Measurements recently made of flow in the infiltration galleries of the Des Moines Water Company are of interest as presenting a somewhat novel application of the chemical method, and showing the adaptability of this method to a situation where satisfactory discharge measurements by the velocity area method are not possible. The chemical method of measuring discharge has been developed principally as applied to stream flow, particularly of mountain torrents, and for turbine testing. The extreme refinement used to secure high precision in turbine testing is apt to obscure the usefulness of this method in many cases where great precision is not required, and where it can be readily and cheaply used to advantage. It has a distinct field for instance, in measuring flow in submerged conduits inconvenient of access, or in which the velocity is so moderate as to be practically outside the range of the current meter, as in the case of the Des Moines infiltration galleries.

Description of Galleries.

The total length of gallery completed and in use at the time of the test was 8,480 feet Manholes are provided about GOO feet apart. These galleries are built at a depth of from 20 to 25 feet in the very extensive and uniform coarse sand deposits of the Raccoon river valley, and are submerged five to ten feet at low water. The new part, 5,160 feet in length, is built of concrete rings and has a cross section of 12.6 square feet. The older galleries are rectangular and vary from 15 to 20 square feet in cross section. When yielding at the rate of the present maximum pumpage, 14 million gallons per day, the average velocity at the lower end of the main gallery is only 0.9 feet per second, the velocity decreasing progressively to practically zero at the far end.

Purpose of Test

In planning for the future extension of this supply, it was important to know the amount of variation in yield between different parts of the gallery from these causes. When the subject came up for study last year it was decided to measure the flow by chemical method.

Method of Insuring Flow.

The test consisted in feeding a solution containing a known amount of salt (sodium chloride) at constant rate into one branch of the galleries near its upper end, and titrating for chlorine in samples of the gallery water collected at all manholes between the dosing point and the pumping station, including a sample at the pumps, just before the water entered the venturi meter. The chlorine determinations were expressed in parts per million. These determinations, corrected for normal chlorine, indicated the dilution at any given point, of the salt fed into the gallery and gave a measure of the volume of flow at the corresponding point. The equation for flow at any point is:

C—c” Q=—C C”-c’

in which Q is the gallery flow, q is the rate dosing, C the concentration of the dosing solution, c c c’ the normal concentration of chlorine in the

•Assistant Engineer with Alvord & Burdick, Consulting Engineers, Chicago, Ill. Abstract of paper read before Illinois Section of American Water Works Association, March, 1917. gallery water, and c”” the concentration of chlorine in the gallery water at the sampling point, after dosing. The volume of the dose was so small in comparison with the gallery flow, that it was ignored except to keep it constant. In other words the salt used in making up the solution was carefully measured and was always added to the same quantity of water, but the exact amount of water in the solution was not determined. Having determined the rate of flow at each manhole, the difference between the flows at successive manholes, gives the total infiltration rate of the gallery intervening, and this divided by the length, gives the average yield per unit length. This enables us to compare in rate of yield the several lengths of the gallery as divided by the manholes. The ideal condition for this test, of perfectly uniform draft on the galleries could not be secured. Des Moines has a direct pressure system and the pumpage fluctuates more or less continually. It was not practicable to overcome this difficulty by taking simultaneous samples at the several sampling points, so the test w’as continued during the greater part of the day, and the result used over selected periods of comparatively uniform pumpage, as shown by the venturi meter chart. In this way the results are based upon observations made under conditions approaching as nearly as possible the ideal of steady uniformly accelerated flow, that would pertain under constant draft. A check on the total yield of the galleries computed by this method w:as furnished by the venturi meter.

Details of Test.

Commercial sodium chloride was used for this work. The solution wras prepared by dissolving a definite weight of salt (57 pounds) in about 35 gallons of water, as needed in each of two barrels from which the solution was drawn alternately. Each barrel emptied in about thirty minutes. The barrels were piped into an orifice box, the water level in which operated a float valve controlling the barrel discharge. A 3-16 inch orifice was provided under about four-foot head, the idea being to minimize the effect on the discharge of the slight fluctuations in head that were unavoidable with the float valve used. A 1-inch drop pipe with funnel at the upper end caught the solution below the orifice and delivered it to the center line of the gallery. The dosing apparatus was not entirely satisfactory. Its principal defect was the imperfect arrangement for maintaining constant head on the orifice, and absolute uniformity of feeding could not be secured with it. Some trouble was also experienced with small splinters in what appeared to be perfectly clean salt, becoming lodged across the orifice. This was overcome by straining the solution through cheese cloth as it entered the orifice box. This trouble with the feeding apparatus was the principal source of error in the work. The quantity of salt used and the size of the orifice were predetermined to approximate the desired chlorine determination in the final sample at the pumping station. The dose was applied in the case of the main gallery test at a point 600 feet from the upper end, with satisfactory results. The average velocity at this point during the test was about 0.113 feet per second. In testing the shorter gallery, the dose was applied only 350 feet from the dead end, and the results seemed to indicate that the velocity at this point (about 0.03 feet per second) was too low for very satisfactory results. Samples were collected at intervals of about one hour through the test, in 200 cc bottles. The samples were titrated with silver nitrate.

Result of Test.

The infiltration rates under constant draft on the galleries varied from 0.20 to 1.25 gallons per minute per foot of gallery. The low rates were found in those sections of the gallery not adjacent to the river. The test was carried out in co-operation with Robert F. Kinnaird, superin* tendent, and E. T. Kirkpatrick, chemist with the Des Moines Water Company, and was greatly facilitated by the admirable laboratory equipment of the company.

No posts to display