FILTRATION AND PURIFICATION
The following are excerpts from a paper by John Don, Assoc. M. Inst. NT. E., on “The filtration and purification of water for public supply”:
If agricultural pursuits have to be tolerated within the catchment area, these should, if possible, be restricted to grazing. In any case, the use of farmyard manure and town refuse as top dressing is very objectionable, and can hardly fail to import polluting matters into the stream. Mineral fertilisers may be allowed if any concession on this point has to be made. Water proceeding from marshy ground is in no case to be recommended, for, in the first place, it is of the nature of stagnant water and holds the products of vegetable decay, and, secondly, it teems with insects and animalculae, dead and alive. Water from peaty grounds is discolored and, from the presence of peaty acids which it invariably carries, it is able to dissolve lead. Certainly these ingredients may be removed by precipitation with alum and other chemicals, and, if that is properly attended to, the supply may lie rendered perfectly wholesome. * * * In all undertakings of this nature the same considerations apply with regard to the area surrounding the wells, as in the case of catchment-districts A knowledge of the geology of the country will enable water authorities to determine the extent of land which they should have under their control.
Pollution of water from underground sources is of frequent occurrence, and to discover the most suitable means of locating the prime cause of contamination the surrounding area must first be searched for rubbish heaps, farmyard manure, cesspools, etc. The suspected spot is then dosed with a chemical substance, as common salt, ammonium chloride, fluorescein, or even with a culture of harmless bacteria, and the well is regularly pumped for some time. If the water gives indications of the presence of the added chemical, the deduction naturally is that there is some avenue of communication between the suspected locality and tin well in question. The interpretation of the results demands great care and experience, for one must particularly note the rate at which the chemical travels from the place wherein it was deposited to the water in the boring. By that means the observer will be able to form an opinion as to whether a fissure does or does not exist. Slow percolation would indicate unfissured strata, and through closely compacted ground bacteria and offensive matters do not penetrate to any great depth. This is why it has been recommended, as a controling experiment, to make use of a culture of bacteria, for, if that can filter into the boring from the suspected spot, other species of the same kind can do likewise. Storage reservoirs fall into several classes, not only in reference to their capacity relative to the daily demand, but, also, in regard to the admission or exclusion of light, and to the arrangements whereby their contents are either left comparatively still, or are forced to travel from compartment to compartment. The time of descent of any given particle depends upon its size, its density, its form, and, above all, upon the speed with which it is being carried forward. Where the sediment is fine and the storage capacity not very great, subsequent filtration must be depended upon to complete the purification. The oldest and the most natural method of forming a reservoir is to dam up a stream which is inclosed by rising ground. Here the chief causes which militate against the acknowledged benefits of impounding are to be looked for, in the growth of water-weeds, algse, plankton. In soft waters plant species of a low type vegetate very actively, and cause many objectionable results. The water carries away products of their decay, which become noticeable to the consumer both from the visible turbidity and from the attendant odors.
Much may be done to obviate these evils by judicious construction of the impounding dani. Roofing over is the perfect remedy, for plants of the species mentioned will not grow in the dark, nor readily in water which is kept in motion. * * * If the reservoir is exposed to the light, the dam should he as deep as possible, not less than about 30 ft., so as to discourage bottom growths. The slope of the sides is to be steep, so as to diminish the shallow margin on which the alga find an attachment. The borders are to be kept clear of grass. To a depth of 10 ft. below the probable water level they should be pitched with set rubble, or lined with concrete, and deeper still with loose stones. * * * The alga may be destroyed by the application of a chemical which need not impair the potability of the water. Salts of copper are algicidal even in very dilute solution. One part of the sulphate in 10,000,000 of the impounded water serves to extirpate the alga and to leave the water clear and limpid. It is usual in America to have three or four compartments arranged either in a rectangular form or in a crescent. From the rawwater basin the flow takes place over a dividing wall into the next, and so on. At the Paris waterworks of the Compagnie Generate the settling tanks are so constructed as to induce continuous and progressive sedimentation. The process here is three-fold. First, the path of the water lies through a series of narrow troughs disposed in zigzag form, in which the heavier particles are thrown down. Following the troughs comes a sequence of larger compartments, and finally decanting basins, and in both sides of these the alternating left and right and also up and down movements induce still further precipitation. Thus the combined effect of these slow and regulated and tortuous movements is that a more rapid sedimentation is conducted within a much less area and with far less depth than if mere stagnation were depended upon to produce a like result. Starting with a filterbed constituted of layers of gravel and sand, graded from coarse to fine upwards, and all thoroughly clean, we know that for a time it intercepts none of the microspocical germs and only an inconsiderable portion of the less minute organisms and of the suspended silt. As its operation continues, the ratio of suspended matters which escape to those which are retained decreases gradually, until after the lapse of a certain time it becomes almost a vanishing quantity. The reason is that sediment accumulates upon the surface of the sand, covers the surface of the granules in the top layer and loosely plugs the intervening channels. Distributed all over this film of silt are numerous minute forms as algae, desmids, bacterial. The upper surface of the sand is transformed into a loose and more or less permeable matting of vegetable filaments, mostly lying quite superficially and only invading the sand to a depth of a few inches. When this feltwork has attained to a certain consistency, the water which escapes through its interstices is denuded of sediment and even of bacteria to a very large extent. From the experience of the “roughing” filters in the Paris waterworks, it appears that a large percenetage of microbes— sometimes as much as 90 per cent.—fails to escape from the fourfold percolation of beds furnished with comparatively large particles, ranging in size from 1 in. in diameter to .02 in. The gravel strainer does not operate at all like a sieve. The pebbles and coarse grains are covered with a sticky film composed partly of dead, partly of living material. Particles of mud as well as germs impinge on the stones and tend to adhere. Hence, the same occurs in the ordinary sandbeds. By means of their slimy covering they constitute one of the agencies by which bacteria are removed. * * * A similar office is performed by the interlacing filaments. The minute strands, with their slimy coating, capture the bacteria and becomes the second agency for intercepting them. A second filtration, therefore, is likely to prove more efficacious in the treatment of refractory filtrates than would have seemed to be the case at first sight.
A third cause of the important check which the sand filter imposes upon bacteria is green algae which are energetic destroyers of minute germs, acting upon them, in all probability, by means ot the nascent oxygen which they evolve. Even matters in solution, more especially colloidal substances, are separable from water by contact with idterbeds. Precipitation of colloidal matter undoubtedly occurs in the filterbeds which receive the fluids from septic tanks, and the marked diminution of organic matter which is effected by certain types of filter designed to purify river waters is very likely attributable to de-solution of colloids.
What are the disadvantages of sand filters? These are chiefly three. First, the rate of filtration is slow. In the second place, the cleaning of a bed which has become clogged entails considerable expenditure of time and labor, and the contact of laborers’ feet with the filtering bed is objectionable. The top layer must be scraped away to a depth of about half an inch, and fresh sand substituted. Thirdly, there is a loss of time and of water after the run is restarted. In cold weather the filtering skin forms more slowly, and in summer the rapid increase of algae necessitates frequent cleaning. How to construct and manage slow-sand filters: 1.—The surface of the filters should be of such dimensions that there is a sufficient reserve to insure the delivery of the necessary supply under all conditions of the raw water. 2.—There should be two or more beds, so that cleaning may be done without wholly interrupting the work of purification. 3.—Each filterbed should be provided with a regulator, to control the velocity of the flow. 4.—The effluent should be readily available for sampling as soon as it leaves the filter. 5.—Bacteriological analysis should be made periodically, more especially after cleansing, after filtered water has been allowed to pass into the service reservoirs or mains, and under all conditions which appear to be abnormal. 6.—There should be experimentally determined the time that ought to elapse between the cleansing of a bed and the delivery of the effluent to the service reservoir. 7.—The thickness of the filtering layer should be sufficient for the local conditions and never less than 16 in. 8.—The walls and floor of the filter must he watertight. 9.—There should be appliances for the proper washing of the sand, and, wherever practicable, the contact of the workmen with the material should be avoided. 10.—Special regulations should be drawn up for the management of the filters in cold weather, more especially in times of frost. In Britain water containing more than 0.1 part of albumenoid ammonia per 1,000,000 would be regarded as suspicious. The presence of nitrites is also looked upon with suspicion in this country. Upland surface water should not hold more than 0.3 to 0.5 parts of nitric and nitrous nitrogen together per 1,000,000. Seine water contains on the average 1 part per 1,000,000 of nitric nitrogen and traces of nitrates. A knowledge of local circumstances bearing on the area of catchment is a sine-qua-non, if the analysis is to be made the test of potability, and we must rely on the bacteriological analysis.
In comparison with slow-sand filters, Jewell filters possess the following advantages: 1.— Capacity to treat very turbid waters and to remove color. 2.—Occupy a relatively insignificant area of ground. 3.—Freedom from risks of objectionable growths, and from the tastes and odors they impart. 4.—Rapidly and easily and cheaply cleansed, without risk of contamination by workmen. In the Bell filter the precipitant employed is the same, and the device for washing the sand when it becomes clogged is only a little more elaborate. Not only is filtered water forced upwards from below, but the revolving arms which stir up the sand are hollow, perforated tubes, which receive a further supply of pure water from above, by way of the revolving shaft, and discharge this in fine streams among the filtering material. The holes in the hollow wash-arms are protected, when not in use, by an automatic back-pressure valve. In the American filter, the head varies from 6 ft. to 14 ft.—i. e., it does not exceed a pressure of 7 lb. per sq. in. The Bell filter pressures of 30 lb., 95 lb., 120 lb. and 300 lb. per sq. in. are employed at Bury, Crewe, Polton and Linwood, respectively. But the filter works with a less powerful head, it needful. The consolidation of the granulated quartz in the Bell filters, owing to the high pressure. does much to perfect the efficiency of the apparatus. The Bell filter removes peaty acids, when these are present, and deals effectively with fine silt. As tested at Banbury, by Prof. Frankland, the filtrates were “of a very high degree of bacterial purity,” the bacillus coli being reduced from 10 per cm.* o 1 in 10 cm.*.
The Reeves (Mather and Platt) filter does not differ greatly from the Bell. Coagulants may be used, if so desired. The filtering material is crushed quartz, and the size is graded, a layer of finer particles being at the top and of coarser below. The filter is constructed to act either with pressure or with a deep head of water. In the pressure variety the form of the filtering cylinder is that of the frustum of a cone, with the wider end upwards. This design prevents water from slipping down the sides without being properly filtered. The filtered water is allowed to run waste for a few seconds only; but it is not state. that the filtrate is satisfactory as regards the bacterial content after that brief run. In this country, at least, there is an absence ot systematic testing of the filtrate after washing. Unless the greatest care is taken to exclude the first filterings until such times as experiment has shown that the filter will have resumed its normal efficiency, great numbers of bacteria may escape into the consumers’ supply. British opinion is that a filtering viscous skin, whether artificial or natural, is the agent which plays the most important part in the elimination of germs, so long as sand or crushed quartz is the material of which the bed is composed. With the Jewell filter about half an hour was considered sufficient for the skin of precipitated alumina to establish its proper efficiency.
Against the use of coagulants there are various objections. There is the question of expenditure, representing an outlay of $1.25 to $2.50 per 1,000,000 gal. filtered—$1.50, on the average, per grain per gallon, per 1,000,000 gal. It is recognised that it may be beyond the ordinary attendant of a filtering plant to determine the exact amount of coagulant that should be added under varying conditions of the raw water. Oxidium—the effective medium of the Candy filter—is a porous composition of iron, oxide, silica, etc., which exhibits properties similar to those of spongy platinum. Consisting largely of iron and silicates, oxidium absorbs oxygen from the air, and yields again to organic matters suspended or dissolved in the water under filtration. Living germs are intercepted as efficiently as in any of the filters yet described. Wet combustion proceeds rapidly as the water percolates. The oxygen is supplied from the pores of the oxidium, wherein it is occluded. It is gradually used up as filtering goes on, but is easily renewed by emptying the cylinder and admitting air. By a simple means a large volume of air is compressed into the dome of the cylinder, so that the descending water is well aerated. In the effluents of the Candy filters the purification in general has been shown to be admirable. Practically, there is no escape for pathogenic bacteria; albutnenoid ammonia is reduced to an insignificant amount; nitrification makes great progress in passing through the layers of oxidium; iron salts in solution are wholly got rid of.
The Puech filter is strictly an adaptation of the older system, in which the sediment-charged water is subjected to graded filtration. There is a series of beds in which the filtering substances range from coarse to tine. First is a battery of “degrossisseurs” or roughing filters, with four compartments. In the first, the whole filtering medium is pebbles of the size of walnuts; in the second, they are of the size of hazel nuts; in the third, of beans; in the fourth, of peas. The depths of the layers are respectively 12 in., 14 in., 16 in., 16 in.; but these thicknesses may be varied with the character of the water treated. The water finds its way more easily through the coarser materials; wherefore, the area of the first “degrossisseur” is smaller than that of the second, and so on. Thus, the speed with which the water percolates decreases rapidly from one compartment to another, being five times slower in the last than in the first, and the suspended impurities are arrested throughout the whole depth of eacli bed. Dr. Kemna stated that the action of the Puech-Chabal gravel-strainers was very effective as a preliminary treatment of river water; that most of the matters in suspension were removed : that there was a great diminution in the number of microbes; that there was a slight fall in the percentage of organic matter and albumenoid ammonia, and a considerable reduction of the free ammonia; the diminution of microbes and of suspended matter is largely independent of the speed, while the chemical reduction—loss of ammonia and organic matter—is inversely proportionate to the rate of filtration.
From the gravel strainers the effluent passes on, by cascades preferably, to the socalled pre-filter, the upper layers of which consist of coarse sand, thus promoting quicker passage of the water. Last of all comes a bed composed of materials very similar to those in the slow-sand filter. In Paris, the efficiency of the Puech-Chabal system is very marked. The Seine water at Nanterre, after treatment with this system, is from week to week superior in freedom from bacteria to that obtained by other processes at Neuilly, Choisy le Roi, and Saint Maur. At Magdeburg, the Puech filters were applied to the purification of the intensely polluted water of the Elbe, and its success has been beyond question. But, while in the Puech filter plant no film forms upon the finishing filters, the water is in summer loaded with vegetable matters which have to be scummed away by manual labor. No expert is satisfied that the filtering film can safely or advantageously be dispensed with, lienee, at YVaelhem, there is to be a considerable modification of the original Puech process. There will be “degrossisseurs,” but no prefilter. The slow-sand filters will be left to do their full share of the work. Neither will there be any aeration from bed to bed; but the strainers will be covered to exclude light and prevent growth of vegetable organisms.
Ozone, as is well known, is best prepared by the silent discharge of high-tension electricity through perfectly dry air. There are many devices lor ozonising air, most of them being adaptations of the Siemen’s ozoniser. That at Wiesbaden was designed by Messrs. Siemens and Halske. The air drawn into the ozoniser is dried by contact with a hygroscopic substance, as CaCL, or, preferably, by refrigeration in a chamber cooled as in the freezing machines. The voltage required to operate this instrument is about 80,u00, and the generator absorbs 1-h.p. The amount of ozone produced varies from 13.5 to 27 grams per hour, the rate of production depending on the dryness of the air which enters the apparatus. This quantity of ozone will purify from 2 200 to 4,400 gal. Nine ozonisers are in operation; thus a large volume of water is being constantly dealt with. The water supply at Paderborn is also purified by ozone, and 12,000 to 16,000 gal. per hour are passed through. The ozone is applied to the water as it descends through gravelbeds, the current or ozonised air being forced up from below. An ozonising instal lation has been set up at Ginnekin in Holland to purify the water drawn from the Mark. Nor mally this stream is discolored, and the number of germs per on.’ is generally about 6,000. its course lies through fields frequently manured from the farmyard. The water from the intake is first passed through filters. Lift-pumps raise the filtered water to a tank—placed at the top of the sterilising tower, with a capacity of 45,000 gal.—ordinarily the pump delivers 4,500 gal. per hour. The whole purifying apparatus is housed in the tower. Electricity is obtained from a local installation, and the 100-volt current is trails formed to operate at a pressure of 65,000 volts. The ozonisers are of the Schneller pattern, and do not differ in principle from those already described. The outer electrode is hemi-cylindrical in shape, and is kept cool by water circulating in a cast-iron jacket. The inner electrode is the insulated one, and it is of the same form as the outer one, but of less radius. Charged with its load of the purifying gas, the air is forced into the sterilisers, where it encounters the descending stream of water. The sterilisers are cylindrical tubes built in short sections. The inner surface is enameled to protect the metal from oxidisation. At each junction of the sections there is a celluloid sieve, resting upon a metallic grid at right angles to the axis of the cylinder. Ozonised air enters from below and rises with much agitation through the down-coming water, and forms a cushion under each sieve, so that the line jets of liquid are thoroughly exposed to the action of the ozone. There is. indeed, an adequate scrubbing of the water at each sieve, air struggling to pass up and the fluid to gravitate downwards. The water is made to pass through two cylinders in succession, and it issues in a limpid stream, to all intents and purposes sterilised. For, out of many samples tested, one-half showed no growth from 1 cm.* after three days, and 20 per cent, no sign of life after six days. Of the remaining 80 per cent., the most developed only one or two colonies of harmless bacterial. and in no case did the number exceed a dozen. Oxidisable organic matter was diminished by about one-half. The cost of working at Ginnekin is about $9.50 per 1,000,000 gal.
Costly as this process of purification may be, it is probably one which is well worth the sacrifice of a little money when it furnishes unexceptionable drinking water from a source like the river Mark, whose waters have be*,n justly described as discolored, and execrable from a hygienic point of view. The experience at many stations with ozone purifiers indicates that a fair proportion of the ozone goes to waste. Different waters, of course, make different demands on the purifying agent, and it is recommended that tests be made to determine the minimum quantity of ozone that can with safety be applied, not only in the interests of economy, but to prevent traces of the ozone being carried to the consumer. The Howard-Bridge apparatus effects this useful purpose, and, considering that it takes one centime to produce a gram of ozone, there is good reason to be as sparing as possible with this, the most expensive of all purifying agents in ordinary use. The Howard-Bridge system has the further advantage over installations which use an air compressor to force the ozone into the sterilisers, that it dispenses with that portion of the apparatus, and causes the flow of the water to suck in the ozone. Ozone acts detrimentally upon lubricants, so that the piston of the compressor has to lit very closely, in order that they may be dispensed with. Suction performs the whole duty ot applying the ozone to the water in the Howard-Bridge plant, and the claim that by it the purification of water can be accomplished more cheaply than by slow-sand filtration awaits more complete proof.
In the de Fries system of ozone purification, which has been tested at St. Maur on a large scale, the arrangements are somewhat similar to those at Ginnekin. The sterilisers are vertical, inwardly enameled, iron cylinders, divided into many sections by celluloid sieves. The water and ozonised air enter together at the base, and the outflow is at the top. The plant can treat 6,000,000 gal. per day at a little more than $10 per 1,000,000 gal., including outlay on capital account—considerably lower than the cost at the German towns already referred to, where the corresponding outlay to cover all expenditure runs to a figure twice as large.
The Vosmaer is a special device for overcoming some difficulties in insulation, and in the es tablishment of a purely silent discharge. Spark ing leads to the formation of oxides of nitrogen, which would be harmful both to the water and to the apparatus. Vosmaer forms his ozoniser of a series of parallel tubes, each of which contains as electrodes, at opposite sides of the inner circumference, two strips of metal, supported upon porcelain insulators. These strips of metal have saw edges pointing inwards, and the air sent through the tubes is subjected to the silent discharge from these saw edges of the strips which are maintained at a very high potential differ once.
Satisfactory results have been obtained with widely differing “concentrations ‘grams per in’, of air—of ozone, and for each installation there should be determined the most economical value of the concentration. This will be a ‘‘constant” of the plant when the data are known. For field filters it is considered best to have a large content of ozone per m*.—say, 3 grams. This means that the air must be drawn along more slowly, or the current increased. More ozone, on the whole, is produced, if the air current is rapid; but the concentration is lower i. e., the same horsepower might produce a sum total of—say, 48 grams of ozone, with a concentration of 0.8, while, with a concentration of 3.5, it could gen crate only 28 grams in the same time. Experiment will determine the best conditions of work ing. When a mixture of chloride of iron and bleaching lime is added to water holding carbonic acid or carbonates in solution, a precipitation takes place, and at the same time a coagulation is formed, which carries down the suspended matters. Free chlorine acts as a bactericide. After the application of these chemicals, the water being led to filterbeds is easily freed from suspended material, owing to the presence of the coagulant. Residual chlorine is eliminated in settling tanks. The amount of chemicals added to the raw water is small—namely, of chloride of iron 8 parts per 1,000,000, and of chloride of lime 0.5 part. Oxide of iron is the coagulant, and, if care is exercised in regulating the dose of the chemicals, the filtrate will contain no trace of it. Chlorine is not so easily got rid of, for it is soluble in water to a very considerable extent, and its elimination ultimately rests upon the formation of chlorides. There is an experimental plant on this system at the Paris waterworks capable of treating 1,000 gal. per hour. The results are stated to be very satisfactory; but the cost is relatively high, being about 1.5 cents per 1,000 gal.—that is, over $15 per 1,000,000.
Of all preciptants the cheapest is iron, suitably brought into contact with the water in a metallic state. In the presence of air and salts of lime metallic iron rapidly forms a proto-salt, which first dissolves in the water and then, becoming oxidised by further aeration, develops the coagulant which carries down organic matter and the silt. This is the Anderson process, and the most imnortant installation is that of the Compagnie Generale des Eaux at Paris. A horizontal cylinder is made to rotate by gearing. The inlet and outlet pipes are directed along the axis, and are connected to the cylinder by watertight collar joints, which permit of the cylinder rotating. The cylinder itself is built of iron plate, and within are many laminar projections, which serve to raise up the fragments of iron and allow them to fall into the water from above, so that a quick oxidation results. The rotation is slow, one turn taking from two to three minutes. Air is forced into the space above the water by way of curved tube and the direction of the air current follows that of the water.
The flow of the water is so regulated that about 3½ minutes are required to traversse the cylinder, and in that time some It grams of iron per cubic metre finds its way into solution. Small as the quantity is, it serves to coagulate the silt and organic matters suspended in the water. It is averred that there is a diminution of dissolved organic matter from the moment the water escapes from the cylinder. Further, it is held that any colloidal alumina in the raw water is coagulated by the salts of iron, when these pass into the ferric state. On leaving the rotators, the water undergoes a preliminary treatment in precipitating basins and decanting compartments, and finally reaches a finishing filter, where the residual iron oxide helps to form a “felting,” which intercepts bacteria and sediment. The speed of filtration is greater than in British slowsand filters, and averages 8 in. per hour. Nevertheless, the purification effected is superior to that of the average sand filter, for the action of the iron salts is in many ways a valuable supplement to the ordinary process; 09.8 per cent, of the bacteria are removed from the Seine waters, and though from too to 2(H) germs per c.c. are >till left, they do not seem to be of a harmful kind, for the district supplied by this installation is fairly free from enteric and similar epidemics. The cost of the whole purification process, including interest on capital, works out o $5 per 1,000,000 gal.
The main question at issue would appear to be whether the sand filter is to be displaced by mechanical substitutes in future water undertakings. l’he latter have many advantages, and are on the whole less likely to permit the escape of unsound water, l’he Puech system indicates a simplification of the method of working, which will possibly be received with increasing favor, especially, if the experiments being conducted on the continent of Kuropc are brought to a successful issue.
The question of mechanical sand washing is attracting much attention at present ; but, taking a broad view of the facts which have come to light in connection with the true action of the filtering skin, it would seem to be a debatable point whether a thorough cleansing of the sand forming the top half inch of the bed is after all so very desirable. Clean sand acts as a strainer merely, and has no decided power of trapping bacteria until it has become coated again with viscous matter, silt, desmids, algae
i. e., with the very matters that have been so carefully washed away. No one has as yet attempted to show that the fresh coating is in any way more pure or “refined” than the old one. Is it not possible that there is a happy medium in the process of sand washing, and that, in clearing a filter that has become clogged, it may be sufficient to break up the growths of algae and wash off part of the dirt? If there is reason in this view of the case, the problem of sand washing in open sand filters would be simplified. Raking of the top layer followed by a backcurrent of filtered water would do all that is necessary. Much time might be saved by sparing a portion of the viscous coating of the particles, tor the filtering skin would be able to re-establish itself more quickly Let it be remembered that it is not clean sand but dirty sand that filters in the true sense of the word. Hence, those who insist upon thoroughly washed sand, and, also, those who put much faith in particular qualities of sand from special localities, have probably overlooked the essential point.