Determination of Oil in Refinery Effluent Waters Report of the Subcommittee on Refinery Effluent Water Analytical Methods, Committee on Analytical Research, Division of Refining, American Petroleum Institute J. 6. RATHER, Jr., Chairman Socony Mobil Oil Co., Inc., Brooklyn 22, N.
Y.
J. A. GRANT, The American Oil Co. V. H. GUNTHER, Shell Oil Co., Inc. C. E. HEADINGTON, The Atlantic Refining Co. J. H. KARCHMER, Humble Oil & Refining Co. S. S. KURTZ, Jr., Sun Oil Co. A. R. RESCORLA, Cities Service Research and Development Co. F. M. ROBERTS, The Texas Co. A. J. STEPHENSON, Imperial Oil, Ltd. W. J. TROELLER, Jr., Esso Research and Engineering Co. E. 6. TUCKER, Standard Oil Co. (Indiana)
A , subcommittee o f the Committee on Analytical Research of the American Petroleum Institute recently has completed a cooperative study o f three procedures for the determination of total oily matter in refinery effluent water. In the test program, an infrared technique and a direct extraction-pycnometer method were comp a r e d with one of t w o approved API methods. The investigation was initiated because the group believes that there i s a need for a more r a p i d method which i s a t least as accurate as the API procedures currently published. The results indicate that the infrared i s the best of the three methods over-all. The direct extraction-pycnometer procedure has certain advantages, particularly where volatile materials are present and where the oil concentration i s low.
S
1948 the Subcommittee on Refinery Effluent Water Analytical Methods of the American Petroleum Institute’s Committee on Analytical Re-?arch has been engaged in investigations aimed a t the development of improved procedures for the determination of oily content of refinery effluent water. Hart ( 2 ) gave a review of the original M ork ivithin the Amwican Petroleum Institute (API) on this subject nhich, in about 1932, led to the development of the U-tube method by the Atlantic Refining Co. The latter procedure is published as method 73253, “Determination of Non-T’olatile Oily Material, Flocculation-Extraction U-Tube Method” ( 1 ) . The importance of this matter to the ISCE
36
ANALYTICAL CHEMISTRY
petroleum refiner was reported hjHeadington, previous chairman of the subcommittee, in 1953 (3)’ who discussed the experiences of several participating laboratorirs using different oil content methods and slion-cd coniparative test data. As a result of this earlier work by the subcommit’tee, the API Committee on Disposal of Refinery \Tastes has publislid a second approved procedure. nietliod 731-53, “Determination of Tolatile and Yon-Tolatile Oily IIaterial, Reflux Distillation-Extraction hlethod” [ I ) . This is superior to rnebhod 732-53 in detrrniining total oily material in n.astc water. ITMe tlie development of nicthod 731-53 represented a definite atlv:mce. the subcommittee has continucd it? work with the objective of perfecting methods which are less time-consuming nithout sacrificing accuracy. A need was felt for more accuracy \rhere volatile hydrocarbons are present in the oil contained in the effluent \rater. Tlie infrared met’hod ( 7 , 8) and the dircct cxtraction-pycnometer method ( b ) ,both of which are discussed in the Headington report (S),offered promise,. dccordinglj-, a statistically designed program n-as undertaken to compare the two methods, against each other and with method i31-53. -211 three methods are intended t o nipasure total oily material in thc cfflucnt. n-hich includes hydrocarbons plus organic materials which are soluble in t h r solvents used.
w a t (lr with L‘a r lion tetra c lilo rid e. Tlie infrared absorption of the rstract is then mc,asured a t selected wave lengths. Tlie method used in the cooperative program diffew somewhat in details from the procedure as published. The modification K B S necessary because the published method descrihcd a procedure for tieteriiiining phenols as n-ell as oil in refinery effluent. The present program \\-as concerned only with the dcterniination of oil content. In addition to the c~liminatioiiof all procedural stcps for p h m d . tlir following modifications \v(cr(>niacle :
PROCEDURES
Direct Extraction - Pycnometer Method, The oil is eutracted from the n-ater with carbon tetrachloride. The weight of an accurately mcasured
Infrared Method. This method involves extraction of thr oil froin the
The size of the s,iniple ut’ w 5 t e n-ater to be tested is 2 . 5 liters instead of a variable quant’ity. Sodium chloride. in a&Iitioii t,o hydrochloric acid. is acltleri to the s:imple prior to extraction. The volume of c:irl)on tetracliloride is fixed at 50 nil. Ten milliliters cif the cnr1)on tetrachloride solution is renio\-etl froni the sample to fill the cell. If available. oil sin>.ilart o that found in refinery effluent ~ h o u l t lhe used for calibrat,ing the inatriiment. If not, t,he mixture oi iso-octane. cet:ine, and benzene is u;ed. dbsorbniice readingc are taken a t approximatelj- 3.42 microns and a t ex:ietly 3..iO microns. An infrared instrunleiit with a sodium chloride prism can be used, whereas formerly :i lithiiun fluoride prism was needed brc:iiise of the phenol determination.
volume of t h e carbon tetrachloride extract is compared to a n equal amountof p,lre solvent. The assumptioll of an average density of t h e oil One to the amount Of Oil* 'Ompared the published g method, there are t h e f o l l o ~ ~ i nmodifications in t h e ilrocedure cooperatively tested:
portions of carbon tetrachloride. This has been replaced by one extraction with 75 ml. of the solvent, niaking the method simpler and more rapid nithout a n y loss in precision. The pycnometer is calibrated for the volume of carbon tetrachloride a t different heights of solvent in the arms of the pycnometer, a s well as for weight of solvent. Only a few simple calculations are necessnrv. The calculation
volume of carbon tetrachloride solution in the pycnometer. The tare used when weighing the pycnometer is prepared by partially filling a 50-ml. volumetric flask Kith carbon tetrachloride, adjusting the weight desired with a little mercury, and sealing off the top of the flask. Originally, the tare was prepared by sealing both ends of a 50-ml. pipet.
includes the ratio,
Copies of the revised versions of the above methods were made available t o
*-
The published method calls for three extractions of the sample with 25-ml.
Table I.
10
~1
nilere
17
=
Oil Content Determinations on Synthetic Samples
Infrared Method Direct Extraction-Pycnometer Method API Method 731-53 Lab- 1OP.P.11. 50P.P.M. 125P.P.11. lOP.P.11. 5OP.P.lf. 125 P.P.M. ' 10P.P.M. 50P.P.M. 125P.PM. oraOil Oil Oil 011 Oil Oil Oil Oil Oil tory Added Found Added Found Added Found Added Found Added Found Added Found Added Found Added Found Added Found Samples Made with Oil 1 1
10 11
10 13
52 53
44 49
127
111
4
12 17 17
12 13 19 19
72 59 45 46
72 59 49 48
120 140 124
126 142 114 135
12
12
GO
57
122
123
11
5
6 7 8
11
10
63 63
7G 51
140 140
147 143
15
13 9
49 47 75
54 42 75
122 139 137 133 125 129 118 145
121 139 147 133 129 130 95 150
16 8
0 0
11
11
123
112 119 91 113
11 13
2 3
12 11 11
19 16 10
10
19 15 20 15
51 52 50 50
44 43 75 82
115
98 100 125 125
69 67 175 187
13 14 1G
17 10 12 13 23 16 25 22 23
7 8 9
11
66
16 15 16 15 6 30
44 46 56 63
45 49 50 64
51 50 36 50
34 56 32 46
5
25 20 20
67
115
101 114
60
12
11 11
48 48
49 47 54 58 41 57
12 12 16
6 8 20 20
52 .~ 52 51 54
53 56 52
110 110 111
86 117 119
16
60
39
121
103
1
56 ..
10
11 11
12 8
66 61 50 46
12 12 16 17
12 12 14 12
45 _. 45 53 53
16
0
14 13 35 36
10 25 16
9 13
18
9
59 53 53
41 __
133 145 126 119 126 116
110 -__
127 131 129 123 119 112
94 .~
Samples Made with Oil 2 1
2 3 4 5 6 7 8
10 11
9 10 13 13 9 8 8 10 10
10 12 10 10 15 12 10
8 8 10 13
51 56 41 51 51 52 60 51 52 50 50
58 71 44 53 55 57 61 50 46 80 62
127 130 139 141 128 124 125 106 97 125 125
142 149 148 147 135 125 122 97 97 125 150
13 13 13 12 13 13 14 14 10 16 10 17 20 37
10 9 8
13 11 8
13
11
11 14 31 31 23 45
63 63 49 52 58 58 53 52 53 51 60 58 42 58
74 72 52 49 54 53 53 34 55
45 71 83 44 57
125 125 106 107 131 121 125 126 137 140 128 136 122 119
117 129 108 105 105 97 128 131 123 114 150 160 130 123
12 10 10 21 21
31 --
115 -
94
64 65 41 31
118
129 105 102
114 137 100 84
47 38 51 50
144 --_
112 -~
133 99 100
118
137 127
81 75 94 98 67
83 89
Samples Made with Oil 3 1
2 3 4 5
6 7 8
10 10 10 10 10 10
9 10 11
10 6
16
13
12
13 34 10 10
14 36 12 10
50 50 48 A6
52 52 50
42 49 50 50
42 43 50
58 43 35
129 129 112 126 125 126
49
110
43 52 55 56
88
92
125 125
123 133 112 135 110 103 106 75 82 130 140
13 13 16 13
17 16 12
11
16 16
11
17 15 23
14 14 12 19 9 11
14 4 12
11
10 17 16 23
63 63 53 55 70 50 54 51 64 64 61 60 39 40
46 37 42 50 51 51 44 30 36 42 78 78 34 34
125 125 104 123 133 110 121 123 183
123 154 141 106 126
81
84 99 110 112 91 86
121
97 67 168 175 100 116
10 10 16 15
11
11 11
16 14
63 64 54 61
53 49 31 43
124 127 122
67 67 44 51
59 53 34 29
130 130 96 108
118
VOL. 30, NO. 1, JANUARY 1958
76
132 100 56 56
37
~______
Table
Infrared Method Laboratory
10
50
P.P.,rn.
P.P.p.
125
p.p.,m.
011
011
110 118
85 93
90
109
89
011
II.
Direct Extraction-Pycnometer Method 10
p.p,m. 011
API Method 731-53
125
10
50
125
P.P,m.
P.P:rn.
P.P:m*
011
011
p.p.m. oil
0 0 0 99 106 102 85 92 88
83 89 86 101 97 99 85 119 102
90 95 92 102 104 103 94 97 96
55 66 60
79 101
85 78 82 106 107 106
50
p.p,m.
Per Cent of Oil Recovered
011
011
Samples Made with Oil 1 1
2 3 4 5
6 7
Av. Av. Av.
106 107 106
Av.
114 112
101 101 101 105 110 108
102
96
110
Av. Av.
94
101
120 81
102 105
106 89 110 100 101 101 101 101 106 104 89 101
104 99 100 100
95 66
63 87 91
89 89 92
91 91 103 97 92 99
125
90 109 114
89
90
96
118
112
85 28-160
98 66-120
99 81-107
74 0-125
99 79-119
10 133 72 109 109
55 65 60 97 107
109 38 57 48
102 67 82
83 97 106 102 83 95
75
89
111
100 107 104 101 103 102 81 101
97 150 200 175
85 150 164
67 71 69 140 149
157
144
Grand av. % ’ recovery % Recovery range
105 94-118
96 84-100
93 67-109
1
100 109 104
114 127
112 114
69 77
117 114
94 103
120
113
73 62 108 85 65 86
116 94 106 100 93 94
99 102 98 100 103 105
76 78 96
94 65 100
87 88 110
83 90 102
104 102 104 103 82 90
100
8a
Av. Av.
84 86
101 105 103 99 109 104
92 91 64 73 80 69 87 78 69 79 74 90 102 96 46 125 130 160 115 28 98
111
98 78-107
Samples Made with Oil 2
2 3 4 5
Av. Av. Av. Av.
101 112
104 108
107 92 116
106 107 109 108
104
the participating laboratories by the chairman of the subcommittee (6). API Method 731-53. T h e sample is refluxed through a t r a p t o collect the volatile oil. The oil collected in the reflux trap is measured volumetrically. The remaining sample is extracted with two successive portions of benzene, using mechanical agitation. The benzene layers are separated from the water following each extraction, combined, and reduced to approximately 5 ml. by rapid distillation in a sample still. The 5-ml. residue is transferred to a special miniature still where the removal of the benzene is completed. The point of complete removal is deter-
38
ANALYTICAL CHEMISTRY
104 106 105 101 105
103
mined by the vapor temperature as measured by a thermocouple and potentiometer. The residue is cooled and weighed, and the calculated results are reported as parts per million by weight. The cooperators used this procedure exactly as published.
81 85
therefore, nonvolatile in characteristics. Oil 2 was a mixture of two thirds of oil 1 and one third kerosine. Oil 3 was a combination of one third oil 1, one third kerosine, and one third gasoline. PROGRAM
SAMPLES TESTED
Three standard samples were prepared for the cooperative study. They were similar to those used in the previous program (S), and were designated as oil 1, 2, and 3, respectively. They covered three volatility ranges. Oil 1 was an SAE 70 heavy motor oil and,
Each laboratory was asked to prepare synthetic refinery effluent samples using the three standard oils. The instructions were to make the samples a t three oil content concentration ranges-10, 50, and 125 p.p.m, The cooperators were cautioned to shake each sample very thoroughly to suspend as much oil
at Various Concentration Levels
Infrared Method 10 50 125 P.P:m. P.Pp p.p.m. 011 Oll oil
Laboratory Av.
011
85 105
81 84
239 115 121
130 98 105
118 103 106
104 71 88
95 94 96
82 84 89
118
102
104
79
95
86
111
103 65-142
102 82-118
89 98
92 100
Av.
100 100 130
94 123 160
96 100 120
Av.
115
142
110
Grand av. Torecovery yo Recovery range
102 92-116
105 89-127
103 92-114
8"
011
102 106
100 100
A
011
86 117 118
99
7
API Method 731-53 10 50 125 Pqm. P.P:m. P.P..rn.
Samples Made with Oil 2 99 96 96 182 118 295 142
93
6
Direct Extraction-Pycnometer Method 10 50 125 p.p.m. P.P+ P.P+ 011 oil 011
62-295
82 10-133
86 55-107
88 81-106
Samples Made with Oil 3 1
2
3 4
Av.
90 100
84 86
95 103
108 108
59 73
65 67
95
85
99
108 75 146
66 79
66 95 89
0
91
102 104
103 104
101 107
110 53 67
71 73
92 83 84
75 77
76 85
59 79 80
Av.
103 60 153
104 68 83
104 82 89
60 36 115
72 59 81
84 71 98
76 44 71
80 57 70
80 53 62
Av.
107
75
84 50 54
58
64
58
93
99
52 109 124
103 107
78 87
79 104
106 108
102 106
s2 93
Av.
107 120 100
104 110 112
88 104 112
Av.
110
111
108
Av.
8.5.
5 6 7
8
59 59
Av. Av.
Grand av. % recovery yo Recovery range
101 60-153
95 68-106
95 82-107
100
96 107
128 85 88
117 92 94
105 94 100
82 57 77
92 52 58
104
86
93
97
65
55
89 53-146
81
57-129
84
50-124
67 0-107
75 57-87
69 52-104
Omitted from recovery average and range.
as possible in the water prior to conducting the determination. It was not imperative that the oil content concentrations be exactly a t the levels just mentioned, b u t it was necessary that each laboratory report the exact amount of oil added in each case. Duplicate determinations were requested a t each concentration on the three samples using the three techniques. RESULTS
Seven of the laboratories reported results using the direct extraction-pycnometer procedure, six by the infrared
method, and five by method 731-53. Table I gives the complete results obtained. I n Table I1 is shown the per cent of oil recovered at the different concentration levels. Table I11 gives the average per cent of oil recovered by all the laboratories a t each concentration level. The results of one laboratory were discarded from the precision calculations because they were consisb ently higher than the other laboratories. Of the three methods tested the infrared method showed the best over-all recovery of oil and precision. The accuracy was determined in parts per million from the data in Table I.
An accuracy standard deviation was calculated as the root mean square of the differences between known and found values. The standard deviation was multiplied by 1.96 to indicate the 95% confidence limits of a single result. These limits are given in Table IV. Thus, at the 10-p.p.m. level on oil 1 a single test by the infrared method would be within 3 p.p.m. of the correct result 95% of the time. No figures are available for repeatability and reproducibility because the check samples were not identical. However, the accuracy data give an idea of the relative precision of the methods. VOL. 30, NO. 1 , JANUARY 1958
0
39
DISCUSSION
I n submitting their data the different laboratories included comments relating t o their experiences in the use of the three techniques. The various points are summarized below.
Table Ill.
Average Per Cent of Recovered
Oil
Direct Extract,ion. ... .
P.P.LI, Oil 1 10 50 125 Oil 2 10
50 125 Oil 3 10 50 125
Pycnom- d P I eter Method Infrared Method Method 731-53 105 96 93
85 98 99
74 99 98
102 105 103
111 103 102
82 86 88
101 95 95
89 81 84
67 75 69
INFRARED METHOD.Two of the laboratories expressed the opinion that the infrared method is superior to the others in all respects. One laboratory reported trouble in trying to extract 123 p.p.m. of oil from 2500 ml. of water with 50 ml. of carbon tetrachloride. They
Table IV.
Accuracy of Oil-in-Water Methods“
Direct Extraction.4PI PycnomMethod eter Method 731-53
Infrared P.P.J1. Method Oil 1 13 3 9 10 14 10 15 50 23 34 17 125 Oil 2 12 2 15 10 24 13 20 50 30 19 31 125 Oil 3 19 5 8 10 32 10 34 50 85 23 71 125 a Figures are 95% confidence limits of a single result.
increased the extraction time from 15 to 30 minutes for such samples and suggested using a smaller sample when the oil content is high. Another laboratory commented that this method may not be as accurate on actual samples of re40
ANALYTICAL CHEMISTRY
finery effluent waters as on the standard samples n-hich were tested. This statement was based on the fact that the absorptivity of the oil is knonn in the instance of the staridard samples, vihereas on plant samplcs an average absorptivity for the effluent oil must be taken. Four of the laboratories thought the infrared nicthod gave the best accuracy, particularly n ith samples prepared from oil 3 containing gasoline and with samples in the 10-p.p.m. range. DIRECT EXTR.4CTIOT-PYCXOMETER METHOD. One of the cooperators feels that this method is as good as the infrared if the oil content of the sample tested is in excess of 10 p.p.m. and the operator is skilled in the use of the piocedure. A number of laboratories reported trouble getting good results the first time they tried the method because it requires appreciable manipulative skill. One laboratory had difficulty in preparing a calibration curve. Several participants pointed out that the pycnometers must be iyeighed accurately, because 1 mg. in error in weighing means about 2 p.p.m. in oil content. API METHOD 731-53. T o eliminate bumping or spattering in the semimicrostill, the still temperature should be kept between 550” and 575’ F. I n one case the volatile matter in the reflux trap did not form a layer but remained dispersed as droplets. Some cooperators mentioned that this method requires bulky equipment that is inconvenient to handle and that the benzene blanks were high, about 50 mg. RAPIDITY. Based on information submitted by the group, a coniparison was made as to the relative rapidity of the three methods.
Method
Over-all Time, Minutes
Infrared method 25 Direct extraction-pycnometer 30 to 35 .4PI 731-53 dbout 240
Operator Time, Minutes 10 12 102
ADDITIOSAL COMMENTS.One laboratory gave results on regular refinery effluents and advised that the infrared method and method 731-53 gave satisfactory results on their plant effluent samples. Hon-ever, einulsions were formed n i t h the direct extractionpycnometer method. This is contrary to experience in the chairman’s laboratory, where the latter method has been used extensively and emulsions are seldom encountered on plant samples. CONCLUSIONS
The group believes that the following conclusions can be drawn as a result of this investigation: The infrared method
is superior to both the direct extractionpycnometer technique and method 73153 in respect to speed, recovery of oil, accuracy, manipulative ease, and in obtaining more accurate results iyhere volatile products are present and n here the oil concentration is very 1017 in the saniple of nater tested. The direct extraction-pycnometer method has an appreciable admntagp m e r method 73153 in analytical time. I n addition, this method involves the use of lower cost laboratory equipment than the other two procedures. Khile the samples tested were synthetic rather than true effluents] the above conclusions undoubtedly are valid for oily matter in actual refinery waste naters. If the procedures are applied to waste water containing significant amounts of materials] other than oil, which are soluble in the solvents used, such substances will be included in the determination as part of the total oily matter. These findings have been transmitted to the API Committee on Disposal of Refinery Wastes (4) 11-ith the suggestion that it publish the infrared method and direct extraction-pycnonipter method in the waste disposal manual. It is understood steps already have been taken to publish the infrared metllod. The use of infrared as an analytical tool has become widespread, but the apparatus continues to be relatively costly. Khile there has been some progress in making available 'lion-' ’ cost infrared instruments, it is hoped that the work reported here will serve as a further stimulus to manufacturers to produce equipment at a price range within the reach of the small laboratories. LITERATURE CITED
(1) ,inierican Petroleum Institute, New York, S . Y., “Manual on Disposal
of Refinery Wastes,” l e t ed., Vol. IT, 1953. (2) Hart, W. B., Proc. A m . PetrolezamInst. 111,36 (1956) in prese. (3) . . Headington, C. E., .kNAL. CHEW 2 5 , 1681 -( 1953). (4) Lake, G. R.. Union Oil Co. of California, Brea, Calif., April 11, 1956, n-ritten communication. ( 5 ) Levine, IT. S., Mapes, G. S., Roddy, 122. J., ANAL. CHEM. 2 5 , 1840 (1953). (6) Rather, J. B., Jr., Socony Mobil Oil Co., Inc., Technical Service Laboratory, Brooklyn, S . Y., May 21, 1956, nritten communication (7) Simard, R. G., Hasegawa, Ichiro, Bandaruk, William, Headington, C. E., h A L . CHEM. 23, 1384 (1951). (8) Ibid., 24, 909 (1952) (corr.).
RECEIVEDfor reviem- June 10, 1957. Accepted July 10, 1957. Division of Refining, 22nd Meeting, -2merican Petroleum Institute, Philadelphia, Pa., May 1957.
