Waste Lubricating Oil Rerefining by Extraction-Flocculation. 3. A Pilot

U = the error square sum defined in eq 9. [] = concentration of species in the bracket, M. Greek Letters u = standard deviation defined in eq 10. Supe...
1 downloads 0 Views 1MB Size
2449

Ind. Eng. Chem. Res. 1991,30, 2449-2456

D = distribution ratio of zinc extracted with the D2EHPATOPO mixture Do= distribution ratio of zinc extracted with D2EHPA alone E = descending ratio of dimeric D2EHPA in the presence of TOPO HR = monomeric form of DSEHPA (HR)2 = dimeric form of D2EHPA Kzm= extraction constants defined in eq 2, M(2-m)/2 K 1 = constant defined in eq 13, M-' K 2 = constant defined in eq 14, M-'12 Kd = dimerization constant of D2EHPA defined in eq 12, M-' U = the error square sum defined in eq 9 [] = concentration of species in the bracket, M Greek Letters u =

standard deviation defined in eq 10

Superscripts

- -- species in the organic phase Subscripts i = initial log = logarithm of base 10 min = minimum value t = total Registry No. DZEHPA, 298-07-7; TOPO,78-50-2; Zn, 744066-6.

Literature Cited Ajawin, L. A,; Perez de Ortiz, E. S.; Sawistowski, H. Extraction of Zinc by Di(2-ethylhexy1)phosphoric Acid. Chem. Eng. Res. Des. 1983, 61 (l), 62-66. Bates, C. F., Jr. The Extraction of Metallic Species by Dialkylphosphoric Acids. J. Inorg. Nucl. Chem. 1962, 24, 707-720. Danesi, P. R.; Reichley-Yinger, L.; Cianetti, C.; Rickert, P. G. S e p aration of Cobalt and Nickel by Liquid-Liquid Extraction and Supported Liquid Membranes with Di(2,4,44rimethylpentyl)phosphinic Acid [Cyanex 2721. Soluent Extr. Ion Exch. 1984,2 (6), 781-814. Huang, T. C.; Juang, R. S. Extraction Equilibrium of Zinc from Sulfate Media with Bis(2-ethylhexy1)phosphoric Acid. Ind. Eng. Chem. Fundam. 1986,25 (4), 752-757. Huang, T. C.; Tsai, T. H. Extraction of Nickel(I1) from Sulfate Solutions by Bis(2-ethylhexy1)phosphoricAcid Dissolved in Keros-

ene. Ind. Eng. Chem. Res. 1989,28 (lo), 1557-1562. Ihle, H.; Michael, H.; Murrenhoff, A. A Spectrophotometric Investigation of Synergic Solvent System Uranyl IonlTri-n-octylphosphine Oxide/Di(2-ethylhexyl)phosphoric Acid. J. Inorg. Nucl. Chem. 1963,25,734-736. Johnston, B. E. Commercial Applications of Phosphorus-Based Solvent Extractions. Chem. Znd. 1988, 17 (Oct), 656-660. Kimura, K. Inorganic Extraction Studies on the System between Bis(2-ethy1hexyl)phosphoric Acid and Hydrochloric Acid (I). Bull. Chem. SOC. Jpn. 1960,33 (e), 1038-1047. Liem, D. H. High-speed Computers as a Supplement to Graphical Methods. 12. Application of LETAGROP to Data for LiquidLiquid Distribution Equilibria, Acta Chem. Scand. 1971,25 (51, 1521-1534. Morel, F. M. M. Principles of Aquatic Chemistry; Wiley-Interscience: New York, 1983; Chapter 6, pp 242-249. Nash, K. L.; Choppin, G. R. The Thermodynamics of Synergistic Solvent Extraction of Zinc(I1). J. Inorg. Nucl. Chem. 1977, 39, 131-135. Paatero, E.; Lantto, T.; Ernola, P. The Effect of Trioctylphosphine Oxide on Phase and Extraction Equilibria in Systems Containing Bis(2,4,4-trimethylpentyl)phosphinic Acid. Soluent Extr. Ion E x c ~1990,8 . (3), 371-388. Peppard, D. F.; Ferraeo, J. R.; Mason, G. W. Hydrogen Bonding in OrganophosphoricAcids. J. Inorg. Nucl. Chem. 1958,7,231-244. Ritcey, G. M.; Ashbrook, A. W. Soluent Extraction, Part I; Elsevier: Amsterdam, 1984; Chapter 4, pp 172-206. Rublev, V. V. Synergistic Systems Involving Alkylphosphoric Acids. Zh. Anal. Khim. 1983,38 (5), 922-930. Sabot, J. L.; Bauer, D. Solvent Extraction of Cobalt(I1) Di(2-ethylhexyl) Dithiophosphate Adducts with Oxygen Donor Ligands. J. Inorg. Nucl. Chem. 1979,41, 767-769. Sastre, A. M.; Muhammed, M. The Extraction of Zinc from Sulfate and Perchlorate Solutions by Di(2-ethylhexyl) Phosphoric Acid Dissolved in Isopar-H. Hydrometallurgy 1984, 12, 177-193. Sato, T. The Extraction of Thorium from Sulfuric Acid Solutions by Di(2-ethy1hexyl)phosphoricacid. J. Inorg. Nucl. Chem. 1965,27, 1395-1403. Sato, T.; Kawamura, M.; Nakamura, T.; Ueda, M. The Extraction of Divalent Manganese, Iron, Cobalt, Nickel, Copper and Zinc from Hydrochloric Acid Solutions by Di(2-ethylhexy1)phosphoric Acid. J. Appl. Chem. Biotechnol. 1978,28, 85-94. Zong, C. L.; Furst, W.; Renon, H. Extraction of Zinc from Chloride and Perchlorate Aqueow Solutions by Di(2-ethylhexyl)phoephoric Acid in Escaid 100. Hydrometallurgy 1986, 16, 231-241. Received for reuiew May 1, 1991 Accepted July 11, 1991

Waste Lubricating Oil Rerefining by Extraction-Flocculation. 3. A Pilot Plant Study M.Alves dos Reis Faculdade de Engenharia d a Uniuersidade do Porto, Rua dos Bragas, 4099 Porto Codex, Portugal

Waste lubricating gils may be rerefined by treatment with an organic solvent that dissolves base oil and flocculates the major part of additives and particulate matter. This operation is intended to substitute the classical reaction with sulfuric acid, which generates an acid sludge, creating difficult disposal problems. The solvent developed by the author, a solution of potassium hydroxide in 2-propanol and a hydrocarbon, segregates an organic sludge from waste oils. This sludge may be used as a component of asphalts or, better, as a component of offset inks,consequently increasing the value of waste oils. This paper describes how the fundamental extraction-flocculation operation is integrated in a rerefining plant. Pilot plant results described here have shown that the proposed technology produces a recycled oil with properties quite similar to virgin oils. Introduction Waste lubricating oils have been rerefined for many years by the classical acid/clay process. In this process, the oil is flash-distilled at atmospheric pressure and at a temperature of about 150 O C to remove light hydrocarbons and water and is then mixed with 5-10% by volume sulfuric acid. After 24-72 h of reaction time, a clarified oil

is decanted and an acid sludge containing additives and other impurities is removed from the bottom of the reactor. The oil is finished by adsorption with activated clays and, in some plants, fractioned into several base oil stocks by vacuum distillation. The acid sludge produced in the acidlclay process is the main problem of this simple technology. All other steps

OSSS-58S5/91/2630-2~49$02.50/0 0 1991 American Chemical Society

2450 Ind. Eng. Chem. Res., Vol. 30, No. 11, 1991

involved in the rerefining plant pose no important environmental problems. In fact, the first step (flash distillation) produces light hydrocarbons that may be used as a fuel which is almost completely free of heavy metals. Spent clays generated in the finishing step may be introduced in Portland cement kilns (Berry et al., 1976). The Portuguese cement industry consumes more than 3 million tons/year of clay and could easily incorporate the 1500 tons/year of spent clays generated in a 30 OOO tonslyear waste oil rerefinery, the capacity required to process all the waste motor oil that can be collected in Portugal (Reis, 1982). It is not expected that this low ratio of oily clay to ordinary clays processed by this industry may be detrimental to clinker properties. Vacuum distillation, if existent, produces lube oil stocks and materials that may be used as fuels. Acid sludge disposal problems are the main incentive for the development of processes based on organic solvents. Organic sludges obtained with polar solvents may be incorporated in asphalts or, better, used as a component of offset inks (Reis and Jeronimo, 1982). In Portugal, where the majority of offset ink components, namely, carbon black and polymers, are imported and very costly, this raw material could compete for a market slice, specifically in the formulation of newspaper ink. The author has demonstrated the possibility of manufacturing an offset ink on the basis of a solvent and waste oil sludge as the unique source of carbon black and polymeric materials (Reis, 1982). This observation opens up the possibility of incorporating variable percentages of this sludge in usual formulations. The dry sludge production of a 30000 tonslyear capacity waste oil rerefinery is 1000 tonslyear, approximately 25% of the Portuguese production of black offset ink. This sludge may be processed (Reis and Jeronimo, 1982) at a cost which is less than 20% of its potential value as an offset ink raw material, estimated to be 1.5 million dollars. However, it is possible that only a fraction of this sludge could find use as an ink component, the rest being incorporated in asphalts. The global profitability of the rerefining operation may be considerably increased if the final ink blends are made at the rerefinery, thus producing finished products which could eventually be easier to sell than the raw sludge. Some polar solvents have been proposed, namely, 1butanol by Brownawell and Renard (1972), butanone by Jordan and McDonald (1973), a solution of l-butanol, 2-propanol,and butanone by Whisman et al. (1978a,b),and a solution of n-hexane and 2-propanol containing 3 g/L of potassium hydroxide by Reis (1982). Reis and Jeronimo (1988,1990) have shown experimentally and theoretically why l-butanol is the best single extraction-flocculation solvent and butanone is, comparatively, poor. Other previous publications (Reis,1982; Reis and Jeronimo, 1979) also explain why the above-cited Whisman and co-workers solvent has good extraction-flocculation capability and explains the method to design other composite solvents, including the above-cited solutions of n-hexane and 2propanol containing 3 g/L KOH. Whisman and co-workers (1974) have initiated the first systematic study of treatment of waste oils by organic solvents. These studies have culminated in the design of a solvent composed of (parts by volume) 2 parts l-butanol, 1part 2-propanol, and 1 part butanone. This process has been tested on a pilot plant (Corlew and Sluski, 1976) and has been the object of two US.patents (Whisman et al., 1978a,b). Detailed economic studies by Bigda et al. (1977a) have shown that a plant capable of processing 10 million gallons of waste oil a year was economically feasible with

a return on investment before taxes approaching 45%. At that time waste oil purchased at 15 cents/gal could be converted into base oil, in such a plant, for 39 cents/gal. This total processing cost was revealed to be very sensitive to waste oil price. A comparison of this process with the classical acid/clay technology (Bigda et al., 1977b),based on the above-mentioned used oil processing capacity, without vacuum distillation, has shown a processing cost of 45.8 cents/gal for the acidlclay process and 37.9 cents/gal for the process proposed by Whisman and coworkers. These results have encouraged the author's attempt to develop a proprietary solvent, having in mind that the economic assumptions of Bigda et al. could, possibly, be transported to Portuguese reality withdut an inversion of their conclusions. A few years later, the Portuguese company CREFOL obtained the license to use one of the solvents developed by Reis and Jeronimo (1979), based on hexane and 2-propanol containing 3 g/L potassium hydroxide, and performed a detailed economic study (CREFOL, 19841, which essentially shows agreement with the conclusions of Bigda et al. Should extraction-flocculation sludge be processed as an offset ink raw material then the economic advantage of solvent processes would be significantly reinforced. This paper describes results obtained from a pilot plant similar to the one described by Corlew and Sluski (1976) insofar as the basic steps are concerned. However, we have designed a pilot plant operating continuously. To achieve this objective, extraction-flocculation has been carried out in a conical tank and adsorption with activated clay has been realized by percolation through a fiied bed of bauxite. Some glass pyrex tubes provide visual monitoring of all fluids circulating through the pilot plant. This is an important detail, since, for instance, oil color is closely related to some aspects of oil composition, namely, the content of oxidized compounds.

Pilot Plant Description The pilot plant capacity is 250 kg/day. This represents, approximately, l/mth the capacity required to process all waste motor oil that can be collected in Portugal, calculated as 50% of motor oil sales in this country (1990 figures). Figure 1 is the pilot plant flow sheet, and Figure 2 is a frontal view of the plant. The plant is divided into five sections: (I) dehydration and light hydrocarbons removal; (11)extraction-flocculation; (111) solvent recovery; (IV) vacuum distillation; (V) percolation through bauxite bed. (I) Dehydration and Light Hydrocarbons Removal. In this section waste oil is fed by gravity from tank TK and heated in a steam heater (SH-1) to about 180 "C.The steam pressure is near 13 kg/cm2 ( t r 192 "C) and is supplied by the refinery boiler of PETROGAL (Oporto, Portugal), where the pilot plant has been built (PETROGAL is the Portuguese state oil company). When the steam pressure is below this value or whenever desired, an electric resistance heater (EH-1) provided with a thermistor sensor connected to a temperature controller makes it easy to set the oil temperature to a fixed value. This resistance has low power dissipation per unit of length (5 W/cm) to avoid oil thermal degradation. The heated oil is fed to a flash distillation column (DC-1) provided at the bottom with an electric resistance (EH-2) which proves to be very useful during the plant start up and during normal operation, especially when the steam pressure is low. The vapor leaves the column at about 170 "C and is condensed at condenser C1. The condensate is fed to florentine decanter FLO,where water

Ind. Eng. Chem. Res., Vol. 30, No. 11, 1991 2451 Sdvat

r

I

Free Water

-jj--J

IT

f

SH-3

R

Drain c i

xite

P1

PC

r4-l

r

Rerefined oil

n I

Distilled oil

I l l

- 1 1

I

P6

Figure 1. Pilot plant flow sheet. Legend (by order of first appearance in processing sequence): TK, waste oil tank;FL, flowmeter; SH-n, steam heater n;S, steam; T, temperature sensor; EH-n, electric heater n; DC-n,distillation column n;W, water; Cn,condenser or cooler n; FLO, florentine decanter; LH, light hydrocarbons; Pn, pump n; M, mixer; TEF, extraction-flocculation tank; ST,sludge tank; F, filter; EV, electrovalve; A, air; V, vacuum; VM, vacuometer; MV, manovacuometer; VT-n, vacuum tank n; PC, percolation column.

Figure 2. Frontal view of the pilot plant.

and light hydrocarbons are continuously separated. The dehydrated oil is cooled in cooler C2 and stored in a 200-L drum. Although the pilot plant has been designed to operate continuously, storage of all intermediate products in 200-Ldrums makes the plant operation very flexible, enabling the independent functioning of a given section for nearly 1 day.

(11) Extraction-Flocculation. In this section the oil is treated with a solvent composed of toluene and 2propanol containing 3 g/L potassium hydroxide. Toluene was selected instead of n-hexane for safety reasons. However, as explained in our previous papers (Reis, 1982; Reis and Jeronimo, 1979,1988,1990), n-hexane is a more economical choice for industrial use. Oil and solvent

2452 Ind. Eng. Chem. Res., Vol. 30, No. 11, 1991 2 -Propanol

Oil

Toluene

Figure 3. Curves of constant sludge removal for the system waste oil/toluene/2-propanol with 3 g/L KOH, 20 'C. The curves plotted are geometric sites of compositions giving sludge removal of 0, 1, 2, 3, 4, 5, and 6 g/(lOO g of waste oil). The phase envelope has been obtained for three temperatures. Compositions in weight fractions.

proportions are those represented by point P in Figure 3, where the phase envelopes for three temperatures and the lines of constant sludge removal were plotted. The method of obtaining Figure 3 has been explained before (Reis and Jeronimo, 1990). The dehydrated oil and the extraction-flocculation solvent are pumped through independent lines to mixer M. In this mixer, provided with agitation, the oil and solvent are completely mixed and flocculation begins. However, large aggregates are not desired and strong agitation prevents the growing of particles above a certain limit. In fact, this mixer has the main function of dissolving the base oil in the solvent but cannot correctly separate a clear solution from the particulate matter. This is done at the extraction-flocculation tank TEF, a conical upflow sludge blanket clarifier. Dispersion obtained at mixer M is introduced at the bottom of TEF. Small particles go along with the liquid moving upward but they generally collide with other particles going upward or downward. Large aggregates are formed, falling down and accumulating at sludge tank ST,where some aging takes place before sludge pumping is initiated. When the flow rate of dispersion fed to TEF is 30 L/h, the corresponding residence time is 2 h. Higher flow rates are possible, up to 120 L/h, with acceptable sludge separation. Due to progressive enlargement of the TEF cross section, only very small particles reach the top of the clarifier, where the clear solution is removed. The small quantity of particles in this solution is removed a t filter F. (111) Solvent Recovery. Reis (1982) has shown by calculation and verified experimentally that the solvent is completely removed by distillation in a system of two flash columns, the first one operating a t tl = 150 "C and P1= 1 atm, the second one a t t2 = 150 OC and P2 = 5 mmHg. An alternative option among many others is tl = 200 "C and P1 = 1 atm and t 2 = 150 "C and P2 = 15 mmHg, respectively. To avoid oil cracking and oxidation all atmospheric operations must be done at the lowest temperatures possible, which is the primary advantage of the first set of operating conditions mentioned above. The oil is heated at steam heater SH-2and if necessary at electric heater EH-3. The temperature of atmospheric column DC-2 is controlled by means of an internal electric resistance EH-4 commanded by a temperature controller receiving the information of a thermistor. The major part of the solvent is removed at this column and condensed

at cooler C3. A solenoid valve controlled by a level controller feeds the vacuum column DC-3 where the remaining solvent is almost completely separated, condensed at condenser C4, and collected at vacuum tank VT-1. Pilot plant vacuum tanks VT-1 to VT-4 are composed of a small (l-L) and a large (12-L) tank, separated by an intermediate valve. The large tank is provided with a glass level monitor. When this tank is full, the separation valve is closed and ita content discharged to a 200-L drum by means of air injection at a pressure slightly above 1atm. Meanwhile condensate is collected at the small tank. When this transfer operation is complete, a vacuum pump is connected to the 12-L tank, and when ita pressure equals the vacuum column operating pressure, the separation valve is opened and a new cycle begins. An internal electric resistance EH-5 simplifies temperature control of the vacuum column. The dark brown oil obtained after extraction-flocculation and solvent recovery will be called pretreated oil. (IV)Vacuum Distillation. Pretreated oil is heated in electric heaters EH-6 and EH-7 to about 320 OC and fed to a flash vacuum distillation column DC-4, operating at an absolute pressure of 5 mmHg. Each electric resistance is 5 m long and has the form of a U letter. Dissipation is 5 Watt/cm at maximum charge. Oil circulates in the annular space between the resistance and a 1 in. internal diameter stainless steel tube. The residence time of oil in each heater is 8 min. The outlet oil temperature is automatically controlled. Inside the distillation column an electric heater EH-8 permits the control of temperature at the column bottom. In our pilot plant, vacuum distillation is not used to fractionate the oil into several lube stocks but as one of the oil finishing steps. The existence of potassium hydroxide particles in pretreated oil contributes to oil purification. In fact, it reacts with acids which are responsible for the dark brown color, producing heavy nondistillable products, removed at the column bottom (Reis, 1982). The distillate is condensed at condenser C5 and stored in a 200-L drum before the finishing adsorption with activated clay. (V)Percolation through a Bauxite Bed. This operation is intended to improve the oil color and odor. Distillate oil is heated to 80 "C in steam heater SH-3. The percolation column PC is surrounded by a steam shell. Steam passes through SH-3 and the steam shell. Since steam available a t the pilot plant is medium-pressure steam, the condensate is discharged to drain. The small flow rate of steam necessary to achieve a 75-80 "C oil temperature is manually controlled, simply by operating a valve. As referred to before, steam is supplied by the PETROGAL refinery power plant. Ita very constant conditions assure a stable operation of the percolating column without any need of automatic temperature control. The flow rate used in this column is 0.5 L of oil per hour and per liter of bauxite bed. Results The extraction-flocculation properties of waste oil treated by the proposed solvent have been plotted at Figure 3. Miscibility of oil and solvent occurs outside the phase envelope. Curves of constant sludge removal are plotted in the miscibility area. For composition corresponding to point P, the settling curve obtained by the method previously explained (Reis and Jeronimo, 1988) is plotted at Figure 4. This curve suggests a sludge residence time of near 30 min for proper sludge aging. This curve also shows that the conical extraction-flocculation

Ind. Eng. Chem. Res., Vol. 30,No. 11, 1991 2453

0

0

25

50

Time, min

-

Figure 4. Settling curve for composition corresponding to point P in Figure 3 waste oil = 0.25; toluene = 0.20; 2-propanol(with 3 g/L KOH) 0.25. The temperature was 20 "C.

tank may be safely designed by imposing a residence time near 30 min. For this residence time, it was observed that the solids amount found by centrifugation in the overflow solution of base oil and extraction-flocculation solvent never exceeded 0.005 % . Table I summarizes some of the properties of the waste oil and the intermediate and final rerefiied oils. The last column shows the properties of a virgin oil obtained by mixing SAE-30 and SAE-10 base oils in order to obtain a viscosimetric specification similar to rerefined oil. Rerefined and virgin oil in Table I also have been compared in a series of tests. (i) In the corrosion test (copper sheet, ASTM D-130), the result was l b for both oils. (ii) The ANTAR cokefication test (Institute Francais du Petrole, 1971) simulates the formation of coke in a motor piston. Test temperature and test time were, respectively, 290 O C and 6 h. The result was somewhat favorable to virgin oil: average deposited mass, respectively, 60 and 50 cg for rerefined and virgin oils. The appearance of both deposits was similar. Elution chro-

Table I. Properties of Warte Oil and Intermediate and Finished Rerefined Oils Compared to Virgin Oil of Similar Viscosimetric Specification0 waste metreated distilled finished Virgin oil oil oil oil 03 color1 black 8+ 5 3 3 specific gravityz 0.904 0.894 0.883 0.002 0.w viscositi(37.8 "c), 80.10 52.89 52.77 57.90 100.48 cst3 9.59 7.51 viscosity (98.9 C), 7.19 7.11 12.27 csta 107 viscosity index' 124 100 104 101 pour point5 -27 -15 -12 -15 -30 140 flash points 214 228 229 166 carbon residue' 0.82 0.10 0.096 0.081 2.70 0.030 0 0 0.70 sulfated ash8 1.46 0.35 TAM 0.040 0.040 0.035 1.92 TBN'" 3.00 0.23 0 3.94 0 metals, ppm" 123 Ca 0 0 1619 0 Ba 0 202 1033 201 zn 0 0 1630 410 121 Pb 0 0 elution chromatography'2 5.0 3.3 2.3 1.3 % polar 61.3 62.8 k saturated 69.0 62.0 5% aromatic 26.0 35.4 35.7 35.9

'ASTM methods (indicated -y superscript nun-$: 1, D-1500, 2, D-1298; 3,4, D-446; 5, D-97; 6, D-92; 7, D-524; 8, D-874; 9, D-664; 12, D-2007. Other methods (indicated by superscript numbers): 10, IP method 276; 11, atomic absorption spectroscopy. Table 11. Results of

ANTAR Cokefication Test rerefiied oil before after test test

virgin oil before after test test

~~

viscosity, cSt at 98.9 "C 7.11 7.73 7.51 7.49 at 37.8 "C 52.77 64.35 57.90 60.33 viscosity index 101 91 100 93 carbon residue 0.096 0.36 0.084 0.33 chromatographic analysis % polar 2.3 11.5 1.3 10.8 % saturated 62.0 61.2 62.8 63.3 % aromatic 35.7 27.3 35.9 25.9 mass deposited in piston, cg 60 50 similar for rerefined and virgin oils appearance of deposit

matography (ASTM D-2007) has been performed on both oils, after this test. Results obtained and changes in viscosity and carbon residue are presented in Table 11. (iii) In the IP-48oxidation tests (Institute of Petroleum, 1976),oils have been submitted to 15 L/h air injection at

MICRONS 4.0

3.0

,I V

w V

z

a

m e

0 v)

m

a

5.0

7.0

6.0

8.0

9.0

I

oll

I

'7 I

R - Rerefined Cell prth:O.l mm NaCL

1.0

: i 1200

2500

16

0.2

5

3000

11

12

01

0.1

I"

3500

10

2000

le00

1%

1400

WAVENUMBER ( CM-'1 Figure 6. Infrared spectra of virgin and rerefined oils in Table I.

1000

800

60(

2454 Ind. Eng. Chem. Res., Vol. 30, No. 11, 1991 Table 111. Results of IP-48 Oxidation Test rerefined oil before after test test 52.77 99.07 viscosity at 37.8 "C,cST 1.88 viscosity ratio (after/before) 0.096 1.77 carbon residue 1.67 increase carbon residue chromatographic analysis 2.3 21.4 90 polar 62.0 61.8 % saturated 35.7 16.8 % aromatic

virdn oil before after test test 57.90 60.86 1.05 0.081 0.33 0.24 1.3 62.8 35.9

10.6 67.8 21.6

200 O C for 12 h. Results are compared at Table 111. (iv) Infrared spectra of rerefined and virgin oils may be seen in Figure 5.

Discussion Results in Table I suggest that rerefined and virgin oils have similar physical and chemical properties. Due to the complexity of base oil chemical composition, no rigorous statement can be made about possible important differences in the chemical composition of these oils without time-consuming and expensive analytic determinations. Chromatographic determinations of polar, saturated, and aromatic components show that passage through motor engines did not, essentially, change the base oil chemical composition. Infrared spectra in Figure 5 supports this conclusion. The polar compound content is proportional to the carbonyl peak at 1700 cm-I and the aromatic content is proportional to the aromatic ring peak at 1500 cm-'. Figure 5 shows that the aromatic contents of virgin and rerefined oils are very similar. Rerefined oil polar content is slightly higher, in agreement with chromatographic determinations shown in Table I. Pretreated oil additive content, as judged by calcium, barium, and zinc analyses, has been drastically reduced by the extraction-flocculation operation. As shown previously (Reis, 1982; Reis and Jeronimo, 1988), extractionflocculation solvents proposed by us (Reis and Jeronimo, 1979, 1982, 1990) and by Whisman and co-workers (1978a-c), as well as single solvents like butanone and l-butanol, are ineffective at eliminating polar macromolecules, such as polymethacrylates, but very effective at flocculating polyolefina and other nonpolar or slightly polar macromolecules. So, pretreated oil may still contain some macromolecules, especially of a polar nature. Direct visual observation and optical microscopy show that pretreated oil contains no black carbonaceous particles. As previous bench-scale experiments have revealed, these black particles coflocculate with nonpolar or slightly polar macromolecules (Reis and Jeronimo, 1988). The vacuum distillation included in the pilot plant is intended to remove additives containing metals and macromolecular additives that have resisted the extraction-flocculation operation. Additionally, the reaction of oxidized compounds with potassium hydroxide particles produces heavy nondistillable salts, thus reducing clay consumption in the finishing operation (Reis, 1982). Comparing columns 2 and 3 in Table I reveals that such a purpose has been achieved: the viscosity and viscosity index have been lowered to values which are typical of nonformulated oils, evidence that polymethacrylates have been removed, metals from additives (Ca, Ba, Zn) and from gasoline (Pb) have been completely eliminated, and the percentage of polar compounds, which is proportional to the percentage of oxidized compounds, has been changed from 5.0 to 3.3. The infrared spectrum of the finished oil in Figure 5 also shows no anomalous deviation from the

base oil "finger print", evidence that remaining additives are retained by the clay. Direct waste oil distillation cannot be performed in conventional distillation equipment due to the fast incrustation of heaters and the distillation column itself, caused by high macromolecular additive content. Shortpath distillation columns (rotary film columns operating at high vacuum) have been proposed for direct waste oil distillation, and a pilot plant built in Switzerland has demonstrated the process (Fauser and Ritz, 1977). In this process, metallic sodium is added to waste oil prior to distillation. The Dutch engineering company Kinetics Technology International (KTI) has developed a similar distillation process, without sodium addition (Goossens et al,, 1976). A pilot plant unit capable of processing 150 kg of waste oil/ h has been built at the Haberland rerefinery, Dollbergen, West Germany. This equipment, designed to directly treat waste oil, would also be ideal to process extraction-flocculation pretreated oil. However, given the high price of short-path distillation equipment, the economic feasibility of such a processing sequence may be questioned. The electric heating equipment in the vacuum distillation section unit caused some oil degradation. The ANTAR cokefication test and especially the IP-48 oxidation test have shown that rerefined oil obtained in our pilot plant has poor oxidation resistance compared to virgin oil. This behavior was not expected since rerefined oil seems to have a similar chemical composition. Vacuum distillation performed in pilot plant conditions has been the prime suspect of observed rerefined oil poor oxidation resistance as compared to that in virgin oil. In fact, electric heaters may be responsible for oil degradation and could be the cause of this problem. To demonstrate this hypothesis, pretreated oil has been directly passed through the bauxite bed, without previous vacuum distillation. Saturation of clay particles is fast. However, the first fraction of percolated oil obtained before the ASTM D1500 color exceeds 3, submitted to the IP-48oxidation test, has shown a performance very similar to virgin oil (Reis, 1982). Bench-scale rerefined oil obtained by the processing sequence proposed in this paper, including vacuum distillation of pretreated oil, at l-mmHg absolute pressure, in glass equipment used in the ASTM method D-1160, has shown an oxidation stability similar to that of virgin oils (Reis, 1982). This equivalence has also been reported by Whisman et al. (1978~).Such data suggest that, in an industrial plant, where appropriate heating equipment, based on heat exchange from high-temperature heating fluids, is used, oil degradation in the vacuum distillation section is, possibly, unimportant. The use of high vacuum, with operating pressures near 1mmHg, might additionally contribute to improving oil quality, including its oxidation stability. An alternative option is the elimination of vacuum distillation from the rerefinery operation. This option may reduce the processing cost, because suppressing the investment and operational distillation costs may be more important than the subsequent increase of the clay contacting cost. The fact of producing a single oil cut (approximately a SAE-20 stock) does not significantly inconvenience the formulation of additivated oils, provided that, for example, SAE-10, SAE-30, and Bright-Stock virgin oils are available to adjust viscosity properties to meet specifications. Reis (1982) has shown that the majority of motor oils produced by PETROGAL could be formulated by using the same additive package and large percentages of rerefined SAE-20 base oil, with SAE-10,

Ind. Eng. Chem. Res., Vol. 30,No. 11, 1991 2455 SAE-30, and Bright-Stock virgin oils. However, the suppression of vacuum distillation, by creating a dependence of external base oil supplies in the manufacture of formulated oils, reduces the rerefinery blending flexibility and forces the use of as much as 3 times more clay to finish the pretreated oil than that required to finish the lube oil cuts obtained by vacuum fractionation. Although distillation reduces product yields, because the vacuum residuum of pretreated oil is, generally, impossible to process into base oil, losses of oil in spent clay may be more important. The disposal of these increased spent clay volumes may also cause problems. However, we propose that the Portuguese cement industry could easily consume this byproduct. Having in mind these arguments, the ultimate decision to include distillation is dependent on a set of variables, including, as two of the most important to this particular economic balance, the rerefinery dimension and the availability of the virgin oils, which may, eventually, be required for formulations. The finishing step of adsorption with activated clays, for convenience carried out in our pilot plant in a percolating column, in order to obtain a continuous operation with a demonstrative visual effect, should be industrially performed in a agitated reactor. Percolation, used in the oil industry to finish transformer oils, is not economically feasible to finish pretreated oil obtained by the extraction-flocculation technology. In the same manner, oil pretreated by sulfuric acid in the classical acid/clay process, cannot be economically finished by percolation. In fact, bauxite regeneration, requires an expensive Herreshoff type rotary furnace, a fact that favors the agitated reactor option. Conclusions Waste motor oil may be rerefined by a sequence of operations, including dehydration and light hydrocarbons removal, extraction-flocculation with solvents composed of a hydrocarbon and 2-propanol containing 1-3 g/L potassium hydroxide; solvent recovery, vacuum distillation (facultative), and finishing with activated clay. Extraction-flocculation is the proposed substitute operation to replace classical sulfuric acid treatment. Its ecological advantages have been demonstrated the sludge produced, instead of causing serious disposal problems, may be incorporated in asphalts or used in the manufacture of offset inks. Pilot plant results have suggested that the rerefined oil has physical and chemical properties similar to those of virgin oils and, consequently, is a valuable resource in the blending of formulated oils. Improper heating to temperatures near 320 "C, in electric heaters preceding our pilot plant vacuum distillation column, have decreased base oil oxidation resistance. However, this behavior was not found in rerefined oil obtained by clay treating the oil produced in bench-scale high vacuum distillation of solvent-treated oil, nor is such effect expected in a properly designed industrial distillation unit. Additionally, it was shown that solvent-treated oil may be directly clay finished, producing an oil with almost the same physical and chemical properties as a virgin oil with similar viscosimetric specification, including the same behavior in oxidation stability tests. The single-cut base oil so produced (approximately a SAE-20 stock) may be incorporated in most commercially formulated oils. There is strong evidence that during vacuum distillation a reaction occurs between oxidized compounds and potassium hydroxide particles, which is favorable to the improvement of the oil color. This phenomenon is similar

to the action of metallic sodium in short-path vacuum distillation of waste oil (Fauser and Ritz, 1977). The final word in rerefined oil quality is the result of expensive engine sequence tests. It was not possible to undertake these tests during the course of this investigation. Nevertheless, available data suggest there is no reason why oils containing large amounts of rerefined base oil and appropriate additives should not satisfy the specifications of such tests. Acknowledgment This work has been supported by-a grant from the "Junta Nacional de InvestigasHo Cientifica e TecnolBgica" and from PETROGAL, for which I express my gratitude. I am also indebted to my collegue M. Jeronimo for helpful discussions and to many PETROGAL workers who contributed to this program. Registry No. PhMe, 108-88-3; KOH, 1310-58-3; n-hexane, 110-54-3; 2-propanol, 67-63-0.

Literature Cited Berry, E.; MacDonald, L. P.; Skinner, D. J. Experimental Burning of Waste Oil as a Fuel in Cement Manufacture; Report No. EPS 4-WP-75-1; Department of Environment: Ottawa, Ontario K1A OH3, Canada, i976. Bieda. R. J.: Associates. Predesien Cost Estimates for Re-refined Lui& Oil 'Plant; Report RI-77711; Bartlesville Energy Research Center: Bartlesville, OK, 1977a. Bigda, R. J.; Associates, Comparison of BERC Re-refining Process with Acid/Clay/Distillation Process; Report RI-77/19 Bartlesville Energy Research Center: Bartlesville, OK, 1977b. Brownawell, D.; Renard, R. Rerefining of Used Oils, U.S. Patent 3 639 229, Feb 1972. Corlew, J. S.; Sluski, R. J. Treatment of Waste Lubricating Oil Using BERC-ERDA Solvent; Report RI-76/11; Bartlesville Energy Research Center: Bartlesville, OK, 1976. Crefol, Companhia Refmadora de Oleos, Ma. Estudo da Viabilidade Econ6mica do Process0 de Regenerapb de Oleos Usados Baseado no Solvente Hexano/Isopropanol; Report presented to Caixa Geral de Depositos and Instituto de Apoio is Pequenas e Medias Empresas as a Candidate to Prize and Financial Support for the Best Innovative Projects, 1984. Fauser, F.; Ritz, W. Recyclon-A New Method of Re-refining Spent Lubricating Oils Without Detriment to the Environment; ASEOL-Leybold Haereus brochure, Leybold-Heraeus: Hanau, West Germany, 1977. Goossens, A.; Westerduin, R.; Suchanek, A. Why Re-refining of Spent Lubricating Oils; KTI brochure; Kinetics Technology International (KTI): 26 Bredewater, Zoetermeer, P.O. Box 86, The Netherlands, 1976. Institut Franpais du PBtrole. MBthode d'Etude de la StabilitB des Lubrifiants sur Banc de CokBfaction ANTAR Report; Geomecanique: 370 Av. Napoleon Bonaparte, 92 Rueil-Malmaison,France, 1971. Institut of Petroleum. IP-48 Test, according to British Standard BS-4704; Heyden and Son Ltd (on behalf of Institut of Petroleum): London, 1971. Jordan, T.; McDonald, W. Method of Reducing the Lead Content of a Used Hydrocarbon Lubricating Oil by Adding Methyl Ethyl Ketone to Separate the Resulting Mixture into a Coagulated Insoluble Phase. US. Patent 3 763 036, Oct 1973. Reis, M. A. Regenerapgo de Oleos Lubrificantes Usados por ExtracpBo-FloculapHo. Ph.D. Dissertation, Universidade do Porto, Faculdade de Engenharia, Porto, Portugal, 1982. Reis, M. A.; Jeronimo, M. S. Regeneracgo de Oleos Lubrificantes Usados por ExtracpBo-FloculapHoRBpidas. Portuguese Patent 69 392, March 1979. Reis, M. A.; Jeronimo, M. S. Fabrico de Tinta de Base Para Fabrico de Tinta de Impressgo a Partir de Oleos Usados. Portuguese Patent 75 702, Oct 1982. Reis, M. A,; Jeronimo, M. S. Waste Lubricating Oil Rerefining by Extraction-Flocculation. 1. A Scientific Basis to Design Efficient Solvents. Ind. Eng. Chem. Res. 1988, 27, 1222-1228. Reis, M. A.; Jeronimo, M. S. Waste Lubricating Oil Rerefining by Extraction-Flocculation. 2. A Method to Formulate Efficient Composite Solvents. Ind. Eng. Chem. Res. 1990, 29, 432-436.

2456

Ind. Eng. Chem. Res. 1991,30, 2456-2463

Whisman, M. L.; Goetzinger, J. W.; Cotton, F. 0. Some Innovative Approaches to Reclaiming Used Crankcase Oil. Report RI-7925; Bartlesville Energy Research Center: Bartlesville, OK, 1974. Whisman, M. L.; Reynolds, J. W.; Goetzinger, J. W.; Cotton, F. 0. Process for Preparing Lubricating Oil from Used Waste Lubricating Oil. U.S.Patent 4 073 719, Feb 1978a. Whisman, M. L.; Goetzinger, J. W.; Cotton, F. 0. Method for Reclaiming Waste Lubricating Oil. U.S. Patent 4 073 720, Feb

1978b. Whiaman, M. L.; Reynolds, J. W.; Goetzinger, J. W.; Cotton, F. 0.; Brinkman, D. W. Re-refining Makes Quality Oils. Hydrocarbon Process. 1978c, 57 (Oct), 141-145.

Receiued for reoiew April 16, 1991 Reuised manuscript received July 8, 1991 Accepted July 17, 1991

Adsorptive Separations Based on the Differences in Solute-Sorbent Hydrogen-Bonding Strengths Nirmalya Maity, Gregory F. Payne,* and Jennifer L. Chipchosky Department of Chemical and Biochemical Engineering and Center f o r Agricultural Biotechnology, Uniuersity of Maryland, Baltimore County, Baltimore, Maryland 21228

Selectivity in adsorptive separations can be enhanced by limiting solute-sorbent interactions to a single or a few specific mechanisms. This work examines the potential of exploiting solute-sorbent hydrogen bonding as a selective adsorption mechanism, for solute adsorption from a nonpolar solvent onto a polycarboxylic ester sorbent. The hydrogen bond is believed to be formed between a proton-donating group on the solute and the carbonyl group on the sorbent. Studies were conducted for three classes of solutes, all of which can hydrogen bond, to determine whether differences in the strengths of adsorption can be exploited for separations. The enthalpies for adsorption from the nonpolar solvent onto the polycarboxylic ester sorbent were determined from calorimetry to be -5.1, -6.4, and -8.2 kcal/mol for the adsorption of N-methylaniline, alcohols, and phenols, respectively. In single-solute-adsorption studies with these solutes, we also observed a strong correlation between the adsorption affinity and the adsorption enthalpy. In studies on the adsorption from mixtures of two solutes, we observed that the solute with the higher adsorption enthalpy was preferentially adsorbed and that the temperature dependency of the separation factor could be related to the difference in the adsorption enthalpies of the two solutes. A simple thermodynamic framework, using data from single-solute studies, was capable of successfully predicting separation factors and the temperature dependence of separation factors.

Introduction Adsorption is gaining wider acceptance for large-scale separation from liquids (Sircar and Myers, 1986). The low-energy nature of adsorptive separation process can be advantageous compared to distillation for the separation of liquid mixtures of low volatilities or for mixtures with relative volatilities close to 1 (Ruthven, 1984). In comparison to extraction, the high concentrating abilities of adsorption have been exploited especially for the more efficient removal of solutes from dilute aqueous solutions (Faust and Aly, 1987). Currently, however, because of the difficulties in establishing multistaged contacting, adsorption operations are generally conducted in semibatch fixed bed mode, although complex equipment can be utilized to stimulate moving bed operations (Broughton et al., 1970). Fixed bed operation is ideally suited for two types of problems. In the first type, fixed bed operation is utilized with a nonselective sorbent to adsorb a wide range of solutes from a stream. An example of this problem is the removal of of organics from wastewater streams using activated carbon (Faust and Aly, 1987). In the second type of problem the goal would be to selectively adsorb one solute from a solute mixture. To achieve this goal, highly selective sorbents are needed. An example of such sorbents are molecular sieves which limit nonspecific adsorption through steric constraints. Thus, by using molecular sieves, it is possible to achieve high separation factors and therefore greatly reduce the need for multistage contacting to achieve separations of one solute from a mixture of solutes (Collins, 1968; Breck, 1974). The use of molecular sieves has thus demonstrated the potential

oaaa-5aa5pi12630-2456$02.50/0

of adsorptive separation when highly selective sorbents are available. These results have also stimulated the development of alternative sorbents which confer selectivity not through steric limitations, but rather by limiting adsorption to specific chemical interactions (Alexandratos and Kaiser, 1990). The development of such sorbents has been greatly facilitated by advances in polymer synthesis methods, which has made it possible to produce polymeric sorbents of well-characterized and uniform chemical surfaces (Albright, 1986). In our research, we have focused on hydrogen bonding as an appropriate adsorption mechanism. Hydrogen bonding is a useful mechanism for separations because the low energy of this bond ensures reversibility (Le., for recovery through desorption), while the directionality and the short range of this "bond" confer selectivity. In initial studies, we observed that solutes capable of hydrogen bonding were adsorbed onto a polycarboxylic ester sorbent (Amberlite XAD-7). Solutes unable to donate a proton for the formation of a hydrogen bond (e.g., benzene, anisole, and N-methylindole) were not adsorbed onto this sorbent (Payne et al., 1989). Subsequently we examined the potential of this polycarboxylic ester sorbent for use in separations and obtained high separation factors when one solute which could hydrogen bond was adsorbed from a mixture containing a second solute which was unable to hydrogen bond. Further, we observed that although the primary and the secondary amines aniline and Nmethylaniline, respectively, were adsorbed with similar enthalpies, the primary amine was selectively adsorbed from the mixture with a separation factor of 2 (Payne and 0 1991 American Chemical Society