FIG. 12.-Regulator designed by Lindley for the Warsaw filters. Fig. 13 shows the method of regulation devised by Gill for the Tegel Works of the Berlin water supply. The outlet from the middle chamber is through a weir; and the depth of water on its crest, and therefore the discharge, is indicated by the height of the float read on the scale a. This is kept constant by means of the gate. The corresponding loss of head is shown by the difference of the readings on scale b. Keeping the water in the filter always at the same level, a constant rate can only be maintained by the gradual falling of the level in the right-hand chamber and a consequent wider opening of the gate. Fig. 14 shows the principle used in the new Hamburg filters. A similar method was recommended by Kirkwood for St. Louis. FIG. 13.-Regulator used in the Berlin (Tegel) works. The scale on the right reads downwards, and its zero corresponds to the level of the surface of the water on the filter, which must of course be kept constant. In the first chamber is a float with a pointer attached. The reading of this pointer on the right scale evidently gives the loss of head. The reading of the same pointer on the other scale gives the corresponding rate. This is accomplished in the following way: The outlet of the first chamber is through the weir, which is movable in a vertical direction. The smaller scale is fixed to this weir as shown in Fig. 14, so that the distance between the crest of the weir and the zero of the scale is the same as that between the pointer and the water line of the float. Both loss of head and the rate may therefore be regulated by lowering or raising the weir. As to the limit beyond which the loss of head should not be allowed to go, the general opinion seems to be that it should not be greater than the depth of water on the bed, though the Lawrence experiments have not shown any bad effects from exceeding this limit. As a general thing it may be stated that, everything else being equal, the higher the rate the poorer the quality of the filtrate. FIG. 14.-Regulating apparatus used in the new Hamburg filters. But with fairly low rates this deterioration is slight, so that entirely satisfactory results can be obtained with rates up to three or three and a half million gallons per acre per day. Probably higher rates could be safely employed if very great care were exercised in the operation of the filter. The relative effects of high and low rates from a financial point of view are discussed further on. SCRAPING THE FILTER, When the clogging of the filter bed has become such as to require a loss of head greater than the prescribed limit, the inlet is closed and the water allowed to drain away until it has sunk some distance below the surface of the sand. When this has become sufficiently firm, workmen enter the bed with planks, wheelbarrows and broad flat shovels. With these they carefully remove the surface layer and pile it up in little heaps, which they afterward remove with barrows. The depth removed varies from one-half to one inch, and averages about eight-tenths of an inch. The surface of the sand is then raked to loosen up the packing caused by the boots of the workmen, and after smoothing down any irregularities the planks are removed and the filter is ready for another period of service. The refilling begins from below by admitting through the underdrains filtered water from another bed in action. The object of this is to drive out the air from the pores of the sand, where its presence in the form of bubbles would cause considerable unnecessary friction. When the water has risen a few inches above the surface of the sand the lower connection is shut off and the refilling is completed by means of the surface inlet. Before filtration proper begins the water should be allowed to stand on the bed for several hours, or the first million gallons or so should be wasted. The amount wasted can be reduced by beginning the filtration at a low rate and gradually increasing it to the maximum. When the scrapings have reduced the sand bed to the minimum allowable thickness the total amount removed, which has in the meantime been thoroughly washed, is replaced at one time. Before doing so the surface of the permanent layer, which is never removed, should be loosened up by being spaded over to a depth of six inches or so. If this is not done there is a liability of subsurface clogging at its junction with the clean sand. When the filter is started again it is, except for the permanent layer, in the condition of a new filter, and so requires extra care in operating it, and the filtrate should be wasted for a much longer time than is required after the scrapings. Piefké, of the Berlin water works, places this period at six days. Considering the labor necessary and the time the bed is out of use, this replacing of the sand is an expensive operation, and should not occur oftener than can be avoided. In most plants the usual period is about once a year. SAND WASHING. Sometimes it is possible to obtain new clean sand at less cost than is necessary to wash the old. But this is rarely the case; hence an important part of the equipment of a fair-sized filtration plant is the apparatus for the washing of the sand. The simplest |