EPA-650/2-75-011




JANUARY 1975
Environmental Protection Technology Series

-------
                  RESEARCH REPORTING SERIES


Research reports of the Office of Research and Development, U. S. Environ-
mental Protection Agency, have been grouped into series. These broad
categories were established to facilitate further development and applica-
tion of environmental technology.  Elimination of traditional grouping was
consciously planned to foster technology transfer and maximum interface
in related fields.  These series are:

         1. ENVIRONMENTAL HEALTH EFFECTS RESEARCH
         2. ENVIRONMENTAL PROTECTION TECHNOLOGY

         3. ECOLOGICAL RESEARCH
         4. ENVIRONMENTAL MONITORING
         5. SOCIOECONOMIC ENVIRONMENTAL STUDIES

         6. SCIENTIFIC AND TECHNICAL ASSESSMENT REPORTS
         9. MISCELLANEOUS

This report has been assigned to the ENVIRONMENTAL PROTECTION
TECHNOLOGY series.  This series describes research performed to
develop and demonstrate instrumentation,  equipment and methodology
to repair or prevent environmental degradation from point and non-
point sources of pollution.  This work provides the new or improved
technology required for the control and treatment of pollution sources
to meet environmental quality standards.

-------
                                      EPA-650/2-75-011
SULFUR  AND  NITROGEN  BALANCES
      IN  THE  SOLVENT  REFINED
              COAL  PROCESS
                        by

                   Charles H. Wright

         The Pittsburg & Midway Coal Mining Company
                  9009 West 67th Street
                 Merriam,  Kansas 66202

                        for

                 Office of Coal Research
              U.S. Department of the Interior
                   2100 M Street, NW
                Washington, D. C.  20037
             IAG No. EPA-IAG-D4-0454, Task 1
                  ROAP No. 21ADD-025
                Program Element No. 1AB013
            EPA Project Officer: Lloyd Lorenzi, Jr.

                Control Systems Laboratory
            National Environmental Research Center
         Research Triangle Park, North Carolina 27711
                     Prepared for

          OFFICE OF RESEARCH AND DEVELOPMENT
          U.S. ENVIRONMENTAL PROTECTION AGENCY
                WASHINGTON, D.C. 2046%

                     January 1975

-------
                        EPA REVIEW NOTICE

This report has been reviewed by the National Environmental Research
Center - Research Triangle Park, Office of Research and Development,
EPA, and approved for publication.  Approval does not signify that the
contents necessarily reflect the views and policies of the Environmental
Protection Agency, nor does mention of trade names or commercial
products constitute endorsement or recommendation for use.
This document is available to the public for sale through the National
Technical Information Service, Springfield, Virginia 22161.
                                  ii

-------
                    ABSTRACT

The purpose of an exact elemental balance study of the
solvent Refined Coal (SRC)  process with the laboratory reactor
was to determine the fate of sulfur and nitrogen fed in
the slurry.  The work was performed in late 1972 as part
of a normal experiment with Kentucky #9 high volatile B
bituminous coal' and a blend of processed anthracene oils
under 1000 psig^ hydrogen pressure.  A variety of tech-
nique studies had been made in preparation, such as in-
vestigation of the effects of sample size on analysis and
methods of handling all samples of input and output
material for maximum recovery and representative com-
position.  Accounting for carbon and hydrogen was accu-
rate, for sulfur good, for nitrogen poor.  A detailed
comparison of conventional Kjeldahl and Dumas analytical
results for nitrogen in coal and solid products revealed
that input nitrogen is not fully reported by Kjeldahl
and that sample size affects nitrogen results reported
by Dumas.  Nitrogen analysis needs further investigation.
The study forced review of sampling and handling tech-
niques as well, with salutary results in laboratory work.
For similar balance study in any scaled up version of the
upflow reactor, a preliminary attempt for the purpose of
discovery and procedure development should precede the
final detailed work.

This report was submitted in partial fulfillment of Agreement
Number EPA-IAG-D4-0454 between the U. S. Bnviromental Protection
Agency and the office of Coal Research, U. S. Department of
the Interior.  Work was performed by the Pittsburg & Midway
Coal Mining Company under Office of Coal Research contract
14-01-0001-496 and was completed during October 1974.
                         111

-------
                        CONTENTS


                                                   Page
                                                   - — —•*- -

Abstract                                            ii

List of Figures                                     vi

List of Tables                                      vii


Sections

I       Conclusions                                 1

II      Recommendations                             2

III     Introduction                  -             3

        III-A  Scope of This Report                 3
        III-B  Background                           4
        III-C  General Objectives of the            6
               Continuous Reactor Program

IV      Equipment and Procedures                    9

        IV-A   Description of the Continuous        9
               Reactor

               IV-A-1  Gas Compression and          9
                       Metering System
               IV-A-2  Slurry Pumping and           13
                       Metering System
               IV-A-3  Reaction and Sampling        15
                       Systems

        IV-B   Reactor Operation                    21

               IV-B-1  Startup                      22
               IV-B-2  Run                          23
               IV-B-3  Shutdown                     24
               IV-B-4  Concurrent Operations        24
               IV-B-5  Laboratory Procedures for    25
                       Control of Feed Slurry
               IV-B-6  Lineout Control              26
               (Continued on Next Page)

                           iv

-------
                       CONTENTS
                       (Continued)
Sections

        IV-C  Workup of Coal Solution                29

              IV-C-1  Stripping                      30
              IV-C-2  Filtration                     33
              IV-C-3  Distillation for Solvent       35
                      Recovery
              IV-C-4  Workup of Wet Filter Cake      36

        IV-D  Analytical Methods                     37

              IV-D-1  Analysis of Gaseous Products   37
              IV-D-2  Analysis of Feed Slurry and    43
                      Liquid Products  '

                      IV-D-2-(a) Feed Slurry         43
                      IV-D-2-(b) Product Solution    48
                      IV-D-2-(c) Water Phases        49
                      IV-D-2-(d) Light Oils          50
                      IV-D-2-(e) Cut 2 Oil           52
              IV-D-3  Analysis of Solid Products     55
        Results of Experiment CU 48                  58

        V-A  Objectives                              58

        V-B  Materials Used                          58

             V-B-1  Coal             .                58
             V-B-2  Solvent                          61

        V-C  Reactor Operation                       71

        V-D  Yields                                  74

        V-E  Gas Analysis                            75

        V-F  Elemental Balances                      75

             V-F-1  Accounting for Material Fed      77
             V-F-2  Weights of Materials Fed         78
             V-F-3  Accounting for Pro-ducts          82
             V-F-4  Composition of Products          83



             (Continued on Next Page)

                         v

-------
                       CONTENTS
                      (Continued)

                                                     Page

VI        Review of Elemental Balance Calculations    87

          VI-A  Material Losses                       87

                VI-A-1  Ash                           87
                VI-A-2  Water                         89

          VI-B  Recalculations                        90

                VI-B-1  The Reporting of Sulfur       92
                VI-B-2  The Reporting of Nitrogen     92

          VI-C  Gains from the Study                  99


Glossary                                              102

Appendix A  Determination of Hydrogen Sulfide in      107
            SRC Product Gas

Appendix B  Molecular Weight of Product Gas           114

Appendix C  Conditions Used for Gas Solid             115
            Chromatograph of Process Gas Samples

Appendix D  Automatically Recorded Potentiometric     116
            Aqueous and Nonaqueous Titration
            Information

Appendix E  IR Spectrum, CU 48A1 Unfiltered Coal      130
            Solution

Appendix F  IR Spectrum, CU 48A1 Stripped and         131
            Filtered Coal Solution

Appendix G  IR Spectrum, CU 48A1 Cold Trap Oil        132

Appendix H  IR Spectrum, CU 48A1 Cut 1 Oil            133

Appendix I  IR Spectrum, CU 48A1 Cut 2 Oil            134

Appendix J  Blackness and IR in Successive Samples,   135
            CU 80

Appendix K  First Trial Elemental Balance, CU 48A1    136

Appendix L  First Trial Elemental Balance, CU 48A2    137
                             VI

-------
                     LIST OF FIGURES
No.                                                Page
1    SRC Process, Simplified Schematic               7
2    Gas Compression and Metering System            10
3    Slurry Pumping and Metering System             14
4    Reaction and Sampling Systems                  17
5    Workup                                         31
6    Stripping Apparatus                            32
7    Filtration Equipment                           34
8    IR Spectrum for Product Gas                    40
9    Gas Density Bulb                               41
10   Nitrogen in Feed Coal  (Dumas Method)           62
11   Effect of Hydrogenation on Absorbance at       67
     3.28 Microns
12   Differential IR Working Curve                  69
13   Working Curve, % H in Anthracene Oil Vs.       72
     TR Ratio
                            Vll

-------
                     LIST OF TABLES
No.                                                Page
1   Product Distribution and Analytical Methods     38
2   Experimental Conditions                         73
3   Operating Data                                  73
4   CU 48 Yields                                    73
5   CU 48 Solvent Compositions                      75
6   CU 48 Product Gases                             76
7   Elemental Balance, CU 48A1                      79
8   Elemental Balance, CU 48A2                      80
9   Product Gas Composition, CU 48A1                84
10  Product Gas Composition, CU 48A2                85
11  CU 48 Vacuum Bottoms                            85
12  Comparative Sulfur Analyses (%)                 93
13  Reproducibility in Feed Coal Nitrogen Results   95
    (Kjeldahl)
14  Alternate Nitrogen Determinations               96
15  Comparative Nitrogen Analyses                   98
                            Vlll

-------
                      SECTION I
                     CONCLUSIONS

The elemental balance study revealed shortcomings in
the handling of feed coal, a need for better suspension
of mineral matter in feed slurry, a need for larger
samples for analysis, a need for changes in sampling
techniques, and a need for a better analytical method
for nitrogen in feed coal and SRC products.

Material recovery, 98% at the time of the study, is
currently 99% or better as a result of subsequent im-
provements.  This is significant because reactor behavior
permits a preferential segregation of mineral, which is
rich in sulfur; laboratory-size, short-interval samples
reveal variations in ash content, forcing a prorating of
lost sulfur to the product distribution.  The loss of
water from the slurry feed vessel has similarly been
controlled.

The balances for carbon and hydrogen are accurate, for
sulfur good.  Accounting for nitrogen from raw data is
awkward because output exceeds input, especially by
conventional Kjeldahl analyses; the effect of sample
size on Dumas analysis of feed coal is demonstrable, and
other analytical methods for products have special limit-
ations .

It is clear that the SRC process removes from coal nearly
all of the ash, most of the sulfur, and less of the
nitrogen.  To remove an appreciable fraction of the nitro-
gen might require liquefaction under conditions that
convert all material to hydrocarbons of appreciably lower
weight than is usual now in the SRC process.
                      -1-

-------
                     SECTION II

                  RECOMMENDATIONS


Insofar as they would apply to the laboratory reactor

operation, many of the recommendations arising from
the study have already been adopted:

     1.  The uniformity of feed coal composition was
         achieved by adopting the ASTM long pile
         sampling method for handling initial lots, by
         improved methods for homogenizing the ground
         sublots, and other procedures.

     2.  The uniformity of feed slurry has been greatly
         improved by modified grinding and sieving
         equipment for the coal, modified stirring of
         slurry in the feed vessel for suspension of
         mineral, and by avoidance of the need to warm
         slurry and hence lose contained water from the
         input.

     3.  The effect of holdup of material in sample
         flasks has been significantly reduced by taking
         larger samples and, when appropriate, leaving
         flasks uncleaned for a relatively constant
         holdup of mineral-rich material between runs;
         so that raw material recovery is good, mass
         balances are regularly closed above 99%, and
         exact elemental balances are facilitated.
For exact nitrogen balances, analytical methods other than
Kjeldahl and Dumas should be investigated.  This is par-
ticularly true for feed coal, solids, and high molecular
weight products.


In any scaled up operation of the SRC process, an exact

element-by-element balance will require good understanding
of mineral behavior in the dissolver and its operating
characteristics.  Recovery of either 100% raw material or
100% ash is unlikely.  Therefore, a preliminary elemental

balance study should be undertaken for the purpose of dis-
covery and procedure development some months before final

detailed study of scaled up SRC process operation.

                       -2-

-------
                   SECTION III
                  INTRODUCTION

III-A  SCOPE OF THIS REPORT
Section I presents conclusions reached from the detailed
elemental balance studies of the SRC process reported in
Section V.

Section II lists those recommendations, chiefly affecting
handling and workup of samples and certain analytical
methods, resulting from the studies.

This section sketches the background for this report and
the general objectives of the continuous reactor program
in the SRC process.

Section IV describes the laboratory reactor in use when the
studies were made.  It includes typical operating pro-
cedure, typical workup of the coal solution samples, and
typical analytical methods then employed.

Section V describes experiment CU 48:  its objectives, the
materials used, run conditions, yields, product distri-
bution and compositions, and the elemental balances.

Section VI evaluates the errors, problems, and gains of
the study.

The glossary contains all special terms and abbreviations
used in this report.  The appendices include certain
detailed analytical methods employed, worksheets, illus-
trations of infrared spectra  for various  solvents, and
similar supporting data.

                      -3-

-------
III-B  'BACKGROUND
Studies in the 1960's related to the solvent refining of
coal had been conducted in a laboratory batch autoclave
and in a continuous unit capable of processing about 23
kilograms (50 pounds) of coal per hour.  In these experi-
ments the material balances had generally been closed to
account for most (often 98-100%) of the materials fed to
the reactor.  However, the balance made for each individ-
ual element had been made with variable accuracy.  Some
suggestion of the range of reported values is found in
OCR R&D Report No. 9  (Gulf Oil):
         GRAMS IN              GRAMS OUT
                          Run B-55   Run B-62
 C         505.4           503.0      519.9
 H          35.16           28.6       35.03
 N           4.96            5.29       6.52
 S           9.48            6.69       5.43
 O          20.98           13.58      20.49
Ash         15.50           14.9       11.14
As a rule the balances for carbon were closed accurately,
and for hydrogen reasonably well; but sulfur was lost
from the system, and the products failed to account for
the amount of sulfur fed.

Another problem was that nitrogen obtained from the
products frequently exceeded nitrogen reported in materials
fed into the reactor, in the examples given above, by 0.3
to 1.6 g.

The problem with nitrogen was also evident in work done
with other coal conversion processes.  OCR R&D Report No.
39 (Consolidated Coal) reported three elemental balances:
                      -4-

-------
1356-1
GRAMS
IN
621.29
52.71
3.81
22.09
8.42
6.99
26.60
OUT
620.98
52.19
3.82
22.68
7.20
6.91
28.14
1356-2
GRAMS
IN
619.92
53.34
4.31
23.24
7.85
7.00
26.64
OUT
621.74
53.06
4.22
21.98
5.55
7.60
27.98
1356-3
GRAMS
IN
625.02
54.19
4.87
25.56
6.59
7.04
26.78
OUT
627.66
54.24
5.38
24.53
5.15
7.03
29.55
c
H
N
O
Organic S
Inorganic S
Ash less
inorganic S
These balances were obtained from a much milder reduction
process than the other two, and presumably output would
look more like input.  One of Consolidated's three balances
and all P&M* balances show more nitrogen in products than  in  feed,

Such results indicated sampling errors and probably some in-
accurate analytical methods.  The Environmental Protection
Agency  (EPA) had received coal process data from reports
by Pittsburg & Midway coal Mining Co.  (P&M) and others; the
EPA was interested in obtaining more accurate results
than those currently available.  The best laboratory data
from the P&M Process Development Plant or laboratory
tended to close the elemental balance for carbon most of
the time, for hydrogen some of the time, for sulfur and
nitrogen not usually.  P&M was therefore asked to make
studies of the technique, sampling, and analytical methods
necessary to close these balances.

This work required,first, that the continuous laboratory
reactor be operated by the best method for making samples,
and, second, that these samples be the basis for analytical
studies designed to develop a closed elemental material
balance.  The problems attacked fell into several cate-
gories:
*The Pittsburg & Midway Coal Mining Company, a subsidiary of
 Gulf Oil Corporation.
                       -5-

-------
     Materials preparation and analysis
     Reactor operating techniques designed to obtain well
     controlled and well validated samples
     Sampling and workup techniques which should produce a
     set of samples capable of producing an elemental balan<
     Analytical methods which produce accurate results

Eventually these balances will have to be repeated in the
Tacoma (Fort Lewis) Pilot Plant studies.  The number of
vessels involved, and therefore the number of accurate
samples needed, will probably be larger.  This in turn will
probably require a considerable study of sampling before
accurate material balances are obtained.  From such
samples the element-by-element balance should be possible.
In addition to studies with the major elements, the Pilot
Plant will eventually conduct a sampling campaign to
produce balances for a number of trace elements which
affect the environment.  The laboratory studies reported
here are thus the leading effort in a more extensive work
plan.

III-C  GENERAL OBJECTIVES OF THE CONTINUOUS REACTOR PROGRAtJ
The goal of all processes to upgrade coal is to produce a fi
high in Btu/lb and low in sulfur and ash.  In the Solvent
Refined Coal process (figure 1), coal is dissolved without
added catalyst under moderate hydrogen pressure in a solver
generated by the process itself.  Most organic matter in
the feed coal dissolves.  The liquid can be filtered to
remove mineral matter and unreacted carbonaceous matter,
and the filtrate can be subjected to vacuum flash distill-
ation for solvent recovery.  The vacuum flash residue is
the upgraded coal.  Some light hydrocarbon byproducts are
obtained also.
                         -6-

-------
                                       SOLVENT REFINED COAL PROCESS
                                                                                               SULFUR
         COAL
                                                          DISSOLVER
PULVERIZER
SOLVENT
RECOVERY
UNIT
                                     PRODUCT
                                     AND SOLVENT
             HYDROGEN
                                                                 SOLVENT REFINED COAL
                                                                                                         t., T«

-------
Figure 1 SRC Process,  Simplified Schematic

-------
The general objective of the autoclave program (OCR Interi

Report No. 6, Vol. II, Part 1) •,  to improve our knowledge o

the chemistry of the coal solution process, is the same

in the continuous reactor program.  During the 1970's

the parts of this study have included:
     1.  Investigation of a reactive initial solvent
         for startup

     2.  The modeling of startup conditions at the
         Tacoma Pilot Plant

     3.  The development of suitable sampling techniques,
         handling methods, and accurate analytical
         methods which collectively should monitor the
         reaction, account for materials fed in, and
         permit valid correlations of results from
         run to run

     4.  Establishment of operating conditions for a
         given solvent and coal at a given ratio which
         produce or reproduce enough reusable solvent and
         other desired products

     5.  Investigation of carbon monoxide and hydrogen
         as reducing gas and the effects of shift re-
         actions in the reactor

     6.  Examination of the effects of catalysts,
         different reducing gases, different retention
         times in the reactor, and similar variables
                         -8-

-------
                    SECTION IV
             EQUIPMENT AND PROCEDURES

This section discusses the continuous reactor, the
operating procedure, typical workup of process materials,
and methods of analysis employed at the time experiment
CU 48 was done.  An understanding of the process and ana-
lytical techniques is prerequisite to evaluation of
results obtained, and to assessment of probable results
to be expected from a large scale reactor.

IV-A  DESCRIPTION OF THE CONTINUOUS REACTOR
The continuous reactor may be considered as several inter-
connected systems.  These systems, repeatedly modified
since CU 48 to model the Tacoma Pilot Plant design or to
improve operation, are described as they existed during
the experiment reported.   (Further information is avail-
able in Office of Coal Research R&D Report No. 53, Interim
Report Nos. 7 and 8.)  The following block diagram shows
the relationship; when samples are not being taken, the
flow through the reactor is directed to the waste receiver
and gas is vented.
       Gas Compression
          & Metering
       Slurry Pumping
          & Metering
                          (Preheater  &J	    .	,
                          |  Dissolverj  p •*   i  ISamplesJ
Waste
 IV-A-1   Gas Compression and Metering System
 This  system  (figure  2) is designed  to  take hydrogen  from
 a  standard gas  storage bottle, compress  and  store  a
 reserve  of this gas  above reactor operating  pressure,  and
 then  meter the  stored gas into the  reactor.   The system
 provides for  calibration of gas  delivery at  operating
                         -9-

-------
                                                   k TO
                                                    REACTOR
                                   GAS SUPPLY CYLINDER
                                   REGULATOR
                                   ROTAMETER
                                   SUCTION RESERVOIR
                                   COMPRESSOR
                                   HP GAJTsTORAGE AUTOCLAVE
                                   RECYCLE CONTROL VALVE
                                   HP REGULATOR
                                   HP INDICATOR
                                   PRESSURE RECORDER
                                   REACTOR PRESSURE INDICATOR

                               12   BACK PRESSURE REGULATOR
                               13   PRESSURE CONTROLLER
                               14   PRESSURE INDICATOR
                               15   NEEDLE METERING VALVE
Figure 2
Gas Compression and Metering System
                      -10-

-------
pressure/ and for recording both delivery pressure and
operating pressure in the reactor.  Hydrogen flow is
adjusted by maintaining a controlled differential pressure
across a metering valve.

In figure 2, commercial size 1A gas cylinders deliver gas
through a high pressure (HP) regulator and HP rotameter
to the suction end of the compressor.  The rotameter/
Brooks 1410-01D1B1A, is operable to 134 kg/cm2 (1900 psig)
at 37.7°C (100°F).  Since the compressor/ Aminco 46-4025SP,
                                          2
operates with a suction in the 56-70 kg/cm   (800-1000 psig)
range/ gas cannot be completely pumped from a cylinder
with this compressor.  The suction reservoir for this
compressor is an autoclave vessel/ Autoclave Engineers
3-3/4-liter  (1-gallon) size.  It maintains gas at regu-
                                          2
lated input pressure/ normally 56-70 kg/cm   (800-1000 psig)/
set by the regulator on the gas cylinder.  The air-
operated compressor takes gas from the suction reservoir,
boosts its pressure/ and stores it in a 1-liter HP gas
                                  2
storage autoclave at 155-190 kg/cm   (2200-2700 psig).
Gas from this storage autoclave passes through an HP
regulator and needle metering valve to the reactor.
Excess compressed hydrogen is recycled to the suction side
of the compressor through a pneumatic recycle control
valve, Annin 5060/ whose setting thereby controls the
pressure maintained in the HP storage autoclave.

The flow of gas to the reactor is metered by the needle
valve; high side and low side regulators govern the  flow.
High side pressure to this valve is set on an HP regulator,
Marotta RH-292GA-35-2222, located between HP storage auto-
clave and needle valve.
                         -11-

-------
Low side pressure is set by the reactor operating pressure,
maintained as described later.  The HP needle metering
valve, though adjustable if necessary upon changing
types of reducing gas, in practice is seldom readjusted.
A pressure recorder, Foxboro Consitrol 5420 F-E, pro-
vides a continuous record of pressures on each side of
the metering valve.  The recorder operates through
differential pressure cells, Foxboro 11 GM Force Balance
Pneumatic Pressure Transmitter units.  Bourdon type in-
dicating gages provide separate checks on recorder
performance.

To calibrate flow through the HP needle metering valve,
gas to the reactor is diverted through a Grove Model
S-91XW back pressure regulator, set at the desired re-
actor operating pressure, to an exit line.  Differential
pressure between the HP regulator and the back pressure
regulator determines gas out; the volume is measured by
a wet test meter at the exit line.  Minor volume adjust-
ments per unit time are controlled by small changes in
differential pressure, and larger adjustments are made by
resetting the needle valve also.  Initial calibration of. hj
rogen flow to desired reactor delivery rate therefore requi
both a valve setting and an operating differential pres-
sure to be established which will result in desired
flow at desired pressure.  Any subsequent calibration is
then accomplished by minor adjustments of the differential
pressure.   (The back pressure regulator is adjusted by
setting a gas pressure in the head of a pressure dome.)
Nitrogen lines, not shown in figure 2, are provided to
pressurize the back pressure regulator for readjustment
when necessary and to purge reducing gas from the re-
                        -12-

-------
actor or gas compression system when necessary.  Air
supply lines for operation of the pressure sensing and
pressure controlling devices have also been omitted from
figure 2.

IV-A-2  Slurry Pumping and Metering System
This system {figure 3) maintains a uniform slurry, pumps
it into the reactor at a uniform rate, and measures the
pumping rate accurately.  In addition, it maintains a
supply of solvent for startup and shutdown.

Reliable performance of this system is essential to satis-
factory reactor operation.  Failure of a pump in this
system is the most common cause of subsequent difficulty
in the other systems.  When flow is stopped in the pre-
heater or dissolver, repolymerization or even coke for-
mation in the high temperature vessels commonly results.
Either condition may permanently plug the preheater.

Coal and solvent in weighed amounts are mixed separately
and then transferred to the slurry feed vessel  (figure 3),
a stainless steel vessel of about 8 liters capacity.  It
is fitted with a stirrer, operated continuously to keep
the coal suspended in the solvent.  Slurry is drawn up
through a dip tube reaching close to the bottom of the
feed vessel, an arrangement allowing room for the scale
to move and so permit weighing of the feed vessel, and
circulated  (during CU 48) by a centrifugal pump.  The
dip tube is stainless steel, rigid enough to stand
securely in place when connected to the feed line for the
centrifugal pump.  The return stream of slurry was made to
run down the dip tube to reduce a tendency to inject air
                         -13-

-------
1  SCALE (to establish feed rates)
2  SLURRY FEED VESSEL AND STIRRER
3  EASTERN CENTRIFUGAL PUMP (used to
   prime Hills McCanna pump head)
4  WASTE CAN (receives excess slurry
   and/or flush oil)
5  HILLS MCCANNA DUAL HEAD PUMP
   (modified)
6  BACK PRESSURE REGULATOR (holds
   pressure on flush oil  system)
   FLUSH OIL FEED VESSEL
        Figure 3  Slurry Pumping and Metering System
                            -14-

-------
bubbles below the surface of slurry in the feed vessel.

Slurry in the recirculating loop provides feed for the
suction of the dual head feed pump, Hills McCanna UM2F.
One pump head is used only for standby and only to pump
flush oil when necessary.  The other standard pump
head was replaced because its internal volume was too
large for pumping slow feed rates of coal slurry and
because coal settled out in the pump head.  The replace-
ment pump head, Autoclave Engineers Speed Ranger, catalog
No. 10-4440, was more suitable.

The slurry pumping and metering system thus contains two
feed vessels, one for coal slurry and one for flush oil.
Piping allows the pumping of flush oil by either pump
head, but coal slurry only with the modified pump head.
By this design, solids may be flushed from the slurry
head when necessary.  In normal operations the reactor
is started up with flush oil pumped by the Autoclave
Engineers pump head, which is primed by the centrifugal
pump circulating flush oil.  Before the experiment,
the operator cuts off the flow of flush oil and substi-
tutes feed slurry.  The auxiliary pump head is normally
kept working against a back pressure regulator in order
to keep the pump ready for rapid switching from the pump
head feeding slurry to the other feeding only flush oil.

IV-A-3  Reaction and Sampling Systems
The reactor has two functions:  to dissolve coal, and  then
to further hydrogenate the initial reaction products.
During initial heating of the coal-solvent slurry, most
of the coal is dissolved by a hydrogen transfer reaction
which is largely completed in the preheater.  The solution
                        -15-

-------
is then kept hot and in contact with hydrogen for some
time to hydrogenate the initial reaction product.  Figure
4 is a configuration diagram of the reactor.

The preheater is a coil of high pressure tubing in a
fluidized sand bath.  The dissolver is a 450-ml autoclave
vessel designed for upflow of material.  The designs
permit operating these vessels at different temperatures,
and thus match the capability of the pilot plant for the
SRC process.

The preheater tube used during CU 48 was a  5.4-meter
(18-foot) length of 0.278-cm (7/64-inch) ID pressure
tubing with a volume of about 33 ml; the combined volume
of the preheater and dissolver/ with allowance for
all void space and the volume of the thermowell, was
estimated to be 485 ml.  For this volume of reactor space/
and with a nominal slurry density of 1.15 g/ml/ a feed
rate of 450 ml/hr of cold slurry corresponds to a liquid
hourly space velocity  (LHSV) of about 1 through pre-
heater and dissolver.  Experiments have been reported
with retention time referred to the dissolver  (LHSV of
1.0 = 520 g/hr in most reporting done).  The LHSV values
are omitted in this report for the following reasons.
The relative volumes of the gaseous and liquid phases
actually present in the reactor are difficult to esti-
mate with precision:  the liquid expands when it is
heated; gas bubbles from injected hydrogen displace the
liquid; and, further/ the solution absorbs reducing
gas, reacts with it, and forms various gaseous products.
The "true retention time" is therefore less than the
reciprocal of LHSV by the factor which relates the
relative volume of gaseous and liquid phases.
                         -16-

-------
HYDRQGEM.
 INPUT |
SLURRY
 FEED
                                               1
                                               2
                                               3
                                               4
                                               5
                                               6
                                               7
                                               8
                                               9
                                              10
                                              11
                                              12
                   PREHEATER
                   DISSOLVER
                   REACTOR PRESSURE  INDICATOR
                   STIRRED AUTOCLAVE
                   PRESSURE CONTROLLER
                   GISMO  VALVE
                   3-WAY  MANIFOLD VALVE
                   RECEIVING FLASK
                   SLOP VESSEL
                   COLD WATER CONDENSER
                   KNOCKOUT VESSEL
                   GAS COLLECTION BAG
                                      Veat
                 Figure 4
Reaction and Sampling Systems
                                    -17-

-------
The temperature of the preheater sand bath is measured
by means of thermocouples in the sand and between heaters
and outer wall of the sand bath.  The temperature of
the fluid in the preheater coil can be estimated by means
of a thermocouple in a block at the top of the coil.
Effluent fluid from the preheater flows through a transfer
line to the bottom of the dissolver.

The dissolver is a stainless steel bolted-closure pressure
vessel.  It is cylindrical, with an internal volume of
about 450 ml, approximately 3.8 cm  (1-1/2 inches) in
diameter and 38.5 cm (15.1 inches deep), with a hemi-
spherical end 4.4 cm (1-3/4 inches) in radius.  The upper
end of the cylinder is 4.4 cm  (1-3/4 inches) in diameter
for a distance of 1.72 cm (1/2-inch), tapering to the 3.8
cm (1-1/2-inch) diameter.  A thermocouple well is attached
to the cover of the vessel and extends 30.5 cm (12 inches)
below the cover into the cavity.  Connections at the
bottom of the vessel and in the center of the cover
accomodate standard cone and thread high pressure fittings
In addition, a hole is drilled through the body of the
vessel near the top for attachment of a rupture disc
assembly.  The enlarged hole near the top is within the
flange provided for the bolted closure.  The vessel OD is
6.3 cm (2-9/16 inches), which results in a wall thickness
of 1.33 cm  (17/32 inches) over most of the length.  The
dissolver is heated by means of electrical heaters wound
around the outside, covered by insulation.

Liquid product is taken from the top of the dissolver
vessel through an electrically heated transfer line to
a 2-liter stirred autoclave.  Product is removed contin-
uously through a dip tube reaching the bottom of the
autoclave.  The main function of the autoclave is to
                        -18-

-------
provide a surge capacity to smooth out pressure fluctu-
ations.  Product is removed from the system by allowing
gas and liquid to flow out through a Fisher Governor
"Gismo" control valve.  This valve is air-operated;
it balances the operating pressure in the reactor against
a bellows containing a controlled air pressure.  The
bellows moves the valve stem in and out to open or close
the orifice in the pressure letdown valve.  All product
made must flow through this orifice.  Both valve stem and
orifice in this valve are tungsten carbide, which pre-
cludes rapid wear from ash minerals flowing through the
pressure letdown orifice.  The sample flows from the pres-
sure control valve and is directed to the collection flask
through an air-operated three-way valve or/ if a sample
is not being collected, into a slop receiver instead.
Samples are collected in 3-liter stainless steel flasks
with suitable standard taper joints.  These are connected
in the system by means of stainless steel standard taper
fittings; sample flasks have two necks, each with a
29/42 standard taper joint.  These joints are sealed with
vacuum grease and secured by clamps tightened with thumb-
screws to prevent pressure surges causing joints to leak.
Gas flows out of the flask through a water cooled condenser.
Condensate is collected in the knockout vessel.  Gas is
collected in a rubberized fabric gas collection bag or
vented, as desired.  Generally it is necessary to collect
liquid and gas samples for different lengths of time to
permit the air to be flushed out of the flask.  The gas
sample is taken for a timed interval near the end of the
liquid sampling period.

To prevent materials from contaminating pressure gages
and the pressure control tap of the Gismo valve, the lines
                         -19-

-------
from the reactor to the gages are flushed with nitrogen.
Nitrogen is pumped into the system by means of a Ruska
pump, model 2284, at a constant rate and delivered to
the top of the 2-liter autoclave used as surge vessel.
Thus, pressure for the reactor is measured at this point,
by measuring pressure in the nitrogen purge line which
is at system pressure.  This pressure is controlled by
the setting of the "Gismo" control valve.

In the usual mode of operation the "Gismo" valve is set
at desired system pressure, and all liquid and gas from
the autoclave goes from the pressure of the reaction
system to atmospheric pressure as the sample flows through
this valve.  Both liquid and gas flow down and enter
either the slop vessel or the sample collection flask,
depending on the setting of the air-operated valve.  The
slop vessel, an alternative receiver, is used whenever a
normal product is not being collected, as when collection
flasks are being changed, to receive lineout materials
or off-specification materials, etc.  Gas is disengaged
from the liquid and flows from the flask through a con-
denser to a vent or to a gas sampling bag as required.
Gas is also disengaged from the liquid in the slop re-
ceiver, and this normally flows to the vent.

As liquid and gas flow from dissolver to autoclave,
liquid which accumulates will normally be removed from
this vessel through the dip tube which reaches to the
bottom of the autoclave.  The retention of liquid in
this vessel is therefore limited to the time required for
liquid to drain to the bottom and be collected by the
dip tube.  This vessel is operated at a temperature
                        -20-

-------
well below reaction temperature and so does not contri-
bute reaction time.  The nitrogen from the gage purge
also enters this vessel/ and the gas sample therefore
contains a fraction of nitrogen which is metered into
the system as a consequence of the purge operation.
In general, the autoclave vessel provides a surge capacity
to smooth out pressure surges.  In the case of plug in
the pressure letdown valve, the system pressure increases
gradually.  This arrangement allows time for corrective
action before the safety rupture discs in the system open.

IV-B  REACTOR OPERATION
This section discusses the general conditions of startup,
run, and shutdown.  The functions of reactor components
have been described; the mechanics of their control in
conduct of experiments is beyond the scope of this report.
                                        /•
The time needed to start up and line out  (establish various
steady state conditions before experimental samples are
taken) is considerable.  Consequently, it is not econom-
ical to start up for a single sample.  Furthermore, when
an operating condition is changed, time must be allowed
for the effect of the change to line out; and this delays
sample collection accordingly.

Laboratory experiments vary in length from about one day
to one week.  Laboratory work needed for reactor control
is carried out during the run, and so are gas analyses.
The workup of liquid products may be partially performed
during the run, or may be deferred, according to time
available.  The conduct of experiments with the con-
 tinuous reactor therefore demands a period of intense
 effort which is followed by a considerably longer  period
 for routine workup and analysis of product materials.
                        -21-

-------
IV-B-1  Startup
At startup the reactor has been left filled with flush
oil from the previous experiment.  This expedient avoids
exposing coal and solvent to heat for a prolonged period
while the reactor cools, and hence to a situation in
which partially processed coal or minerals may settle
and plug the preheater tubing.

The first step is to fill the suction reservoir  (figure 2)
with hydrogen and adjust the back pressure regulator to
the desired operating pressure; if necessary, the needle
metering valve is set to deliver the desired volume of
gas per unit time.  These gas metering adjustments, made
by trial, are completed the day before the experiment.

The next step is to turn on the heaters and pumps to circu-
late flush oil while operating temperatures are  adjusted.
During this period the  flow of reducing gas is also
started.  The specified operating temperatures,  flow rates,
and operating pressures are usually obtained within
three hours.

This initial lineout period affords time  to prepare feed
slurry.  Cans of  solvent  are  placed in a  drying  oven, at
75-100° C for anthracene oil,  where crystalline material
is melted.  Partially  used cans  may be kept warm until
used, or reheated and  mixed if allowed to cool.  The same
stock provides  oil for flushing  and for slurry;  in both
uses the oil is thoroughly mixed before any is removed.
A stainless steel beaker  is weighed;  and  oil is  added to
the weight desired, with  a precision  of about 1  g in
total weights of  2-4 kg.  The charge  of powdered coal
is weighed with similar precision, carefully scooped into
the beaker, and blended by  spatula to work out air in the
                        -22-

-------
wetted powder.  The mixing adds  considerable air, which
must be eliminated before the  feed pump can perform re-
liably.  A motor-driven  stirrer  assists in air removal
 (stirring is required  for about  half  an hour at moderate
rates); air is more fully eliminated  when the slurry is
heated above operating temperature and allowed to cool
to the temperature desired.  This temperature has an effect
on the behavior of the feed pump.  The viscosity of the
oil determines the size  opening  between ball check and
seat as the pump operates.  For  anthracene oil, the
slurry pot temperature is above  the point at which oil
may tend to crystallize.  In long runs, makeup batches
of slurry must be added  to the feed vessel with care to
avoid mixing in air bubbles, which is facilitated by
adding before residual slurry  drops to two kilograms,
and by bringing new slurry to  a  matching temperature.

IV-B-2  Run
At the end of the condition lineout,  feed slurry is
substituted for flush  oil.  No product is collected for
some considerable time;  the initial coal solution mixes
with flush oil, and the  back mixing dilutes it progress-
ively; so that several reactor volumes are made before
flush oil is eliminated.  Gas  is vented and lineout
slurry goes to the slop  vessel (figure 4).   (So-called
11 lineout" products during this period on coal solution
may be taken for measurements, described in section
IV-D, which govern the decision  to conclude lineout, or
which indicate a need  to await effects of an operating
change.)   The entire procedure is intended to insure that
a valid sample is collected from reacted feed slurry,
with no admixture of flush oil.
                        -23-

-------
A sample of product solution is collected for a timed
interval.  Toward the end of this interval, perhaps
the last 45 minutes, a gas sample is collected; this
segment is selected to insure that air in the sample
flask has been completely flushed out before product
gas is sampled.  Several sets of operating conditions
may be studied before shutdown, with one or more samples
collected for each.  If products are not taken, the
liquid is diverted to the slop vessel and gas is vented
(figure 4) by means of control valves.
       i         ,
IV-B-3  Shutdown
After the final sample, feed slurry is replaced by flush
oil for shutdown.  Heaters are turned off, but pumping
continues until the reactor cools; when appreciable
reaction can no longer take place, the gas flow is also
stopped.  Pumping continues until the reactor temperature
is below 200°C.

IV-B-4  Concurrent Operations
Throughout the run, the operator records certain data
and makes indicated adjustments.  The most important data
recorded are:
     1.  Feed rate - measured by recording scale readings
                     at 15-minute intervals and computed
                     on an hourly basis
     2.  Temperatures of preheater and dissolver - re-
                     corded continuously and observed by
                     the operator
     3.  Pressure readings across needle metering valve -
                     recorded continuously; the operator
                     confirms that values remain the same
                     as those at the calibration of gas
                     flow
     4.  Start time for liquid sample period, with slurry
                     weight, nitrogen purge pressure and
                     volume, and a verification that
                     temperature and pressure readings
                     are at specified values

                        -24-

-------
     5.  Start time for gas samplfe period
     6.  Stop times for sample periods, with readings for
                     nitrogen pressure and volume,
                     scale reading for feed vessel weight
The operator adjusts feed rate by changing the length of
the piston stroke in the high pressure feed pump.  He
takes a feed slurry sample from the recycle loop at the
beginning and at the end of each liquid sample period.  He
adjusts temperatures by variac and controller settings.
Following each liquid sample collection, the operator
removes the sample flask from the system and stoppers
it; he also drains water from the knockout vessel into a
weighed bottle and weighs it.  The operator takes the gas
sample bag to the laboratory for analysis of the gas and
measurement of the gas volume.

IV-B-5  Laboratory Procedures for Control of_ Feed Slurry
Feed slurry is prepared from materials analyzed to establish
their composition:  a coal ground to a controlled particle
size distribution and a well-characterized solvent, in
accurately weighed amounts.  The average initial compo-
sition is presumably known.  However, coal or coal and
mineral phases settle in the feed vessel, water evaporates
from the feed vessel, and some of the solvent evaporates
also.  To maintain a uniform slurry of correct composition
over an extended period, and to draw this suspension down
to small residual volumes without affecting the average
composition, various procedures are adopted.  The methods
of stirring slurry, .warming the feed vessel, and returning
circulated slurry down the dip tube to reduce the mixing
with air, have been mentioned.  Feed slurry composition is
monitored by periodic ash analysis and by occasionally drawing
samples for water determination if the feed coal contains a
significant amount  (excess of 2 or 3%) of water.
                        -25-

-------
To estimate the amount of coal in a slurry, a weighed
sample is taken from the slurry recycle loop, ashed for
a measure of mineral content, and compared with the ash
in feed coal.  For a 2:1 solvent to coal ratio, ash from
a slurry sample should be 1/3 the concentration in feed
coal if the suspension is good.  When stirring is marginal
samples tend to increase in ash as slurry is used up, or
to run below theory if taken from a full pot.

The Karl Fischer method for determination of water is
adapted for use in analyzing feed slurry for water.
Slurry is sampled by inserting a small bottle in the re-
turn stream to collect 10-20 g from the slurry recycle
loop.  Feed slurry is also inspected for density; it has
also been tested for evaporation rates in the light oils.
All of these procedures for control of feed slurry are
detailed in OCR Interim Report No. 7.

IV-B-6  Lineout Control
One general problem in experiments with the reactor is
to develop methods which establish the reproducibility
of the product, methods which can indicate when the pro-
duct is equilibrated with feeds and with the effect of
reaction conditions.  Four such methods are mentioned here
details of the analyses are given in section IV-D along
with product analyses.

In early studies it was observed that the infrared spectru
of a coal solution had several bands whose relative in-
tensity had altered from similar bands in the initial
solvent.  The absorbance ratio*of bands at 3.28 microns
and 3.42 microns changed progressively after the switch
from flush solvent to feed slurry.  For comparison of
                        -26-

-------
samples, the ratio of absorbance for the 3.42 micron
band divided by the absorbance at the 3.28 micron band
is computed.  The ratio, called IR hereafter, is plotted
as a function of operating time.  The IR ratio  for raw
anthracene oil is smaller in the initial flush  oil than
in the lined out sample, which has been hydrogenated
during the reaction and also had hydrogen components
added from the dissolving coal.  Because the mineral
phase in the coal exerts a catalytic influence  on pro-
duct composition, the settling of mineral in the reactor
causes long swings in reactivity; minerals may  settle,
then classify by gravity separation processes.  In this
condition the reactivity oscillates periodically; a lined
out reactor produces slightly different composition
products from time to time because of this mechanism.

A "blackness" test is also applied to coal solutions.
The absorbance of light by coal solutions in pyridine,
when observed by a spectrophotometer, provides  a rapid
method for measuring a property of a coal solution which
relates to the concentration and reduction state of coal
dissolved in the solution.  The color can be related  to
the absorbance by a colorimetric factor to yield this
concentration.  Within observation limits, it is a
sensitive index of the state of the product in  the reactor.

Ash analysis is a third method of observing lineout.  If
the concentration of mineral in the coal solution should
account for all of the mineral fed into the reactor,  pre-
 sumably the product solution would be fully lined out.
But sampling is difficult since mineral settles, and  re-
 suspending it for a representative sample is awkward. If
 ani entire product sample is filtered and the residue
 ignited to determine the ash present, the finest mineral
matter will have passed through even fine analytical  paper,
                        -27-

-------
With small samples, it is preferable to ash the whole
sample.  An alternative procedure is to collect a reason-
able volume of sample, warm it, and stir it properly
while subsamples are removed for ash analysis; this is the
same analysis as is used for slurry feed.  Precise dup-
licate values are reasonable evidence of reliability in
the samples ashed.  The buildup of mineral in the product
solution, as noted with the IR measurements, is influenced
by the tendency of mineral to settle in the dissolver;
and times required for equilibration of mineral may
differ readily from time required for equilibration of
the organic phases.

Assuming that iron is catalytic, the concentration of iron
must reach steady state in the dissolver before the re-
actor can be considered completely lined out.  An underr-
standing of mineral behavior appears essential to developing
the most reliable results possible.  In run CU 48, the,
first sample was taken after a normal lineout interval,
the second after an  extended lineout interval.  Changes
between samples could indicate the effect of changes in
mineral concentration or composition in the dissolver.

The composition of gaseous products is a good method of
observing approaching lineout.  The appearance of hydrogen
sulfide especially and its change as a function of
operating time follow a repeatable pattern.  However, gas
composition does not relate directly to the amount of
pyrite fed, the amount of organic sulfur fed, or the amount
of conversion produced.  This is because the hydrogen
sulfide and carbon dioxide react with traces of ammonia
formed and then dissolve in the water condensate.  The
equilibrium is temperature-dependent and influenced
appreciably by minor variations in ammonia yield.  These
                        -28-

-------
characteristics differ with flash vessels operated at
different temperature and pressure conditions.  The
relations between gas composition, operating conditions,
and yield should be more direct in flash vessels operating
at higher temperatures and pressures.

Such observations account for some of the difficulty in
making complete element-by-element balances; the time
required to line out the composition of any given com-
ponent needs to be satisfied in samples taken.

IV-C  WORKUP OF COAL SOLUTION
Along with the gas sample, two others are taken from the
reactor  to account for all of the feed  slurry and hydro-
gen fed  in.  The  smaller of  these, perhaps  0.1% of  feed
slurry,  is water  recovered from  the  knockout  vessel be-
tween the liquid  sample  flask and the gas collection bag,
water which  has condensed in the gas line.  Because the
weight of sample  is  so  small, this product  is normally
reported in  the water yield, which is  largely from the
 stripping operation.   In fact,  the knockout vessel water
 contains traces  of  oil,  as  well  as ammonium carbonate,
 ammonium sulfide, and a small amount of phenol.   A
 refinement  of yield results by analysis of  this fraction
 is possible, but the influence on overall results would
 be very small.

 This section discusses methods used for evaluation of the
 liquid product sample.  These are separation procedures,
 conditioned by handling losses which must be accounted in
 order to afford a handling-loss-free product distribution
 within  the sample.  These methods are  typical and were
 employed during  the experiment reported in section V.
                         -29-

-------
However, since that time, the modification of equipment
and the increase in sample size available from the
reactor, refined handling techniques, refined sampling
techniques, and accumulated experience, have all reduced
the size of corrections required and are a real consid-
eration in reporting results.

Before workup, the coal solution has been collected in a
tared, 3-liter, stainless steel, round bottom flask.
Flask and contents have been weighed.  The first step
in workup of the product solution in the laboratory
version of the SRC process (figure 5) is stripping it
of water and light oils.

IV-C-1  Stripping
The flask of coal solution is attached to the stripping
apparatus shown in figure 6.  Stripping is a preliminary
vacuum distillation designed to permit recovery of the
product without uncontrolled loss of volatile substances
in the filtration step which follows.  The flask is
warmed while vacuum is cautiously applied to preclude
bumping or foaming.

The more volatile fraction passing through the condenser
as vapor and condensed at dry ice temperature is the
cold trap product.  The fraction condensed by the water
cooled condenser  in CU 48 was 0.1 g per sample,  included
with the cold trap oil.  Both products contain both
oil and water.  The oil and water phases are separated
after stripping is stopped at a head temperature of
50°C  (later, 100*C, or adjusted as necessary to obtain
a balance yield of solvent for recycle).

The vacuum reached by the end of the distillation is ^
 3 mm  Hg.  The weights of water  and cold trap oil  are
 recorded.
                         -30-

-------
         UNFILTERED COAL SOLUTION
         FROM REACTOR
                   I
         VACUUM STRIPPING (50°C)
                           WATER
                           COLD TRAP  OIL
         STRIPPED COAL SOLUTION
         VACUUM FILTRATION
WET FILTER CAKE
         \
    FILTRATE
       J,
 PYRIDINE
 EXTRACTION
VACUUM DISTILLATION
 CUT 1 OIL (50-100°C)
. CUT 2 OIL (100-230°C)
»HEAVY OIL (230-270°C)
 VACUUM BOTTOMS
                   EXTRACT (Discard)
 PYRIDINE INSOLUBLES
 (For analysis)
                            Figure 5  Workup
                                  -31-

-------
to
 I
VIGREAU'X.
DISTILLING.
COLUMN
                                               THERMOMETER
                     HEATING
                     MANTLE   i*
                                                                WATER COOLED-
                                                                - CONDENSER
                                                 3-LITER
                                                 STAINLESS STEEL
                                                 FLASK  '
                                                      & DISTILLING RECEIVER
                                                      K
  :• (Tb vacuum)

COLD TRAP
                                                                                                                    DRY ICE BATH
                                                    Figure  6 'Stripping Apparatus

-------
IV-C-2  Filtration
The flask is removed from the stripping apparatus and
weighed before filtration.  To afford quantitative re-
sults, all equipment components  (figure 7) for this
step are separately weighed:  Whatman #4 filter paper,
steam-heated Buchner funnel, vacuum adapter which directs
the filtrate, and round bottom receiver flask containing
five small boiling stones*

Sometimes a measured amount of cut 1 oil is applied to
the filter paper, enough to hold it in place over funnel
orifices.  This procedure has been adopted because solu-
tion occasionally passed under the paper when dry.  Some-
times also a cold trap is inserted in the vacuum line to
condense volatiles which escape  the filtrate.  Since the
condensate is less than 1 g, this practice was later
abandoned.  However, during CU 48, the collection of cold
trap condensate was made.

The hot stripped solution is poured into the funnel, under
a vacuum of approximately 90 mm  Hg.  When drained, the
collection flask is reweighed for a 'record of material
filtered and material held up by walls of the flask.

The products are a filtrate and  a wet  filter cake.  The
receiver flask is weighed again  to obtain weight of
filtrate.  The cake, perhaps 0.32 to 0.64 cm  (1/8 to
1/4 inch thick), is removed from the funnel and weighed
as a unit with the filter paper;  usually  these make a
compact unit which can be handled without difficulty.
Finally, the funnel and  associated glassware are weighed
to establish transfer losses.   (In later work, the funnel
has not been weighed nor cleaned since a  fairly constant
holdup in the funnel is  demonstrated;  however, small
                         -33-

-------
                            KJCHNJiR fUNSEL
                            WHATKAN #4
                            FILTER PAPER
                               (To  cold  trap
                         y    and vacuu:a)
                 VACUUM ADAPTJbR
               2-LITER
               KLC^IVER FLASK
Figure 7  Filtration Equipment
             -34-

-------
variations in holdup occasionally give an output weight
more than input weight.)  As a check, the weights of wet
filter cake, filtrate, and transfer losses are summed for
comparison with the weight of stripped liquid transferred
from the original collection flask.

Wet filter cake is removed from the paper and bottled
for further workup.

IV-C-3  Distillation for Solvent Recovery
The filtrate contains both solvent and vacuum bottoms,
or solvent refined coal.  These are separated by distill-
ation with equipment used for stripping  (figure 6).  Dis-
tillate is collected in a 1- or 2-liter  flask.  For experi-
ment CU 48, "cut 1" oil is that fraction distilling be-
tween 50 and 100° at less than 3mm Hg; solvent  ("cut 2")
is the cut distilling between 100 and 230°C at less than
3mm Hg.  The use of a fraction collector permits changing
receivers without interrupting distillation.  Heavy oil,
during CU 48, was collected between 230  and 270°C; later,
heavy oil was allowed to remain with the vacuum bottoms,
although the fraction distilling between 230 and 250°C
was included with cut 2 oil.   (Cut points in many experi-
ments were arbitrary and might be conditioned by the need
to obtain a breakeven quantity of solvent.)

In this second distillation, the cold trap also condenses
about 1 g of oil, which is included with cut 1 oil in
weighing products.  Generally, the weight of input com-
pares well with that of products.

The weights of distillates and residue are obtained by
weighing all flasks before and after distillation.  As
a check, the weights of vacuum bottoms,  heavy oil,
                        -35-

-------
cut 2 oil, and cut 1 oil are summed and compared with
the original weight of sample taken.

Typically, the second distillation is started with a flow
of cool water through the condenser.  As vapor temperature
increases, the distillate tends to solidify.  At this point
the water flow is cut off and, if necessary, the condenser
is drained.  Toward the end of distillation, wax-like
products often appear in the condenser or. fraction collector.
To make the distillate flow into the receiver flask, a
blast of hot air is directed onto the area affected by
an electric heat gun.

Before sampling, cut 2 oil is warmed and mixed to homo-
genize it.  Distillation residue, or vacuum bottoms, is
poured while molten into a metal pan, cooled to solidify
it, and then broken up and bottled.  In CU 48, however,
vacuum bottoms were allowed to solidify in the flask,
then broken out.  This product is carefully ground in a
mortar, sieved to remove boiling stones  which are necessary
to prevent bumping during distillation and thoroughly mixed be-
fore analysis.

IV-C-4  Workup of_ Wet Filter Cake
An aliquot of wet filter cake is taken for analysis.
The remainder is sampled for determination of the amount
of soluble and insoluble material present.

The insoluble fraction was determined by extraction with
pyridine; the insoluble residue was rinsed with benzene,
then with acetone, and then dried.  The dry residue was
weighed.
                         -36-

-------
The soluble material was assumed to be the same compo-
sition as filtrate.  The insoluble residue is further
analyzed by ash analysis for determination of MAP conver-
sion; for CU 48 the insoluble residue was also analyzed
completely to obtain C, H, N, Sf and ash values.

IV-D  ANALYTICAL METHODS
This section describes the analytical techniques applied
to feed slurry, product solution, and the gaseous, liquid,
and solid products of continuous reactor operation.  A
summary of these methods  for  the products is  shown in
table 1; standard  procedures, such as those for deter-
mining basic or acidic functions in water, are not de-
tailed here.

IV-D-1  Analysis of Gaseous Products
Toward the end of  each sample period in a run, gases
are collected for  a timed interval, usually 45 minutes,
in a rubberized fabric gas bag.

First, an infrared spectrum is  obtained for each  gas
sample as a qualitative indication of the gas compo-
sition.  The quantitative distribution is obtained
usually by gas solid  chromatography.

Spectra are obtained  with a Perkin-Elmer  267  Grating
Infrared Spectrophotometer.   Either a 5-cm or 10-cm gas
cell with sodium chloride windows  is used; the smaller
cell gives more satisfactory  results.  The cell is
flushed with sample gas and then filled with  sample gas
to ambient pressure.   In  very humid weather Irtran-2
windows are used to overcome  the problem  of water con-
densing on the windows, but their  lower transparency
                         -37-

-------
   Table 1.  PRODUCT DISTRIBUTION AND ANALYTICAL METHODS
 PRODUCT
 Gaseous
Liquid
 Solid
 FRACTION ISOLATED
1.  Untreated gas
    Cold trap oil
    and water
               Cold trap oil
           3.  Cut 1 oil
               (condensate IBP
               to 100°C @ 3mm
               Hg)

           4.  Cut 2 oil
               (Condensate IBP
               100 to 230°C @
               3 mm Hg)
5.  Heavy oil
    (230-270°C @
    3 mm Hg)

1.  Vacuum distilla-
    tion residue
    (vacuum bottoms)
            2.  Wet filter cake
 ANALYSIS AVAILABLE
                                 a.
                                 b.
                                 c.
                                 d.

                                 e.

                                 f .
a.
b.
c .
d.

a.
b.
c.
d.

a.
b.
c.
d.

a.
b.
c.
d.
e.

a.
b.
                                 b
                                 c
                                 d
                                 e
                      a
                      b
IR spectrum

Molecular weight
Composition by gas
chromatography
H_, % (estimated from
Mw correlation)
Water (trapped when
gas sample is vented)

Total sulfur, %
Basic functions
Acidic functions
Waste water analyses

Density
IR spectrum
Elemental analysis
Functional groups

Density
IR spectrum
Elemental analysis
Functional groups

Density
IR spectrum
Elemental analysis
Total nitrogen
Functional groups

IR spectrum
Elemental Analysis


Fusion point  (gradient
bar)
Elemental analysis
Ash, % and composition
Coking properties
Potential utility
specifications

% pyridine insolubles
% ash in pyridine
insolubles
Potential utility
specifications
                         -38-

-------
requires increased gain settings and gives less satisfactory
results.  A typical gas product spectrum is shown in figure 8.
The areas of interest include the hydrocarbon absorption in
the 2800-3200 cm"  region/ the carbon dioxide bands 2340-
2360 cm  , and the carbon monoxide bands around 2120-2170
cm  .   Concentrations of methane, carbon dioxide, and carbon
monoxide can be estimated, when desired, from the spectrum.
In some cases, small amounts of ammonia can be detected by
absorption in the region of 900-1000 cm"1.

Second, the volume of hydrogen  sulfide  is established by
withdrawing a sample, such as 500 ml, from the gas bag with
a syringe, bubbling  it  through  an ammoniacal zinc sulfate
solution  to absorb H2S, and determining the amount iodometri-
cally.  The procedure is  detailed in appendix A.

Molecular weight of  the product gas is  determined by the gas
density method; a worksheet is  illustrated in appendix B.
A gas density bulb of thin wall design  (figure 9) is used;
it  can  be evacuated  and weighed repeatedly within 0.3 rog on
a semi-microbalance.  The bulb  is evacuated to less than 1
mm  Hg and weighed.   The stopcock is opened to fill the bulb
with air, and the bulb  is reweighed.  It  is then reevacuated,
reweighed, and  filled with product gas  and weighed again.
The weight of gas and weight of air are determined in trip-
licate.   The  ratio of gas weight to air weight times 28.95,
the calculated  molecular  weight for air,  provides the mol-
ecular  weight reported  for product gas.

Fourth, the gas composition is  generally  analyzed by gas
solid chromatography.   The development  of this procedure has
been presented  in OCR Interim Report No.  6; conditions are
listed  in appendix C.   This analysis provides concentrations
of  methane, ethane,  propane, n-butane,  iso-butane, carbon
monoxide, carbon dioxide, hydrogen, nitrogen, and oxygen.
                           -39-

-------
 PAGE NOT
AVAILABLE
DIGITALLY

-------
            Figure 9  Gas Density Bulb

Water concentration in the output gas is calculated from
the weight condensed in the cold trap when gas bag contents
are vented later.  Results are normalized to 100% by
making any required small correction  (under 3%) to the
hydrogen content.  Any oxygen present is presumed to be
from air contamination during sampling; therefore, results
are corrected for the oxygen content, usually under 0.4%;
and the amount of nitrogen which would be associated with
this level of oxygen is determined.  The results are then
reported on a normalized basis.  Output gas compositons
as  determined by gas chromatography  for CU 48 are given
in section V.
                           -41-

-------
Finally, for mass balance information, the volume of gas
collected in the timed interval is determined.  Gas remaining
in the gas bag is vented through a trap cooled by dry ice to
condense water, which is determined by the weight gain in
the cold trap, and then through a wet test meter.  The gas
volumes used in analysis are added to this vented amount
to give the total volume which has been collected.

This total may be converted to moles of product gas by
multiplying volume of gas in liters by the molar equivalent
of one liter of gas at the temperature and pressure of the
vented gas.  Molar equivalents for one liter of gas in tabular
form are shown in appendix A; they may also be calculated
from the equation
                                          1
             moles/liter = ^  X ^ X ^^40
where P is pressure  in mm of Hg and  T is temperature in
degrees absolute  (=  °C + 273).  The  temperature used to
establish total gas  volume is that from the wet test meter
thermometer; the pressure used  is ambient barometric,
corrected for the vapor pressure of  water at the wet test
meter temperature.   The moles of product gas for the timed
interval are then converted to  a moles-per-hour figure.  The
weight of product gas is calculated  simply by multiplying
moles by molecular weight.

In the mass balance  report for  CU 48 (section V), total gas
output calculated as just described  includes unreacted
hydrogen, but not nitrogen from the  nitrogen purge; in the
yields reported, the yield of gas does not include nitrogen,
water, or unreacted  hydrogen, but only gases arising from
the process  (hydrocarbons, hydrogen  sulfide, carbon monoxide,
                            -42-

-------
and carbon dioxide), or total gas weight less hydrogen and
nitrogen.  Yields of individual gases are not usually re-
ported but are obtained when desired, as with CU 48, from
output gas compositions.  Examination of product gas com-
positions shows the bulk of gas yield to be hydrocarbons.

A reasonably good correlation is found between hydrogen
concentration in product gas and product gas molecular
weight, for the range of conditions of interest here.
Hydrogen output in moles is readily calculated from moles
of output gas and mole % hydrogen, which is estimated
from the molecular weight.  Hydrogen consumption is hydrogen
in, which is calculated by the method described in section
IV-A-1, less hydrogen out.  Hydrogen consumption on a
moles-per-hour basis is converted to a weight % consumption
based on slurry and on raw feed coal.

IV-D-2  Analysis of_ Feed Slurry and Liquid Products
(a)  Feed Slurry —
Slurry is sampled by inserting a small weighed bottle or
dish in the return stream from the recycle loop to the feed
slurry vessel.  Two determinations are routinely made, one
for ash and one for water.  The ash determination is dis-
cussed in section IV-B-6 as a means of establishing lineout.
The method is as follows:  a weighed platinum evaporation
dish is passed rapidly throught the return stream of
slurry to collect 5-10 g.  Dish and sample are reweighed
on an analytical balance to determine exact weight of
sample.  The dish is placed on a sand bath which is heated
by low flame from a Fisher burner; oil can be evaporated
without ignition, although burning the oil does not seem
to introduce significant error.  After most of the volatiles
                            -43-

-------
have been removed, the dish is placed just inside the door
of a muffle furnace adjusted to operate at 750°C.  At this
position most residual oil and organic matter in the coal
burn off without igniting.  Shortly the dish can be moved
to the back of the furnace and residual ash ignited.  If
a deep ash layer is observed, the ash is stirred with a
platinum wire to surface any 'unburned carbonaceous matter.
The percentage of ash in feed slurry is reported; for
a 2:1 solvent-coal ratio,  the % ash in a well suspended
slurry will be 1/3 of the % ash in the feed coal.

Water content in feed slurry is determined by adaptation
of the Karl Fischer method.  The whole slurry sample taken
is used for this analysis because the water settles out
rather quickly.  The tared bottle and sample are weighed
accurately on an analytical balance. The sample is trans-
ferred, through a powder funnel into a dry 200-ml volu-
metric flask placed in the hood, by use of a stream of dry
pyridine from a wash bottle.  The volume of solution is then
brought to the mark by adding dry pyridine.  The flask is
stoppered, and the solution thoroughly mixed and allowed to
stand for an hour or so to extract water from the coal.
(This procedure seems to remove water effectively from
lignite and subbituminous coal, as well as water which may
be added to solvent.)  Water content varies appreciably in
the various grades of reagent grade pyridine; so the grade
with lowest water content is chosen when possible.

The following adaptation of the Karl Fischer method for
titration of water uses a dead stop indicator and commer-
cially available Karl Fischer reagent.  This reagent is put
into an automatic buret of 25 ml capacity; the titration
                           -44-

-------
vessel is stirred by a magnetically driven stirring bar.

Coal slurry is diluted with reagent grade pyridine, and an

aliquot of this solution is titrated.  The procedure is
scaled for slurries containing about 10% water; changes in

detail are needed for different water concentrations or
for different sample sizes.
     Apparatus:  Automatic buret, 25 ml capacity with
                 1-liter reservoir
                                   1
                 dead stop indicator and platinum elec-
                 trodes

                 magnetic stirrer and Teflon-coated
                 stirring bars

                 3-neck standard taper flask,  200 ml,
                 round bottom  (or equivalent airtight
                 titration vessel)

     Reagents:   Karl Fischer  reagent  (Fisher  Scientific
                 Co.f SDK 3)

                 Karl Fischer  diluent  (Fisher  Scientific
                 Co., SOK  5)

                 reagent grade pyridine  (low water content,
                 or dried over mol sieve)

     Procedure:

     a.  Weigh clean dry sample bottle with cap which will
         contain 20 ml of liquid.  Record weight.  With slurry
         pump running, collect 10-20 g of slurry.  Cap
         bottle and reweigh to determine amount of sample
         taken.

     b.  Dilute slurry with dry reagent grade  pyridine and
         transfer quantitatively to 200 ml volumetric flask.
         Fill to the mark with dry pyridine and stopper
         flask.  Mix contents of flask thoroughly and
         allow flask to stand, with occasional remixing for
         about an hour.  Allow powdered coal to settle out
         after final remixing.

     c.  Put about 75 ml of Karl Fischer diluent in titra-
         tion vessel.  Titrate until microammeter of dead
         stop indicator reads  a constant 50-60 microampere
         current.  Pipet 25 ml of dry pyridine into titra-
         tion vessel, and titrate with Karl Fischer reagent
                           -45-

-------
          to  establish a blank for pyridine.   The end
          point is taken when the same indicator current
          is  obtained as originally used to pre-titrate
          the solvent.

      d.   Use 10 ml pipet to transfer solution from 200
          ml  volumetric flask to 100 ml volumetric flask.
          Dilute with dry pyridine and adjust to the mark.
          Mix thoroughly.

      e.   Transfer 10 ml of solution from second flask to
          titration vessel.  Titrate to same indicator
          current used to pre-titrate the solvent.  De-
          pending on exact water content, the aliquot which
          can be titrated with one buret of reagent may
          allow use of a larger pipet.  With a titer of
          1.0 mg of water per ml, the 10 ml pipet is a
          suitable volume, assuming the, original sample
          contained 10% water.

      f.   The reagent blank must be calculated in proportion
          to the sample taken for titration, compared to
          the 25 ml portion of dry pyridine used in the blank
          determination.  Since the same pyridine is used
          for all dilutions, its contribution to the water
          content remains constant in concentration.  Thus
          the water content of the solvent is proportional
          to the sample taken.

Use the following procedure for standardization of the re-
agent :

      Weigh a 200 ml volumetric flask on an analytical
      balance.  Add about 250 mg of distilled water.
      Stopper the flask and reweigh to determine the
      exact weight of water taken.  Dilute to the mark with
      dry pyridine and mix thoroughly.  Titrate 10 or 15
      ml aliquots of this solution, adding the solution
      to pretitrated solvent as described in the deter-
      mination above.  Calculate the titer in milligrams
      of water per ml of reagent.
   Titer
               (mg of water taken)  (ml of sample)
                               200
               (ml of reagent - ml of blank)

Calculate the sample as follows:


     %H20  =    (ml of reagent - ml of blank) (titer)(100)
                       sample size in mg
                            -46-

-------
Note:  Sample size is
                              Total  Sample  (V^ (V£)
                                             ~
                                      (F-) (F-)

     where ^ and V2are pipet volumes taken  and F. and F
     are the sizes of the volumetric flask used in the
     dulution sequence .

Since it is necessary to titrate a 25 ml portion of the pyri-
dine to determine the water blank for the reagent, the pyri-
dine must be dried for any amount of reagent more than a
few milliliters.  The most convenient means  is to pass the
pyridine through a column of molecular  sieve.  A suitable
apparatus for this procedure consists of a  76 -cm  (30-inch)
length of 2.54-cm (1-inch) ID tubing sealed  to a flask
which will hold 2 liters of liquid.   The lower end of the
column is closed by a Teflon stopcock and filled with
Linde type 4A molecular sieve.  The  reservoir is filled with
reagent grade pyridine, and the stopcock adjusted to per-
mit pyridine to trickle through at about 5 ml a minute; dry
pyridine is collected in a clean dry reagent bottle.  A
column of well activated molecular sieve of  this size will
process a 3.7-liter  (8-pint) bottle. The reagent blank for the
dry material should be only 2-3 ml of Karl Fischer reagent
for 25 ml of dry pyridine.

The slurry sample to be titrated is  dark colored, which
necessitates use of an electrical device to  locate the
end point; and for this the conventional dead stop indi-
cator is satisfactory.  In the procedure given, the orig-
inal sample is diluted in a volumetric  flask and mixed
before an aliquot is taken for dilution in a second volu-
metric flask; obviously various combinations of aliquot and
volumetric flask volumes may be used to obtain suitable
                            -47-

-------
samples for titration.  To draw samples within narrow weight
ranges is difficult, and variation in sample size must be
compensated by the dilution sequence or variation in the
amount finally taken for titration.  In a small pipet, pow-
dered coal may plug the tip; precision-is improved by using
larger volumes.

(b)  Product Solution —
The unstripped coal solution (liquid sample) is monitored
during the run.  The infrared spectrum is measured, as for
other oils (section IV-D-2-d).   A weighed sample of solution
is also examined by visible spectrophotometry for a quality
called "blackness."

The sample is diluted to a measured volume with pyridine,
thoroughly mixed, and filtered.  The absorbance of the
filtered solution is read in a 1.72 cm (1/2-inch) cell
with pyridine in a similar cell for the reference, in a
Bausch and Lomb Spectronic 20 Colorimeter.    Since the
relative intensity of various wavelengths does not change
greatly from solution to solution and absorbance in the
400 to 700 nanometer region declines along a smooth curve,
the absorbance at 550 nanometers was arbitrarily chosen
for one-point comparisons.
     blackness = absorbance per gram of coal solution per
                 100 ml, measured in 1/2-inch cell at 550
                 nanometers
Blackness and IR in combination have been correlated with
vacuum bottoms concentration, but the relationship is com-
plex and not fully understood.   As a sensitive and repro-
ducible measurement, blackness is useful to indicate
whether a reaction is lined out or not.  Blackness values
exhibit a continuing drift until lineout.  As all compo-
sitions reach steady state, blackness values become stabil-
ized but exhibit small changes in response to mineral
                           -48-

-------
changes in reactor drifts.  This oscillation may be caused
by temperature, pressure, feed rate, or catalytic changes
within the limits of reproducibility of the controls.

 (c)  Water Phases —
Liquid products are categorized as water phases and oil
phases.  Most water is isolated in the cold trap during
stripping; some condenses and forms a separable phase of
cut 1 oil, and a much smaller amount is obtained from the
knockout vessel.  Water phases are usually combined for
analysis; the composite phase is mostly water but does
contain sulfide, carbonate, ammonia, and small amounts of
phenol (1/2 to 1%).

As water samples age, they turn yellow, indicating the
formation of ammonium polysulfide.  When yellow solutions
are acidified, a whitish precipitate of colloidal sulfur
is commonly observed.  Sulfur in the water phase is deter-
mined as follows.  Sodium hydroxide is added to make it
alkaline, and hydrogen peroxide is added to insure oxi-
dation of all sulfur to sulfate.  The solution is boiled
to remove excess hydrogen peroxide and acidified with di-
lute hydrochloric acid.  Sulfate is precipitated by
barium chloride.

Ammonia in the water is determined by adding magnesium
oxide to a weighed sample in a Kjeldahl flask, distilling
the ammonia out, and titrating by the standard Kjeldahl
method.  By subtracting an assumed amount of phenol, the
ammonia, sulfur, and hydrogen are then calculated from the
remaining water.
                           -49-

-------
Water trapped from the gas bag and the knockout vessel is
usually too small for analyses.

Phenol can be estimated by ultraviolet spectroscopy of
acidified water phases.

(d)  Light Oils —
Laboratory workup procedures provide two light oils/ cold
trap oil and cut 1 oil; both are byproducts of the SRC
process.  As they are removed from the process, a thorough
investigation of their composition is not usual.  The cut
1 oil in present practice is that cut boiling below 100°C
at <3 mm Hg during stripping/ condensed in the water
cooled condenser; during CU 48/ however/ cut 1 oil was
obtained as the fraction boiling below 100°C at ^3 mm
Hg at distillation after filtration.  Cold trap oil is
collected at -78°C under the same vacuum at stripping.
The boiling range of cut 1 oil is within the boiling range
of kerosene; the boiling range of cold trap oil is within
the boiling range of gasoline.

The density of these oils is measured by weighing in 10 ml
volumetric flasks.  Their IR spectra are analyzed and their
elemental compositions (C, H, S, and N) are established by
the methods outlined below for cut 2 oil.  However/ in the
carbon and hydrogen determination of light oils, because
of their volatility/ the samples are placed in weighed
quartz glass capsules instead of the open boat and are
distilled more slowly to prevent explosion.

For all oil samples, IR spectra are obtained by either of
two methods.  By the neat sample method/ a few drops of oil
are placed on a rock salt plate, covered by a similar plate,
                            -50-

-------
 and compressed by a brass  holder.   The  oil  film must be
 of  reasonably uniform thickness, without water droplets or
 air bubbles.   The spectrum is  obtained  with the spectro-
 photometer used in obtaining the IR spectrum for gas
 (section IV-D-1) .  Alternatively,  the spectrum may  be
 obtained from the sample in solution:   a known weight  of
 oil, such as  2.000 g, is transferred to a  100-ml volu-
 metric flask  and diluted to the mark by carbon disulfide
 for a weight/volume concentration  of 2.000% (for cold
 trap oil, a weight/volume  solution of 1.000% is better).
 A rock salt cell of 0.5 mm thickness is filled with the
 solution; this and a matching  reference cell containing
 only carbon disulfide are  placed  in the spectrometer.
 Representative IR spectra  for  cut  2 oil and other mater-
 ials for CU 48 are illustrated in  appendixes E through I.

 A quantity termed the IR ratio is  determined by a method
 described in  the Division  of Fuel  Chemistry Preprints,
 American Chemical Society  (vol.16, No.  2,  pages 68-72).
 By  the general techniques  of quantitative  analysis  as
 presented in  ASTM E 168, a baseline is  drawn from the
 vicinity of 3800 cm'1 to the vicinity of 2000 era" . The
 absorbance of the bands at 2920 cm" and 3040 cm" are
 measured, and the ratio
                 Absorbance at 2920
                 Absorbance at 3050 cm
is calculated.  (The IR symbol denotes this ratio, and not
the inverse, which is the method of the reference.)  This
ratio serves for ready comparison of samples.

The IR is useful in evaluating variables in reactor oper-
ating conditions.  If the coal solution is prepared with raw
anthracene oil, the initial TR value may be about 0.56;
                            -51-

-------
as coal derived material dilutes and replaces the initial
solvent, the value rises to approximately 0.75.  If the coal
solution is prepared with a well hydrogenated solvent, the
initial IR may be higher than the value at lineout, partic-
ularly if solutions are made at high operating temperatures
and at short retention time in the reactor.  The IR values
may be related to the hydrogen transfer capacity of the
solvent; but as coal derived material is added to the system,
a significant amount of saturated structures is introduced,
and these interfere  with the observation of transfer re-
actions.  With only IR spectra as a basis for analysis, it
is not possible to resolve the hydroaromatic contributions
from the contribution of the saturated functions.

TR values have a separate use after reactor lineout.  The
mineral phase in the coal exerts a catalytic influence on pro-
duct composition.  The settling of mineral in the reactor
causes long swings in reactivity as minerals settle, then
classify by gravity separation processes.  In this condition
the reactivity oscillates periodically, and a lined out
reactor produces material of slightly different composition
from time to time in response to this mechanism.  In solvent
recycle studies, the solvent composition is progressively
altered by the addition of oil from the coal and by the
thermal decomposition of the original solvent.  The obser-
vation of TR values is useful in recording these long term
changes.

 (e)  Cut 2_ Oil ~
Cut 2 oil is usually that fraction obtained from distilling
filtrate between 100 and 230°C at ^3 mm Hg.  When stripping
has been stopped at 50°C, the fraction of filtrate distilling
between 50 and 100°C is "cut 1" oil and is sometimes included
                            -52-

-------
with cut 2 oil for analysis.  "Heavy oil" is the fraction
distilling between 230 and 270°C for this report; in later
practice oil distilling between 230 and 250°C has been in-
cluded with cut 2 oil, and that distilling above 250°C has
been allowed to remain with vacuum bottoms.  These cut points
will be adjusted in plant operation to obtain the volume
of solvent needed for balanced recycle of new slurry.

Besides the IR ratio obtained for cut 2 oil, as described
above, this sample is weighed in a 10-ml volumetric flask
to establish its density; and its elemental composition
is determined as follows.

For the most part, the methods outlined in ASTM D 271 are
employed for the carbon, hydrogen, nitrogen, and sulfur
determinations.  For C and H, the macro combustion train
is modified by the substitution of lead peroxide, operated
at 185°C, for lead chromate.  in addition, a plug of silver
screen is used just in front of the copper oxide filling.
For increased precision and accuracy, sample weights are
increased from the usual 200 mg to 400-500 mg, a modifi-
cation adding time to the analysis but justified in the
improved precision of results for hydrogen.  When light
oils are analyzed, their volatility produces a risk of fc-rming
explosive mixtures, with the chance of blowing the stopper
from the combustion tube; consequently, the liquid is placed
in a small Vycor tube for weighing and evaporating it in the
combustion tube.  The higher boiling oils are weighed in con-
ventional Alundum boats of the type used for steel analysis,
or in platinum boats; samples burn without difficulty.
                            -53-

-------
Total nitrogen content was routinely determined in oil
samples; for run CU 48, the detailed elemental results ob-
tined through the Kjeldahl-ASTM method are reported in section
V.  Basic nitrogen is obtained by the procedure detailed in
appendix D and briefly described here.

For basic nitrogen determination, a 2-g sample of cut 2 oil is
weighed in a beaker, to which is added 150 ml acetic acid.
Titration equipment consists of:
     A Sage Instruments Syringe Pump, model  341, which controls
            flow rate of titrant solution  (0.1 N acetous
            perchloric acid) into the covered beaker; the
            beaker is set on
     A Labine Magnestir, Cat. No. 1250; electrodes are
            connected to
     A Leeds and Northrup pH meter, set at 1400 mv, connected
            to
     A Sargent Recorder, model SR, set for a chart speed of
            2.54 cm (1 inch) per minute
The resulting chart is examined for its inflection point, that
point in the middle of the break in the curve where change of
slope occurs.  The shape of the curve resulting for basic
nitrogen determination is similar to that exhibited for the
THAM sample in appendix D.  The inflection point indicates
quantity of titrant required to titrate the  sample.  Time in
minutes is measured between this equivalence point, repre-
senting neutralization of the amine in the sample, and the
"start" point marked on the chart when the syringe pump
started to pump.  Basic N concentration in the sample is
calculated from time required for titration  of the sample
and the titer of the titrant in meq/min., which is determined
against standard tris  (hydroxylmethyl) aminomethane  (THAM).
                            -54-

-------
Sulfur determination in cut 2 oil (and other oils) is by
the bomb washing method of ASTM D 271.  A 1-g sample is
weighed into the platinum crucible  (or, for the volatile
cold trap oil, in a gel capsule) and placed in a Parr bomb
with 5 ml distilled water for gas absorption under 30
atmospheres of oxygen.  After ignition by the usual wire
method, the bomb is cooled in its bath to permit water in
the bomb to absorb sulfuric acid.  The bomb is thoroughly
washed with distilled water.  Bromine water is added to the
washings, which are boiled to release excess bromine.
Sodium hydroxide is added to make the washings slightly
alkaline; the washings are filtered to remove metallic
hydroxide.  Dilute hydrochloric acid is added to acidify
the filtrate to a methyl orange indicator.  Barium chloride
is added to precipitate sulfate as barium sulfate.  The
solution is digested, filtered, ignited, and weighed, and
sulfur is calculated by the standard method.

IV-D-3  Analysis of Solid Products
Solid products are vacuum bottoms and wet filter cake.
Most of the basic information for evaluation of both pro-
ducts is provided by elemental analysis (C, H, S, and for
CU 48 also N).  The ash content is determined for each, and
the fusion point for vacuum bottoms.

The methods for carbon and hydrogen are the same adaptations
of ASTM D 271 as have been described for liquid products.
The method there described for sulfur is also identical,
except that  a 1-g sample of vacuum bottoms or wet filter
cake is melted in the platinum crucible to insure complete
combustion at firing of the Parr bomb.  The Kjeldahl method
is used for nitrogen.
                             -55-

-------
Fusion point for vacuum bottoms samples is customarily
obtained only to provide a check on distillation procedures;
it permits sample comparison from run to run.

The ash content of vacuum bottoms, ususally 0.1% or less,
or of wet filter cake (25-30%) is determined by igniting
samples in platinum evaporating dishes.  For accuracy,
vacuum bottoms samples are relatively large, 5 to 10 g.
Wet filter cake samples can be smaller.  The dish is placed
on a sand bath heated with a Bunsen burner, which drives
off volatile matter and causes partial combustion of the
sample.  The dish is then placed in an electrically heated
muffle furnace at 750°C to complete ignition, then cooled
and weighed.  Residue is reported as ash.

In order to determine the amount of coal converted, wet
filter cake is analyzed.  A weighed portion is put in a beaker,
heated with pyridine, and stirred to extract any solubles.
The hot pyridine suspension is filtered and washed with hot
pyridine until washings come through colorless, then sequen-
tially rinsed with benzene and acetone.  The residue is dried
and weighed to determine % pyridine insolubles, which con-
sist of mineral matter  (unignited ash) and unconverted carbon-
aceous material (insoluble organic matter).  The % ash is
determined on this fraction.

The amount of coal converted to gaseous, liquid, and solid
products in the process is determined from pyridine insol-
ubles data.  By the Insolubles Difference Method, this is
     % feed coal converted = 100 - % pyridine insolubles,
                                     on loss free basis in
                                     reference to feed coal
An alternative method estimates amount of coal converted based
on the increased concentration of ash observed in pyridine
                            -56-

-------
insolubles.  By the Ash Enrichment Method/ this is
     100 (% ash in feed coal) = (100 - X) (% ash in pyridine
                                 insolubles) , where X is the
                                 % feed coal converted
Both methods are subject to some error.  The first depends
on quantitative collection of unreacted material; if handling
losses are corrected for in the calculation, materials lost
must contain the proper ratio of liquids and solids, which
is rarely the case.  The second method requires that the
mineral and carbonaceous material, not dissolved, remain
in the proper ratio; settling at any point may segregate
mineral preferentially, and this will cause error.  The true
percentage of coal converted probably lies between results by
the two methods.
                                    /
Since different coals may have different moisture and ash
contents, it is often desirable to reduce conversion to the
moisture-ash- free or MAP basis; the MAP basis represents the
organic fraction of the coal:
     % MAP conversion = <%
Both % ash and % moisture are % in feed coal.
                            -57-

-------
                         SECTION V
                RESULTS OF EXPERIMENT CU 48
V-A  OBJECTIVES
The objectives of CU 48 were:
     1.  To demonstrate that the solvent  (section V-B-2) ,
         which produced an unsatisfactory solution with Elkol
         coal, would behave in a normal way with Kentucky coal
     2.  To study the reproducibility of the coal solution
         production over an extended time; to learn if the
         solution and product characteristics would match
         those predicted by interpolation of data acquired
         through use of the middle fraction of raw anth-
         racene oil as solvent, and through use of a more
         hydrogenated oil as solvent
     3.  To prepare samples for detailed elemental balance
         studies of interest to EPA
     4.  To prepare materials for use in various projects
         which require coal derived materials (minerals re-
         claimed by retorting wet filter cake, for Washington
         State University; water from cold trap oil and
         knockout vessel, for Gulf Degremont waste treat-
         ment studies)
V-B  MATERIALS USED
The four input materials to the reactor are hydrogen, nitrogen,
solvent, and coal.  The first two are chemically pure gases.
The nitrogen purge serves only to prevent other material
from plugging lines to pressure gages; it does not affect the
reaction (any nitrogen in the products is that released
from, or retained in, the coal).

V-B-1  Coal
The coal used was Kentucky #9 from the P&M Colonial Mine near
Madisonville, Kentucky.  It is ranked as high volatile B
bituminous coal.  The substantive analysis for the sublet used
                            -58-

-------
in the feed slurry in CU 48 is reported on the as-received
basis:
     Carbon      62.66%     (Macro combustion train method)
     Hydrogen     4.22%     (Macro combusition train method;
                            corrected for H in H_ O reported)
     Sulfur       4.63%     (Oxygen bomb method)
     Nitrogen     1.26%     (Kjeldahl method)
     ELO          3.25%     (Karl Fischer method & oven-
                            drying loss method)
     Ash         15.25%     (Ignition at 750°C in platinum)
     Oxygen       8.73%     (by difference)
     Fe in ash   24.97%     (Ash dissolved in hydrofluoric
                            acid and  perchloric acid.  Iron
                            determined volumetrically.)
Separate analyses of this  sublet revealed small  variations  in
percentage, under 0.5% except for ash content  (up to  1%).
These differences are explained almost wholly by the  procedure
which prepared the coal  for use.

Mine samples drawn for laboratory use have been  more  or less
uniform and well averaged  samples according  to the  skill  and
care of the sampler.  At the laboratory/ the only control of
quality of sample which  can be exercised is  drawing the sub-
samples carefully.  The  sample actually used was received
in a 55-gallon drum, as  run-of-mine in chunks up to 6 inches
in diameter.  Half of the  sample had been removed and used
in earlier experiments.  When the remainder  was  used, it  was
broken up by hand to be  accommodated by the  grinder and was
ground in a Mikro SH Bantam pulverizing mill; a  subsample
was retained from the ground coal for analysis.  The  powdered
coal was placed in polyethylene bags, sealed, and stored  in
5-gallon cans  (nominal 10-kg contents), where it had  little
tendency to gain or lose moisture.  Coal for feed slurry
makeup was then dipped from a can as it was  needed.   Analysis
of each sublet showed that ash varied from 9.4 to 12% from
the average analysis for the first half of sample coal removed
                             -59-

-------
from the drum? and in the subsample used for the experiment
reported here, the ash content ran quite high (15.25%).
This variation in ash content among the subsamples suggests
that the mineral and organic phases had not been well mixed.
An attempted blending of raw coal in a rolling drum, instead
of homogenizing the blend, led to classification with a
nonhomogeneous ash distribution and similar variation of
ash level in coal used; a more satisfactory processing,
ultimately adopted, entailed blending before grinding by
the ASTM long pile sampling method, and other improvements.

As a check on the Kjeldahl method of determining nitrogen
content in feed coal, the Dumas method was also employed.
This was considered necessary because, when all materials
were analyzed by the Kjeldahl method, the recovery of nitrogen
in coal reaction products exceeded input nitrogen from feed
coal and solvent.  The obvious cause could be the failure of
this method to report all nitrogen in feed coal.  Circum-
stances which cause such failure are often met in analysis of
functional groups containing nitrogen; and it is not sur-
prising to meet the difficulty in a complex polymeric
system such as coal.  The Dumas method, available commercially,
is often successful on compounds which do not yield good results
in Kjeldahl digestions.  While it may still be suspect, it
does yield higher values, and thereby suggests the qualitative
accuracy of the diagnosis that nitrogen in feed coal is not
fully reported by the Kjeldahl analysis.  This subject is dis-
cussed more fully in section VI-B-2.

At the time the Dumas analyses were obtained for the input coal,
it was also resolved that the effect of sample size on nitrogen
                             -60-

-------
content reported should be established.  The feed coal, well
pulverized for microanalysis and considered homogeneous, was
therefore analyzed in a range of sample sizes.  The results
are shown below:
     SAMPLE SIZE (mg)      %N
        40.345             1.25
        34.928             1.30
        30.460             1.37
        20.290             1.39
        13.461             1.64
        13.772             1.51
This range of values, rather larger than the usual precision
of the Dumas procedure allows, may demonstrate a real effect
of sample size on value.  It was assumed/ anyway, that the
larger samples deposit larger amounts of coke residues which
retain nitrogen.  If this assumption were correct, it would
be reasonable to plot all data and extrapolate to zero sample
size for an estimate of the probable % N in the feed coal.
Figure 10 plots these data; the precision of the analysis
is suggested by data at 13 mg, where the band bounded by two
straight lines including all data has a mean deviation of 0.065%.
The center of the band at zero sample size may be taken as
average value, or 1.79% + 0.04% N.  The average dry basis N
would be 1.88%.

V-B-2  Solvent
The solvent used was taken from a blend of solvents reclaimed
from earlier experiments.  This blend had been prepared as a
150-gallon batch for future work as follows.

In preparation for run CU 40, three 55-gallon drums of raw
anthracene oil had been distilled at atmospheric and reduced
pressure to obtain the 550-800°F middle fraction.  After heating
                             -61-

-------
0
 I


                                                                      25        30
                                                                SAMPLE SIZE, MILLIGRAMS
                                                                                                   40
'45
50
                                           Figure 10   Nitrogen  in Feed Coal  (Dumas Method)

-------
and blending of the middle fractions, received as three sepa-
rate cuts, the distillation range for the blend was 70-256°C
at 3 mm Hg; about 93% was within the desired range of 100-230°C.
This middle fraction raw anthracene oil (MFRAO) had been found
suitable in many of its characteristics for reactor operating
conditions which would remove sulfur from coal efficiently.

CU 40 was a hydrogenation run; that is, it was an operation in
which only oil and hydrogen were admitted to the reactor, for
the purpose of adjusting the hydrogen content of the oil.  A
concentration of 6.04% hydrogen would be near the equilibrium
hydrogen content estimated for an anthracene oil solvent in
single-pass steady state equilibrium with Kentucky #9 coal in a
reactor under 1000 psig hydrogen pressure.  For this run the
reactor configuration and operating conditions were therefore
modified as follows.

The dissolver vessel of the reactor was dismantled and cleaned
out.  A wire mesh basket about 1-1/2 inches in diameter and 14
inches long was made from fine mesh stainless steel screen.
This just filled the volume of the dissolver vessel.  The basket
was filled with 150 grams of NALCOMO 471 cobalt-molybdenum-
alumina extruded catalyst (1/16- x 1/8-inch extrudates).  The
preheater was made double the normal length and the pumping
heads were modified to allow pump rates of up to 10 kilograms
of feed oil per hour*  In order to simplify feeding of larger
amounts of oil, the slurry recycle pump was arranged to draw oil
directly from a 55-gallon drum of material.  The drum was
heated electrically to maintain a temperature over 73°C, the
point at which all crystalline material appeared to melt in a
laboratory test.  Crystalline material was present in this lot
of anthracene oil because it had not been fully weathered at the
                             -63-

-------
time it was needed for use.  This material was less fluid and
generally more difficult to process than the raw anthracene oil
stock used, previously.

The reactor was started with a hydrogen pressure of 1000
psig, and the operating temperatures were slowly advanced
to start the reaction.  Samples were taken for measurement of
infrared spectra to control the hydrogenation reaction.
The sand bed for the preheater was advanced to 450°C (although
at these flow rates it is likely that the fluid in the pre-
heater did not reach this temperature) .  With catalyst bed
temperatures ranging from 360°C upward to 430°C, the reaction
with hydrogen progressed at useful rates.  Feed rates, temp-
eratures, and hydrogen pressures were varied as required to
produce liquid product with an IR of about 0.77.  Samples
were taken about every two hours for inspection by infrared,
and the operating variables were adjusted as required to
produce material with the desired composition.  In several
instances the reactor was upset, and the off-specification
materials made in these intervals were set aside for use in
developing an infrared working curve for solvent analysis.

The materials which were considered usable were blended to
make a composite sample.  This material consisted of about
120 gallons of hydrogenated stock.  About 30 gallons of Pilot
Plant Process Reclaim Solvent (LR 2023), which had been
retained from previous work, was then added to this stock.
This material had an IR ratio of 0.83 and was considered
suitable for this use.  The resulting blend was then filtered
through candle filters of about 30 to 40 micron retention
to remove any insoluble materials present.  Upon cooling it
was found that excessive crystalline material was deposited;
the filtration was repeated at room temperature to correct
this problem.
                             -64-

-------
A sample of the blended oil was then stripped to remove the
fraction boiling below 100°C at a nominal pressure less than
3 mm Hg.  Analytical data for LR 2023, the anthracene oil
fraction processed, and the CU 40 final blend after stripping
are as follows:
                                               Stripped, Hydro-
                                               genated Middle
                 Middle Frac-  Middle Fraction Fraction of Raw
                 tion of       of Raw Anthra-  Anthracene Oil
                 LR 2023       cene Oil (MFRAO)(SHMFRAO)
Carbon %           91.07
Hydrogen %          6.12
Nitrogen %          0.82
Sulfur %            0.56
Oxygen %            1.43
(by difference)
Total             100.00
H/C Atomic Ratio    0.800
IR Ratio            0.83
91.68
5.77
0.99
0.54
1.02
100.00
' 0.75
0.51
90.58
6.04
0.60
0.36
2.42
100.00
0.79
0.75
The SHMFRAO was the solvent mixed with coal for the CU 48
experiment reported here.

Some explanation of the rather particular determination of
solvent characteristics may be gained from the following
discussion.  From analysis of many dozen samples of solvents
prepared in anthracene oil experiments, it had been possible
to construct a working curve useful in assessing the degree
of reaction for any given set of operating conditions.
Equally important, the same data affords a means of selecting
input solvent characteristics or, by extrapolation, the pre-
diction of output solvent characteristics.

A sample of preliminary calibration data for the development
of the working curve is presented below.  The IR ratio for
                            -65-

-------
product gas and for oils has been discussed  (section IV-D) .

The samples below were run at  slow  speed on  a Perkin Elmer

237 grating spectrophotometer,  from which  the ratios of

absorbance were obtained:

          •=£  _  absorbance at 3.41 microns  (2933 cm"  )
                 absorbance at 3.28 microns  (3049 cm"1  )
The relative intensity of these absorbances  is  found to shift

with hydrogenation :  raw anthracene oil tends to be lean in

hydrogen, and reclaimed solvent contains more hydrogen.  In

the following data hydrogen was determined by combustion an-
alysis.

          Solvent                          %  C     % H  IR Ratio

Middle fraction,  hydrogenated reclaim
     solvent
Run PA 39 redistilled reclaim solvent
UNO 373 reclaim solvent
UNO 377 reclaim solvent
UNO 376 reclaim solvent
Run PA 74 reclaim solvent
UNO 277 reclaim solvent
Run PA 72 reclaim solvent
Continuous Run 2  reclaim solvent
Process Development Plant Run 66
     reclaim solvent
Continuous Run 8  reclaim solvent
Run PA 73 reclaim solvent
Continuous Run 11 Sample 17
     reclaim solvent
UND anthracene oil redistilled
     at atmos. press.
Continuous Run 7  reclaim solvent
Continuous Run 12 reclaim solvent
UND 357 reclaim solvent
Middle fraction of raw anthracene oil,
     sample 1
Middle fraction of raw anthracene oil,
     sample 2
91.42
91.38
90.64
90.59
90.85
90.46
90.33
90.78
90.65
90.41
90.93
90.90
6.901
6.616
6.412
6.428
6.407
6.281
6.132
6.158
6.092
5.987
5.925
6.151
2.32
1.48
1.26
1.26
1.22
0.97
0.94
0.85
0.78
0.77
0.75
0.74
                                         91.35   6.024   0.71
90.91
91.05
91.20
91.09

91.32

91.01
5
5
5
5
  916
  895
  896
  828
5.731

5.741
                                                         0.69
                                                         0.65
                                                         0.61
                                                         0.57

                                                         0.56

                                                         0.56
The change in absorbance with hydrogenation may also be seen
in figure 11, which shows the same maximum absorbance in

three spectra for comparison of the single strongest band.
                            -66-

-------
u
u
a
o-
o
.
-------
 The 3.28 micron region is predominantly aromatic material.
 The ratio includes  hydroaromatic hydrogen and hydrogen  on any
 saturated radicals  as  substituents,  or any saturated  hydro-
 carbons  present as  diluents in the solvent at 3.41 microns.
 In  solvents  derived from anthracene  oil,  little  saturated
 diluent  is observed since most of the solvent dissolves read-
 ily in concentrated sulfuric acid;  in cut 2 reclaim solvents
 the saturated oil is less than 1% by volume,  too small  to
 isolate  from a  typical sample.   While the increase in hydro-
 gen can  be observed by means of the  IR ratio,  the difference
 between  hydroaromatic  and saturated  structures is not re-
 solved.   In  a single pass,  the  change in  ratio is mostly
 caused by change in hydroaromatic content;  as  a  solvent is
 recycled,  this  is progressively obscured  by saturated radicals
 derived  from the coal.   (Graphs including ratios for  petroleum
 derived  solvents show  the hydroaromatic structures completely
 obscured  by  saturated  radicals.)   It therefore remains  a
 problem  to isolate  this interesting  function.

 Nevertheless, experience  with IR spectra  for various  solvents
 affords an almost immediate  qualitative picture  of the  re-
 action.  Quantitative results are also  obtainable.  The sol-
 vents discussed are  soluble  in  carbon tetrachloride or
 carbon disulfide; the spectra can be run  by a  solution  in
 one cell with the pure  solvent in a matching cell in
 the reference beam.   If equal concentrations of  raw solvent
 solution are placed in one cell  and reacted solvent solution
 in the other, a sophisticated differential measurement  is
obtained.  Here, the gain in hydrogen is  linear with  increased
absorbance at 3.41  microns.  The following data,   including
recycle solvents from autoclave  studies,  is plotted in
 figure 12.  While most of the scatter may arise  from errors
inherent in measurements, the last entry  in the data below
                             -68-

-------
  1.4
  1.2
  1.0
00.8
2
<
ffi
  0.4
 0.2
          0.2    0.4     Q6     0.8     1.0      1.2
                           %  HYDROGEN  GAINED
1.4     1.6
                 Figure 12  Differential  IR Working  Curve
                                     -69-

-------
represents catalytic hydrogenation under severe conditions.
This hydrogenation resulted in removal of sulfur, oxygen,
and nitrogen.  It also caused formation of alkyl groups,
which indicates that more than one kind of structure con-
tributes to absorbance at 3.41 microns.
             Solvent                  %    Gain    Differential
                                  Hydrogen in %     Absorbance
Continuous Run 14 Sample 11         5.736   -00-    Reference
Continuous Run 15 reclaim solvent   5.938   0.202     0.135
Continuous Run 11 Sample 17         6.024   0.288     0.195
UND Run 360 reclaim solvent         6.112   0.376     0.314
Continuous Run 18 Sample 7          6.148   0.413     0.298
Run PA 74 reclaim solvent           6.281   0.545     0.362
Continuous Run 19 Sample 40
     reclaim solvent                6.267   0.531     0.396
UND Run 376 reclaim solvent         6.407   0.671     0.467
Run PA 39 reclaim solvent           6.616   0.880     0.670
Middle fraction of hydrogenated
     reclaim solvent                6.901   1.165     1.13

The foregoing data formed the basis for evaluating solvent
and coal solutions up to the time of CU 40.  IR ratios were
plotted for all reclaimed solvents on all runs, and the
results were checked by combustion analysis.  The results
were regarded as estimates of trends, useful in rapidly
evaluating the effect of a variable change in the process.
Alternatively, changes in the operating conditions could
be made to maintain the ratio desired, as in the hydro-
genation procedure of CU 40 for solvent production.

Three samples of the CU 40 production were selected to be
the basis for a new working curve:  original raw anth-
racene oil, a moderately hydrogenated product, and a well
hydrogenated or over-hydrogenated product.  Each was care-
fully analyzed for carbon and hydrogen; blends were
prepared on an analytical balance in large sample weights
to minimize blending errors.  IR spectra for all material
                           -70-

-------
are tabulated below and plotted in figure 13; IR values
were computed by dividing measured absorbance at 3.41
microns by measured absorbance at 3.28 microns.
     Sample                     % C          %H          IR
Raw anthracene oil           91.41+0.16   6.341+0.008   1.04
Moderately hydrogenated AO   91.60+0.08   5.770+0.0001  0.55
Well hydrogenated AO         91.69+0.18   6.603+0.076   1.41
Blend 1                        Calculated 6.001         0.73
Blend 2                        Calculated 5.933         0.66
Blend 3                        Calculated 6.169         0.89
Blend 4                        Calculated 6.450         1.21

A plot with good fit of points to the average trenc! is
obtained from this blending procedure.  For coal solution
studies, the reclaimed solvents may contain variable
proportions of functional groups or radicals.  Here,
usually, the ratios of. functional groups do not match
exactly, and small drifts of data from an averaged line
must be expected; but the curve is helpful for inspection
of reaction products and is often as accurate as a routine
combustion analysis.

V-C  REACTOR OPERATION
The operating conditions for CU 48 were chosen to produce
a maximum of vacuum bottoms together with effective desul-
furization.  The operating time for the run was about 40
hours, of which 7 hours was  allowed for heating the re-
actor and bringing slurry feed and reactor contents into
a nominal steady state condition.  The analytical sample,
CU 48A1, was an hour-long collection taken during the
ninth hour; CU 48A2 was also an hour-long sample, taken
toward the end of the run.  The general conditions are
                                                        2
summarized in table 2.  Hydrogen pressure was 70.3 kg/cm
 (1000 psig); the coal and solvent analyses are given in
section V-B.
                            -71-

-------
     6S
     G6L
     64-
% H
     62r
     6.0-
     5.6
      02     04    0.8    OB     1-0     12
18     18    20    22
                             IR RATIO
                =TT   Absorbance at  3.41 microns
                 IR =
                      Absorbance at  3.28  microns
                      Figure 13  Working Curve, % H in Anthracene
                                 Oil Vs. IR Ratio
                                  -72-

-------
            Table 2.  EXPERIMENTAL CONDITIONS
CONDITION
Coal
Solvent
Solvent: coal ratio in feed slurry
Reactor pressure, psig
Reactor temperature,  °C
LHSV
GHSV
l/LHSV:hr
Solvent:MAF coalrwater             12.50:1:0.08
	___	1	
           CU 48A2
  Kentucky No.  9
  CU 4OB cut 2  oil
   2:1
1000
 450
   1.96
 304
   0.510
      2:1
   1000
    450
      1.96
    304
      0.510
2.50:1:0.05
               Table 3.  OPERATING DATA
ITEM
Blackness of solution
IR ratio of feed solvent
IR ratio of solution
IR ratio of cut 2 rec!
% recovery
% MAF conversion




m oil


CU 48A1
11.65
0.745
0.935
0.744
98.6
82.85
CU 48A2
12.26
0.745
0.940
0.758
99.2
85.09
                 Table 4.  CU 48 YIELDS
MATERIAL
Carbon monoxide
Carbon dioxide
Hydrogen sulfide
Hydrocarbon gas
Water
Excess solventa
Vacuum bottoms
Insoluble organic matter

              Total








% MOISTURE-ASH-FREE COAL
CU 48A1
0.26
1.37
2.33
6.54
2.33
2.88
CU 48A2
0.23
1.12
2.32
5.28
3.60
5.36
! 68.51 • 68.12

17.15 ; 14.91
 101.37
    100.94
a
 Under the method of reporting, product solvent is not
shown; "excess solvent" is that amount exceeding the input
quantity.  The excess varies considerably with operating
conditions and workup techniques, and is significant only
to show that solvent for recycle should be available at
the operating conditions studied.
                           -73-

-------
During the run occasional IR and blackness checks were
made to study the reproducibility of the solution being
made.  For the samples, these values are included in table
3.  Because at room temperature the slurry contained the
sticky crystals mentioned earlier, the slurry was heated
to 60-70°C.  The effect of heating slurry to keep the feed
pump system from plugging was probably the loss of some
water from the slurry and a small loss of the most volatile
solvent fraction.

Except as noted above, the reactor was operated as
described in section IV-B.  IR spectra for most types of
the liquid products are presented in appendix E through I.

V-D  YIELDS
Yields (table 4) of most substances followed the expected
pattern, but hydrocarbon gases and insoluble organic matter
were quite high.

The yield of hydrocarbon gas is largely derived from the
solvent; the increase indicates that the input solvent may
be less stable than that used earlier  (e.g., raw anthracene
oil).  The difference in behavior may possibly be explained
by the increased substitution of saturated radicals and
functional groups in the process reclaim solvent blended
with CU 40 solvent.  For comparison, the composition of
the input solvent is shown in table 5 with solvents re-
covered from samples.  The results indicate that the
solvent has been brought into balance with respect to
net hydrogen, but that some residual reactivity remains.
Oxygen elimination and possibly other molecular cleavages
consume chemical energy which would otherwise be avail-,
able to increase the conversion of coal.  On the basis
                           -74-

-------
           Table 5.  CU 48 SOLVENT COMPOSITIONS
CHARACTERISTIC
% Carbon
% Hydrogen
% Nitrogen
% Sulfur
% Oxygen
Total
H/C Atomic Ratio
IR Ratio
INPUT SOLVENT
(CU 40B Cut 2 Oil)
90.58
6.04
0.60
0.36
2.34
99.92
0.795
0.775
CUT 2 OIL RECLAIMED
CU 48A1
91.31
6.02
0.765
0.40
1.50
99.995
0.785
0.766
CU 48A2
91.73
6.03
0.734
0.377
1.129
100.000
0.783
0.758
of a lower conversion, increased insoluble organic matter
would be present at the end of the reaction time allowed.

Duplicate run conditions were compared  (CU 28 and CU 48) ,
in which the gas space velocity only was modified (from 304
to 200-230).  Results indicated that the reaction is not
particularly sensitive to hydrogen flow rate per se.  Changes
of operating pressure affect the reaction more directly.
Generally the yield data appear normal.

V-E  GAS ANALYSIS
The product gases are shown in table 6.  The difference in
distributions is explained in the note.  Compared with
prior runs, the values are representative for the oper-
ating conditions.

V-F  ELEMENTAL BALANCES
As a result of efforts to close material balances care-
fully, duplicate samples were obtained which each
accounted for 98% of the input material.  Given the solvent
composition currently available and operating conditions
simulating a Pilot Plant run, the yield pattern should
represent that of the Pilot Plant.
                            -75-

-------
              Table 6.  CU 48 PRODUCT GASES
GAS
Hydrogen
Hydrogen sulfide
Nitrogen
Methane
Ethane
Propane
n-Butane
Isobutane
Carbon dioxide
Carbon monoxide
MOLECULAR %3
CU 48A1
73.94
3.64
5.70
9.32
2.91
1.45
0.75
0.10
1.68
0.52
CU 48A2
77.7
3.09
7.19
6.46
2.32
0.97
0.53
0.09
1.24
0.40
     ^formalized on an air- and water-free basis.
     Sample CU 48A1 was taken in the ninth hour,
     CU 48A2 about 30 hours later; the gas differ-
     ences are partially-explained by a pressure in-
     crease of 1.8 Jcg/cm   (25 psig) later in the run.

The material balance was obtained in steps:  balancing
feed and primary products, then distributing these pri-
mary products in the proper weight ratios as the separa-
tions were made.  The basic data on worksheets detailed
all handling losses and % yield for the various distill-
ation, filtration, and solvent reclaim operations.
Materials are first reported on a handling loss-free
basis which accounts for all material in the primary
reactor samples; finally the weights of these samples are
used to adjust the yield pattern to a completely loss-
free basis.  The adjustment, applied to the weight of
liquid product obtained, was prorated in proportion to
yield throughout the product distribution obtained in
workup of the sample.  This final adjustment to loss-free
basis required only a small correction, since the primary
product sample was good.  Products were sampled after
filtration and after final separation, as an internal
check useful in the study of sampling errors or technique
                           -76-

-------
errors in the analytical laboratory.

V-F-1  Accounting for Material Fed
Feed slurry was made by weighing charges of analyzed
coal and analyzed solvent, initially 1500 g and 3000 g
respectively, on a balance with a precision of +0.1 g
for each.  Additions were mixed in substantially the same
way to maintain volume in the feed vessel.

The feed vessel was weighed, while an air motor stirred
the slurry for uniform feed in the circulation loop to
the high pressure pump, on a balance with a precision of
+10 g in weighing 0-20 kg.  Since the heat of stirring
tended to drive water out of the feed slurry, an accurate
feed analysis for water content was needed.

The composition of both coal and solvent has been given
in section V-B.  The other feed components are hydrogen,
fed from a calibrated high pressure system  (section IV-A-1),
and nitrogen, fed from a Ruska pump to the stirred auto-
clave  (figure 4) as a gage line purge.  The weights of
gases fed, or number of moles of hydrogen and nitrogen,
were determined accurately by application of standard
gas law calculations.

Thus, feed components were reduced to weights of materials
by measuring weight of slurry and noting sample time
during which a constant flow of hydrogen and a measured
volume of nitrogen were introduced.  The required values
for each feed material are expressed in grams.  Nitrogen
from this source is considered inert in the balance and
is therefore not reported.
                           -77-

-------
Feed slurry was  sampled at the beginning and  at  the  end
of each timed sample  period, each was analyzed by  the
Karl Fischer method for Wlfer*, -aliS-:^He-Toeaii- -values
computed for water content:
         CU 48A1 2.05%'
         CU 48A2 l;.4£%i
For reasons to be .discjisafis&ifltlaeQew \BBnaafesl5are probably
incorrect.

V-F-2  Weights of_ Materials gesk
For the timed one-hour .sample intervals, the  weights of
materials fed in were:
                                CU 48A1     CU  48A2
            Hydrogen, g          11.2         11.2
            Slurry, g          102-O^jfl       1000.0
            Slurry composition
                 Dry  coal, g    332.5        324.9
                 Solvent, g     68.7itS        675.1
The input by elements is-shown iiiktatrles ? and  8.   In
these tables the "primary sepaxaifcid^irKp'resents the
compositions from the sample taken after filtration,
the "secondary separation-" the3 cqinpositipns- from- the
sample after distillation* ^GEaaastoDfrrtCv'fH', N, S, ash,
and O recovered  at each stage .aji^iShQKiB with %  fed into
the reactor.  The steps in the elemejitsajL1-^balance cal-
culation are listed below.
     1.  Slurry  fed:   from prel-i^rvsanfr trial calcu-
         lations it is likely that H? O, in  slurry as fed
              ,»  --  .   i *n* 	• i? L£? _T O _Wi_ • i	• 	'	
was 0.0% to 1.08%  maximum^.0  -The best approx-
imation therefore- assisnejsi^aanzry to be made
with wet coal  at 3.25%. H0pbujt_d^ried_ Before
use.  The adjustmeiri
of coal and oil  fedi
         Mineral loss:  from prrflftix&S:3r'*ttfiai tiaTcu-
         lations it is establis3ffifiid4Kt
          occurred.   To obtain a sulfur  balance, it is
          essential  that the mineral  accounting be
          forced and that product  distribution be made
          with the forced balance  yield and product dis-
          tribution.
                             -78-

-------
 I
-J
vo
 I
Table 7. ELEMENTAL BALANCE, CU 48A1

MATERIAL
"INPUT : I"

Coal
Solvent
Hydrogen
Total
WEIGHT
__{£) _

% C

% H

% N

% S

% ASH % 0
coal assumed to have 3.25% ^0; slurry assumed dry when fed
332.5
687.5
11.2
1031.2
64.76
90.58


4.36
6.04


1.85
1.07


4.94
0.36


PRIMARY SEPARATION - Loss free basis, ash balance forced
Dry gas 35.9, See table 9
'HgO (knockout)
hbO (cold trap)
Cold trap oil
Filtrate
Uet filter cake
Total
% Recovery
2.6°
23. 9b
8.0
770.3
190.5
1031.2


0.57
82.68
90.36
62.32




12.04
6.25
3.89


0.24
0.38
0.212
1.17
1.08


0.547
0.821
0.65
0.45
4.70


SECONDARY SEPARATION - Loss free basis, ash balance forced
Dry gas 35.9 See table 9
H20 (knockout)
H20 (cold trap)
Cold trap oil
Cut 1 oil
Cut 2 oil
Heavy oil
Vacuum bottoms
Pyridine
insolubles
Total
% Recovery
2.6
23. 9b
8.0
29.9
631.6
15.3
185.5
98.5

1031.2


0.57
82.68
87.47
91.31
86.35
87.50
33.61





12.04
7.99
6.02
5.15
5.23
1.67



0.24
0.38
0.212
0.0665
1.14
1.87
2.20
0.775



0.547
0.821
0.65
0.20
0.40
0.815
0.882
8.68


15.76
_








27.51










0.179
52.88


i
8.33
1.95




88.9
88.0 ,
4.42
1.77
0.50



-
83.9
88.0
4.42
3.67
1.13
5.82
4.01
2.39



c
(g)

215.3
622.7

838.0

15.14

0.14
6.61
696.04
118.72
836.6
99.8

15.14

0.14
6.61
26.15
576.7
3.2)
162.31
33.10

833.4
99^4
u
(g)

14.50
41.53
11.20
67.23

11.68
fl.29
2.35
0.96
48.14
7.41
70.83
105.3
'
11.68
0.29
2.35
0.96
2.39
38.02
0.79
9.70
1.64

"67.82
100.9
N
(q)

6.15
7.36

13.51


0.01
0.09
0.02
9.01
2.06
T17T9
82.8


0.01
0.09
0.02
0.20
7.20
0.29
4.08
0.76

12.65
93! 6
s
«J
(g)

16.43
2.48

18.91

5.83
0.01
0.20
0.05
3.47
8.95
TO6
98.1

5.83
0.01
0.20
0.05
0.06
2.53
0.13
1.64
8.54

19.04
100.7
0
V
(q)

27.70
13.41

41.1

- 3.13
2.31
21.03
0.35
13.63
0.95
41.40
100.7

3.13
2.31
21.03
0.35
1.10
7.14
0.89
7.44
2.35

45.74
111.3
ASH
rWJ| 1
(0)

52.4


52.4






52.4
527?









0.33
52.07

5274"

                a   99SS H20
                b   982 H20

-------
 I
00
o
 I
Table 8. ELEMENTAL BALANCE, CU 48A2

MATERIAL
WEIGHT
(g)
INPUT - Coal assum
Coal
Solvent
Hydrogen
Total
324.9
675.1
11.2
1011.2

% c

% H

X N

% S

% ASH

% 0
sd to have 3.25% H20; slurry assumed dry when fed
64.76
90.58


4.36
6.04

>
1.85
1.07


4.94
0.36


PRIMARY SEPARATION - Loss free basis, ash balance forced
Dry gas
H20 (knockout)
H20 (cold trap)
Cold trap oil
Filtrate
Wet filter cake
Total
% Recovery
32.2 See table 10
3.6a
19. 4a
6.8
779.4
169.8
Tom


0.57
86.41
89.68
59.27




11.28
5.94
3.54


0.55
0.485
0.219
1.28
1.10



1.25
0.841
0.886
0.46
5.14


SECONDARY SEPARATION - Loss free basis, ash balance forced
Dry gas
H?0 (knockout)
H20 (cold trap)
Cold trap oil
Cut 1 oil
Cut 2 oil
Heavy oil
Vacuum bottoms
Pyridine
insolubles
Total
% Recovery
32.2 See table 10
3.6a
19. 4a
6.8
30.9
627.6
17.7
182.1
90.9

ion .2


0.57
86.41
88.05
91.73
87.26
87.32
32.03





11.28
7.71
6.03
6.24
5.11
1.73



0.55
0.485
0.219
0.69
0.91
1.92
2.27
0.90



1.25
0.841
0.886
0.244
0.377
0.842
0.944
9.14



15.76









30.16










0.133
56.07



8.33
1.95




88.0
88.0
1.21
2.64
0.79




88.0
88.0
1.21
3.36
0.95
3.74
4.22
0.13



C
(q)

210.4
611.5

BTTT

12.08

0.11
5.88
699.00
100.64
817.7
99.5

12.08

0.11
5.88
27.21
575.7
15.45
159.00
29.11

824.5
100.3
H
(
-------
 3.   Sulfur  and nitrogen:   selected values  are used.
     For nitrogen,  except  in  water phases and cold
     trap oil,  these are extrapolated  Dumas values.
     Kjeldahl results are  shown  for cold trap oil,
     distillation of ammonia  from MgO  results for
     water phases.   Knockout  vessel water and water
     from the gas sample bag  are assumed to have the
     same composition.  NH3 composition here is
     calculated as  though  NH4SH  were present, since
     only enough sample for the  sulfur determination
     was available.

 4.   Sulfur  results:  these are  gravimetric sulfurs
     from oxygen bomb combustions on all low-ash
     products.   High-ash products have been analyzed
     gravimetrically after Eschka decomposition
     of samples.

 5.   Gas compositions:  these are slightly  revised
     for calculation of dry volumes and weights.
     Cold trapped HO has  been added to H_0 from the
     knockout vessel and is assumed to have the same
     composition.

 6.   Wet filter cake:  the weight is computed  from
     the ash content of the filter cake and the
     theoretical ash input.

 7.   Pyridine insolubles:   the weight  is computed  from
     the ash content of the pyridine insolubles and
     the theoretical ash  input.

 8.   Stripping weights:   the  weights of gas, water,
     and oil taken  at stripping are taken  relative
     to the whole sample  collected.  Except for
     holdup losses  in the  distillation equipment,
     these are not  corrected.

 9.   Remainder:  the sum  of filtrate and wet filter
     cake must equal the  remainder  of  the  whole  sample,
     Thus filtrate  is calculated by  subtraction of
     light products and theoretical wet filter cake
     from the weight of sample to be accounted.

10.   The elemental  analysis of each product is then
     entered.  The  weight of  each component element
     is computed.  In most cases, the  oxygen by diff-
     erence closes  all accounts, and  the  sum of
                        -81-

-------
         elements reported equals the product weight
         entered.  The gas has a residual of a few
         grams not acounted for, probably material
         measured in the gas but not calculated as
         elemental matter (e.g., oxygen associated
         with CO and COJ .


     11.  Mean values for carbon and hydrogen analyses
         are shown.  Extrapolated Dumas nitrogen values
         for most samples are shown; data for input sol-
         vent scatter, and an average nitrogen figure
         is given.  For the molecular weight and
         heterocyclic content, the extrapolation should
         not have a steep slope in any case.  Sulfur
         values in wet filter cake and pyridine insol-
         ubles are from the Eschka method.  Other re-
         sults are by oxygen bomb combustions for oils
         or from oxidation by bromine or alkaline per-
         oxide for aqueous samples.
V-F-3  Accounting for Products

Reactor products were collected in three vessels:  the

bulk of liquid in the tared stainless steel flask, a

small amount of liquid condensate in the knockout vessel

(obtained during gas collection), and gas in the rubber-

ized fabric bag.  During the first quarter hour of liquid

sample collection, gas was vented to the air to purge the

liquid sample flask of air.  The gas sample was then

collected for 45 minutes.  The volumes are calculated on

an hourly basis.


The initial accounting was made by adding the weights of

the three samples.  The weight of gas is the only complex

measurement and is described below.  The weights were:

                                CU 48A1     CU 48A2

         Liquid sample          977.1 g     953.7 g
         Knockout vessel liquid   1.1 g       1.6 g
         Gas sample              37.0 g      34.2 g
         Total                 1015.2 g     989.5 g
         % Recovery of feed      98.5        97.8
                            -82-

-------
The distribution of liquid product weights on a loss-
free basis is an intermediate step in calculating yields
(table 4).  The grams of each product made from each
sample are taken from these data and presented in tables
7 and 8 without the details of necessary corrections for
handling losses and products taken as analytical sub-
samples.

V-F-4  Composition of Products
The weight of product gas is obtained by calculation
from the gas analysis data and gas volume data.  Pro-
duct gas is reduced to an air-free basis on the assumption
that any oxygen present is caused by incomplete flushing
of the sample flask in the first 15 minutes of the
sample period.  It is also assumed that any air present
replaces an equal volume of air-free gas, which remains
in the liquid sample flask.  The analytical methods are
described in section IV-D-1.  Water is measured by cold
trap weight gain when the gas bag is vented; gas volume
is measured by venting the bag through a wet test meter.
Tables 9 and 10 summarize product gas composition.

Each of the materials reported in the summary tables,
tables 7 and 8, was analyzed for carbon, hydrogen,
nitrogen, and sulfur.  The composition of vacuum bottoms
is shown in table 11.
     1.  Water was determined in some samples, by Karl
         Fischer titration, and phenol by ultraviolet
         spectrophotometry.  Traces of phenol in water
          (usually 3/4%) were neglected in the results
         computation since the contribution to the
         carbon balance is trivial.  The ammonia and
         and sulfide figures are included.
     2.  Ash was determined in appropriate samples,
         though not in filtrates  (usually 0.03 or 0.04%
         ash).  The primary separation reported in
         tables 7 and 8 includes ash from the wet
         filter cake only.
                            -83-

-------
     Table 9.  PRODUCT GAS COMPOSITION, CU 48A1'
SUBSTANCE
H2
CH4
C2H6
C3H8
C4H1Q
CO
CO2
H2S
H2Qk
N2C
Total
MOL.
WT.
AIR-FREE
VOLUME %
2.0161 73.94
16.04
30.07
44.09
58.12
28.01
44.01
34.08
18.02
28.02

9.32
2.91
1.45
0.85
0.52
1.68
3.64

(5.70)
100.01
WEIGHT
(g)
7.511
7.532
4.409
3.221
2.489
0.734
3.726
6.250
(1.47)
(8.047)
35.87
C
(g)

5.640
3.522
2.632
2.057
0.314
1.016



15.14
H
(g)
7.511
1.892
0.887
0.588
0.432


0.370


11.68
S
(g)







5.88


5.88
0
(g)





0.42
,2.71



3.13
a Hourly volume:  132.3 liters, measured by wet test
  meter at 25.5°C and 733.5 mm Hg.  Corrected to dry
  gas at STP:
         (132.3) (273.1) (733.5 - 24.5)
            U98.6) (760)  (22.4)
=  5.039 moles
b Obtained as gas is vented through cold trap and wet
  test meter; this water is included with knockout
  vessel water in balance.

c N2 is from gage purge; not entered as input for
  material balance nor reported in gas output.
                         -84-

-------
    Table 10.  PRODUCT GAS COMPOSITION/ CU 48A2
SUBSTANCE
H2
CH4
C2H6
C3H8
C4H10
CO
C02
H2S
H2°b
N2C
Total

MOL.
WT.
2.016
16.04
30.07
44.09
58,12
28.01
44.01
34.08
18.02
28.02


AIR-FREE
VOLUME %
77.70
6.46
2.32
0.97
0.62
0.40
1.20
3.09

(7.19)
99.95

WEIGHT
(g)
8.719
5.767
3.882
2.380
2.006
0.624
2.939
5.861
(2.000)
(11.209)
32.18

C
(g)

4.318
3.101
1.944
1.656
0.267
0.802



12.08

H S
(g) 1 (g)
8.719J
1.4481
0.7811
0.436J
0.350!
i

0.357; 5.504


11.39 i 5.504
t
0
(g)





0.36
2.13



2.49

a Hourly volume:  146.4 liters, measured by wet test
  meter at 26.50°C and 736 mm Hg.  Corrected to dry
  gas at STP:
(146.4)  (736  -26.0)  (273.1)  =
  (760)  (299.6) (22.4)
                                                 x
                                               moj.es
b Obtained as gas is vented through cold trap and
  wet test meter; this water is included with knock-
  out vessel water in balance.

c N2 is from gage purge; not entered as input for
  material balance nor reported in gas output.
           Table 11.  CU 48 VACUUM BOTTOMS
MATERIAL
Carbon
Hydrogen
Nitrogen
Sulfur
Oxygen
Ash

% IN CU 48A1 %
87.50
5.23
2.040
0.882
4.17
0.179
100.001
IN CU 48A2
87.32
5.11
1.909
0.944
4.58
0.133
99.996
                          -85-

-------
3.  Cut 2 oil composition includes carbon from
    combustion and hydrogen from the IR correlation.
    Hydrogen from the combustion method will check
    the IR value within the precision of a single
    trial; but since IR correlation uses the
    average of many trials and can be reproduced
    with precision, it is considered more precise
    than the single combustion result.  Combustion
    results are used for the hydrogen of all other
    products.

4.  Sulfur was determined repeatedly in wet filter
    cake and pyridine insolubes by the oxygen bomb
    method in order to establish sulfur values
    presented.  The insolubles must be diluted
    considerably with benzoic acid to insure com-
    bustion; the proportions are critical, and
    enough fuel to insure complete decomposition
    is required.  The analyses are therefore
    duplicated by the Eschka method:

                          % S  (oxygen bomb) % S  (Eschka)
    Wet filter cake
        CU 48A1                 4.66            4.70
        CU 48A2                 4.772           5.14
    Pyridine insolubles
        CU 48A1                 8.128           8.680
        CU 48A2                 8.930           9.14
                      -86-

-------
                      SECTION VI
       REVIEW OF ELEMENTAL BALANCE CALCULATIONS

This section discusses both the reasons for failure to
make an exact element-by-element balance and the gains
made from the study.  It includes alternate analyses,
with some comment on the reporting system, to assist in
interpreting and evaluating the results.  Failure to
make an exact elemental balance was caused by operating
errors, practical difficulties in treating raw data, and,
in one case, the limitations of available analytical
techniques.

VI-A  MATERIAL LOSSES
The chief practical difficulties in obtaining a material
balance were caused by settling of mineral, holdup of
sample material in collection vessels, and loss of water
from feed coal.

VI-A-1  Ash
Mineral contains much of the sulfur to be reported.  But
raw samples did not contain as much ash as the feed
slurry should have contained.  Since the mineral balance
must be closed to account for all sulfur, the ash balance
was forced mathematically.  Such an operation is normal
to loss-free yield calculation; and because sulfur re-
sults for the individual samples were satisfactory, the
data based on a forced ash balance do account for input
sulfur very well.

The need for this forced balance arises from a fact
frequently mentioned earlier:  mineral  in this process
                             -87-

-------
tends to settle in the feed vessel, in the dissolver, and
in the sample flask, which results in preferential loss
of ash and associated unreacted organic matter.  The sig-
nificance of this fact was better understood during CU 48
than the way to handle it.  Since CU 48, the refinements
in both mixing of feed slurry and ash suspension have per-
mitted better reporting of ash input.  For better reporting
of ash output, several measures have been taken.

It will be noted that the primary and secondary separations
in CU 48 both account for the same weight of ash.  The
loss could be the amount settled in the reactor tube or
the amount clinging to the inside of the sample flask,
probably the latter since the time allowed to line out
the reactor should have been long enough to equilibrate
the mineral bed settled in the dissolver.  For CU 48, the
liquid sample flasks had been cleaned.  After stripping,
each was emptied into a filter.  Material held up on the
sides of the flask, and lost from the sample for further
analysis, was 14 g for CU 48A1 and 12.5 g for CU 48A2,
or more than 1%.  Because the holdup is richer in mineral
than the material filtered, and because ash is about 8%
sulfur, prorating lost material to the liquid product
distribution for loss free calculation will build in some
error.

The more recent practice permits better reporting.  In
later work, the same coal and a recycled solvent make up
the slurry for long series of runs; sample material tends
toward uniform viscosity; larger samples improve recovery
percentages; and the holdup in sample flasks is not
cleaned out between runs so that net holdup on successive
uses is negligible.  Material accounting is within 1/4%
at this stage of workup.
                             -88-

-------
The precise accounting for materials also suggests a long
sample interval to account for ash.  The 1-hour samples
taken during CU 48 might better have been a single 40-
hour sample for two reasons.  First, lineout would have
been assured.  Second, the long collection would tend to
overcome the effect of mineral variation found in the
samples taken more frequently.  (The characteristic of
the upflow reactor to discharge mineral contents in cycles
is described in section III; its effect on blackness and
IR measurements is indicated --in appendix J, representing
20 successive 4-hour samples in a later experiment.)

To allow for the loss of ash, CU 48 material balances were
forced by calculation of weights of insoluble material
and wet filter cake, from weight of input ash, analysis of
wet filter cake, and composition of pyridine-washed in-
solubles from the filter cake.  The holdup of material in
sample flasks or water vapor in the gas collection bag
is a small percentage, but it forces a prorating of the
elemental constituents to the product distribution in
loss-free accounting.

VI-A-2  Water
Feed coal was sampled and subjected to ovendrying and also
Karl Fischer titration for water at the time the coal
was initially ground and when the  sublot was taken for use
in CU 48.  A later check of water  in a retained sample
of the coal gave  3.25% by the titration and 2.0% by ovendry-
ing.  Assuming the input solvent contained no water,
after its stripping to 50°C at  3 mm Hg, the slurry evidently
had up to 1.08% water at the  2:1 solvent:coal ratio.
                             -89-

-------
Feed slurry was analyzed by Karl Fischer titration for
water; the results were too high.  Feed coal could not
account for water indicated by titration of slurry.
Feed slurry was operated warm  (60-70°C), probably causing
preferential loss of some water and the lightest fraction
of solvent, perhaps 1% of slurry weight.  Since the feed
vessel was on a scale and total weight was charged to
slurry fed, the weight of water and oil evaporated was
recorded as material fed.  The effect on mass balance and
oxygen balance is discussed with the recalculations .later.

The output water is mostly cold-trapped.  The collective
weight is determined by weighing water from several
trapping operations.  The efficiency of these traps
therefore influences the accuracy of the oxygen balance.
As a check on the procedure, an additional cold trap has
occasionally been used to collect any vapor overflow
from the trap collecting water and cold trap oil at
stripping.  It may collect an additional 5-10%; thus, the
error in collecting light materials may permit an in-
crease of 1-2 g water and 0.5 g oil in the samples.
Holdup of moisture in the gas collection bag can only be
estimated.  It can be concluded, however, that the un-
certainty of the oxygen by difference is large enough to
create larger average errors than the loss by cold trap
overflow or inefficiency and other verifiable losses.

VI-B  RECALCULATIONS
The elemental balances for CU 48A1 and CU 48A2 were re-
calculated several times.  The first trial balances are
shown in appendixes K and L.  The first trial used the
analysis of that lot of feed coal from the averaged sample
                             -90-

-------
taken at the time the coal was ground, with results of Karl
Fischer titration for water in the feed slurry  for the
estimate of water input.  After the check on the retained
sample of feed coal and the discovery that the coal could
not account for water indicated by slurry titration, the
composition of the feed slurry was reviewed.  It was
assumed that the only water present was that in the coal.
An oxygen balance was calculated, by difference for all
samples/ from the results of the best available ele-
mental analyses.  This balance indicated more oxygen in
the feed than in the products.  Since the warmed feed
vessel had actually lost water to evaporation during
recirculation in the feed loop, a second trial balance
was attempted.

The second trial assumed the coal to contain 3.25%
moisture as weighed, and also assumed the slurry to have
lost this water by the time slurry entered the reactor.
This procedure does result in the best accounting for
oxygen, although complete loss of water is unlikely.
The probable failures of Karl  Fischer titrations to
determine the facts leaves no numerical value for water
in the slurry.  Qualitatively, the value would best be
approximated between 0.00% for a dry sample, and 1.08%,
which would represent the slurry resulting from a 2:1
blending of moisture-free solvent with coal with 3.25%
water.

Since total loss of water from the feed vessel would in-
crease the weight of organic matter fed about 1%, the high
final C and H yields become reasonable.  For inputs
considered, the 1% is 10.2 g H2O or 9.1 g 02.  If this
                            -91-

-------
oxygen is used to correct the oxygen balance, nearly
all material is accounted for.  Incomplete recovery of
water from the gas bag, cold traps, condenser, etc.
is the probable cause for residual error.

The summary tables, tables 7 and 8, show that carbon and
hydrogen  are accounted for within a percentage
or so.  The correction to 100% recovery of feed required
the addition of 1.5% and 2.2% of material prorated to
product liquid distribution in the two balances.  This
practice is suspect wherever material lost is not rep-
resentative of material retained for analysis.

VI-B-1  The Reporting of Sulfur
The elemental balance study, then, enforced a full re-
view of sampling techniques.  In addition, the effort
to confirm balances by alternative and duplicate analyses
has revealed limitations in the available methods, which
suggest separate causes for imprecise results.

For sulfur, the difficulty of decomposing mineral-rich
samples (wet filter cake and pyridine insolubles) by
standard ASTM oxygen bomb methods has been mentioned.
Results by the Eschka method have been compared.  Table
12 shows the higher results obtained in CU 48 by the
second method.

VI-B-2  The Reporting of Nitrogen
The determinations of carbon, hydrogen, and sulfur were
satisfactory or good.  This is not true for nitrogen.
                            -92-

-------
     Table 12.  COMPARATIVE SULFUR ANALYSES (%)
MATERIAL
Coal (dry basis)
Solvent
Knockout H2O
Cold trap I^O
Cold trap oil
Filtrate
Cut 1 oil
Cut 2 oil
Heavy oil
Vacuum bottoms
Wet filter cake
Pyridine insolubles
OXYGEN BOMB
CU 48A1
4.94
0.36
0.547*
0.821*
0.65
0.45
0.20
0.40
0.815
0.882
4.66
8.128
CU 48A2
4.94
0.36
1.25a
0.841*
0.886
0.46
0.244
0.377
0.842
0.944
4.772
8.930
ESCHKA
CU 48A1










4.70
8.680
CU 48A2










5.14
9.14
     a  No decomposition necessary; treated with
        bromine water and gravimetric finish
For coal, solids, and materials of high molecular weight,

the reliability of nitrogen determinations has been a

question for some time.  The Kjeldahl method is often re-

garded in Germany as inaccurate for coal.l

     The Kjeldahl or Dumas method is generally used
     for the determination of nitrogen in solid fuels.
     There appears to be a considerable difference of
     opinion among fuel analysts as to which method
     gives the most reliable results .... Bunte and
     Schilling applied the Kjeldahl method to the an-
     alysis of solid fuels shortly after it was pub-
     lished and found it yielded very consistent re-
     sults . . .  [yet]  The values obtained by the
     Kjeldahl and Dumas methods (the latter including
     a subsequent combustion with oxygen) may differ
     by as much as 46% for coal.2
x W. M. Ode.  "Coal Analysis and Mineral Matter."
The Chemistry of Coal Utilization, ed. H. H. Lowry.   (New
York:  J. H. WlTey, 1963).  P. 214.

2 W. R. Kirner.  "Microchemical Analysis of Solid Fuels."
Industrial and Engineering Chemistry.  Vol. 7, No. 5
(July, 1935).  P. 296.
                           -93-

-------
The reasons for this difference, already indicated by
the present study, are primarily that Kjeldahl does not
report enough of the nitrogen actually present, however
reproducible Kjeldahl analyses may be.

In preparation for the elemental balance study here,
a technique study with various standard catalyst systems
and various digestion times had been made.  Obtaining the
same result from trial to trial presented little or
no problem; the results obtained were repeated with great
precision, and with standard Kjeldahl reagents and diff-
erent digestion times.  An example for coal is illus-
trated in table 13.  With such consistency in Kjeldahl
analytical results, it was naturally disturbing later on
to demonstrate an excess of nitrogen out of the reactor
over that amount put in.  (The first trial balances,
appendixes K and L, show 114% and 113% nitrogen recovery.)

The initial sample set for the elemental balance study
was analyzed for nitrogen by Kjeldahl, with mercury and
selenium to catalyze the digestion.  None of the modi-
fications of the method seems to produce significantly
higher nitrogen results than the ASTM method.  To confirm
initial results, the samples of all significant products
were analyzed by the Dumas method in the Schwarzkopf
Microanalytical Laboratory.   The-results, table 14, for
replicate Dumas analyses give values from 1.25% to 1.64%
in feed coal.  Because of the laboratory location, it was
not possible to search for systematic errors in the
measurement; but these results are not considered sus-
pect.  Results with the Coleman automatic Dumas nitrogen
analyzer have followed the pattern also:  Dumas results
are always higher than Kjeldahl.
                           -94-

-------
Table 13.  REPRODUCIBILITY IN FEED  COAL NITROGEN RESULTS (KJELDAHL)
SAMPLE
WEIGHT (g)
Set la
1.0373
1.0235
1.0572
K0245
1 . 0908
Blank

Set 2a
1.0722
1.0341
1.0083

1.0673
1 .0453
Blank

Set 3a
1.0040
1.0108
1.0064
1.0687
1.0499
Blank

•mi H2^U4
(0.10034)

10.30
9.90
10.55
1'0.12-
10.25
0.20


10.35
10.10
6.55

10.40
10.20
0.05


9.80
9.75
9.75
10.15
10.48
0.60

% Nb

1.369
1.332
1.376
1.361 '
1.295
(1.347)
average

1.350
1.366
0.806
lost
1.363
1.365
(1.366)
average

1.358
1.342
1.348
1.322
1.390
(1.352)
average
DEVIATTnN
d

40.022
-0.015
+0.019
+Q.OI&
-0.052



-0.011
+0.0005


+0.0002
+0.004



+0.006
-0.010
-0.004
-0.030
+0.038


d2

0.00048
0.00023
0.00036
-MQ02Q
0.00270



1.21 X 10"4
2.5 x 10-7


0.8 x 10~3
1.6 x TO'7



3.6 x 10~5
1 x 10-4
1.6 x lO"5
0.9 x 10-3
1.488 x ID"3


\l=>d2
\ n-1



1.347+0.03







1.366+0.0064








1.352+0.0512




Samples  in set 1 were allowed 2 hours digestion  after  the green color
appeared; set 2 allowed 3 hours; and set 3 allowed  4 hours.  Regents
were 0.7 g mercuric oxide (red) selenium and kelpack plus 30 g ^$04
(concentrated).

The same coal was analyzed by the Dumas method in various milligram
sample sizes, with these results:
     42.780  mg 1.09 %N
     33.930  mg 1.11 %N
     24.830  mg 1.30 %N
     23.080  mg 1.21 %N
     13.220  mg 1.58 %N
     13.138  mg 1.58 %N
                                 -95-

-------
           Table 14.  ALTERNATE NITROGEN DETERMINATIONS

SAMPLE
Feed coal





Cut 1 Oil






Cut 2 Oil






Filtrate



Heavy Oil



Vacuum bottoms



Wet filter cake



Pyridine insolubles



CU 48Ala
SAMPLE SIZE (mg)
40.345
34.828
30.460
20.290
13.461
13.772
25.110
21.100
15.100
1 1 . 005
C 7.878
C 7.706
C 4.704
24.928
18.440
13.665
C 11.858
C 10.451
9.427
C 6.760
25.810
21.760
14.120
9.390
20.645
16.060
12.035
8.510
21.855
17.812
12.365
8.885
26.834
19.828
14.611
10.585
39.330
31.710
20.062
15.180
%N
1.25
1.30
1.37
1.39
1.64
1.51
0.67
0.68
0.65
0.66
0.54
0.54
0.59
1.06
0.99
1.01
0.76
0.76
1.14
0.81
1.03
1.01
1.05
1.15
1.54
1.63
1.70
1.75
1.84
1.86
2.13
2.03
0.905
0.978
1.02
1.04
0.701
0.724
0.734
0.756
CU 48A2a
SAMPLE SIZE (mg)


(Same Coal)



30.315
26.643 •
21.869
16.790
C 14.563
C 8.015
C 5.039
22.792
16.820
12.570
C 11.201
C 10.782
8.310
C 5.340
24.580
19.492
14.610
9.820
17.580
12.115
9.343
7.705
20.551
16.370
11.553
8.740
24.565
19.510
15.070
8.840
40.452
30.665
22.048
16.048
%N






0.58
0.61
0.63
0.55
0.55
0.53
0.54
0.84
0.80
0.91
0.75
0.74
0.77
0.76
1.06
1.11
1.03
1.19
1.56
1.69
1.77
1.78
1.91
1.97
2.17
2.20
0.994
1.07
1.03
1.12
0.704
0.663
O'.SOO
0.783
All determinations were  by Dumas  except as  noted:  C  indicates colorimetric
determination.  Analyses were made between  23 and  25°C at 760-778 mm Hg,
at Schwarzkopf Kicroanalytical  Laboratory.
                                -96-

-------
As an added refinement in the search for an explanation,
the practice was to run several Dumas samples of different
sizes and then to extrapolate to zero size, on the theory
that larger samples deposit larger amounts of coke resi-
dues containing nitrogen.  If nitrogen values in table
14 for duplicate analyses of each sample are plotted by
sample size, the values may fall on a straight line or
scatter or form an array forcing an arbitrary choice be-
tween extrapolation or calculated average, to afford the
probable nitrogen content in the product.  The Dumas
extrapolation for feed coal (table 15) materially in-
creases the nitrogen estimate.  If an elemental balance
is made with this result and with the Kjeldahl analyses
of products, the nitrogen input substantially exceeds
the nitrogen output.

This condition is partially correctable by using the
extrapolated Dumas method for all input materials and
for all product samples; the final balance then indicates
some loss of nitrogen in workup or collection of materials.
This situation is more reasonable than the original, in
which nitrogen in the products exceeds nitrogen input.
But extrapolation of the Dumas results to zero sample
size is at best a dubious procedure.  There appears no
particular reason to suppose that the result is accurate.

Apart from conscientious applicaton of established methods,
it was considered that nitrogen by difference in liquid
products and ash-free solids might be determined if an
accurate oxygen analysis were available.  Since the carbon
and hydrogen analyses are sufficiently accurate for trac-
ing these elements, and since judicious use of oxygen
                            -97-

-------
         Table 15.  COMPARATIVE NITROGEN ANALYSES
MATERIAL
Coal (dry basis)
Solvent
Knockout II2O
Cold trap H2O
Cold trap oil
Filtrate
Cut 1 oil
Cut 2 oil
Heavy oil
Vacuum bottoms
Wet filter cake
Pyridine insolubles
KJELDAHL
CU 48A1
1.30
0.60

0.38a
0.212
0.981
0.566
0.765
1.611
2.04
0.895
0.644
CU 48A2



0.485
0.219
0.979
0.560
0.734
1.627
1.909
0.868
0.590
EXTRAPOLATED DUMAS
CU 48A1
1.85
1.07-1.11



1.17
0.665
1.14-1.19
1.87
2.20
1.08-1.10
0.775
CU 48A2





1,28
0.69-0.74
0.91-1.00
1.92
2.27
1.10
0.90
OTHER
CU 48A1

0.77b
0.24C
0.38a


0.56b
0.74b




CU 48A2


0.55C
0.485a


0.54b
0.75b




   a Ammonia  in H2O phase, distilled over MgO
   b Colorimetric N (Schwarzkopf)
   c H2O in the knockout vessel was used up by the analysis
    for sulfur (as NH^SH), from which N values are taken.
bomb and Eschka decompositions give accurate sulfur
results, only oxygen  or  nitrogen must be measured pre-
cisely; the other  is  then determined by difference.
Thus, the conventional nitrogen analysis might be checked
by the neutron activation analysis for oxygen.  Unfor-
tunately, the method  would have limited use; neutron
activation is difficult  to apply in substances with
mineral because the oxygen in mineral is also deter-
mined along with oxygen  in the non-mineral material,
and prorating oxygen  to  the separately reported ash
would be artful.   Therefore,  a method for the simul-
taneous analysis of ash  from the coal and analysis of the
coal sample would  be  desirable.   It was concluded from
these investigations  that separate determinations are
preferable, since  any result obtained by difference will
sum the errors in  all determinations involved.  The
manipulations that gave  an oxygen balance were justified
only by achieving  carbon,  hydrogen, and sulfur balances
in the data sets;  yet the accuracy of the oxygen balance
could be affected  materially by the accuracy of the nitro-
gen balance.
                             -98-

-------
Neither functional group analysis nor IR spectroscopy
accounts for all nitrogen or oxygen in a dependable way.
Analyses for the light fractions are better than for
high molecular weight fractions.  For coal, the
information from functional group determinations
generally supports the conversion of part of the
nitrogen to amine functions, with hinderedKor neutral
functions accounting for an appreciable fraction of the
total nitrogen in the higher molecular weight materials.
Application of the Kjeldahl method to the higher
molecular weight fractions is thus subject to increasing
risk of error, while coal itself appears to be the
material most prone to cause difficulty.  In general,
the use of the Dumas method has failed to produce precise
results; methods for correction of the Dumas method are
also a question.

Input solvent and reclaim solvent cuts have been routinely
analyzed by infrared spectroscopy.  In these materials
absorption bands are found at the usual locations for OH
(phenolic) and NH_  (amine) functions.   (The spectra for
oils and coal solution are shown in appendixes E through I.)
These materials have also been titrated by conventional
nonaqueous procedures which determine even weakly basic
functional groups and weakly acidic functional groups,
such as hindered phenols.
When basic material is determined by titration, the result
can be related to the nitrogen present in the sample.  For
                          -98A-

-------
low boiling products from the SRC process, the basic
nitrogen function accounts for most of the nitrogen
reported by the Kjeldahl method.  As molecular weight
increases, the fraction of the total nitrogen accounted
for by titration of basic functional groups declines.
For cut 2 reclaim solvent, this may account for only
30 to 50% of the total nitrogen present.  Efforts to
titrate vacuum bottoms in mixtures of tetralin and
acetic acid resulted in similar estimates for the
proportion of the nitrogen as basic functions in this
material.  The latter results may involve solubility
problems and are perhaps less accurate.

In lower boiling fractions the determination of acidic
functions by titration gives results in which the oxygen
as phenolic functions and oxygen-by-difference were
similar.  As higher molecular weight products are studied,
this result also fails to account for the oxygen completely.
The location of inflection points, as in the titration of
vacuum bottoms dissolved in pyridine, becomes difficult.
Problems with solubility and perhaps with hindrance of
functional groups may be involved.

Evidently another method should be used to check the re-
sults or to replace the Dumas extrapolation method.  Of
                          -98B-

-------
established methods, probably the only remaining is gas-
ification, reported in the article by W. M. Ode just noted,
as an alternative determination for nitrogen in coal.  Ode
refers to Deutscher Normenauschuss, DIN 51722.  The de-
tailed method is not readily available; it is given in
brief form in English by Ode and also ii> brief form in
German elsewhere.    From these two sources, it appears that
the decomposition of 0.5-1.0 g coal for quantitative
yields of contained nitrogen in ammonia is practical.
Application of the gasification method to SRC products
would entail additional studies.  Its application to the
more volatile materials could be troublesome; so the
Kjeldahl method here seems reasonable.

At the present time, therefore, the selection of best
analytical methods and reporting systems for nitrogen in
the heavier materials of the SRC process is, at the least,
in need of further study.

VI-C  GAINS FROM THE STUDY
After the elemental balance study, several facts emerged.
First, the performance of the continuous reactor could be
improved.  To improve uniformity in slurry actually fed,
a campaign of mechanical corrections and control analysis
was conducted.  Sample collection and workup procedures
were both optimized, especially by collecting samples
 3 W. Radmacher.   "Brennstoffe,  feste, flussige."  Ullmans
 Encyklopadie  der  technischen Chemie, dritte  auflage.
 (Munchen:Urban  und Schwarzenberg, 1953).   Band  4, P.  701.
                           -99-

-------
over longer intervals and learning to work up these
larger samples.  As a result, raw material balances
usually close at better than 99% recovery, often better
than 99.5%.

Second, a better control of coal sampling and preparation
was needed.  The considerable attention given has greatly
reduced the risk of segregation or nonuniformity of coal
lots from can to can.  Long experiments are now made with
slurry composition under good control.

Third, the fate of sulfur and nitrogen in the SRC process
seems clarified.  As yet, the interdependence of solvent
composition, solvent reactivity, coal composition,
operating conditions, and equipment design is not fully
elucidated.  But an appreciable part of the sulfur in
the vacuum bottoms fraction, the large source of fuel
value recovered, can be removed.  Removing an appreciable
fraction of the nitrogen seems improbable, perhaps re-
quiring liquefaction under conditions that convert all
material to hydrocarbons of appreciably lower molecular
weight than is usual now in the SRC process.

In addition, certain assessments of reactor behavior and
analytical methods have been assisted by the elemental
balance study.  The problem of mineral behavior was not
solved by mechanical corrections.  When mineral output
has been measured frequently as a function of time, its
concentration in the effluent product has always varied
in a cyclic manner.  For the upflow type of dissolver
being studied, this seems a normal characteristic; the
effect of scaleup is not known, but mineral behavior
                           -100-

-------
may be influenced by initial particle size of the coal
or by such mineralogical factors as the dispersal of
crystallite sizes in the coal minerals.  It further
appears that to collect a set of samples for an exact
element-by-element balance will require good understanding
of the operating characteristics of the larger dissolver
and its auxiliary equipment items.
                            -101-

-------
                        GLOSSARY

aromatic - liquid aromatic hydrocarbons in solvent or derived
           from the reaction with coal; substituted benzenes,
           indene s,naphthalenes, acenaphthylenes, phenan-
           threnes, anthracenes.  The term includes phenols,
           anilines, quinolenes, and pyridines.

ASTM -     American Society of Testing Materials

blackness - determination of visible absorbance; applied
           to coal solutions as an empirical measure of the
           coal which is in solution

cm -       centimeter(s)

coal converted - the amount of coal changed to gaseous,
           liquid, and solid products.   The percentage is
           determined from pyridine insolubles data, by
           either of two methods:

           A.  Insolubles Difference Method

               % feed coal converted = 100 - % pyridine in-
                                             solubles, loss
                                             free basis in
                                             reference to feed
                                             coal

           B.  Ash Enrichment Method

               100  (% ash in feed coal)=(100 - X)  (% ash in
                                                  pyridine
                                                  insolubles)
               where X is % feed coal converted

           Both methods are subject to some error.  The
           first depends on quantitative collection of un-
           reacted material; if handling losses are corrected
           for in the calculation, materials lost must con-
           tain the proper ratio of liquids and solids,
           which is rarely the case.  The second method
           requires that mineral and carbonaceous material,
           not dissolved, remain in the proper ratio; settling
           at any point may segregate mineral preferentially,
           and this will cause error.  The true percentage of
           coal converted probably lies between results by
           the two methods.
                            -102-

-------
           Since different coals may have different
           moisture and ash contents, it is often desir-
           able to reduce conversion to the moisture-
           ash-free (MAF) basis, representing the
           organic fraction of the coal:

           % MAF conversion = A* coi?yer8*?!; on.feed coal basis)( 100)
                              100 -  (% ash+% moisture)

cold trap oil - During vacuum stripping of the reaction
           product, liquids are collected in a cold trap
           cooled with dry ice and acetone.  After thaw-
           ing, a two-phase product is obtained.  The
           upper phase is a hydrocarbon mixture consist-
           ing of materials with an atmospheric pressure
           boiling range from approximately 20-175°C
           with a density of approximately 0.80.  It is a
           mixture of aliphatic and naphthenic hydro-
           carbons, essentially a stabilized naphtha.
           The lower phase is mostly water.

cut 1 oil - a distillate obtained during stripping of
           the reaction product in recent work  (100°C
           at less than 3 mm Hg); during CU 48 it was
           obtained as the 50-100° <3 mm Hg cut after
           filtration; boiling range approximately 175-
           288°C at atmospheric pressure, density
           0.9-1.0.  In the Pilot Plant this cut will
           be used to wash the filter cake; any excess
           may be used as solvent or disposed of as
           product.

cut 2 oil - the vacuum distilled filtrate  (100-230°C at
           <3 mm Hg) reclaimed for solvent; at atmos-
           pheric pressure, the boiling range is 288-
           425°C; density of 1.1 if derived from anth-
           racene oil.  Cut points may be adjusted to
           obtain breakeven quantity for continuous
           operation of the reactor.

g -        gram  (s)

heavy oil - any distillate boiling above 230°C to the
           end point  (maximum 325°C) at <3 mm Hg.  The
           theoretical distillation range at atmos-
           pheric pressure may be 425-560°C; density is
           near 1.2.  This oil is viscous and may be
           solid at 20°C; it may not be isolated in
           the continuous plant  (the cut 2 end point
           may be raised to utilize a fraction as sol-
           vent, leaving the rest in the vacuum bottoms).
                            -103-

-------
hydroaromatic liquid - typical compounds are reaction
           products of aromatic liquids and hydrogen:
           indenes, dihydronaphthalenes, tetralin, di-
           hydrophenanthrenes, tetrahydrophenanthrene,
           dihyroquinoline.  Normally this
           term implies a partly hydrogenated aromatic
           compound.

IR -       infrared

rR -       infrared ratio  (section IV-D-2-d); absorb-
           ance at 2920 cm~Vabsorbance at 3040 cm~l

kg/cm^ -   kilogramXs) per square centimeter

1 -        liter(s)

light liquids - see "cold trap oil"

LHSV -     liquid hourly space velocity  (section IV-A-3)

m -        meter(s)

MFRAO -    middle fraction, raw anthracene oil

meq/min. - milliequivalents per minute  (appendix D)

mg -       milligram(s)

ml -       milliliter(s)

mv -       millivolt(s)

MAF -     . moisture and ash free

OCR -      Office of Coal Research

ppt -      parts per thousand

polymer  - a  compound of high molecular  weight,  derived
           either by  the addition or condensation  of
           many smaller molecules.   In  this context
           coal is  a polymeric condensation product
           with a wide variety of monomeric units.
           For this study, polymer may  result from
           interaction of  solvent and coal  derived
           fragments  to product  higher  molecular
                            -104-

-------
           weight, less soluble materials.  This pro-
           duction is induced by high temperatures or
           too low partial pressures of hydrogen.


pyridine insolubles - the residue of unconverted carbon-
           aceous material (insoluble organic matter) and
           mineral matter from feed coal, which results
           from treating filter cake.  To determine the
           amount of coal converted, the wet filter
           cake must be washed free of imbibed coal
           solution (or residual wash oil, if appli-
           cable) .  A weighed portion of filter cake
           is digested in hot pyridine, filtered, washed
           with hot pyridine until washings come through
           colorless, rinsed sequentially with benzene
           and acetone to remove pyridine, and then dried
           and weighed.  See "coal converted."
R&D -
Research and Development
saturates - typical compounds are cyclohexane, decalin,
           and aliphatic hydrocarbons from ring openings
           or side chain removals
SRC -
Solvent Refined Coal; vacuum bottoms
SHMFRAO -  stripped hydrogenated middle fraction, raw
           anthracene oil

vacuum bottoms - end product after solvent extraction, the
           residue from vacuum distillation of filtered
           coal solution.  It is actually the high molecu-
           lar weight fraction of the coal dissolved in
           the process.

water phases - water collected in the cold trap with light
           liquids, or collected with cut 1 oil and com-
           bined to obtain a consolidated value simulating
           total water yield from the coal process.
           Water originates as water fed with the coal or
           produced by the reaction of hydrogen with
           oxygen-containing functions in the coal.  Water
           phases may contain variable amounts of ammonia,
           carbonate, sulfide, phenol, or other water
                           -105-

-------
           soluble organic compounds available for
           extraction.  A small amount of water also
           collects in the knockout vessel and is
           cold-trapped in the gas collection bag
           as gas is vented and sampled.

wet filter cake - the insoluble materials retained on
           the filter as a cake of solids after fil-
           tration of the stripped coal solution, con-
           taining 40-60% imbibed solution
                            -106-

-------
                   APPENDIX  A
   THE PITTSBURG & MIDWAY COAL MINING COMPANY
            RESEARCH DEPARTMENT
           Analytical Laboratory

 DETERMINATION OF HYDROGEN SULFIDE IN SRC PRODUCT GAS

Data;  This method covers the determination of hydrogen
sulfide in product gas from SRC lab unit.  It is
applicable on a concentration range of about 0.1 to 7%
V/V H2S.  A measured volume of product gas is bubbled
through ammoniacal zinc sulfate solution to remove
hydrogen sulfide.  The amount of hydrogen sulfide in the
absorber is then determined iodometrically.

Special Apparatus;
(a)  A 250 ml Erlenmeyer flask with a two-hole rubber
stopper carrying  (1) a 7-mm diameter glass tube, with a
drawn down tip, extending nearly to the bottom of the
flask and (2) a short 7-mm diameter glass tube/ ex-
tending just a short distance on either end of stopper/
for exit of excess unabsorbed gas.  On the inlet end of
the  (1) inlet-tube, a 1-ft long rubber injection tube
is attached.  On the outer end of the  (2) exit tube,
a small rubber bulb having a small slit cut in it can be
used as a 'flap1 valve to restrict the rapid flow of gas
and to exclude the entrance of air.
(b)  Gas Syringes, 500 or 1000 ml - Hamilton Super
Syringes No. S-0500 or S-1000.

Reagent Solutions;
(a)  Ammoniacal Zinc Sulfate.  Dissolve 50 grams of
zinc sulfate heptahydrate in 250 ml of water, and then
slowly add 250 ml of concentrated ammonium hydroxide
while stirring.  Filter off any precipitate that may
form upon long standing.
                          -107-

-------
                      -2-
 (b)  Hydrochloric Acid 1;1.   Dilute  concentrated HCl
 with an equal  volume  of water.
(c)   iodine Solution (0.05N)  - Weigh  12.8 g oi  resublimed
 iodine crystals  into  a 250 ml beaker.   Add 40  g of
 potassium iodide (KI)  and 100 ml of  water.  Stir until
 solution is complete,  dilute to 2000 ml, mix thoroughly,
 and  store in a brown-glass reagent bottle.  (No need  to
 know iodine solution  normality  exactly if it does not
 change and the same exact amount volume (25.0ml) is used
 in reagent blank and  sample  determinations).
 (d)   Sodium Thiosulfate Standard Solution (0.05N) -
 Dissolve 25 g  of sodium thiosulfate  (^28203 •  5H90)  in
 500  ml water and add  0.01 g  sodium carbonate (Na2co3) to
 stabilize the  solution.  Dilute to 2000 ml and mix
 thoroughly. Standardize verus  potassium dichromate or
 potassium iodate by usual techniques  to accuracy of
 + 2  ppt.
 (e)   Starch Solution,  2%. To 250 ml of boiling water,
 add  a cold suspension of 5 g of soluble starch and  0.025 g
 mercuric iodide. Boil for a few minutes to clear.
 Store in glass-stoppered bottle with undissolved
 mercuric iodide  on  bottom.

 Sampling
 The  product gas  must  be analyzed for H2S as soon as
 possible after receipt.  Hydrogen sulfide can  react
 with any condensed  water vapor  or ammonia and  can come
 out  of the gas mixture.  It  may also dissolve  in the
 sides of the bag and  be lost.  The effect of it possibly
 combining with carbon monoxide  to form carbonyl sulfide
 (COS)  is not known.  It would be preferred if  the gas
 could be sampled at the source.
                          -108-

-------
                   -3-

Procedure;
Transfer 30 ml, by graduate, of the ainmoniacal zinc
sulfate solution to a 250 ml Erlenmeyer flask.  Dilute
with water to about 150 ml and add a 1-1/2 inch stirring
bar.  Put in inlet tube-valved stopper.  Attach filled
gas syringe to injection tube by means of a short piece
of glass or stainless steel tubing.  While magnetically
stirring, slowly inject 100 ml to 1000 ml of product gas
depending on H2S percentage.  (Watch for heavy turbidity
formation as guide to the volume of gas to use in first
test.)   (Caution:  This part of test should be done in
area free of open flames or sparks.  Also if gas contains
much carbon monoxide, it should be done in good fume hood I)

Remove syringe from injection tube and record volume of
gas injected into flask.  Raise rubber stopper just
enough to bring end of inlet tube out of solution and
wash down the injection tube with about 1 ml of 1:1 HC1
and a little water from a wash bottle.  Transfer the
stirring bar, by means of a thief, from the 250 ml E
flask to a 500 ml Erlenmeyer flask containing 25.0 ml
of 0.05N iodine solution  (by pipet) and 40 ml 1:1 HC1
solution  (by graduate).  While stirring continuously,
very slowly pour the contents of the absorbing 250 ml
flask into the 500 ml E flask.  Rinse the 250 ml flask
with about 100 ml water into the 500 ml flask.

As quickly as possible, titrate the solution, while being
stirred, with standardized  0.05N sodium thiosulfate
solution until the solution is yellow.  Then add 2 ml
of 2% starch solution and continue titration to a per-
manent colorless end point.  Record volume of sodium
thiosulfate required for titration.   Sample solutions
are usually turbid at end of titration, blanks are clear.
                          -109-

-------
                      -4-
Run through above procedure, leaving out gas sample, for
reagent blank.  The nature of this test is that the
blanks are equal to or usually higher than the titration
volumes obtained for samples.

The blank and sample tests should be run in duplicate.
The blank values do not change very much; therefore
only a weekly check is necessary if there have not been
any changes in reagents, room temperature, etc.  If
sample titration volume is less than half of the reagent
blank, test should be rerun using less gas sample.
                          -110-

-------
                    -5-
Calculations:
     (Vb- Vs) x Nt x F X 100
                                      (mole %)
               V gas
Where:
     Vfc = Volume (ml)  of sodium thiosulfate used in blank
     V_ = Volume  (ml)  of sodium thiosulfate used in sample
      s
     Nt = Normality of sodium thiosulfate
    V gas = Volume  (ml)  of product gas used in test
     F = Factor in milliliters of hydrogen sulfide per
         milliequivalent of sodium thiosulfate.  It
         is one half of the reciprocal of the molar
         equivalent of one liter of gas  (moles/liter)
         at temperature and pressure of product gas at
         time of  testing.  See "Molar equivalent of
         one liter of gas at various temperatures and
         pressures"  table for ease of calculations.
         An average value of 12.5 can be used in the
         Kansas City Area.
Revised:   July 1971  - REP
                          -111-

-------
               MOLAR EQUIVALENT OF ONE LITER OF GAS AT VARIOUS TEMPERATURES  AND PRESSURES
Values in this
an "ideal" gas,
               table give the number
                calculated by use of
of moles in one
the formula:
              moles/liter =
liter of gas.   Values

             P
                                                                       273
are based on the property of

    1
  22.40
where P is the pressure in millimeters of mercury and T is the temperature  in  degrees  Absolute  (=  °C  +  273)
Pr,-,ur.

610
600
601
670
675
CS3
CSS
C-S3
655
703
702 " '
705
v;-s
710
712
714
710
713
720
724
720
71'3
730 - ...
73"*
*7- A
735
738
740
742
744
745
743
7W
712
7M
710
713
760
7C2
704
700
7CS
770
772
774
775
778
7S3

10'
6 03712
3731
2708
37C5
3825
0.03313
3910
3138
33C7
O.C3;78
4000
4Z-2
4K3
0.04035
4^57
4CC8

O.C'-OSI
41C3
4125
413C
• 0.04U8
4113
4171
 . 2963
0.03979 .
3030
4001
4012
	 4023 	
0.04034
4045
401-3
4037 —
4C7S
0.04089
4100
4111
4122
4133 .
O.CM144
4115
4166 :
4177 ,
• • , • 41C3 •'
0.04199
4210 •
4221
4232
4243
0.04214
4205
427G
4233

20* {
O.C3035
3012
. 3640
35U7
3095
0.03094
3749
3776
3304
3331
"0.03C42 "
3513
3D54
3S75
3SSO
0.03C97
391 a
3530.
3941
0.03D52
3303
3973
3354

0.04005
4017
4028
4039 ' :
'.050 *
i
O.O-'-MI j
4072
4033
4094
4105
0.04115
4127
4133
4149
4!DO
0.04171
4'31 ' -
4192
4M3
4214
0.04W5
4236
4 '.'15
Prcwmre
inrrt-ury
611
660
605
C70
675
CEO
685
650
•695
700
702
704
705
703
710
712
714
716
713
720
722
724
716
723
730
732
734
735
738
740
742
744
746 .
748
760
712
714
7C-3
713
760
762
704
706
763
770
772
774
770
770
730

::'
0 )3'j61
35S3
. 3014
3642
3669
0.33697
3724
3751
3778
3805
0.3381$
3827
3333
3349
3800
0.33370
3031
3392
3502
3914
0.33925
3936
3947
3957
3908
0.33979
3330
4001
4012
4023
0.34033
4044
4015
40C6
4077
0.340(58
4093
4110
'121
4151
0.04142
4113
4104
4175
41BG
0.3'! 197
4207
4718
4240

24'
0.03137
3D51
31S1
3618
3645
0.03072
3099
3710
371?
3780
O.C3790
3001
3812
3823
3834
0.03844
3615
3B36
3SJ77
3883
0.03898
3903
3S20
' 3931
3541
0.03D02
3913
. 3974
JOGS
3995
0.04COS
4017
4C20
4039
4049
0.04060
4071
4052
4033
4103
4114
4125
4130
4147
4153
0.04168
4179
41-30
4201
4211
Trmp«rn
25*
0.03115
3103
3595
3022
0.03049
3070
37C2
37?9
3710
0.03767
3777
3738
3810
0.03S70
3K31
3342
3S13
3SC3
0.03E74
3835
3850
35*30
3417
0.03323
3533
3049
390-3
3971
0.03D81
» 3592
4033
4014
4024
0.04C35
4040
4013
4007
4073
4039
40S9
4110
4121
4132
' 0.04142
4153
4114
4175
41(15
Vure'C
28'
0.034:3
3143
3169
3595
0.03C23
30-19
307S
3720
0.03740
37M
3701
3772
37S3
0.03793 '
3004
3813
3S25
3336
0.03547
3507
3i68
3S78
3CD9
0.03500
3310
3321
33-12
0.03953
3574
33S5
3390
0.04000
4017
4o.'3
40J3
4049
4003
4070
4CS1
4052
4102
0.04113
4124
4;34
.  7
3520
3146
3572
0.03535
3025

37CJ
0.03715
37J6
3735
37<7
3713
0.03703
3779

3S30
3311
O.Or-321
3C32
3342

3303
O.C3374


3315
0.03327
3533
3943
3313
3509 .
0.03930
3301


4CC2
4033
4043
4014
40C5
4075
0.04CS6
4C-C-5
4107
4117
4123



3455
3123
3149
0.03575

3CSO
0.0303!
3701
3712
37?2
3733
0.03744
37M



O.C375G
SC07
MI7

*~~'d
o.c:£<5>


M91
O.CJ001
3512
3-i?2
3533
3543
0.03CM
^ .-. 04


— '*5
40M
4017
40?7
4C38
4043
0.04059
4070
4080
41101

-------
  STANDARDIZATION OF SODIUM THIOSULFATE VERUS POTASSIUM
  IODATE

1.  Weigh, to nearest 0.01 mgm, about 0.065 to 0.070 g
    of pure potassium iodate  (KIO.J , dried at 180°c for
    2 hours.  Record weight as W.   (see Note #1).
2.  Transfer to a 250 ml Erlenmeyer flask.  Add 2.0 g
    of potassium iodide  (KI)  and 50 ml water.  Swirl to
    completely dissolve salts.
3.  Add 10 ml of 1:10 Hydrochoric acid  (HCl) solution.
    Solution will turn dark brown from liberated iodine.
4.  Titrate immediately with  sodium thiosulfate with
    constant swirling.  When  the yellow color of the
    iodine has almost disappeared, add 1 ml 2% starch
    solution.  Continue the titration dropwise to  ,-
    disappearance of the blue color.   (Run a blank deter-
    mination, leaving out the KIO-,) .
5.  Calculation:
                        ^   x 28.03608 = N

    where:  W = Weight, in grams, of KI03
            V = Volume, in ml,  of  sodium thiosulfate solv,

6.  Do in triplicate.  Widest difference should be + 2 ppt,

Note #1  - Instead of weighing three small weighings,
           about 0.65 to 0.70 grams of KI03, weighed to
           nearest 0.1 mg,  may be transferred to a
           250 ml volumetric  flask.  Add water to dissolve
           salt, then dilute  to mark, and mix  throughly.
           Take out 25.0 ml aliquots, by pipet, and
           transfer to 250 ml Erlenmeyer flasks.  Add
           KI and only 25 ml  water and continue above
           procedure.  Use aliquot weight in calculations.
 July  1971 - REP
                          -113-

-------
                    APPENDIX B





        MOLECULAR WEIGHT  OF  PRODUCT  GAS
Run:  	             Date:





Wt of Bottle + Air 	  	



Wt of Bottle       	  	



Wt of Air
Wt of Bottle * Gas



Wt of Bottle



Wt of Gas







Average Wt of Air
Average Wt of Gas
                                     —  3
                                      e-
Molecular Wt=  (average gas wt)	X   28;95=_

               (average air wt)
                          -114-

-------
                  APPENDIX  C
  CONDITIONS USED FOR GAS SOLID CHROMATOGRAPH OF PROCESS

                 GAS SAMPLES


Instrument:  Hewlett-Packard Model 700-231 dual column
             Gas Chromatoaraoh with dual gas sampling
             valves No.  1902OA on each column.
     Conditions
Column dimensions and
  packing(s):
                            Column A
                       4M x 1/4"SS tubing
                       of Linde Molecular
                       Sieve 5A, 60/80
                       mesh
                                      Column B
                                 4M X 1/4"  tubing
                                 of Waters  Associates
                                 Porasil A, 100/150
                                 mesh
Gas Sample Loops:

Flow Rate:

Carrier Gas:

Carrier Gas Pressure:

Detector:


Bridge Detector Current:
                               l.Occ

                           60 ml/min.
                                         2.0cc

                                      30 ml/min,
                                        Argon
                                        50 psig
                           Thermal Conductivity with
                           WX Filaments

                                     100 milliamperes
Temperatures:
    Injection Port
    Oven  (isothermal)
    Detector Block

Chart Speed:
                                     125°C.
                                     100°C.
                                     0.5 inehes/minute
Column A;  Resolves H
              2'  LN2'  ^"4' & co-  (Ethane is
a later interference).
                     2/ o2, N2, CH
Column B;  Resolves H2  (air, CH   & CO not resolved well),
           Ethane, C02, Propane, Iso- & n-Butanes.
                         -115-

-------
                   APPENDIX D

    AUTOMATICALLY RECORDED POTENTIOMETRIC AQUEOUS AND
         NONAQUEOUS TITRATION INFORMATION

In this section, the preparation, standardization, general
techniques, and use of aqueous and nonaqueous titrants
utilizing automatic titration equipment are described
in detail.  In general, the standardization procedures
are the same as for the sample determinations, except that
                   t
primary standard materials are used.  These methods are
used to obtain information for relating the distribution of
nitrogen in amine or other basic functions to total nitrogen
content of solvent and light oil products.  In addition
phenolic compounds can be similarly related to the oxygen
content of such materials.

APPARATUS
See attached "Automatically Recorded Potentiometric
Titration Assembly" drawing.  (The recorder and syringe
pump are set near each other.)  Substitution of eqivalent
pH meters or recorders to make use of existing equip-
ment is usually practical.  The sage pump and an appro-
priate syringe are not expensive items, and substitution
for these items is not recommended.  The same equipment
is used for all of the potentiometric titrations.  Dif-
ferent solvents and titrants are used as described in
the specific procedures.  Automatic titrators, such as
the Sargent Model D titrator or the Precision Dow
"Recordomatic" titrator, can be purchased and should
function as well as this assembly.

NONAQUEOUS TITRANT SOLUTIONS
      General
Working temperatures of the nonaqueous titrant should
approximate that of its temperature during standardization;
                         -116-

-------
                         -2-
otherwise the high thermal coefficient of cubic expansion
of organic solvents will cause the volume delivered to
be in error.

Acetous Perchloric Acid - 0.1 N
To 100 ml glacial acetic acid in a 1000 ml volumetric
flask add 9 ml perchloric acid (72%), followed by 20 ml
acetic anhydride.  Dilute to mark with glacial acetic
acid and let stand overnight before standardizing
versus THAM.  If gels or precipitation occurs when
titrating samples, use methanesulfonic acid.

Acetous Methanesulfonic Acid - O.I N
To 100 ml glacial acetic acid in a 1000 ml volumetric
flask add 6.8 ml of methanesulfonic acid  (Eastman
Organic Chemicals No. 6320), followed by  2 ml acetic
anhydride.  Dilute to mark with glacial acetic acid,
mix and allow to stand overnight before standardizing
versus THAM.   (Reference:  Caso and Cefola, Anal. Chim.
Acta. 21, 374-9   (1959))

Tetra n-Butylammonium Hydroxide  (TBAH or  TBAOH) - 0.1 N
Methanolic l.OM TBAH  (25% in methanol), titration grade,
can be obtained from Southwestern Analytical Chemicals,
Austin, Texas, Cat. No. 304.   (An aqueous l.OM TBAH
solution is also useful in certain cases  (SAC Cat. No. 108.).)
To obtain 0.1 N solutions, 100 ml of the  l.OM solution is
diluted to 1000 ml with a desired solvent or solvent
mixture.  Usually for very weak acids where the alcohol
should be kept at a minimum, a 9:1 benzene-methanol
solution is obtained by diluting 100 ml of the l.OM TBAH
to 1000 ml with reagent grade benzene.  This is the
solution usually used for most titrations.
                         -117-

-------
                   -3-
Tetra n-Butylammonium Hydroxide (TBAH or TBAOH) - 0.1 N
Continued

Occasionally, other solutions are used for specific
problems.  All-alcohol solutions are made by diluting
the l.OM TBAH with either methanol, ethanol, isopropanol,
etc., for special titrations.  Also, a useful solution
of 7:2:1 benzene-isopropanol-methanol is made by dis-
solving 100 ml of l.OM TBAH in 200 ml of isopropanol,
then diluting to 1000 ml with benzene.  Do not use
old solutions that have absorbed carbon dioxide.
(Remarks:  TBAH is not soluble in toluene, so do not
substitute it for benzene.)  Standardize versus benzoic
acid.


Alcoholic Potassium Hydroxide - 0.1 N
To 1000 ml of aldehyde-free anhydrous ethanol, add 10 ml
of 45% W/W aqueous KOH solution.  Standardize versus
benzoic acid or potassium acid phthalate.  Do not use old
solutions that have absorbed carbon dioxide.
Any aldehydes present in the alcohol will interfere by
causing the solution to darken and produce a resinous
precipitate upon standing.  The anhydrous ethanol is
purified as follows:  Reflux 4 liters of anhydrous
ethanol for 3 hours with 30 to 40 g.  of NaOH or KOH
pellets and 25 g.   of aluminum foil cut into strips and
then distill.  Collect the first 100 ml of distillate
separately for some other use.  Distill down to a low
pot volume, collecting the distillate in a brown glass
bottle.  The last portion coming over is collected separately
                          -118-

-------
                            -4-
for some other use.  (Note:  Other alcohols or blends
thereof can also be used in special cases.)

AQUEOUS TITRANT SOLUTIONS
Use usual 0.1 or 0.5 N NaOH or H2S04  standard aqueous
solutions.  Standardize versus potassium acid phthalate
or THAM.  (Other acids and bases at different strengths
versus special standards are also used in special cases.)

STANDARDS
Tris  (hydroxylmethyl) aminomethane  (THAM), primary
standard grade.   (This material is also referred to as
TRIS, or 2-amino-2-(hydroxymethyl)- 1,3-propanediol.)
Obtain from Fisher Scientific Company/ No. T-395 or
Eastman Organic Chemicals No. 4833).

Dry at 60°C in vacuum oven for 2 hours.

Equivalent Weight 121,14.

Benzoic Acid, primary standard grade.   (Matheson, Coleman
& Bell, No. CB1008).  Dry at 60°C in vacuum oven for 2
hours.  Equivalent Weight - 122.12.

Potassium Acid Phthalate  (KHP), primary  standard grade.
 (Mallinkrodt No.  6704).  Dry at 60°C in  vacuum oven for
2 hours.  Equivalent Weight -  204.22.

SOLVENTS
All solvents used should be reagent grade.  They should
be protected from acidic or basic fumes.   If  a solvent,
such  as pyridine  or DMF, should give a high blank it
should be purified before use.  Other  special titrants,
standards,  solvents, etc. needed will be discussed  in  de-^
tail  in their  specific procedures.

                            -119-

-------
                       -5-

GENERAL TITRATION PROCEDURES

(See attached "Automatically Recorded Potentiometric
Titration Assembly" drawing.)
Recorder Settings
Set recorder range at 50 mv  and chart speed at 1 inch
per minute. Other adjustments are as per manufacturer's
instructions so that whole span of meter just covers
chart span.  (Different conditions may be necessary
for special cases.)
Meter Settings

Set at 700 mv  setting for small inflection points, at
1400 mv  setting for large inflection points, or for pH
readings if this type of information is desired.
Other settings as per manufacturer's or subsequent
instructions.
Titration Techniques

Turn on pH meter and recorder and allow to warm up.
Remove 50 ml syringe with attached needle from syringe
pump.  Insert the Teflon needle into required titrant
solution and fill syringe by slowly pulling plunger out
with a slight rotating motion.  After end of plunger has
passed the 50 ml mark, invert syringe so that the tip is
upward.  Slowly push the plunger upward to push all air
bubbles out of syringe and needle.  Replace syringe
in   syringe pump holder.  Position driving carriage on
pump platform to engage mating racks and gears and so the
end plate of carriage is next to outer end of syringe
plunger.  Turn on syringe pump until titrant starts to
flow from needle, then turn it off.
                         -120-

-------
                            -6-
Place a 250 ml beaker containing a magnetic stirring
bar on the magnetic stirrer.  Add ca. 150 ml of specified
solvent, using a graduate.

Weigh sample, liquids in suitable hypodermic syringe and
solids in weighing scoop, and add to measured solvent in
beaker.  Some pasty,   sticky semi-solid samples may have
to be weighed in the beaker directly, then stirring bar
and measured solvent added.

Place the rubber stopper containing the electrodes in
the beaker.  Turn on stirrer and stir to dissolve sample.
Insert Teflon needle through glass tube until end is
submerged to midway solution.  Adjust the pH meter
by releasing the "Zero" button and use "Standardization"
knob to set meter pointer at desired end of scale.
For acid samples, move pointer to about pH 1 position
and for basic samples, move pointer to about pH 13 position,
Adjust the recorder so that pen is near the left edge
of the chart for acid samples and near the right edge of
chart for basic samples.   (See attached example
standardization curves.)  Start the chart drive of the
recorder and when the pen is on a major  (heavy) line
on the chart, simultaneously start the syringe pump.
Mark this major line "Start."  The recorder will now
produce a chart showing the EMF as a function of titrant
added.  Run the chart until all of the inflection points
have been passed; turn off stirrer and pump, and turn
recorder to "Stand-by1.1  Remove Teflon  needle from
beaker, turn pump on for a few seconds to wash out any
solution that may have diffused into the titrant in the
needle.  Remove electrodes and rinse clean with a
suitable solvent and store in water.
                            -121-

-------
                         -7-
Inflection Points

The chart will have one or more inflection points which
indicate the quantity of titrant required to titrate
the components of the sample.  The significance of several
inflection points is discussed in detail in the specific
procedures.
The inflection point of the curve is taken as the end point
of the titration.  With experience, the inflection point
 (i.e., change of slope in the middle of the break) is
readily found by visual inspection.  Count the minutes
of time from "Start" of titration to each inflection
point.  Designate first inflection point time as T, t
second inflection point as T2, etc. Tlf therefore,
represents the time to titrate the first component
titrated and T^-T-I represents the time to titrate the
second component, etc.
Notes

1.  With titrations which run several inches of chart
length, results are precise and accurate to better than
1%  of  error.   This  error  can be  reduced  to  a  few parts
per thousand if the sample size and rate of titrant
delivery is properly set.
2.  With too small a sample or with trace quantities,
precision will suffer.  With major and minor quantities
it is likely that one precise and one approximate result
will often be obtained.  This can be corrected by
running another sample with appropriate adjustment in
sample size if the importance of the result justifies the
extra labor.
                         -122-

-------
                       -8-
3.  Because of the mechanics of the titration, some
reagent must be injected and stirred into the beaker
to produce an inflection.  In the case where the true
result should be zero, the inflection is obviously late.
It is possible to differentiate between a nil and a trace
by increasing the sample size until a distinctive curve
is produced.  In non-critical applications, trace or
nil curves should be reported nil to the maximum indicated
by the inflection observed.

4.  The usual titrant delivery rate is about 3 to 4
ml/min.;  however in certain cases  (e.g., small amount of
sample available, low solubility, slow reaction, etc.)
it may be necessary to adjust to a slower rate.

STANDARDIZATIONS
Standardization of titrant is done under the same
conditions under which the sample determination is to
be done.  To eliminate any solvent blank influence, the
solvent and volume of solvent is kept the same for both
standards and samples.   (Note:  If solvent has a very
high blank, it should be pretitrated to a specified
indicator color change for both standards and purified
samples if possible.)
If the titer of the titrant is determined in the
units of milliequivalents  per minute  (meq/min.), the
delivery rate of syringe or true timing of chart speed
does not need to be determined; they only need to be
constant.
The dried primary standard is weighed and brushed in the
titration beaker containing the measured, selected solvent.
It is titrated in the method described above, with the
titrant to be standardized.  The weight of standard
should be such as to give at least  a 10 to 12 minute
titration. Titrant solutions should be restandardized
daily, each shift, or when any changes occur.
                          -123-

-------
                       — Q —
The titer of a titrant solution is calculated as
follows:
   W
T   E w—  = Titer in milliequivalents per minute

             (meq/min.)


where:

   W     =Weight, in milligrams, of primary standard
   T     =Time, in minutes, to inflection point
   E.W.  =Equivalent weight  (mg/meq) of primary standard
   f     =Titer in (milliequivalents/min.)  meq/min.
                          -124-

-------
                       -10-
Remarks
There are times and conditions where conventional primary
standards do not give comparative standardization
results.  For this reason, purified samples of the
sought component are used as standards.  The titer is
then calculated as  mg   of material per minute which is
easily calculated as follows:
           	mg  of material             .
             Time of inflection point=   m9/™-n. titer

Standardization Examples
Tetra-n-butylammonium hydroxide solution is standardized
versus benzoic acid.  Weigh accurately about 400 to 420
mg of primary benzoic acid and brush into solvent in
beaker.  Titrate as per general procedure for acid
samples.  See attached "Standardization Curve of TBAH
versus Benzoic Acid."  The, equivalent weight of benzoic
acid is 122.12.
The TBAH solution may also be standardized against
standard O.1000 N sulfuric acid.  Accurately pipet 10 ml
of standard 0.1 N sulfuric acid into a beaker.  Dilute
to 150 ml with reagent grade acetone.  Titrate as per
general procedure for acid samples.  Measure distance
between the first and second inflection points and
determine the titer of the titrant in meq /min.
Perchloric acid solution is standardized versus THAM in
acetic acid.  Weigh accurately about 380 to 400 mg  of
primary grade THAM and brush into acetic acid in beaker.
Titrate as per general procedure for basic samples.
See attached, "Standardization Curve of Acetous Perchloric
Acid versus THAM."  The equivalent weight of THAM is 121.14
                          -125-

-------
              SELECTED BIBLIOGRAPHY



The literature of non-aqueous titrations has become so

enormous in recent years that a complete bibliography
is beyond the  scope of this discussion.  The following

textbooks and review articles and the references contained
therein should enable the interested reader to pursue
further any branch of non-aqueous titration technology.


Textbooks

     Fritz, James S.,  Acid-Base Titrations in Nonaqueous
     Solvents.   The G. Frederick smitn Chemical Co.,
     Columbus, Ohio, 1952.

     Kucharsky, J. and Safarik, L.,  Titrations in Non-
     aqueous  Solvents.   American Elsevier Publishing Co.,
     New York, 1965.


Literature Review Articles

     Analytical Chemistry  (Reviews), "Acid-Base) Titrations
     in nonaqueous Solvents."
     Riddick, J. A., 24, 41  (1952); 26., 77   (1954);
     2£ 679  (1956); 3_0, 793  (1958); 32_, 172R  (1960).

     Streuli, C. A., 34_, 302R  (1962); 3£, 363R  (1964).

     Harlow,  G. A. and Morman, D. H., 38, 485R  (1966).

     "Potentiometric Titrations" are also reviewed in a
     like manner starting in 1954.


These series  of reviews of  analytical interests are to

Be continued  every other year.
                          -126-

-------
               TITRATION ASSEMBLY
 Meter
             Covered ground-shielded 2 wire connection
Recorder
METER, pH-millivolt - Leeds  & Northrup,  Model  7664  or
7401, or equivalent.
MAGNETIC STIRRER - Magnestir, or  equivalent.
MAGNETIC STIRRING BARS - Teflon-coated,  1 7/16"  x 3/8",  or
equivalent.
BEAKERS - 300 ml, tall-form  without spout.
RUBBER STOPPER - No. 13 with suitable holes cut  for electrodes
and glass tubes.  Glass tubing  is 6mm OD and 4 mm ID,  one
straight for Teflon needle and  a  "L" shaped one  for nitrogen
inlet tube.
GLASS ELECTRODE - Leeds & Northrup #Std. 119-30  or 117163.
Store in water, not organic  solvent, when not in use.   (A
platinum electrode may be used  here instead).
CALOMEL REFERENCE ELECTRODE  - L & N #Std. 119-31.  This elec-
trode is modified as follows:   Prepare an agar-KNO3 gel by
placing 1.5 gm. of powdered  agar  in a 100 ml beaker, add 50 ml
distilled water, heat on a  steam  bath with stirring until
                         -127-

-------
dissolved.  Add 5 gms of potassium nitrate to the beaker and
stir until dissolved.  Cool the mixture until solidified, and
store in a clean capped bottle.  Unscrew, remove, and wash
the salt bridge tube of the electrode.  Place a few small
pieces of the gel in the tube, hold with test tube clamp in
steam or hot water bath until get is liquidified.  Depth of
gel should be about 5mm.  Cool to solidify gel.  Add satur-
ated potassium chloride to tube and screw into electrode as
per manufacturer's directions.  Store in water, not organic
solvent, when not in use.
SYRINGE PUMP - Sage Instrument Co., Model 249-2 with alternate
gear set Model 259.  Using a 50 ml syringe, the rate of de-
livery is approximately 3 ml per minute.
SYRINGE - Hypodermic, glass, 50 ml.  Any conventional brand
with lock tip or glass tip and reducer.  The "0"-ring type
should not be used with organic solvents.
NEEDLE - Teflon tubing 2 ft. long with Kel-F hub.  Kontes
Glass Co. #K-30818.
RECORDER - E. H. Sargent Co., Model SR or SRL is preferred,
however the Heath Built Servo Recorder #EUW-2OA with EUA-20-2
motor has been found satisfactory.  Chart speed is set at 1
inch per minute by selective drive motor assembly or chart
speed control.  Recorder is connected to meter by a suitable
covered ground-shielded 2 wire cable.  Meter, stirrer, syringe
pump, and recorder are to be connected to a common ground,
such as a water pipe.
                        -128-

-------
   Standardization-Curve of Tetra n-Butylajtrronium Hydroxids  (TBftH) vs.   Benzoic Acid

"• ~ys^wii.^Sl^i^i^..'Cl6i^f.. Anaiyiicsi K<,.j$fa4**A'.   •. ._^.':;.V;,..': :.:x/::^.^::i;:j  '

  -Wc.™

          ••?7n,ff.      •   • •-*•
       32.12 mg/Meq X 7.22. WLH.    ..
                   ai  of Bewtilc Acld'fwp 10

                            "
                                   ltvolts         '
                                              -129-

-------
Standardization Curve of Acetous Perchloric Acid   vs. Tris  (Hydroxymethyl) AmincmetharK

  O)
  
-------
 PAGE NOT
AVAILABLE
DIGITALLY

-------
                                                           K-tr 10 X '9 TO 5/C INCH . T i*? X )7 I
                                                            £ KCUTCL • C'.-C" CO. •ori.wi.
 i
• •
:
                   BLACKNESS 3 |f
                                                       mim          in I.-
                                                                           '
                                                                                                •:;  .;; ;• | :- :.:        • | , i , i ;:: | j f;!
                                                                                                                     ,
                                                                                              rHj7! I rr|Trj'!7

                                            :,;>::•;: \~^-,'\ •'•", r; L;::j:':r1.;.M.U:.;.f|;.]4-|-i.
                                 " -H ' ' -i-i H-l-'-1'•'•:•• i ;-- • • .-•; -| •-^-•-• Ir!'.-.-; H :'M '•': ^
.,.;.;.!,:!:''::; j.,,,
(-U...J.I i .  I f '-I-! I . r'..
                                                    . . I .    I • I  ' I i I . . 1
                                                     iii  i    11!   i

                                                                                                  14   15   16   17   18   19    20   21
                                                                                                                                                    3.0
                                                                                                                      IR
                                                                                                                   2.0
                                                                                                                                                    1.0
                                                Appendix  J   Blackness  &  TR in Successive .Samples,  CU  80

-------
 I
H
CO
                                                                        APPENDIX K
                                                          FIRST TRIAL ELEMENTAL BALANCE, CU 48A1

MATERIAL
IKPUT.
Feed Coal
Feed Solvent
Feed Water
Feed Hydrogen
TOTAL
Wt.
GRAMS

322.3
676.8
20.9
11.2
1031.2

% C

56.34
90.58




% H

4.57
6.04
11.19
TOO.


% N

1.29
0.60




% S

5.12
0.36



OUTPUT - PRIMARY SEPARATION - CORRECTED TO 100% RECOVERY OF FEED
Gas
Knock Out Pot
Cold Trap HpO
Cold Trap Oil
Filtrate
V.'et Filter Cake
TOTAL
2 RECOVERY
37.03 See table 9
1.1 1
24.2
8.0
763. 0
192.9
1031.2
100


82.68
90.36
62.32




12.04
6.25
3.89




0.38
0.212
0.981
0.895



0.547
0.821
0.65
0.45
4.66


OUTPUT - SECONDARY SEPARATION - CORRECTED TO 100% RECOVERY OF FEED
Gas
Knock Out Pot
Cold' Trap H?3
Cold Trap Oil
Cut 1 Oil
Cut 2 Oil
Heavy Oil
Vacuurr. Bottoms
Pyridine Insol.
TOTAL
'/:• RECOVERY
37.03 See table 9
1.1
24.7
8.0
29.9
631.6
15.3
185.5
97.8
1030.9
100
*•

82.63
37.47
91.31
85.35
87.50
33.61



12.04
7.99
6.02
5.15
5.23
1.67


I

0.33
0.212
0.566
0.765
1.511
2.040
0.644


0.547
0.821
0.55
0.20
0.40
0.815
0.832
8.128



% ASH .

16.04










27.51










0.179
52.88
GRAMS
C

- 213.8
613.0


826.8

15.27


5.78
693.96
120.21
835.22
101

15.27


5.78
26.33
575.53
15.19
159.07
30.52
1 828.78
• 100.2
GRAMS '
H

14.72
40.90
2.34
11.20
69.16

11.83
0.123
2.707
0.843
48.00
7.50
71.00
102.6

11.83
0.123
2.707
O.S43
2.405
38.010
0.906
9.508
1.516
67.90
93.2
GRAMS
N

4.157
4.051


8.212



0.092
0.017
7.535
1.726
9.370
114



0.092
0.017
0.169
4.830
0.245
3.782
0.632
9.770
118.9
GRAMS
S

16.501
2.436


18.937

5.908
0.006
0.199
0.052
3.456
8.989
18.61
98.3

5.908
0.006
OJ99
0.052
0.060
2.526
0.125
1.635
7.973
18.542
97.9
GRAMS
ASH

51.70



51.70






53.10
53.10
102.7








0.331
51.875
52.206
101

-------
 I
H
W
00
 I
                                                                             APPENDIX L
                                                              FIRST TRIAL ELEMENTAL BALANCE, CU 48A2
MATERIAL
INPUT
Feed Coal
Feed Solvent
Feed Hater
Feed Hydrogen
TOTAL
ttt.
GRAMS
317.9
667.5
14.6
11.2
Torn
OUTPUT - PRIMARY SEPARATION
Gas
Knock Out Pot
Cold Trap H20
Cold Trap Oil
Filtrate
Wet. Filter Cake
TOTAL
% RECOVERY
34.23
1.6
20.0
7.0
780.6
168.2
1011.2
ICO
% C
66.34
90.58
- CORRECT!
See tabl<
85.41
89.63
59.27
(Probable
X H
4.57
6.04
11.19
100.
:o TO 106
j 10
11.28
5.94
3.54
i Error -
% N
1.29
0.60
I RECOVER1
0.485
0.219
0.979
0.868
DifficuH
% S .
5.12
0.36
t OF FEED
1.25
0.841
0.886
0.46
(4.772)
t Decorcposl
OUTPUT - SECONDARY SEPARATION - CORRECTED TO 100% RECOVERY OF FEED
Gas
Knock Out Pot
Cold Trap HpO
Cold Trap Oil
Cut 1 Oil
Cut 2 Oil
Heavy Oil
Vacuum Bottoms
Pyridine Insol.
TOTAL
2 RECOVERY
34.23 .
1.6
20.6
7.0
30.1
627.3
17.6
181.8
90.8
TO 11. 2
100
See table
86.41
88.05
91.73
87.26
87.32
32.02
5 10
11.28
7.71
6.03
6.24
5.11
1.73
0.485
0.219
0.560
0.734
1.627
1.909
0.590
1.25
0.841
0.886
0.244
0.377
0.842
0.944
8.930
% ASH
16.04
30.16
tion of W
0.133
56.07
GRAMS
C
210.9
604 -.6
8T575
12.136
6.03
700.0
99.60
loois
.F.C. Mine
12.136
6.05
26.50
575.42
15.36
158.74
29.07
823.3
100.9
GRAMS
H
14.53
40.317
1.633
11.2
12.60
0.176
2.238
0.789
46.37
5.95
68.12
100.6
rals)
12.60
0.176
2.238
0.789
2.321
37.82
1.10
9.289
1.57
67791T
100.3
'GRAMS
N
4.100
4.005
8.105
0.097
0.015
7.642
1.459
9724T
114
0.099
0.015
0.163
4.603
0.286
3.470
0.536
9.177
113
GRAMS
S
16.21
2.403
5.515
0.020
0.168
0.062
3.590
(8.026)
93^5
5.515
0.020
0.173
0.062
0.073
2.364
0.148
1.716
8.108
97'.7
GRAMS
ASH
50.99
50.99
50.73
50771
99.7
0.24
50.91
5T7T5
100.4

-------
                                TECHNICAL REPORT DATA
                          (Please read Instructions on the reverse before completing)
 . REPORT NO:.
EPA-650/2-75-011
                           2.
                                                      3. RECIPIENT'S ACCESSION NO.
4. TITLE AND SUBTITLE
Sulfur and Nitrogen Balances in the Solvent Refined
Coal Process
                                   5. REPORT DATE
                                   January 1975
                                   6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S).
         C. H. Wright, The Pittsburg $ Midway Coal
Mining Co. ,  Merriam, Kansas  66202
                                                      8. PERFORMING ORGANIZATION HEPOHT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Office of Coal Research, U.S.  Dept of the Interior
2100 M Street NW
Washington, DC 20037
                                   10. PROGRAM ELEMENT NO.
                                   1AB013; ROAP 21ADD-025
                                   11. CONTRACT/GRANT NO.
                                   EPA IAG D4-0454
                                   (Task 1)
12. SPONSORING AGENCY NAME AND ADDRESS

EPA, Office of Research and Development
NERC-RTP, Control Systems Laboratory
Research Triangle Park, NC  27711
                                   13. TYPE OF REPORT AND PERIOD COVERED
                                   Final (Task)	
                                   14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
16. ABSTRACT,
         The report gives results of an exact elemental balance study of the Solvent
Refined Coal (SRC) process that was conducted with the laboratory reactor to deter-
mine the fate of sulfur and nitrogen  in the SRC process. The work was performed in
late 1972 as part of a normal experiment with Kentucky No.  9 high volatile B bitu-
minous coal and a blend of processed anthracene oils under  1000 psig hydrogen pres-
sure. A variety of technique studies had been made in preparation,  such as investi-
gation of the effects of sample size on analysis and methods of handling all samples
of input and output material for maximum recovery and representative composition.
Accounting for  carbon and hydrogen was accurate, for sulfur good,  and for nitrogen
poor. A detailed comparison of conventional Kjeldahl and Dumas analytical results
for nitrogen in  coal and solid products revealed that input nitrogen is not fully repor-
ted by Kjeldahl and that sample size affects nitrogen results reported by Dumas.
Nitrogen analysis needs further investigation.  The study forced review of sampling
and handling techniques as well, with salutary results in laboratory work.
 7.
                             KEY WORDS AND DOCUMENT ANALYSIS
                 DESCRIPTORS
                                           b.lDENTIFIERS/OPEN ENDED TERMS
                                                c. COSATI Field/Group
 Air Pollution
 Coal Preparation
 Liquefaction
 Sulfur
 Nitrogen
 Analyzing
Cleaning
Solvents
Air Pollution Control
Stationary Sources
Solvent Refined Coal
13B,  13H
081, 11K
07D
07B

14B
18. DISTRIBUTION STATEMENT

 Unlimited
                       19. SECURITY CLASS (This Report}
                       Unclassified
                         21. NO. OF PAGES

                              149
                       2C

                      1
20. SECURITY CLASS (Thispage)
"nclassified
                         22. PRICE
EPA Form 2220-1 (9-73)

-------