0
Carbonyl Compounds as By-products
C. M. FINIGAN, H. V. FULLERTON, AND G. W. TAYLOR Technical Division, Polymer Corp., Lfd., Samia, Onfario, Canada
S
UBSEQUENT to the use, by Polymer Corp., of a calcium-nickel phosphate catalyst containing 2% of chromiuin sesquioxide for the commercial production of 1,3-butadiene by n-butene dehydrogenation, it was discovered that approximately 0.75% of the hydrocarbon feed was being converted to ratersoluble carbonyl-type compounds. An investigation of this side reaction was initiated by (1) the substantial loss of hydrocarbon involved and (2) the potential value of the by-products produced. A description and flow plan of the general features of the commercial dehydrogenation process (Jersey process) have been published by Russell, hIurphree, and Asbury (2). .il recent publication by Britton, Dietzler, and Noddings includes complete operating details for the use of a calcium-nickel phosphate cntalyst in the Jersey process (1). The present study furnishes information on total carbonyl production during the life of the catalyst and provides a quantitative distribution of the individual carbonyl constituents formed. The data were obtained from t x o commercial units operating on alternating l-hour periods of dehydrogenation and air regeneration.
Reaction Temperature. During the initial start-up on a fresh charge of catalyst, mixed feed temperatures are raised slowly from approximately 1000" F. to the normal operating temperatures of 1160' to 1200O F. Carbonyl production for this period, as plotted in Figure 2. increased with mixed feed temperatures up to a temperature of 1130" F. Above this point very little increase occurred and a maximum conversion of about 1% was indicated. These data were obtained over an 18-day period, during which time other reaction conditions were maintained as constant as possible. Age 'of Catalyst. Carbonyl conversion levels were determined over complete catalyst runs and the information obtained is plotted in Figure 3, for two lots of catalyst in two dehydrogenation units. Each graphical line is an average of two reactors on
Production of Carbonyl-Type Compounds Coincident with the use of a calcium-nickel phosphate catalyst in these dehydrogenation units, abnormal operation of the butadiene extraction section occurred. Analyses of various plant streams for extraneous materials revealed the presence of substantial amounts of mater-soluble carbonyl compounds in the products from the dehydrogenation reaction. These carbonyl compounds were then determined quantitatively a t regular intervals over the complete life of the catalyst.
T O W E T TEST
KNOCXOUT
POT
Experimental Work The sampling apparatus used for the determination of total carbonyl compounds in the products from the dehydrogenation reaction is shown in Figure 1.
HYDROCHLQRIDE
Figure 1.
Sampling Apparatus for Determination of Total Carbonyl
The hot reactor products from the waste heat boiler were passed through a water-cooled condenser into a 1-liter knockout pot and then through three consecutive bubblers to a wet-test meter. The first two bubblers each contained 100 ml. of distilled water and the last bubbler contained 100 ml. of hydroxylamine hydrochloride solution (20 grams per liter). Condensed steam was collected in the knockout pot and bubblers, while noncondensable gases were measured by the wet-test meter. IneachcaselOO liters of noncondensable gas were passed through the train over a period of 40 minutes (10 minutes after start of reaction to 10 minutes before reaction completion). The water in the knockout pot and bubblers was analyzed for total carbonyls by the hydroxylamine method, and the carbonyls were calculated as weight per cent acetone based on the noncondensable gas. This value approximates the percentage of hydrocarbon feed converted to carbonyls and was used as such in this investigation, Carbonyl Production Carbonyl production, as determined a t regular intervals over 7-month operating periods, was found to be dependent on both the reaction temperature and the age of the catalyst.
1070
1080
1090
I100
Ill0
1120
ill0
MlXED F E E D TEMPERATURE
Figure 2.
1356
1140
1150
1160
'F
Carbonyl Production vs. M i x e d Feed Temperature
June 1953
INDUSTRIAL AND ENGINEEgING CHEMISTRY
1357
column, 4 feet in height, 1 inch in inside diameter, and filled with Heli-Grid packing. One hundred grams of water chaser were added to facilitate complete carbonyl recovery. The overhead was removed a t an average rate of 20 grams per hour with the column operating a t a reflux ratio of 30 to 1. This distillation permitted separation of major constituents and provided fractions of known boiling range for mass spectrometer analyses. The observed distillation curve is shown in Figure 5, where weight per cent of net recovery is plotted against overhead temperature. The 14 fractions denoted numerically were those chosen for mass spectrometer analyses. The portion boiling below 27" C. was separated into three fractions by further distillation in a low temperature Podbielniak column.
-
_c
REACTOR PRODUCT FROM WASTE HEAT BMLER CATALIST
Figure 3.
n
AOL IN MONTHS
Carbonyl Production vs. Catalyst A g e
-WATER
OUT
-WATER
IN
FINTUBE EXCHPiNGER
the same lot of catalyst. The maximum conversion of approximately 1% of the hydrocarbon feed to carbonyls gradually decreased to about 0.5% after 6 months' continuous catalyst operation. Mixed feed temperatures for this period were raised from 1160' to 1200' F. and this increase possibly prevented carbonyl production from decreasing more rapidly. Other operating conditions were relatively constant. These data have been obtained from a large scale operation where reaction conditions were difficult to keep constant. The information is presented only to indicate the general trend in carbonyl formation as experienced in two commercial units under a specific set of operating conditions. Conditions more favorable to carbonyl formation are possible and this subject could form a basis for further investigation.
"
*
The reactor product gases were drawn off immediately doanstream of the waste heat boiler, and passed through a watercooled &-tube condenser, and into a knockout pot. The condensate was collected from the bottom of the pot in 40-gallon drums, and noncondensed gases were vented to the atmosphere. Twenty drums of condensate were collected at an average condensate temperature of 80" F. Carbonyl recovery for this operation as determined by analyses before and after condensation approximated 85%.
Concentration of Carbonyl Constituents. Concentration of carbonyl constituents in the 20 drums of reactor product water was effected by continuous distillation in a column, 20 feet in height, 3 inches in inside diameter, and packed with a/,-inch Berl saddles. The feed was preheated and entered the column 5 feet below the top a t a constant rate of 0.30 gallon per minute. The column was operated to distill a minimukn of 95% of the carbonyls overhead by continuously discarding the bottoms if the carbonyl content was 0.005 weight % or less. Forty-five hundred grams of carbonyl concentrate were collected a t an overhead temperature of 55' to 70" C. and a t an average reflux ratio of 25 to 1. Upon completion of feed addition, the tower temperature was raised and residual carbonyls were collected up to a n overhead temperature of 96" C. Total distillation time approximated 50 hours. Fractionation of Carbonyl Concentrate. Five hundred grams of the carbonyl concentrate were charged to a Hypercal
b
WDROCARBOH
KN~ouT/
Y 4 5 G A L L O N DRUM
Figure 4.
Equipment for Collection of Reaction Product Condensate
Figure 5.
Hypercal Distillation of Carbonyl Concentrate
100
Component Distribution of Carbonyl Constituents A quantitative component breakdown of the carbonyl compounds produced is a requisite to the estimation of their potential value as by-products and provides data necessary for the design of a removal and/or a recovery system for these compounds. The component breakdown was obtained by the following sequence of operations: Collection and Condensation of Sample. A dilute aqueous solution (about 0.15%) of carbonyl compounds was obtained by condensation of the products from the dehydrogenation reaction. A sketch of the equipment used is shown in Figure 4.
TO PURIFIChTION
90
80
* 70
$ 9 60 f
z 50
40
30 WElQHT % OF MET
RfCOVERl
Mass Spectrometer Analyses of Distillation Fractions. The 14 fractions from these distillations were quantitatively analyzed by mass spectrometry using a Consolidated instrument, Model 21-102. Hypercal samples were introduced in the liquid phase and pod fractions in the gas phase. Calibrating liquids were run where available in the same manner as the Hypercal samples. Other calibrations were obtained from the library of mass spectra provided by the American Petroleum Institute. As only estimations of water contents could be made using a mass spectrometer with a glass inlet system, this component was quantitatively determined by a Karl Fischer titration method. Identification of all components in a sample was considered complete when the entire mass spectrum of the sample was substantially accounted for as the sum of the spectra of pure coni-
I N D U S T R I A L AND E N G I N E E R I N G CHEMISTRY
1358
Vol. 45, No. 6
Table I. Mass Spectrometer Analyses of Distillation Fractions Fraction N o . Boiling range, a C. Net recovery, % Component Acetaldeh de Propionadehy de Acetone Butyraldehydes Methyl ethyl ketone Methyl vinyl ketone Methacrolein Butenes Butadiene Dimethylacetylene Benzene Toluene Water Other components
Table 11.
1 otal. The curve rises to a second plateau, which represents the methyl ethyl ketone-water azeotrope and includes a major portion of the methyl vinyl ketone. The curve then rises rapidly to the boiling point of water, indicating a minimum of higher boiling constituents. The over-all analysis of the total carbonyl mixture is presented in order of decreasing abundance in Table 11. The t'wo major carbonyl constituents, acetone and methyl ethyl ketone (ill an approximate ratio of 2.5 t'o 1)) constitute over 83% of the total concentrate. Other carbonyl compounds include 2.4% methyl vinyl ketone, 0.7% acetaldehyde, 0.2% butyraldehyie, 0.2% methacrolein, and 0.1 % propionaldehyde. Two points should be noted from these analyses: (1) The main distillation fraction was 99.4 weight % acetone and contained 88% of the total acetone present; and ( 2 ) the second major cut contained approsiniately 93% of the total methyl ethyl ketone present in a purity of over 80%.
..
..
2.4
11
74-76
.. 0:4 0.3
..
..
..
10 71-74 26.8
..
..
Composite Analysis of Carbonyl Concentrate
Component Acetone Methylethyl ketone Water Methyl vinyl ketone 1.3-Butadiene Butenes Acetaldeh de But yraldex yde Methacrolein Toluene Benzene Dimethylaoetylene Propionaldehyde Other components Total Overhead not analyzed Bottoms
6 55-56 5.6
Test No. 1
2
3
Acetylene5 in Feed Mole Total Description Yo mole Blank Trace Trace Blank + m e t h y l 1.7 0.077 acetylene Blank ethyl. 2.3 0.077 acetylene
+-
180 grams per hour 545 grams per hour 147 grams 1157' F. Carbonyls in Product, Total Mole 0.0077 0.0290
Acetylene Couversion t o Carbonyls,
0.0223
19.2
%
27 ,' 7
lune 1953
INDUSTRIAL AND ENGINEERING CHEMISTRY
approximately 2 mole % methyl and ethyl acetylene. The data, given in Table IV, indicate an approximate threefold increase in total carbonyl production over that obtained from butene feed containing only traces of acetylenic compounds. This is a definite indication that the methyl- arid ethylacetylene have contributed to an increase in carbonyl formation. These data are preliminary and are presented only to suggest a logical mechanism by which carbonyl-type constituents might be formed during the dehydrogenation reaction. Acknowledgment The authors appreciate the courtesy of Polymer Corp. in extending permission to publish this paper. The cooperation
1359
and assistance given by chemical engineering personnel, the Research and Development Division, and various laboratory groups are gratefully acknowledged. literature Cited Britton, E. C., Dietrler, A. J., and Noddings, C . R., IND.ENG. CHEM.,43,2871-4 (1951). ( 2 ) Russell, R. P., Murphree, E. V., and Asbury, TT. C., Trans. Am. Inst. Chem. Engrs., 42, 1-14 (1946). (1)
RECEIVEDfor review December 8, l E 2 . A C C E P ~ EFebruary D 1 6 , 1953. Presented before the Division of Industrial and Engineering Chemistry a t the 122nd Meeting of the AMERICANCHEMICALSOCIETY,Atlantic City, N. J.
0 0
Hydraulic Analog for Studying Steady-State Heat Exchangers
0
0
1.
s. JUHASZ' AND
F.
C. HOOPER
Deportment of Mechanical Engineering, University o f Toronfo, Toronfo, Ontorio, Canada
HE use of hydraulic analogy models for the solution of unsteady-state heat transfer problems is established (1-6). Such analogs permit relatively simple solution of problems in which direct mathematical solution is very difficult. These models are built of vertical glass tubes which are connected by capillary tubes. Liquid flows through the capillaries and the change in level with time in the vertical glass tubes is observed. Time and liquid level in the model correspond to time and temperature, respectively, in the prototype. There does not appear to have been any application of the hydraulic analog to cases of steady-state heat transfer in heat exchangers. There are two apparent reasons for this. First, many steady-state problems are relatively simple and yield readily to the methods of solution of the calculus and analogy methods have therefore not been necessary. Second, because time is not a variable in the steady-state case, the time-to-time correspondence of the conventional hydraulic analog is not applicable. There are, however, many steady-state problems in heat exchangers which are not sufficiently simple in a mathematical sense to be solved by simple calculus. To deal with these problems a new dimensional equivalency has been introduced, following a personal suggestion made by Matts Backstrom of the Royal Institute of Technology in Stockholm, Sweden. In this analog, time in the model is equivalent to area in the heat exchanger or more precisely to relative area of the heat exchange surface. It might be a little difficult to grasp immediately the meaning of this equivalency, but its significance will be made clear in this paper. The level of the liquid in the tubes is equivalent to the temperature in the actual heat exchanger fluids, and quantities of liquid are proportional to heat quantities in the exchanger as in the unsteady analogs. By using capillary tubing to connect the vertical tubes, advantage is taken of the direct proportionality of pressure head to flow which results from the laminar flow in the capillary passages. To date the andog has been applied only to parallel flow and 1 Present
Maas.
address, Massnchusetta Institute of Teohnology, Cambridge,
modified parallel flow heat exchanger problems, including those with change of phase. However it is hoped to extend it to cover cases of counterflow, crossflow, combination flow (6),and other complex problems. Derivation of Equivalency The equivalency for parallel flow is established in the following derivation, where the symbols correspond to the arrangement shown in Figure 1.
Figure 1.
Diagrammatic Layout
of Simple Parallel Flow Analog
Choose the conversions: tl = ntz1
(1)
12tZ.2
(2)
y = n,B
(3)
t2
=
Now for any heat exchanger -dtiwicp, dZ/AU (ti or
- tz)
(44
