New Technique for Waste Pickle
Liquor Neutralization RICHARD D. HOAK AND CHARLES J. SINDLINGER Mellon I n s t i t u t e , Pittsburgh, Pa. plants. The settling rate of the slurry made by this process is high; sedimentation is virtually complete in less than an hour, the time depending somewhat on the alkaline agent used. I n addition, the filtration rate is much higher than that of conventional slurries.
Where waste pickle liquor is neutralized with lime according to conventional practice, the low settling rate of the slurry does not justify provision of means for discharging the iron-free supernatant liquor. In addition, vacuum filtration or centrifugation of the slurry is generally impractical. This paper describes a new technique whereby a substantial reduction in sludge volume is effected, settling is complete in less than an hour, and the vacuum filtration rate is increased markedly. The new method has been applied to magnesia, high calcium lime, and dolomitic lime to provide flexibility of adaptation. The process provides a feasible solution for plants with inadequate lagoon space and offers over-all operating economy.
THE NEW PROCESS AND ITS OPERATION
Controlled oxidation of the ferrous hydrate to ferrosoferric oxide is the fundamental principle on which the process operates. The alkaline agents used in this investigation were magnesia, high calcium quicklime, and dolomitic quicklime; the last two agents are the ones commonly employed for pickle liquor neutralization. Wilson ( 3 ) recently described a similar process whereby ferrosoferric oxide is produccd using ammonia as the alkaline agent. Basically the process consists of feeding pickle liquor a t a predetermined constant rate to a bath in a reactor, where the temperature is maintained above 75 O C. and means is provided for efficient aeration. The experimental reactor, illustrated diagrammatically in Figure 1, consisted of a section of steel pipe bolted t o a steel base. Aeration was supplied by three shrouded impellers, mounted equidistant on a shaft, with a separate air supply for each. The impellers were driven at a peripheral speed of 700 feet per minute. A vertical baffle, extending the full depth of the reactor, furnished necessary turbulence. Waste pickle liquor was admitted from a reservoir through a tube discharging below the lowermost impeller. Details of the reactor design are shown in Figure 2. All the runs reported here were made by the cyclic continuous method because suitable equipment was not available for feeding the alkaline agents accurately at a sufficiently constant rate t o collect data on continuous operation. Supplemental runs, in which the alkaline agent was added manually, revealed that with proper feeding devices the process could be operated continuously without difficulty. The operating procedure was the same for each of the agents. For each run, a quantity of alkaline agent equivalent to, or somewhat in excess of, the stoichiometric requirement of 8 liters of pickle liquor, was slaked in water and mixed thoroughly with about 8 liters of slurry left in the reactor from a previous run. The mixture was heated externally t o the proper temperature, and the air feed set at the desired rate. The three needle valves were adjusted to split the air feed into three substantially equal streams. The pickle liquor feed was then started a t a predetermined rate. Pickle liquor was fed to the bath at a constant rate until 8 liters had been added. A sample of slurry was then filtered, and the filtrate was tested for soluble iron; if iron was present, the air feed was continued until the iron oxidized and was precipitated. Where the variables were properly adjusted, all the iron was precipitated at the end of the pickle liquor feed period. Half of the slurry in the reactor was then drained off, and t h a t remaining was used immediately to start another cycle. Although the characteristics of the sludge produced in succeeding cycles was affected to some extent by the quality of the slurry present at the start of a run, this effect was normally small, even when the variables were changed over a relatively broad range.
T
HE widespread demand for effective treatment of waste waters before they are discharged to streams, lakes, or tidewater has caused those who produce wastes t o re-examine their treatment methods t o determine where improvements in operation can be made. The steadily rising cost level also has necessitated inquiry into means for reducing operating charges, and this is especially the case where waste treatment involves disposal rather than by-product recovery. Where pickle liquor is treated with lime, the supernatant liquor t h a t separates is iron-free, and i t should be possible t o discharge it directly t o a stream or t o a municipal sewer. In practice, however, the volume of this liquor is usually so small, relative t o the total sludge volume, t h a t all the slurry is stored in lagoons with no provision for drawing off the clear liquor. Solar evaporation and percolation into the subsoil are depended on t o dissipate the liquor that separates. Lagoons frequently occupy valuable land, and, in many instances, inadequate lagoon space is available. Also, pickling plants located in metropolitan areas cannot resort t o lagoons for sludge disposal; in such cases, sludge handling is indeed costly. T h e sludge produced in conventional treatment plants usually cannot be dewatered successfully on vacuum filters, and i t is necessary to pump or truck the full volume of slurry t o a disposal site. A simple method for reducing the final sludge volume would have great value in solving the sludge-handling problem. PICKLE LIQUOR SLUDGE DIFFICULT TO DEWATER
Cooperative studies with manufacturers of various types of phase-separation equipment have been in progress for some time with the object of determining the suitability of existing devices for sludge concentration. This work has resulted in a moderate amount of data on performance characteristics, but i t has indicated that the capacity of commercial devices is generally too low for them t o be attractive economically, or the quality of the separated liquor is too poor t o permit discharge without supplemental treatment. This situation prompted a study of means for modifying the physical form of the precipitate t o increase the rate at which i t could be dewatered commercially. The investigation has resulted in the development of a process whereby a sludge can be produced t h a t settles to a much smaller final volume than can be attained in conventional neutralization
65
66
INDUSTRIAL A N D ENGINEERING CHEMISTRY
Vol. 41, No. 1
Pickle Liquor SUPDlY
RzyD::;: Figure 1. Schematic Arrangement of Apparatus
Figure 2. Details of Reactor Design
The successful operation of the process depends primarily on balancing the rate of iron precipitation with the rate of oxidation of ferrous hydrate. These rates must be so controlled that the iron in the precipitate will have a ferric t o ferrous ratio between 2 and 5, preferably betv-een 2.5 and 3 4 while the temperature of the bath is maintained high enough to promote the formation of ferrosoferric oxide. Obviously, there are many variables that should be considered in selecting the optimum operating conditions-for example: reactor design, including aeration and agitation; reactivity of alkaline agent,; composition of pickle liquor; reaction temperature; excess alkdine agent; and pickle liquor and air feed rates. DetaiIed study of each of these variables would be an undertaking of considerable magnitude. Fortunately, all are not of equal importance. Experiments have demonstrated that reactor design is not a critical factor except that efficient dispersion of air must be provided. Although shrouded turbines were used in this study, a turbomixer gave satisfactory results in a trial run, and there is no reason why any of the aerator-agitator devices on the market that provide vigorous dispersion of air could not be used. The lower temperature limit is governed by the conditions for formation of ferrosoferric oxide, choice of alkaline TABLE1. agent and composition of pickle liquor will be fixed by local conditions. I n a commercial installation, then, the only variables to be decided on would be excess agent and the rate of feed of pickle liquor and air. These could easily be est,ablished by a few trial runs. The data presented below shox the feasibility of the process and the improvement in sludge properties that may be expected from it. Magnesia was used to develop t,he basic principles of the process t o avoid interference from other coprecipitated compounds, such as calcium SUIfate. High calcium lime then r a s studied because of it's popularity as a neutralizing agent, and also because there as considerable doilbt that it could be adapted to the process in view.
of the amount of calcium sulfate that would be coprecipitated. Finally, dolomitic lime was used to represent a neutralizing agent n-hose reactivity was intermediate between that of the first two mat,erials investigated. Actually, the
dolomitic lime was found to be less reactive than the magnesia employed in the study, but the magnesia chosen was the chemical grade which is a highly reactive form of the compound. The pickle liquor used was made by dissolvillg steel mil: copperas and sulfuric a5id in water to yield a solution contaidng 60 grams of iron and 200 grams o f sulfate per liter; this position WTas chosen to represent the waste liquor from continuous strip picuers. For ea& a&aline agent, studied a series of F U ~ E mas made with the standard liquor diluted with an equal of water. Another series then w&s made u,ith the full-strength liquor. This procedure was adopted primarily to facilitate process development, but it served also to illustrate the conditions to be expected where the waste liquor is combined witt rinse water. MAGNESIA A S A NEUTRALIZING AGENT
Magnesia has not found wide application as a neutralizing agent for n-aste pickle liquor because of its high price. The
MAGNESIA.
Agent,
%
Run NO. 1
3 5 8 9 10
l1 12 13
Stoichiometrio
Resmt. 120 115 110 105 105 105 105 105
a Air rate, 9 . 2 5
Waste Pickle Liquor Rate,
AS
ALXALINE ,4UF,r;T-FX.4LF-SrR~,NGTH LIQUOR" Fp++i.
Temp.,
~ i . / m n ~ 0 C. 270 70 250 73 260 75 260 73 260 73 270 77 270 72 270 77 270 80
Slurry pH
I _
8.6
Fe" 1.57
8.0 7.8 7.9 8.5 7.8 6.8 7.3 6.9
2.60 2.41 3.04 4.02 3.61 3.54 4.23 8.36
Fe in Supernatant,
G./L.
0.00 0.03 0.00 0.00 0.00 0.00 0.04 0.03 0.03
Initial Settling,
In./,Min. 0.028 0.054 0.075
0.100 0.341 0.608 0.091 0.388 0.850
yo ,
Run No. 2 3 4 5 6
* Air
Stoichiomotric Reqmt. 112 112 104.5 104.5 104.5
19
'E"
18 8
I./min.
TABLE11. MAGNESIAAS ALKALINEk3ENT-FULIrSTRENOTK Agent,
Fina! Sludgr 1'01.. % of Original 45 26 26
Waste Pickle Liquor Rate, Ml./Min. 117 117 117 117 117
rate, 8 . 7 I./min.
__ Fe*' Fg++'
Tfmp., Slurry C. PH 80 7.8 80 80 80 70
8.2
7.6 7.4 7.4
2.71 2.40 2.67 2.65 2.73
Fe in
Supernatant, G./L. 0.00 0.00
0.06 0.04
0.03
Initial Settling. In./AIin. 0.03 0.04 0.27 0.58
0.13
LIQUOR" Final Bludge Val.,
70o! Originai . i
..
18 16 25
INDUSTRIAL AND ENGINEERING CHEMISTRY
January 1949
Run
67
reveals the adverse effect of operating at temperatures below 75"c. The effect of the ferric to ferrous ratio is not clearly shown in either of these tabulations but other data, especially those in the following tables, bring out that this ratio must be held between 2 and 5 for best results. The small final volume of sludge produced with magnesia makes it possible t o wash it free from sulfates by continuous decantation. For this reason filtration studies were not made on these slurries.
No.
.-I
x -3 0-9 A- 13
HIGH CALCIUM LIME IN NEUTRALIZATION
Quicklime, properly slaked, was used for all the runs with this agent but hydrated lime suspended in water could be substituted without changing the method of operation or the results obtained. Data from a number of runs with half-strength liquor are given in Table 111, and settling rates from some of them are plotted in Figure 4.
1 I
Figure 3.
1 2
I
1
I
4 6 SETTLING TIME, HRS.
3
I
Over Night
Magnesia and Half-Strength Liquor
present process may offer an opportunity for this compound in some areas, however, because it provides a way to recover relatively pure iron oxide and magnesium sulfate. The results from a number of selected runs with half-strength liquor are given in Table I to illustrate the progressive improvement in the sludge as variables were adjusted. Settling data for several of these runs are plotted in Figure 3. Magnesias vary widely in reactivity, as has been discussed in a prior paper ( 8 ) . The principal object in runs 1 to 8 was t o establish the optimum pickle liquor feed rate and the excess of agent required for complete precipitation of iron and a ferric to ferrous ratio in the desired range. Improvement in sludge settling rate was the criterion by which the results were judged. Settling rates were measured in 100-ml. graduated cylinders. The remaining runs in the tabulation were made t o stpdy the effect of reaction temperature, and it may be seen that this is an important variable. Increasing the temperature t o 77" c. in run 10 resulted in a striking improvement in settling rate over run 9. where the temDerature was 73" C. Reducing the temperature io 72" C. in run 11 destroyed
,
the rapid Of the but raising zhe temperature to 77" and 80" C. in the next two runs restored it. These runs, and other similar ones, showed that the reaction temperature must be maintained above 75" C., and preferably as high as 80" C., or above. The results obtained with full-strength liquor are shown in Table 11. Two runs were made with a greater excess of magnesia than had been shown to be desirable from the results Kith half-strength liquor; under this condition it was impossible to produce a rapid-settling oxide. Run No. 6 again
40tu i
1
0.
0.25 0.50 1.0 SETTLING TIME; HRS.
Figure 4.
Over Night
High Calcium Lime and HalfStrength Liquor
The high reactivity of high calcium lime suggested t h a t i t could be used in the process in stoichiometric quantity but, as the data reveal, it was necessary to provide a small excess to ensure complete precipitation of iron. A higher rate of pickle liquor addition could have been used in runs 8 and 9 with maintenance of an iron-free supernatant, but this would have reduced the
TABLE 111. HIGHCALCIUM LIME AS ALKALINEk 3 D " H A L F - s T R E N G T H LIQUOR= Agent,
%
Run No. 1 3 5 6 7 8 9
Stoichiometric Reqmt. 100 100 100 100 100
Waste Piokle Liquor Rate, Ml./Min.
270 288 288 285 285 285 lo5 285 105 310 11 lo 105 310 a Air rate, 9 . 2 5 l./min.
Temp., Slurry C. pH 76 72 73 80 80 73 79 74 80
,
..
5.2 5.1 5.5 5.5 9.1 9.0 9.1 9.0
Fe in Super-
Final Settling
Sludge Vol., % of
Fe++
G./L.
In./>Iin:
Original
4.66 2.04 3.09 2.54 2.24 2.09 2.53 1.84 1.55
1.01 2.10 0.76 0.70 1.45 0.00 0.00
0.042 0,217 0.280 0.480 0.698 0,248 0,580 0,387 0,222
E natant,
0.00
0.00
Initial
.. ..
35 34 35 35 35 35 35
I N D U S T R I A L A N D E N G I N E E R I N G CHEMISTRY
68 I
I
I
Run 4 shows the adverse effect of a low ferric to ferrous ratio, and 5 of a low temperature. Runs 6 and 7 demonstrate good operating practice. The final sludge volume is only about 65% larger than that from the half-strength liquor. This emphasizes the significant role played by the crystalline form of the calcium sulfate that coprecipitates Lvith the oxide. If these results are compared with those for magnesia, i t nil1 be seen that the calcium sulfate is responsible for a major portion of the bulk of the sludge. For comparison, a sample of pickle liquor was neutralized rvith lime slurry a t room temperature without aeration, to simulate conventional practice. A filter test identical to those given above showed that' this slurry filtered a t a rate of 0.055 gallon per square foot per minute When the slurry was diluted m-it,h 2.5 parts of water, the rate became 0.12 gallon per square foot per minute. This comparison demonstrates the marked superiority of the new procedure. TKOfactors are responjible for the high filtration rates of the slurries made by the new process. The iron is prccipit,at,cd as a finely divided, nongelatinous, hydrated ferrosoferric oxide, and the coprecipitated calcium sulfate is distributed through the slurry as needle-shaped crystals that act as a filter aid. These crystals are visible as an interlaced mat when a filter cake is broken open.
I
Run No. A -1 0 -4
-5
I
I
0.25
Q50
Figure 5.
Vol. 41, No.. 1
DOLOJlITlC LIME 1.0 2.0 SETTLING T I M E , H R S .
C
b
Doloitiitic Lime and Half-Strength Liquor
ferric to ferrous ratio and thereby would have decreased the settling rat,e of the slurry. The effect of the ferric t o ferrous ratio on the settling rate may be seen by comparing run 9 with runs 10 and 11. In the latter runs the ferric to ferrous rat,io was purposely reduced by increasing the pickle liquor rate; this resulted in a substantial decrease in settling rate. These results emphasize the necessity for holding the ferric to ferrous ratio above 2 for rapid settling slurries. The effect of temperature appears in a comparison of runs 8 and 9. The necessity for holding the reaction temperature above 75" C. is clearly illustrated in these runs: Many other runs showed that the slurry-settling rate could be altered a t will merely by changing the tcrnperature of precipitation. Regardless of the initia,l settling rate of the slurries produced with high calcium lime, the final sludge volume !vas always about 357, of the original volume. While this is a much smaller final volume than could be attained in conventional neutralization practice even Kith a long settling period, it represents a considerable volume to be stored in a lagoon. In many instances an inadequate lagoon area is arailable, but the sludge disposal problem could be solved if the slurry could be filtered a t a rapid rate. Filtration tests on the slurries from these runs gave the data in Table IV. These tests were made with a filter leaf of 0.1-squarc foot) filtering area under a vacuum of 54 em. of mercury. The slurries with the highest settling rates also had the highest filtration rates; this finding has been substantiated by numerous filtration tests. I n run 7, where no precoat n-as used, the filtrat,e 1%-as cloudy a t first but it cleared rapidly. In practice, then, a precoat would not be necessary where a slurry of good quality was produced. A sample of the cake from this test was dried a t 105" C.; it had a moisture content of 37y0 on the wet basis. Moisture was not determined for the ot,her tests, but the filter cakes appeared t o be similar to that from run 7. The results obtained with full-strength liquor are given in Tables V and VI. A 57, excess of lime was used for each run.
Dolomitic quicklime, after slaking to a slurry, was ubed in all of the runs with this agent, but the properties of the atmospheric hydrate are so similar that this material could be substituted with no change in the method of operation The results from a series of runs with half-strength liquor are given in Table VII, and settling data are plotted in Figure 5. These figures show clearly the lower reactivity of dolomitic as compared with high calcium lime. A greater excess was required, and the pickle liquor rate had to be reduced niaterially to get satisfactory results. The higher basicity of dolomitic lime, however, permits the use of a considerable excess v ithout
TABLE IV. FILTRATIOS TESTS OF HIGHCALCIGM LIVESL~JKEYHALF-STRICNGTII L t Q U O R R u n No. 7 8
Time for First. Liter Filtrate, hlin.
AV. Filtration Rate Gal./Sq. Fi./Min. 1.76"
Cakc Thickness, In.
9/16 0,53 1/2 1.32 7/16 2.0 0.53 7/16 6 .O 4.3 0.62 1/2 a No precoat: other tests used l/r-inoh preaoat of diatomaceous earth. 1.5 5.0
9 in 11
TABLET'.
HIGH CALCIUM LIXE .4s ALKALINE AGEXT-FULI~ QTREXGTH LIQUOR'"
Waste Pickle Liquor Kun Rate Temp., Blurry KO. > X ~ . / M ~ L pH C. 4 140 81 9.5 72 9.5 5 140 6 140 82 9.5 s2 9.5 7 140
Fe in SuperInitial natant, Settling, G./L. In./hIin.
!?%??!
Be++ 1.61 2.30 2.60 2.34
0.00
0.00
0.00 0.00
0.23 0.12 0.40
0.40
Final Sludge Vol., % of Original 00
62
5s 5s
* Air rate, 16.4 l./min.
TABLE VI. FILTRATIO?; TESTSOF HIGHCALCIUV LINESLURRYFULL-STREXOTH LIQUOR
Ruii No. 4
5 0
7
'1
Filtration Ratea Lb. wet cake/sq. ft.,/hr. 278 , 0.7;
Gal./sq. ft./min. 0.44 1.06
1.13
161 388 407
Precoat, 0.25 inch of diatomaceous earth.
Cake Moisture, '% 41.4 40.9 40.2 39.8
I N D U S T R I A L A N D E N G I N E E R I N G CHFMISTRY
January 1949
TABLEVII. DOLOMITICLIME Agent,
%
Run No. 1 2 3 4 5 6
Stoichiometric Reqmt. 110 110 110 110 110 110
Waste Pickle Liquor Rate MI./Mh 76 96 108 108 108 108
AS
Air Rate, L./Min. 3.32 0.98 0.98 1.45 1.87 1.87
ALKALINEAGENT-HALF-STRENGTH LIQUOR
Temp., Slurry O C. pH 88 5.2 84 7.7 88 7.9 84 8.0 82 6.9 83 7.5
Fe in Initial natant, Settling G./L. In./Mid. 0.48 0.171 0.04 0,738 0.00 0.736 0.01 0.775 0,326 0.03 0.03 0.326
~ e + + +Super-
___ Fe++ 6.78 4.31 2.64 3.45 8.53 10.05
thirds greater volume with high calcium lime and strong liquor. The improvement in slurry quality as the ferric t o ferrous ratio was lowered may be seen from the data.
Final Sludge Vol.,
% of
Original .. 22 22 20.5 23 23
AIR REQUIREMENTS
The amount of air required for oxidation is based on the reaction 3Fe0
TABLE VIII. FILTRATION TESTS OF DOLOMITIC LIMESLURRYHALF-STRENGTH LIQUOR
Time for First Liter Filtrate, Min.
Av. Filtration Ratea, Ga!./Sq. Run No. Ft./iMin. 3 0.56 4 0.54 5 0.40 6 0.41 a Precoat (0.25-inch) used for all tests.
Cake Thickness, In. 0.25 0.25 0.25 0.25
TABLEIX. DOLOMITIC LIME
AS ALKALINEAGENT-FUL~ STRENGTHLIQUOR"
Waste Fe in Pickle Liquor SuperInitial Run Rate Tzrn&., Slurry E + natant, Settling No. MI./M'in. pH F e + + GJL. In./Mid 82 6.8 8.62 0.06 0.15 2 75 82 7.1 3.87 0.04 0.45 3 75 82 6.9 4 75 2.53 0.02 0.60 a Air rate, 3.6 l./min. for run 2; 2.5 I./min. for runs 3 and 4.
Final Sludge
Val,
% of Original 35 31 30
69
+ '/zO~= FeO.FezOa
A calculation from this reaction shows t h a t the minimum theoretical oxygen for a ferric t o ferrous ratio of 2.25 is 0.0131 pounds per liter of pickle liquor converted, or 0.772 cubic foot of air a t 20' C. and 1 atmosphere. The actual volume of air fed to the reactor will exceed the minimum requirement] under all practical conditions, because oxygen cannot be absorbed completely a t any reasonable operating rate. I n general, where the oxidation rate is the controlling factor, as with high calcium lime, it will be advantageous to use a high rate of air feed at low oxygen efficiency. Similarly, where the rate of neutralization is controlling, as with dolomitic lime, it is desirable t o restrain the oxidation by using a low air feed rate with concommitant high oxygen efficiency. This point is illustrated in Table X I which tabulates average oxygen efficiencies for the various combinations of reagent, liquor strength, and air feed rate. DISCUSSION OF PROCESS
TABLEX. FILTRATION TESTSOF DOLOMITIC LIME SLURRYFULL-STRENGTH LIQUOR
a
Run No. Filtration Ratea, Gal./Sq. Ft./Min. 2 0.20 3 0.36 4 0.41 Precoat, 0.25 inch of diatomaoeous earth used.
The neutralization process presented here provides a marked improvement over conventional practice, and the three alkaline agents studied give the process enough flexibility to fit most local conditions. A decision on the best agent to use will depend
I
TABLE XI. Reagent xgo hlg0
CaO CaO CaO.MgO CaO.MgO
I
I
I
OXYGENUTILIZATION UNDER VARIOUS CONDITIONS Liquor Strength Weak Strong Weak Strong Weak Strong
Air Rate, L./Min. 9.25 6.70 9.25 16.40 1.45 2.50
Oxygen Efficiency, % 35 39 34 19 89 75
esceeding the cost of high calcium lime, as was pointed out in a previous publication (1). There is no doubt that the pickle liquor rate could have been increased by increasing the excess of alkaline agent. I n addition, dolomitic lime yields a slurry that settles somewhat faster and to a smaller final volume than high calcium lime. The data in Table VI follow much the same pattern as those in Table 111. The adverse effect of excessively high ferric to ferrous ratios is shown clearly in runs 5 and 6. Filtration data were collected by the same procedure as WM used for the high calcium lime slurries. The data are given in Table VIII. Here again the slurries with the highest settling rate had the highest filtration rate, but this material filtered at a much lower rate than the high calcium lime slurries. The lower rate was attributable to the smaller proportion of calcium sulfate in these slurries. Data on the full-strength liquor appear in Tables IX and X. A 10% excess of alkaline agent was used for all runs. Here the final sludge volume is about a third larger than that from half-strength liquor. This is in agreement with the two-
0 HIGH CALCIUM LIME DOLOMITIC LIME X MAGNESIA
I
0.125
Figure 6.
I I 0.25 0.375 SETTLING TIME, HRS.
I
0.50
