:LOW TEMPERATURE

thus compensating for part of the de- method to achieve maximum possible

"
i
krll Ilne: beginning of warm period
, dotted line: end of warm period
warm end of the regenerator
I
sorption heat which causes the N, to purification of the air in full-scale
be cooled over the gel. These are the plants.
reasons why the temperature curve in Method 5. Total combustion of the
the regenerator at the gel layer is hydrocarbons before the air enters the
' B cold end of the regenerator S-shaped, Figure 3. An appropriate separator should reduce the C3Hs to
"
6
B
length of the regenerator
C layer of adsorbent
quantity of heat storing material must 0.02 ppm (10% of the concentration
be placed adjacent to the gel layer that would pass through the old end
for the exchange of adsorption and of the regenerators). This requirement
desorption heat. The quantity of 0, is high, but is ~ossibleif the initial
additional filling mass taken up by the gel and given off content of C3H, in the air does not
to the N, stream amounts to about exceed 10 ppm.
0.5 to 0.75% of the production. ~f the bulk of the C,H, is to be con-
TOcheck the behavior of the hydro- verted, the temperature must be set
carbons in regenerative adsorption and 50 to 100°C higher than usual for the
desorption, air at a pressure of 5 atm.; combustion of C,H,, CSHR, CZHP
a temperature of 100°K; and contain- This will not be difficult if heat ex-
ing about 1 ppm C,H,, 1 ppm C,H,, changers for outlet and inlet air can
0.3 ppm C3&, 0.7 ppm C,H,, and a be built cheaply enough.
surplus of CO, was passed over a layer A great advantage of the precom-
of alumina gel or sovabeds in one bustion method would be that the
direction for a period of 15 min. Sub- air sepsation processes could be sub-
sequently the same quantity by weight jected to changes in gas and liquid
of pure N, at the same temperature flow to obtain desired results, without
and about atmospheric pressure was having to devise new safety measures
figure 3. Temperature curve in the re- passed over in the opposite direction. in each case. The safety measures dis-
generator with a gel layer. These operations were repeated alter- cussed under previous methods do not
nately over the course of several fulfill this requirement.
rough after a short time, and the weeks. The air remained practically
,,HA a little later. Operational free from hydrocarbons. The hydro- LITERATURE CITED
eriods of several months can be carbons' with the of 1. ~ ~ E,, Chem,~ ~ Progr.,
Eng. ~ 51, tNo. ,
hieved up to the break-through of amounts of C,H,, were found in the lo. p. 96-101, (1968').
,H, and even longer periods for recycled N,. We concluded from these 2. German Patents 633 277 and 534, 565, on
March 7, 1930: anh 8.h. 106, NOV.11.
experiments that it is possible by this 1949.
Method 4. The most effective way
air impurities is by
*,,,nnnnn
11111*1.1'.11*111**IIt*I**,,I,,,,,,,,,,,,,*,,,,,,,,,,,,,,,,,,,,,,,*I,,,,,,,,,,,
I,,,,,, "
,,,I*,*,,'I,,"." *,,*I,,*,,,.,,,",,
I.II*.*....***.UQ

located near the cold end

enerators (2). The impurities 6. T . WRIGHT
from the air during the Dominion Foundries and Steel L t d . (Canada)
riod are taken up during the
penod by the stream of N, pas-
over the gel. This method of
egenerative adsorption and desorp-
on was investigated in the labora-
Oxygen plant
tory and in full-&ale tests. The chief
measuremepts on the full-scale regen-
erator concerned the thermal varia-
vaporzzer exploszon
tions resulting from the introduction
of gel, while chemically, only the
adsorption and desorption of CO, An analysis of the causes led to plant -
was observed. The laboratory mea- changes for improving safe operation.
surements dealt with the behavior of
the hydrocarbons.
In the regenerator, the gel layer Hrai
PURITY OXYGEN SUPPLIED by split cycle-type. The intake air is
achieves equilibrium with the pres- three units with a total capacity of filtered through a 10-ply paper filter
surized air during the warm period 350 tons/day is required by the Do- and then compressed to 75 Ib./sq. in.
und with the stream of N, &ring the minion Foundries and Steel, Ltd. for gauge in a turbo compressor. It is
cwld period. In the warm period, an producing steel. At 8:10 P.M., Sep- water scrubbed to remove soluble im-
0,-rich 0,-N, mixture is taken up tember 30, 1959, the vaporizer in No. purities. Two thirds of this air passes
hv the gel, this mixture being given 2 unit exploded, wrecking the vapor- through aluminum-filled regenerators,
off again when the pressure is reduced izer and surrounding parts of the cold then into the high pressure column.
and during the rnsuing cold period. box, Figures 1 and 2. No one was in- The rest of the air is compressed
The respective quantities of adsorp- jured and production loss was mini- to 350-550 Ib./sq. in. gauge, goes
tton and desorption heat are distri- mized because the No. 3 spare 150 ton through caustic scrubbers, heat ex-
h t e d between the gel and the gas. unit had gone on stream three weeks changers, then through an expansion
lleat taken up after leaving the gel earlier. engine, and into the high pressure
laver is given off by the air to the column. Rich liquid, produced in the
ailing mass, and from there it is Plant description high pressure column, then passes
t~kmup again by the recycled N?, The No. 2 plant is a 100 ton/day through silica-gel filters into the low

HtMlCAL ENGINEERING PROGRESS, (Vol. 57, No. 4)

Figure 1. Damage to plant before any repair work was
started. Vaporizer displaced from normal position can be Figure 2. View of plant after cleanup commenced shows
seen in foreground. the low pressure vessel (dented) in the foreground.

pressure column where further rectifi- sheet and are slotted to distribute the purity was very low. Subsequent ex-
cation takes place. Liquid 0, is then liquid 0, around the inside periphery aminations showed one vaporizer tube
pumped to &e main <aporizer which of the tube. N, at 65 lb./sq. in. con- had ruptured.
is flanged on top of the high pressure denses on the outside of the tube and
column. Gaseous 0, is withdrawn boils the 0, inside. Explosion at 8: 10
from the vaporizer through heat ex- The plant started in August, 1956, On the day of the explosion, the
changers to reciprocating compressors. and was derimed in November, 1956, wind had blown 5 to 15 mi./hr. from
The vaporizer is a vertical shell and because of mechanical trouble with the direction of our coke oven battery
tube unit about 5 ft. in diameter and the main blower. It operated continu- and rain had fallen in the afternoon.
25-ft. high. It is a failing-film type ously until April 23, 1958, when a At about 5 P.M. the flexotimers,
and has 1600 X-in. copper tubes in minor explosion occurred. At this time which control the reversal of the re-
a stainless steel shell. These tubes the booster compressor stopped mo- generators, stuck, and one regenerator
project 6 in. through the top tube mentarily, and after start-up, the o? warmed up at the mid point from

Figure 3. Close-up of tubes in vaporizer and bottom tube

sheet at right. The line of detonation can be seen to be Figure 4. A close-up of an individual tube shows a fracture
approx. 6-8in. above the tube sheet. characteristic of acetylene explosions.

CHEMICAL ENGINEERING PROGRESS, ('401. 57. No. 4)

 LOW TEMPERATURE
-150" to +60°. This also lowered
1e liquid levels in the low pressure
olumn. After 28-33 min., the regen-
rators began reversing and the liquid
wels began building up. At 7:30 the
dant was operating more or less
omally until 8:10, when the explo-
ion occurred.
Approximately 30 tubes in the
aporizer burst 6 to 8 in. above the
ottom tube sheet, Figure 3. The
xce of tlie explosion ruptured the
tainless steel outer shell and blew out
he side of the cold box and building
losest to the vaporizer.
An independent research organiza-
ion, The Ontario Research Founda-
ion of Toronto, was given the job of
tudying the data to try and deter-
line the causes of this explosion and
he previous tube detonation in 1958.
Judging from the appearance of
he fracture of the tubes, Figure 4, the
xplosion was apparently caused by
n accumulation of hydrocarbons in
he vaporizer, probably C2H2,which
ras triggered by the warm up of the Figure 5. Catalytic filter installed on No. 2 plant. Air is heated 320" by the
lant. steam heat exchanger at the right.
:auses of the explosion removing all the C2H, from the rich should the regenerator temperature go
The following factors contributed, liquid. above -50". The plant will be shut
3 the explosion: 7 . An air pollution study indicated down should this occur.
I. The warming up of the regener- there were several sources of contam- 2. An auxiliary vaporizer has been
tor released hydrocarbons which ination in the district. The most criti- installed in the No. 2 plant which
rould go into the high pressure cal was Dofasco's battery of 105 coke can be derimed without shutting
ovens. down the main plant.
olumn. A subsequent analysis of the
8. In spite of the fact that some continued on next page
erime exhausts from the regenerators
f No. 1 plant showed that large intake contamination was evident, no 1111111*11'1'1"1111IIIIII,IIIIIIIII,II'IIII....1IIIIIaI'II~tIIIIIttII,IIII~

olumes of CO, would also pass into positive C,H, test had ever been re- Table 1. Acetylene contaminant levels
he high pressure column. The warm in plant operation.
corded after the rich-liquid filters. A
p of the regenerator also caused the test of the rich-liquid filters on the No. 1 No3
emperature of the vaporizer to fluc- No. 2 and No. 3 plants after one LOCATION PLANT PLANT
'uate. month's operation showed less than After compressor 0.026 pprn 0.024 pprn
0.2 g., Table 2. This indicates the After catalyticfilter no filter 0.015 pprn
2. The foundation of the cold box
rich-Equid filters were operating at After watkr
llad previously heaved from frost, and 0.032 pprn 0.013 pprn
:awed hie vaporizer to be pitched 99%efficiency. scrubber
After regenerator 0.028 ppm 0.015 ppm
tway from the side where the explo- Plant safety changes After exchanger 0.055 ppm
iion took place. It is possible this The following changes were made Bottom of G.P.
taused some tubes to be starved of to improve the operation of the-~lant Cof. 0.12 ppm 0.051 ppm
liquid. and prevent any further explosions: After rich-liquid
3. The wind was very gentle and filter 0 0
1. Alaims have been installed on Bottom of L,P.
from the direction of the coke ovens. the three plants to warn the operator col. 0 0
It is probable that a higher than
normal level of contamination existed. filter 0 0
4. A freshly regenerated rich-liquid G. T. Wright is Vaporizer 0 0
Mechanical Superin-
filter had been put into senice' 12 hr. tendent of Domin-
before the explosion. This Sovabead ion Foundries and Table 2. Contaminants on the rich-
showed traces of C,H, on the inlet Steel, Ltd., Hamil- liquid and liquid oxygen fiiters in NO.
ton, Ontario, where 2 and No. 3 olants.
-
side of the filter, but none on the he has worked in
exit side. RICH-LIQ. LIQUID0 2
maintenance a n d FILTER FiLTEU
5. Because of production require- eng~neering since
No. 2 No. 3 No. 2 No. 3
ments, the plant had been on stream 1939..His depart-
ment IS responsible
----
for 17 months which is longer than NO,, g. 1.1 0.27<0.015 <0.04
for the mechanical maintenance of the KO, g. <200 700 700 900
usual. plant and generation of steam, com-
6. The contaminants after each unit pressed air, and oxygen. Wright received co,, g. 1500 4500 130 800
a B.Sc. in Mechanical Engineering from CZHZ, g. 52 0.46 <0.01 <0.18
in the plant .in normal operation were Hydro-
Queens Univ., Kingston, Ontario. He is
; checked, Table 1. Everything was a member of the Association of Pro- car-
normal and the silica-gel filter was fessional Engineers of Ontario. bons, g. <80 120 1000 600

CHEMICAL ENGINEERING PROGRESS, (Vol. 57. No. 4 ) April 1961 II

 3. The capacity of the liquid 0, and act as an oil filter on the dis- 10. Level recorders have been in-
pumps has been increased 25%to give charge of the booster compressor. stalled on the separator after
more liquid for flushing the vaporizer 7. An additional silica gel filter has auxiliary vaporizer and low pressure
hbes. been installed on the discharge of the column. Warning lights indicate ab-
4. The heaved foundation was re- liquid 0, pumps. normal levels.
placed with a ventilated slab to pre- 8. The cold box has been simpli-
vent frost heaving. 11. A derirning schedule of 9
fied with the removal of an argon months has been set for No. 1 and
5. M.S.A. catalytic filters have been column and the installation of stain- No. 2 plants, and 12 months for NO.
installed on all three plants, Figure 5. less steel purge lines instead of carbon 3 plant. The No. 3 plant can operate
6. Activated alumina driers were steel piping. longer safely because it has 100%
installed in place of the lump caustic 9. The cold box casing has been product purge and has consistently
desiccator. This will give a lower dew made air tight and continuously operated at a lower hydrocarbon level.
point going into tlie warm-exchangers purged with N,.

Questions and answers up the tubes, opening these holes

along the tubes until they reached
N. H, WALTON-SunOlin Chem. Co.,
Claymont,Del.: I am not quite certain
the interface level where, as you have as to what your conclusion was on
HEW-Sun Oil Co., Marcus Hook, indicated, there was no liquid wash- the initiator of the explosion. The
Pa.: I couldn't help but think of the ing. When it reached this point there reason I'm confused, perhaps, is be-
similarity between this explosion and was a large accumulation of hydro- cause you said that you never got a
our own of four years ago. The dam- carbons which erupted with the positive test for C,H, at that point,
age to the reboiler is remarkably major force of the blast. and yet it appears, that there was an
similar. We reached a little different We were able to conclude to our initiator there. The wind from the
conclusion, I believe, about these dam- satisfaction that there were two dis- coke oven battery makes me wonder
aged tubes. We found a lot of tubes tinct explosions, a small triggering whether oxides of nitrogen might be
that were split open, and in our case explosion followed by the main energy the initiator here.
where we had a long enough reboiler, release, and they were at entirely WRIGHT: That might be possible.
there were a number of tubes which separate places in the reboiler. The reason for blaming C,H, in spite
had more than one opening per tube. WRIGHT: We did notice one or two of the fact that we had no positive
These were very equally spaced (about tubes which had been broken in more test, was Ontario Research Founda-
14 in.) and it appeared to be a phe- than one place. tion's statement that this fracture was
nomenon of a pressure wave going KEZTH-Hydrocarbon Research, N. Y., definitely characteristic of C,H,. Cer-
through the tube which eventually N. Y.: I do not recall whether you tainly, we have no proof that there
exceeds the stress of the tube and mentioned at what pressure this sys- were not other contaminants in there
opens it up at these points. tem was operating. Also, I want to -in fact, we are fairly certain that
You perhaps might not have had know if there was any indication of there were.
enough length to observe this, but we any formation of acetylides of copper. J. E. HART-Dow Chemical Co., Mid-
found apparently the same mecha- WRIGHT: There was no indication of land, Mich.: What is the frequency
nism that you noted. There was a any acetylides formation. The system of your acetylene tests?
minor explosion in the bottom of the operates at 65 to 75 lb. in the vapor- WRIGHT: The vaporizer and low
reboiler under the tube sheet, which izer. In the O2 side there is a pressure pressure column are checked every 8
buckled the tube sheet. Then there of from 3 to 4 lb., outside the tubes hr, by the B.O.C. method and the
was a progressive pressure wave going N, is at 65 to 75 lb./sq. in. Ilosvay method.

L. G. MATTHEWSdispensary and released. A twelfth em-

ployee, who was outside the building,
Linde Co. received the most exposure and was
hospital-treated for minor bums. He
0
- plant expbsion was back at work the following morn-
ing.
Cause of explosion
Hydrocarbon concentrations in main It has been concluded that this
condenser believed cause of detonation. flash was probably caused by concen-
tration of a quantity of hydrocarbon
AN ExpLOslON Tm '
Linde's on-site oxygen plant serving
and compressors. Except for the con-
unit of denser and column of the No. 1 unit,
material in the main condenser. The
identity of the exact hydrocarbon or
hydrocarbons involved has not been
amage to production equipment was
DuPont at West Oc- not extensive. There was no damage ascertained, nor has the mechanism
curred at about 2:20 PSm. On to the No. 2 unit, though the flash of ignition been pin-pointed. The con-
20, 1960. The flash resulted in des- in N,, 1 resulted in its shutdown, ~h~ centration of hydrocarbon material in
truction of the main condenser (re- No. 2 unit was withheld from service the condenser was probably due to
boiler) and a portion of a fractionating until after a complete check. impaired circulation between the con-
column. Externallv. an insulation silo denser and a silica gel trap, designed
was severely damiged, and there was Eleven employees, who were in- to keep the hydrocarbon concentration
some damage to the transite-sheathed side the building, received minor in- in the condenser at an acceptable
building housing the operating panels juries and were treated at the plant level.

12 April1961 CHEMICAL ENGINEERINGPROGRESS,(VOI. 57, NO. 4)