a . ""' 1\ SEL US D 103.24/2: EL-84-5 TECH NICAL REPORT EL-84-5 vvc~ ~L-~-s-BISON )• APPLICATION OF PROCEDURES FOR TESTING AND EVALUATING WATER DISTRIBUTION SYSTEMS by Thomas M. Walsk i Environmental Laboratory U. S. Army Engineer Waterways Experiment Station P. 0 . Box 631 , Vicksburg, Miss. 39180 Ap ril 1984 Final Rep'ort Approved For Public Release; Distribution Unlimited ' c. €a ce 0 En~i ~ -nn Lbrary, Prepared for Office, Chief of Engineers, u. S. Army Wash ington, D. C. 20314 under Work Unit 31794, Water System Operation, Maintenance, and Rehabilitation Destroy this report when no longer needed. Do not return it to the originator. The findings in this report are not to be construed as an official Department of the Army position unless so designated by other authorized documents. The contents of this report are not to be used for advertising, publication, or promotional purposes. Citation of trade names does not constitute an official endorsement or approval of the use of such commercial products. Unclassified SECURITY CLASSIFICATION OF THIS PAGE (When Dete Entered) REPORT DOCUMENTATION PAGE I. REPORT NUMBER rGOVT ACCESSION NO. Technic-;:,1 RPnort F.l-RL.-c; 4. TITLE (and Subtitle) APPLICATION OF PROCEDURES FOR TESTING AND EVALUATING WATER DISTRIBUTION SYSTEMS 7. AUTHOR(e) Thomas M. Walski 9. PERFORMING ORGANIZATION NAME AND ADDRESS u. s. Army Engineer Waterways Experiment Station Environmental Laboratory P 0 Rox 611 Vic-k~hnro Mi"'"' ':\Q1RO 11 . CONTROLLING OFFICE NAME AND ADDRESS Office, Chief of Engineers, u. s. Army Washington, D. c. 20314 14. MONITORING AGENCY NAME a AODRESS(II dlllerentlrom Controlling Olllce) 16. DISTRIBUTION STATEMENT (of thla Report) READ INSTRUCTIONS BEFORE COMPLETING FORM 3. RECIPIENT'S CATALOG NUMBER 5 . TYPE OF REPORT a PERIOD COVERED Final report 6. PERFORMING ORG. REPORT NUMBER 8. CONTRACT OR GRANT NUMBER(o) 10. PROGRAM ELEMENT. PROJECT, TASK AREA a WORK UNIT NUMBERS Water Supply and Conservation Program, Water System Operation, Maintenance, and Rehabilitation Work Unit (No. 31794) 12 . REPORT DATE Anril 1984 13. NUMBER OF PAGES 113 IS. SECURITY CLASS. (olthlo report) Unclassified ISe. DECL ASS I Fl CATION / DOWN GRAOI NG SCHEDULE Approved for public release; distribution unlimited. 17. DISTRIBUTION STATEMENT (of tho abatract entered in Block 20, II dlllerentlrom Report) 18. SUPPLEMENTARY NOTES Available from National Technical Information Service, 5285 Port Royal Road, Springfield, Va. 22161. 19. KEY WORDS (Continue on reverae .tide II neces.tcry and Identify by block number) Infrastructure Pipe network models Manometer Pitot tube Pipe flow Water distribution 20. ABS~ACT(C"DDtmua -,.,.r.. ala II nece.a..,. -ldenJity by block number) This report is made up of eight individual papers describing solutions to specific problems in evaluating water distribution systems. The topics range from general discussions on deterioration of water systems and the need for pipe network model calibration to examples of actual reports on pipe break analysis and model calibration. Other papers give examples of how pipe roughness tests and gage calibration can be performed. Two additional papers (Continued) FORM EDITION OF I NOV 65 IS OBSOLETE DO I JAM 73 1473 Unclassified SECURITY CLASSIFICATION OF THIS PAGE (IIO'It.., Data Entered) Unclassified SECURITY CLASSIFICATION OF THIS PAGE(WII., D•t• Entered) 20. ABSTRACT (Continued). describe criteria for selecting differential pressure measuring devices for use with pitot tubes and a computer program for determining flow rates given these differential pressure readings. Unclassified SECURITY CLASSIFICATION OF THIS PAGE(When Date Entered) L PREFACE The work described in this report was performed to demonstrate the tech niques of field testing and evaluation, as presented in Engineer Technical Letter (ETL) 1110-2-278 "Evaluation of Existing Water Distribution Systems." Both the ETL and the report presented herein were prepared at the U. S. Army Engineer Waterways Experiment Station (WES), Vicksburg, Miss., under Work Unit No. 31794 (Water System Operation, Maintenance, and Rehabilitation) of the Office, Chief of Engineers, Water Supply and Conservation Research Program. The principal investigator for this work was Dr. Thomas M. Walski of the Water Resources Engineering Group (WREG), Environmental Engineering Division (EED), Environmental Laboratory (EL), WES. Mr. Sidney (Buddy) Ragsdale of the Sedimentation Branch, Estuaries Division, Hydraulics Laboratory, WES, assisted in the field testing and equipment development. Mr. Ragsdale's field testing capabilities were substantially responsible for the success of that portion of the work. Dr. Walski and Mr. Ragsdale were assisted by Mr. Michael Evans, WREG. Reviews of portions of the report were provided by Drs. Roger W. Burke and Joe Miller Morgan, WREG. Work performed with the Washington Aqueduct Division (WAD) of the Balti more District of the Corps was done under the purview of Mr. Harry C. Ways, Chief, WAD, and Mr. C. C. Peterson, Chief, Engineering Branch, WAD. Mr. Keith Copeland of WAD also provided valuable assistance in this study. Testing conducted in Vicksburg, Miss., was conducted under the purview of Mr. John Kelly of the Vicksburg Water Department. The work unit technical monitor at the Office, Chief of Engineers, was Mr. James Balli£ (DAEN-ECE-BU). Part VI is based on a paper "Why Calibrate Water Distribution System Models?" which appeared in the October 1983 issue of Water Engineering and Management. Part IX is based on a paper "How Water Systems Age" which appeared in the July-August 1983 issue of The Military Engineer. Both were written by Dr. Walski. The study was conducted under the direct supervision of Mr. Michael R. Palermo, Chief, WREG, and under the general supervision of Mr. Andrew J. Green, Chief, EED, and Dr. John Harrison, Chief, EL. The Commander and Director of WES was COL Tilford C. Creel, CE. The Technical Director of WES was Mr. F. R. Brown. 1 This report should be cited as follows: Walski, T. M. 1984. "Application of Procedures for Testing and Evaluating Water Distribution Systems," Technical Report EL-84-5, U. S. Army Engineer Waterways Experiment Station, CE, Vicksburg, Miss. 2 CONTENTS Page PREFACE . . . . . . . . . 1 CONVERSION FACTORS, U. S. CUSTOMARY TO METRIC (SI) UNITS OF MEASUREMENT 5 PART I: INTRODUCTION .. 6 Background and Purpose 6 Overview . . . . . . . 7 PART II: PIPE BREAK ANALYSIS 9 Introduction 9 Historical Records 9 Soil Corrosivity Tests 15 Costs of Breaks and Remedial Measures 17 Economic Analysis of Breaks 17 Summary . . . . . . . . . . 24 PART III: DEVELOPMENT OF A WATER DISTRIBUTION SYSTEM COMPUTER MODEL 25 Introduction 25 System Description 25 Calibration Process 29 Summary . 35 PART IV: USING PITOT TUBES TO CALIBRATE AND TEST GAGES 36 Introduction 36 Calibration . . . . 36 Summary . . . . . . 40 PARV V: DETERMINING PIPE ROUGHNESS AND HEAD LOSS 42 Background . . . . . . . . 42 Conducting Head Loss Tests 43 Illustrative Example 49 Common Pitfalls . 62 Summary 64 PART VI: WHY CALIBRATE WATER DISTRIBUTION SYSTEM MODELS? 66 Introduction . . . . . . . . . . . . . 66 Scenario (Can This Happen in Your System?) 66 Calibration Data (How Good is Good Data?) . 67 Adjusting the Model (How Do You Use the Data?) 68 Excuses, Excuses, Excuses ..... 68 Summary . . . . . . . . . 70 PART VII: PROGRAM FOR REDUCING PITOT TRAVERSE DATA 71 Purpose . . . . . . . 71 Input . . . . . . . . 71 Calculating Velocity 71 Determining Flow 72 Program and Variable Listing 77 Example Problems . . . . . . 77 3 Page PART VIII: SELECTING DIFFERENTIAL PRESSURE DEVICES FOR USE WITH PITOT TUBE . . . . 89 Introduction 89 Alternative Pressure Measuring Devices 89 Evaluation of Differential Pressure Measuring Devices 90 Construction of Air-Filled Manometers 93 Operation . . . . . . . . . . . 95 Summary ........... . 97 PART IX: PREVENTING DETERIORATION OF WATER DISTRIBUTION SYSTEMS 98 Introduction . . . . . . 98 Pipe Breaks . . . . . . . 98 Loss of Carrying Capacity 100 Malfunctioning of Appurtenances 101 Summary 102 REFERENCES 103 APPENDIX A: DIPRA SOIL INVESTIGATION REPORT AI 4 CONVERSION FACTORS, U. S . CUSTOMARY TO METRIC (SI) UNITS OF MEASUREMENT U. S. customary units of measurement used in this report can be converted to metric (SI) units as follows: Multiply By To Obtain feet 0 . 3048 metres feet per second 0.3048 metres per second feet per second 0 . 3048 metres per second per second per second gallons per minute 3.785412 cubic decimetres per minute inches 25.4 millimetres miles (U . S. statute) 1.609347 kilometres pounds (force) per 6894 . 757 pascals square inch pounds (mass) per 16 . 01846 kilograms per cubic foot cubic metre square feet 0.09290304 square metres 5 APPLICATION OF PROCEDURES FOR TESTING AND EVALUATING WATER DISTRIBUTION SYSTEMS PART I: INTRODUCTION Background and Purpose 1. With the publication of Engineer Technical Letter (ETL) 1110-2-278 (Headquarters, Department of the Army 1983), the Corps of Engineers has guidance on testing, making decisions concerning rehabilitation, and developing and calibrating computer models for water distribution systems. While this ETL provided engineers with tools for testing and evaluating water distribution systems, it also raised more specific questions. Some examples of the questions left unanswered included: what kind of equipment should be used to conduct head loss tests, how do water systems deteriorate, how can flow measuring devices be calibrated, and why is it important to calibrate pipe network models? Similarly, there was the need to demonstrate application of the techniques presented in the ETL on analysis of pipe breaks and network model calibration. 2. During fiscal year 1983, the U. S. Army Engineer Waterways Experiment Station (WES) addressed many of the above issues under the Office, Chief of Engineers, Water Supply and Conservation Research Program. The purpose of this report is to document the field testi ng and other analyses conducted as part of this work. 3. Unlike most WES reports which address a single issue, this report is actually a compendium of eight separate reports, each with its own purpose, target audience, and level of detail. In general , the target audience consists of engineers conducting water distri bution studies. Whi le Part VI describes the need to calibrate pipe ne two r k models and is oriented toward study managers, Part III illustrates the calibration procedure in an actual system and is oriented toward the engineer actual l y performing the work. 4 . The primary lesson learned i n this work was that there is no substitute for collection of substantial, accurate data. It is hoped that the description of the collection a nd use o f such da t a i n water di stribution system eval uation will serve as a mod e l fo r good data collecti on practices for 6 future studies. This report contains not only a description of successful test results but also highlights the pitfalls awaiting those who might want to use haphazard methods for system evaluation. Overview 5. Each part of this report addresses individual issues involved with evaluation of water distribution system. The respective issues are listed in the following paragraphs. 6. Part II gives the results of an analysis of pipe breaks in the Federally Owned Water Main (FOWM) System operated by the Washington Aqueduct Division of the U. S. Army Engineer District, Baltimore. The process used to determine whether to replace or rehabilitate water mains, based on an analysis of prior breakage, is illustrated. 7 . Part III illustrates the development and calibration of a water distribution system model for the FOWM System. Special emphasis is placed on collection and use of model calibration data. 8. In Part IV the problem of calibrating a meter using a pitot tube is addressed for a situation in which a pitot tube was used to calibrate an orifice plate. 9. Loss-of-head tests are conducted to measure pipe roughness and carrying capacity. In Part V, the techniJues for conducting these tests are demonstrated in six field tests. 10. Some engineers and utility managers q estion whether the precise model calibration methods described in E~~ 1110-2-278 and demonstrated in Part III are necessary. Part VI discusses the need for accurate model calibration. 11. In measuring flow rate using a pitot tube, it is necessary to convert the differential pressure readings made using a manometer or differential gage into point velocities and then integrate the velocity distribution to determine flow. Part VII describes an easy-to-use computer program which can be used to solve this problem. 12. As mentioned above, pitot tubes must be attached to some diffe~ential pressure measuring device. Part VIII discusses the relative merits of manometers, differential gages, and transducers for making these readings and concludes an air-filled manometer is best for field use. Instructions for 7 constructing and using such a manometer are also presented. 13. Part IX is a general description of how water distribution systems deteriorate with time. It discusses in fairly broad terms how to analyze and correct these problems. 8 PART II: PIPE BREAK ANALYSIS Introduction 14. The Federally Owned Water Main System has been serving Federal facilities in the northern part of Arlington County, Virginia, since the 1940's. As with any water main, these mains are subject to breaking, and there is a cost associated with each break for repair, water loss, damages, and inconvenience. If the cause of the breaks can be identified and corrected by replacement or repair of the main, the rate of breakage for those mains can be reduced to virtually zero with the elimination of costs associated with breaks. The purpose of this analysis is to determine whether it is economical to replace water mains or take other remedial action in the FOWM System due to frequent main breakage. 15. In this part, the historical record of breaks is reviewed to identify problem areas and breakage trends. The results of soil corrosivity tests are presented and the costs of breaks and remedial actions are estimated. Finally, the costs and break projections are combined to develop recommendations for remedial actions. Historical Records Problem areas 16. To evaluate the economics of pipe replacement, it was necessary to assemble records on main breaks in recent years. These records were grouped into two data files: a pipe inventory and a break record file. The pipe inventory consists of the pipe identification number, length, diameter, type, and a brief description of the pipe as shown in Table 2-1. Since all mains in Table 2-1 are cast iron except for the one 30-in. ~·( steel main, the type of pipe is not listed. Table 2-2 gives data on breaks during the period 19701981. The data in Tables 2-1 and 2-2 were assembled by the Washington Aqueduct Division (WAD) and stored in data files on the Boeing Computer Service (BCS) computer by WES personnel. * A table of factors for converting U. S. customary units of measureme nt t o metric (SI) is presented on page 5. 9 Pipe Number 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Length ft 6,000 14,200 500 4,400 3,200 1,200 1,400 1,500 1,600 600 3,910 1,600 1,600 1,000 700 2,600 60 1,400 1,000 800 Diameter in. 30 16 16 24 16 10 8 8 12 18 16 24 8 8 6 8 10 6 8 6 Table 2-1 Pipe Inventory Description Main--Key Bridge to Pentagon Main--Key Bridge to V291 Main--Marshall Drive Main--Pentagon Loop to 15th and Eads Sts. Main--15th and Eads to V306 Airport Main--V229 to V235 Navy Annex Fire hydrant loop-Navy Annex Main--V233 to pump, Henderson Hall Main--from V291 to V103 Pentagon South Park Main--V108 to V104A Pentagon Loop Main--V121 to 107 Pentagon Loop Main--Vl21 to V108 Pentagon Loop Fire hydrant feed from V113A (Pantagon Loop) Supply to Pentagon heating plant Supply to heating plant from Arlington County Fire hydrant line V131 to V123 Line--D.C. side Key Bridge Line--feed to Columbia Island and sewage plant Fire hydrant loop around Pentagon heating plant Fire hydrant feed to FH1, FH2, FH3, Pentagon Note: V =valve, FH = fire hydrant. 10 Table 2-2 Break Record Type of Break-l• Description 4 Corrosion hole 1 Shear-approx. 15 ft of pipe abandoned-blank flange insta11ed 1 Line sheared 4 Corrosion hole 2 Lead joint repair 4 Hole in pipe 1 Line sheared where concrete footer 1 2 Lead joint repair it was resting on 1 Lead joint on fire hydrant 1 Sleeve over shear 4 Hole in pipe 2 Lead joint 5 2 Lead joint 5 5 5 2 Lead joint Army Navy and Eads 5 4 1 Sheared, sleeve install 1 Rupture in cemetery 4 Hole in pipe near Key Bridge 5 Line sheared where resting footer short piece and two dressers installed 1 Shear split sleeve installed 1 Ord and Wetzel Gates in cemetery 2 D.C. end of Key Bridge 4 Hole in pipe 1 Light pole bearing on pipe 2 Lead joints 1 Sheared repaired with dressers 1 4 Hole near FH33 1 Loop Heating Plant between FH21 and FH22 1 1 1 Split in pipe near Rt SO and George Washington Parkway 4 Two patches welded break in pipe; 2 = joint leak; 3 = hydrant leak; Pipe No. 11 17 15 11 4 11 15 15 4 16 15 11 5 7 4 6 2 15 4 14 11 7 2 1 15 8 2 2 11 4 4 18 14 11 19 20 15 2 11 Month/Year of Break 7/70 2/71 3/71 7/71 8/72 9/72 11/72 1/73 7/73 2/74 8/75 6/76 8/76 12/76 2/77 8/77 8/77 9/77 6/78 8/78 8/78 3/79 6/79 8/79 9/79 12/79 12/79 2/80 8/80 8/80 9/80 10/80 12/80 12/80 12/80 1/81 2/81 2/81 11/81 ~·· Break type: 1 = 4 =other; 5 =unknown. 11 Breakage trends 17. Before analyzing the economics of breaks, it is important to identify trends in break rates . Evaluation of these trends is necessary since the projection of future breaks depends on identifi cation of break causes and determination of how these causes are affected by time. For examp l e, if breaks are caused by corrosion, the break rate will increase with time, while for breaks due to improper bedding, the rate may remain constant or even decrease with time. 18. Data were provided by WAD for 20 pipes with a length of 9.33 miles. In the 12-year period under consideration, 39 breaks were reported giving an overall break rate of 0.35 breaks/year/mile. With this rate the FOWM would have the third highest break rate out of 15 cities for which O'Day (1982) reported break rates. This high rate is probably due to the large amount of construction activity which has taken place around the FOWM mains and the presence of corrosive soil in the area. Some mains have been subjected to loads in excess of design loads during construction and operation of the George Washington Parkway, I-95, and the Washington Metro. 19. One obvious trend in the breakage rate data is the increase in break rates after 1976 . For the years 1970-1975 the break rate was 0.196 breaks/year/mile, while for 1976-1981, the rate was 0.500 breaks/year/ mile. The apparent dramatic increase in break rate may be due simply to better record keeping after 1976. However, if the quality of the records is consistent, then the break rate in any year can be given by (ETL 1110-2-278): ~ J (1 + b)t-1970 0 (2-1) J =0.28 exp [o.13(t-1970) J ~ 0 . 28(1.13)t~ 1970 where J =break rate, breaks/year/mile J =break rate in 1970, breaks/year/mile 0 b = rate constant, 1/year t =year It is doubtful that such a steep increase in break rates as shown above would continue into the future. However, if it did, the break rate in the year 2000 would be 13.8 breaks/year/mile, which is higher than any rates reported in the 12 literature. While an experimental growth rate b of 0.13/year may be inordinately high, the data show that breaks are increasing. If the data set is accurate and complete, the most likely explanation is the heavy construction that has taken place in the area in recent years. Some of the notes in Table 2-2 indicate that breaks have been caused by contact between pipes and other structures. With new construction there are simply more opportunities for contact to occur. Despite the possible explanation for the increase in breaks given above, the increased break rate may be misleading or exaggerated since a few missing entries in the break data for a system with a fairly small number of breaks may drastically change the calculated break rate. This appears to be the case here . 20. Another phenomenon shown by the data is that breaks tended to be repaired in August (25 percent) while no breaks were repaired in April and May. Most of the joint leaks were detected and repaired in the summer. This appears to be primarily due to the dry weather in t he summer which makes it much easier to locate these breaks. Breaks are most likely to be more evenly spread throughout the year than indicated by the repair dates. 21. While the temporal trends associated with pipe breaks are interesting, the primary purpose of this analysis is to identify bad sections of pipe to be replaced or repaired. This was done using break rate data files prepared early in this work. The break rates are summarized by pipe and type of break in Table 2-3. 22. From Table 2-3 it is clear that a small number of pipes accounted for most of the breaks. Two thirds of the breaks can be assigned to pipes 2, 4, 11, and 15 which have overall break rates of 0.17, 0.65, 0.98, and 4.80 breaks/year/mile, respectively. Note that pipes 7, 17, and 20 also have high break rates but this is primarily because they are fairly short (e.g. pipe 17 is 60ft long and had only one break). The next step in the analysis was to examine each pipe with a significant number of breaks or break rate to identify the causes of the breaks. 23. Pipe 2 is the 14,200-ft-long, 16-in. pipe from Key Bridge to Valve 291. It broke five times, which is not unduly high for such a length. There are no significant patterns to the breakage except that all of the breaks have occurred since 1977. 24. Pipe 4 is the 24-in. pipe from the Pentagon to Eads St. Five of its six breaks have been lead joint leaks, possibly indicating poor thrust 13 Table 2-3 Summary of Breaks by PiEe and Type of Break Pipe Number --- Pipe Length ft Pipe Diameter in. PiEe Breaks Rate Number break/ 1970-81 year/mile Joint Leaks Rate Number break/ 1970-81 year/mile Unknown and Other Rate Number break/ 1970-81 year/mile Overall Break Rate break/year/mile 1 2 3 4 5 6,000 14,200 500 4,400 3,200 30 16 16 24 16 0 3 0 1 0 0.00 0.10 0.00 0.11 0.00 0 1 0 5 1 0.00 0.03 0.00 0.55 0.15 1 1 0 0 0 0.08 0.03 0.00 0.00 0.00 0.08 0.17 0.00 0.65 0.15 ...... ~ 6 7 8 9 10 1,200 1,400 1,500 1,600 600 10 8 8 12 18 0 1 1 0 0 0.00 0.34 0.32 0.00 0.00 0 0 0 0 0 0.00 0.00 0.00 0.00 0.00 1 1 0 0 0 0.40 0.34 0.00 0.00 0.00 0.40 0.68 0.32 0.00 0.00 11 12 13 14 15 3,910 1,600 1,600 1,000 700 16 24 8 8 6 0 0 0 1 5 0.00 0.00 0.00 0.48 3.43 0 0 0 0 0 0.00 0.00 0.00 0.00 0.00 8 0 0 1 2 0.98 0.00 0.00 0.48 1.37 0.98 0.00 0.00 0.96 4.80 16 17 18 19 20 2,600 60 1,400 1,000 800 8 10 6 8 6 1 1 1 1 1 0.18 8.00 0.34 0.48 0.60 0 0 0 0 0 0.00 0.00 0.00 0.00 0.00 0 0 0 0 0 0.00 0.00 0.00 0.00 0.00 0.18 8.00 0.34 0.48 0.60 restraint or backfill combined with waterhammer events in the area. If the breaks were in the same general area, then it may be economical to reinforce thrust blocks or install tie rods to restrain the pipe in critical areas. 25. Pipe 11, which is the 16-in. pipe around the east side of the Pentagon, had eight breaks. Most were corrosion related, indicating corrosive soil or stray current. The soil survey described in the next section indicates that corrosive soil is the problem in this area. 26. Pipe 15 has the worst break record with seven breaks in 12 years over a 700-ft length (i.e. 4.8 breaks/year/mile). This 6-in. pipe lies directly under I-395 and connects the Arlington system to the Pentagon heating plant. The primary problem with this pipe is line shear which accounted for all of the breaks for which the causes were reported. These breaks appear to be caused by the loads from the interstate highway. Soil Corrosivity Tests 27. The corrosivity of soil to metal pipe can be determined by taking a sample of the soil and performing tests such as redox potential, pH, soil resistivity, and sulfi de which indicate the corrosivity. The Ductile Iron Pipe Research Association (DIPRA) was asked to perform these soil tests for the WAD system . They provided the auger shown in Figure 2-1 which was used to obtain samples from the level of the pipe as opposed to taking samples at the surface. 28. A copy of DIPRA's test results is presented as Appendix A. In summary, pipe 11, located on the east side of the Pentagon, was found to lie in a corrosive soil . A ball of this organic clay soil which has low resistivity and negative redox potential is shown in Figure 2-2. This explains the large number of corrosion-related breaks along that pipe. Samples taken along pipes 2, 4, and 15 indicated that the soil was a fairly noncorrosive fill material. Potential corrosion problems exist in this fill material in that some fill material can contain flyash which results in a fairly corrosive soil. 29. The implication of these tests on the pipe break rate is that the rate should increase with time along pipe 11 which lies in the highly corrosive clay soil . Any replacement pipe should be made of inert material or wrapped to protect the pipe. 15 Figure 2-1. Auger used to take soil samples Figure 2-2. Organic clay found at pipe 11 16 Costs of Breaks and Remedial Measures 30. Before an economic analysis can be performed, it is necessary to develop cost estimates for repair of breaks and installation of new pipe or some other remedial measure. The cost of remedial measures is described separately for each pipe. Those estimates are very rough "planning level" estimates. Before any contracts are let for work, detailed estimates must be made. The cost of breaks (cost of repair plus indirect costs) is discussed below. 31. The WAD provided WES with work orders from five breaks repaired since 1981. While these work orders only contain estimates of the cost, they can at least be regarded as approximations of the actual cost. The average cost was $2848 which will be rounded up to $3000 to account for inflation and uncertainty for use in economic analysis. 32. The cost to repair a break is only one portion of the cost associated with a break. There are also indirect costs due to interruption of service, flooding, disruption of traffic, and ice, all of which are difficult to quantify. In some cases these costs are negligible, while, on the other hand, if a pipe break should flood a Metro station or cut off water to the Pentagon, the costs can be enormous. For the purpose of this study, the cost to repair a break will be doubled (to $6000) to account for the indirect cost. Economic Analysis of Breaks Data required 33. The economic analysis of breaks consists of comparing the cost of repairing breaks with the cost of remedial action such as replacement of pipe, improvement of supports and restraints, and cathodic protection. The data required to perform the analysis include: existing break rate, cost to repair a break (plus damage/inconvenience factor), cost to replace pipe (or institute other remedial measure), interest rate, and fraction of pipe to be replaced. Three pipes in the FOWM System are candidates for this analysis-pipe 11, the 16-in. main around the Pentagon; pipe 4, the 24-in. main to the east of the Pentagon; and pipe 15, the 6-in. main connecting the Pentagon heating plant to the Arlington County system. Because each pipe has different problems, a slightly different analysis is required in each case. 17 Formulas 34 . The formula for decidi ng when t o replace the ma i ns (or take other remedial action) is essentially: Replace if J > J* (2-2) where J =actual break rate, breaks/year/mile J* =critical break rate at which it i s economical to replace or correct weak pipes, breaks/year/mile 35. In the case in which the corrective action involves a unit cost in dollars per length, J* can be given by (ETL 1110-2-278): 5280 crL 1n [o + b)/(1 + R)J (2-3) 1 + b)m cb 1+R -1 ( where c = cost to replace pipe or correct problem, $/ft r L = fraction of pipe to be replaced b = rate constant for increase in breaks, 1/year R = interest rate, % cb =cost to repair break, $ m = study period, years For this study, a time frame of 20 years was used with an interest rate of 7-7/8 percent, which is the interest r a te for Federal water resources projects for fiscal year 1983. 36. In the case in which the cost was a single amount rather than a dollar per foot cost, J* can be gi ven by (2-4) 18 where C =single amount to correct problem, $ r Q = length of pipe, miles An analysis of each problem pipe is presented below. Pipe analyses 37. Pipe 11. Pipe 11 has primarily corrosion-related problems. The two most likely solutions are replacement and cathodic protection. Equation 2-3 is first used to determine whether it is economical to replace pipe 11. The cost of break is given as $6000 and the cost to replace the pipe with polyethylene encased pipe is $80/ft. It is assumed that approximately one half of the pipe is in corrosive soil and will need replacement (i.e. L =0.5). The primary uncertainty is the value of b (0 to 0.13/year) . J* can be given by = 5280(80)(0.5) ln [o + b)/1.07875] 31, (2-5) (6ooo) [ ( :.~7~75f-1] where = $6000 cb c = $80/ft r L = 0.5 R = 7-7/8% m = 20 years Q = 0. 74 mile $6000 cb = c = $30,000 r Values of J1• for different values of b are given below: J•'· b " liiE_ breaks/~ear/mile 0 0.66 0.04 0.46 0.13 0.20 These results indicate that cathodic protection is economical if it can be provided for $30,000. 38. Cathodic protection is only useful for protecting pipes which are in still good condition. If the pipe has already deteriorated significantly, 19 cathodic protection cannot restore its strength. Deterioration shows up most dramatically as pitting. The next time pipe is excavated due to a 11 break or new construction, it should be examined for pits. If the pits have penetrated significantly into the pipe, then cathodic protection can be eliminated as a remedial measure. 39. Pipe 4. Pipe 4 is the 24-in. main west of the Pentagon to Eads St. Most (five out of six) of its breaks are lead joint leaks, which indicates poor jointing or bedding or inadequate restraints . These problems can be corrected by identifying specific locations where the breaks are concentrated and replacing tie rods or reinforcing thrust blocks in those areas. Unfortunately, there is no easy way to identify joints or bends which may need additional restraints. The only way to permanently correct the problem is to install new pipe. The cost to install new 24-in. pipe in a congested area is $120/ft. Assuming all of the pipe will be replaced (L =1) and the break rate due to leakage is not increasing, Equation 2-3 becomes (for large m ) 5280 C ln (1 + R) r J* = (2-6) Substituting the actual values of cost and interest rate gives =5280 (120) ln (1.07875) = J* 8.0 (2-7) 6000 Since the current break rate is only 0.55 breaks/mile/year, it is clearly not economical to replace the entire 24-in. pipe. This does not mean that the problem of joint leaks in this pipe can be ignored. Joint leaks tend to erode bedding material and result in major breaks. In a high value district, this can be catastrophic. 40. The WAD needs to pinpoint the cause and location of the joint leak and take remedial action to prevent fixture leaks. One way to prevent the problem from becoming worse is to conduct a sonic leak detection survey in the area. In this way the leaks can be detected and repaired in one pass at lower cost than waiting until they reach the surface. If the leaks are concentrated in a small area, it can be economical to install tie rods or bell joint clamps to adequately prevent the joints from moving. 20 Four values of b were inserted into Equation 2-5 to determine possible values of J* . They were 0 (no change in breaks), 0.014/year (observed in Binghamton, N.Y. (Walski and Pelliccia 1981), 0.04/year (observed in Buffalo, N.Y. (U.S. Army Engineer District, Buffalo 1981)), and 0.13/year as determined by Equation 2-1. The critical break rates determined using Equation 2-5 are summarized below: b J* 1/yr breaks/:~ear/mile 0 3.42 0.014 3.06 0.040 2.46 0.130 0.96 Comparisons of these results with the calculated break rate for pipe 11 of 0.98 breaks/year/mile indicate that only at the highest value of b is it economical to replace the main. 41. The present worth cost of cathodic protection program for pipe 11 is $30,000 based on 50 anodes, 1 sacrificial anode placed every 40 ft at a cost of $100 per anode, with installation cost of $400 per anode plus $5000 for miscellaneous and contingencies (i.e. (100 + 400) x 50+ 5000 =$30,000). Equation 2-4 gives J* as $3o,ooo ln [o + b)/1.07875] J* = (2-8) 1 + b )20 ] (0.74) 6000 1.07875 -1 [( 42. Pipe 15. The pipe with the worst break record is pipe 15, a 6-in. main connecting the Arlington County system to the Pentagon heating plant. It is only 700 ft long but has had seven breaks in the past 12 years, which is a break rate of 4.8 breaks/year/mile. The breaks are primarily line shear. Such breaks are usually due to beam loading on a pipe with poor bedding or insufficient wall thickness. The only remedy is to replace the section of the pipe with new pipe. Using a value of b =0 since this type of break usually does not increase with time and a replacement cost of $50/ft gives: J-!:: = (5280)(50)(1.0) ln (1.07875) = 3 3 (2-9) 6000 · 21 The above value of J* is based on r epl acing the entire 700-ft main (L =l). If, for example, closer i nvesti gation shows that the breaks were confined to a 200-ft section, then J* =3.3 (200/700) =0.94 and it would be economical t o replace that section. Furthermore , i f b =0.04 , a conservati ve value, then: (5280)(50) (1. 0) ln ( 1 . 07875) 1. 04 = 1.6 J* = (2-10) ( 6000 ) which indicates that the entire pipe could be economi cally rep l aced . 43. The breaks in pipe 15 may be due to construction of I-395. Breaks in the line may decrease now t hat t he hi ghway ha s be en in p l ace f or some time. I f b r eaks conti nue a t thei r current rate, t he weak s e ctions s hould be pinpointed and replaced . Recommenda tions 44 . Pipe 11 (east s ide of Pentagon) i s the most likely candidate for rep l acement or cathodic protection. During the first opportunity to vi sually inspect the pipe, i t shoul d be examined for pitting. If pitting is s i gnificant, the pipe should be replaced. If p i tting is not significant, a fi rm specializing i n corrosion protection should be hired to design a cathodic protection system . 45. I f breaks continue along pipe 15 (heati ng plant), t hat pipe should be replaced. Ideally, the replacement wi ll be limi ted to t he portion r esponsible for the break rather than the entire pipe . 46. The WAD s hould conduct a sonic leak detection survey along pipe 4 (24-in. pipe between t he Pentagon and the airport). If leak problems are conf ined t o a small section of pipe, then t he fea sibility of remedial measures in that section s hould be investigated . 47 . The WAD should keep better records of breaks and l eaks in t he FOWM System . A first step is to des i gnate a map of the system a s a map of b reaks and use col ored pins or some symbol to i dentify preci se location of each break (e.g. red pins for main breaks, green for joint l e aks, yellow for hydrant or hydrant l ateral leaks). Wi th such a map, i t will be easy to identify "hot spots" in the system needing additiona l investi ga tion to determine the fe a s ibility of remedia l measures. A second step is to i dentify a leak reporting system. Figure 2-3 s hows a sample of a leak repair repor t . Such records will make it easy t o m~ke rlecisions as to remedial meas ures . 22 ------------------------, --·------ - SANTA ROSA W.l.\TER UTILITY LEAK REPAIR REPORT : LOCATION : IMAP NO.: DATE 8 TIME REPAIRED: jw.o. NO. !FOREMAN: -DESCRIPTION OF DAMAGE- WHAT PART WAS DAMAGED ? TYPE OF BREAK: 0 Pipa barrel 0 Flange nuts, bolts , tie rods Osplit 0 Crushed Pipe 0Joint 0 Other -(describe)• 0Hole 0 Crocked bell 0 Valve 0 Circumferential split 0 Broken couplingIN YOUR OPINION, WHAT CAUSED THE DAMAGE ? 0 Service pulled 0 Crocked at corporation stop 0 Gasket blown OWATER IosERVICE I IDEPTH TO 0 Other (describe) SIZE: TOP OF PIFE: MAIN LATERAL PIPE MATERIAL: LOCATION OF LEAK: (circle number closest to leok) 0 Golv. Iron 0 Ductile iron 0 Copper 0 Black iron 0 Cost Iron P.V.C. 0 0 Steel 0 A.C.P. 0 Polybutylene 3 FIFE 3 0 Other ·~· 4 4 EXAMINE BROKEN EDGE OF CAST 0~-DUCTILE IRON PIPE: ~ 6 . 5 Original Min. thickness Deieriorotion is on: thickness: of good grlo/ SIZE OF LEAK: II] Ooutside 0 Inside metal remo~ning: (circle ·one) A ~ @] Is there evidence of previous leo k No of previous leok Ilost repair dole 1 DYES 0NO repairs in some general oreo ? repair clamps p-esenl (if known) TYPE OF SOIL: EXISTING BEDDING: IN YOUR OPINION, SHOULD PIPE BE REPLACED 7 0Rocky 0 Sandy 0 Adobe 0 Grovel/ Sand 0Notive DYes 0No 0Do not know Delay 0Hord pan 0 Loom 0 Peo grovel soil IF YES, EXPLAIN EXTENT ON REVERSE SIDE. -DESCRIPTION OF REPAIR- I It replaced, what DAMAGED PART YIAS: D_RepGired 0Reploced material was used ? IF REPAIRED, WHAT REA\IRS WERE MADE? D leak clomp 0 Repacked valve 0 Other (describe) 0 Welded 0 Recoulked joint FILL IN THE FOLLOWING : l Street nome, north arrow; 2. Draw in main and hydrants in shutdown orea; 3. Show oil valves c!osed and •;olve numbers; 4. Locale leak to nearest intersection or house with address. Show dimensions to property lines or street centerlines. ATTACH THREE PHOTOS : I. Straight down over leok or damage; 2. Close up of leak or domoge ; I 3. Any other photo which you feel will help. If possible, tog and Iobel any ~ re ploced port which you feel describes the general level of corrosion or deteriorat ion ond de liver pan toWORK ORD ER, LEAK REPORT, 8 PICTURES wate r f ie ld office wi th this form. I.~UST BE SUSI,'i TTEO AS SOO~J AS POSS!f;LE I ! ~--·-·-I ·----- - Figure 2-3 . Example of a leak repair report 23 48. Another potential problem area exists where the 30-in. steel main lies close to the metro tracks. There is often stray de current associated with such a system, and this current can cause serious corrosion problems. Because of the importance of the 30-in. main, it is recommended that an "electrolosis survey" be conducted to determine the seriousness of the problem. Summary 49. In general, the FOWM System is in good condition from the standpoint of main breaks. There are a few mains, however, for which the present worth of the cost of break repairs is comparable to cost of taking corrective action to prevent breaks. The effectiveness of the remedial measures depends on the precision with which the measures can be applied. For example, if only 200 ft of pipe 15 needs to be replaced, it is economical to replace the pipe, but if breaks are spread over the entire pipe, replacement is marginal. Better record keeping will make it much easier to make replacement decisions. 24 PART III: DEVELOPMENT OF A WATER DISTRIBUTION SYSTEM COMPUTER MODEL Introduction 50. The Federally Owned Water Mains (FOWM), operated by the Washington Aqueduct Division (WAD), serve many important facilities in north Virginia, including the Pentagon, Washington National Airport, the Navy Annex, Fort Myer, and Arlington National Cemetery. There is some concern over the reli ability of the system in an emergency situation such as a fire failure of or a line. Testing the response of the system to emergencies by shutting down part of the system or discharging water in the amount needed to fight fires would interfere with service. Therefore, in order to study these events, it is necessary to simulate them with a computer model. 51. With a model of the system it is possible to quickly and easily simulate such events as valving off segments of pipe, changes in water use, fire events, and new piping. Of course, for the model to be valid it is necessary to calibrate it carefully with observations from the field. Calibration is achieved by adjusting water use magnitude and distribution and pipe roughness (Hazen-Williams C-factor) until the heads predicted by the model agree with those observed in the field. In the following sections, the overall model is described, and the calibration process is documented. System Description 52. The area to be modeled extends from the Dalecarlia Water Treatment Plant and Foxhall Tank in the District of Columbia to Washington National Airport, and is part of the "First High" service area served by the WAD . Pipe sizes range from 48 in. near the source to 6 in. in some of the mains near the extremity of the system. The flow in the First High service area is divided between the FOWM System and the remainder of First High system on the east side of the Key Bridge. A 30-in. and a 16-in. main cross the Key Bridge to serve the FOWM System. The 30-in. and 16-in. mains follow roughly parallel paths from the Key Bridge to the Pentagon. Past the Pentagon, the most important main is a 24-in. main which reduces to a 16-in. main as it approaches National Airport. 53. In order to study the behavior of the distribution system. it is 25 not necessary to include every pipe in the system. Rather, it is possible t o analyze a skeletal system which includes only the major mains and still produces accurate results since the mains no t included in the skeletized system do not carry a great deal of the flow. A map of the skeletized system to be modeled was prepared on a 1:7200 scale. It contained circled three-digit numbers corresponding to node numbers in black and elevations of the nodes in green. The one-or two-digit circled numbers referred to test numbers for model calibration, the distance between nodes (in feet) in red and pipe diameter (in inches) in red. 54. Pipes with diameters less than 12 in. were not included in the model of the skeletized system. There were a few exceptions, such as the pipes to the Navy Annex which were 8-in. pipes. The skeletal system did not include many of the small pipes in Arlington National Cemetery. The model also could account for opening of interconnections between the Arlington County system and the FOWM System. 55. The computer program used to calculate flows and pressures in the system is the MAPDIST program, which is the water distribution system analysi1 portion of MAPS (Methodology for Areawide Planning Studies)--a program developed at the U. S . Army Engineer Waterways Experiment Station. It uses the Hardy-Cross method as applied to loop equations. The user's guide and documentation for the program are provided in Chapter 17 of Engineer Manual (EM) 1110-2-502, Change 1. The program can be run on both the Boeing Computer Services system and the CDC Cybernet system. 56 . A schematic drawing of the network showing the location of node~ is given in Figure 3-1. The characteristics of each line and node in the 405 NOT TO SCALE 118 131 FT . MYER ARLINGTON CEMETERY 114 HENDERSON HALL Figure 3-1. Skeletal pipe network for FOWM System 26 system are given in Table 3-1, which is a Usting of a data file for a run of the MAPS computer program. There is one entry in the table for each pipe in the network as designated by a "LINE" card. Each of these cards contains numbers of the nodes connected by the pipes, the pipe diameter in inches, and the length in feet. The "ELEV" cards contain the elevation of each node in the system. The "TANK" card contains the water level above the bottom of the tank in feet. The "COEF" card contains the Hazen-Williams C-factor. The "OUTPUT" cards contain the water use rate in gallons per minute for each node. Table 3-1 ExamEle of Data for FOWM Model Pipe Pipe Pipe Pipe Diameter Length Diameter Length Keyword Node in. ft Keyword Node in. ft LINE 501 502 48 11900 LINE 401 402 16 1220 LINE 502 503 48 1200 LINE 314 401 30 1830 LINE 502 301 48 7420 LINE 403 404 8 1120 LINE 502 302 30 7320 LINE 401 407 16 1225 LINE 302 130 16 1820 LINE 404 405 6 1170 LINE 302 309 30 1580 LINE 403 406 16 1940 LINE 309 310 30 2590 LINE 201 202 24 1860 LINE 310 311 30 3340 LINE 202 203 24 2800 LINE 311 312 30 340 LINE 203 121 16 1730 LINE 312 313 30 1890 LINE 121 115 16 1540 LINE 313 314 30 1960 LINE 115 116 8 1250 LINE 314 315 8 1210 LINE 115 117 16 1650 LINE 315 402 8 1600 LINE 117 118 8 1100 LINE 131 102 16 3450 LINE 117 119 16 570 LINE 130 131 16 2360 LINE 119 120 8 1330 LINE LINE 311 101 101 102 16 16 530 63 Keyword Node Elevation, ft LINE 102 103 16 1900 ELEV 101 65 LINE 103 104 16 1390 ELEV 102 65 LINE 104 105 16 1260 ELEV 103 60 LINE 105 106 16 880 ELEV 104 50 LINE 106 107 16 670 ELEV 105 60 LINE 107 108 16 240 ELEV 106 60 LINE 108 109 10 420 ELEV 107 65 LINE 109 110 8 66 ELEV 108 65 LINE 110 113 8 1460 ELEV 109 65 LINE 113 114 8 1150 ELEV 110 85 LINE 110 111 8 720 ELEV 111 150 LINE 111 112 8 1400 ELEV 112 150 LINE 108 201 12 1610 ELEV 114 150 LINE 201 406 18 700 ELEV 115 30 LINE 406 401 24 1700 ELEV 116 30 (Continued) 27 Table 3-1 ( Concluded) Keyword Node Elevation 1 ft Keyword Node Elevation 1 ft ELEV 117 40 ELEV 501 150 ELEV 118 39 ELEV 502 155 ELEV 119 40 ELEV 503 235 ELEV 120 18 Keyword Node Water Level 1 ft ELEV 121 30 ELEV 130 50 TANK 503 10 ELEV 131 10 Keyword C-factor ELEV 201 45 ELEV 202 45 COEF 70 ELEV 203 50 Water Use Rate, ELEV 301 75 ELEV 302 Keyword Node ~n~m/node 75 ELEV 309 15 OUTPUT 403 200 ELEV 310 10 OUTPUT 406 600 ELEV 311 35 OUTPUT 101 200 ELEV 312 20 OUTPUT 111 100 ELEV 313 30 OUTPUT 112 100 ELEV 314 30 OUTPUT 117 100 ELEV 315 12 OUTPUT 114 100 ELEV 401 30 OUTPUT 116 100 ELEV 402 30 OUTPUT 118 100 ELEV 403 30 OUTPUT 119 100 ELEV 404 10 OUTPUT 120 100 ELEV 405 10 OUTPUT 103 50 ELEV 407 50 OUTPUT 104 50 ELEV 406 30 OUTPUT 404 100 OUTPUT 301 5000 Table 3-2 Water Use in FOWM System User Node Number Percent of Water Use Pentagon 406,403 40 Washington National Airport 116, 117' 118, 119' 110 25 Fort Myer 101 10 Arlington National Cemetery 103, 104 5 Pentagon Heating Plant 404 5 Navy Annex 111, 112 10 Henderson Hall 114 5 100 28 57. Assigning water use to the nodes in the network is quite difficult in this study because of the lack of metering. Table 3-2 shows how water use was divided among major users served by the FOWM System. For model calibration, flow of 5000 gpm between the Foxhall Reservoir and the District of Columbia portion of the First High service area was used. 58. In actually running the program, the data were divided into two computer files: MAIN, which contained a description of the pipes and nodes in the system; and USE, which contained water use data. These files were merged along with data on C-factors and fire flows to make up data for a run of the program. Calibration Process Data 59. In order to calibrate the model, it is necessary to collect data on pressures and flows in the system during a period of time in which the operation of pumps, tanks, and valves is constant (i.e. pumps are not switched on and off; tank levels do not fluctuate widely; and valve settings are not changed). Such data were collected on 18 November 1982 between 10:20 a.m. and 1:45 p.m. and throughout the day and on 12-13 April 1983. On 18 November the pumps at the Dalecarlia Pumping Station to the First High service area were off, no valve settings were changed, and the water level at Dalecarlia Tank changed only from 245.8 ft to 244.7 ft. That set of data is used for calibration while the set collected on 12-13 April is used for later verification. 60. One problem with the data collection for this work was that it was not possible to meter the flow to the FOWM System as opposed to the remainder of the First High service area. It was, however, possible to determine the drawdown on the Foxhall Tank over a known period of time. Given that the water level dropped by 1.9 ft from 10:00 a.m. to 2:00 p.m. on 18 November and there was a change in storage of 0.877 MG per foot of elevation change, the outflow from the tank was 1.66 MG/4 hr or 6917 gpm during the time testing was conducted. Since it was highly unlikely that any flow was occurring from the other First High reservoir (near U.S. Soldiers Home) to Key Bridge, this flow represents an upper limit of the flow to the FOWM System. 61. Pressure readings were taken at numerous locations in the FOWM System on 18 November 1983. At several of these locations fire flow tests 29 were also conducted. The results of the pressure readings and fire flow tests are given in Table 3-3. Some additional pressure readings were made on 12-13 April 1983, but these were not included in the table since they correspond to a different set of water levels in reservoirs and pump operation. The data shown in Table 3-3 correspond to the Foxhall Road Tank being filled to elevation 245 ft and the pumps at Dalecarlia, which serve the First High Service Area (to which the FOWM System is connected), being off. 62. During the testing conducted on 13 April 83, the flow measured at the meter on the 30-in. mains in the vault under Key Bridge was compared with flow measured using a pitot tube. Over a 21.6-min period, the meter recorded a flow of 40,000 gal, which corresponds to an average flow rate of 1850 gpm. A pitot traverse of the 30-in. main, which is actually 24 in. in diameter in the vault immediately upstream of a gate valve and meter, indicated that the flow was 1680 gpm. The 10 percent difference in readings was most likely due to the fact that the pitot tube had to be inserted into the pipe very close to a bend. The meter reading in this case was probably more accurate than the pitot tube reading. Since the flow measurement was made during a weekday , the reading of 1850 gpm (plus roughly 300 gpm, which is the flow in the parallel 16-in. line) can be considered a typical daytime water use in the FOWM System. 63. Another type of measurement conducted during the 12 April 83 field work was head loss. This measurement was conducted by connecting a manometer between two hydrants along a major transmission main. The hydraulic gradient in feet/1000 feet could then be measured by reading the manometer to determine head loss and dividing by the length of pipe over which the head loss occurred in thousands of feet. The tests were conducted along the west side of the Pentagon between hydrants 13 and 12 (210 ft of 24-in. pipe) and between hydrants 57 and 58 along the George Washington Parkway in front of Washington National Airport (720 ft of 16-in. pipe). The results showed that the hydraulic grade line had a slope of 1.0 ft/1000 ft on the west side of the Pentagon and 0.4 ft/1000 ft in front of Washington National Airport. Results 64. Calibration of the model of the FOWM System presented some unusual problems in that there was no metering of individual water users; the quality of available elevation data was poor; and there was very little head loss in the system at normal flows. The lack of metering meant that the di stribution 30 Table 3-3 Summary of Pressure Test Results Test Number Location and Pipe Size Time-- Hydrant No. Pressure psi Elevation ft Head ft-- Hydrant Pressure psi Test Head ft -- Hydrant Discharge ____gp_m 1 Vault below Key Bridge (30 in.) 10:21 - 100 9 240 2 Park Service hose bib (16 in.) 10:30 - 92 30 242 3 Along Metro (30 in.) 11:00 136A 93 25 240 4 Pentagon north parking lot (8 in.) 11:18 15 83 48 240 78 228 1300 5 South side Pentagon (16 in.) 11:43 5 92 34 246 85 230 1140 w ,_. 6 Riverside p~rking lot (8 in.) 12:00 36 95 12 231 65 162 1620 7 Heating plant (6 in.) 12:20 22 98 12 238 68 169 800 8 National Airport (16 in.) entrance 12:40 56 95 20 239 80 205 1140 9 National Airport south terminal (8 in.) 1:00 41 89 18 224 10 Navy Annex (8 in.)* 1:15 46 62 150 293 45 254 650 (Continued) * Connected to Arlington System. Test Number Location and Pipe Size 11 H&S Headquarters (past Henderson Hall) (8 in.) 12 Arlington Cemetery Visitors Center (16 in.) 13 Lee Mansion (4 in.)t w N Table 3-3 (Concluded) Time -- Hydrant No. Pressure ____p~ Elevation ft Head ft -- Hydrant Pressure psi Test Head ft Hydrant Discharge gpm 1:30 - 60-80<'••'• 150 289 335 1:45 - 82 48 239 2:10 - 47 210 318 25 267 440 ,..,., Booster pump cycling on/off. t Downstream of Fort Myer booster pumps. -1 of water use had to be estimated based on size of the installations and water use had to be adjusted during calibration. 65. The head loss across the FOWM System during normal afternoon use was on the order of 0.3 ft/1000 ft, which is extremely low. This meant that pressure readings taken during normal use were not very helpful for calibration since head loss was of the same order of magnitude as errors in the elevation data. Therefore, it was not possible to use the method of model cali bration described in ETL 1110-2-278, Inclosure 5, for calibration. (At one step in that procedure it is necessary to divide by the head loss at normal use, but dividing by a number such as 4 + 10 is meaningless.) Therefore, calibration calculations had to be based primarily on the comparison between observed and predicted heads during fire flow tests. 66. The results of the calibration are shown in Table 3-4. The first two columns give the water use and C-factor used in a particular set of model runs. Columns 3 through 7 give the heads predicted at Key Bridge, along the Metro about 2000 ft north of Memorial Bridge, at the north entrance to the Pentagon, at the entrance to Washington National Airport, and at the Pentagon heating plant, respectively. Columns 8 through 11 give the drawdown in the hydraulic grade line during fire flow tests at the North Pentagon entrance, the airport entrance, Pentagon heating plant, and Riverside parking lot (Pentagon), respectively. The final two columns give the hydraulic gradient predicted by the model along the west side of the Pentagon (between nodes 401 and 406) and along the George Washington Parkway at the airport entrances (between nodes 115 and 117). Table 3-4 Summary of Calibration Results Drop In Head During Hydraulic Gradient Water Use Head at Nodes at Normal flow, ft* Fire Flow Test ft** ftL!OOO ft in FOIIM Along North Airport Heating North Airport Riverside \lest AirportSystem Key Bridge Metro Pentagon Entrance Plant Pentagon Entrance Heating Parking Pentagon EntranceC-Factor &Em Vault (309) 1lli2 ~_lillL (Q=800) (~1620) (24 in.) ~~~ ~ 120 1000 245 245 244 244 0.1 120 2000 243 243 243 242 242 0.2 244 5 0 . 1 4 0.1 80 1000 244 244 244 243 243 4 14 0.2 0.2 80 2000 242 241 240 239 239 6 18 39 51 0.4 0.3 70 2000 242 240 239 237 238 9 23 50 65 0.5 0.4 60 2000 241 238 238 235 235 17 34 66 84 0 .7 0.5 Actual· MeaSUred 240 240 240 239 238 u Q.4 34 69 69 r:o * Parens indicate node number . ** Q.is the hydrant discharge, gpm . 33 67. The two parameters to be adjusted during calibration are the water use in the FOWM System and the Hazen-Williams C-factor. Each row in Table 3-4 corresponds to a pair of values of C-factor and water use tested using the model, except for the final row which contains the measured values of head or change in head. In some of the early test runs both the water use and Cfactor were adjusted. Since water use has less effect on head, especially at high flow, than C-factor, and field data showed that water use should be approximately 2000 gpm, use was set at 2000 gpm and final adjustments were made on the C-factor only. 68. The heads recorded at normal flow do not provide much guidance on which parameter to adjust, and by how much, since Table 3-4 shows the correlation between model and field results is good for virtually any reasonable value of C-factor and water use. The decrease in head during a fire flow test gives a much better indication of the effect of C-factor and water use on the model results. The hydraulic gradient tests do give some guidance on accuracy of calibration at lower flow rates. 69. In general, the best correlations between observed and predicted heads occurred at a C-factor of 70 although in a few cases a C-factor of 60 produced better results. The only value that was not in agreement was the hydraulic gradient along the west side of the Pentagon. It was tw i ce the value predicted by the model. Rearranging the Hazen-Williams equation shows that, for the hydraulic gradient to be 1.0 ft/1000 ft in a 24-in.-diam pipe, the flow and C-factor must be related by 0 54 2 63 0 54 2 63 ~ =0.0067(h/L) · n· =0.0067(1) · (24) · =28.6 (3-1) where Q =flow, gpm C =Hazen-Williams C-factor h/L =hydraulic gradient D =pipe diameter, in. The model predicts a flow in that pipe of 1365 gpm which would require a Cfactor of 48 to give a head loss of 1.0 ft/1000 ft. Similarly, a C-factor of 70 would require a flow of 1995 gpm to yield calibration. Another more likely explanation is that valve number V120 may not be completely open and is therefore causing minor head loss of approximately 0.5 ft. Nevertheless, since 3 4 other indicators show that the best calibration occurs for a C-factor of 70 and for water use of 2000 gpm, the model can be considered calibrated with those values . 70. A C-factor of 70 indicates that the pipes in the FOWM System are quite rough. This would correspond to an appreciable amount of corrosion in an unlined, 40-year-old pipe. This indicates that some of the critical lines in the system may need cleaning and lining, and that the quality of the water produced at the Dalecarlia should be checked to ensure that it is not corrosive. Summary 71. Good quality data are required for developing a water distribution system model. Even with the extensive data collected for this model, calibration is not completely accurate because of the lack of data on water use distribution and a combination of poor elevation data and very low head loss. Nevertheless, the techniques for model development described in this section can serve as a guide for others faced with the problems involved with developing models. 35 PART TV: lJSTNG PTTOT TUBES TO CALTRRATE ANU TEST GAGES Introduction 72. Calibrating or testing a gage or meter used to determine flow in a closed conduit requires comparing the reading from the gage or meter with the actual flow rate calculated using another method. For small flow rates, a "bucket and stopwatch" technique can be used to determine the actual flow rate. Tracer methods are also available which involve feeding a tracer into the flow at a known rate and then measuring the concentration downstream. If the pipe characteristics are known, ultrasonic flowmeters can be used, or small propeller or magnetic velocity meters can be inserted into the flow. 73. Generally, the easiest and most reliable method to measure flow in closed conduits (except possibly the "bucket and stopwatch" method) is to insert a pitot tube into the flow. The pitot tube senses velocity at points in the fluid. The readings can be integrated across the conduit to give the flow rate and average velocity. 74. The following paragraphs describe how a pitot tube was used to DIRECTION calibrate an orifice plate used to 10" STEEL OF FLOW PIPE measure flow through a machine de « signed to test concrete for cavitation. A schematic diagram of the apparatus is shown in Figure 4-1. During pro- PI TOT CAVITATION duction runs of the cavitation machine, TUBE CHAMBER the orifice plate will be used to DIFFERENTIAL measure flow. However, the orifice PRESSURE ORIFICE GAGE plate must be calibrated against PLATE measurements made using the pitot GATE VALVE tube before production runs can be made. Calibration WET WELL Relevant equations 75. Average velocity in the pipe can be related to the head drop across Figure 4-1. Apparatus used to calibrate an orifice plate the orifice by: 36 (4-1) where v = velocity in pipe, ft/sec 2 A = area of orifice, ft 0 2 area of pipe, ft A1 = c = orifice coefficient 2 g = 32.2 ft/sec LlP =pressure drop across orifice, psi y =specific weight of water, 62.4 pcf 2 Since A /A= (D/D) , Equation 4-1 can be rewritten for D = 5 in. (the 01 010 opening of the orifice) and D=10 in. (the full pipe diameter) as 1 v =3 . 0SC ..{iiP (4-2) In order to calibrate the orifice, it is now necessary to determine the velocity in the 10-in. pipe for an array of values for LlP , and calculate the average value of C from 0 . 328v c = (4-3) ,JLlP The LlP value can be determined by connecting a differential pressure gage to taps on each side of the orifice. Velocity v can be determined using a pitot tube. 76. The formula for determining point velocity from the pressure difference detected by a pitot tube at a point in the flow is v =Ak...{i:lP (4-4) p where v = velocity at pitot tube tip, ft/sec p A = constant (12.2 psi or 2.32 in. H0) 2 k =pitot tube constant t.P =difference in pressure between legs of pitot tube, psi or in. 37 In thjs study, differential pressure was measured in two ways: (a) an airwater manometer for which the readings were made in inches of water, and (b) a differential pressure gage reporting in pounds per square inch. These point velocities were integrated using the method described in Part VII to give average velocity. Procedure 77. Unlike a typical water distribution system which is always under pressure, flow in the cavitation equipment can be turned off except during testing, so it was possible to tap the pipe when it was empty. The pitot tube was inserted into the pipe through a l-in. corporation cock which was specially drilled and threaded to accept the pitot tube. 78. The procedure for testing consisted of: (a) turning on the pump, (b) filling the manometer with water and bleeding air from the lines to the manometer and differential pressure gage, (c) recording the differential pressure from the gage connected across the orifice plate, (d) recording the differential pressure from the manometer or differential gage connected to the pitot tube at 11 points from the bottom of the pipe to the top of the pipe at l-in. increments, and (e) changing the flow rate by adjusting the values and repeating steps (c), (d), and (e) until a wide range of flows had been measured. Results 79. Once the measurements were made, the velocity in the 10-in. pipe was calculated using the computer program PITOT (developed at the U. S. Army Engineer Waterways Experiment Station) which converts the differential pressure readings from the pitot tube into point velocities and integrates the velocity profile to determine flow and average velocity. 80. The results of the tests are summarized in Table 4-1. For fairly large velocities the orifice flow coefficient C was fairly constant. As the velocity dropped, the value of C decreased. This was caused by the fact that in the later runs at low velocities the ball valve was throttled substantially and the gate valve was open. Therefore, it was doubtful that flow extended across the entire pipe downstream of the orifice plate in runs 7, 10, and 11. Therefore, the results from these runs were not used in determining an average C. Discussion 81. The C value of 0.49 is lower than typical values for C 38 Table 4-1 Results of Calibration of Orifice Plate Velocity by Differential Pressure 0.328v Run Pitot Tube v at Orifice Plate tu> c = No. ft/sec ,ESi .ff 7.1 0.52 1 4.2 6.9 0.46 2 3.7 5.4 0.49 3 3.5 0.49 4 2.5 2.8 5 4.3 (p) 7.7 0.51 6 3.3 (p) 4.7 0.50 0. 43~· 7 2.3 (p) 3.1 7.7 0.44 8 3.7 0.50 9 3.2 4.3 0. 34~" 10 1.3 1.3 0.46"~• 11 2.0 2.0 Ave. 0.49 Note: (p) indicates readings made with differential pressure gage; other readings made with manometer. Indicates value was not used in averaging. (approximately 0.62) for a (D /D) =0 . 5 and a Reynolds number greater than 0 104 . This is primarily due to the fact that the ports at which the pressure drop across the orifice was measured were located a significant distance up stream and downstream of the orifice. In a standard orifice the ports are located at the orifice. 82. Some problems were encountered in measuring the velocities with the pitot tube as the velocities would change somewhat over time. This was course apparently caused by the ball valve moving slightly during the of the test. Some additional problems occurred when the cavitation device would leak and suck air into the lines, thus producing unpredictable readings from the pitot tube. These readings were not used. 83. The most significant possible error was caused by the pitot tube and orifice plate being preceded by 10 and 15 ft of straight pipe, respec tively. This was significantly less than the 30 pipe diameters of straight 39 pipe usually recommended for such installations. This proximity to a bend resulted in velocity profiles which deviated from what would be expected in a smooth, straight pipe in that maximum velocity was not at the center of the pipe. This was apparently due to secondary currents caused by the bend and could not be avoided because of the configuration of the apparatus. This error was probably small since the measured flows agreed fairly well with flo\vS determined independently using head produced by the pump and the pump head curve for that pump. 84. Overall, the air-water manometer and differential pressure gage produced the same results when attached to the pitot tube. The manometer is superior for field use because: (a) it is less expensive and delicate; (b) it is slightly more precise (manometer could be read to 0.01 in. H0, while gage 2could be read to 0.03 in.); (c) it is more rugged; and (d) it works if the pitot tube is rotated 180 deg without reconnecting any pipe (i.e. works for flows in either direction). Summary 85. Velocity in a 10-in. line can be determined from the pressure drop across the orifice plate using v = 1 . 49~ (4-5) Equation 4-5 is plotted in Figure 4-2 showing all data points. At low veloc ities when the gate valve is open, Equation 4-5 may overestimate velocity slightly. This should not be a problem since the cavitation apparatus will not be operated for this range of velocities. 86. In terms of flow rate QS , Equation 4-5 can be rewritten Q(cfs) =0.81~6P (4-6a) Q(gpm) =364~6P (4-6b) for flow in cubic feet per second and gallons per minute, respectively. 40 10 r-----------------.----------r------,-----,---~--~---r--~~ u w ~ ,_· 5 3 0 ...J w > 3 4 8 10 DIFFERENTIAL PRESSURE. PSI Figure 4-2. Plot of relationship of velocity and differential pressure 41 PART V: DETERMINING PIPE ROUGHNESS AND HEAD LOSS Background 87. Water utility engineers need to know the internal roughness of pipes in order to analyze flows and pressures in their systems. The HazenWilliams C-factor (Williams and Hazen 1920) is the most commonly used indicator of pipe roughness. Even though it actually depends slightly on the Reynolds number (i.e. velocity and diameter) (Walski 1983b), the C-factor is primarily a function of pipe roughness and is a value for which water engineers have considerable experience. 88. Typical values for C for a wide variety of pipes are available from many sources. One of the best summaries is Lamont (1981). There is, however, a great deal of variance about these typical values (e.g. while a typical pipe in a given category may have a C of 90, C-factors for an individual pipe may range from 60 to 120). It is, therefore, necessary to measure C-factors in a few pipes in a system before a detailed hydraulic analysis of the system can be performed. 89. Measuring this roughness is not easy since an engineer cannot see into a pipe, and even if this were possible, relating the size of randomly spaced and sized roughness elements to C is far from simple. The engineer must measure C indirectly by measuring the other quantities described by the Hazen-Williams equation and solving for C as (5-la) or (5-lb) where Q =flow, gpm L =length, ft h =head loss, ft D =diameter, in. v =velocity, ft/sec 90. In this part, procedures for conducting head loss tests are 42 presented including a description of measurements that must be made in the field and techniques for making these measurements (including techniques which do not involve tapping of pipes). The procedures are illustrated using the results of actual field tests conducted by the U. S. Army Engineer Waterways Experiment Station (WES) and potential problems are discussed. Conducting Head Loss Tests Measurements 91. Conducting a test to determine C involves measuring each of the unknowns on the right side of Equation 5-l. Since, in most cases, there is more than one way to measure each parameter, the implications of the measuring techniques must be considered. The decisions that must be made are described below. 92. Selection of test section. The line selected for testing should have a constant diameter pipe with uniform internal characteristics (e.g. all 8 in. unlined cast iron). There should be no change in flow along the pipe due to junctions. It is best not to have a signi ficant loss of head along the pipe due to valves and bends since this makes interpretation of the test results difficult. 93. Head loss. There are two approaches for measuring head loss: (a) parallel pipe method, and (c) two gage method. In the parallel pipe method, the head loss is measured directly using a differential pressure gage (or manometer) that is connected to both ends of the section of pipe to be tested as shown in Figure 5-l. It is critical that the head detected by each side of the differential gage be exactly the same as the head at each end of the pipe test section. That means there must be no leakage or use along the parallel pipe (actually a hose is usually used instead of a rigid pipe). For greatest accuracy, the range of the differential pressure gage should be only slightly greater than the expected reading. Pressure snubbers may be required on the gage to dampen vibration due to surges in the system. 94. In the two gage method, pressure is measured at each end of the test section along with the difference in elevation, and head loss is calculated as (5-2) 43 DIFFERENTIAL PRESSURE INDICATOR Figure 5-l. Head loss measured by parallel pipe method where h =head loss, ft pl = pressure at beginning of test section, psi p2 = pressure at end of test section, psi zl = elevation of gage at beginning of test section, ft z2 = elevation of gage at end of test section, ft The two gage method can produce results accurate to !2 ft, and as such is significantly less accurate than the parallel pipe method (which is accurate to +0.1 ft or better depending on the differential gage). Therefore, the two gage method is only applicable to test sections where there is very large head loss (>20 ft). The elevation difference (z -z ) should be measured using 1 2 surveying equipment as opposed to being read from a contour map. 95. Length. The length to be used in determining C is the length over which the head loss occurs. This is not necessarily the length of the parallel pipe. If minor losses are not negligible, the C-factor calculated based on actual length will be lower than the C-factor in the pipe alone (as it will 44 include loss in bends and fittings). To determine the C-factor for the pipe alone, an equivalent length for bends and fittings which accounts for minor losses should be added to the actual length. 96. Diameter. Since, in many instances, the actual diameter of the pipe is not necessarily the same as the nominal diameter, significant errors can result in using a C-factor based on actual diameter to predict head loss based on nominal diameter. Since in most cases engineers will be using the values for C in situations in which they will only know nominal diameter, it is best to calculate and report C using nominal diameter. 97. Velocity. The most straightforward way to measure velocity or flow is to insert a flow-measuring device in the pipe. On very rare occasions, there may be a flow-measuring device (e.g. venturi meter) permanently mounted at one end of the test section. Usually a pilot tube is inserted into the pipe to measure velocity while head loss is being recorded. If the characteristics of the pipe (in particular wall thickness) are known, a clamp-on ultrasonic gage can also be used . There are also some propeller type flowmeters on the market that can be inserted directly into the pipe through a corporation stop. Regardless of the exact method used, directly measuring flow or velocity involves excavating the pipe with the associated costs for shoring, repaving, and traffic control . Directly measuring flow does however provide the most accurate results. 98. Velocity from hydrant discharge. When the pipe test section can be isolated from the remainder of the system such that virtually all of the flow through the pipe is discharged from one or more metered hydrants, the total discharge from the hydrant can be used in the formula for C . This can be done on dead-end lines or pipes which can be valved off from the remainder of the system except for one end. (In Figure S-2, this would mean closing valve B.) This method is only workable for fairly small pipes (<18 in.) in which discharge from a hydrant(s) alone can create measurable head loss. Indirect method for determining C 99. In many cases a large pipe cannot be valved off without adversely affecting service, and the cost of excavating and tapping a pipe are prohibitive. For these cases, an indirect method for calculating C and flow, described below, can be used. 100. The problem is one of how to measure C and Q without having to excavate or tap the pipe. In order to do this it is necessary to know the 45 ------------~A~------------~--------------~------------~0---SOURCE NO. 2 HYDRANT B SOURCE NO. 1 ~ERVICE AREA Figure S-2. Methods for isolating pipe for testing head loss in the line both at a flow rate Q and when that flow rate is changed by a known amount (say by opening a hydrant). 101. The key to this procedure i s that the only way water can get to the service area downstream of the test pipe is through the test pipe. Storage tanks, other sources, and other paths must be shut off. This can be accomplished in one of two ways as shown for the sample service area in Figure S-2. Either valve B can be shut and Q can be measured at the hydrant as described in the previous section or valves A, C, and D can be shut in which case Q is the actual water use in the service area. The bottom line is that there should be no "back door" through which water could get to the flowing hydrant(s). 102. The indirect procedure for determining C is based on measuring head loss at two flow rates and the difference in flow rate (caused by opening a hydrant), and then simultaneously solving the head loss equations fur the test section at both flow rates to yield C and the flow rate in the pipe. The information that will be known is: 46 a. Diameter D of pipe (diameters if pipes are connected in series). b. Length L of pipe(s). c. Discharge from a hydrant(s) Qf in the service area. d. Head loss in length L at background flow hand while 1 hydrant is open h 2 e. A pressure reading in the service area at background flow and while hydrant i s open P. The term "background P1 2 flow" refers to the flow in the pipe when no hydrants are open. 103. The head loss in a single pipe for background and hydrant flow can be determined by rearranging Equation 5-l as (5-3) (:1)" (5-4) (Q2: Qfr where h1 = head loss background flow, f t L = length of pipe (including equi valent pipe for minor losses), ft D = diameter of pipe, in. m = exponent on diameter i n head loss equation (4.87) = flow when hydrant(s) closed, gpm Q1 C = Hazen-Williams C-factor n =exponent on flow in head loss equation (1.85) h2 = head loss when hydrant(s) open, ft Q2 = flow to service area when hydrant(s) open, gpm = hydrant di scharge, gpm Qf The decision on whether to include equivalent length of pipe for minor losses in L depends on whether the C to be determined will be for the pipe or for the pipe plus losses. If it is for the pipe only, L must include the minor losses. The unknowns to be determined are Q(and Q) and C 1 2 104. Solving Equations 5-3 and 5-4 for Qand C gives 1 47 1 C=(10.41) l/n Qf (5-5) Dmh [I -hl/h2 1/nQr] 2 and hl Dm Ql =( (" c (5-6) 10.41 where Q2 Qr = Q 1 105. The term Qr in Equation 5-5 reflects the fact that, when the hydrant is opened, flow out of orifices and nozzles in the service area decreases. In some cases the effect is offset simply by individuals opening valves on faucets wider, so that the flow is unchanged (Q = 1). The biggest r change in flow occurs if no faucet settings are changed during the test and flow can be given by the orifice equation (5-7) where q. =flow out of i-th orifice, gpm c. =discharge coefficient of i-th orifice 1 1. P. =pressure at i-th orifice, psi 1. Using an effective pressure for the service area, Equation 5-7 can be approximated by Q =Lci JP and (5-8) This is the lower limit on Q , except for some special cases in hilly terrain. r It represents the maximum impact of the test on water use. The upper limit is the case with no impact. Therefore, (5-9) 48 There is no way to know exactly but in most cases i s not too far Qr p2 from p1 so that the error involved with this uncertainty is small. Since the period during which the hydrant i s open is s hort and hence not many water users will adjust valves to maintain flow rate, Qr will be closer to ./P2/Pl than to 1. In general, can be given by Qr (5-10) where W=weighting factor (0 ~ W~ 1). 106. To provide the reader with an idea of the sensitivity of W to other values, consider a typical hydrant test when P= 70 psi and P 1 2 = 55 psi: 65/80 = 0.90. If W=1 , Q =0.90 ; if w =0 Q = 1 · and r ' r ' if w =0.5 ' Q =0.95 . It is also i mportant to note that (h /h ) 1/n is r 2 1 on the order of 3. So an error of 0 . 05 in Qr will not result in a large error in C . In general, as h /hbecomes large in Equation 5-5, the uncer 12 tainty in C caused by uncertainty in Qr becomes small. In most instances, W will be near 1 for large pipes for which the indirect method is appropriate and will decrease with pipe size. Illustrative Example 107 . The procedures described above were applied to several test problems. In each case head loss was measured using the parallel pipe method; that is, a differential pressure gage was connected between the two ends of the test section. Gages with 2 psi (4.6 ft) and 10 psi (23.1 ft) range were used in this study. The differential pressure gage is shown in Figure 5-3 and the connection with a fire hydrant is shown in 5-4. The pressure gage connected to the hydrant in Figure 5-5 measures line pressure for the indirect method. Discharge from hydrants was measured using a pitot gage. Figure 5-5 shows how the gage is mounted in the hydrant while Figure 5-6 shows the tip of the pitot gage which actually detects the velocity. 108. Ideally, the differential pressure gage can be located near any flowing hydrant so that both devices can be read simultaneously by a single individual as shown in Figure 5-7. In some of the tests, velocity in the pipe 49 Figure 5-3. Differential pressure gage Figure S-4. Pressure gage and one end of parallel pipe 50 Figur e S-5 . Pitot gage connected to hydrant Figure S-6. Tip of pitot gage 51 Figure 5-7. Measuring hydrant discharge and head loss was measured using a pitot tube inserted into the pipe. Figure 5-8 shows a pitot tube inserted into a clear pipe. 109. The following paragraphs describe the results of six tests. Each one presented different problems in applying Equations 5-1 and 5-5 to determine C. These illustrative tests should also give the reader insight into how the tests can be conducted in still other situations. Test 1 110. Test 1 was conducted for a 6-in. nominal diameter, 345-ft-long pipe located behind the WES Structures Laboratory. A plan view is shown in Figure 5-9. The differential pressure gage is connected between a fire hydrant and a 2-in. hose bib. This test is complicated by the fact that the head loss measured when the hydrant is not flowing occurs primarily in the 2-in.-diam line. This head loss of 0.3 psi (0.69 ft) was substracted from the head loss measured when the hydrant was flowing as this was the head loss attributed to the 2-in. line. 111. The 10-year-old pipe was a lined cast iron pipe; therefore, an internal diameter of 6.4 in. was used in calculating C . Head loss and hydrant discharge were measured for a wide range of discharges and the C-factor was calculated using Equation 5-l; the results are shown in Table 5-1. The average value of 118 is based on a diameter of 6.4 in. If a nominal diameter of 52 Figure 5-8 . Pitot tube inserted into pipe 6 in. was used, C would be 140. This illustrates that care must be exercised in deciding whether to use actual or nominal diameter to calculate C in cases where the two differ. 112. An important lesson learned during the test is that head loss in the piping connecting the differential pressure gage with the test section can be significant. If this is not taken into account, a lower C will be determined because head loss that actually occurred in the 2-in. line would have been attributed to the 6-in. line. Test 2 113. Test 2 was conducted in a 262-ft length of nomimal 6-in.-diam. pipe (actual diameter approximately 6.4 in.) located alongside the Headquarters Building at WES as shown in Figure 5-10. The tests were conducted on two 53 STRUCTURES LABORATORY --21N. 61N. --------_,---,--------~--------------------------~~~-21N. -------"~~R~~~E~---------------~?sSE HYDRANT) PIPE -'-..DIFFERENTIAL GAGE Figure S-9. Plan view for Test 1 days--5 May 83 and 8 June 83. One or both of the hydrants were flowed during the tests so that head loss could be observed over a wide range of flows. There was a head loss of O.I psi (0.23 ft) when no hydrants were open. This value is used as The line pressure when the hydrants were closed was hi 90 psi. II4. The results of the tests are shown in Table S-2. Since hi is Table 5-1 Results of Test I Head Loss Hydrant Discharge ft c BErn 9.5 IOlO I2I 8.9 980 121 5.4 7IO 114 120 I0.3 1100 I25 1.8 4IO 3.9 570 110 116 Ave. 118 0.4 175 54 ----------------------------................. HEADQUARTERS BUILDING PARALLEL 61N. PIPE--_.. I I I I I I I I I I I I DIFFERENTIAL.. GAGE • Figure 5-10. Plan view of Test 2 Table 5-2 Results of Test 2 Head Loss Line Pressure Hydrant Discharge c c ft ESi BErn (Eq 5-1) (Eg, 5-5) 14.0 45 820 107 113 7.2 57 556 104 115 5.5 75 500 108 126 14.5 35 960 123 127 13.6 so 820 108 116 21.4 30 1105 114 118 15.0 35 960 121 126 5.7 74 530 112 130 13.6 42 920 123 130 12.2 45 840 118 125 5.5 75 480 103 121 20.5 37 1060 112 117 Ave. 113 122 55 fairly small, one is tempted to ignore that head loss and use Equation 5-l. The C-factor was calculated with Equation 5-1 which ignores hand Equa 1 tion 5-5 which accounted for hand the results, shown in Table S-2, indi 1 cate that neglecting the head loss caused by flows other than the hydrant, even if those flows are small, can cause a significant error. Using an average C of 122 and Equation 5-6 shows that a flow as small as 106 gpm (1 ft/ sec) caused the difference. Being able to calculate Qusing Equation 5-6 1 is an extra benefit if using the indirect method for determining C . 115. The spread in calculated C values is due to both measurement error and variation in water use other than through the hydrants. Hydrant discharge readings from a pitot gage a r e the least accurate of the data values as the needle vibrates considerably (on the order of 100 gpm) so that obtaining an ac~urate reading is difficult. Since most of the parameters were measured to two significant digits, it is improbable that C values would be reproducible beyond the first two digits . The second source of error is due to the fact that it is impossible to control other water users on the line and changes in their use will change head loss and flow during the test. Since a constant value was used for h in calculating C (when actually h varied 1 1 slightly), some error will occur. 116. The importance of using accurate gages was illustrated during this test as the C values calculated varied greatly depending on which of two pitot gages was used to measure hydrant discharge. When the gages were tested, one was shown to read low. Readings taken with this gage were adjusted and the calculated C's became consistent. Test 3 117. Test 3 was conducted for a 2-in. line feeding some hose bibs in a grassy area at WES. The 120-ft-long test section is shown in Figure 5-11. Since the flows from the hose were small, the discharge was measured with a calibrated bucket and stopwatch. In this test, there was no flow in the line except the controlled discharge through hose 1 and hose 2, so there was no need to use Equation 5-S or correct the head loss. The results are shown in Table 5-3. 118. Because of the simplicity of the problem, it is possible to illustrate how Q1 and C could be determined using Equation S-5. Suppose the J flow (Q =17.0 gpm) was not known. Instead, all that was known was that when 1 an additional 16.3 gpm was discharged, the head loss changed from 0.85 ft to ~ 2.93 ft. Equation 5-5 gives 56 j ..___ -~DIFFERENTIAL GAGE ' '~PARALLEL PIPE ',~ Figure 5-11. Plan view of Test 3 Table 5-3 Results of Test 3 Head Loss Hydrant Discharge ft gpm c 0.85 17.0 141 2.93 32.5 139 c = (10.4) (120) l6.3 87 [ (2.93)24 · ] 0.54 [ 1 -Qr (0.8~/2.93)0 · 54] (5-5, bis) In this case Qr is 0.96 (i.e. 16.3/17.0) which would give a value for C =137 and Q=16.6 gpm. This demonstrates that the indirect method for 1 determining C and c, as embodied in Equation 5-S, will produce accurate 2 results when Qr is estimated accurately. A more important point is that 57 Qr is usually not known. If it was estimated incorrectly (say 1.0, 0.9, or 0.8), the values of C would be 142, 128, and 117 which shows that, even for small errors in Qr (<0.05), C is reasonably accurate. Usually one would not apply this method to cases with Qr less than 0.9 since reducing pressure by this amount can adversely affect customers. 119. The C-factors observed for this pipe were fairly high. This indicates that the pipes are quite smooth and/or the actual diameter is slightly greater than nominal diameter. Test 4 120. Test 4 was conducted on a 229-ft length of old, 6-in. cast iron pipe located on Security St. in Vicksburg, Miss . A plan view of the test is shown in Figure 5-12. This test di ffered from earl ier tests in that: (a) velocity was measured using a pitot tube inserted into the pipe and (b) the downstream end of the parallel pipe was connected to the test section through a meter box instead of a hydrant or hose bib. HYDRANT .... .... PARALLEL"'~ PIPE~-'-.,,\ ........ ........ ..... DIFFEREN~~:... GAGE FLOWED HYDRANT Figure 5-12 . Plan view of Test 4 121. Since the pipe was a dead-end line with virtually no flow, it was necessary to open a hydrant on the downstream end to cause sufficient velocity and head loss to produce reasonable readings. In order not to waste water or cause any flooding, the hydrant was only flowed for a short period of time. Because the high velocity was maintained only briefly, a four point velocity traverse was made instead of ten points as is customary. An average velocity of 5.25 ft/sec was observed when the head loss was 5.54 ft. Substituting this information into Equation S-1a gives a C-factor of 58 54 63 c = (8.70)(5.25)(229) 0 · (5.54)-0 · 54(6)-0 · (5-1a, bis) =110 122. This C-factor is fairly high for an old, small diameter pipe. This may be due to the fact that the test was not run for a sufficiently long period of time for the differential pressure gage with pressure snubbers to reach steady-state. Hence, the pressure loss recorded may be lower than actual and the C-factor may be too high. This points out the need (a) to run head loss tests for a long enough period of time for gages to respond, and (b) to select snubbers that do not excessively dampen the response of the gage. 123. Another lesson learned during this test was the need to use highpressure hoses and tubing. The line pressure of 100 psi ruptured two hoses connecting the pitot tube and manometer. Only when these hoses were replaced by braided tubing could the test be completed. Test 5 124. Test 5 was conducted on a 424-ft length of 16-in. cast iron pipe located on Lee St. in Vicksburg, Miss. The line is a dead end and serves a mill as shown in Figure 5-13. The test was originally conducted with the parallel pipe connected between the hydrant and the hose bib, located downstream of the meter on the mill property. The head loss in the 6-in. hydrant lateral, meter, and 2-in. service line for the mill was so large it caused the differential pressure gage to read ful l scale even when there was virtually no head loss in the 16-in. pipe. Since the 16-in. pipe was the one to be tested, a tap was made in the 16-in. pipe immediately downstream of the pitot tap so that the head loss readings would not be affected by losses in the 2-in. line and meter. In this way only head loss in the 16-in. pipe would be measured. 125. A pipe caliper was used to measure the internal diameter of the pipe which was found to be exactly 16 in. The flow was measured both with the pitot tube and at the discharge from the hydrant. Because the line pressure in this part of town is in excess of 100 psi and the hydrant is tied in to a 16-in. line, it was possible to produce significant velocity and head loss by only opening one hydrant. 126. When the hydrant was fully opened, the head loss was 2.08 ft and the velocity was 2.7 ft/sec measured with the pitot tube (which corresponds to a flow of 1680 gpm in the pipe). Discharge from the hydrant was measured as 59 FLOWED HYDRANTS MILL Figure 5-13. Plan view Test 5 ~ 1720 gpm which, when added to the 60-gpm use at the mill, gives a pipe flow of 1780 gpm, a 5 percent difference. The C-factor was calculated using Equation 5-1 to be 76 based on 1780 gpm and 71.5 based on 1680 gpm (use 73.7 based on 1730 gpm). 127. As a check of the results, the test was repeated with the differential gage replaced by a manometer. Because of the limited range of the manometer, the discharge from the hydrant was limited to 1185 (1245 gpm in pipe) which resulted in a head loss of 1.0 ft. Substituting this into Equation 5-1 gives a C of 79. 128. This test points out the fact that, for rough pipe flow, the C-factor will depend slightly on the velocity (or flow) at which it was measured. Walski (1983b) stated that C is inversely proportional to velocity to the 0 . 15 power for hydraulicly rough pipes. That is 60 ~~ ;G~) 0.15 (5-11) Using C =73.7 at 1730 gpm for the high flow test and C = 79 at 1185 gpm, 15 0 15 gives c;c= 73.7/79 =0.93 and (Q /Q )0 · = (1185/1730) · = 0.94 . 12 2 1 These data tend to support Equation 5-11. The other pipes tested in this work could be categorized as being in the transition zone between hydraulically rough and smooth flow . In those situations C would decrease with increasing velocity but not as dramatically as predicted by Equation 5-11. Test 6 129. Test 6 was conducted on a 700-ft section of 16-in. cast iron pipe along Eisenhower Drive in the Arlington National Cemetery (Figure 5-14). Head loss was measured by the parallel pipe method with a manometer rather than a differential gage. One valve had to be closed so that the pipe could be made a dead-end line. The flow fluctuated during the test because of a booster pump and hydropneumatic tank which were located downstream of the test section. Since it was not possible to control flow in the test section, the indirect method for calculating C was used. BOOSTER PUMPS 10/N. /MANOMETER ________..,__ L PARALLEL -------PIPE -----.... ~--......~ Figure 5-14. Plan view of Test 6 130 . A significant problem in measuring head loss resulted from a leak in the hose bib at the downstream end of the test section. This meant that pressure detected at the downstream side of the manometer was lower than the 61 actual pressure at the downstream end of the line. It was necessary to correct the manometer readings for this leak. Flow in the line could only occur in one direction so the lowest flow rate was zero. Where flow was zero the head loss would be zero as the manometer reading would be due only to leakage. This would correspond to maximum negative reading on the manometer which was 12 in. (1 ft) of water column. Therefore, 1 ft was added to each manometer reading to correct for this leakage. 131. Manometer readings fluctuated from 0 to 0.73 ft when the hydrant was not open. The average was 0.45 ft which is used as hWhen the by 1 drant was opened, its discharge was 1090 gpm and head loss increased to 1.97 ft. The C-factors (and flow when hydrant is closed) determined for Q =1.0, 0.95, r and 0.90 are shown in Table 5-4. 132. Because there is considerable uncertainty concerning the magnitude of the correction due to leakage in the parallel pipe, a sensitivity analysis was conducted to determine the effect of different corrections. The correc tion was changed from 1 ft to 1.2 ft and 0.8 ft and it was found that this estimated value significantly affects C and Qas shown in Table 5-4. 1 Therefore, very little confidence can be placed on the results of this test. The lesson learned is that no leakage can be tolerated in the parallel pipe. Table 5-4 Results of Test 6 Correction to c and (Flow Q1 gem)Manometer Qr = 1 Qr = 0.95 Qr = 0.90 Reading 1 ft 1.0 114(890) 110(859) 106(827) 1.2 102(580) 99(563) 97(551) 0.8 124(1181) 118(1124) 112(1067) Common Pitfalls 133. While the procedure for measuring C-factors appears fairly simple, there are a number of problems that can occur to reduce the accuracy of the results. Some of these pitfalls include: (a) leak in parallel pipe, (b) other sources of head loss, (c) fluctuations due to surges, (d) selecting differential pressure gage with wrong scale, (e) insufficient change in flow, and (f) lack of correct fittings. 62 Leak in parallel pipe 134 . If there is a leak in the parallel pipe used in the parallel pipe method, the head loss measured by the differential pressure gage will be inaccurate. Since a fairly small parallel pipe is usually used, a small leak can produce a significant error. No leaks should be tolerated along the parallel pipe. Other sources of head loss 135. Usually the ends of the parallel pipe are connected to the pipe test section at hydrants, and it is a safe assumption that there is no head loss between the test pipe and the hydrant. In some cases, however, hydrant laterals are used as service lines and may have flow (with its associated head loss). This problem is even more significant if head is measured at a point along a small service line. Any flow in that line will produce head loss that will make the differential pressure measurements useless. This can be avoided by making certain there is no flow in these lines. Fluctuations due to surges 136. The differential pressure gage or manometer readings will fluctuate fairly wildly in most systems due to surges which regularly occur in most systems. This makes obtaining an accurate reading of head loss difficult. These fluctuations can be dampened out by installing pressure snubbers on both sides of the gage. The snubbers should be selected so that the gage can respond to a change in pressure in about 10 sec. Since flowing a hydrant wastes water and poses potential flooding problems, a gage should be able to respond to changes in flow in a fairly short period of time so that hydrants need not be left open long . Selecting differential pressure gage with wrong scale 137. The pipe diameter is fixed, the length of the test section is determined by the layout of the site, and the flow rate can only be controlled to a limited extent during the test. Selection of a gage with the correct scale is therefore essential to obtaining accurate results. To select the scale of the gage, solve Equation S-3 for head loss as 63 10.4 L (~) 1.85 h = (5-12a) D4.87 1.85 54.7 L (~) h = (5-12b) Dl. 16. Using a high estimate for Q and V and a low estimate for c· will give an upper limit for head loss and, hence, the differential pressure gage that can be used. It is advisable to have several differential pressure gages on hand, each with a different range (e.g. 0-2 psi and 0-10 psi). Insufficient change in flow 138. For the indirect method of determining (Equation 5-5) C to be accurate, the hydrant discharge Qf should be significant compared with Q 1 This will make the h2/h1 term in the denominator of t he formula large, which reduces the effect of error in estimati ng Q . For small pipes, it is ade r quate to open one hydrant, but for larger pipes (>18 in. ), two or more hydrants may need to be opened . Lack of correct fittings 139. When conducting a test in the field, one encounters a wide variety of fittings (hydrant outlet, hose bib, meter box, tubing, etc.), threads (machine, hose, hydrant-male and female), and sizes (from 4 in. to 1/4 in.). An entire test can be ruined if one connection is not available. Careful planning before a field trip is necessary to ensure that the right fittings are available. Swnmary 140. The procedure for conducting head loss tests must be tailored to fit the conditions under which the test is being run. Head loss should be measured using the parallel pipe method. Flow in the pipe can be most accurately measured using a pitot tube, although this means the pipe must be excavated and tapped. If the pipe can be isolated properly, the flow can be measured at a hydrant. If this is not possible, some indirect methods exist to measure flow and hence C 141. Flow measurement is usually the largest source of error in the tests, although there can be significant problems in measuring head loss if 64 the measurement is not made carefully. Confusion can arise in using C-factors if the individual reporting the test results does not specify whether the results are based on nominal or actual diameter (if available) and equivalent or actual length. In general, nomimal diameter should be used since this may be all that will be known when the C-factor is used, and test sections should be selected so that minor losses are negligible. 142. Head loss tests can be conducted accurately at a reasonable cost if care is exercised in planning, conducting, and reporting the tests. 65 PART VI: WHY CALIBRATE WATER DISTRIBUTION SYSTEM MODELS? Introduction 143. Water distribution system models, whether they depend upon the traditional Hardy-Cross method or more sophisticated techniques, provide essential information on pressures and flows that may be used in evaluating existing systems, or designing new or enlarged systems. In applying these, or any other models to a specific system, it is necessary to make certain that the model used actually describes the behavior of the existing system before using it to analyze different conditions. The process of checking flows and pressures predicted by the model against field observations and adjusting model values to make the results agree is known as "calibration." 144. Distribution system models are fairly easy to calibrate by adjusting pipe roughness (or Hazen-Williams C-factor) and water use estimates to make the model behave like the real system. There is, however, a temptation to use haphazardly gathered sets of pressure readings or, worse yet, "litera ture values" for C-factors or water use, and to then claim the model is cali brated. This type of "calibration" can lead to serious problems when the model is used to design pipes and pumps. These problems are illustrated below and are followed by a discussion of how to avoid common pitfalls and how to answer someone who claims calibration with good data is unnecessary. Scenario (Can This Happen in Your System?) 145. Suppose a new subdivision is being built at the end of Elm St. in Town X, so an expansion to the water distribution system at Elm St. is needed. Town X has a model developed to study the expansion and some other problems. Instead of gathering data to calibrate the model, the consulting engineer Z asks the town engineer, "What is the pressure at the end of Elm St.?" and is told, "about 65 psi." Assuming a Hazen-Williams C-factor of 100 and the fact that the nearest storage tank is half full when the pressure is 65 psi, engi neer Z uses the model to predict a pressure of 70 psi at Elm St. He then concludes, "The model works," and begins sizing the expansion. (Do not laugh, readers; this really happens.) 146. If engineer Z had actually measured the pressure at the end of 66 Elm St., the results would have shown the pressure to be 60 psi. Furthermore, if engineer Z had opened up a hydrant there and measured pressure, the Cfactor could have been back-calculated to be 80. But engineer Z does not want to bother opening up hydrants and measuring pressures (after all, the consulting firm has a computer program to calculate pressures), so engineer Z specifies 3000 ft of 10-in. pipe and says "the pressure at peak use will be 62.1 psi." 147. A decade later, the 10-in. line is in the ground, the subdivision is built, on a good day the pressure is 40 psi, and the fire insurance inspectors are talking about raising insurance premiums in the subdivision. The town or facility engineer loses sleep and can only say, 0 1 do not understand what happened, the consultants used a computer model to size that line." 148. What happened? There is a saying among computer professionals-"Garbage in, garbage out." There was probably nothing wrong with the computer program that was used. Engineer Z simply did a haphazard job calibrating the model. To avoid the problem, engineer Z could have (a) specified an oversized pipe to be on the "safe side" even if it did cost the utility twice as much, or (b) carefully calibrated the model over a range of flows using good quality data. Calibration Data (How Good is Good Data?) 149. The data collected for calibration must be accurate and should be gathered in an organized manner. Some rules are given below. 150. Flows and pressures in a water distribution system change over time, but in order to back-calculate C-factors and water use rates for a model, it is necessary to get a snapshot of the system at one point in time (i.e. one set of boundary conditions), knowing which pumps were running and the current water level in the tanks. This means that data for calibration should be collected over a short period of time, and settings of pumps and valves should not be changed during the data collection period. 151. A model must accurately predict pressure (actually static head) at specific points in the system. The observed heads must be accurate if the model is to be of any value. Pressures should be read with a type A, or better, gage (i.e. corresponds to American National Standards Institute B40.1-1974, which specifies an accuracy of +1 percent of full scale for 67 type A gages). For the observed head to be meaningful, the elevation at which the reading is taken must be accurately known. If the best available contour map has contour intervals of greater than 5 ft, then it may be worthwhile to survey the elevations of the point where the pressure reading was made. 152. If only static pressures are measured, it is virtually impossible to determine whether C-factors or water use estimates should be adjusted to achieve calibration. (It is like trying to decide which knob to turn to adjust the color on a TV set.) Engineers can argue at length about the merits of adjusting C-factors and water use rates, but actually the correct parameter to adjust is the one in error in the initial runs. This problem can be circumvented by dramatically changing the flow rate and observing how the pressures change. The easiest way of doing this is to open up a nearby fire hydrant and measure the discharge with a pitot gage while, at the same time, observing the pressure. 153. If the model can predict pressures at high and low flow rates, considerable confidence can be placed in its capability to simulate future uses, shutdowns, or expansions. On the other hand, by failing to calibrate the model over a . range of flow rates, a careless individual can make the model appear to be calibrated when actually errors in C-factors and water use rates are merely compensating for one another. Adjusting the Model (How Do You Use the Data?) 154. It is not unreasonable to expect the model to predict pressures to within !3 psi (!7 ft). Eggener and Polkowski (1976) showed this to be possible when the data are of good quality. The obvious question is, "By how much should each parameter be adjusted?'' This can be determined using some basic hydraulics, common sense, and trial and error, but Meredith (1982) and Walski (1983a and 1983c) give some procedures for zeroing in on the right C factors. Simply stated, if the error in calibration is the same at high flow and low flow, the water use estimates need to be adjusted. If the error is worse at high flow, the C-factor should be adjusted. Excuses, Excuses, Excuses .... 155. If it's so simple to calibrate a water distribution system model 68 properly, why don't more people do it? This author does not understand why, but here are a few excuses, and some suggested replies. 156. "It takes a lot of time and costs the client a lot of money." Actually, it takes less than a day for a small system. A larger system can be subdivided into several smaller ones and studied over a couple of days. Considering that a modeling study takes 1 to 3 man-months, this extra time in data collection is a small price to pay for correct results. The testing provides the engineer with the opportunity of seeing the system first-hand. The only problem is that the engineer's shoes may get wet when measuring hydrant discharge. 157. "We don't have any pressure gages or pitot gages." All of the gages needed to collect this type of data will cost only a few hundred dollars. If the gages are handled properly, they will last almost forever. 158. "Literature values for C-factors are good enough." Definitely not. The California American Water Works Association Committee on Loss of Capacity in Water Mains (1962) found 30-year-old, 6-in. cast iron mains to have Cfactors ranging from 40 to 75. Lamont (1981) states that 60-year-old, 24-in. cast iron mains can have C-factors of from 56 to 107. This amount of uncertainty is unacceptable . Remember also that head loss varies as a function of C to the 1.85 power, meaning that an error of 20 percent in a C-factor will result in an error of almost 40 percent in head loss. 159. "The utility has data in its records." Even the best run utilities only have a rough idea of the pressures and flows in the system. They cannot provide the modeler with the "snapshot" of the system needed for model calibration. If given precise guidance, utilities can collect the data, but the engineer should be wary of gages provided by the utility if they have not been recalibrated recently. These gages may have been mishandled (e.g. tossed into the back of a pickup truck), changing the accuracy from !1 percent to+ (you name it). Be cautious about accepting data from the utility without checking its validity. The utility may say, "That pressure reducing valve is set at 60 psi" and it may actually be set at 50 psi, or "The gate valve at First St. is always closed," and it may have been open for the last 3 years. This author has found some "surprises" in virtually every study with which he has been involved. These surprises can only be discovered by field observations. 160. When a utility is hiring a consultant to do modeling work, the 69 utility should get assurances that the model will be calibrated over a range of flows with good quality data. If the consultant gives excuses instead, the utility should look for another consultant. Summary 161. Mathematical models of water distribution systems can only be considered calibrated when they can predict pressures and flows over a range of water use rates. The pressure data used must consist of accurate information collected over a short period of time, and not a set of haphazard observations . 70 PART VII: PROGRAM FOR REDUCING PITOT TRAVERSE DATA Purpose 162. In measuring flow through a pipe using a pitot tube, the engineer must read the difference in pressure between the two tips of the pitot tube at a number of points in the pipe, convert these readings to velocities, and then integrate the velocity over the pipe cross section to obtain the flow. These calculations are fairly tedious and have therefore been computerized in a program called PITOT written in Fortran IV. The program is described in the following paragraphs. Those wishing to use the program should contact the author of this report at (601) 634-3931 or FTS 542-3931. 163. This program calculates the flow in a pipe given differential pressure readings and the location of the reading as measured from the bottom of the pipe . The program can accept up to 25 readings from a differential pressure gage (psi) or manometer (in.), and converts velocity to flow using both a 10 point equal area method and integration of either a power law or normal (semi-log) law velocity profiles. 164. The data and format required by the program include: a. Title of run (25A1). b. Type of differential gage (IUNIT =0, pressure gage; IUNIT = 1, liquid manometer; IUNIT =2, air-filled manometer); if liquid manometer is specified, the second value on the data line is specific gravity of manometer fluid, while if air is specified, the second value is the line pressure in pounds per square inch (I1, F5.0); c. Diameter of pipe (inches) and pitot tube coefficient (2F5.2); d. Distance from bottom (inches) and reading (pounds per square inch or inches) (2F5.0) for each point in the traverse, terminating with negative number. Calculating Velocity 165. The velocity can be determined from differential pressure using the following formula 71 v=k ~ (7-1) where v = velocity, ft/sec k = factor for pitot tube 2 g = acceleration due to gravity (32 ft/sec ) Llli = difference in pressure head, ft For a differential pressure gage in which the readings are in pounds per square inch (1 psi= 2.31 ft), Equation 7-1 becomes for the pressure gage: v = 12.2 k ..Jl;.P (7-2) where ~ = differential pressure, psi . For a liquid manometer, the head is expressed in inches of displacement of the water columns d , so Equation 7-1 becomes (for a heavy liquid (7-3a) then for air (7-3b)): v =k /2gR(SG-i) =2.32 k ~R(SG-1) (7-3a) " 12 (7-3b) where R = displacement of manometer column, in. SG = specific gravity of manometer fluid P = line pressure, psi 166. The program can be modified to accept different types of data by adding new values of IUNIT to accept data in metric units for examp~e. This part of the program produces a velocity value for every differential pressure reading. Determining Flow 167. The next step is to convert the individual voint velocities into the flow and average velocity. This is done by integrating the vei~city over the area using 72 (7-4) Q =J vdA A where Q = flow, ft3/sec 2 A = pipe cross-sectional area, ft There are two overall approaches to determining Q The first is to approximate the integral in Equation 7-4 with a summation over small annular areas. The second is to convert the velocities into a continuous function and integrate the function analytically. 168. In the following sections, the approximation techniques are first used to numerically integrate Equation 7-4 followed by the analytical integration for two velocity functions: power rule and normal rule. Numerical integration 169. The integral in Equation 7-4 can be replaced by n n Q = L = L v. 1 M. 1 (7-5) i=1 i=1 where V . = average velocity in i-th annular area, ft/sec R. = outer radius of i-th annular area, ft 1 1 = inner radius of i-th annular area, ft R. 1 1-2 M. = area of i-th annular area, ft 1 If the R. values can be selected in such a way that all of the M. 1 values 1 are identical, then it is possible to replace Equation 7-5 by n Q = A V. (7-6) n L 1 i=1 In order to use Equation 7-6, the V. values must be measured at specific 1 locations in the pipe. When n = 10, these dimensionless locations can be given as follows: 73 Fractional Fractional Distance from Distance from i Bottom of PiEe Center of PiEe 1 0.021 0.958 2 0.082 0.838 3 0.148 0.704 4 0.228 0.544 5 0.349 0.302 6 0.651 0.302 7 0. 772 0.544 8 0.852 0.704 9 0.919 0.838 10 0.979 0.958 170. Since the velocity readings are not taken at precisely the loca tions given above, it is necessary to interpolate using velocities at nearby points V. This is done with the following formula J d.) J (7-7) d. J where v. =velocity at i-th plotting position, ft/sec 1 v. = velocity for j-th observation, ft/sec J velocity for j+1-th observation, ft/sec vj+1 = d. = distance from bottom of pipe for i-th plotting position, in. 1 d. = distance from bottom of pipe for j-th observation, in. J = distance from bottom of pipe for j+l-th observation, in. dj+1 and d. 1 is close to dj and dj+1 Note that the j subscripts refer to the actual observations while the i subscripts refer to the interpolated values for velocity based on Equation 7-7. Once the v. terms are calculated, they can be substituted into Equation 7-6 1 to give flow. Integrating EOWer rule 171. The velocity profile in a pipe in fully developed turbulent flow can be approximated by a power function of the form v(r) =a(R -r)b (7-8) 74 where r = distance from center, ft a,b = regression constants R =pipe radius, ft The a and b can be determined by plotting (R -r) versus v on log-log graph paper. Such a plot is a straight line and, for any two points on the line, a and b can be given by (7-9a) (7-9b) In the program, a and b are determined in subroutine CURVE, given a table of log (R -r) and log v from the actual velocity profile, using a linear regression formula. 172. To determine flow, it is necessary to insert Equation 7-8 into Equation 7-4 and integrate 2n R Q = f f b a (R -r) r dr d8 (7-10) 0 0 This yields 27t R r)b+2 R(R-r)b+l] [ (R -d8 (7-11) a (b + 2) (b + 1) Q =! 0 0 which gives Q = 27t Rb+2 a [ 1 1 J (7-12) b+1-b+2 Equation 7-12 is evaluated in the program to give Q . Integrating normal rule 173. Another way of describing the velocity profile in the pipe is a normal velocity profile which can be expressed as v(r) = a + b log (R -r) (7-13) 75 The a and b can be determined by plotting (R -r) versus v on semi-log graph paper (R-ron log scale). From any two points on the straight line, a and b can be determined as (7-14a) (7-14b) In the program, a and b are determined using regression from the sub routine CURVE given log (R -r) and v . 174. To determine flow, it is necessary to insert Equation 7-13 into Equation 7-4 and integrate 2n ~R-t: . [a Q = + b log (R -r) rdrd~ (7 -15) J 0 0 where t: is very small. The upper limit is R -t: since Equation 7-13 is undefined for r =R(O < t: < < R). This yields 271 2 2 Q =f rar + bh_-r -(-R) 2 log (R-r) 0 [ 2 \2 (7-16) which gi ves, taking limit as £ approaches 0, (7-17) which is the formula included in the progfam. Note that the logarithms in the program are base e logarithms. Conversi on factors 175. Flow is determined in cubic .feet per second in the above formulas. It is converted to million gallons per day using 1.54 (ft3/sec)/MGD, which is converted to gallons per minute using 694 gpm/MGD. The average velocity in 76 feet per second is determined by dividing the flow in cubic feet per second by the pipe area in square feet. Goodness of fit The subroutine CURVE prints two indicators of goodness-of-fit, 176. the correlation coefficient (R) and the standard error of the estimate (SE). These pertain to the linear transformation of the original velocity and location data. For the goodness-of-fit to be acceptable for either the power or normal laws, R should be greater than 0.9 and SE should be less than 0.1. 177. The methods for determining flow which rely on curve fitting should only be applied when the equation actually fits the observed velocity If the model is not very good (e.g. near an obstruction or bend, or profile. simply because of unusually shaped roughness in the pipe), then only the solu tions corresponding to interpolation of the actual velocities should be used. Program and Variable Listing 178. A listing of the PITOT program is given in Table 7-1 while a listing of all the variables in the program along with definitions and units is given in Table 7-2. Example Problems 179. Two example problems are provided to illustrate use of the proThe first consists of six observations taken in a gram (Figures 7-1 and 7-2). 20-in.-diam pipe with a differential pressure gage attached to the pitot tube. Note that the velocity profile is symmetric since the input data is symmetric. Note that there are some differences in the average velocity because of the way the three methods interpolate the data. 180. The second example uses actual data from a 16-in. pipe taken with a liquid manometer with a specific gravity of the indicating liquid of 1.27. Because there are more readings, the results from the three methods are in closer agreement. Output The output from the program, shown in the example problems, con 181. sists of several tables which are described as follows: 77 a. Title. b. Pipe diameter and pilot coefficient. c. Type of manometer or gage and specific gravity or line pressure. d. Table with the distance from bottom, center, and pipe wall, and velocity for each observation. e. Table with velocity and distance from bottom corresponding to interpolated points. f. Average velocity and flow by interpolation and summation. g. Coefficients (a and b), correlation coefficient (R), and standard error (SE) for power law approximation. h. Average velocity and flow by integrating power law. i. Coefficients (a and b), correlation coefficient (R), and standard error (SE) for normal law approximation. i· Average velocity and flow by integrating normal law. 78 Table 7-1 Listing of Computer Program PITOT 0 HOGRA M PITOTCI~PUT,CUTPUT,TAPES=INPUT,TAPE6=CUTPUT) C THIS PROGRA~ CALCULATES FLOW IN A PIPE FROM ?I~C T TESTT ~ATA C ThE FLOW IS CALCULATt:O USING A TEr• ?OINT E~U~L AREA fviETHCD, C A~ INTEGRATION OF THt: POW~R LAW, AND TYE NO~MAL LA~ OF THt WALL. 5 C THE DATA RE~UIRED INCLUDES: C 1. A TITLE, 2.A 0 IF DIFFERENTIAL PKESSUR~ GAGE IS USED AN D A 1 AND C TH E S?ECIFIC GRAVITY OF THE ~ ANO~ETER LIWUID, C A~D A 2 IF AND AIR/WATER MANOMETE~ IS USED C 3 • f t1 E DI Arr, E T E k I . ·~ TE k r'l A L 0 F TH E P I PE ( I •II • ) , 1 Cl C 4. ON E LIN~ FOR EACH R~ADlNG GIVING THE DISTANCE FR0M THE 8 0iTC M (IN) C .; r·~ D P RESS U R E ( P S I ) u~ ~·i A ~ ~ 0 rv. i:: T E R R E A D I N G ( I N ) , TERMI NAT ED wI T H A C NEGATIVE ~uvaER. THERE MUST ~EAT LEAST ·4 AND NO MOHE THAT 25 READINGS. C IT I~ POSSI ? LE TO RUN SEVEKAL DATA SETS AT A TI ME. -..J DIME.\J.3ICI\ RX(25),DXC25),P~(1Q),V/..(2)),V(1Q),P R cS(25) \0 1 5 1 , XL V ( 25 ) , ;< LS ( 2 5 ) , D ( 1 u ) , .:) ( 2 5) , I T I T L E ( 2 5 ) DAT .ll P D I . 0 2 1 , • Q8 2 , • 1 4 6, • 2 ~ 8, • 3 4 9, • 6 51 , • 7 7 2, • ~ 52, • 91 9, • 't 7 9 I PI2=2.*3.141~ c:_, READC5,6)ITITLE 6 FOR:"1AT<25A1) 20 IF.llT.EQ.1)WRITE(6,4)SG 4 F 0 R rt1 AT ( 2 2 H LI~U I ) ,>1 A{lj 0 oH TER, S G =, F 6 • 2) WRITE(6,9) 40 ~ FORMATCSX,25H DX,I~ RX,FT S,FT VX,FPS) RAD=DIAM/2 00 c 0 c RA~ DATA ~ANIPULATIONS c 45 DO 10 I==1,2S REA D ( 5 , 11 ) D X ( I ) , PK ES ( I ) 11 FOR~AT<2F5.J) so IF(DX(I).LT.O)GO TO 19 I F (I U1'41 T • E Q • 0) V X (1 ) =1 2 • 2 * X K * ~ .,. k T ( t-' R E S ( I ) ) I F (I UN I T • E Q • 1 ) V X (l)=(RAD/12)-k,( (!) IF(S(l) .LT.0.001 )~(I)=.001 (Continued) (Sheet 2 of 5) Table 7-1 (Continued) XLSCI)=ALOG(S(l)) Wf-1 IT E ( 6 , 1 2 ) I , D X ( I) , R X ( I ) , S ( I ) , V X ( I ) 60 12 FCR~AT(I4,4F6.2) 10 CONTINUE I=I+1 c c VELOCITIES AT PLOTTING PC~ITI0~5 65 c 1 <;-NR=I-1 I~ ~1 =1'4 R-1 IF(NR.LT.4)G0 TO 31 WRITE(6,13) 70 1 6 FOR~AT(I5X,14H POS,IN V,FPS) AVc=O ;)8 20 1=1,10 D(I)= PD (I) *D IAI''i CXl 1-' IF(D(l).LT.IJX(1J)G0 TC 21 75 IF(D(l).C:T. DX ([\jR))GO TO 22 DC 23 J=1,NR1 IF(D(I).GT.DX(J+1))GO TO ~3 2 5 V (I ) =1/ X ( J ) + ( V X ( J +1 ) -V'I ( J ) ) * (;) (I ) -DX ( J ) ) I ( DX ( J + 1 ) -DX ( J ) ) GO TO 27 80 2.3 CONTi i•U€ 2 2 1/ ( I ) ::. IJ X ( N R ) + ( 'J X (1\ K ) -V X ( ,~ 1\ -1 ) ) * ( D ( I ) -D X ( ~~ r\ ) ) I (DX (N K ) -D X (-N R -1 ) ) GO TC 27 21 'J (I)=V X (I)+ ( V X ( 1)-V X ( 2) ) * ( D(I) -D X( I) ) I ( D X ( 1 ) -D X ( 2)) 27 WRITE(6,26)I,D(I),V(I) 85 26 FORMAT(I4,2F6.2) AVE..=AVb·V(I) 2 0 CC~-T I N U E AII~=AIIEI10. 1 ~CF~=AVE*.~REA c;o ~,VJ:;D=QCFS/1 .54 (Continued) (Sheet 3 of 5) Table 7-1 (Continued) ~3P~=J~G)*694. WRIT~(6,29)AVE,QCFS,~~G),JGP~ 29 FCR~ATC3H AVE V~L ,F8 .2,4H FPS/5H FLO~/F8.4,4H CFS/ 1F8.4,4H MSD/F8.2,4H GP M) 'iS c C I~T~G~A7I~G ~~7HC~ c 31 WKIT~Co ,32J 3 2 F 0 "i•i AT ( 11 H PC -..J E r< R U i.. E ) 10J ~: F=f<.AD/12. CALL CGr Y2= J DO 1 I=1,N 1 0 SY=SY+Y(l) SYX=SYX+~(I)*X(l) SX=SX+X (I) SX2=SX2+X(!)**2 SY2=S Yt:+Y( I) **2 CONTl~UE 1 5 EN=~ (X) w A=(SY*SX2-SX*SYX)/(N*SX2-SX**2) :: = ( i~ * S Y X -S X * S Y ) I (N * S X 2 -SX ** 2 ) R= ( N* SYX-SX* SY) I ~J f<. T ( C~ *S X 2-SX** 2) * ( N* SY2-S Y * *2) ) 20 SE=~JRT((SY2-A*SY-9*SYX)/E~) WkiTE(6,2)A,o,R,~~ 2 FORMAT(3H RETLJRt-. A=,F8.~/3H E=,F~.2/3H R=,F3.4/4H SE=,F8.4) E;·~ D (Sheet 5 of 5) Variable A AREA AVE B DIAM DX EN I TITLE IUNIT NR NR1 p PD P12 PRES QCFS QGPM QMGD R Table 7-2 List of Variables in Computer Program PITOT Definition Units Regression coefficient in A + Bx ft/sec Pipe cross-sectional area ft2 Average velocity in pipe ft/sec Regression coefficient in A + Bx Pipe diameter in. Distance from bottom of pipe in. Floating point value of NR Counter on loops Title of run 0 if differential pressure psi 1 if liquid manometer reading in. 2 if air manometer reading in. 1 Number of observations NR-1 Line pressure psi Dimensionless plotting point for 10-point profile expressed in distance from bottom Differential pressure psi if IUNIT = 0 in. if IUNIT or 2 = 1 3 Flow ft /sec Flow gpm Flow MG/day Correlation coefficient (Continued) (Sheet 1 of 3) 84 Table 7-2 (Continued) Variable Definition Units RAD Pipe radius in. RF RAD/12 ft RX Distance from center to observation ft s Distance from wall to observation ft SE Standard error of the estimate SG Specific gravity of manometer fluid sx N I X. 1. i=l I SX2 N 2 X . 1. i=l SY N I Y. 1. i=l SY2 N I y~ 1. i=l SYX N I X. Y. 1. 1. i=l ft/sec v Velocity at plotting point ft/sec vx Observed velocity X Transformed distance from wall in curve (Continued) (Sheet 2 of 3) 85 Table 7-2 (Concluded) Variable XK Manometer Definition constant Units XLS ALOG (R -r) XLV ALOG (VX) y Transformed velocity in curve (Sheet 3 of 3) 86 TEST DATA SET #1 G 24. o.a3 2. • 389 6. • soa 9. • 566 15 ••566 1!:1 • • 5 08 22. .389 -<;<;. a. Input TEST DATA SET #1 POS,IN V,FPS DIAMETER= 24.00 IN. 1 .so 5.98 PI TOT COEFFICIENT . 830 2 1.97 6 .31 DIFFEREt-.TIAL GAGE 3 3.55 6.67 DX,IN RX,FT S,FT VX,FPS 4 5.47 7.10 1 2.00 .33 .1 7 6. 32 5 8.38 7.53 2 6.00 .so • 50 7.22 6 15.62 7.53 3 9.00 .25 .75 7.62 ., 18.53 7.10 4 15 .00 .25 • 7 5 7.62 8 20.45 6.67 5 18.00 • 50 .so 7. 22 9 2 2. 06 6 0 30 6 22.00 .83 .1 7 6.32 10 23.50 5.98 AVE VEL 6.72 FPS FLOW 21.1300 CFS 13.7208 1%D 9522.22 GP1'1 POWER RULE A= 2.06 B= • 1 2 R= .9998 SE= .0016 EXP(A) 7.88 AVE VEL 6.59 FPS FLOW 20.7447 CFS 13.4706 MGD 9348.59 GPM A/P VEL .8366 NOR."'AL LAW A= 7.34 o= .o6 R= • 9991 SE= .0227 AVE VEL 6.55 FPS FLOW 20.5998 CFS 13.3765 MGD 92o3 .3Cl GPi'. A/P VEL .8.350 b. Output Figure 7-1. Example 1 input and output 87 LIQUID ~A~OMET tR DATA 5~T 1 1 . <.7 1 b. u.63 .37S 1.15 . 75 1.5 <. . 25 2 . u <.e 2 . 21 3.75 2.35 6 . 0 2.72 6.1 2. d 6.5 3 . c; . 3. s . 5 .) • 1 1 2. ~ .6 1 4. 2. 62 1 5 . 2 . 35 1 6. 2 2 . 16.6 1.75 17.21.~ -s.s. a. Input LIGil.ID I'A ·o•47 s::= . ::>203 AVE VEL 1 . 4..1 FPS FLOW 2 . 5239 CFS 1.6369 r•Go 1137. 37 G?,v, A/F ~EL . 8162 b. Output Figure 7-2. Example. 2 input and output 88 PART VIII: SELECTING DIFFERENTIAL PRESSURE DEVICES FOR USE WITH PITOT TUBE Introduction 182. Velocity in water distribution systems is routinely measured by using pitot tubes which are inserted into the pipe as shown in Figure S-8. The velocity can be determined from the difference between the total pressure and static pressure as detected by the pitot tube. This difference in pressure ranges between 0.1 in. and 20 in. of water for velocities normally encountered in water distribution systems with line pressures of roughly 80 psi. Many ways of measuring this differential pressure exist, but each has certain undesirable characteristics. In this paper, each of the individual methods for determining differential pressures is evaluated, and an inexpensive, easy-touse manometer developed for this purpose is described. Alternative Pressure Measuring Devices 183. Several types of devices can be connected to pitot tubes for measurement of differential pressure. These range from manometers in which the difference in pressure is balanced by different heights of fluid columns, to bellows-type differential pressure gages in which the change in position of the bellows is converted into a dial reading, to electronic pressure transducers where differential pressure is converted into an electrical signal. (Differential pressure indicators of the spring or bellows-type are referred to as "gages" in this paper whether they are connected to a pointer or chart recorder.) Manometers to be used with water can be further subdivided into heavy and light liquid manometers, depending on whether the manometer liquid is heavier or lighter than water, and air-filled manometers in which air is used as the manometer fluid. 184. Differential pressures can also be obtained by measuring the pressure using two pressure gages and subtracting the readings. For water distribution systems, gage pressure is usually much higher than the differential pressure and, therefore, this two gage approach is not sufficiently accurate. 89 Evaluation of Differential Pressure Measuring Devices 185 . The differential pressure measuring devices described above are evaluated below based on criteria which are important in applications with a pitot tube. The evaluations are oriented toward use of the gage for field testing rather than in a permanent installation. Accuracy and range 186. Devices to be used to measure differential pressures for pitot tubes should be able to sense velocities in the range of 0.65 to 8.20 ft/sec. The equation for velocity as measured by a pitot tube is v = k "Vzg Ah (8-1) where v = velocity, L/T k = pitot tube coefficient (usually about 0.8) 2 g = acceleration due to gravity, L/T Ah = difference in pressure head, L For the desired range of velocities, the differential pressure gage must measure Ah values of 0.12 to 19.6 in. in terms of water column height. In terms of differential pressure, this corresponds to a range of 0.0043 to 0.708 psi. This accuracy is within reach of pressure transducers and manometers and is at the lower limit of accuracy for differential pressure gages. 187. One advantage of manometers is that they can record flow in either direction. Gages and transducers can be equipped with positive and negative scales with some loss in accuracy. (Accuracy is a function of range on these devices.) This bi-directional feature of manometers enables a user with a symmetric pitot tube to reverse the direction of the tube and repeat the measurements as a check of the readings. Manometers have the additional advantage of giving more accurate readings at low differential pressure if they are inclined. If the manometer is tilted, the Ah in Equation 8-1 should be multiplied by cos (a) where a is the angle between the manometer column and the vertical. Durability 188. Pressure gages with ranges of 19.2 in. H0 tend to be fairly deli 2cate. Pressure transducers are fairly rugged, but the electronic equipment 90 required to display and record the readings is considerably more delicate and must be kept dry. Pressure gages and transducers can be mounted in portable cases (several firms make cases for their gages as stock items) but these cases significantly increase both cost and weight. Glass manometers tend to be too delicate for use in the field, but plastic manometers are virtually indestructible. Overpressure 189. There are three pressures of interest in selecting a differential pressure device: the range of the instrument, the maximum line pressure (i.e., gage pressure in the distribution system), and the maximum overpressure. All components (i.e., hoses, fittings, valves, as well as the gage itself) must be able to withstand line pressures of at least 145 psi for use within water distribution systems. Overpressures (i.e., pressures in excess of the range of the gage) arise from two sources: velocities in the line higher than the gage can measure, and inadvertent opening of one side of the gage to the atmosphere. The later can result from opening a valve or from a break in the hose from the pitot tube to the differential pressure device. 190. Since pressure transducers can normally withstand overpressure to only about twice the pressure scale, a single mistake in the field can destroy the transducer. Good quality pressure gages have overpressure protection at pressures up to line pressure. Safety 191. If a hose should break on a manometer, water and indicating liquid are sprayed on the ground (and workers), which is messy but does not destroy the manometer. A more significant problem occurs if the differential pressure exceeds the maximum which can be recorded by the manometer, or if the manometer should fall over onto its side. In either case, manometer fluid is sucked into the water lines. Since mercury and halogenated hydrocarbons (e.g., carbon tetrachloride, bromoform) are commonly used heavy manometer fluids, these toxic chemicals, which in some cases also cause color, taste, and odor problems, can be introduced into drinking water supplies. Since for drinking water systems it is best to avoid using potentially toxic chemicals, an air-filled manometer becomes attractive. Power requirement 192. Most commercially available transducers and associated equipment to produce a digital readout generally require an a-c power source, although 91 some transducers can operate on d-e power from batteries. Gages and manometers require no power. Even the recorders available with some gages can be equipped with spring-wound or battery-powered drives. Type of output 193. Manometers must be read directly in the field since there is no convenient way to record manometer readings. Gages can be attached to chart recorders to measure differential pressure over time. Since they produce an electrical signal, transducers can be connected to ~ digital readout, an electronic recording device, or a transmitter that sends the signal to some distant location. Cost 194. Transducer systems are generally the highest priced of the alternatives. The cost is not so much for the transducer itself but for the associated electronic equipment required to produce a readout, plus a manifold and a carrying case. Gages are slightly less expensive than transducers, and several manufacturers make complete kits including gage, manifold, bleed valves, and carrying case. With transducers, a user must assemble the system from individual components. 195. Manometers are much less expensive and can usually be constructed from materials found in any hardware store or laboratory. High pressure fittings and hoses should be used and the manometer tube itself should be at least 0.25 in. in diameter so that air bubbles will not be trapped in the manometer. A manometer can be constructed from materials costing only a few dollars. Size and weight 196. Gages are by far the heaviest of the devices with some weighing as much as 10 lb, including the case and manifold. The size of a transducer system depends on how efficiently the components can be interconnected. Since 197. Gages are the easiest of the above devices to operate, and transducers, once the power source is connect~d, are not much more difficult. Liquid-filled manometers are somewhat trickier to operate properly since it is very important to purge air from the manometer. Air-filled manometers are the size of a manometer depends on the manometer fluid, a mercury manometer can be fairly compact while a light oil manometer must be fairly tall. An air-filled manometer must be about 2 ft tall. Manometers and transducers can be fairly light. Simplicity 92 somewhat easier to operate in that they only require use of some extra piping to store air which is compressed during use on high pressure lines, and they do not require any special manometer liquids. Summary of evaluation 198. In general, transducers are the best choice for permanent installations where power is available, a recorder is used, and the unit is housed. Gages are best for more remote locations where power may not be available and a record of differential pressure is required. For field installation without recording, manometers are best. Light fluid manometers tend to be extremely large but can be filled with nontoxic fluid. Heavy fluid manometers pose the threat of contamination of drinking water. Air-filled manometers circumvent these problems, are easy to use, and are inexpensive. 199 . This author has found air-filled manometers to be ideal for field applications with pitot tubes for velocities in excess of 1 ft/sec. The remainder of this paper describes how to construct and operate an air-filled manometer, and how to calculate differential pressure using such an apparatus. Construction of Air-Filled Manometers Construction 200. Figure 8-1 is a picture of an air-filled manometer and Figures 8-2 and 8-3 are a schematic of the manometer and a close-up of the required valves. The vertical tubes in the manometer are 0.5-in.-diam, clear, rigid polyvinyl chloride (PVC) tubing with a height hmax of 24 in. Valves A, C, D, E, and G are flow-through petcock valves while valves B and F are three-way bleed valves. Container H is an air chamber used for storing air as it is compressed by line pressure. It is not needed for line gage pressures less than 1 atmosphere (i.e., 14.7 psi) Its volume should be at least (P -l)V (8-2) max where pmax =maximum line pressure in atmosphere v = internal volume of the vertical tubes in the manometer The junction at the top of the manometer can be made using a group of fittings, a block of plexiglass, or other material which has been drilled and threaded. 93 Figure 8-1. An air-filled manometer ~~~~·~,f~~~~r~H~o T h AIR WATER _j_ I I I I I I I I I I I I I I I I I I I I I I I I II I I I I I I I I II I I I 1 I I II II II II I I II I I I I II II II II I I I I I I 1 I II II \\ ------ '~::::! H' ( ~ I I '--------I Figure 8-2. Schematic of a manometer 94 Operation 201. Before using the manometer , the user should test the valves under line pressure to understand how the valves operate and to ensure that there are no leaks. The manometer need not be connected to a pitot tube for this testing. The procedure to use the manometer is given as a step-by-step procedure below: a. All of the valves should be closed and air chamber H should be placed at an elevation higher than valve D. b. The pitot tube should be in serted into the line to be tested. c. Valves A, G, B, and F should be opened until all the lines from the pitot tube to the manometer Figure 8-3. Close-up of have been purged of air. valves for operating a manometer d. When this is done, valves A and G should be closed, and, optionally, a few drops of nonstaining food dye can be placed into the manometer through valves B and F. e. Valves B and F should be closed and valves A and G opened. f. Valves C, E, and then D should be opened to compress the air in the manometer into air chamber H. ~· When flow has stopped, valves C and E should be closed and the level of air chamber H lowered to allow compressed air to flow back into the manometer. It may be necessary to tilt the manometer to allow water to flow into the air chamber. h . When the vertical tubes of the manometer are roughly half full of water, valve D should be shut. (The air chamber may be disconnected at this time.) If line pressure is expected to drop significantly during the test, the manometer should be more than half full of water or the air will expand as the pressure drops. i. Valves C and E should be opened and the water levels in each leg of the manometer should 0qualize at one half of the height of the column when the pitot tube is oriented perpendicular to direction of flow. If this does not happen, check for closed valves or trapped air between the manometer and pitot tube. l· The line pressure should be recorded. k. The pitot tube is ready for operation. 95 Interpretation of results 202. The manometer measures a di fference between the static and total pressure sensed by the pitot tube . Th i s difference in pressure can be determined as llp =llhy =h(y -y ) (8-3) w w a where llp =difference in pressure, F/12 llh =pressure difference in terms of water column, 1 3 Yw =specific weight of water, F/1 h =difference in water level between legs of manometer, 1 3 Ya =speci fic weight of air, F/1Dividing by yw gives llh = h(l -r) (8-4) Since the density of air is much less than the density of water, the term r can be ignored for most applications with less than a 1 percent error. In these cases, llh =h and the manometer reading corresponds directly to the difference in head. 203. For greater accuracy, the weight of air can be accounted for in Equation 8-4 by treating r as a function of temperature and pressure. Since r varies only slightly with temperature (i.e., in the fourth decimal place), temperature effects can be ignored for all practi cal purposes. Similarly, the density of water only changes very slightl y with pressure. By far, the most significant impact on r is caused by the compressibility of air. Since air is practically an ideal gas, the relationship between r and pressure can be given by the ideal gas law r = = p (8-5) Y/Yw abs Ya1/Yw where p = absolute pressure, atm abs 3 = specific weight of air at 1 atm, F/1 Ya1 96 At 20°C, ya1/yw =0.00120, so Equation 8-4 can be given, for pressure in gage pressures, as r =0.00120 (1 + P /U) (8-6) g where p = gage pressure, F/L2 g (1 for pressure in atmospheres U = )101.3 for pressure in kilopascals )14.7 for pressure in pounds per square inch {407 for pressure in inches of water Therefore Equation 8-4 becomes Ah =h[1 -0.0012 (1 + P /U)] (8-7) g Table 8-1 provides insight into the magnitude of the correction that must be applied to h (i.e., the quantity in brackets in Equation 8-7) the compressibility of air in determining Ah corresponding to several pressures are shown. Values for r to account for and Ah/h Table 8-1 Effect of Gage Pressure on Correction for Compressibility of Air Pressure KPa ~ r Ah/h 0 0 0.0012 0.9988 200 29.0 0.0036 0.9964 500 72.5 0.0071 0.9929 1000 145.0 0.0130 0.9870 2000 290.0 0.0249 0.9751 204. Table 8-1 clearly shows that the corrections are quite small for pressures typically encountered in water distribution systems. Summary 205. An air-filled manometer is the best choice for use with a pitot tube in the field for measuring velocity in water supply pipes because of safety, cost, durability, and accuracy. 97 PART IX: PREVENTING DETERIORATION OF WATER DISTRIBUTION SYSTEMS Introduction 206. Out of sight, out of mind--water distribution systems may be the most easily forgotten component of America's infrastructure. The average American knows that when a water tap is opened, water flows out, and concludes, sometimes incorrectly, that all's well with the water distribution system. Yet, these systems, many of which have been in use for over 100 years, are deteriorating, and if nothing is done, Americans may not be able to count on having water when they need it. The nearly 300,000 residents in Jersey City, N.J., learned this lesson in the summer of 1982 when the combination of a main break and faulty valves shut off water to that system for several days. Deterioration of water distribution systems is not limited to urban systems, but is occurring at military installations as well. 207. It is rarely economical to simply rebuild entire systems. Therefore, a wise engineer or water system manager must be able to pinpoint the weak links in a system. Furthermore, since funds for infrastructure rehabilitation are limited, it is necessary to focus primary attention on those components for which rehabilitation or replacement will more than pay for themselves. 208. There are three basic types of problems that arise in water distribution systems as a result of aging: frequent breaks and leaks, loss of carrying capacity, and malfunctioning of appurtenances. The purpose of this paper is to describe these problems, present some methods by which they may be detected and corrected, and discuss how to decide when it is economical to take corrective action. Pipe Breaks Problem 209. The term "pipe breaks" is commonly used to describe several different conditions by which water may be lost from a system including: major pipe ruptures, joint leaks, and service leaks. Pipe breaks usually result from external causes such as contact with other structures, improper bedding, expansive soils, frost loads, brittle failure, live loads, and accidents. 98 Joint leaks generally result from motion of the pipe due to inadequate re straints or poor bedding. Unprotected metal pipes become more susceptible to these problems as they remain in the ground longer and are weakened by corro sion. The rate at which corrosion occurs is, however, highly variable, de pending upon pipe material, nature of the soil, location of the water table, and presence of stray electrical currents. O'Day (1982) and Morris (1967) describe the causes of breaks in some detail. Detection 210. Large main breaks are easy to locate as water will usually reach the surface. For example, a break in a 72-in.-diam conduit in St. Louis in 1973 resulted in the loss of 40 million gallons of water in only a few hours and washed out a rail spur (Fletcher 1982). Most leaks are, however, considerably smaller and not as dramatic. Some fairly large leaks can go undetected if the water can find its way to a sewer without reaching the surface. Leaks can be located using sound detection devices ranging from simple old fashioned telephone earpieces to sophisticated electronic devices. Cole (1979), Heim (1979), and Laverty (1979) describe several different methods for leak detection. One especially useful approach is to conduct a water audit (Cole 1979) in which all of the water flowing into an area is monitored for a 24-hr period. If nighttime flows are more than SO percent of average flows, there is reason to suspect significant leakage in the area. Correction 211. Repair of leaks usually involves placing a sleeve or clamp around the leak. In some instances, however, it is more economical to replace a section of the pipe. In congested urban areas, it is important to schedule "advanced replacement" of water mains to coincide with road repair or sewer installation so that inconvenience to citizens is minimized. In any case, it is also necessary to correct the cause of the leak (corrosion, poor bedding, inadequate restraints, etc.) or else the problem will simply reoccur. Evaluation 212. When considering a leak detection/repair program, there are two types of decisions the water system engineer/manager must make. The first regards whether or not to conduct a leak detection survey, consisting of a water audit, systematic use of leak detection devices, or both. Guidance on this is provided by Beckwith (1964), Brainard (1979), and Moyer (1983). Assuming that a survey is made and some bad sections of pipe are located, the next 99 decision is whether it is more economical to repair or replace the pipe. This decision should be based on cost (including damages and interruption of service), past break history of the pipe, and a projection of the change in break rates in the future. Useful guidance on decisionmaking is provided by Shamir and Howard (1979) and Walski and Pelliccia (1982). Loss of Carrying Capacity Problem 213. New pipes are generally fairly smooth inside, but as they remain in place they become rougher because scale or layers of corrosion form in the pipe. New pipes often have Hazen-Williams C-factors, an indicator of the pipe's carrying capacity, around 140, but as pipes age, this number can drop to less than half of its original value. This means that, for a given hydraulic gradient, the pipe can carry less than half of the flow that it did when it was new. If water use does not decrease, the result is low pressures in the system, especially during high use periods, or dramatically increased energy costs to maintain an acceptable pressure. Detection 214. Loss of carrying capacity becomes evident with complaints by users of low pressure, poor results in fire flow tests (American Water Works Association (AWWA) 1962), or, if early indications are not heeded, inadequate water to fight a fire. One way in which low values for the Hazen-Williams C-factor can be detected is during the process of calibrating a model of the water distribution system (Walski 1983a). If the model can only be calibrated for Cfactors of 60 to 80, then the system has lost much of its carrying capacity. To precisely determine a C-factor, it is necessary to measure flow, diameter, length, and head loss in a given section of pipe . Head losses can be readily determined using a pitot tube and pressure gages (Hudson 1954; AWWA 1962). Correction 215. If a pipe is found to have low carrying capacity, it can be rehabilitated by cleaning and lining. This process consists of scraping the inside of the pipes with mechanical scrapers or hydraulic pigs, and spraying a thin cement mortar lining in the pipe. In most cases, this can restore the C-factor to approximately 120. For smaller pipes, especially in developing areas, it is more economical to install a parallel pipe than to clean and line an old pipe. 100 Evaluation 216. To determine the most economical approach to solving the problem, it is necessary to compare the cost of cleaning and lining (or installation of a parallel pipe) with the costs of the alternatives of additional pumping energy and equipment to deliver the required flows. A useful comparison procedure has been described by Walski (1982). In general, cleaning and lining will be most economical for large mains which carry water at fairly high velocities. Malfunctioning of Appurtenances Problem 217. Most valves and hydrants are only used in emergency situations, and, since it may be many years between uses of these appurtenances, they are often neglected. Therefore, when called upon to perform in an emergency, valve and hydrant stems may be frozen, valves may not seat adequately, and valve boxes may have been paved over, or be impossible to find for other reasons. Detection 218. Valves and hydrants should be inspected and tested regularly, and records of the condition of these appurtenances should be kept (AWWA 1961, 1980). Annual inspection, maintenance, and testing will prevent most problems and identify the few that do occur. A file should be kept on the results of this inspection and testing. Kuranz and Barrett (1982) and Franklin (1982) describe how such programs can be set up. Correction 219. Once defective valves and hydrants have been located, they should be repaired or replaced as soon as possible. Valves that have been paved over can be located with metal detectors and the valve box cover can be raised. Valves are often found in the wrong position (e.g., closed when they should be open) and should be set properly. Gate valves should be either completely open or closed (i.e., not used for throttling which usually results in damage to the valve). Evaluation 220. Determining the frequency with which valves and hydrants should be inspected and tested involves consideration of the trade-off between the 101 cost of the program and the benefits of being assured that the system is operating properly. The American Water Works Association (1961) recommends large valves (>12 in.) be tested annually while smaller valves (<12 in.) be tested every 3 years. While such a program may at first appear costly, it is inexpensive compared with shutting down the system because valves did not operate properly. Summary 221. Water distribution systems, like many other infrastructure components, are gradually deteriorating. In most cases, it is less expensive to detect and correct problems early than to allow the problem to continue, or to replace major sections of systems. 102 REFERENCES American Water Works Association (AWWA). 1962. "Water Distribution Training Course," AWWA Manual M8, Denver, Colo. 1980. "Installation, Field Testing and Maintenance," AWWA Manual M17, Denver, Colo. Beckwith, H. E. 1964. "Economics of Leak Surveys," Journal of the American Water Works Association, Vol 56, No. 5, p 575. Brainard, F. S. 1979. "Leakage Problems and the Benefits of Leak Detection," Journal of the American Water Works Association, Vol 71, No. 2, p 64. California American Water Works Association Committee. 1962. "Loss of Capacity in Water Mains," Journal of the American Water Works Association, Vol 54, No. 10, p 1293. Cole, E. S. 1979. "Methods of Leak Detection: An Overview," Journal of the American Water Works Association, Vol 71, No. 2, p 73. Eggener, C. L., and Polkowski, L. 1976. "Network Models and the Impact of Modeling Assumptions," Journal of the American Water Works Association, Vol 68, No. 4, p 189. Fletcher, S. T. 1982. "Designing and Installing Large Steel Water Mains to Prevent Failures," Journal of the American Water Works Association, Vol 74, No. 11, p 568. Franklin, B. W. 1982. "Maintaining Distribution System Valves and Fire Hydrants," Journal of the American Water Works Association, Vol 74, No. 11, p 576. Headquarters, Department of the Army. 1980. "Methodology for Areawide Planning Studies (MAPS)" Engineer Manual EM 1110-2-502, Washington, D.C. Headquarters, Department of the Army. 1983. "Evaluating Existing Water Distribution Systems," Engineer Technical Letter ETL-1110-2-278, Washington, D.C. Heim, P. M. 1979. "Conducting a Leak Detection Search," Journal of the American Water Works Association, Vol 71, No. 2, p 66. Hudson, W. D. 1954. "Flow Tests on Distribution Systems," Journal of the American Water Works Association, Vol 46, No. 2, p 144. Kuranz, J. H., and Barrett, B. S. 1982. "Valve and Hydrant Maintenance," Water Engineering and Management, October, p 33. Lamont, P. 1981. "Common Pipe Flow Formulas Compared with the Theory of Roughness," Journal of the American Water Works Association, Vol 73, No. 5, p 274. Laverty, G. L. 1979. "Leak Detection: Modern Methods, Costs and Benefits," Journal of the American Water Works Association, Vol 71, No. 2, p 61. Meredith, D. 1982. "Optimal Parameter Estimates for Water Distribution Network Analysis," 4th World Congress on Water Resources, International Water Resources Association, Buenos Aires, Argentina. Morris, R. E. 1967. "Principle Causes and Remedies of Water Main Breaks," Journal of the American Water Works Association, Vol 59, No. 7, p 782. 103 Moyer, E. E. 1983. "The Economics of Leak Detection and Repair -A CaseStudy," Journal of the American Water Works Association, Vol 75, No. 1, p 29. O'Day, D. K. 1982. "Organizing and Analyzing Leak and Break Data for Making Main Replacement Decisions," Journal of the American Water Works Association,Vol 74, No. 11, p 588. Shamir, V., and Howard, C. D. 1979. "An Analytical Approach to Scheduling Pipe Replacement," Journal of the American Water Works Association, Vol 71, No. 5, p 248. U. S. Army Engineer District, Buffalo. 1981. "Urban Water Study," Buffalo, N.Y. Walski, T. M. 1982. "Economic Analysis of Rehabilitation of Water Mains," Journal, American Society of Civil Engineers, Water Resources Planning and Management Division, Vol 108, No. WR3, p 296. 1983a. "A Technique for Calibrating Water Network Models,"Journal of Water Resources Planning and Management, American Society of CivilEngineers, Vol 109, No. 4, p 360. 1983b. "How Constant is the Hazen Williams Constant?", AmericanWater Works Association Distribution Division Seminar, Birmingham, Ala. 1983c. "Using Water Distribution System Models," Journal of theAmerican Water Works Association, Vol 75, No. 2. Walski, T. M., and Pelliccia, A. 1981. "Water Main Repair/Replacement forBinghamton, .N.Y.," Miscellaneous Paper EL-81-1, U.S. Army Engineer WaterwaysExperiment Station, CE, Vicksburg, Miss. 1982. "Economic Analysis of Water Main Breaks," Journal of theAmerican Water Works Association, Vol 74, No. 3, p 140. Williams, G. S., and Hazen, A. 1920. Hydraulic Tables, John Wiley & Sons,New York. 104 APPENDIX A: DIPRA SOIL INVESTIGATION REPORT J SOIL INVESTIGATION REPORT WASHINGTON AQUEDUCT.DIVISION U.S. ARMY CORPS OF ENGINEERS WASHINGTON, D. C. APRIL 29, 1983 BY: APPROVED BY: J. Richard Page, P.E. Michael R. Redmon Regional Engineer Director of Regional Engineers (D I P R A) DUCTILE IRON PIPE RESEARCH ASSOCIATION 245 Riverchase Parkway, East, Suite 0 Birmingham, Alabama 35244 205/988-9870 A3 INDEX Title Synopsis Introduction Project Data Procedures Test Results Observations Conclusions Recommendations Project Map Test Results Page No•. 1 1 1 2 2 2 3 3 E X H I B I T S I II A4 S Y N 0 P S I S This soil investigation was conducted on existing water mains operated by the Washington Aqueduct Division of the U.S. Army Corps of Engineers. Soil Samples #l and #2, along the Jefferson-Davis Highway near the Pentagon, are potentially corrosive to ductile iron pipe. Should a replacement line to the existing 16" diameter steel main be installed, it is recommended to polyethylene encase through this area. Soil Samples #3, #4 and #5 were found to be nonaggressive to cast iron and ducti~e iron pipe. As all of these samples showed e~idence of fill material, replacement of existing water mains through these areas should be polyethylene encased if large quantities of cinders or slag material is encountered. Introduction Requested by: Dr. Thomas M. Walski, Waterways Experiment U.S. Army Corps of Engineers P. 0. Box 631 Vicksburg, Mississipp601/634-3931 P.E. Station i 39180 The survey was conducted in ~ spirit of service and cooperation for the purpose of identifying potentially corrosive conditions relative to ductile iron piping systems. Project Data Location: This investigation was conducted on existing water mains in the vicinity of Pentagon and Arlington National Cemetery in Washington, D.C .. Pipe Quantities: Several miles of 6" through 16" diameter water mains. Date of Survey: April 26 & 27, 1983 Funding: Federal Conducted by: J. Richard Page, P.E. and Allen H. Cox, P.E., DIPRA Regional Engineers. -lAS Procedures Representative test locations were selected along the route of proposed pipe installation. Each was assigned a number which corresponds to a number appearing in the soil analysis listing and on the project map, Exhibit I. A small diameter boring to approximate proposed pipe depth at each location facilitated field testing and removal of soil specimens for laboratory analysis. All field and laboratory procedures were completed in accordance with Appendix A of ANSI/AWWA Standard Cl05. Test Results Specific soil analysis results are listed in Exhibit II of this report. Observations No major oil and gas pipeline crossings were observed. The areas between the Potomac River and the Pentagon were originally marshland, and have been filled in for the construction of parks and roadways. -2- A6 Conclusions Due to a positive sulfides reaction, a neutral pH, and a low or negative redox reading, Soil Samples #1 and #2 are potentially corrosive to ductile iron pipe. Again, it should be noted that the existing uncoated steel water main does not have the corrosion resistant properties of either cast iron or ductile iron pipe. Soil Samples #3, #4 and #5 had negative sulfides reaction, high positive oxidation reduction potential, and moderately high resistivities. These conditions present an environment that should be non-aggressive to cast iron and ductile iron pipe. Recommendations Replacement ductile-iron water mains for the existing 16" diameter steel mains should be polyethylene encased through the gray clays found along the-Jefferson-Davis Highway, north of the Pentagon, contained in sc;l Samples #1 and #2. Although Soil Samples #3, #4 and #5 were found to be nonaggressive to ductile iron pipe, they did show evidence of containing some fill material. Upon replacing water mains through these areas, the new water mains should be polyethylene encased through areas of cinders and slag fill material, should they be encountered. The installation procedures and material specifications for polyethylene encasement are outlined in ANSI/AWWA Standard -3- A7 The foregoing report and recommendations are based upon examinations and tests which were made in accordance with generally accepted professional engineering standards and considered necessary in the circumstances. APPROVED BY: -Jo/~~ Richard Page, Michael R. Redmon gional Engineer Director of Regional Engineers -4- A8 .·. .., · ~ I .· "" ~.i I .' . _I j -.....;:,::: . PEN T'-GON LAGOON I: YACHT BASH~ '' \ ! I' \: I PENTAGON STREET. ON EXHIBIT I S",..,.,.,Pt. I-J.tJ e tf-77, .. A9 EXHIBIT II TEST RESULTS - Resis- Sample tivity Redox Sulfides Soil Description No. Location ohm-em mv pH l" Jefferson-Davis Highway 2,800 -20 6.8 Positive Gray Clay-Saturated ;l> Near Pentagon """"' 0 2 Jefferson-Davis Highway 3,360 + 20 7.0 Positive Brown Silty Clay W/Fill-Near Pentagon Saturated Power Plant Drive 3,320 +200 6.7 Negative Light Brown Clayey Sand W/Fill 3 Material-Moist 4 South Eads Street 2,720 +220 6.8 Negative Brown Clay W/Fill Material-Moist 5 Arlington Cemetery 2,880 +180 6.3 Negative Brown Clay W/Fill Material-Moist L-~ ·-----~ .....----~-.-l" r --