< 262-7538 i>TAT£ WATER SURVEY DIVISION LIBRARY CORY 06T 5 IS6S I'ns^rrrp- ^30 - SL U5XI ARS- H l- l I 8 An Electric Analog For Computing Direct Surface Runoff April 1966 ARS 41-118 Agricultural Research Service UNITED STATES DEPARTMENT OF AGRICULTURE ■ CONTENTS Page INTRODUCTION. c 1 TERMINOLOGY . X ANALOG DEVICES. 2 WATERSHED ANALOGY. 3 TEST MODEL OF ELECTRIC ANALOG . 8 Electric Analogy . 9 Calibration of Analog . X7 Input to Analog. 19 Operation of Analog .. 22 EXAMPLES OF HYDROGRAPHS . 23 CONCLUSIONS. 23 ACKNOWLEDGMENT. 27 LITERATURE CITED. 01 1 / AN ELECTRIC ANALOG FOR COMPUTING DIRECT SURFACE RUNOFF By 2 / J. Marvin Rosa INTRODUCTION The purpose of this study was to demonstrate the feasibility of an electric analog device for routing runoff to generate hydrographs on experi¬ mental watersheds,, A small, inexpensive analog model was desired for quickly routing runoff from hourly records of precipitation on headwater areas,, Even the most simple routing through multiple stages, as if the watershed and channels act like a series of reservoirs, is adequate if time intervals are very short. The use of many increments of time and storage would complicate any long-hand or graphical computation previously known, but this is no problem with fast electronic devices. A rapid solution could be obtained by utilizing the analogy between the flow of electricity in a passive network of resistors and capacitors as used in oil-reservoir simulations and soil- moisture movement studies. A simplified, schematic explanation of the hydrology of watersheds is first outlined to show that they function like a series of reservoirs. After derivation of the analogy to a reservoir in electrical units, a description of some construction and of calibration details completes the discussion of electronic equipment. Finally, the operation of this analog model is shown to simulate the time-consuming routing computations previously required. Some actual hydrograph computations illustrate the applicability of this model at a cost of about $100. The term analog "model" is more descriptive of this network for solving the storage function by the use of common radio parts than is the term "computer." TERMINOLOGY Symbol Unit Electrical I , g , = Current (ma = milliamperes)--- amperes L^ , g , 3 = Inductance--- Henrys R , , = Resistance (K = 1,000 ohms) - ohms V , 2 , 3 = Voltage- volts C , 2 , 3 = Capacitance (/if. = microfarads) - farads Qc = Charge on capacitors - coulombs T s = Time of storage-*•- seconds 1 / Soil and Water Conservation Research Division, Agricultural Research Service, USDA, in cooperation with the Idaho Agricultural'Experiment Station. _2/ Research hydraulic engineer. Soil and Water Conservation Research Division, Agricultural Research Service, USDA, Moscow, Idaho. 1 Symbol TERMINOLOGY--Continued. Unit Hydraulic Ii Io Qo Qt t e X k = Inflow - = Outflow - = Initial discharge --- = Discharge at time - - Time - = 2.7183 = Relative effect of inflow and outflow, when = "Muskingum" routing coefficient Hydrologic = Direct storm runoff - = Accumulated runoff for periods - = Increment of runoff --— = Mass direct runoff - = Storm rainfall - = Accumulated rainfall for periods = Rainfall for periods - = Mass precipitation - = Storage - = Recession constant c « f . s • C . f . S . c.f. s. c.f.s. seconds inches inches inches inches inches inches inches inches inches ANALOG DEVICES Analog devices depend on--and are named for--the use of analogies, which are similarities of properties of relationships. As stated by Rogers and Connolly (13 )' simulation results whenever a physical model is represented in such a way that its characteristics, parameters, and behavior can be easily identified, adjusted, and studied as accomplished in the use of analog models. The term "analog computer" refers to operational amplifiers, which are as¬ sembled into a computer for general purposes. Such a computer must be capable of (1) multiplying constants, (2) adding variables, (3) multiplying variables, (4) integrating variables with time, or (5) generating a function of a variable. All analogs could be called computers, but all computers do not operate as analog models. One major type of computer is called digital--which deals in numbers according to an ordered sequence of simple arithmetic operations. It computes by repeatedly refining an approximation so that its accuracy is potentially unlimited. Another type of computer is some sort of physical model that is set in operation to generate a solution. This analog unfortu¬ nately is limited in accuracy by the physical elements of the system. The most direct analog "model" (not usually called a computer) performs like a scale model under experimentation in a hydraulic laboratory. It warrants being in a class by itself--the passive network. 3 / Underscored numbers in parentheses refer to Literature Cited at the end of this report. 2 As utilized by Harder and others (6), electric analogs are models of a dynamic system where flow networks can be integrated by instantaneous electri¬ cal currents. Such electric analog models have been used by the petroleum industry, according to Scheidegger (14), and should have a similar application to ground-water problems. Problems involving ground-water behavior, as studied by Skibitzke and Brown,—' can be solved by using electric analogs where solutions are unattainable in any other way. There is no limit to the application of such electric analogs to hydrologic problems, according to Tribus (15), since it is possible to construct analogous electrical networks for every dynamic system. Electric analogs for routing streamflow to generate hydrographs on small watersheds are a recent development. Linsley, Foskett, and Kohler (9) in 1948 first utilized an electronic routing machine to route floods from basins of up to 15,000 square miles. In 1952 Paynter (11) demonstrated the analogy between flood routing by admittance methods and transients in electrical transmission lines. Glover (4) in 1953 presented an electronic circuit for an analog model which provided for the square-law resistance. And in 1956 Rockwood and Hildebrand (12 ) described an electrical circuit for multiple-stage storage routing on large rivers. Messerle (10 ) first reported an electronic high-speed simulator developed at the University of Melbourne in 1953. Japanese developments were described in 1955 and 1957 by Ishihara and Ishihara (7) and Ishihara (8). At about this same time, foreign articles by Dziatlik (3) and Halek (5) mentioned analog apparatus. By 1958 electric analogs were mentioned in the literature of most countries. WATERSHED ANALOGY Basically, the watershed analogy can be represented as an exchange of water between storage and flow or losses, the time pattern of precipitation is a variable input following the trace of a recording gage. The first losses are intercepted by vegetation, which is shown as part of evapotranspiration in figure 1, The first increment of runoff could be called channel interception or the rain falling directly on the wet surface area of the open channel. Direct evaporation from water surfaces could be extracted here and combined with other losses. Other increments of overland flow follow each other down the channel after separation at the soil surface by infiltration. All increments of runoff to this point are routed through storage; First through detention on the rough land surfaces or humus (A 0 ) layer; then as channel storage, including bank storage, less other riparian losses downstream. These losses and storages are shown diagrammatically in figure 1. Moisture entering the topsoil (A ) is available for percolation or is eventually extracted by the process oi evapotranspiration in proportion to temperature or other controlling factors. As far as immediate direct runoff 4/ Skibitzke, H. E., and Brown, R. H. Analysis of hydrologic systems. U.S. Geol. Survey unpublished paper, 10 pp. 1961. 3 is concerned, the characteristics of the detention volume in the soil profile determine the percolation rate of free water into the next incremental volume of storage, shown as subsoil (B 1 ). Here again, retention is trapped and re¬ leased gradually either to direct runoff laterally if there is a limiting layer (Rg), or to percolation downward into the rock mantle (C ), as shown in figure 1. 1 Rainfall c o Overland Detention a, \ Flow \ ^ 1 hour C7 -**- s C 9 - . 500 /if, 200 WVDC capacitor 40 /if, 600 WVDC capacitor 25 K ohm potentiometer - 10 K ohm potentiometer - 43.20 2.50 2.50 3.75 , + s (15) where Q 2 and are the total runoffs (accumulated runoff) after the second and third time periods, respectively. The runoff increment for the third time period is given by: p + p + r- s (Pi + p s +p 3 ) 1 - r P + P . s i (16) J (P x + P s + p 3 ) + s J L 1 s ( Pl + P 3 ) + s J reduced to: % = p 3 - s (p,+ P2 +p 3 ) + S(p x + P 2 ) (IV) (P! + P 2 + P 3 ) + S (p x + p 3 ) + s In developing an electrical analogy to equation 17, the assumption was made that the currents labeled I in Branch 1 and Branch 2 of figure 12 are equal, constant, and independent of any changes of resistance, power supply voltages, or brush positions. The reason for the constant current devices is to insure this. Another assumption was that both resistances labeled = £p are zero. Referring to the circuit in figure 12 and using equations 11 and 12, the equivalent resistance between points A and D is found to be R eqAD R + (%+ Re) R (R 1 + Bg) + R and between points B and D to be ^BD (%+ Re+ Rj) R («!+ V + R (18) (19) Applying equations 10 and 18, the potential difference between points A and D is found to be Vt = I [~IL + (Ri + Rg) R 1 L 3 (R 1+ Rg) + R J and between points B and D to be v 2 . i r (R ^ + w R i L (R l + Rg+ ft,) + R J 21 As can be seen from equations 10 and 19, the potential difference between points A and B, which is one of the output voltages of the computer, is V x - V Q = = i [% + («1+ 1*8) R (%+ Bg) + R ]- X [- (Ri+ Be+ %) R (R x + FgH- Rj) + R ] (20) This can be written as (8!+ Rg) R («!+ Eg) + R (R x + Rg+ R;) R (R 1 + 1^+ Rj) + R ( 21 ) If the pairs of resistors labeled R^^, Rg, and , and R are chosen such that R = Pi» "Pg* ^ = p 3 , and R = S, equation 21 may be written as = 1 [P 3 + (Pi + P 3 > S (P!+ P 2 ) + S . (Pi+ p s + p 3 ) s -| ^ P l + P S + p 3 ) + S ( 22 ) An examination of equations 17 and 22 shows that V 1 - V 2 is proportional to - Q 2 . This fact is the basis for the design of the computer. The above derivation is for the brush positions shown in figure'12. If, for example, the brushes labeled C and A are on commutator segments 5 and 6, respectively, in Branch 1, and if at the same time the brush labeled B is on commutator segment 6 in Branch 2, then the output voltage is proportional to Q6"Q5® Sweeping the brushes at a uniform rate will, therefore, produce a voltage output which is a step function. The height of the individual step is the runoff increment, and the width of the step is the time unit. This is the same as the time unit over which the inputs are averaged. In deriving equation 22 it was assumed that the resistances labeled Rjl = £p in figure 12 were zero. This would be the case if the analog were being programmed to start at the beginning of a rainfall. When the computer is programmed to start at a time other than the beginning of the rainfall, the two resistances Rjj are set equal to the accumulated P at this time. This feature also extends the total rainfall time for which the computer can be us'ed. Operation of Analog Operation of the analog model is very simple for reproducing recorded hydrographs from small watersheds. Given the time distribution of rainfall for routing of excess rainfall into a downstream hydrograph, output can be computed from input by the proper choice of only three values: 1. The potential storage for solution of equation 9. 22 2. The time of storage , T s , to determine the RC value so that resistances can be set on the analog. 3. The number of stages or the lag time which may be estimated from the length of channels, slope, area, etc. By plugging a recorder in the output jacks of the analog, the computed hydrograph is quickly obtained and after a few trial-and-error solutions the best choice of arbitrary values is soon indicated. Only one choice of RC values, number of stages, and potential storage will best duplicate the avail¬ able hydrographs in general shape and time of occurrence. These values can be said to have scaled the analog so that it is now a model of that particular watershed. Thus, any other rainfall record could be converted into a hydro¬ graph from that watershed or any other ungaged drainage with similar soil, cover, geology, drainage density, slope, area, etc. EXAMPLES OF HYDROGRAPHS The practical application of this analog model can be illustrated by a sample computation (table 1) and by typical hydrographs from the South Fork of the Palouse River from Moscow, Idaho, to Pullman, Wash., as shown by figures 13 and 14* Hourly rainfall values from recording gages on the watershed were first tabulated, then converted to runoff (i.e. excess precipitation) by use of equation 9, in the form of EQ = EI4-S, after estimating the potential storage (S). Finally, the increments of runoff (AQ) were entered into the analog after conversion to milliamperes and routed through a number of stages and time of storage so that the final downstream generated hydrograph agrees with the actual. For reproduction of hydrographs on tributaries of the Palouse River, a value of T s ranging from 5 to 10 hours and the use of 5 to 10 stages are required for drainage areas of 34 to 84 square miles. The physical signifi¬ cance of the time of storage, T s , may be characteristic of land form, soil mantle, and cover type. The number of stages indicates the length of channel-- that is, the number of routing reaches. CONCLUSIONS The application of electric analog methods to the solution of a complex network of channels and reservoirs is relatively new in the United States. Because the future of these devices seems promising for both ground water and surface routing, a test model of multiple stages of linear reservoirs was assembled simply and inexpensively. Recommendations resulting from testing this model analog are: 1. The number of stages should be increased to 10 or more. 2. Leakage currents could be decreased by greater capacitance or by more expensive capacitors. 3. The time of storage should be varied exponentially. 23 TABLE 1.— Analog hydrograph computations on Palouse River , Fourmile Creek at Shawnee, Wash., April 20, 1937 Watershed Analog Hydrograph Time t Rain Runoff-^ Input—/ Output—^ Computed^/ EP 2Q • Z\Q lX 50 AQg -f 50 In. In. In. Ma. Ma. In./hr. 0.00002 0.00 0.00 0.0000 1000 0.01 0.00002 .00017 .01 .00 .0000 1100 .03 .00019 .00034 .02 .00 .0000 1200 .05 .00053 .00082 .04 .00 .0000 1300 .08 .00135 .00165 .08 .00 .0005 1400 .12 .0030 1500 .17 .0060 .0030 .15 .01 .0007 s .0022 .11 .02 .0009 1600 .20 .0082 S) .0066 .33 .03 .0011 1700 .27 .0148 .0166 .83 .04 .0013 1800 .40 .0314 .0149 .79 .07 .0019 1900 .49 .51 .0463 .0037 .19 .11 .0026 2000 .0500 .0020 .10 .18 .0041 2100 .52 .0520 —— .. .24 .0053 2200 •• •“ m mm .. .27 .0059 2300 * •" mm mm .27 .0059 2400 -- -- .24 .0053 1/ EQ EF 5 " EP + 4.68 2/ Factor for full-scale deflection of one milli-amp. = 50. 3/ 10 stages at 2.5 seconds = 25 seconds, RC = 2.5 seconds. 4/ AQg/50 + 0.0005 in./hour. T s = 5 hours. 24 Discharge -Whr. Ppt . . ln . /hr> Figure 13.—Hydrographs: Fourmile Creek at Shawnee, Wash. (71.6 sq. mi.) 25 Discharge - in./hrs. Ppt. - in./hr Figure 14.—Hydrographs: South Fork Palouse River at Pullman, Wash. •(84.4 sq. mi.) 26 The analog should be thought of, not as a mathematical computer, but as a "tool to think with" in terms of the physical system. The operation of the analog model is not based on man-devised algebra, as is that of the digital computer, but is governed by the physical laws of the electrical circuits. An outstanding advantage of the analog model is that it can be con¬ structed of standard radio parts and can be operated by personnel with only routine engineering training. The use of the analog model is a never-ending game; there must always be feedback between the user and the system. Some advantages of this multiple-stage, storage-routing analog are: 1. Flexible operation to meet variable storage-discharge relations. 2. Application to all routing where analytical or graphical solutions were previously required. 3. Success in routing through channels or basin storage for rain or snowmelt. 4. A saving in operating time by replacing tedious graphical methods, thus making possible many trial routings that otherwise would not be attempted. 5. Simplicity and cost of about $100 for parts. An improved analog model would be particularly useful for: 1. Calibrating experimental watersheds before new rainfall or runoff data are available. 2. Forecasting water supplies or estimating design floods. 3. Demonstrating routing procedures to local groups. ACKNOWLEDGMENT The author wishes to express his appreciation to Professor A. D. Robinson and L. A. Beyers, Physical Sciences Department, University of Idaho, Moscow, for their assistance in development of the model analog. LITERATURE CITED (1) Bruce, W. A. 1943. An electrical device for analyzing oil-reservoir behavior. Amer. Inst. Mining and Metall. Engin. Trans. 151: 73-85. (2) Chow, Ven Te [ed.]. 1964. Handbook of applied hydrology. McGraw-Hill, New York. Chapt. 29, pp. 10 and 11. (3) Dziatlik, H. 1955. Electrical model for a river. Przeglad Elektrotechniczny 31(7): 447-455. (4) Glover, R. E. 1953. Electrical analogies and electronic computers. Amer. Soc. Civ. Engin. Trans. 118: 1010-1027. 27 (5) Halek, V. 1958. Hydraulic problems using electric analogy. Vodni Hospodarstvi, No. 10, pp. 297-305. (6) Harder, J. A., Mockros, Lyle, and Nishizaki, Ray. 1960. Flood control analogs. Water Resources Center, Cont. No. 24, University of California, Berkeley, 40 pp. (7) Ishihara, T., and Ishihara, Y. 1955. On the electronic analog computer flood routing. Japan Soc. Civ. Engin. Trans., No. 24: 44-57. (8) Ishihara, Y. 1957. On the application of an electronic analog computer for flood routing to actual rivers. Japan Soc. Civ. Engin. Trans., No. 43:43-47. (9) Linsley, R. K., Foskett, L. W., and Kohler, M. A. 1948. Electronic device speeds flood forecasting. Engin. News Record 141(26):64 and 66. (10) Messerle, H. K. 1953. Electronic high speed simulation of hydraulics. Inst. Engin. Jour. 25(3):35-41. (11) Paynter, H. M. 1952. Methods and results, M.I.T. studies in unsteady flow. Boston Soc. Civ. Engin. Proc. 39(2):120-165. (12) Rockwood, D. M., and Hildebrand, C. E. 1956. An electronic analog for multiple stage reservoir type storage routing. U.S. Corps of Engin. Tech. Bui. No. 18, 12 pp. (13) Rogers, A. E., and Connolly, T. W. 1960. Analog computation in engineering design. McGraw-Hill, New York, 450 pp. (14) Scheidegger, A. E. 1960. The physics of flow through porous media. Macmillan, New York, 313 pp. (15) Tribus, N. 1958. The use of analogs and analog computers in heat transfer. Okla. Engin. Expt. Sta. Pub. 100, p. 2. (16) U.S. Soil Conservation Service. 1964. National engineering handbook. Sec. 4, Hydrology, Part 1, Watershed Planning, U.S. Soil Conservation Service, p. 10.3. 28 UNIVERSITY OF ILLINOIS-URBANA 3 0112 099059351