LP GAS CONTAMINATION REMEDIATION AT

BULK TERMINALS PROJECT

FINAL REPORT
Prepared for the
Propane Education & Research Council
(PERC Docket #11649)

August 18, 2006

By: Nolan Sambrano, Jared Meyer, and Alex Spataru

ADEPT SCIENCE & TECHNOLOGIES, LLC

51 Rover Boulevard
Los Alamos, NM 87544 USA
Telephone: (505) 672-0002
Fax: (505) 672-0003

decont046_final report rev5a.doc

 Table of Contents

I. Abstract .........................................................................................................3
II. Glossary and Abbreviations ...........................................................................3
III. Introduction ................................................................................................4
IV. Residue Testing Methods...........................................................................5
V. First Small Scale Decontamination Test ........................................................6
A. Summary....................................................................................................6
B. Objectives ..................................................................................................6
C. Experimental Set-up and Materials.........................................................6
D. Procedure ...............................................................................................8
E. Results and Discussion ............................................................................11
F. Proposed Next Steps ...............................................................................13
VI. Second Small Scale Propane Decontamination Test...............................14
A. Summary..................................................................................................14
B. Objectives ................................................................................................14
C. Experimental Setup and Materials........................................................14
D. Procedure .............................................................................................15
E. Results and Discussion ............................................................................17
F. Proposed Next Steps ...............................................................................20
VII. Full Scale Column & Process Design.......................................................20
VIII. Economic Analysis ...................................................................................22
IX. Field Service Recommendations..............................................................23
A. Pre-process Wetting and CO2 Pre-loading...............................................23
B. Flowrate ...................................................................................................23
C. Breakthrough and Activated Carbon Replacement...............................23
D. Filters....................................................................................................23
E. Ethyl Mercaptan Caution ..........................................................................24
X. Operating Options........................................................................................24
XI. Conclusions..............................................................................................24
XII. Future work ..............................................................................................25
XIII. Acknowledgements ..................................................................................25
Appendix 1..........................................................................................................27
Appendix 2..........................................................................................................29
Appendix 3..........................................................................................................50
Appendix 4..........................................................................................................53

2
I. Abstract
Adept Science & Technologies, LLC (ASCENT) has tested activated carbon
adsorption columns as a means to remove soluble heavy-end contaminants from
LP Gas. Such contaminants lead to operational problems in some LP Gas
applications. The composition of such LP Gas contaminants was also
determined as part of this project. Process design parameters (i.e. expected
flowrates and bed dimensions) were subject to constraints imposed by industry
practices (to facilitate subsequent field implementation).

Initial results show effective removal of contaminants representative of those

commonly found in commercial LP Gas. Also examined was the effect of the
process on odorant (i.e. ethyl mercaptan) content. Results are presented along
with an assessment of the tested processes. Background information on
analytical methods required for these tests is also provided. Follow-on steps are
necessary to build on and test commercial applications of this work.

II. Glossary and Abbreviations

AC Activated carbon
ASTM International An international voluntary standards organization
Bed-Volume The adsorption column volume occupied by packed activated
carbon or other adsorbent material.
Breakthrough Curve A plot of contaminant concentration in the product stream (after
treatment) vs. time (or volume or bed volumes).
C20+ Heavy-Ends Refers, in this report, to hydrocarbons with 20 or more carbons
or which boil at or above the boiling point of normal C20H42 per
ASTM D2887 simulated distillation, and are the primary
component of vaporizer deposits.
CEC Cosmo Engineering Company, based in Japan
DOT U.S. Department of Transportation. U.S. Federal agency which
specifies and oversees design and on-road transportation
requirements.
DSI Dixie Services, Inc. in Galena Park, TX - the laboratory used for
all analytical work in this project
Fines Very small activated carbon particles (carbon dust) which may
escape the column.
Heavy-Ends Residue Any soluble LP Gas hydrocarbon impurities larger than
pentane. The term includes soluble large organic materials
such as phthalates and similar plasticizers and materials
extracted from gaskets, sealants, and hoses in LP Gas
systems.
micron 1 x 10-6 m (3.9 x 10-5 in). Metric unit equivalent of mesh size
used as a rating of filter porosity.
“Special Duty” LP Gas Category term used in ASTM D1835 to identify LP Gas for
automotive applications. “Special Duty” specifications are
similar to those listed in GPA 2140 (also known as “HD-5”).

3
Mixed measurement unit systems are used throughout this report. In some
cases, for ease of comparison alternate units are provided, but generally values
are reported in units as originally measured or specified.

III. Introduction
A growing LP Gas industry concern is the effect of heavy-end residues
contamination of LP Gas streams on the proper functioning of fueling systems
and on other LP Gas distribution system components. Premium quality LP Gas,
with low heavy-ends content is costlier and scarcer than most commercially
available LP Gas. Sources for such cleaner LP Gas (mainly from natural gas
processing as opposed to refineries) are geographically limited in number.

This project is to quantify residue content in commercial and/or “Special Duty” LP

Gas and to design and test an adsorption column/filtration process to remove
“heavy-end” contaminants to make “on-site” a premium quality LP Gas. The goal
of subsequent work is to have a commercially viable bulk storage plant heavy-
ends decontamination unit.

Activated carbon has been tested and used before as a means to lower residue
content in LP Gas. One example is a patented process using a packed activated
carbon column to effectively remove residual oil while retaining mercaptans.1
Another, a Japanese government funded, multi-million dollar project conducted
by Cosmo Engineering Co. (CEC) yielded a 2 ton/h (16 gal/min) demonstration
plant to remove residues from LP Gas.2 Principals from both projects were
consulted for this work. Petrolane has also used (for a limited period of time
~1980) activated carbon to reduce residues in LP Gas. No data was available for
the Petrolane project.

To properly address the challenges to remove LP Gas heavy-end contaminants,

it is necessary to know what the levels of these residues are likely to be, how
they are to be measured, and to what average level they must be consistently
reduced. International LP Gas maximum residue limitations (Table 1) provide
one measurable and accessible basis to assess the level to which the
decontamination unit should reduce the residues content (for current and near-
term applications like LP Gas-powered forklift trucks).

Table 1: Summary of LP Gas Residues Limits

Evaporation
Nation LP Gas Specification Residues Limit (ppm) Test Method
Temp. (°C/°F)
3
US HD-5, ASTM D 1835, GPA 2140 500 (and oil stain from 0.3 mL of ASTM D 2158
residue-solvent mixture) 38/100
California HD-10, CCR §2292.6 “ “
4
Japan Utility Grade 12 JLPGA-S-05T -
Australia ALPGA Automotive Spec. 2000 20 JLPGA-S-05T/86 105/221
100 (proposed reduction to 50
Europe EN-589 EN ISO 13757 105/221
ppm limit)

When comparing residues limits, it is essential to note the temperature of the test
method used to measure residues content. Methods calling for higher
temperatures during evaporation of the LP Gas will drive off more of the lighter
compounds (such as gasoline and lighter diesel fuel components), which is likely
4
to result in underreporting of the total residue levels and of these lighter heavy-
ends. However, these higher temperature methods can adequately indicate the
presence of C20+ heavy-ends which are the primary cause of some types of
operational problems in applications such as forklifts.

As present and near-term applications (forklift trucks, microturbines, and fuel

cells) will require cleaner LP Gas to function properly and to meet emissions
reduction mandates, it was preliminarily decided that the unit to be designed and
tested is to consistently achieve average residues content below 25 ppm in the
cleaned LP Gas product (which is over one order of magnitude lower than the
current allowable American standards limit). Of note are reported field problems
caused by as little as 3 – 10 ppm of “oily residue”.

IV. Residue Testing Methods

The selection of a suitable residue testing method is crucial to this project. To
yield meaningful results, an applicable test method must have repeatable
resolution on the order of 10 ppm; be relatively easy and safe to conduct; should
require a minimal sample volume; and be reasonably priced.

The evaluations of five (5) standard in-use, and near-term proposed pertinent
testing methods for use in this project are summarized below:
(1) ASTM D 2158 is the current method prescribed by ASTM D 1835. It
was found to underreport total residues, to have undesirably low
resolution, and no means to determine residue composition.
(2) Dixie Services Inc. (DSI) P397 provides resolution of 0.1 ppm. Its
drawbacks are that it requires a large sample volume and the
heating of residues (leading to loss of lighter residues).
(3) EN ISO 13757 reports insufficient repeatability at high resolution,
and requires heating of residues.
(4) CEC method5 uses gas chromatography after extraction of residues
with a solvent under pressure. The method, currently under review
by the D02.H Subcommittee of ASTM, calls for results reported to
0.1 ppm. However, this method currently does not have published
repeatability and reproducibility data.
(5) prEN 15470:2006 (E) [previously referred to as CEN/TC19 N 1221
– Annex E] (a proposed replacement for EN ISO 13757), is a gas
chromatographic method with results reported to 1 ppm, with 95%
repeatability and reproducibility at 20 ppm of 5 ppm and 12 ppm
respectively. Heating of residues in this method underreports
lighter contaminants. This method is also under preliminary
consideration by ASTM Subcommittee D02H.

After evaluating the above test methods, prEN 15470:2006 (E) and the CEC
method were considered the most suitable methods for subsequent work.
Because of the immediate availability of the necessary equipment, with the
advice of DSI, it was decided to use prEN 15470:2006 (E), but with a lower
evaporation temperature of 25 °C (77°F) (See Appendix 4 for test method).

5
V. First Small Scale Decontamination Test
A. Summary
Tests were conducted on one type of activated carbon to treat a liquid phase
propane feedstock containing four (4) selected contaminants representative of
those previously found in commercial LP Gas. “Propellant grade propane”
(~99.6% pure C3H8 with a heavy-ends specification of less than 1 ppm∗ and
already processed by molecular sieve media to eliminate water and sulfur) was
used as a base to which pure, known contaminants were added. A fifth
contaminant, n-octane, was selected as representative of lighter contaminants
(such as gasoline-range compounds) which have been problematic in the
propellant industry. Two (2) polyester filters of different mesh sizes were placed
downstream of the activated carbon to obtain an initial assessment of the amount
of effluent activated carbon fines that may be entrained by LP Gas flow through
the test vessel.

B. Objectives
(1) Determine the effect of a specific type of activated carbon on
removal of selected concentrations of “heavy” contaminants from
propane.
(2) Measure effluent activated carbon fines generated at a given
process flowrate.
(3) Measure process temperatures and pressures to better predict
process conditions (particularly during wetting procedure) in follow-
on experiments.

C. Experimental Set-up and Materials

1. Activated Carbon Vessel
A 26.5-in (67.3 cm) long column with an inner diameter of 2.87 in (7.29 cm)
[internal cross-sectional area ~0.045 ft2 (41.7cm2)] with removable end caps, and
rated for 250 psig (17.2 bar) was designed and constructed (Figure 1). The
length of this vessel allows for variation of bed lengths in experiments (ranging
from a bed size occupying nearly the full length of the vessel to a shorter one
likely to be more representative of the proportions used in a full-scale column).
The activated carbon bed length can be shortened by use of additional inert
support balls to occupy space. The column was fitted with external pressure and
temperature gauges and pressure relief valves. Filter housings were added
downstream of the column. The design also includes interior mesh screens
which support the column packing materials (inert support balls and adsorbent).

∗
This residue specification is the requirement of the propellant manufacturer for its incoming
feedstock propane. The analysis method used by the propellant manufacturer is ASTM 1945 with
a modified injection technique using a 4-port liquid valve (Appendix 4).
6
 a) Vessel b) Filter Stand Setup c) Top Valve, Instrument, and Filter Detail
Figure 1: Pilot Activated Carbon Filter & Experimental Setup

2. Activated Carbon
CEC previously developed a metal-impregnated activated carbon designed to
remove heavy-ends from LP Gas and which could then be regenerated. This
activated carbon mix, because of its precious metal content, was determined to
be prohibitively expensive.

Three (3) commercially available activated carbons types were obtained for
possible use. Of these, a 12 x 40 mesh (1680 x 400 micron) bituminous coal-
based granular activated carbon was chosen for initial tests based on its cost
effectiveness and reported past effectiveness to remove C12 – C21 range
contaminants from LP Gas (where LP Gas storage vessels were contaminated
with diesel). Additionally, the activated carbon manufacturer reported that the
selected activated carbon may work well to remove heavy-ends contaminants
while the removal of sulfur compounds (including ethyl mercaptan) would be
negligible. Regeneration or disposal by the activated carbon manufacturer is
currently the preferred method to replace the spent adsorbent.

3. Support balls
1/8” (0.318 cm) and 1/4” (0.635 cm) ceramic balls were used to support the
activated carbon and to evenly distribute LP Gas flow inside the vessel.

4. Filters
100 micron (149 mesh) x 0.100” (2.54 mm) thick and 5 micron (2500 mesh) x
0.080” (2.03 mm) thick polyester felt filter sheets were cut to fit two (2) modified
Impco VFF30 filter housings. The filters were used to asses the approximate
size range and amount of effluent activated carbon fines which were entrained
during the test.

5. Contaminants
The following five (5) compounds (98-99+% pure) were added as contaminants
for the feedstock (Table 2).

7
 Table 2: Description of Added Contaminants for First Test
Contaminant Description
n-Octane (C8H18): a representative compound of gasoline contamination
Pentadecane (C15H32) a representative compound of diesel contamination
Methyl linoleate a plasticizer and known LP Gas contaminant
Dioctyl adipate a plasticizer and known LP Gas contaminant
Butyl benzyl phthalate a plasticizer and known LP Gas contaminant

These contaminants are representative of the type and range of compounds that
can be found in commercial LP Gas streams6. Gasoline and diesel contamination
can occasionally occur when LP Gas is shipped through multi-product pipelines.

D. Procedure
1. Column Packing
The activated carbon vessel was loaded (from bottom up - the direction of
process flow is up) with: (1) a 100 micron (149 mesh) filter used as a support
material to prevent activated carbon fines from dropping into the upstream line
(this filter should not to be confused with the two filters used to trap and measure
activated carbon fines downstream of the column); (2) a layer of 1/4” support
balls; (3) a layer of 1/8” support balls; (4) the activated carbon; (5) a layer of 1/8”
support balls; and (6) a layer of 1/4" support balls (Table 3 and Figure 2).

Table 3: Column Loading

Layer Mass (lbs./kg) Height (in/cm)
1/4-inch Support Balls 0.8/0.36 2.5/6.4
1/8-inch Support Balls 0.6/0.27 1.5/3.8
12x40 Activated Carbon 2.2/1.0 20/51
1/8-inch Support Balls 0.2/0.09 1.0/2.5
1/4-inch Support Balls 0.4/0.18 1.5/3.8

a) Layer of 1/8” Support Balls b) Activated carbon c) Layer of 1/4" Support Balls
Figure 2: Pilot Vessel Packing

2. Preparation of Contaminated LP Gas Supply

The contaminants were pre-measured by weight into a clean beaker. The
contaminants were added to the clean propane in proportions and levels similar
to those in previously analyzed commercial LP Gas streams. The actual
concentrations in the feedstock used for comparison with the results of the
treated LP Gas were determined by a gas chromatographic based method (see
Section V.E.: Results and Discussion).

A clean 25-gallon (95-L) propane tank was partially filled with propellant-grade
propane. The contaminants were then poured into a clean propane delivery
hose and the hose was reconnected to the tank. The contaminants were flushed

8
from the hose into the tank as it was filled (to ~80% full) with propellant grade
propane. The 25-gallon tank was subsequently shaken and rolled to thoroughly
mix the contaminants throughout the filled vessel.

3. Wetting
Pre-process “wetting” of the activated carbon is required to remove air from the
activated carbon pores7. The wetting procedure consisted of two steps: (1)
carbon dioxide (CO2) pre-loading to minimize the amount of heat released during
wetting with the process fluid, and (2) wetting with propellant-grade propane.

The intent of CO2 preloading is to reduce the heat of adsorption when propane is
introduced. The activated carbon first warms from the exothermic loading of
CO2, then cools over time. CO2 was introduced into the activated carbon bed and
pressurized to 14.7 psig (2.0 bar). The temperature of the bed initially increased
as expected, and was allowed to stabilize over ~1.5 h. This procedure was
recommended by the activated carbon manufacturer, but if it is found
unnecessary in future tests, it should be eliminated to facilitate field
implementation.

Next, propellant grade propane was introduced. The propane flow is to be fast
enough to allow the propane to act as a heat sink, (although the CO2 also acts to
reduce heat as it desorbs). There was an initial heat rise of ~60 °F (15.6°C) as
propane vapor flowed through the bed, followed by a rapid temperature drop as
liquid filled the system (Figure 3). The vessel valves were then closed and the
activated carbon bed with liquid propellant grade propane was left undisturbed
for ~1 h.
140

120
Temperature (°F)

100

80

60

40
0 50 100 150 200 250 300 350 400 450 500
Time (seconds)

Inlet Temperature Outlet Temperature

Figure 3: Column Temperature During Wetting Procedure

4. Decontamination and Sampling

After the wetting procedure, the contaminated supply tank was attached to the
vessel containing the activated carbon bed, and flow of the contaminated LP Gas
stream through the bed was initiated. Based on discussions with project advisors
the process flow was directed upwards through the column against gravity. The

9
flowrate was calculated by monitoring the drop in weight of the contaminated
supply tank over time. Adjustments were made as needed to maintain a
predetermined optimal flowrate. This desired flowrate was calculated based on
project advisors’ recommendations and on prior exchanges with CEC. The
process was scaled down based on cross-sectional area to maintain the same
superficial velocity of LP Gas flow in the smaller pilot vessel. The flowrate is
reported per cross-sectional area of the column to provide a means for
comparison of flow between two different sized columns and to more easily
indicate the contact time with the activated carbon through the length of the bed.

The first sample was drawn after flowing through approximately one bed-volume
of contaminated supply. Over the course of the first pilot test, nine (9) samples
were drawn at regular time intervals (Figure 4). Two (2) additional samples - one
of the contaminated supply and one of the propellant grade propane - were also
taken.

The temperature of the system was monitored, but not controlled throughout the
course of the experiment. There were no significant temperature changes during
the experiment. Ambient temperatures remained at ~70°F (21°C) until night time
when the temperature dropped to ~62°F (17°C). Pressures were constant at 170
psig (12.7 bar).

25.00

Sample #1 Sample #5
Sample #6

20.00
Sample #2
Sample #7
Sample #3
Volumetric Flowrate (gal/min-ft²)

15.00
Sample #4

10.00

5.00

0.00
0 2 4 6 8 10 12 14 16 18 20
LP Gas Processed (gal)

Figure 4: LP Gas Flowrates During First Test

5. Activated Carbon Fines

100 micron and 5 micron filters were weighed (dry) prior to their installation
downstream of the activated carbon. These filters were later removed after the
contaminated propane testing was completed, and re-weighed (dry).

10
 E. Results and Discussion
1. Contaminant Removal
The analytical method [prEN 15470:2006 (E)] was modified to use heptane as
the internal standard because octane was one of the selected contaminants.
Some samples were not analyzed. In-line (or on-site) analysis was also not
possible. Sample #4 was selected for analysis because it had the lowest flowrate
during sampling (closest to the predicted optimal flowrate and presumably with
the greatest potential for contaminant removal). Sample #5 was selected for
analysis because it had a higher flowrate during sampling and was taken shortly
(1.3 gal / 4.9 L) after Sample #4 was completed. Sample #6 was selected for
analysis as an intermediate flowrate sample taken shortly (1.3 gal / 4.9L) after
Sample #5.

The expected Contaminated Propane concentration (based on what was

weighed and poured into the propane transfer hose) differs significantly from the
concentration measured with the GC method (Table 4). Several possible factors
could contribute to this discrepancy. One is that not all of the contaminant
mixture was transferred from the weighed beaker to the transfer hose. The
mixture was viscous and a visible amount was unavoidably left on the beaker
wall. Given that the total contaminant volume added was 19.3 mL (0.65 oz), this
loss of contaminant to the beaker may have been a significant error source.
Another possible source of contaminant loss was inside the transfer hose. Some
of the contaminant may have either been absorbed by the hose or adhered to it.
A third factor may have been whether the contaminants, once introduced to the
tank, were actually mixed homogenously throughout the propane (as intended).
Some of the contaminants may have been adsorbed by the propane tank wall.

Table 4: Weight and GC Based Concentrations of Each Contaminant in First Test

Contaminants (order of Concentration (mass ppm)
increasing boiling pt.) Expected (by weight) GC % Diff.
n-Octane 119 19 84%
Dioctyl adipate 95 27 72%
Methyl linoleate 59 19 68%
Pentadecane 119 42 65%
Butyl benzyl phthalate 86 32 63%
Total∗ 479 141 71%

Although the GC test method itself may have been a source of some loss, it is
not likely that the heavy contaminants would have evaporated off to a large
extent. The test method was modified to keep the evaporation temperature
relatively low (25°C / 77°F) to avoid such losses. Although there is an apparent
correlation between the boiling points of the contaminants used and the amount
of loss of each contaminant, the contribution of evaporation to residue loss is
undetermined.

∗
“Total” value varies slightly from the sum of the listed contaminants as it is an independent
measurement of total residue.
11
Based on these results, a new method to introduce contaminants with minimal
loss was used in subsequent tests.

Because the same GC method was used to measure both the Contaminated
Supply and the processed samples, the above described concentration
discrepancy was not considered a significant factor in the evaluation of the tested
activated carbon’s ability to remove heavy-ends. It should also be noted that on-
site immediate analysis of these samples during the course of the tests was not
possible, and therefore the results could not be known until after all the tests
were completed. The possibility of the Contaminated Supply concentration
changing over time should be addressed the in future tests (See Section VI. F).

The results from Samples #4, #5, and #6 clearly indicate effective removal of the
contaminants selected for this initial protocol (Table 5). The results indicate that
either lower flowrate results in better n-octane removal or that between Samples
#4 and #5 the activated carbon became loaded to the point that it was unable to
further adsorb n-octane. Another possibility is that other larger contaminants
began to displace n-octane in the adsorption sites. The additional analysis of
Sample #6 does not provide any additional support to the idea that a lower
flowrate would result in better n-octane removal. However, the analyst indicated
that there was some evidence of a small leak in the sample vessel. If this is the
case, it is likely that the lighter molecules (i.e. propane) evaporated, leading to an
erroneously high concentration of the contaminants (i.e. octane) left in the
sample vessel. Whether n-octane could be further adsorbed by the activated
carbon is uncertain at this point.

Table 5: Concentration of Contaminants in Samples from First Test

Concentration (mass ppm)
Average
LP Gas Sample Butyl
Processed Flowrate Methyl Dioctyl benzyl
Sample (gal.) (gal/min-ft²) Octane Pentadecane linoleate adipate phthalate Total
Propellant-
grade N/A N/A 0 0 0 0 0 1
Supply N/A N/A 19 42 19 27 32 141
#4 8.4 1.9 11 0 0 0 0 12
#5 11.7 10.7 21 0 0 0 0 22
#6 13.0 7.5 20 0 0 0 0 24

Adsorption capacity of the activated carbon for the added contaminants was not
determined. However, a predicted adsorption capacity based on manufacturer
estimates and from project advisors’ experience suggests that the volume of
contaminated propane processed in this experiment was far less than what could
be expected to be treated before the activated carbon pores are loaded.
Breakthrough (the point at which the activated carbon no longer adequately
removes selected contaminants) was determined in subsequent tests. This
would provide an indication of how much contaminated LP Gas could be
processed by a given amount of activated carbon (given that no channeling
occurs in the adsorption bed).

12
Removal of the lighter contaminant (octane) was not the primary focus of this
test, but as octane is representative of some LP Gas contaminants, this test
reveals that additional work beyond the scope of this project may prove beneficial
if higher removal rates of such contaminants are desired.

It can be concluded that full removal of the four other contaminants, over a
processed volume of ~11.7 gal (44.3 L), could be achieved using this bed
dimension and a flowrate of 10.7 gal/min-ft2 (436 L/min-m2). There was no
measured distinction between the concentration of the contaminants in Samples
#4, #5, and #6 (all are 0 ppm), so there is no indication of the effect of a change
in flowrate (faster or slower than 10.7 gal/min-ft2) on contaminants removal.

2. Activated Carbon Fines

Both the 5 micron and 100 micron polyester filters had visible activated carbon
fines, and a measurable increase in weight after drying (Table 6). The filters,
however, were not visibly clogged with activated carbon fines. Also, some of the
additional mass can be attributed to metal pipe shavings and thread compound
which were also clearly visible on the filters. It was not possible to separately
measure to what extent this metal and thread compound contributed to the mass
deposited on the filters. It was expected and later determined that the amount of
metal and thread compound found on the filters significantly decreased in future
tests with the same test vessel.

Table 6: Activated Carbon Particles Collected in First Test

Filter Size Mass Collected (g / oz)
5 micron 0.10 / 0.0035
100 micron 0.17 / 0.0060

a) 100 micron (149 mesh) Filter b) 5 micron (2500 mesh) Filter

Figure 5: Filters (Post-Process) Showing Activated Carbon Fines and Other Debris

F. Proposed Next Steps

The initial tests successfully demonstrated contaminant removal, and supported
further experiments (See Section VI).

13
VI. Second Small Scale Propane Decontamination Test
A. Summary
Based on the results and know-how gained from the first test, this second test
was designed and conducted to determine the effect of activated carbon on
odorized HD-5 LP Gas that contains actual LP Gas residues and compounds
representative of heavier residues (C20+). The activated carbon bed was also
shortened to be in proportion to the design of the proposed full-sized column and
to better characterize the breakthrough of each contaminant. Also addressed
were the reduction of the apparent loss of contaminants added to the supply
tank; better flow control at a low flowrate; and eliminating metal debris as a
source of error from the polyester filters.

B. Objectives
(1) Determine the effect of a specific type of activated carbon on the
removal of selected concentrations of “heavy” contaminants from
LP Gas processed at a pre-determined flowrate.
(2) Measure effluent activated carbon fines at a given process flowrate.
(3) Measure process temperatures and pressures to better predict
process conditions in future experiments.
(4) Achieve breakthrough of contaminants.
(5) Use a lower length to diameter activated carbon bed ratio (scalable
to the full-sized DOT transportable full-sized vessel). (See Section
VII)
(6) Determine the effect of the activated carbon adsorption process on
mercaptan sulfur (i.e. odorant) concentration in the tested LP Gas.

C. Experimental Setup and Materials

1. Activated Carbon
The same type of activated carbon was used as in the first experiment.

2. Support balls
1/8” and 1/4” ceramic balls were used to support the activated carbon and to
evenly distribute flow inside the vessel.

3. Filters
New 100 micron x 0.100” (2.54 mm) thick and 5 micron x 0.080” (2.03 mm) thick
polyester felt filter sheets were cut to fit the modified Impco VFF30 filter
housings. The filters were used to determine the approximate size and amount
of effluent activated carbon fines which are entrained at specific flow conditions.

4. Contaminants
The following compounds were obtained and added to HD-5 LP Gas (Table 7):

14
 Table 7: Description of Contaminants Added for Second Test
Contaminant Description
n-Octane (C8H18) a representative compound of gasoline contamination
Dioctyl adipate (C22H42O4) a plasticizer and known LP Gas contaminant
American Welding & Tank residues (a blend 7% recovered at n-C8 boiling point; 54% recovered at n-C11 boiling point;
of many contaminants from LP Gas tanks) 95% recovered at n-C20 boiling point (Determined by ASTM D2887)
Mobil Rarus 427 compressor oil ISO 100 (viscosity) with 100% recovered above n-C20 boiling point

The proportions added were weighted to ensure higher concentration of heavy

(C20+) compounds. The actual concentrations in the feedstock used for
comparison with the results of the treated LP Gas were determined by a gas
chromatography based method.

The American Welding & Tank residues were collected from used and recovered
LP Gas tanks from throughout the eastern U.S. These residues are thought to
be representative of contaminants found in commercial LP Gas streams,
although not necessarily those found in fuel system residues. ASTM D2887
analysis was used to provide the above characterization. Additionally, the total
sulfur concentration (using ASTM D 4294) was 6.56 mass %.

D. Procedure
1. Column Preparation
All valves to the column were vented, and the top cover and filter housings were
removed. A flowmeter was installed downstream of the vessel. After removing
support balls and activated carbon from the previous experiment, all exposed
parts (e.g. filter housing interior, fittings with excess pipe compound) were first
wiped clean with a cloth, and then the entire system was filled and flushed with
pure propane. After flushing, the system was vented.

Prior to loading the activated carbon vessel, the apparatus was placed on a
scale. The activated carbon vessel was then loaded (from bottom up – the
direction of process flow is up) with: (1) a 100 micron filter used as a support to
prevent activated carbon fines from dropping into the upstream line, (2) a layer of
1/2” support balls, (3) a layer of 1/4” support balls, (4) a layer of 1/8” support
balls, (5) the activated carbon, (6) a layer of 1/8” support balls, and (7) a layer of
1/4" support balls, and (8) a layer of 1/2” support balls (Table 8). The extra
support balls were used as a means to shorten the activated carbon bed length
within the same column used previously.

Table 8: Column Loading

Layer Mass (lbs./kg) Height (in/cm)
1/2-inch Support Balls 0.8 / 0.36 3.5 / 8.9
1/4-inch Support Balls 1.4 / 0.64 4 / 10
1/8-inch Support Balls 1.0 / 0.45 2.5 / 6.4
12x40 Activated Carbon 0.6 / 0.27 7 / 18
1/8-inch Support Balls 1.0 / 0.45 3.1 / 7.9
1/4-inch Support Balls 1.0 / 0.45 3.5 / 8.9
1/2-inch Support Balls 0.8 / 0.36 2.9 / 7.4

2. Wetting
As before, a wetting procedure was conducted (CO2 pre-loading followed by
wetting of the activated carbon with propellant-grade propane). After flowing

15
CO2, through the column, the outlet valve was closed and the system was
pressurized to 14.7 psig (2.0 bar). There was no measurable temperature
change. The system was then kept under pressure for ~1.5 hours. Propellant-
grade propane was then introduced, but unlike the first test, there was no
measurable temperature rise. This likely occurred because the bed in this test
was smaller, which would result in less heat released, and also less time to fill
with liquid phase LP Gas. Additionally, because of the smaller bed size, the
temperature probes were further away from the activated carbon. Because there
was no temperature differential, the outlet and inlet valves were closed once it
was determined that liquid had entirely filled the column. The wetted system was
left closed overnight.

3. Preparation of Contaminated LP Gas Supply

The contaminants were pre-measured by weight into a clean glass beaker. A
clean 100-gallon (379 L) propane tank was then partially filled (~20%) with HD-5
LP Gas. A short metal pipe with a fitting for the LP Gas delivery hose was
attached to the tank inlet valve. The contaminants were then poured into the pipe
and flushed into the tank as it was filled (to ~85% full) with HD-5. The beaker
used to pour the contaminants was re-weighed to account for contaminant loss.

The 100-gallon “contaminated supply” tank was then shaken and rolled to
thoroughly mix the contaminants throughout the filled vessel. The tank was left
undisturbed overnight. The next morning, the contaminated supply tank was
again rolled and shaken. A nitrogen supply line was then connected to the vapor
valve of the supply tank, and a pressure head of ~185 psig (13.8 bar) was
applied to the contaminated supply.

4. Decontamination and Sampling

Following the wetting procedure, the contaminated supply tank was attached to
the activated carbon vessel, and flow of the contaminated LP Gas stream
through the bed was initiated. Adjustments were made, if necessary, to maintain
a predetermined optimal flowrate. The flowmeter installed for this test did not
function properly; so it did not improve flow monitoring and control vs. the first
experiment. Flow was measured by monitoring the increase in weight of the
receiving vessel over time. Flow adjustments were made manually by
decreasing the vapor pressure in the receiving vessel and/or opening the vessel
outlet valve. This receiving vessel was not used for sample collection.

Each aluminum sample vessel was first cleaned with propellant grade propane,
vented, and then filled with nitrogen to ~80 psig (6.5 bar) as a means to control
the pressure differential created when the LP Gas sample line is first connected
to for filling. Aluminum vessels were used at the recommendation of project
advisors. The flowrate was controlled in the same way as in the first test. The
first sample was drawn after flowing through approximately one bed-volume of
contaminated supply. Over the course of the pilot test, seven (7) samples were
drawn at regular time intervals. An eighth sample was being collected as the
contaminated LP Gas supply ran out. A sample of both the contaminated supply,
and the HD-5 spec LP Gas were also taken.
16
 5. Activated Carbon Fines
100 micron and 5 micron filters were weighed (dry) prior to their installation
downstream of the activated carbon. These filters were later removed after the
contaminated LP Gas propane testing was completed, and re-weighed (dry).

E. Results and Discussion

1. Contaminant Removal
A modified prEN 15470:2006 (E) method was again used for sample analysis,
with heptane as the internal standard and a lower evaporation temperature
(25°C).

Table 9: Weight and GC Based Concentrations of Each Contaminant in Second Test

Contaminants Concentration (mass ppm)
Expected (by weight) GC Determined % Difference
n-Octane 53 15 72%
Dioctyl adipate 57 28 51%
American Welding & Tank Residues 211 98 54%
Mobil Rarus 427 Compressor Oil 346 196 43%
Total 667 337 50%

The expected contaminated LP Gas concentration (based on what was weighed

and poured into the tank) again differed significantly from the concentration
determined with gas chromatography (Table 9). This difference is a concern that
should be addressed as test methods for this purpose are further developed.
There was, however, a significant improvement in the efficiency of addition of the
contaminants from the previous test. In this test, the amount of residual
contaminants remaining on the beaker walls was accounted for by weight, but
this resulted in only a slight reduction (0.6%) in the expected concentration.
Although the shorter, steel transfer pipe likely led to better transfer of the
contaminants to the tank, some of the contaminant may have still adhered to the
pipe surface or, once introduced to the tank, settled or been adsorbed onto the
tank walls instead of dispersing homogenously throughout the LP Gas. One
possible improvement to in future work is to pre-dissolve the contaminants in a
lighter solvent such as pentane or toluene prior to addition to the LP Gas.

Results show that for this flowrate and bed dimensions the activated carbon
removed significant amounts of all the added contaminants. The amount of
contaminant removed and the time to breakthrough varied for each contaminant
(Figures 6 & 7 and Table 10). As LP Gas is processed, contaminants with lesser
affinity for activated carbon may be displaced by compounds that are more
readily absorbed. The process was effective in removing the compressor oil and
dioctyl adipate (which contained the majority (~98%) of C20+ compounds used in
this test). The concentration of these contaminants remained at below 10% of
their initial concentrations (>90% removal) and below a combined 15 ppm until
after 241 bed volumes (47 gal.) of LP Gas were processed. This achieves the
goal initially set for these tests of reaching less than 25 ppm C20+ heavy-ends
content in the decontaminated product.

17
 Table 10: Concentrations of Heavy-Ends Contaminants and Ethyl Mercaptan (Ethanethiol)
for Second Test
Sample Flowrates
LP Gas Processed (gal/min-ft²) Concentration (mass ppm)
Sample Bed Dioctyl AW&T Compressor Total
No. Volumes Gallons Average Max Octane Adipate Residue Oil Residues Ethanethiol
Supply N/A N/A N/A N/A 15 28 98 196 338 14.3
1 11 2 1.8 9.6 2 0 7 7 17 2
2 62 12 1.5 7.5 2 0 12 7 21 8.9
3 119 23 2.0 6.4 4 0 35 7 46 11.9
4 177 34 2.7 5.3 11 1 59 13 85
5 241 47 2.0 4.3 5 2 47 11 65
6 297 58 2.2 4.3 14 16 91 32 153
7 332 65 1.9 5.3 11 25 86 35 157

Because contaminated LP Gas flow during sampling varied more than

anticipated, for each sample are listed both the average flowrates during
sampling and the maximum flowrate which occurred as the sample was being
taken (Table 10). Both these flowrates are relevant as a high average flowrate
may not allow enough contact time with the activated carbon to achieve
maximum contaminant removal, or alternatively one or more flowrate “spikes”
during a 2-gal (7.6 L) sample may significantly impact the concentration taken for
the whole sample. This is a likely reason that the concentrations of most
contaminants in Sample #4, which had the highest average flowrate, were higher
than those found in Sample #5.

Table 11: Concentration of Sulfur Compounds for Second Test

LP Gas Processed Concentration (mass ppm as S)
Sample Bed 1- Diethyl Diethyl 1- 1- Isopropyl
No. Volumes Gallons Ethanethiol Butanethiol Sulfide Disulfide Heptanethiol Octanethiol Disulfide Total
Supply N/A N/A 7.4 0.2 3.6 8.8 0.7 1.2 0.2 22.1
1 11 2 1.0 0.2 3.1 9.0 13.3
2 62 12 4.6 0.3 3.2 6.3 0.5 1.1 16.0
3 119 23 6.1 0.2 3.2 8.5 0.5 1.0 19.5

Analysis of sulfur compounds in the LP Gas suggests that they are less readily
removed by the selected activated carbon than the heavy-ends residues (Table
11). Of particular note, ethyl mercaptan (ethanethiol) is initially reduced to 14%
of its original level (Sample #1), then returns to 62% (Sample #2) within 62 bed
volumes, and 82% within 119 bed volumes (Sample #3). This suggests that
heavy-ends removal can continue long after the ethyl mercaptan is no longer
removed by the activated carbon.

18
 12 1.00

0.90

10
0.80
Average Cross-Sectional Flowrate (gal/min-ft²)

0.70
8

Normalized Contaminant Concentration

0.60

6 0.50

0.40

4
0.30

0.20
2

0.10

0 0.00
0 10 20 30 40 50 60 70 80
LP Gas Volume Processed (gal)

Octane Dioctyl Adipate AW&T Residue Compressor Oil Total Residues Ethanethiol

Figure 6: Flowrate Overlaid by Breakthrough Curves (Normalized) for Second Test

12 180

160

10

140
Average Cross-Sectional Flowrate (gal/min-ft²)

Contaminant Concentration (ppm)

8 120

100

80

4 60

40

20

0 0
0 10 20 30 40 50 60 70 80
LP Gas Volume Processed (gal)

Octane Dioctyl Adipate AW&T Residue Compressor Oil Total Residues Ethanethiol

Figure 7: Flowrate Overlaid by Breakthrough Curves (Concentration) for Second Test

19
 2. Activated Carbon Fines
Both the 5 micron and 100 micron polyester filters had visible activated carbon
fines, and a measurable increase in weight, but no evidence of metal particles or
other debris. There was no measurable pressure drop at any point in the test
that would indicate that the process was affected by these fines.

Table 12: Activated Carbon Particles Collected in Second Test

Filter Size Mass Collected (g / oz)
100 micron 0.03 / 0.001
5 micron 0.05 / 0.002

F. Proposed Next Steps

The second tests demonstrate effective removal of heavy-end contaminants from
LP Gas. Suggested next steps based on these results are:
(1) Pilot scale tests using a higher flowrate (desirable for larger
commercial applications);
(2) Pilot scale tests with a lower total heavy-ends concentration (~200
ppm) and lower C20+ heavy-ends concentration (~100 ppm)
(3) Full-scale tests with LP Gas contamination levels as in #2 above.

As flow control continued to be problematic in the second tests, a crucial piece of

subsequent work will be a better flow control system. Future tests should also
determine if the concentration in the Contaminated Supply changes over time.

VII. Full Scale Column & Process Design

A full sized column was designed and manufactured (Figures 8 & 9) to help with
future steps in follow-on tests. The vessel was designed to be DOT approved
transportable, while providing adequate LP Gas flow design, and was subject to
the constraints of Table 13. DOT approval allows for the option to switch out the
entire column when the activated carbon no longer effectively removes
contaminants (rather than regenerating or replacing the activated carbon). (See
Sections IX.A and X).

Figure 8: Full Sized Column Design

20
 Table 13: Full Sized Column Design Constraints
Footprint 48”x48” (1.2 m x 1.2 m)
Weight <3,200 lbs. (1,450 kg) when full (able to be lifted with standard equipment)
Length ~72” (able to fit inside a standard freight truck)
Pressure rated to 250psi at 650°F (to allow for the option of future work with high
temperature in-situ activated carbon regeneration)

The design also includes interior screens and a diffuser plate at the inlet
designed to disperse the LP Gas evenly throughout the adsorption medium as
well as to incorporate inert support balls for the adsorbent.

Figure 9: Process Flow Diagram

After CEC engineers calculated expected flow rates and made recommendations
on the tank re-design dimensions based on their prior work, the initial vessel
design was further refined with a reduced outer diameter (24” / 61 cm). This
smaller diameter design reduces manufacturing cost and is likely to reduce
undesirable channeling of the LP Gas through the activated carbon bed. The
desired flow rate for the reduced diameter vessel was accordingly lowered. The
length to diameter ratio in this large scale vessel was modeled in the pilot
experiments, and the flowrates tested in the pilot tests were scaled to maintain
the same cross-sectional flowrate as in the full-scale column (Table 14). The
pilot unit to full-scale flowrate ratio is 1:67 (1:560 by volume), which is well within
the 1:1000 max ratio commonly used and accepted for process scalability.

21
 Table 14: Comparison of Full Scale and Pilot Columns Dimensions
Full-Sized Column Pilot Scale Column
Bed Length (in) 56.4 6.8
Bed Diameter (in) 23.6 2.87
Length to Diameter Ratio 4.8 4.9
2
Cross-sectional Area (ft ) 3.0 .045
Volumetric Flowrate (gal/min) 6.0 .09
2
Cross-sectional Flowrate (gal/min-ft ) 2.0 2.0

VIII. Economic Analysis

Last fall, the additional cost paid in Southern California for premium LP Gas with
low heavy-ends content was ~$0.10/gallon at $2.10 to $2.25 retail pricing. For a
decontamination process to be economically viable, the total cost must be
sufficiently below this $0.10/gal premium to achieve payback over a reasonable
period.

The above described full-sized column was used for the below economic
analysis (Table 15, see Appendix 3 for full analysis). Additionally, the following
assumptions were made:
(1) Total heavy-end contaminants in the supply stream are 338 ppm,
including ~229 ppm of C20+ based on boiling point analysis of the
contaminants in second test (see Table 10 and Appendix 2). Lower
concentrations of these contaminants in the LP Gas will likely result
in a greater volume of treated LP Gas prior to breakthrough, and
thus a lower cost per volume of clean product.
(2) Replacement of activated carbon is required after the C20+
residues content exceeds 25 ppm. In the experiments used for this
analysis, the last measured sample with C20+ content below 25
ppm was at 13 ppm (or <10% of the initial concentration) at which
point 241 bed volumes had been processed. The total residues
concentration at this measurement was 65 ppm (or <20% of the
initial concentration). For this reason, 241 bed volumes was used
as the capacity of the activated carbon before replacement.
(3) The process flowrate is 6 gal/min.
(4) For this analysis, the per-gallon revenue to be generated by the full
scale plant is taken to be the known additional cost recently paid for
low-residue LP Gas ($0.10/gal)

Table 15: Economic Analysis Summary

Annual Production (gal) 1,057,536
Fixed Capital $ 89,874
Working Capital $ 8,987
Total Capital Cost $ 98,862
Annual Manufacturing Cost $ 64,948
Annual Revenues $ 105,754
Project Life (years) 15
Tax Life (years) 15
Payout time (years) 2.74

22
The total cost per gallon of clean product is $0.061. This analysis yields a payout
time of 2.74 years.

IX. Field Service Recommendations

Pending full scale implementation and testing of this activated carbon LP Gas
decontamination process, the following recommendations can be made based on
initial tests:

A. Pre-process Wetting and CO2 Pre-loading

The activated carbon must first be properly wetted to achieve optimal results.
This is an essential step in the process. One option envisioned for field
implementation is that “module” columns are pre-packed and can be switched
out when breakthrough is reached (when the activated carbon must be replaced).
In this case, the carbon dioxide preloading step can be performed at the column
packing site. If future tests determine that CO2 pre-loading is not necessary, this
step can be eliminated. Wetting is conducted by flowing LP Gas through the
column until it is filled with liquid and the temperature differential between inlet
and outlet streams is less than 10°F (5.5°C) (if necessary, the LP Gas used for
this procedure can be recycled back through the column). At this point, the
vessel valves should be closed and the LP Gas is left undisturbed for at least one
hour7.

B. Flowrate
The flowrate used to process the contaminated LP Gas is critical to achieve
optimal contaminant removal. It is expected that an increased flowrate will result
in lower removal of contaminants. (Note: The precise effects of such changes
must be tested to better understand the effects of this key variable).

Care must be exercised so that upon start-up, the flow of propane is not started
abruptly in a manner which may lead to disruption of the packed bed and
possible loss of activated carbon or formation of “channels”.

C. Breakthrough and Activated Carbon Replacement

The quality of LP Gas upstream and downstream of the column must be
monitored regularly. The concentration of various contaminants in the supply
stream will affect the amount to which the activated carbon can remove
contaminants, and how much product may be treated before the activated carbon
must be replaced. Based on the above results, it can also be anticipated that an
increase in the effluent concentration of lighter residue components will likely
precede breakthrough of the heavier heavy-end residues. The above heavy-
ends removal capacity of the activated carbon is provided as an estimate based
on pilot tests conducted to date at specific conditions.

D. Filters
Initial experiments show that some fine activated carbon particles are entrained
in the product stream and can be captured by filters downstream of the column.
In tests, these particles were not sufficient to impact the flow of product by

23
clogging the filters; however, the pressure difference across these filters should
be monitored to determine if filters need to be replaced.

E. Ethyl Mercaptan Caution

Because ethyl mercaptan content is initially reduced, caution should be
exercised. LP Gas downstream of the activated carbon column during this initial
period of operation may not be sufficiently odorized to indicate a potential leak.

X. Operating Options
Two possible options for the replacement of spent activated carbon
decontamination process are:
(1) Spent activated carbon from LP Gas dealers is regenerated at a
propellant manufacturer who already has equipment necessary for
regeneration on-site. This propellant manufacturer would supply
the LP Gas dealer with regenerated activated carbon.
(2) Spent activated carbon would be collected, regenerated, and
replaced by the activated carbon manufacturer.

Refinery
LP Gas
Clean-up Site

LP Gas Dealer LP Gas Dealer

LP Gas
Clean-up Site Propellant
Manufacturer
Spent Medium (catalyst impregnated AC)
Regeneration Site

Option 1: Regeneration of Spent AC at Propellant Manufacturer’s Site

Refinery
LP Gas
Clean-up Site

LP Gas Dealer LP Gas Dealer

LP Gas
Clean-up Site
AC Provider

Option 2: Spent AC Regenerated or Disposed of by AC Provider

Figure 10: Activated Carbon Regeneration Schemes

XI. Conclusions
These activated carbon tests have demonstrated that certain heavy-end
contaminants can be significantly removed from LP Gas. The latest tests show
the process to be particularly effective in removing heavier contaminants (i.e.
C20+ heavy-ends).

Additionally, the results show that ethyl mercaptan concentration is reduced by

the process, but reappears in the product stream more quickly relative to other

24
contaminants – suggesting that heavy-ends may continue to be removed while
mercaptans are not. However, addition of more ethyl mercaptan may be
required to ensure that the final product has a suitable (and legally defensible)
concentration of odorant.

Initial economic analysis suggests that this process can be viable. Further
development is recommended.

XII. Future work

As stated above, proposed immediate next steps based on the results of the
latest tests are:
(1) Pilot scale tests using a higher flowrate (desirable for larger
commercial applications);
(2) Pilot scale tests with a lower total heavy-ends concentration (~200
ppm) and lower C20+ heavy-ends concentration (~100 ppm)
(3) Full-scale tests with LP Gas contamination levels as in #2 above.

Additionally, future work may include a comparison of analytical methods (i.e.

CEN and CEC) or variations on these methods.

ASCENT has also discussed the possible implementation of a novel method to

achieve reasonably priced in-situ regeneration. This aspect of the
decontamination process may be developed as an eventual follow-up if the
disposal option of the spent activated carbon by the activated carbon provider
turns out not to be viable.

XIII. Acknowledgements
ASCENT is grateful to PERC for support and funding of this project. For their
ongoing assistance, we thank our advisors: John Ehlers, Bob Falkiner, George
Maes, John O’Connor, Andy Pickard, Arnie Smith, Jean-Paul Trespaille, and
Quan Zhuang.

For their invaluable help ASCENT is also indebted to Ernie Reed (Aeropres
Corporation); Larry Osgood (Consulting Solutions), Mical Renz (Dixie Services,
Inc.); Junichi Yodotani and T. Kioyosawa (Cosmo Engineering Co.); Jerry Gilliam
and Walker Ogden (American Welding and Tank); Steve Moore (Expo Propane),
Dan McCartney (Black & Veatch), and Bob Myers (PERC).

1
J. O’Connor, Phillips Petroleum Company, “Process for Removing Oil from Liquefied Petroleum
Gas,” U.S. Patent 5,474,671, (Filed April 11, 1994, Issued December, 12, 1995)

25
2
T. Hideo et. al., Cosmo Engineering Co. Ltd., “Method of Removing Residue in Liquefied Gas
and Regenerating Method of Activated Carbon,” Japan Patent 2001-294415, (Filed April 13,
2000)
3
ASTM D 2158, “Standard Test Method for Residues in Liquefied Petroleum (LP Gases),” (West
Conshohocken, PA: ASTM International, 2002)
4
Q. Zhuang and A. Pickard, CANMET Energy Technology Center, “Accurate Measurement
Method for Residues in LPG” (presented to CGSB Committee on Petroleum Test Methods,
November 18, 2004)
5
Q. Zhuang, J. Yodotani, and M. Kato, “Accurate measurement method for the residues in
liquefied petroleum gas (LPG),” Fuel Vol. 84 (2005): 443-446
6
M. Mamakos, J.Meyer, A. Spataru, “Discussion of LP Gas Fuel and Residues Collected from
Microturbine Powered Hybrid Electric Vehicle Sites,” (The ADEPT Group, Inc, March 2005)
7
S. Carr, “CO2 Pre Loading Followed by Organic Wetting,” (Pittsburgh, PA: Calgon Carbon
Corporation, 1999)

26
Appendix 1
First Test Log
Time Comment T1 T2 P1 P2 P3 P4
(min) (°F) (°F) (psi) (psi) (psi) (psi)
0:00 Begin test 150 160
1:30 170 170 170 170
5:30 Sample 1
7:30 Receiving tank outage gauge opened
briefly
13:00 Testing stopped briefly
14:00 Sample 2 170 170 170 170
16:25 68 68
20:00
27:30 Sample 3
36:30 70 68 175 170 170 170
44:00 Begin Sample 4
67:45 End Sample 4
70:15 Testing stopped briefly
73:30 Sample 5
80:45 Sample 6
94:30 Nighttime – ambient air cooler 62 62 175 175 170 170
96:00 Testing stopped briefly
99:30 Sample 7
110:45 Testing ended
Samples 8&9
Temperature and pressures remained steady between times noted.
T1: Temperature at column inlet
T2: Temperature at column outlet
P1: Pressure at column inlet
P2: Pressure at column outlet
P3: Pressure after 100 micron filter
P4: Pressure after 5 micron filter

Sample ID Volume Processed Prior to Sample Average Flowrate During Sampling

(gal.) (gal/min)
Propellant-grade Propane N/A N/A
Contaminated Propane 0.0 N/A
#1 1.76 0.76
#2 3.86 0.38
#3 6.10 0.38
#4 8.38 0.08
#5 11.71 0.57
#6 13.00 0.38
#7 15.57 0.57
#8 Receiving Tank N/A N/A
#9 Receiving Tank N/A N/A

28
 Appendix 2

Certificates of Analysis
Dixie Services, Inc.
Galena Park, TX
CERTIFICATE OF ANALYSIS Number: 122255

Client: The ADEPT Group, Inc. Date: January 6, 2006

10866 Wilshire Boulevard, Suite 350
Los Angeles, CA 90024

Attention: Alex Spataru

Sample: Three samples of propane, submitted 27 Dec 05

Marks: Pure C3H8; Test #1 Baseline; Test 1, #4

The subject samples were analyzed by test method CEN/TC19 N 1221 – Annex
E, “Liquefied petroleum gases – Determination of dissolved residues – Gas
chromatographic method,” except that the maximum evaporation temperature was
limited to 25 °C and n-heptane was substituted for n-octane as the internal standard for
the gas chromatographic analysis. The results were as follows:

Pure Test #1 Test 1

C3H8 Baseline #4
Total residue, mg/kg 1 141 12
n-Octane, mg/kg 0 19 11
Pentadecane, mg/kg 0 42 0
Methyl linoleate, mg/kg 0 19 0
Butyl benzyl phthalate, mg/kg 0 32 0
Dioctyl adipate, mg/kg 0 27 0

Dixie Services Incorporated,

Mical C. Renz

MCR/s

30
CERTIFICATE OF ANALYSIS Number: 122255
Supplement

Client: The ADEPT Group, Inc. Date: January 14, 2006

10866 Wilshire Boulevard, Suite 350
Los Angeles, CA 90024

Attention: Alex Spataru

Sample: Propane, submitted 27 Dec 05

Marks: Test 1, #5 Fast

The subject sample was analyzed by test method CEN/TC19 N 1221 – Annex E,
“Liquefied petroleum gases – Determination of dissolved residues – Gas chromatographic
method,” except that the maximum evaporation temperature was limited to 25 °C and n-
heptane was substituted for n-octane as the internal standard for the gas
chromatographic analysis. The results were as follows:

Test 1
#5 Fast
Total residue, mg/kg 22
n-Octane, mg/kg 21
Pentadecane, mg/kg 0
Methyl linoleate, mg/kg 0
Butyl benzyl phthalate, mg/kg 0
Dioctyl adipate, mg/kg 0

Dixie Services Incorporated,

Mical C. Renz

MCR/s

31
CERTIFICATE OF ANALYSIS Number: 122255
Supplement 2

Client: The ADEPT Group, Inc. Date: February 1, 2006

10866 Wilshire Boulevard, Suite 350
Los Angeles, CA 90024

Attention: Alex Spataru

Sample: Propane, submitted 27 Dec 05

Marks: Test 1, #6, 48455

The subject sample was analyzed by test method CEN/TC19 N 1221 – Annex E,
“Liquefied petroleum gases – Determination of dissolved residues – Gas chromatographic
method,” except that the maximum evaporation temperature was limited to 25 °C and n-
heptane was substituted for n-octane as the internal standard for the gas
chromatographic analysis. The results were as follows:

Test 1
#6
Total residue, mg/kg 24
n-Octane, mg/kg 20
Pentadecane, mg/kg 0
Methyl linoleate, mg/kg 0
Butyl benzyl phthalate, mg/kg 0
Dioctyl adipate, mg/kg 0

Notes
The sample cylinder was exhausted in the conduct of the test, and the total
amount of sample available from the cylinder was less than from the previous test of
Cylinder #5. The analyst observed that an LPG-like odor was detected before the test
was started. Thus it seems likely that this cylinder had a very small leak

Dixie Services Incorporated,

Mical C. Renz

MCR/s

32
CERTIFICATE OF ANALYSIS Number: 123314

Client: The ADEPT Group, Inc. Date: May 20, 2006

10866 Wilshire Boulevard, Suite 350
Los Angeles, CA 90024

Attention: Alex Spataru

Sample: Eight samples of propane, submitted 19 Apr 06

Marks: Listed in tables below
Date: 3/30/2006

Seven of the subject samples were analyzed for residue by test method
CEN/TC19 N 1221 – Annex E, “Liquefied petroleum gases – Determination of dissolved
residues – Gas chromatographic method,” except that the maximum evaporation
temperature was limited to 25 °C and n-heptane was substituted for n-octane as the
internal standard for the gas chromatographic analysis. Four of the samples were tested
for sulfur compounds by ASTM D 5623, “Sulfur Compounds in Light Petroleum Liquids
by Gas Chromatography and Sulfur Selective Detection. The results were as follows:

Contaminated Sample Sample Sample

Supply #1 #2 #3
Total residue, mg/kg 338 17 - 46
Sulfur compounds, mg/kg as S
Ethanethiol 7.4 1.0 4.6 6.1
Ethanethiol, as C2H5SH 14.3 2.0 8.9 11.9
1-Butanethiol 0.2 0.2 0.3 0.2
Diethyl sulfide 3.6 3.1 3.2 3.2
Diethyl disulfide 8.8 9.0 6.3 8.5
1-Heptanethiol 0.7 - 0.5 0.5
1-Octanethiol 1.2 - 1.1 1.0
Isopropyl disulfide 0.2 - - -

Sample Sample Sample Sample

#4 #5 #6 #7
Total residue, mg/kg 85 65 152 157

Dixie Services Incorporated,

Mical C. Renz

MCR/s

33
CERTIFICATE OF ANALYSIS Number: 123314
Supplement

Client: The ADEPT Group, Inc. Date: May 22, 2006

10866 Wilshire Boulevard, Suite 350
Los Angeles, CA 90024

Attention: Alex Spataru

Sample: Eight samples of propane, submitted 19 Apr 06

Marks: Listed in tables below
Date: 3/30/2006

As requested, additional integration intervals have been applied to the

chromatograms from the Total Residue test (method CEN/TC19 N 1221 – Annex E) to
obtain estimates of the four components used to prepare the synthetic residue. The
results were as follows (sum of individual components may not equal the total column
because of rounding):

Dioctyl LPG
n-Octane Adipate Residue Oil Total
Contaminated Supply (mg/kg) 15 28 98 196 338
Sample #1 2 <1 7 7 17
Sample #3 4 <1 35 7 46
Sample #4 11 1 59 13 85
Sample #5 5 2 47 11 65
Sample #6 14 16 91 32 153
Sample #7 11 25 86 35 157

Dixie Services Incorporated,

Mical C. Renz

MCR/s

34
CERTIFICATE OF ANALYSIS Number: 123314
Supplement 2

Client: The ADEPT Group, Inc. Date: June 20, 2006

10866 Wilshire Boulevard, Suite 350
Los Angeles, CA 90024

Attention: Alex Spataru

Sample: Propane Sample #2, submitted 19 Apr 06

Marks: Sample #2
Date: 3/30/2006

As requested, Sample #2, which was previously tested for sulfur compounds, has
also been analyzed for total residue (method CEN/TC19 N 1221 – Annex E). Estimates
of the concentration of the four components used to prepare the synthetic residue were
also obtained. The results were as follows:

Dioctyl LPG
n-Octane Adipate Residue Oil Total
Sample #2, mg/kg 2 <1 12 7 21

Dixie Services Incorporated,

Mical C. Renz

MCR/s

35
CERTIFICATE OF ANALYSIS Number: 123914

Client: The Adept Group, Inc. Date: June 20, 2006

10866 Wilshire Boulevard, Suite 350
Los Angeles, California 90024

Attention: Nolan Sambrano

Sample: Mobil Rarus 427 air compressor oil, submitted 7 Jun 06

Marks: Grainger # 4Zf21

D 2887 Boiling range, °C, % Recovery

IBP 337.0 n-C20 = 344
1 355.5 n-C21 = 356 26 462.0
2 372.0 n-C22 = 369 27 463.5
3 383.0 n-C23 = 380 28 465.5
4 391.5 n-C24 = 391 29 467.5 n-C32 = 468
5 398.0 30 469.0
6 404.0 n-C25 = 402 31 470.5
7 409.0 32 472.5
8 413.5 n-C26 = 412 33 474.0 n-C33 = 479
9 417.5 34 475.5
10 421.5 n-C27 = 422 35 477.0
11 425.0 36 478.5
12 428.0 37 480.0
13 431.0 n-C20 = 431 38 481.5 n-C34 = 481
14 434.0 39 483.0
15 437.0 40 484.5
16 439.5 n-C29 = 440 41 485.5
17 442.0 42 487.0
18 444.5 43 488.5
19 447.0 44 489.5 n-C35 = 489
20 449.5 n-C30 = 449 45 491.0
21 451.5 46 492.0
22 453.5 47 493.5
23 456.0 48 495.0
24 458.0 n-C31 = 458 49 496.0 n-C36 = 496
25 460.0 50 497.0

36
 Boiling Range, °C, % Recovery, continued
51 498.5 76 532.0
52 499.5 77 533.5 n-C42 = 534
53 501.0 78 535.0
54 502.0 79 536.5
55 503.5 n-C37 = 503 80 538.5
56 504.5 81 540.0 n-C43 = 540
57 506.0 82 542.0
58 507.0 83 543.5
59 508.5 n-C38 = 509 84 545.5 n-C44 = 545
60 509.5 85 547.5
61 511.0 86 549.5
62 512.0 87 551.5
63 513.5 88 553.5
64 515.0 89 556.0
65 516.5 n-C39 = 516 90 558.5
66 517.5 91 561.0
67 519.0 92 564.0
68 520.5 93 567.0
69 522.0 n-C40 = 522 94 570.5
70 523.0 95 574.0 n-C50 = 575
71 524.5 96 578.5
72 526.0 97 583.5
73 527.5 n-C41 = 528 98 589.5
74 529.0 99 599.0
75 530.5 FBP 606.0 n-C60 = 615
(n-Cnn = xxx entries above denote the boiling points of normal paraffins as reported in D
2887)

Dixie Services Incorporated,

Mical C. Renz
MCR/s

37
 Appendix 3

Economic Analysis
Annual Production Capacity
Process Flowrate 6 gal/min
Daily Production Hours 10 hr
Annual Production Days 306 days
Operating Capacity 96%
Daily Production 3456 gal
Total Annual Production 1,057,536 gal

Activated Carbon
Total Column Volume (gal) 130 gal
Bed Volume (gal) 107 gal
Bed Volume (cu ft) 14.3 ft^3
Activated Carbon Density 29 lb/ft^3
Filled Bed Weight 414.7 lb
Lifetime before Replacement 241 bed volumes
Lifetime before Replacement 25787 gal
Usage Rate per Gallon Product 0.016 lb/gal produced
Activated Carbon Cost 1.88 $/lb

Filter Cartridges
Unit Cost $ 18.00 $
Rate per Gallon Product 0.0001 units/gal

Carbon Dioxide
Cost $ 1.25 $/ft^3
Volume Per Pre-load 27.81 ft^3
Rate per Gallon Product 0.0011 ft^3/gal produced

Product Quality Analysis

Test Cost $ 150.00 $
Frequency of Test 0.0001 tests/gal produced

Rent & Utilities

Pump Size (hp) 3 hp
Pump Size (kW) 2.2 kW
Annual Power Consumption 6732 kW-hr
Electricity Rate 0.081 $/kW-hr
Annual Electricity Cost $ 545.29 $
Monthly Rent $ 60.00 $
Annual Rent $ 720.00 $

Labor
Daily 0.5 hr
Annual 153 hr
Supervision & Payroll (20% of Labor) 30.6 hr
Labor Rate 24 $/hr
Annual Total $ 4,406.40 $

Capital Costs
Fixed Capital Quantity Unit Base Cost Module Factor-1 Cost (Installed)

51
 Activated Carbon Column 1 column $ 4,060.00 3.2 $ 12,992.00
Pump 1 pump $ 2,500.00 2.4 $ 6,000.00
Storage Tank (Horizontal) 1 tank $25,000.00 2.2 $ 55,000.00
Filter Housing 2 housing $ 560.00 1.2 $ 1,344.00
Support Balls 3.1 cu ft $ 45.00 1 $ 138.36
Gas Chromatograph 1 GC $12,000.00 1.2 $ 14,400.00
Subtotal $ 89,874.36
Working Capital (10% of Fixed)
Subtotal 10.00% $ 8,987.44
Total Capital Cost $ 98,861.80

Manufacturing Cost
Consumables Cost/gal Product Rate Annual Cost
Activated Carbon $ 0.0302 $ 31,973.21
Filter Cartriges $ 0.0014 $ 1,476.38
Carbon Dioxide $ 0.0013 $ 1,425.40
Utilities $ 0.0005 $ 545.29
Rent $ 0.0007 $ 720.00
Labor $ 0.0042 $ 4,406.40
Analysis $ 0.0150 $ 15,863.04
Maintenance 5.00% of Fixed Capital $ 4,493.72
Supplies 1.50% of Fixed Capital $ 1,348.12
Taxes, Insurance 3.00% of Fixed Capital $ 2,696.23
Total Manufacturing Cost $ 0.061 $ 64,947.78

Annual
Price/gallon Revenue
Revenues $ 0.10 $ 105,753.60

NPV Calculation
Required Rate of Return 0.1
Salvage Value $ 19,500.00 $
Project Life 15 years
Tax Life 15 years
Tax Rate 0.5
Straight Line Depreciation $ 4,691.62 $/year
NPV $ 80,986.47
Annualized NPV $ 10,647.60
Rate of Return at NPV=0 22%
Payout time 2.74 years

Yearly revenues, expenses, and depreciation are taken to be constant so NPV is calculated based on the
following expression: NPV = -(CI+Cw) + (R-X)(1-t)[1-(1+i)-n/i + Dt[1-(1+i)-nt]/i +(Cs+Cw)/(1+i)n
Where:
R = Annual revenues n = Project life (years)
X= total Expenses nt = Tax life
CI= Fixed capital investment i = Required rate of return
Cs = Salvage Value t = Tax rate
Cw = Working capital

52
Appendix 4

Test Methods
Determination of Part Per Million Trace Components.

Purpose: To analyze for part per million levels of contaminants in feedstock.

Such as unsaturated compounds, and hexane+ hydrocarbons.

All Incoming and outgoing product is required to have the trace

contaminate analysis performed. The product must meet the
predetermined specification of either the customer or Aeropres.

A 300cc stainless steel sample container equipped with toggle valves

on each end is to be used for sample procurement. The cylinder is
attached to the vessel to be sampled via a rubber hose. The sample
cylinder is to be purged a minimum of three times. Fill the sample
cylinder until it is completely full. Disconnect from the vessel being
sampled. Blow off a small portion of product from the sample cylinder
to allow room for expansion.

In the laboratory the cylinder is to have a helium head pressure

applied of 180-200 psi. The sample cylinder is then connected to a
bulkhead and then injected into the gas chromatograph. Aeropres
uses a 4-port valve with a 1.0µl sample size. The instrument used is
equipped with a flame ionization detector. The parameters are as
follows:

Oven Temp: 20-30ºC, isothermal

Carrier gas: Helium 15-35cc/min.
Column: 25’ x 1/8” x 0.012” wall, stainless steel, P/W 25% Bis(2-
methoxylEthyl) Adipate on Chromsorb NAW, 80/100 mesh.
Detector Temp: 110-225ºC
Air Bottle Pressure 40-60 psig
Hydrogen Flow Rate 20-30 cc/min

The method used by Aeropres is a modified version of ASTM- 1954-64

The Flame Ionization detectors are calibrated monthly using only

standards that are N.I.S.T. traceable.

54
 LP GAS CONTAMINATION REMEDIATION AT
BULK TERMINALS PROJECT

FINAL REPORT ADDENDUM

Prepared for the
Propane Education & Research Council
(PERC Docket #11649)

October 24, 2006

By: Nolan Sambrano, Jared Meyer, and Alex Spataru

ADEPT SCIENCE & TECHNOLOGIES, LLC

51 Rover Boulevard
Los Alamos, NM 87544 USA
Telephone: (505) 672-0002
Fax: (505) 672-0003

decont055_final report addendum.doc

 ADEPT SCIENCE & TECHNOLOGIES, LLC
DOCKET #11649 FINAL REPORT

Table of Contents

I. Explanation of Changes to Work Originally Proposed ...................................3

A. Multi Bed Sieve vs. Activated Carbon Column ...........................................3
B. Design and Calculations for Full-Scale and Pilot Columns ........................4
C. Pilot Scale vs. Full-Sized Column...........................................................4
D. Ethyl Mercaptan......................................................................................5

2
 ADEPT SCIENCE & TECHNOLOGIES, LLC
DOCKET #11649 FINAL REPORT

I. Explanation of Changes to Work Originally Proposed

This Addendum is to provide further insight into how the scope of this project
evolved; how decisions were reached regarding design elements, experimental
set-up, and procedures; and why certain initially intended steps were modified.

A. Multi Bed Sieve vs. Activated Carbon Column

When the original proposal was written, it was not yet known what strategy would
be most effective to reduce the contaminants of interest. ASCENT’s proposal
references a multi-bed sieve (i.e. a column containing more than one type of
adsorption media in separate layers or beds) and molecular sieve media
because it was then thought that this approach would have a high likelihood of
success to remove multiple contaminants. Much of the pre-award proposal was
inspired by a GPA research report (#991) dated June 11, 2001. However, soon
after beginning work, it was found that activated carbon had been used with
some success in the past to remove heavy ends contaminants from LP Gas.
The other filtering media mentioned in the same GPA report were eliminated
from consideration due to characteristics less desirable for use by LP Gas bulk
plant operators. Additionally, conversations with molecular sieve manufacturers
regarding their use in this application were not encouraging. It became apparent
from the prior work of Cosmo Engineering Co. (CEC), Phillips Petroleum Co.,
Petrolane, other project advisors’ experience, statements from others in the LP
Gas industry, and the experience of technical experts at Calgon Carbon in
removal of diesel contamination from propane, that activated carbon alone was
the most promising solution to remove the contaminants of interest (i.e. heavy-
ends residues). This was supported by a summary of CEC’s project results
which showed that in their two-column process [one entirely of activated carbon,
the other of molecular sieves (zeolite material)], the activated carbon fully
removed the bulk of heavy-ends residues, while the molecular sieves removed
sulfur and water. Removal of contaminants such as sulfur and water with
molecular sieves is also already well characterized in the aerosol propellant
industry, and not part of this project’s focus on heavy ends. At this point, it was
decided that zeolite materials would not be the tested adsorbents and that using
a multi-bed column would be unnecessary.

While there was then consensus that activated carbon would be the way to
proceed given past successes, available documentation and data from previous
tests was neither accessible nor comprehensive, particularly with regards to the
removal of a wide range of heavy-ends residues as opposed to simply “lube oil”
or No. 4 fuel oil as in the cases of the CEC project and the data available from
Mr. John O’Connor’s patent, respectively. Additionally, access to CEC’s project
reports alone turned out to be prohibitively expensive (in excess of the entire
PERC project budget) which was contrary to earlier expectations.

Economic feasibility of field implementation was also a consideration in the

selection of the particular activated carbon eventually tested.

3
 ADEPT SCIENCE & TECHNOLOGIES, LLC
DOCKET #11649 FINAL REPORT

B. Design and Calculations for Full-Scale and Pilot Columns

The full scale column was designed and constructed after extensive research
into prior work including CEC’s multi-million dollar decontamination project. This
step was undertaken given earlier expectations of affordable cooperation from
CEC. CEC was willing to cooperate with initial work on this project and provided
basic parameters (i.e. column size and flowrates) from their prior projects. At the
same time, ASCENT was working with project advisors to determine design
constraints for the full-sized column. These constraints and the rationale behind
their use are provided in Section VII of the Final Report. After determining the
maximum sized vessel based on these constraints, and including design
elements such as instrumentation ports, and a means to add and remove
activated carbon, ASCENT then submitted the revised design to CEC. CEC
recalculated and provided ASCENT with new estimates for the expected process
flowrates. This information from CEC was considered along with project
advisors’ flowrate recommendations including those of Mr. John O’Connor from
prior work at Phillips Petroleum, Co. An average of these inputs was used as the
target flowrate for the full-sized column and the basis for pilot-scale work. The
full scale column was constructed at a time when further work with CEC
appeared likely.

Scaling to the smaller sized column (560:1 by volume/ 67:1 by flowrate) was a
straightforward calculation to maintain the same flowrate per cross-sectional area
(i.e. velocity). The dimensions, scale, flow orientation, and flow rates were then
reviewed and approved by our advisors.

C. Pilot Scale vs. Full-Sized Column

Prior to proceeding, the work plan (as initially conceived) was reevaluated. It is
crucial to understand that prior to the two pilot plant tests, there was no first-hand
knowledge that the system would work to remove the heavy-ends of interest. It
was considered prudent, therefore, to use small scale tests to determine if we
were on the right path. To have not done this and proceeded directly to a full
sized 60-day field trial (as intended at the onset of the project) would have been
extremely risky and irresponsible. By conducting well thought-out experiments at
a smaller scale, ASCENT gained valuable information within the resources
available to this project. Such critical path data collection would not have been
possible by conducting a full-scale test as a first-and-only project step. Pilot
scale work, for example, allowed for taking the column to breakthrough and
exhaustion. It also allows for further development to be completed without using
highly costly resources (e.g. the large volume of LP Gas needed for full-scale
tests). This commonly used practice and research strategy was approved by the
project advisors. Even though the size of the testing vessel was scaled down, it
should be noted that the number of samples drawn and samples analyzed in
these tests far exceeded the four (4) sample analyses originally proposed (17
drawn, 13 analyzed). Thus, more activated carbon heavy-ends removal data
was collected than earlier anticipated.

4
 ADEPT SCIENCE & TECHNOLOGIES, LLC
DOCKET #11649 FINAL REPORT

D. Ethyl Mercaptan
The effect of this activated carbon process on ethyl mercaptan concentration,
although not part of the contract, was explored in the second series of tests to
provide useful data to address practical industry concerns. The results show the
relative rates at which ethyl mercaptan and other contaminants elute from the
column. Because analysis indicates that ethyl mercaptan is initially removed, the
panel of expert advisors suggested that a cautionary statement be included.

The data collected will help with both future developments of this
decontamination process and to take precautionary steps, if and when they may
be needed.