| Rob Speht's Homepage........(Under construction) | ||||||||||||||||||||||||
|
| ||||||||||||||||||||||||
| Renewable Energy | ||||||||||||||||||||||||
|
|
EXTENDING THE LIMITS OF THE SALT GULP DILUTION METHOD OF STREAM FLOW MEASUREMENT R. SPEHT Centre for Renewable Energy Systems Technology, Dept. of Electronic and Electrical Engineering, Loughborough University, UK Indexing terms: salt dilution, salt gulp, discharge measurement, Flowstream�, undersalting, high background conductivity, temperature compensation, mixing length, sewage effluent Abstract The salt gulp dilution method is well established as a means of measuring river flow rate. However, existing operating recommendations impose constraints on the use of this technique. A study has been made of extending the operating limits of the salt gulp method, with particular reference to the Flowstream� integrating conductivity meter. A review of mixing length was carried out, and the effects of thermal shock, undersalting and high background conductivity were studied experimentally. Recommendations are made for extending the operating limits of the method. 1. Introduction Many streams and small rivers may be utilised by small, micro and run-of-the-river hydro-turbines for electricity production. In order that a site may be assessed, it is very important to know what flow rates may be expected. There are several techniques for measuring the flow rate of a potential site. The simplest method involves diverting the entire flow into a bucket or tank of known capacity. The time taken for the bucket to fill is timed and the flow rate is readily found by dividing the volume by the time taken. In the float method the average cross sectional area is calculated for a particular stream and multiplied by the velocity of a float and an empirical factor. This method works particularly well for very regular (man made) channels, in which it is possible to obtain a reliable value for the average cross-sectional area. Propeller devices (also known as current meters) are velocity indicators that display the speed of rotation at a particular point by a mechanical counter when the propeller is placed at the required depth. A formula provided with the device relates the rotational speed to the stream velocity. This can then be multiplied by the cross-sectional area in order to give a estimate of the flow rate. A flow measurement weir may also be used, which has a notch through which all the water in the stream flows. Upstream of the weir the head relative to the bottom of notch is measured and the volumetric discharge is determined using an empirical formula. The Salt Gulp Dilution Method The salt gulp dilution method for flow estimation has been used for many years as a simple, cheap and robust technique for use in small, turbulent upland streams, and as such it is already in wide use for environmental studies. The rigorous application of the salt gulp method, taking account of the various sources of error and uncertainty, can be a demanding and skilful, even tedious task. However, by adopting good practice, with an experienced operator and the right conditions, the accuracy and precision of this technique can be quite high. Salt gulp dilution gauging is particularly good for survey work, where a large number of measurements need to be carried out over a short period of time, either on a single stream or on a number of streams. However, it is generally limited to low and moderate flows because of the inordinate amount of salt required in large flows and because the mixing of tracer at high flows may be more inefficient for a given path length as the flows become more laminar and less influenced by channel roughness. For example at 1,000 l/s flow rate and background conductivity of 500�S, 35 Kg of salt would be required and 140 litres of water (or about 9 buckets) Although simple in principle, the method does require careful application, especially to avoid systematic errors. The most common systematic errors include: the initial estimation of weight of salt required, the measurement and mixing of salt tracer prior to injection into the stream, the manual recording of conductivity readings, the estimation of mixing length and the post-processing of the conductivity data to give flow. The technique relies on the fact that the conductivity of water in a stream will rise when salt is added. A known mass of salt is mixed with some of the stream water in a bucket and tipped carefully in. The mass of salt used depends on the flow rate and trial runs. For accurate results, pure table salt is used, and carefully weighed. Further downstream, a short time after the salt water has been injected, the conductivity of the water is recorded at five second intervals once the conductivity starts to rise as the salt cloud passes the meter. The measuring continues until the conductivity has returned to the background level and is then plotted against time. The flow can then be calculated from the mass of salt and the area under the curve. Flow rate, Q, is given by; where : Q is the flow rate in m3s-1 MSALT is the mass of salt in kg, ?t is the sampling interval in seconds,. C is the conductivity in �S CO is the background conductivity in �S K is the salt concentration per unit conductivity By examination of the above equation it is clear that when the mass of salt is large relative to the area under the curve, the flow rate is also large and if the area is large then this corresponds to a smaller flow. This corresponds with the qualitative description that if the same mass of salt was added first to a river and then to a small stream, the conductivity of the passing salt cloud would be greater in the stream and hence the area under the curve would be bigger. 2. Description of the Flowstream device Dulas Ltd have developed an integrating conductivity meter, the Flowstream�, which calculates the flow rate automatically. The device uses the salt gulp principle. By first measuring the background conductivity and using the known mass of salt (value inputted by operator) calculates the area under the conductivity - time curve and then displays the flow rate using the formula given above. The conductivity is measured every 0.1 seconds and then averaged over 10 values. The curve above shows a typical curve obtained by the Flowstream�. The instrument has potentially many different applications, but the recommendations for use are quite constraining at present and, as will be described later, perhaps somewhat little misleading. Various aspects of the dilution technique in general and more specifically the Flowstream� instrument were studied in order to obtain a better understanding of the operational limits. The overall aim was to be able to use the device to its full potential. 3. Mixing Length The success of river flow gauging by the dilution method mainly depends upon proper mixing of tracers / salts. The salt dilution method of discharge measurement consists of adding a concentrated solution of common salt to a flowing channel and measuring the dilution concentration sufficiently downstream, where efficient mixing is assumed to have taken place. This is particularly suitable for flow measurement in open channels under steady flow conditions where the degree of turbulence is high enough to ensure thorough mixing of the salt solution with the water. 3.1 Literature Review A literature review was carried out in order to assess what previous work had already been completed. Many papers exist, which cover the subject of mixing length (distance between salt injection and point of conductivity measurement). As proposed by Gilman [7] the simplest formula is that of Day (1977) L = 25B. Patra and Bhunia [19] reviewed many techniques in 1984 [17] and recommended the following formula; L=94B-160. And similarly they found Days' formula (1977) L=25B, Elders' formula (1990) L=15B for mountainous rivers. [17] Patra and Bhunia [19] suggested in their paper 'Discharge Measurement in Hilly Streams by Dilution Methods' (1994) that in general mixing length is given by L=KB C which is valid for central and sudden injection Where L = mixing length B = breadth, this was stated as being valid for the range of breadth (2 The two constants C and K were found to vary for different types of river. For smooth regular rivers and sand beds K = 94.0 C is the Chezy constant which equals -161.8. And in hilly/mountainous areas K=77 and C = 120. Although these may be accurate specific empirical results, as a n observation, they are no use for predicting the mixing length for a given river. An analysis of data published by Day in 1976 yielded the following formula L=100Qe1/3 where Qe is the estimated flow. [17]. This requires unfortunately a reasonable estimate of the river flow initially, which may not always be possible. Using the data provided by Patra and Bhunia (1984) Kite derived the useful relationship L=-0.01202 S-1 + 261 A1/2 or simply L = 260 A1/2 where A is the cross-sectional area. [17] Gilman [7] went on to recommend the use of an estimation of mixing length from nature of the banks and river beds. That of Rimmar (1953) L = 0.13C (0.7C + 2g1/2) (B2 / gd) where C is the Chezy constant which depends on the roughness of the river. B is the breadth of stream d is the depth and g = 9.81. 3.2 Comparison of different prediction formulae Most of the prediction theories, and experimental theories are based around the breadth of a river or cross sectional area, with one notable exception, that being Rimmar's (1953) formula. This is very useful because C varies between 15 and 50 as the roughness of the stream bed varies from rough to smooth (50 is for a smooth plastic or glass channel and 15 represents a very rocky stream) The various formulae were compared against each other for a 'typical' river of depth 0.5m and varying breadths with the exception of Day's theoretical prediction formula developed in 1976 which is based upon an estimation of the flow. 3.3 Usage recommendations The various different formulae give quite a variety of predicted mixing lengths for different scenarios. However only one makes allowance for the fact that the streambed may be of a certain roughness. Obviously, for a very smooth channel such as a concrete channel, the mixing length will be much longer than for a very rocky stream. Using Rimmar's formula a table was produced for varying breadth and depth of river and for differing values of Chezy constant (which is dependant on roughness). A complete graphical comparison of the various formulae may be found at the end of this paper. A simplified from of the table is shown below, and a more detailed version of the table may be found in the appendix at the end of this paper. h b 0.25m 0.5m 0.75m 0.25m Very Smooth 10 12 12 Sandy Riverbed 6 7 7 Slightly rocky 3 3 4 Very Rocky 1 1 1 0.5m Very Smooth 21 23 25 Sandy Riverbed 12 13 14 Slightly rocky 6 7 7 Very Rocky 1 1 1 0.75m Very Smooth 31 35 37 Sandy Riverbed 17 20 21 Slightly rocky 9 10 11 Very Rocky 2 2 2 For a given depth (h) of stream and given breadth (b) the mixing length may be found by reading across the table. For each combination four mixing lengths are given from very smooth to very rough with two intermediate value. The equation may naturally be used for any intermediate values, by substituting in the prediction formula. 4. Temperature Effects Flow stream measurement by the salt dilution method may sometimes be required to be carried out in adverse temperature conditions. In colder climates it might well be the case that flood water due to snow melt might require an extra measurement of a river flow. Water at such low temperatures (below 5OC) does effect the performance of salt dilution methods. The amount of salt, which may be dissolved in water, is dependent upon temperature. Similarly at high ambient temperatures the difference between the air temperature and water temperature effects the probe performance as the probe temperature must settle to that of the water temperature before it can provide accurate readings. 4.1 Background Built into the Flowstream� probe is a Hanna thermistor temperature sensor, which measures the stream flow temperature. The programming of the Flowstream� then compensates for the fact that a different conductivity will register. The temperature dependency is about 2% per degree Celsius. Dulas ltd used a four-term polynomial to approximate the temperature compensation necessary. 4.2 Laboratory tests Laboratory tests were carried out to determine how the Flowstream� reacted to thermal shock, and to determine how well it behaved at very low temperatures (below 5OC). 4.3 Discussion The reaction of the Flowstream� to thermal shock may be separated into two parts. Firstly, the amount of time the system takes to settle down, i.e. how long before which the probe may be used reliably. Although a variety of settling times were observed (depending upon the ambient air and water temperatures) the conductivity reading had always settled down within 5 minutes. Secondly there was found to be an approximately linear relationship between the temperature difference (?T) and the sttling time required. A general reccommedation to users of the Flowstream� would therefore be to leave the probe in the stream water for 5 minutes prior to testing to allow the temperatures to equalise. At low temperatures (below 5OC) the probe worked adequately. However the probe was found to be unstable below 0.5OC, it being unable to settle down to a contant reading. Its recorded background conductivity continually rose, even after several hours . this may have been due to local icing. 5. Undersalting The use of less salt than that recommended by the manual, to achieve an estimation of the flow rate in the usual manner. 5.1 Background If less salt may be used for larger flows or higher background conductivities, this would mean that the range of potential sites open to testing by the Flowstream could be widened and thus the marketability of it. At present operation is limited by the amount of salt to be carried to a river or the amount which the operator(s) can dissolve practically. Other considerations include how much salt can be thrown into a river in a short space of time and the environmental effects of doing so. To a lesser extent, the cost of salt used may also be kept down. 5.2 Lab Based Flume Tests A series of tests were carried out in a test flume in the Department of Civil Engineering at Loughborough University to determine the effect of undersalting on the accuracy of flow measurement. For various flow rates the Flowstream� was used with varying peak conductivities (as a percentage of background conductivity) to measure the flow rate. The results given below show the accuracy as a percentage difference. 5.3 Discussion It may be seen from the graph that for levels of peak conductivity above 200% the accuracy is about 3 %. If however one is prepared to tolerate an error of 10% then a peak of only 150% of background conductivity need be used. This may be sufficiently accurate for some purposes. It may also be noted that there is no improvement by going above 200%. At 300% and above the same error of 3% exists. This may not appear to be a large saving but because the quantity of salt needed rises in a logarithmic fashion large quantities of salt may be saved. This depends mostly on the estimated flow of the river. Physically mixing the salt and water in a bucket may also present problems. The maximum amount of salt that may be dissolved in pure water is about 3.6 Kilograms per 10 litres at 25OC. (In practice however, dissolved impurities already exist in river water and no more than 2 or 2.5 Kilograms of salt may be dissolved). 6. Operation with High Background Conductivity & Undersalting for high Background Conductivity Operation Normally in the Loughborough area tap water has a conductivity of about 500�S and river water somewhere in the region of 1,000 �S. In certain situations, however, the background conductivity may be higher than this. If the run-off from roads in the Winter time is to be measured it may be found to have a much higher conductivity due to the dissolved rock salt which is sprinkled on the roads to melt ice. Out-flow waters from a range of industries might contain salts or other chemicals which in themselves are innocuous but which affect the conductivity of the water to be measured.A significant opportunity for the Flowstream� as a device to be used to measure the flow of (treated) effluent that flows out of sewage treatment works. 6.1 Background It has already been established that at lower (common) levels of background conductivity the Flowstream� instrument is accurate. However, it is difficult to predict its performance at higher levels of background conductivity, using theoretical methods alone. It may be expected that at some point the device output would saturate and thus render the result inaccurate. Also, for high levels of background conductivity, such large quantities of salt are required to achieve the minimum 'doubling' that for many applications the test would become impractical. 6.2 Lab based Tests Two sets of tests were carried out in a small flume to determine for a fixed flow rate what different levels of accuracy would prevail. First the background conductivity was raised up to very high levels and then to compare those with what sort of accuracy could be expected from using the undersalting method. The first set of results show the percentage error attributable to varying levels of background conductivity And the second set of results shows a comparison between the percentage error using the recommended salting levels and the percentage error which occurs when using undersalting. Each value of accuracy for the recommended level of salting a corresponding parallel value was also determined. 6.3 Discussion It can be seen from the first graph that with rising background conductivity that using the recommended level of salt produces error bands that increase very rapidly. At 2,000�S the extra error inherent is negligible but by 4,000�S it is at 10% and by 6,000�S it is 40%. In the second graph in may be seen that by comparison the use of undersalting (at 150%) is relatively more accurate. By using undersalting the peak of the curve does not push up into the region where it starts making the calculation inaccurate (which is estimated to be above about 10,000�S) at 150% one may go up to about 6,500�S background conductivity before problems start arising. 7. Probe Tests in Sewage Processing Plant It was expected before the tests were carried out, that the background conductivity of sewage would be higher than normal stream water. This was not entirely the case. While stream water varies around the 1,000�S mark and household tap water varies about the 500�S mark, the conductivity of the raw sewage varied from 700�S to 2,500�S. While these values fall well within the capabilities of the Flowstream� the problem was that they were very variable in the intermediate sewage processing stages. So the device was unable to find a constant background conductivity level. It was thus unable to determine the flow rate. The final effluent leaving the sewage was found to be of comparable conductivity to that of stream water and, more importantly, was found to be very stable. Flow rate measurement of the effluent leaving the sewage plant is therefore no different from measuring that of a stream. 8. Conclusions & Recommendations The settling time increases with temperature difference,. However, the probe was found to settle to within 1�S within 5 minutesfor temperature differences up to 20OC. A general recommedation to users of the Flowstream� would therefore be to leave the probe in the stream water for 5 minutes prior to testing to allow the temperatures to equalise. At temperatures between 5OC and 0.5 OC the probe worked adaquately. However the probe was unable to settle down to a consstant conductivity reading when operated below 0.5 OC. This may have been due to local icing. Undersalting to obtain a conductivity peak of 150% instead of the reccommennded 200-500% range produced results which are accurate to 10% compared with 3% for the recomended range. For many applications this level of accuracy is acceptable and would enable larger flow rates to be measured, by avoiding the need for excessive quantities of salt. High background conductivity levels in the stream to be measured present two problems. It requires even larger quantities of salt to be used which restricts the range of streams which may be practically tested. Also, for really high background conductivities the 200-500% peak may exceed the upper operational limit of the conductivity measuring equipment. The flowstream device has an oparerational maximum of about 10,000�S. To achieve the 200-500% of background as a peak, maximum background conductivity would be 3,000�S. If undersalting would be used to obtain a peak of 150% of background conductivity, flows of up to 6,000�S may be measured. Sewage flow at intermediate processing stages has a very uneven background conductivity. It would therefore not be possible to use the salt gulp method for the raw sewage flowing into the treatment works or in the intermediate stages of treatment. However the final effluent from the sewage works visited had a conductivity comparable to that of stream water and so flow rate measurement should be quite feasible. 9. References 1. Aastad, J., and Soegnen, R.: 'Discharge Measurements by Means of a Salt Solution "The Relative Dilution Method"' (IAHS Gen. Ass. 1954) 2. Bjerve, L., and Groeterud, O.: 'Discharge Measurements by a New-Formed Relative Salt-Dilution Method in Small Turbulent Streams' (Nordic Hydrology 11, 1980, 121-132) 3. Bronge, C., and Openshaw, A.: 'New Instrument for Measuring Water Discharge by the Salt Dilution Method' (Hydrological Processes Vol. 10, 463-470 (1996) 4. Cobb, E.D.: 'Evaluation of Uncertainty in Time-of-Travel Measurements resulting from Truncating Tracer Response Curves'(USGS, 1991) 5. ELE International, The Flowstream� Manual, 6. Gilman, K.: 'A Recommendation Relating to the Sampling Interval for Dilution Gauging by Sudden Injection Method' (Private Circulation Document, Unit of Fluvial Geomorphology, Institute of Hydrology, 1984) 7. Gilman, K.: 'Field Applications of Dilution gauging' (Nordic Hydrol. Prog. Seminar on Methods for flow measurement with emphasis on new methods, Trondheim, Oct 16-18, pp65-82, 1984) 8. Gilman, K.: 'Errors and Uncertainties in Dilution Gauging' (Nordic Hydrol. Prog. Seminar on Methods for flow measurement with emphasis on new methods, Trondheim, Oct 16-18, pp83-101, 1984) 9. Golterman, H.L., Clymo, R.S., and Ohnstad, M.A.M.: 'Methods for Physical & chemical Analysis of Fresh Waters' (Blackwell Scientific Publications, 1978) 10. Groat, B.F.: 'Chemi-hydrometry and its application to the precise testing of hydroelectric generators' (Am. Soc. Civ. Engrs. Proc., Vol41, No. 9, 2103-2427, 1915) 11. Gwillim, J.: 'Assessing the best cut off value for stopping the integration of salt gulp readings' (J. Gwillim, Dulas Ltd. 1996) 12. Harvey, R. A., Kidd, C. H. R., and Lowing, M. J.: 'Automatic Dilution Gauging In Storm Sewers' (Institute of Hydrology, Report No.75 1980) 13. Herschy, R.W.: 'Hydrometry' (John Wiley & Sons Ltd., 1978) 14. Hongve, D.: 'A Revised Procedure for Discharge Measurement by Means of the Salt Dilution Method' (Hydrological Processes, Vol. 1, 267-270 (1987) 15. Hudson, J.A., and Gilman, K.: 'Field testing of the Flowstream� Integrating Conductivity Meter for River Flow Estimation Using the Salt Gulp Dilution Method' (Natural Environment Research Council, 1996) 16. Institution of Water Engineers: 'Proceeding of Symposium on River-flow Measurement' (Institute of Water Engineers, 1969) 17. Kite, G.: 'Computerised Streamflow Measurement Using Slug Injection' (Hydrological Processes, vol 7, 227-233, 1993) 18. Oestrem, G. 'A Method of Measuring Water Discharge in Turbulent Streams' (Hydrologisk Avdeling, 1964) 19. Patra, A. K., and Bhunia, A.K.: 'Discharge Measurements in Hilly Streams by Dilution Method' (IE 9I) Journal-CI Vol. 65, July 1984) 20. Sellin, R.H.J.: 'Flow in Channels' (Macmillan, 1969) 10. ACKNOWLEDGEMENTS Acknowledgements and thanks to Dr Maha Soundra-Nayagam (CREST, Loughborough University), Jo Gwillim (Dulas Engineering), Will Perrott (ELE International) and Kevin Smith (Severn Trent Water) | |||||||||||||||||||||||
|
|
|
Favourite links
|
|
| ||||||||||||||||||||
|
This page has been visited
|