242 or 246. This high deactivation is related to the very low effective diffusivity of this sample.

Catalyst Design for Coal Liquefaction

The design of a coal liquefaction catalyst is an extremely challenging problem. Before embarking into

the discussion, it is helpful to reexamine the triangular diagram for key catalyst properties, in Figure 3,

that must be balanced.

Due to the large heat release associated with coal liquefaction, the reactor contents should be mixed

sufficiently so that temperature gradients are maintained at some acceptable level. This usually means,

we use a reactor (typified by the ebullated H-Coal type) with a recycle pump to circulate the exit steam

back to the reactor inlet.

This sets the physical constraint on the catalyst, its geometry, density, and abrasion resistance. The

catalyst particles are suspended in the upwardly flowing coal liquid. The particles have two key forces

acting on them, the buoyant force of the coal liquid they displace and the fractional drag from the flowing

liquid. Typically, extrudates about 1/8" in diameter are used in the H-Coal reactor. In contrast, the H-Oil

reactor (for petroleum resid upgrading) can use the smaller 1/16" particles with little trouble since the

liquid is less dense and provides less buoyant force for a given volumetric displacement. The minimum

fluidization velocity (Kuni 1969) can be estimated as:

The second aspect is activity of the catalyst. This implies a highly porous support which provide plenty

of active surface. We select the particular combination of catalytic metals based on which of the

reactions we need to promote.

Levy and Cusumano (Levy 1977) have noted, "Before a material is tested for catalytic coal liquefaction,

2its chances of survival in the liquefaction environment should be examined. The presence of H S poses

the most severe problem. A large number of compounds that may ordinarily be considered promising

candidates sulfide in this environment. It is therefore fruitless to spend considerable effort in the testing

of these materials. Compounds that are expected to resist sulfidation include a number of oxides,

nitrides, borides and silicides."

For the tried and true hydrotreating catalysts, we include the following table for reference.

Table 14Examples of Typical Sulphide Catalysts Practically Used in the Industry.

(Weisser et al. 1973

Catalyst type Catalystnotation

Catalystcomposition

in the fresh(nonworking) state

Origin ofthe catalyst

Application Reference

For destructive hydrogenationof the heaviest fuels andpetroleum residues

4 2 310927 FeSO + Na CO ondust from Winklergegenerator (5%Fe)

I.G.Farbenindusstrie

Destructive hydrogenation ofthe heaviest fuels (I.G. F.,Varga-process

[317] [1107] [1172]

3- CoO + MoO onaluminosilicate(?)

Gulf Research Dev.Co.

Hydrocracking of residualpetroleum fractions in theGulf-HDS process

[155] [2137]

3 2 3- CoO + MoO on Al O HydrocarbonResearch

Hydrocracking of residualpetroleum fractions in the H-Oil process

[1618] [1621] [2137] [2148]

For hydrocrack- ing tar andpe-troleum distil-lates

26434 10%WS on clayactivated with HF

I.G.F. Hydrocracking of tardistillates (vapor phase)

[1101]

- NiS onaluminosilicate

CaliforniaResearch Corp.

Hydrocracking of mediumpetroleum distillates by theIsocracking process

[2148]

Refining and destructive-refining catalysts

33510 53.5% MoO + 30%ZnO + 16.5% MgO

BASF Hydrotreating and hydrocacking [793] [1101]

25058 WS I.G.F. Hydrotreating and hydrocrackingof medium oils

[793] [1101]

39062 7.5% MoO + 7.5%3 2 3WO + 35% Al O +

250% SiO

VEB "Leuna Werke" Hydrotreating and hydrocrackingof medium oils

[1429]

23076 NiS:WS = 2:1(unsupported)

I.G.F. Hydrogenation catalyst of highefficiency (hydrogenation ofolefins)

[317]

2- NiS-WS (40% W and25% Ni,nonsupported)

Shell Oil Co. Unsupported desulphurationcatalyst

[1743]

28376 27% WS + 3% NiS2 3on (-Al O

I.G.F. Hydrotreating of tar-oils invapor-phase, MTH-process

[1101]

38376 ox. 26% WO + 4.5% NiO2 3+ 69.5% Al O

VEB "Leuna Werke" Hydotreating of tars andpetroleum distillates, MTH-process etc.

[1101] [1429]

8376 ANTIAS Type 8376 withincreased Nicontent

Chemical WorksZáluží (�SSR)

Hydrotreating of arsenic-containing tar

[2015]

37846 10% MoO + 3% NiO2 3on Al O

I.G.F. resp. BASF Hydrotreating of medium oils [1429]

Refining and destructive-refining catalysts

38197 17% MoO + 4.5%2 3NiO + 78.5% Al O

VEB "Leuna Werke" TTH-process {1429}

3 3U 63 12% WO + 8% MoO+ 4.5% NiO + 75.5%

2 3Al O

VEB "Leuna Werke" Hydrotreating of tar andpetroleum distillates

[1429]

BASF 0852Nalco 471Nalco 810Aero HDS-2OroniteGildler G-35BHoudry 200-AComoxP.SpenceCyanamid HDS1Co-Mo (USSR)7362 (�SSR)

For composition,see Table 78

Different Americanand EuropeanCompanies

Hydrorefining anddesulphuration catalysts fordifferent purposes

[1743]

BR 86 2.3% CoO + 0.8%3NiO + 17% MoO +

2 380% Al O

VEB "Leuna Werke" Improved refining Co-Mocatalyst

[1429]

- 1.0% Co + 0.5% Ni2 3+ 8.3% Mo on Al O

Davison Chem. Co. Improved hydrorefining Co-Mocatalyst

[1743]

2Dehydrogenation catalysts IG 5615 NiS:WS -1:2(nonsupported)

I.G.F. Dehydrogenation of cyclanes toaromates

[1347]

2- NiS:WS -2:1 Shell Dev. Co. Dehydrogenation of cyclohexaneand methlycyclohexane toaromatic substances

[1049]

Finally we must consider catalyst stability or how well it survives the coal liquefaction environment. Pore

structure, specifically a bimodal pore size distribution of macro (> 1000 Å diam) and meso (20 Å < d <

600 Å) pore is the key to acceptable activity maintenance. Coking of the catalyst causes a rapid initial

decline in the activity. Then a more gradual deactivation occurs due to sintering and additional coking.

The inclusion of macropore volume is accomplished in the catalyst forming step such as extrusion. The

drawback of the macropores is that the catalyst is "softer" and can sometimes grind up when it is in the

reactor being ebullated. Thus, there is a practical upper limit (25% of total) of macropore volume that

one aims to stay under.

In summary, design of a coal liquefaction catalyst represents a balance of properties. The type of

reactor, coal type, reaction severity, and desired product quality will all enter into the design of the

catalyst. As a base point, the properties of the Amocat™ family of catalysts is given below for use in

catalytic hydroliquefaction of coal. They have enjoyed considerable success in Wilsonville, AL and the

H-Coal pilot plant at Cattettsburg, KY.

Table 15

Catalyst Name Amocat™ 1A Amocat™ 1B Amocat™ 1C

@ Composition, wt%

3MoO 16±0.9 16±0.9 16±0.9

CoO 3±0.4

NiO 3±0.4

CaO #0.1 #0.1 #0.1

4SO #0.5 #0.5 #0.5

2 3Al O balance balance balance

@ Surface Properties

BET Surface Area, M /g 170 min ))))))))))))))))>2

Pore Volume, cc/gm (<1200 Å diam) 0.65 min ))))))))))))))))>

Total pore volume, cc/gm 70.82 ))))))))))))))))>

Pore size distribution bimodal ))))))))))))))))>

Total pore volume in macropores (>1200 Ådiam), %

18±5 ))))))))))))))))>

Pore Volume in micropores <40 Å diam) <2 ))))))))))))))))>

@ Physical Properties

Shape extrudate ))))))))))))))))>

Size, in. 1/16-1/10 ))))))))))))))))>

Dry Attrition, 24 hr, 30 mesh, wt% <3 ))))))))))))))))>

Bulk density, compacted, lb/ft 37±3 ))))))))))))))))>2

Fines content, 16 mesh, wt% <1 ))))))))))))))))>

@ Suggested Use

Liquefaction X X X

Desulfurization X

Denitrogenation X

Aromatic saturation X X

LIQUEFACTION REACTOR TYPES/MODELS

Ideal Reactors - A Brief Review

Many types of reactors enter into the field of coal liquefaction. The researcher may be using a simple

tubing bomb reactor which is operated batch wise. In another situation, coal liquids may be pumped over

a fixed bed of catalyst in a plug flow trickle bed reactor. The continuous pilot plant or commercial

liquefaction reactor will frequently be well mixed, by either mechanical agitation or high recycle; this

reactor type is the continuous stired tank reactor (CSTR). The reactor performance equations for these

three ideal reactors are given below:

Reactor Type Batch Plug Flow CSTR

Shematic

Rate Equation

Reactor Equation

Coal liquefaction is a complex reaction process comprised of both homogenous and catalyzed

heterogeneous reactions. Thus, we must use a reactor performance equation which combine accounts

for both thermal and catalytic reactions for the depletion or formation of a reaction species.

There are four primary tasks that involve an understanding of liquefaction reactors and include:

1. Reaction model development

2. Evaluation of catalyst performance

3. Evaluation of coal liquefaction behavior

4. Scale-up from one level to the next size up (i.e. commercial)

We will discuss each at these situations in turn and give some examples.

Reaction Model Development

The development of a reaction model to describe coal liquefaction is a common situation that a research

engineer encounters. The data are generated from small scale tests with a batch or continuous reactor,

operated isothermally. It is obviously important to measure the intrinsic kinetics and not to have the data

disguised by mass transfer limitations.

A micro-batch reactor (-5 ml) such as the tubing bomb reactor is a common, inexpensive device to

develop the data. The coal, donor solvent, and optionally, catalyst are changed into the small reactor,

sealed and then pressured with hydrogen gas to about 500 psi. To start the reaction, the tubing bomb is

immersed in a heated fluidized sand bath for a specified length of time with agitation. Shortly after

immersion in the heated sand bath, the reactor pressure is increased to a final level, close to commercial

conditions. To stop the reaction, the micro-reactor is pulled out of the heated bath and rapidly quenched

in a cooling fluid.

A more expensive, but superior, batch reactor is a stirred high pressure autoclave (-300 cc). The

internal mixing by the spinning turbine blade is far better than the gentle up-and-down agitation of the

micro-batch unit. The dispersion of small gas bubbles into the liquid phase enhances mass transfer and

eliminates stagnant zones, where coking might occur. The problem with the large size reactor is in

defining time zero, when the reaction starts. The reactants cannot all be changed into the cold reactor

and then heated up without some of the reaction taking place during heat-up. A solution to this is to put

only slurry oil (2/3 of total) in the stirred autoclave and slowly heat to slightly over the desired reaction

temperature. When the desired temperature is reached, the coal and remaining slurry is rapidly charged

into the top of the reactor with a pressured bomb fitted with ball valves on the top and bottom. The

liquefaction reaction is the allowed to proceed and then thermally quenched at the desired time by

circulating cooling fluid (water or oil) through internal cooling coils.

For either of these batch reactors, the rate equation and reactor performance equations are:

The data are in the integral form for a batch reactor and do not lend themselves well to model

discrimination or parameter estimation. A technique developed by Churchill (Churchill 1974) translates

integral data to differential rate data where it can be manipulated conveniently.

The difference table above illustrates how the integral data on the left graph are translated to rate data

on the right graph. A smooth curve drawn through the rectangular tops (so that area in triangles above

and below are equal) gives the instantaneous rate at time, t.

The continuous stirred tank reactor (CSTR) is a frequently a better tool to obtain kinetic data if you can

afford the time and expense of setting one up. The reaction rate can be derived directly from

experimental data as pointed out by Mahoney, Robinson, et al (Mahoney 1978). The reactor has no

concentration or temperature gradients and conversion is controlled by changes in slurry feed rate or

reactor temperature. An overflow tube controls the level and an annular catalyst basket is fitted near the

wall. The reactor schematic for the multiphase CSTR is given below with more details in Autoclave

Engineers Bulletin 1200.

The reactor performance equation for the CSTR is:

A AThus the reaction rate, -r or -r ', can be directly calculated from conversion data and process conditions

and then used to develop a kinetic rate expression

Two familiar test reactors are the Robinson-Mahoney (stationary catalyst basket) and Mahoney-Robinson

(spinning catalyst basket). In Figure 28, these two reactors are shown.

Evaluation of Catalysts or Coals

The next two tasks that involve liquefaction reactors are catalyst performance and coal liquefaction

behavior. Compared to kinetic modeling, the experimental constraints are not as demanding. It is not as

critical to define time zero, if the reactor can be heated up in the same way each time. Usually coals are

run at the same severity and a relative comparison made on the conversion level, and product quality.

Catalyst evaluation/ranking, however, is best done when compared at constant conversion level. For

continuous reactions, the relative catalyst activity is defined as the ratio of space velocities as shown

below. If it is not possible to keep conversion constant, then the constant reaction severity mode will

certainly suffice.

Coal reactivity is frequently compared with a micro-batch reactor. This reactor was described earlier and

the important aspect is to keep temperature- time history constant. Researchers at HRI (lb) developed a

scheme to translate coal conversion data at various reactions times and temperature a common basis.

The reaction severity parameter they developed is called to Standard Temperature-Time Unit or STTU.

It is defined as:

Liquefaction Catalyst performance should almost always be in a continuous flow reactor. Catalyst life is

a critical factor in coal liquefaction and fresh catalyst activity is only important to establish where the

catalyst starts out. Coke deposits rapidly form on the catalyst the first few days that it is on-stream and

then a more gradual increase occurs. Catalyst is usually evaluated for at least 10 days and preferably

around 30. A CSTR high pressure autoclave is preferred and these are described by Autoclave

Engineers (Autoclave Bulletin 1200). A typical plot at constant reaction severity is shown below:

Commercial/Pilot Plant Reactor Design and Scale-Up

The design of larger liquefaction reactors provides a significant challenge since heat release is

substantial and the flow regime deviates from the ideals of plug flow and perfect mixing. We will

examine the commercial reactors in increasing order of complexity.

Exxon Donor Solvent (EDS) Reactor

The liquefaction reactor used in the EDS process is a network of tubes with no catalyst inside. Coal

slurry and hydrogen gas are introduced in the inlet with a nominal residence time of 30-45 minutes. The

key issue is establishing the coal slurry residence time since gas bubbles occupy some fraction of the

g lreacter volume (gas holdup, , + liquid holdup, , = 1).. Although hydrogen is consumed along the length

of the reactor, light gases are also formed, so it is not easy to predict the gas and liquid holdup volumes.

The "axial dispersion" model, a one-parameter model, is a good approximation for the EDS reactor. Its

flow regime has a small deviation from plug flow conditions. The reactor coils are immersed in a heat

transfer fluid so that the temperature rise along the length of the reactor can be controlled by transferring

heat out. The material and energy balances are the following:

The above equation are solved simultaneously with a numerical differential equation solver. Estimation

of the Dispersion coefficient can be accomplished by the following equation by Shah (Shah 1981).

Liquid and gas holdup are also estimated from formulas in Shah's book.

l l lThis sets the superficial velocity (u = Q/,A)and, in turn the residence time (J = L/u ) of the liquid in the

reactor.

H-Coal Reactor

The H-Coal reactor is rather unique and called an ebullated bed catalytic reactor. A recycle pump,

located either internally or externally, circulates the reactor fluids down through a central downcomer and

then upward through a distributor plate and into the ebullated catalyst bed. The reactor is usually

insulated well and operated adiabatically. Frequently the reactor mixing pattern is defined as backmixed,

but this is not strictly true. A better description of the flow pattern is dispersed plug flow with recycle.

Thus the equations for the axial dispersion model are modified appropriately to account for recycle

conditions.

The schematic below shows the key elements of recycling a portion of the exit stream with the feed

stream and how that affects the feed concentration(s). The recycle increases the superficial velocity, u,

and changes the feed concentration due to dilution with the product stream.

Although the H-Coal reactor is loaded with catalyst, not all of the reactions are catalyzed; some are

thermal reactions, like thermal cracking, which depend on liquid holdup and not on how much catalyst is

present. Thus the material balance equations need to be divided into two categories, one set for the

non-catalytic thermal reactions and another set for the catalytic reactions. A convenient parameter to

use is the thermal/catalytic ratio, T/C, which is the ratio of liquid holdup to catalyst volume. In a

commercial ebullated bed, this ratio is close to 1.0 under ebullation conditions. Consequently the

material balance equations for the catalytic reactions with no recycle is:

The final set of equations for the H-Coal reactor can now be written to account for the recycle situation,

and thermal reactions in concert with catalytic reactions.

The model for the H-Coal reactor introduces two complications beyond the axial dispersion model. First

the boundary conditions are modified to account for the recycle and second the catalyst in the reactor

means that both thermal and catalytic reactions are occurring simultaneously. The equations are solved

numerically with a differential equation solver. This allows the reactor size to be determined by

Aintegrating along dimensionless distance z until concentration, C , equals the desired final

concentration .

Two Stage Liquefaction

A two-stage liquefaction reactor system represents one additional level of complexity beyond the single-

stage H-Coal reactor. Now the exit stream from the first reactor becomes the feed stream for the second

reactor. The reactor system may be visualized as below.

The approach to solving the above reactor system is to first develop the exit values for Reactor 1 and

then to use to these outlet variables as input for Reactor 2. The reactors will usually be operated at

different temperature levels with stage 2 cooler than stage 1 so that the equilibrium for aromatic solutions

reactions is favorably shifted. The interstage cooler is shown in the diagram for this reason.

Furthermore the second stage catalyst may be different with more activity for denitrogenation and

aromatic saturation. The basic algorithm to solve the 2-stage reactor system is listed below:

1. Define the operating conditions for stages 1 & 2

1 2i.e. temperature, T & Tcatalyst, A & B

1 2recycle reaction, R & R

feed concentration, ....

2. Target outlet concentration of stage 2 for reaction species that you are monitoring.

i.e. ....

3. Select outlet concentrations leaving stage 1 which are midway between inlet and final values.

i.e.

4. Calculate diluted feed concentrations to stage 1 by the recycle equation.

5. Solve the set of differential equations for stage 1, stopping when the outlet concentrations satisfy

your criteria. This defines reactor volume and/or catalyst charge for stage 1.

6. Calculated dilated feed concentrations to stage 2 by the recycle equation.

7. Solve the set of differential equations for stage 2, stopping when the outlet concentrations satisfy

your criteria, set in step 2.

8. Compare relative size of stage 1 and stage 2. If imbalanced too much, modify interstage

concentration or temperature level, and then resolve system of differential equations, starting at

step 2 of algorithm.

Table 16Liquefaction Reactor Guide

Reactor Type Positive Negative Best Application

@ Research and Development Scale

Micro-batch- Inexpensive - moderate mixing - coal screening

- small feed charge - difficult product work-up - catalyst screening

Stirred Batch autoclave- Good mixing/contacting - more expensive

- Product work-up no problem - hard to set time zero

Mahoney-Robinson continuous stirredautoclave (spinning basket)

- excellent mixing, bubble dispersion - smaller catalyst charge - catalyst aging

- expensive - coal comparison

- heavy feeds may not work well - kinetic studies

Robinson-Mahoney continuous stirredautoclave (stationary basket)

- Handles heavy viscous feeds - expensive - kinetic studies

- Holds most catalyst - catalyst life studies

Trickle Bed- simple to operate - isothermal operation difficult - catalyst aging

- simple to design - flow maldistribution common - coal liquid upgrading

@ Commercial Scale

H-Coal- Extensive operating history. - Tricky to operate - commercial operations

- Handles exothermic reactions well

Exxon EDS- Simpler operation - Heat release hard to control - commercial operation

- hydrogen starvation possible

Two-stage

- Better contol of reactions - More complex to operate - commercial operations

- Improved yield of distillates

- Builds on H-Coal technology

Pilot Plant Processes

The following section contains "short-form" process descriptions for some of the leading technologies in

direct coal liquefaction. There are no direct coal liquefaction processes currently in commercial use. Not

only are the production costs expensive, the disposal of waste streams is a big issue. For example, the

residue cannot be burned in a conventional boiler due to its high ash content (-230 %) and sulfur level.

Options have included gasification to generate hydrogen but current thinking leans toward fluid bed

combustion for process heat. Natural gas and process-derived fuel gas appear to be the better raw

materials for generating hydrogen for use in the process.

The four processes described in this section include:

1. H-Coal Process

2. Exxon Donor Solvent (EDS)

3. Solvent Refined Coal (SRC I & II)

4. Two Stage Liquefaction

Process: H-Coal

Description: Direct catalytic hydroliquefaction process developed by HRI, Inc using a novel ebullated

bed reactor (Merdinger et al. 1982?). Typical reactor conditions are 3000 psi, 454°C, and 75 lb coal/hr-ft3

for fuel oil mode or 30 lb coal/hr-ft for syncrude mode.3

Background:

@ Patented process (US 3,321,393) developed by HRI, Inc.

@ Extension of H-Oil process for residuum hydrotreatment

@ Bench scale and process development unit (PDU) operated at the Trenton, NJ laboratories

@ H-Coal pilot plant (200 TPD syncrude or 600 TPD fuel oil mode) built and operated at

Catlettsburg, KY

- Operator: Ashland Synthetic Fuels, Inc.

- Cost: $320 MM

- Participants:

US Department of Energy(DOE)

Kentucky Department of Energy

Ashland Oil

Electric Power Research Institute (EPRI)

Mobil Oil

Standard Oil (Ind)

Conoco Coal Development

Ruhrkohle AG

Process Schematic:

Reactor Conditions/Schematic:

Total Pressure, psi 3000

Temperature,°F 845

Slurry Concentration,% 38

Space Velocity,lb coal/hr-ft 323

Table 17

Process Yields (lb/lb mf coal)

Illinois

Bituminous

Wyoming

Subbituminous

Product Syncrude Fuel Oil Syncrude

2H (5.3) (3.4) (6.2)

2 2H O, CO, CO 7.1 6.5 20.0

2 3H S, NH 3.6 2.2 1.6

1 3C -C 11.2 6.8 12.3

Naphtha (C4-204°C) 18.7 13.4 25.8

Fuel Oil (204-524°C) 29.1 20.8 18.6

Bottoms (524°C+) 35.6 53.7 27.9

TOTAL 100 100 100

Design Considerations:

-Reactor Performance Equations

- Liquefaction Catalysts

Table 18

Name Amocat 1A Amocat 1B HDS 1442ATM TM

Composition 3 3 33%CoO/10% MoO 10% MoO 3.3%CoO/14.7%MoO

Surface Area, m /g2 154 167 323

Avg Pore Diam, Å 123 122 58

Pore Distribution bimodal bimodal bimodal

Size, in diam 1/16 1/16 1/16

- Start-Up/Shut-Down

Start-up of reactor consists of slow heat-up of "cold wall" insulated vessel with

internal refractory. Alternate "hot wall" design requires a high alloy steel liner and is

more susceptible to temperature excursions.

- Reactor Process Control

Catalyst bed level measured by gamma ray detectors which will either raise or lower

ebullation pump rate by speed control on motor. Temperature controlled by temperature

of fresh slurry entering reactor through feed preheater.

- Safety Aspects

Temperature runaway is controlled by quench oil and replacement of hydrogen gas in

reactor with methane to stop the exothermic hydrogenation reactions.

Process: Exxon Donor Solvent (EDS)

Description: The Exxon donor solvent (EDS) process was developed by Exxon Research and

Engineering (ER&E) and is a thermal liquefaction process where the donor solvent is catalytically

hydrotreated in a fixed bed reactor external to the liquefaction zone (Epperly et al. 1981; Hsia et al. 1981;

Ansell et al. 1980). Typical liquefaction reactor conditions are 840°F and 2000-3000 psia with 45

minutes residence time.

Background:

@ Process developed by Exxon Research and Engineering

@ Two pilot plants (75 lb/day) and a third (1 ton/day) at Baytown Research

and Development Division in Houston, TX set the design scale-up basis

@ Larger EDOS process (250 ton/day) built and operated at Baytown, TX

- Operator Exxon

- Cost $341 MM

- Participants

US Department of Energy (DOE)

Exxon

Electric Power Research Institute (EPRI)

Japan Coal Liquefaction Development Company

Phillips Coal Company

Anaconda Minerals Company

Auhrkohle AG

ENI (Italian National Oil Company)

Process Schematic:

Reactor Conditions/Schematic:

Process Yields:

Table 19Product Yields of EDS Process

(2500 psig, 840°F)

Illinois No. 6bituminous(Monterey)

Wyomingsubbituminous(Wyodak)

Texaslitnite(Big Brown)

Residence time, min 40 60 25-40

Yields, wt% maf coal;

2H -4.3 -4.6 -3.9

2 xH O+CO 12.2 22.3 21.7

2 3H S+NH 4.2 0.9 1.7

1 3C -C gas 7.3 9.3 9.1

4C -1000 oF liquid 38.8 33.3 33.3

1000 oF + bottoms 41.8 38.8 38.1

Design Considerations:

- Donor Solvent Hydrotesting Catalyst

- Start-Up/Shut-Down

- Reactor Process Control

- Safety Aspects

Process: SRC - I & II

Description: The Solvent Refined Coal liquefaction process referred to as SRC-II is a thermal

liquefaction process; it is an outgrowth of an earlier Solvent Refined Coal process tested by Gulf Oil in

the 1960s (Rao et al. 1983; Moniz et al. 1984; Gir and Rhodes 1983; Nalitham et al. 1985; Moniz and

Nalitham 1985). The earlier process, known as SRC-I, was aimed at boiler fuel production of an ashless

low-sulfur solid fuel. In contrast, the SRC-II is geared to the production of synthetic liquid fuels. Typical

reactor conditions are 870°F, 1500-2500 psig, and 15-45 minutes with no catalyst.

Background:

@ Two major pilot plants

- 50 TPD at Fort Lewis, Washington under Gulf Oil and DOE

- 6 TPD at Wilsonville, Alabama originally built by

Catalytic, Inc. under sponsorship of Southern Company

Services, Inc. eventually set up as:

Sponsors Participants

US Department of Energy Catalytic, Inc.

Electric Power Research Institute Kerr-McGee Corporation

Amoco Corporation Hydrocarbon Research, Inc. Southern Company

Services, Inc.

Integration of H-Oil hydrotreater into process employed at Wilsonville so that some of recycle slurry oil

came from catalytic hydrotreater; it led to the process being called two-stage liquefaction with the

standard integrated mode (ITSL) and double integrated mode (DITSL).

Control Solvent Deashing (CSD) by Kerr-McGee eliminated filters and allowed recycle of heavy residue

as major component of slurry oil

Process Schematic:

Reactor Conditions/Schematic:

Table 20SRC II With Interstage Deashing

ITSL DITSL

run no. 246F 246G 246A 246B 246C 246DE

THERMAL STAGE*

average reactor temperature (°F) 826 813 812 812 804 825

inlet hydrogen partial pressure (psi) 2,050 2,040 2,040 2,130 2,190 2,160

coal space velocity [b/hr-ft (>700°F)] 24 17 23 27 30 243

solvent-to-coal ratio 1.8 1.8 1.8 1.8 1.8 1.8

solvent resid content (wt %) 34 34 0.7 1.0 1.0 0.9

CATALYTIC STAGE

reactor temperature ( F) 623 623 624 643 606 624o

space velocity (lb feed/hr-lb cat) 1.1 1.0 0.7 1.0 1.0 0.9

feed resid content (wt %) 34 34 22 14 24 22

catalyst age (lb resid/lb cat) 360 496 69 128 204 252-279

2 3*additon of Fe O at 2.0% MF coal and DMDS at 1.1 x stoichiometric requirement for conversion of2 3Fe O to FeS.

Table 21Process Yields

ITSL DITSL

run no. 246F 246G 246A 246B 246C 246DE

yield* (% MAF coal)

1 3C -C gas (total gas) 8(19) 9(19) 7(17) 6(14) 5(13) 9(17)

water 12 11 12 11 12 12

4C +distillate 51 53 51 50 49 55

resid 3 1 -2 -1 0 -1

hydrogen consumption -5.1 -5.4 -5.0 -4.6 -4.1 -5.4

hydrogen efficiency4 2(lb C +dist/lb H consumed

10.0 9.8 10.3 11.0 12.0 10.2

distillate selectivity1 3 4(lb C -C /lb C +dist)

0.17 0.18 0.14 0.12 0.10 0.15

energy content of feed coalrejected to ash conc.(%)

22-24 22-24 26-29 29-32 30-35 21-24

3*elementally balanced yield structures, SO -free ash

Process: Two Stage Coal Liquefaction

Description: Compared to SRC-I & SRC II, two-stage liquefaction tightly couples the dissolver and

catalytic hydrotreater so that the time between the two stages is minimized. Ash removal occurs after

the second catalytic stage rather than interstage deashing. This This accomplishes higher liquid yields

with less hydrogen consumption with two stages operating at different temperature. Second stage

always has catalyst in it but the first stage can be either catalytic or thermal. Typical reactor conditions

are 2500 psig, 2/1 Slurry Oil/Coal, 45-70 lbs/hr-ft cat (each stage) and 650-775°F first stage/810-825°F

second stage (Comolli, MacArthur, and McCleary 1985).

Background:

@ Original concept introduced by Chevron Research in late 70s (US)

@ Pilot plant tests by HRI, Inc. in mid 80's with exceptional success noted

by high distillate and low residuum yields

@ Wilsonville pilot plant operating as SRC was modified to tightly couple

the 1st stage dissolver with 2nd stage H-Oil hydrotreater in 1986-87 and

called RITSL (Reconfigured Integrated Two-Stage Liquefaction)

operating mock.

@ Two stage process under some debate with high/low and low/high

temperature staging not showing yields too different. HRI favors

low/high staging while Amoco prefers high/low to favor aromatic

saturation in 2nd stage.

Process Schematic:

Reactor Conditions:

Table 23CATALYTIC TWO-STAGE LIQUEFACTION PROCESS (HRI tests)

WYODAK COAL YIELDS AND PROCESS PERFORMANCERun 227-22

Process Yields (lb/lb mf coal)

RANGE(1)

REPRESENTATIVEYIELDCONDITION 7

YIELD, W % M.A.F. COAL

1 3C -C 5 - 9 8.1

4C -390 F 17 - 27 25.0o

390-650 F 30 - 47 36.5o

650-975 F 0 - 11 4.2o

975 F Resid 0 - 5 1.9o +

Unconverted Coal 7 - 21 10.7

2 2H O, CO, CO 17 - 23 20.0

2 3H S, NH 1.4 - 1.8 1.6

Hydrogen Consumption 6.3 - 8.1 8.1

4C -975 F Liquid 54 - 68 65.7o

PROCESS PERFORMANCE

Coal Conversion 83 - 92 89.3

975 F Conversion 74 - 90 87.4o +

Hydrogen Efficiency 8 - 9 8.2

Table 24RITSL yield structures (thermal/catalytic) (Wilsonville pilot plant)

run no. 247C-II 247D 247E**

yield*(%MAF coal)

1 3C -C gas (total gas) 7(12) 6(12) 5(11)

water 10 9 11

4C +distillate 60 62 61

resid 5 3 3

hydrogen consumption -6.1 -6.1 -6.3

hydrogen efficiency4 2(lb C +dist/lb H consumed)

9.8 10.2 9.7

distillate selectivity1 3 4(lb C -C /lb C +dist)

0.11 0.10 0.09

energy content of feed coalrejected to ash conc. (%)

22 22 22

*elementally balanced yield structures**catalytic stage feed sensitivity studies

Design Considerations:

- Liquefaction Catalyst

HDS - 1442A, Amocot 1A(Co-Mo), and Amocot IC(Ni-Mo) have been

evaluated by HRI, Inc. at Trenton, NJ laboratory.

- Start-up / Shut-down

- Reactor Process Control

- Safety Aspects

References

Ansell, L.L., J.W. Tauton, and K.L. Trachte. 1980. Bottoms recycle studiesin the EDS process development, Proceedings of the Fifth Electric Power Research InstituteContractor's Conference on Coal Liquefaction. May 1980, Palo Alto, CA.

Autoclave Engineers Bulletin 1200. Autoclave Catalytic Reactor SelectionGuide.

Carberry, J.J. 1976. Chemical and Catalytic Reactions Engineering. McGraw- Hill.

Churchill, S.W. 1974. The Interpretation and Use of Rate Data: The RateConcept. Script Publishing Company.

Comolli, A. G., J. B. MacArthur, and J. B. McCleary. 1985. Two-StageCatalytic Conversion of Wyodak Coal. Proceedings: 10th Annual EPRI Contractor's Conference onClean Liquid and Solid Fuels EPRI AP-4253-SR October, 1985.

Epperly, W.R., K.W. Plumlee, and D.T. Wade. 1981. Exxon donor solvent coalliquefaction process: Development program status VI. Chemical Engineering Program 7 (5):73-79.

Fischer, F. 1925. The Conversion of Coal Into Oils. New York: Van NostrandCompany.

Fulton, J.W. 1986. Testing the Catalyst. Chemical Engineering.

Given, P.H. 1986. Course Notes-Coal Origins and Characteristics,Pennsylvania State University.

Gir, S. and D. E. Rhodes. 1983. Recent Developments in the Integrated Two-Stage Coal Liquefaction Program at Kerr-McGee Corporation. Proceedings of DOE Direct CoalLiquefaction Contractors' Review Meetings, 16-17 November 1983.

Hsia, S.J. et al. 1981. Liquefaction of low rank coals with the Exxon donorsolvent coal liquefaction process, Proceedings of the 11th Biennial Lignite Symposium, June 1981,San Antonio, TX.

Kim, et al. 1979. Catalyst Development for Coal Liquefaction. EPRI Contract 408 (1):AF-1235.

Kuni, D., and O. Levenspiel. 1969. Fluidization Engineering. John Wiley and Sons.

Levenspiel, O. 1972. Chemical Reaction Engineering. John Wiley and Sons.

Levy, R.B., and J.A. Cusumano. 1977. New Materials for Coal Liquefaction.Liquid Fuels from Coal, ed. Elliot --------- Academic Press.

Lumpkin, R.E. 1988. Recent Progress on the Direct Liquefaction of Coal.Science 239.

Mahoney, J.A., K.K. Robinson, and E.C. Myers. 1978. Catalyst Evaluation with the GradientlessReactor. Chemtech.

Merlinger, M and A. Comolli. 1982? Recent Developments in the Processing of Western Coal in the H-Coal Process. Sixth Annual EPRI Contractor's Conference.

Moniz, M. J. et al. 1984. Process Studies of Integrated Two-Stage CoalLiquefacton at Wilsonville. Presented at the Ninth Annual EPRI Contractors' Conference on CoalLiquefaction, 8-10 May 1984.

Moniz, M. J. and R. V. Nalitham. 1985. An Assessment of the Impact ofDelayed Deashing and Subbituminous Coal Processing on Hydrotreating Catalyst Performance. Presented at the Tenth Annual EPRI Contractors' Conference on Coal Liquefaction, 23-25 April1985.

Nalitham, R. V. et al. 1985. An Experimental Evaluation of the ResidenceTime Distribution in the Wilsonville Dissolver Using Radioactive Tracers. Presented at the TenthAnnual EPRI Contractors' Conference on Coal Liquefaction, 23-25 April 1985.

Rao, A. K. et al. 1983. Recent Developments in Two-Stage Coal Liquefactionat Wilsonville. Proceedings of DOE Direct Coal Liquefaction Contractors' Review Meeting, 16-17November 1983.

Satterfield, C.N. 1970. Mass Transfer in Heterogeneous Catalysis. Colonial Press.

Schlesinger, M.D. 1980. Past Trends in Synthetic Fuels R&D. PETC/TR-80/4.

Schobert, H.H. 1987. Coal-The Energy Source of the Past and Future.American Chemical Society.

Shah, Y.T. 1981. Reaction Engineering in Direct Coal Liquefaction. Addison- Wisley.

Spackman, W. 1986. Course Notes-The Nature of Coal and Coal Seams. Pennsylvania StateUniversity.

Stohl, F., and H. Stevens. 1985. Catalyst Deactivation in Direct CoalLiquefaction. A Comparative Study of Wilsonville Runs. 10th Annual EPRI Contractor's Conferenceon Clear Liquid and Solid Fuels, ed. -------------. EPRI-AP-4252-SR.

Weisser, O., and S. Landa. 1973. Sulfide Catalysts, Their Properties andApplications. Pergoman Press.