A
compression still has been devised which contains a rapidly rotating eva porator-condenser of sheet metal. Under centrifugal force the liquid films on both sides o f the rotor are maintained so thin that over-all heat transfer coefficients of U = 2500 to 6000 B.t.u./ft./hr./O F. are routinely developed. Many tons of sea water have been distilled from a pair of copper cones 18 inches in diameter without development of scale and without the heat transfer coefficient departing from an initial U = 3600. A temperature differential as low as 3’ F. has been realizable. Experiments have been extended successfully with disks 4.5 feet in diameter and performance extrapolated to larger sizes. So successful were tests that a still with 8-foot rotors has been constructed. The more significant parameters of this kind o f still have been measured and recorded in this article. It would appear that the relative area of evaporator-condenser surface can be diminished by a factor of 5 to 10 times and the total power requirement o f a water producing installation, given suitable ancillary equipment, reduced to less than 5 watt-hours per pound of distillate or about 42 kw.-hr. per 1000 gallons of fresh water distilled from sea water. This is about 17 times greater than the absolute, “second law” minimum postulated in a previous article (4) but less than half the power requirements of existing compression stills if they are operated at an acceptable ratio of output to first cost. 786
K. C. D. HICKMAN
136 Pelham Road, Rochester 10, N. Y.
Editor’s Note: Feature of the March 1957 Scientific American was an article b y David S. Jenkins, director of the Office 04 Saline Water, U. S. Department of the Interior. This article, which described generally the various salt-water conversion methods being investigated under Interior’s sponsorship, said partially o f the Badger-Hickman still, “In the past three years interest in compression distillation has been heightened b y an exciting new system. In essence what Hickman has added is a simple device for increasing phenomenally the rate of heat transfer to the water: namely spreading it out in a thin film. The salient feature of this device is a rotating drum, shaped something like a child’s musical top.” I&EC’s article on the succeeding pages describes in detail the development of this still. As this article went to press, the editors were advised that the semicommercial No. 5 still, pictured on the cover, had reached its predicted operating capacity.
T H E FIRST recompression still, complete with falling film evaporator and steam compressor, was patented by Harrison (2) in England in 1856, soon after the steam engine had been developed. Many variants were proposed in the next 80 years, including tubular flash boilers ( 9 ) and slowly rotating drums ( 7 7) and, in 19 10, Soderlund and Boberg (70) drew attention to the importance of low heat differential in the conservation of power. Large stills were constructed in Switzerland for the concentration of cane sugar solutions (72) and, at about the commencement of World War 11, Kleinschmidt (6-8) designed for the U. S. Navy the first rugged, generally useful still for the separation of potable water from sea water. Studied further by engineers ( 7 , 5) at Arthur D. Little, Inc., and manufactured by E. B. Badger and Sons (now by the Badger Manufacturing Co., Cambridge, Mass.), thousands of these stills have been placed in operation, as have many related stills by other makers.
INDUSTRIAL AND ENGINEERING CHEMISTRY
Method of Operation There are three essential parts to a compression still (Figure 1) : the still proper, the vapor compressor, and the feed-effluent heat exchanger. Steam is withdrawn from the surface of the distilland, compressed, and returned to the outside of the boiler tubes, where it condenses, passing the latent heat of evaporation through the tube walls and into the feed water, which rises slowly toward the surface if there is no submerged nucleation, rapidly if there is. I n either event: there develops a temperature gradient, At, of 12’ to 20’ F. for useful rates of evaporation and, in addition, there is a degree of superheat, caused by the inefficiency of the compressor and the adiabatic cycle under which it operates, somewhat in excess of the minimum indicated by the Mollier diagram. The boiling point difference (or elevation, BPE) between pure water and the 5 to 10% brine (concentration depending on rate of blowdown) circulat-
,
ing at steady state in the boiler is 2’ to 3’ F. a t atmospheric boiling point, so that a thermal driving force 5 to 10 times the boiling point difference is being employed to effect the transference. The machine itself, ‘generally manufactured of heavy admiralty bronze or copper-nickel, follows high pressure boiler design, though the pressure relative to the atmosphere and pressure differential between the evaporating and condensing sides are both negligible. When operated on sea water at temperaturesabove 180’ F., a hard scale of low thermal conductivity deposits on the evaporator tube after a few days’ service, necessitating remedial measures. There is, thus, ample room, functionally and structurally, for improvement. I t seemed (4) that if the conventional boiler tubes could be replaced by a rapidly rotating circular metal pan and if the water could be spread on one side by centrifugal force and vapor from the water be conveyed to the other side by a fan, it should condense and immediately be flung off-as suggested diagrammatically in Figure 2. Advantages such as the following might accrue. Vapor would be evolved from a continuously renewed, unobstructed liquid surface of progressively increasing solute concentration (would observe to the full the principle of “least mixing”). The length of path and resistance to
Figure 3.
Figure 1. Conventional vapor recompression still 1. 2. 3.
Figure 2. Centrifugal evaporator-condenser or phase barrier
Boiler-condenser Roots type steam compressor Feed-effluent heat exchanger
flow of heat through the liquid films would be brought to substantially a n irreducible minimum. The temperature differential, 4t, and pressure differential, Ap, could be lowered correspondingly, reducing both the
Sectional elevation of No. 1 still
Single rotor 15-inch diameter with positive displacement steam compressor
Figure 4.
work of compression and the mechanical strength required by the phase barrier which separates the distilland from the distillate. The low Ap would permit the customary positive displacement compressor
Rotor and residue scoop of No. 1 still
Thermal lagging and accessories removed
VOL. 49, NO. 5
M A Y 1957
787
to be replaced by a centrifugal or axial flow blower. The blower could handle relatively larger volumes of steam, allowing operation a t low temperatures and pressures. The combination of low temperature, low heat gradients, and large, centrifugally generated shear forces in the distilland layer should discourage the deposition of scale. Low temperatures would diminish or eliminate the approach heat exchanger ordinarily needed to recover heat from the effluent brine and distillate.
Early Experiments
Figure 5. No. 2 still, accessories removed Centrifugal blower with motor in center foreground
Figure 6.
Sectional elevation of No. 2 still
Parabolic double rotor 18-inch diameter
788
A program to test these assumptions was initiated, and since June 1952 four stills have been constructed and tested. A fifth, the largest, of a commercial type, is now entering its test program. The KO. I still is shown in Figure 3 and in Figure 4, which presents a front view with the cover off and accessories removed. The essential working part is the conical evaporator-condenser or phase barrier, A , which is rotated by the shaft. B. -4second cone, C, integral with the first, made friction closure with the vertical steel back plate on which the assembly was mounted. The salt feed water was applied to the center of the rotor and whirled out1vard in a thin film, to collect in the peripherai gutter, whence it was withdrawn by the scoop tube, D . The steam liberated by the water film was collected by the positive displacement pump located at the upper rear OF the back plate and returned under slightly greater pressure to the condensing vessel formed by the tlvo cones. The condensate, flung from the back of the phase barrier cone A, collected at the periphery and was removed by the scoop tube, E. A primitive feed-effluent heat exchanger, F, and water manometer, G, completed the working parts. The apparatus was wrapped with removable thermal lagging. Reduced pressure operation was secured by a water aspirator attached to the pipe and gage H,when desired. No, 1 Still Performance. The still operated successfull), at atmospheric pressure and, with the rotor turning at 1000 r.p.m. and the feed and steam pumps actuated, distillate was produced when the casing temperature rose above 212' F. The performance, relative to similar devices, was assessed in terms of heat transferred across the conical phase barrier. Here, three heat transfer coefficients may be recognized-kl, flow across the distillate layer, k l : across the metal barrier, and ks, across the evaporating layer. A composite figure of
INDUSTRIAL A N D ENGINEERING CHEMISTRY
1260 B.t.u. (hi.)!(sq. ft.)(" F.)
CENTRIFUOAL BOILER COMPRESSION STILL was obtained during the first week and above 1900 when means had been provided for removing the noncondensable gases that accumulated in the condensing compartment. The values may be compared with established procedures as suggested in Table I.
vessel. I n this way tap water, brine, or sea water, sometimes colored with red dye, was recirculated indefinitely under substantially stable conditions of temperature and composition. The temperatures requiring measurement were those of the saturated steam in the casing, obtained by the mercury thermometer, Y , and the feed and residue streams by thermometers Z and Z', placed in tees in the piping.
Table 1. Comparison of Composite Heat Transfer Coefficients
The more significant variables were the electrical energy supplied to the internal blower and the average temperatures, io and t, of the evaporating feed surface and the surface of the condensate across the radius of the rotor, leading to the mean temperature difference, At, and the pressure differential, Ap, created by the blower. The temperature of the product distillate corresponds exactly with the temperature on the condensing surface and the temperature measured in the casing. This exact temperature correspondence follows from the fact that the distillate, passing through the steam as a fine spray and afterward spread over a large area of the walls and
1 l / k l + I/kl = = 600 to 900 B.t.u./(hr.)/(sq. ft.)(' F.) Tubular compression still, 1 = U = 650a l/ki 1/kz l/ks No. 1, compression still, at start, 1 = U = 1250 l/kj l/kz 1/ka No. 1 compression still, with condenser purging, U > 1900
Power-house condenser,
c
+
+
+
+
a Average of values accumulated over many years by Badger Mfg. Co.
No. 2 Still. A preferred way to build a centrifugal evaporator-condenser was evidently to mount two dished spinnings face to face (ultimately a succession of paired spinnings as in the No. 5 still) on one shaft, as indicated in Figures 5 and 6.
P
Here, a large bearing block, A, secured in the end of a cylindrical casing, B, accommodated a long shaft, C, rotated by a pulley, D. A rotor spinning, E, fastened to an Oilite bronze bearing ring, F, made turning closure with block A , while the periphery was sealed to a similar rotor spinning, G. This second spinning terminated in a bronze ring, H, supported by a three-legged spider, J, on shaft C,which in turn was supported in the bearing, K , mounted on a spider system, L, L', fastened to B. I n place of a positive displacement compressor a vacuum cleaner motor and blower was adapted and, surprisingly, operated well a t high speeds and without overheating in the steam partial vacuums later employed. Closure with the rotor shaft was made by a conventional stuffing box, Q, screwed into A, and feed was admitted and of residue withdrawn through holes drilled in this block. The experimental source of feed was a tank warmed with a thermostatically controlled immersion heater, S,and feed passed by suction therefrom into the still through a valve and a degassing sight bulb, T. Residue and distillates were withdrawn through tubes U and V, by the oversize twin sump pumps, U', V', which also doubled as vacuum pumps. A laboratory vacuum pump was employed for quick or supplementary evacuation of the still. Residue and distillates were usually returned to the feed tank, except when transferred to a weighing
and could be duplicated precisely by duplicating the wattage fed to the blower motor from a variable transformer. The effective area of the rotor-pair was 3.85 sq. feet, and the output of distilled water ranged from 20 to 50 pounds per hour, or about 50 to 125 gallons per 24-hour day. The comparative performance was calculated according to the example, below, for which 4 t was determined as follows:
A. The condensing temperature was read from the thermometer in the still casing and the corresponding saturated steam pressure determined from the steam tables (Keenan and Keyes). B. The differential manometer was read, and the pressure so indicated added to the pressure found under A. From this pressure and the steam tables, the corresponding evaporating temperature was found. C. Subtracting the calculated evaporating temperature from the observed condensing temperature furnished the difference, At. D. The over-all heat transfer coefficient was calculated by the equation
u = (lb. of distiliate)(hr.)
X enthalpy, steam = B.t.u./(hr.)(sq. ft.)( F.) At X area of rotor, sq. ft
bottom of the casing, is always in contact with saturated, completely desuperheated steam. The temperature of the composite residue stream differs from the average evaporating temperature in progression across the rotor. While this is important for heat conservation, it vitiates slightly the calculation for a true 44 and the over-all heat transfer coefficient. For these reasons, 4b was measured directly and At calculated therefrom by reference to the steam tables. OPERATING No. 2 UNIT. The briefest experience showed that the unit was self-adjusting and relatively insensitive to operating conditions. If the rotors and blower were turning while feed solution was supplied and residue withdrawn, distillate appeared as soon as the vacuum pumps had reduced the pressure in the casing to the saturation pressure of water a t the temperature then prevailing. Distillate was running freely at 72' F., rapidly at 120' F., with little further increase above 140' F. (The reversal of temperature trend was eventually laid to the characteristics of the blower.) Rate of feed proved to be unimportant over a 5 to 1 range and production of distillate from city water could be varied from 10 to 60% of feed with little influence on power consumption per pound of distillate. The blower speed varied from 10,000 to 18,000 r.p.m., according to the temperature of the steam and output of distillate selected. This output was determined
For example, from Table 11, line 9, water distilled equals 31 pounds per hour, condensing temperature 135.2' F., whence condensing pressure equals 70.61 inches of water column, absolute. Measured Ap equals 4.1 inches of water column whence evaporating pressure equals 66.51 inches and temperature equals 132.9'; giving At = 2.3' F. If the mean latent heat of evaporatioil and condensation is taken as 1017.4 B.t.u. per pound, the over-all heat transfer coefficient is given by 31 X 1017.4 U = 2.3 X 3.85 3560 B.t.u./(sq. ft.)(hr.)( O F.)
where 3.85 equals areas of two rotor halves. Typical data obtained with the No. 2 still are given in Table 11, where all listings except the last show great consistency. The final item derives from a barrel of contaminated Boston Harbor water sent to Rochester for a continuous run through the still, and the yield of distillate was somewhat curtailed. The extremely high yields of water in the other experiments, in relation to temperature differential, reflected by heat transfer coefficients in the range of U = 2000 to 4000, removed all doubt as to the utility of the early concepts. The question now emerging was: Can a rotating boiler-condenser compare favorably with a stationary machine of conventional design operating with one fifth the heat transfer coefficient? T o VOL. 49, NO. 5
0
MAY 1957
789
examine this question three more stills were built. No. 3 Still. Substantially a duplicate of the No. 2 in size and function, it was redesigned for prolonged operation under precisely determined conditions. Copper rotors were substituted for the aluminum, the motor for the blower was mounted outside the still with connection through a high speed stuffing box, the rotor was turned by a variable speed motor, and the differential manometerwas improved. The still and a typical flow sheet are illustrated in Figures 7 to 9. Precise Steam Manometer. The calculated heat transfer coefficient cannot be more accurate than the measurements of steam pressure differential. The water manometer relies, for accuracy. on the quantitative reaction of steam pressure against the ends of the two water limbs. If the limbs are closely connected to the steam space, steam condenses and water soon fills the manometer; if the limbs are at a distance, the pressure is not properly transmitted. A preferred arrangement for the manometer shown in Figure 10 has the top of the taller limb of the glass U-tube connected with the still casing, while the other limb is divided and both branches are connected with the inside of the rotor through the bearing block. Steam from the rotor cavity, owing to the high purity, streams constantly into these two branches, the condensed water overflowing through the block back into the rotor. A meniscus becomes established at a fixed level in the other branch and remains in equilibrium with the steam pressure conveyed through the upper pipe of this branch. The steam communicated through the taller limb is less pure because it contains noncondensable gases left behind by the mass transfer of product water, and it soon fills the limb with cold gas, by which the true steam pressure in the casing is transmitted.
A Figure 7. Rotor drive end No. 3 still, lagging and accessories removed Figure 8. No. 3 still, 18-inch-diameter
4 rotors of various interchangeable contours Figure 9. One of various arrangements of piping of No. 3 still
Figure 10. Glass manometer for measuring steam pressure differential in No. 3 still
Table II.
Performance Data for No. 2 Still At Distillate Rate, Lb./Hr. 13.5 15
Heat Transfer Coeff. = U 2 120 2410
Watts on Blower Motor 70 95
18 20.5 22 23 24 31 38 41 11 17.8
2550 2740 3180 3070 3650 3560 3760 4180 3550b 3180b
95 100 100 100 100 210 250 300 70 170
5.28 4.88 4.55 4.35 4.17 6.00 6.58 7.32 6.36 9.55
1000 44 12.9 2720b 2.1 1.3 SeaC 99 1.60 Residue from still contained 7.4% NaC1, average of feed plus residue 6.57'0, b.p. rise 1.4' F. * Calculated from At adjusted for b.p. rise. Boston harbor water analyzing 2.8% NaCl equivalent. Average of feed and residue 4% NaC1, b.p. rise 0.8' F.
140
10.85
Feed Water Rochester city water, dyed pink
5.2Yo5NaCl
Temp. of Casing Condensing Steam, ' F. 78 80
AP, Inches Water 0.73 0.78
93 100 110 125 137 135.2 144 155 107 109
1.20 1.55 1.85 2.80 3.25 4.1 6.00 7.00 1.75 2.7
1.74 1.695
Speed of Rotor, R.P.M. 1000 1000
Feed Rate, Lb./Hr. 50 50
1.905 2.03 1.85 1.97 1.735 2.3 2.68 2.6 1.88 2.74
1000 1000 1000 1000 1000 1000 1000 1000 1000 1000
50 50 50 50 50 80 80 80 44 50
A$, O
F.
a
790
INDUSTRIAL AND ENGINEERING CHEMISTRY
Corrected for B.P. Rise, F.
0.83 1.50
WattHr./Lb. Dist. 5.18 6.13
CENTRIFUGAL BOILER C O M P R E S S I O N STILL No. 4 Still. A centrifugal device, such as the present type, could be highly sensitive to change in diameter. Area and forces increase as the square, volume and blower capacity as the cube, throughput in relation to volume as the z/8 power, and throughput to weight of structure as about the square root. A search for optimum size and greatest practical size for the rotors goes to the heart of the matter. Quantitative measurements with the No. 3 machine and its 18-inch-diameter rotors are properly related to measurements with much larger rotors and are thus presented with the experimental data from the No. 4 still. The diameter chosen was 4.5 feet, leading to an area nine times that of the No. 3 rotors and 0.3 times the area of the 8-foot rotors proposed for the 25,000gallon-per day, No. 5 utility still. Because of its experimental purpose, the mechanical drive and all but one of the steam and water connections were attached to the lower half of the rotor pair in the No. 4 unit. The upper half could be removed rapidly for inspection or adjustment, and both halves could be changed without touching the drive mechanism. This dictated the vertical axis and removable cover (Figures 11 and 12).
lt
*
The unit comprised a cover, A , resting on a circular dished base, B , which supported a central bearing tube, C, 9 inches in internal diameter and stout enough to carry ball bearings And a stuffing box, not shown. Cooperating with these was the tubular hub, D , on which was mounted the lower rotor, E, and the spur gear, F, which was driven by the gear, G, shaft and pulley, H, and speed changer and motor indicated a t J. Steam, withdrawn through pipe C, was compressed by the five-stage Spencer turbine, L, and returned to the casing. A manometer, M , measured the pressure difference between the inside of the rotors and the casing, employing the same water overflow system for the rotor limb as devised for the Nos. 2 and 3 machines. Feed solution, withdrawn by the suction in the still from a tank, not shown, was routed through the feed-effluent heat exchanger, N , the shutoff and adjustment valves, 0, and the flowmeter, P, to the preheater, Q, and degasser, R,whence it flowed by gravity through a second trim heater, S, into the rotor to be distributed by the nozzles, T , T'. Residue was picked up from the periphery by the two scoops, U, U', and forced by the small centrifugal withdrawal pump, V, through the heat exchanger, N , and thence either to waste or back to the supply tank for recycling or a second tank for storage and a later further pass through the still. Similarly, the distillate was withdrawn from the casing, D, by the pump, W , and passed through exchanger N a n d routed either to waste or the supply tank, for recycling. Because
Figure 1 1.
Sectional elevation and flow sheet of No. 4 still Rotor 54-inch diameter on vertical axis
of the sprawling arrangement of the apparatus and consequent difficulty in lagging, extra resistance heaters, not shown, were applied to the still casing and larger pieces of piping. Thermometers were installed a t various points. To preserve the appropriate vacuum and prevent gases dissolved in the feed from entering the still, a flexible pipe, Y , connected the cover of the still with the base of the packed degasser, R. Purge steam passed upward and out a t the top by pipe Z to the small surface condenser, ZA, which was exhausted by the 27 CPM Kinney vacuum pump, ZB. Condensate collecting in the surface condenser and the exhaust tank, ZC, were measured at Z D , while the total of leakage and purge gas was determined periodically a t ZE. The only power which generates distilled water at a rate greater than a single-effect evaporation is that truly
Figure 12.
expended in compressing steam. All other power, including that lost in the blower, is dissipated in friction and lost in radiation, convection, and singleeffect distillation, represented, for instance, by condensate collecting in Z D . The objective, then, is to secure vacuum and the purest operating steam with the least possible bleeding of purge steam through pipe Z A to ultimate loss as singleeffect condensed water at ZD. Various means can be devised for reaching this end. I n the present experimental setup it suffices to maintain the temperature of the feed in the degasser, R,at 3' to 10' F. below the temperature in the casing of the still, and close off the valve, ZF, and adjust the bleed valve, ZG, to the point that further constriction would cause diminished still performance. With practice, less than 1% of the product need be collected as purge distillate
Rotor casing, blower, and variable-speed rotor drive of No. 4 still Thermal lagging and instruments removed VOL. 49, NO. 5
M A Y 1957
791
of distillate to & l % , and time to 1 second in 300, the precise values were accurate within =t570. Routine values were those not regularly referred back to datum conditions and could vary 10% on an absolute scale, though they were often consistent to within 1%, and repeatable during a day's run to O.5yG. The method of calculating the heat transfer coefficients has been described.
*
Properties a n d Parameters of Stills
Figure 13. rotors
Contours of experimental
No. 1 half-rotor 15-inch diameter No. 2 and 3 rotor-pair, parabolic, stepped 18inch diameter No. 3 rotor-pair, conical, included apex angles a t center of rotation of 1 15', 138', and 158' No. 4 rotor-pairs, 54-inch diameter, apex angles a t shaft of 168" and 1 3 5 '
Feed Water Supply
Four qualities of feed Lvater were used: Rochester, N. Y.. city water; Cambridge, Mass., city water; Boston Harbor water, drawn from Atlantic \Vharf; and ocean water drawn from Gloucester, Mass. The saline waters were trucked to the laboratory in carefully cleaned tanks and during the longer runs one tank truck was set aside solely for the purpose. The incoming sea water was transferred to a 1000-gallon wooden tank, and fed from there to the stills. Distillates and residues could be sent to a second 1000-gallon tank for re-use ~ 1for ' collecting a stronger brine for feed purposes. Method a n d Accuracy of Measurements
The procedure for all the stills was substantially similar. The casing was closed, the vacuum pump started, the rotor set in motion, and feed water admitted, and the withdrawal pumps were immediately started. Preheaters or adjustment (trim) heaters were energized and as soon as the temperature and pressure converged to a steam saturation point the blower was started and distillate shortly appeared. Operation was then leveled off at the selected values of temperature, rotor and blower speeds, etc. Two classes of experimental measurements were adopted : precise and routine. Values designated as precise were obtained when a still had been checked during operation as able to reproduce an optimum yield recorded for a selected set of datum conditions. such as casing temperature 125O, Ap 6.85-inch water, At 5' F.. rotor r.p.m. 1200 and 400 for the Nos. 3 and 4 stills, respectively. With the casing thermometer providing readings to an accuracy of 0.25' F. and the manometer to 0.05 inch, the weight
792
Performance us. F e e d Rate a n d Frac-
Rotors. The shapes and sizes of the rotors so far experimented with are shown in Figure 13. The N o . 1 still used a half-rotor 16 inches in diameter spun from sheet aluminum with a deep peripheral gutter. The No. 2 still rotor was roughly parabolic, \vith two concentric steps to receive feed water at two points along the radius and was spun from 'jaz-inch aluminum. A similar rotor, spun from I/32-inch copper sheet, was made for the No. 3 still, and this was followed by three other copper rotors spun as straight cones having included angles at the peripheries of 65', 42O, and 22'. The performance of all the No. 3 rotors 18 inches in diameter was similar, and for this reason simple conical rotors were chosen for the Nos. 4 and 5 stills. The relative performances are shown in Table 111, all except the first value being determined with Cambridge citv water as feed. Means have not been available for ascertaining the distribution of feed water on the rotors nor the film thickness at any point. However, the increase of heat transfer coefficient was measured as the feed progressed over successive concentric zones of the surface. Attempts to use the outer distilling zones only, by applying the feed at points progressively away from the center, failed because splashing from the residue scoop wetted the central zones to a considerable but unmeasured degree. At the suggestion of Roger Newton, concentric zones on the condensing side of the rotor were painted with thick layers of poorly conducting
INDUSTRIAL A N D ENGINEERING CHEMISTRY
Table 111. Rotos Still Diam., No. Inches Material 2 18 Aluminum Copper 3 18 Copper 3 18 Copper 3 18 Copper 4 54 4 54 Copper a Peripheral included angle.
Table IV, Eff e'ctive Operating Area Full cones Outer Outer Outer
814 1/2
1/4
paint, so that only the remaining uncovered annuli were in effective operation. The data, recorded in Table IV, show the trend in increase of heat transfer efficiency with distance from the center of the rotor. The comparison is between the whole rotor and varying proportions of outer zones only. No comparison has been made between the relative efficiency of an inner zone and an outer zone, but the presumption is that differenceswould be considerablygreater.
tion Distilled. The No. 3 still with 18inch rotor at (datum) 1200 r.p.m. and temperature 125' F. was insensitive to feed rate over a 4 to 1 range. With constant blower input, the yield of distillate from fresh water feed remained constant. M'ith salt water, the output varied in accordance with the mean boiling point rise of the distilling layer, but the heat transfer coefficient adjusted (for BPE) remained constant. The distilling capacity of the 4.5foot rotor of the No. 4 still, taking approximately 10 times the volume of water, was somewhat dependent on the absolute rate of feed and, hence, the ratio of distillate to residue. Yield with fresh water was greatest with the smallest residue that would keep the outer zones of the rotors wetted-e.g., with a 60% distillate cut. Jt'ith salt water and low rotor speed (250 r.p.m.), the potentially increased performance from diminished feed vias almost exactly compensated by the encroachment on At bv the increasing boiling point of the discharged brine. so that the yield varied only slightly over a range of 10 to 60% distillate cut (at constant blower input). The trends are shown in Figure 14. Performance us. Rotor Speeds. The relationship between yield of distillate and rate of rotation of an 18-inch copper rotor in the KO. 3 still is recorded in Figure 15 for three different power inputs to the steam blower; and pertinent data are replotted as heat transfer coefficients in Figure 16. However. because the work done in spreading the
Experimental Rotary Phase Barriers U = B.t.u./(hr.)(sq. ft.)(' F.) Form Parabolic Parabolic 22' conea 65' conea 22' conea 45' conea
400 r.p.m.
... ... ...
...
Rotor Speed 1000 r.p.m. 3300 3320 3700 3200
2800 3000
... ...
1500 r.p.m.
...
4400 4400 3650
...
. . e
Relative Efficiency of Concentric Rotor Zones Diameter, InchesInner Outer 5.12 18.5 18.5 10.25 18.5 13.6 18.5 16.2
Average Radius 5.9 7.2 8.0 8.7
Relative Heat Coefficients 1.00 1.04 1.18 1.50
C E N T R I F U G A L B O I L E R C O M P R E S S I O N STILL
FEED WATER POUNDS PER HOUR
+
ROTOR SPEED rpm
Figure 14. Variation of heat transfer coefficient with rate of supply of feed water for city and sea water
feed water is proportional to the square of the revolutions per minute and this work can increase from a negligible quantity to a serious charge against the economy of the process at high rates of rotation, the data are plotted again against the square of the revolutions per
Figure 15. Relation of yield of distillate to rate of rotation of 18-inch diameter copper rotor in No. 3 still, city water
in Figure 19. Here the heat transfer coefficients of ocean water, adjusted for mean boiling point elevation, are identical with those of fresh water and only one point lies off the line drawn through the other points. Performance us. Centrifugal Force. I t can be deduced from Figures 15 to 19 that for a given tip speed the larger rotor develops a lower over-all heat transfer coefficient than the smaller. Acceleration forces, expressed as times
minute in Figure 17. Probable values for the work expended accelerating the feed water (ratio of feed to distillate, 2 to 1) are shown in the upper abscissa. Similar data were accumulated for the No. 4 rotors at a condensing temperature of 143' F. and the compressor turning at the datum speed of 2700 r.p.m. Yield is plotted against rotor speed over the range 113 to 407 r.p.m. for both city water and ocean water in Figure 18 and heat transfer coefficient us. r.p.m.
noioI) r ~ m
Figure 16. Relation of heat transfer coefficient to rate of rotation of 18inch copper rotor in No. 3 still, city water
320
500
ROTOR SPEED r l m
Figure 17. Relation of heat transfer coefficient to square of rate of rotation of 18-inch copper rotor, city water PERIPHERAL TIMES GRAVITI
IW
SPEED OF ROTOR, rpm
Figure 18. Yield v5. r.p.m. of 54-inch copper rotor in No. 4 still, city water and sea water
ZW
300
4w
300
4 5' ROTOR SPEED r m
Figure 19. Heat transfer coefficient vs. r.p.m. of 54-inch copper rotor in No. 4 still, city water and sea water VOL.49, NO. 5
MAY 1957
793
IWO
I
1
0
40
IW
400
expenditure of z units of energy, 2% pounds should be transferred under 2y head, using 4t units of energy. The yield should thus be proportional to the square root of the load on the compressor. The relationship is well demonstrated in Figure 21, computed from Nos. 2 and 3 still data for small throughput ratios. Thus, Aluminum rotor, 100 watts, distillate = 22.5 pounds per hour, and 200 watts, distillate = 34 pounds per hour By calculation.\lg X 22.5 = 33.5 pounds Similarly, copper rotor, 250 watts, distillate = 45 pounds per hour, and 450 watts, distillate = 61.75 pounds per hour PRESSURE OIFFERENTIAL,lNCHES OF WATER
Figure 22. Variation of yield with steam pressure differential, No. 4 still, city water and sea water
gravity = G, diminish with increasing diameter for a given tip speed and it was thought that the heat transfer coefficient might be directly proportional to these forces, so that U = cG, where c is a constant for given still conditions. A convenient formula for calculating centrifugal force in terms of times g = Gis given by: TS( r.p .m. )2r G =
900 g
where r = radius in feet. The curve in Figure 20 relates G at the periphery of the rotor (G for intermediate positions varies as r ) with heat transfer coefficients for the No. 3 machine. Similar values from the 4.5-foot rotor data, added as crosses and squares, fit the curve of Figure 20 remarkably well. The precise relationship between centrifugal force and transfer coefficient is thus established. Even so, the data suggest that the larger rotor has a slight advantage over the smaller in this respect, due to secondary effects such as greater ventilation and turbulence in the condensing steam caused by the greater tip speed for a given centrifugal force. This is the main discernible advantage; other factors are in favor of smaller rotors. The mechanical work done on
794
the discharged fluids-steam, residue, and distillate-is proportional to the tip speed of the rotor, whereas centrifugal force is proportional to the tip speed squared divided by r. As the rotors are made larger, they must withstand larger collapsing forces from the steam pressure differential and must rotate relatively faster, thus approaching more rapidly the limiting strength of the material. A preliminary mathematical analysis shows that even with copper rotors, having high specific weight and poor yield strength, the maximum useful diameter is not less than 10 feet. Summarizing the experimental data, we have (Table V). Performance us. Loading. COUPRESSOR VARIABLE.As the speed of the steam blower is increased, more steam is transferred at higher pressure and more water is distilled. If x pounds of steam are transferred under y head with the
INDUSTRIAL AND ENGINEERING CHEMISTRY
Table V. Rotor Diameter, Inches 18 54
The power data accumulated from the highly experimental No. 4 still have little quantitative meaning. This applies to the sotor drive, external pumps, and blower and drive. With a feed rate ot 1000 pounds per hour and a sotor speed of 400 r.p.m., the power required to accelerate the water over the surfaces, computed conservatively at twice mass times peripheral velocity squared, is barely 150 watts, whereas the power actually consumed was 1 kw. at the rotor bearing and 1.5 kw. at the driving motor the excess power being taken up by the speed changer and the 10-inch diameter shaft closure. Similarly, the gross power consumed by the steam blower varied from 2 to 8 kw., with 1.1 to 5.5 kw. actually delivered to the impellers. The five-stage Spencer turbine had an average rated adiabatic efficiency of 160jo0,compared with the 65 to 75% that could be expected from the singlestage high velocity impeller that would be included in a utility-type machine. Nevertheless, power data have been
Relation of Rotor Diameter to Centrifugal Force Peripheral G = 62 R.p.m. u 400 287
2600 2750
Peripheral G = 125 R.p.m. L7 700 500
3150 3400
CENTRIFUGAL BOILER C O M P R E S S I O N STlLL Table VI.
Nature of Feed Water Cambridge, Mass., city
.
400
Gloucester, Mass., ocean
Nature of Feed
Water Cambridge, Mass., city Gloucester, Mass., ocean a
-.
Input to Rotor Output Speed to Rotor Changer, Rotor, R.P.M. Kw. Kw.
Calcd. Evap. Temp. of Feed, O
F.
118.34 119.38 120.52 121.53 117.14 118.76 116.47 120.45 122.46
1.65 1.65 1.65 1.65 1.65 1.65 1.65 1.60 1.60
a n d S e a W a t e r Feeds . Apparent Obsd. Temp. Log Steam DX. Mean Input Pressure Evap. Boiling to output Differand Point Steam Comto Casing Temp.* ential, ConElevation Compressor ComCorrected Ap, denser of Feed pressor, Changer, pressor, (conInches, At, B.P.E., R.P.M. Kw. Kw. Apparent densing) H20 O F. O F.
Experimental Data on No.
1.01 1.01 1.01 1.01 1.01 1.01 1.01 0.93 0.93
2700 2240 1830 1410 3140 2700 3170 2210 1460'
5.4 3.5 2.62 2.12 7.25 4.75 7.36 3.24 2.0
DisLb. Lb. Lb. Distil- tillate as % ' Feed/ Residue/ late/ Hour Hour Hour of Feed 958 967 962 968 968 946 931 938 962
580 702 788 877 458 619 459 740 920
U = B.t.u.(hr.)(sq. ft.)(O F,).
included in the complete log of the readings taken during this part of the investigation, and are reproduced in Table VI as an example for the many graphs which, for brevity, have been given without supporting data. The more useful compressor function from Table VI, plotted against still performance for the No. 4 still, is the speed of the impellers. Figure 22 relates r.p.m. and Ap with yield, the curve (a straight line) for city water feed passing exactly through the origin. The line for sea water meets the Ap axis a t exactly the differential corresponding with the boiling point elevation (BPE) for ocean water at zero rate of distillation. The curves for blower r.p.m. us. yield, of Figure 23, are less satisfactory because the operation of the blower became unstable below 1000 r.p.m. The substantially linear dependence of yield on Ap suggests that there is only minor change in heat transfer coefficient with loading (see Figure 24). This is in sharp contrast to normal evaporation phenomena promoted by turbulent boiling, where the proportionate rate falls off sharply with load. Evaporation of dilute solutions on the centrifugal rotor is thus dependent not on the usual disturbance of the liquid caused by below-surface ebullition but on the forced renewal of the surface in re-
378 265 174 91
510 327 462 198 42
39.4 27.4 18.1 9.4 52.7 34.6 49.6 21.1 4.4
3.95 2.39 1.64 1.27 5.52 3.44 5.5 2.16 1.12
NaCl
Equiv. of
Feed, P.P.M.
4 Still, City
124.5 124 124 124 125 125 125 125 125
123.1 122.6 122.6 122.6 123.6 123.6 123.6 123.6 123.6
6.3 4.3 2.8
1.5 8.5 6.5 9.3 4.3 1.6
0.2 0.2 0.3 0.2 0.3 0.6 0.6 0.6 0.6
2360 2440 2480 2520 2350 2010 1920 1860 1275
At
(-B.P.E.) F. 4.76 3.22 2.08 '1.07 6.46 3.97 6.09 2.36 0.43
... ... e..
*..
0.87 1.04 0.79 0.71
Total Watt Delivered Hrs./Lb. Approx. Watt-Hr./Lb. DistilExtra Watt Dislate, tilDisComHr./Lb. late, tilpressor Conoomlate, and sumed by pressor rotor Rotor Salt Feed
Apparent Over-all Heat NaCl TransReal Equiv. fer Transfer of CoeffiCoeffiDist., cient cient" P.P.M. = Uobsd. (-B.P.E.)
60 60 60 60 60 32,400 32,400 32,400 32,400
...
4.76 3.22 2.08 1.07 6.46 4.84 7.13 3.15 1.14
Corr.
2360 2440 2480 2520 2350 2450 2250 2480 3380b
10.5 9.0 9.45 14.0 10.73 10.5 12.13 10.9 26.7
2.67 3.92 5.8 11.1 1.98 3.09 2.15
5.1 24.0
13.17 12.92 15.25 25.10 12.71 13.59 14.28 16.0 50.7
1.2 1.5
1.5
Value unreliable due to small percentage of distillate.
F t g u r e 23. Yield vs. 500 blower r.p.m., No. 4 still, city 3 I water and s e a $ 400 water 0
CONDENSING TEMP 123 "F ROTOR SPEED 400 rpm X .CITY WATER FEED 0:SEA WATER FEED
v)
3
5w 300 w
5
200
-
100
-
Lo
/ I
1000
2OCO STEAM BLOWER
3000
F i g u r e 24. Variation of heat transfer coefficient with steam pressure differential, No. 4 still, city water and s e a water CONDENSING TEMP .123.F
I-
pm
-CITY X I
2
WATER FEED SEA WATER FEED
9 5 6 7 PRESSURE DIFFERENTIAL. INCHES Of W4TER
3
VOL. 49, NO. 5
B
9
MAY 1957
795
tigure 26. Heat transfer coefficient! temperature, No. 3 still, city water
CENTRIFUGAL BOILER COMPRESSION STILL oissoiveo gas can be p tering the still by preliminary degassing of the feed and by countercurrent washing with the purge steam. Air that leaks in cannot be so removed and is a direct charge against the purging operation. Experiments with purposely leaked air, made with the No. 4 unit, are recorded in Figure 29, which relates yield of distillate to the quantity of air withdrawn from the still. The residual gas discharged from the Kinney pump was piped to a large bottle of water and the water displaced under constant head in unit time was measured. When the still was under steady-state vacuum but not in operation, the discharge was less than 1 liter per hour, suggesting that inleakage of air was negligible. When water was flowing through under operating conditions, the gas withdrawn varied from 10 to 15 liters per hour, less for recirculated tank water, and more for certain batches of sea water, but ranging from 2 to 4 % of the volume of the feed at atmospheric pressure. As this is the recorded solubility of air in water a t ambient temperatures, the gas withdrawn evidently originated from the feed water. Most of the dissolved gas was liberated where the feed first met the still vacuum, at the head of the countercurrent degasser, (R, Figure I I ) , and what proportion passed into the still is not known The experimental method thus consisted solely in admitting controlled quantities of air to the still casing and measuring the total quantity expelled by the vacuum pump. It is this total quantity that is plotted as the solid line, against yield, in Figure 29. The dissolved gas ejected at the time of the experiment, when there was no admitted air, was 10.4 liters per hour and this has been subtracted from all valnes to yield the dashed curve of Figure 29, which reflects the effect of added air on yield. The first small quantities of leak air exerted the greatest proportional reprep. sion, which leaves a choice of suppositions that substantially all foreign gas had been removed from the still during normal operation; or that removaloffurther minute concentrations of gas presumably still remaining under normal operation would give a further sharp increase in yield. Which of these is correct only further extensive expesmentation can reveal. In spite of these uncertainties, an approximate calculationcan be made of the purity of the steam in the condensing side of the still. To assume probable values, let feed rate be 1000 pounds per hour the distillate be 500 pounds per hour condensing temperature of 125"
the and at a (an
._ Figure 30. Log of 300-hour sea water test, No. 3 still, heat transfer coefficient vs. time absolute pressure of 1.942 pouods per square inch). The specific steam volume is 178.6 cu. feed per pound and the total volume of dissolved gas liberated a t still pressure is
proper, the steady-state contamination would be
'Oo0 3% 14" = 3.64 cu. feet 100 X 62.5 X 1.942
If 90% were liberated in the degasser the impurity in the still would amoun to 0.014%; if 99%, then 0.0014'%.
Let the feed enter the top of the degasser a t 7' F. below still temperature and be warmed up to 2O below still temperature during passage; then 5000 B.t.u. will be withdrawn and 5 pounds of steam (approximately) will leave the still, condense, and return as hot water under vacuum. Let a further 10 pounds of purge steam collect in the Kinney pump and line condenser per hour; then a total of 15 pouods of steam, occupying 15 X 178.6 cu. feet, will leave the still each hour and carry with them, a t steady state, 3.64 cu.feet of gas. If t h i s gas had all originated in the still
The balance of evidence would appeal to be that the operating steam has beer. exceptionally pure during these tests and little further gain is to he expected from further attempts a t steam purification. Applying the same calculation to the added leaks, it is seen that 123 liters added directly to the still per hour reduces the yield from a sea water feed 320 to 160 pounds per hour. The MIume of 123 liters at still temperature and pressure is approximately 36.4 cu. feet
Table VII.
Summary of 300-Hour Test Data
Duration of test, h a s Quantity of sea water fed, Ib. Total quantity of distillate, Ib. Condensingtemperature, F. Condensing pressure, lb./sq. inch abs. Steam compreemr input, watts Pressure differential, A#, inches H.0 Temperature differential,Ai, ' B.
Yield of disiilhte, Ib./lu. Comprcasor, watt-hr./lb. distillate Heat transfer coefaoients = U = B.t.u./(hr.) bq. ft.)(' P.) Equlmlsnt sslt contents, feeds, p.p.m. Chloride content of all distillates, p.p.m.
Bacterid content, feeds Bacterial content, distillates Aooemnee. distillates
loo 3'99 = 0.136% , 15 X 178.6
315 18,900 (3 tal& Car$ 5670 105 1.1 11&117 1.1-1.6 1.3-1.8 1st Tank Car 2nd Tank Car 3rd Tank Car 18.5 12.5 14.5 6.0 9.0 7.9 3250 13,600
2200 27,748
3400 34.348
0.62.5
Bacterial coli aerogem-9 ColoniedML. Groups Index in loo0 MI in 24 Hr. st 35' C. 700
