Document JrMrxjm1ebQk4bNOXZzkbyODe

tA*Af<5rt. XCO RKOSiCtj pxrmthna) roK ak cofyughtlo m a fam of JoU**AL AUfcMCAN WaTM WokkS ASSOCIATION Vol. 52, No. 8, August, 1960 frt-i-4 UU.S.A. Experience With Main Breaks in Four Large Cities --------------------------Panel Discussion-------------------------- A rand Discussion presented on May 17, I960, at the Animal Confer ence, Dal Harbour, Fla. Philadelphia--Gerald E. Arnold A paper presented by Ccrald E. Arnold, Gen. Supt., Water Dept., Philadelphia, Pa. Philadelphia has one of the oldest large water systems in the United States. The first part of the system was built in 1799, with wooden mains, beginning in 1819, east-iron pipe was laid, and by 1850 all of the wooden mains had l>een al>aiidoncd. The more than 3,000 mi of pipe within the'city limits varies in age from a few months to more than 130 years. Main breaks occur frequently in a sysrm of this size and age. As many as 1.000 breaks have been recorded in a ringle year, with a maximum of 300 C month during the colder months. Ilrcaks have occurred in all sizes of pipe, from 3 in. to 48 in., at various locations in tlie city, at different pres sure levels, and under varying condi tions of traffic and other external forces. Temp^rotur* Chang** The majority of water main breaks in Philadelphia occur during the three coldest months of the year, January, February, and December. During this period the water temperature readies its lowest point of the year. The tem perature at the intakes from the rivers is generally close to 32 F, and the tcmiieraturc throughout the system seldom rises above 35"F during the winter months. Frost penetration in Philadelphia varies from 12 in. to 3() in., depending on the duration and severity of the cold weather. It has been noted that the greatest number of breaks usually occurs at the timi of the winter when the temperature i.. dropping most rapidly for the first time. It is believed that the sudden chilling of the water and the surround ing earth results in shrinkage of the pipe metal, thus setting up sufficient longitudinal stresses actually to pull the metal in two at points where it has already weakened. The. majority of the breaks found under these condi tions are circumferential cracks, with little longitudinal deviation. Most of these breaks can be repaired by in stalling a split clamp around the crack. Some cracks of this type occur at other times of the year, but those arc be lieved to be due to beam action from blocking under the pipe or to other external forces. An unusual occurrence during April 1960 is worthy of note. The month 1041 CAPCO JEN 0019597 1042 CERALD E. ARNOLD lour. AW WA of March I960 was one of the coldest of record, and April was one of the warmest of record. tem|>cratures ex ceeding 90F on several occasions, which is unusual for Philadelphia in April. During April 1960j there were 50 water main breaks, compared to 23 for April 1959. The 1960 record tem perature was 30 per cent above the average for that month. It is believed that these breaks were occasioned by rapid and unusual temperature rises. Proasure Chcmqo* Changing pressures in water sys tems can cause main breaks, particu larly if those- pressure changes are sudd.en. such as those associated with water hammer or rapid changes in operating pressures. A number of \cars ago, a pressure change was made in part cf the Philadelphia distribution system v-hich raised the pressure only about 10 psi but resulted in eighteen main breaks in a single hour. The p/oints at which the mains broke with this slight pressure change v.ere al ready weakened by some other cause. The author had an experience a numl>er of years ago with another water system in which day and night pres sures varied as much as 100 psi owing to unfavorable ojrerating conditions. This resulted in some serious breaks in a 16-in. supply line. SupporL!n<j Ground Another common cause for water main breaks is shifting or settling ground. Pipe that is normally sup[xtrted throughout its length shuuld lie subjected to few bending strains. If the ground shifts nr settles unevenly the transmittal' of beam forces to the pi[>e!ine can set up strains that exceed the tensile strength of the metal. Serious difficulty has frequently 1-ccn encountered with a 4S-in. line on Broad St. in Philadelphia, where the line is carried on concrete piers over the top of a subway. Uneven settle ment of these piers has resulted in a distortion of the pipe, transmitting un usual stress to the metal. Frequent breaks have occurred in this line, and, in some instances, it has been found that the pipe was not resting on the piers, which would normally support it from below, either at the point of break or at an adjacent .pier. Surveys of this pipeline have indicated wide deviations from a straight alignment. Most of the breaks have been at the bell with a circular crack running onehalf or two-thirds of the length of the joint and returning again to the bell, with the result that a large segment of the pipe is hroken away completely. This pipeline is also subject to vibra tion of traffic from above ar.d subway trains from Wow. In addition to breaks in the pipe itself, there have been frequent joint leaks, indicating that the pipe is moving. Electrolysis Another detrimental action that weakens pipe metal is soil corrosion, or stray-current electrolysis. Corro sive soils or stray currents tend to eat away the metal from the outside of the pipe, usually in concentrated locations, thus producing a thin wall or com plete breakthrough in small areas. Removal of metal from the pipe wall, of course, also weakens the wall so that any unusual strain may result in a break. The problem of stray-current electrolysis is becoming increasingly serious in cities where there has been a partial abandonment of the streetcar system. Often the abandoned tracks, when they are no longer used for streetcars, are paved over and used as a return path for the current supplying n[>erating tracks. These old rails CAPCO JEN 0019598 mu MAIN HKKAKS--nill.AliELI'lIJA 1043 being paved over, are not properly O maintained.-.md open joints fref|uently develop resulting in discharge of large (|u;mtilies of current to the ground and underground structures, such as pij>elines. When one recalls that 1 amp of current will remove 20 II) of metal in a single year, there is a definite rea son for concern when as much as f>00 amp is found traveling on an under ground piping system. Traffic Traffic conditions have materially changed in recent years. When many of the pi|>elines were laid 50-- 100 .years ago. street traffic consisted of horse-drawn vehicles. Today, modern motor trucks, weighing as much as 30 tons, pound the depressions in streets and transmit shock waves to all under ground structures. The standard prac tice in Philadelphia has been to bury water pi|>e with at least 4 ft of cover. In the old days, this was sufficient, but with modern traffic loads it is felt that more cover should be provided or that there should be a very heavy concrete base under the pavement to take the impact of traffic shock. It has l>een O noted some main breaks have followed the development of chuck holes in the street where heavy trucks [winded into these holes, setting up noticeable vibrations. Age of Pipo Extreme age docs not .necessarily indicate |xior condition of pi|>e. Many miles of pipeline in. the Philadelphia system are more than 100 years old and are still rendering satisfactory service with comparatively few brinks, whereas other jiij/e 50-75 years of age has had more frequent breaks. A study of the pipe removed from the system indicates that there is very little weakening of the tncta! due to internal corrosion. Tulicrculation of the interior surface is evident, hut pit ting of the metal is negligible. Ex ternal forces or corrosion factors are much more important than internal corrosion. A study of pqie removed after many years of service indicates that wall thicknesses are materially reduced from the outside. In most instances metal has been removed in s|iots. Icaving'tmticeablc cavities and a few fairly extensive areas of thin metal. Construction and Operation Gixid construction is the liest way to avoid water main breaks. A study of breaks in Philadelphia has led the author to Iielieve that much of the 'trouble was occasioned by improper construction methods. Blocking was. utilized in Philadelphia for many years, and it is the author's firm conviction that this was re>]xir.sib!c for many breaks. Blocking should be prohibited. The standard practice in Philadelphia now is to encase the new main in 6 in. of clean sand. The trench is excavated to 6 in. Iielow the bottom of the pipe and liackfillcd with sand; the pipe is firmly bedded in the sand. and. after the joints have liecn made, sand is placed around the pipe to a full thick ness of 6 in. The balance of the trench is then filled with cxacavnted material, excluding broken concrete, stone, or other debris. The backfill is thor oughly tanqicd in thin layers to the surface of the ground and then allowed to settle and l/ecume thoroughly com pacted before [laving is placed." To lessen the shock of traffic impact, a base tourse at least 6 in. thick should lie made of poured concrete and topped off with asphalt or some other finish coat. Excavations for repairs and maintenance work should lie Ilack filled in the same manner. CAPCO JEN 0019599 1044 GERALD E. ARNOLD Jour.AWH'A All pipe should he thoroughly in spected at the plant and again on the job site before installation. Old rec ords of the Philadelphia Water De partment indicate that pijie inspected on the job 50-60 years ago was sub ject to about 20 per cent rejects. Manufacturing is probably much hetter today than it was at that time, hut piping is still subject to damage after it leaves the factory and l>cforc it is installed. Tests at pressures at least 50 per cent higher than normal work ing pressure should he made before the line is accepted front the contractor. Cities having surface streetcars should make a study of the electroly sis problem, particularly if some of the .streetcar lines have ltccn altondoned and the rails are still in place and used for negative return current. Continuing electrolysis studies should 1* made to sec that the water system is not being damaged hv stray cur rents. Soil surveys should l>e made in advance of construction to deter mine whether or not sjwcial protection is necessary to prolong the life of the new pipeline. The installation of rec tifiers. insulating, or landed joints may l>e highly desirable, and their cost is hut a fraction of the rn't of replacing a pipeline. Care should lie exercised to avoid frequent wide fluctuations in pressure. A pipeline carrying water at pressures \ar\ing as much as 50-100 psi actually deflects with the pressure changes, and metal fatigue may result in serious breaks. Whenever a break occurs, a study of the attendant conditions should lie made before the trench is backfilled. This will enable the opera tors to ascertain to .some degree the cause of the break and assist in design ing to avoid repetition of the failure in the future. In some instance: breaks take place for no apparent rea son. On several occasions the author has seen cast-iron lines in which a sec tion of the pipe broke out on one side or the liottom in the middle of a joint without cracking the hell or the spigot. The metal surrounding the break ap peared to be sound and there were no unusual operating conditions that could account for the break. Flexibility at the joints is another factor which, it is believed, can reduce the number of breaks. With the new type of rubber ring gasket, there is much more flexibility than there was in older types of joints. . Movement at the joint relieves stress imthc inctal and should be helpful in reducing the number of breaks. Conclusion Water main breaks usually occur at a point where the metal has lieen weak ened by corrosion or some other cause, hut the actual breaking of the metal occurs as a result of some unusual external force such as temperature changes, shifting ground, foreign ob jects, or traffic impact. Weakening o' the metal is a result of age. Imt age is not necessarily the most important fac tor. Once the metal is weakened, an unusual strain may result in a fracture of the metal. In the author's opinion, sudden temperature changes are re sponsible for most breaks, although other factors may lie involved. Water main breaks should be studied as they occur hy a qualified engineer, to deter mine, if possible, what caused them.. By studying this information and fol lowing good construction practices, many of the water main'breaks exl>erienced in the jiast can be avoided in the future. CAPCO JEN 0019600 Aun.JVM MAIN BREAKS--NEW YORK 1045 -----------------New York--Edward /. Clark----------------- .4 faper, presented by Edvard J. Clark, Chief Eur/r,, Bureau of ll'atcr Supply. Dept, of lYatcr Supply, Gat Cr Eleelrieity. Xew York, .V.1'. Water main breaks have always presented considerable problems to water supply engineers, not only be cause they disrupt the supply and are expensive, but also because of damage caused by the flooding of subways and basements, the upheaving of pave ments, interference with vehicular traffic and transit service, and other inconveniences suffered l>v the public. Causes of Breaks Breaks in cast-iron mains may be caused by flaws in pipe and castings; internal pressure in excess of that for which the pipe was designed; deterio ration of metal due to corrosion; elec trolysis; freezing of exposed pipe; excessive earth load due to depth of cover over pipe; unbalanced pressures in the line such as at (tends and caps; unequal settlement of the soil in which the main is laid; settlement of other 'uhsurface structures crossing the iter pipe, s.uch as electric conduits or Ogas and steam mains: vibrations due to blasting in close proximity to the main; and other causes. In the Xew York system, the num ber of hreaks due to defective pipe and castings, excessive internal pressure, deterioration of metal, electrolysis, or freezing is negligible. Many water main breaks have lieen caused by set tling of the main over short stretches. This causes the pijse to act as a beam, for which it is not designed, ami re sults in fractures or loose jciiVits. Settling of a water main over short stretches may be caused by imprn]>cr bedding or by the gradual w.earing away of soil under the main as a result of a small leak front a pi|ic joint. A substantial numlier of breaks in Xew York ha- been caused by unequal settlement of water mains along suliway routes and the settlement of sub surface structures, such as ekvtrical conduits or gas and steam mains, cross ing the water pi|ie and renting on it. It is sometimes difficult, if not im]>ossible. to determine the exact cause of a spec*fie break, ns the subsurface conditions existing when the broken main is exposed may l>e entirely dif ferent from those prior to the break. There is no way of knowing what the prior conditions were. The New York distribution system comprises about 5.SOO mi of mains incorporating some 4.000.000 pipe joints. 450.000 pipe fittings. 105.000 gate valves, and 90.000 hydrants. For the past 10 years, there has lieen an average of 330 breaks each year (al most one a day). 20 of which have been in 20-in. pipe and larger, with the majority in the 6-S-in. sizes. Al though there is no sjiecific |iattern as to time of day breaks occur, it apjiears that many of them take place during the early morning hours. This may l>c <lue to the sudden opening of exist ing splits in the pipe as a result of slight increases in internal pressures during the early morning hours when the water demands are low. Preventive. Measures The steps that have Iteett taken to minimize breaks include improved con struction methuds and strict inspection CAPCO JEN 0019601 1046 EDWARD J. CLARK Jour. A M' WA of materials and installations. The use under the main and thereby result i. of steel pipe for all mains larger-than unequal settling of the pipe and possible 20 in. I>as been standard practice in rupture. Approximately 600 leaks New York for the past 30 years. , from joints, split pipe, taps, services, Specifications for water main instal and gate valves are detected yearly, lations require: thus effecting an average saving of 25 1. Well tamped bedding of clean mgd. earth or sand under, around, and over As mentioned before, the use of steel the pipe pipe for mains larger than 20 in. is 2. Consolidated backfill to distribute another safeguard adopted to curtail the traffic loads breaks in large mains and thereby re 3. Installation below the frost line duce flood damages, disruption of *o prevent heaving traffic, and other inconveniences. To 4. Use of rods on Ivends and fittings minimise the effects of electrolysis op to help the pipe withstand unbalanced steel pipe, insulating joints are in pressures stalled in the line, and, in some cases, 5. Prn|>er assembly and calking of cathodic protection is provided. lead joints to prevent leaks 6. .Hydrostatic field testing of pipe Remedied Acaon (Alt mains 20 in. or larger are sub New York has had many breaks in jected to hydrostatic field tests at 12S large mains which have resulted in a psi, and in certain other cases, such great deal of flooding and disruption as bridge crossings, similar tests are of subway traffic. required on mains smaller than 20 in.) When breaks do occur, the first ac .7. Reiniorced-concrete mat, with tion taken is to shut down the feeds cradles or pile supports, in areas of to the main in the shortest possible fresh fill, swamps, and other places time in order to minimize damages. where the bearing capacity of the soil For this purpose, emergency crews are is poor maintained on a 24-hr basis and have 8. Concrete cradles under the pipe trucks equipped with motorized valve- in rock trenches. 0[>erating devices as well as short-wave In an effort to reduce the numlier radios to s{>eed up ojierations. of leaks and breaks, an inflection force is maintained at the foundries where Narrows Siphon Incident the pi[>e and castings are made, to Perhaps the most serious break, ami make sure that the materials are sound certainly the most difficult one to re and are being manufactured in accord pair, was the break in a 36-in. cast-iron ance with specifications. Also, all submarine pipeline crossing the nar water main installations have enni|>e- rows in New York Ray and supplying teut .ius]>cctioniil supervision to insure the Horough of Richmond with water. the strict enforcement of all contract The nature of the pipe fracture indi sjiecifications. The water department cated that it was caused by one of the also has a social force engaged in spuds of a large dredge. which had systematic surveys to detect and elimi Ijccn working directly over the pqieline nate underground leaks. Such leaks, at the time of the break. which do not appear at the surface, When the break occurred, a gage may cause the gradual erosion of soil located in one of the pumping stations CAPCO JEN 0019602 Aug. I960 MAIN BREAKS--NEW YORK 1047 recorded a sudden drop in pressure Ofrom 70 to 28 psi. Within IS min, it was determined that the dropin pres sure had to have been caused by a break in the narrows siphon, as a ven turi meter showed that a large quantity of water was flowing in the pipe at the Brooklyn end, but no water was reaching Richmond. Backflow from Richmond was prevented by a 'check valve in the pipeline near the water front. Within an hour after the break occurred, the Brooklyn end of the line was shut down, stopping the 28-mgd flow that had been wasting into the hay. Siphon Repairs Because the narrows siphon fur nished more than two-thirds of the supply for Richmond and local sources could not possibly have met the de mand once the storage in the distribu tion reservoir had been consumed, im.mediatc action to effect repairs had to be taken. Under the emergency powers of the commissioner of the de partment, a dredging company was en;aged to examine the line and make repairs. Three hours after the break,' tl'.e company had some of its plant on Qthe job and started pumping air into the pijjclinc to locate the leak. After 2 hr, the air liegan to show at the sur face of the water, indicating the break to be about 170 ft from the end of a pier on the Richmond side. At that point the pipe was under 30 ft of water and 20 ft of harlmr muck and fill. The contractor started immediately to exca vate the material over the pipe by bucketing and pumping. Approxi mately 4,000 cu yd of material had to be excavated to provide enough space to remove the pijie and effect repairs. Meanwhile, pumping of local sources in Richmond was increased to the full capacity of the pumping stations. This was not sufficient to meet the demands, however, and 7 mgd had to he drawn from storage in the distribution reser voir in Richmond. The reservoir could supplement the supply from local sources for only about 60 days. To help alleviate the emergency, steps were taken to stop waste and leakage and the public was apprised of the necessity of conserving* water. Pressures were reduced by 13 psi during the day and 17 psi at night. At first, there was a 3-mgd (IS per cent) drop in consumption, but the de mand gradually increased as the con sumers found that there was no im mediate likelihood that the supply would be curtailed. Within a week after the break oc curred. the pipe was uncovered suf ficiently to permit the removal of the broken section. In order to permit a diver to enter and make repairs, it wax found necessary first to enlarge the hole in the broken pipe. Cutting of the broken section for removal was done mostly from the inside by means of electric torches and was continued night and day until the work was com pleted and the broken pipe could lie brought to the surface. This took 10 days. While the work of cutting was in progress, designs were made and or ders placed for a solid cast-steel sleeve about 5 ft long and for" a split sleeve of similar length, it being expected that one of the .sleeves could lie found suit able to effect repairs. The solid steel sleeve was used. The.nui\-k which had settled in the excised cuds of the pipe ujmn removal of the broken section was cleaned out by the divers and timlier blocking was laid at the bottom of the trench to support the new pqie. The solid steel CAPCO JEN 0019603 1048 GERALD J. REMUS Jour. AW WA sleeve was then placed over the end of the cut pipe. A new length of flexiblejoint pipe, which had been turned down to the normal outside diameter at the spigot end, was lowered in place, and the sleeve was drawn over it, clos ing the gap. The two sleeve joints and the flexible bell joint were then calked with lead wool. Before the 'siphon was restored to service it was cleaned by introducing water from the Brooklyn end and discharging througli a lilowoff at the Richmond end. Chlorine was injected at the Brooklyn shore by means of a temporary connection and chlorination plant iii order to disinfect the interior of the pijie. which had l>ecn subjected to contamination during repairs from the sewage-laden waters of the hay. The treatment was continued for 7 hr, after which unchlorinated water was passed through the pipeline for 3J hr. A sample of water taken and 'examined at the end of this period proved to lie of excellent sanitary qual ity. Tests showed the new joints to he tight. A month after the break had oc curred. the siphon was again in serv ice supplving water to Richmond. Backfilling under, around, and over the main was continued for a week until a depth of about 10 ft over the pipe was secured. Conclusion While the siphon was under repair, 200 mil gal was drawn from storage in the Richmond distribution reservoir, depleting it to half of its capacity. Al though there was no immediate danger of exhausting the supply in this reser voir. the situation caused considerable uneasiness for a time, and soon after this incident an additional 42-in. pipe line was installed across -the narrows. With duplicate feeds to Richmond, there no longer is much danger of a serious interruption of service to that borough. Contract plans are being prepared presently for constructing a 10-ft water tunnel to Richmond, to take care of the .future requirements of the borough, with the population growth expected to be occasioned by the construction of a new bridge con necting it with Brooklyn. When this tunnel is completed, the two mains crossing the narrows will be kept on a standby basis and Richmond will be assured of an unfailing supply of water. The narrows siphon incident points up the urgent need of installing dua transmission mains to isolated areas, so as to assure a supply of water in the event of a break. New York pro vides such feeds to several islands in the East River on which hospitals and penal institutions are located. -------------------Delzoit--Gerald J. Remus------------------- A paher presnilcd by Gerald J. Rentes, Gen, Mgr., Dept, of Water Supply, Detroit, Mich. The Detroit Department of Water Supply provides water for Detroit and 49 adjacent communities, and contract negotiations are underway to provide service to twelve additional communities. At the present time 3,400,000 people, or 42 per cent of the population of Michigan, get their water from the Detroit system. Detroit has three major pumping stations, with the fourth under construction, and plans are already underway for a fifth CAPCO JEN 0019604 r*3BSHEHR Aug. I960 MAIN BREAKS--DETROIT 1049 major station. Water is transmitted Oand distributed to the system through a network of steel, asbestos-cement, reinforced-concrete-lined steel cylinder, prestressed-concrete cylinder, and castiron pipe, totaling more than 6,600 mi. A breakdown according to type is given in Table 1. AH water service contracts between Detroit and its suburban municipal customers require that installations must meet Detroit's standards in every characteristic. The establishment of these standards requires extreme care and sound engineering judgment. Equally important is the public rela tions job of winning acceptance and application of these standards. In addition, the annual repair and mainte nance expenses, with service reliability, became major factors in the cost of water. Therefore, there is a twofold purpose in studying water transmission and distribution systems: to control .costs and reliability, and to guide the development of an ever enlarging, reli able, and efficient water system, involv ing many units of government. The Detroit Water Board main tains the mains within the Detroit city O limits, and good records are available. The suburban communities maintain TABLE I Types and Lenilks of Detroit Area Water Mains, I960 fiscal Tun Fig. 1. Cut-Iron Main Ernie* is Detroit, 1920-60 Because of the increase in the number of breaks in cast-iron mains (the total miles of main remains relatively constant), Class ISO cast-iron pipe is no longer ac ceptable in Detroit, their own distribution systems and the Board does not have their records. Therefore, this analysis of water main troubles is concerned only with the 3,085 mi of pipe within the Detroit city limits. Steel Pipe Type Steel Asbestos-cement Reinforccd-concrcte- lined steel cylinder Prestressed-concrete cylinder Cast iron Total Insulted Length Detroit Suburban Tout 52 3 55 3 19 22 V52 80 132 13 24 37 2,965 3,420 6,385 3,085 3,546 6,631 The 52 mi of steel pipe have oeen in service 25-30 years. Most coatings that were used were of insignificant protection, and failures due to corro sion have been in the form of leaks rather than breaks. During the period 1950-59, an average of 46.9 leaks per year has occurred in steel mains. The sizes of the mains in which trouble occurred varied from 12 in. to 60 inBecause there has been an increase in pitting and deterioration of steel pipe. In tltlM < Mi u i i 1,000 mi- l CAPCO JEN 0019605 1050 GERALD J. REMUS Jnur.AWWA the smaller sizes have gradually been replaced and the larger sizes are being prejvirod for relining. Ashes les-Ccmont and Concrete Asbestos-cement pipe has been in use only since 1937, and in moderate amounts.. An evaluation of the merits of this type of water carrier can, there fore. be only a calculated guess at best. Coat Iron .The hulk of the 3.085 mi of main within the Detroit city limits is cast iron. This pijse has been used since 1838. Of the cast-iron pipe in service, 2,750 mi is largely of Class 150, and 215 mi is Class 250 pipe. Class 250 pipe has not caused trouble, but it has been in use only since 1941, and only TABLE 2 Record of Brooks in Detroit Cast-Iron Water Mains, 1940-60 Year Intuited 6 in. 8 in. 10 In. | 12 in. 16 in. 24 in. 20 in. 56 In. 42 in. 48 in. Total N'o. of Breaks 1940 140 152 9i i 303 1941 176 173 83 ii i 363 1942 147 151 18 6 t 323 1943 169 158 3 82 ii l 2 t 346 1944 206 175 3 53 i 3 i 397 1945 228 204 95 i. i 443 1946 133 167 10 310 1947 225 188 16 2 2 2 435 1948 274 275 16 2 2 3 572 1949 227 230 1 17 4 2 481 1950 260 270 9 2 6i 548 1951 269 243 18 5 2 i 538 1952 296 255 1 13 3 2i 1 i 573 1953 410 322 2 20 5 1 l 2 763 1954 300 322 1 19 7 1 2 642 1955 342 376 1 23 7 6l 756 I9S6 405 356 5 28 4 4 i i 1 SOS 1957 357 362 2 2 33 n i il 770 1958 476 459 1 30 9 4 i 980 1959 479 419 3 44 15 22 1 1 966 I960* 217 219 17 3 1 1 4S8 * Estimated (or May and June. The performance record of asbestoscement pipe does indicate, however, that its use will be continued and mod erately increased. Rcinforccd-concrcte and prestressedconcrcte water mains, totalling 167 mi of large pipe, "have furnished satisfac tory service. These mains have not faced the test of time, but the record developed thus far indicates that their use will be continued. in 6- to 24-in. sizes. The record of cast-iron main breaks (practically all in Class 150 pipe) is given in Table 2. The curves in Fig. 1 illustrate the total cast-iron main breaks since 1920, the average age of the pipe, and the number of miles of mains in service. The total main breaks gradually in crease, and as the total miles of mains remains relatively constant, the only conclusion that can be arrived at is CAPCO JEN 0019606 aup. im MAIN HRKAKS--INtMANAfOMS 10S1 tat as time elapses the maintenance of existing-;.iains is going to cost more and more, service of distribution sys tem will laconic more unreliable, and a replacement program will have to continue. The causes of breaks in these castiron water mains cannot lie identified exactly. There are many related vari ables. such as water and ground tem perature changes, thinness of 'pipe walls, poor construction materials and practices, water hammer, excessive street loads causing street settlement which could he amplified by shock and insufficient thickness of concrete in the roadbed, sewer washouts, inflexible pipe joints, poor soil conditions, and other utility obstructions. All of these factors affect results, but how much each should l>e held accountable for is hard to evaluate. Water temperature changes and substandard pipe wall thicknesses apjtear to be the most im portant factors. Present Practice* The following precepts are now being followed in Detroit: 1. Class 150 cast-iron pipe is un acceptable, and those sections now having an excessive breakage rate must lie replaced. 2. Class 250 cement-lined cast-iron pi|K* is acceptable, and it is the only cast-iron pipe that will be used. 3. Use of Class 200 asbestos-cement pipe will gradually lie increased, and it will, lie allowed to prove its worth, in 6- and 8-in. sizes, and under special installation practices. 4. Class 150 asliestos-cemcnt pipe is unacceptable. 5. Mains 24 in. in' diameter in either standard cylinder pijie or prestressed emliedded-cytinder pipe and mains If) in. and 20 in. in diameter in prestressed cylinder pipe will be accepted. Speci fications on each job will spell out the construction standards, however. Conclusion Improved laying practices of aH pipehave lieen adopted. More flexibility in the joints and better embedding practices have lieen adopted, and closer inspection in the field is necessary. The sjxrcifications applied to the ex istent construction and improved .main tenance practices will, it is hoped, keep the system under reasonable trouble . control and improve reliability. --------------Indianapolis--Howard W. Nlemeycz----------- A paper presented by Howard IV. Niemeyer, Snpt. of Distribution, . Indianapolis Walct Co., Indianapolis, lnd. The record of pipeline failures in the Indianapolis distribution system during the period 1926-54 was re viewed in the May 1955 issue of the Journal in a discussion of an jtrticle by Leo V. Garrity.1 In his article Garrity concluded that breaks in castiron pipe in Detroit were caused pri marily by a combination of tempera ture stresses imposed by the restraint of sulfur comtxiund joints and a pro gressive loss of pipe strength from corrosion. Although Indianapolis at that time was experiencing only three pipe breaks per year per 100 mi of main, compared to '20 in Detroit, analysis of those failures did lend sup port to Garrity's conclusion. The ex perience with pipeline failures since 1954 has provided additional support ing information, and it now appears that more than half of all the failures CAPCO JEN 0019607 1052 HOWARD W. NIEMEYKR Jnur.AWWA that have occurred in pipe and joints in the Indianapolis system over the past 10 years can be directly associated with the use of sulfur compound as a jointing material. Experience has also indicated that centrifugally cast pipe in 18-ft lengths has not performed as well as pit-cast pipe in 12 ft lengths, both being laid with sulfur compound joints. Indianapolis System To evaluate the Indianapolis experi ence properly, the physical character istics and certain environmental condi tions of the system need to l>e known. The water supply for the city comes from a surface source, except that well fields formerly used have been main tained for emergency purpose. Water 4 3? so . *o 3.11 r- i U6 to =tlF It--4\ JZ-- II 1941 1946 1951 1956 I960 Ye*f Fig. 2. Cast-Iron Main Failurts Other Than Those Caused by Accident* or Freezing, 1926-59 The dashed curve represents that portion of the failures that was caused by circum ferential breaks. Figures accompanying the curves represent overages' per 100 ?i of main per year. CAPCO JEN 0019608 A uq. 19M) MAIN BREAKS--INWANAPOMS 1053 temperature, therefore, has a seasonal variation which amounts to ... much as 50F between winter and summer. During prolonged cold waves, water temperature approaches 32 F. Ground frost penetration has varied over the years from a slight crust to as much as' the 54-in. depth that was reached in the severe 1935-36 winter, the maximum penetration on record.- The soil is for the most part clay, and, in general, there has been little evidence of corrosive action on the exterior of pipes, except in some isolated cases. It has rarely been necessary to install mains over filled ground. Cast-iron pipe has been used throughout the sys tem, except for some bridge crossings ers. The system has been successfully operated to avoid water hammer. Pipeline Failures In `spite of the care used in con structing and operating the system, miscellaneous failures have occurred as a result of some occasional material and construction faults and because of conditions that have developed beyond control of the distribution engineers. Over the years the number of such failures has not been considered exces sive; they were, therefore, accepted as normal operating problems. An in creasing incidence of failure since 1940, however, has indicated presence of some factor or factors that appar- TABLE 3 Types and Lengths of Indianapolis Waicr Mains, 1924--60 When Installed Before 1924 1924-38 1938-50 1950-60 Type pit cast pit cast ccrHrifugally cist; unlined ccntrifugally cast; lined . CU*s Section Lcnrib A Jmnitne Waterul InAallr-d Lmcth - B* 12 lead 482 B* 12 tulfur 187 150 18 sulfur 191 ISO IS rubber 352 * CU^i C or D umS in lire? Iranimimon rruiru. and recently installed large feeder, mains. The schedule of material types used is shown in Table 3. Mains were installed with 48-in. cover prior to the 1936 frost and 54 in. .ever since. Installations have al ways been made in conformance with the best construction practices known with respect to pipe inspection and handling, trench conditions, and clear ances with other structures. Service connections have been installed in a manner to avoid stresses from this source. Average distribution pressure is maintained near 55 psi with some maximums of 120 psi occurring in limited areas adjacent to one high-level pumping station and some of the boost- ently did not exist previously. The reason for any one failure is seldom obvious, and as it has been neither economically feasible nor practical to investigate each' job for all factors present, no record of failures by cause has been accumulated for study. Much can be learned, however, by examina tion of job records as to the kind of failures that have occurred and the types of materials involved. The quantitative record of pipeline failures during the period 1925-59 is summarized graphically in Fig. 2, showing pipe breaks, and Fig. 3, show ing joint failures. These records ex clude accidental damage to the sys tem by adjacent construction and show CAPCO JEN 0019609 10S4 HOWARD W. KIEMEYER Jnnr.AWWA only those failures from normal causes. The exclusion is not intended to underrate the problem of damage failures which create emergency situ ations, disrupt normal operating pat terns, and create added work loads for maintenance personnel. In the In dianapolis system, where damage has probably been no more extensive than drant in place of the solid-barrel type has been one solution to this problem. The reader's attention is directed to the relatively small number of pipe breaks prior to 1941 (Fig. 2). In this early period of the record the inci dence of pipe failures averaged only little more than one break per year per 100 mi of main--certainly a favor- lr n Ir-l ii ll It - t 010 f"1 II cj ,-,LI I II II 1 01S 017 r-* r .I 008 | I ------- rr -Jr-iIt. ii 005 -f-- L- I I I r~* --]--I I -J 0.05 1926 1936 1941 1946 Fig. 3. Joint Failures, 1926-59 The solid end dashed curves represent failures of sulfur compound and lead joints, respectively. Figures accompanying the curves represent averages per 1,000 joints per year. in other systems, construction activity, largely for sewer construction, has been responsible for 223 pipe breaks and sixteen joint failures in the past 10 years. Also omitted from the records in Fig. 2 and 3 are nineteen emergency situations created by pipe breakage in hydrant leads from automobile colli sions. The use of the breakoff hy- able record. There were no circum ferential breaks in the years 1931 and 1939. The absence of any effect- of severe winter temperatures on the incidence of breaks is to be particularly noted for the years 1936 and 1940. Since 1940, however, the incidence of pipe failures has steadily increased to a high of six failures per 100 mi of CAPCO JEN 0019610 Aug. I960 MAIN BREAKS--INDIANAPOLIS 1055 main in 1959, or six times the average prior to 1941. This increase has been almost entirely flue to circumferential breaks which have accounted for 78 per cent of all breaks in the last 10 years. Other kinds of pi(>e failures have held a rather consistent relation ship with miles of mains, amounting to an average of only about one per year for each 200 mi. An increase in the latter types of failures did occur in 1959 because there were nine in stances of hole development in pipe from corrosion. The corrosion was later found to l>e the result of electroly sis in the vicinity of rectifier stations installed by the telephone company for cathodic protection of their underground cables, which paralleled the water main. joint restraint where thrust blocking is required. The use of kicker blocks liehind fittings is a crude and some times awkward method of countering thrust. Indianapolis is currently using stud type follower rings in addition to blocking on the mechanical joints of fittings as added assurance against thrust failures. J * Jointing Materials Figure 3 shows the performance of jointing material, for joint failures sh.are equally with pipe breaks in the maintenance problems of distribution systems. It can also l>e shown that sulfur coni|xiund failures can be re lated to the temperature stresses im posed on pipelines hv their restraint. No rubber gasket joint failures are recorded, as the trouble experienced with this type of joint has !>een very limited and h3S invariably resulted from impro[)er tightening of the bolts in mechanical joints at time of instal lation. There has l>een no instance of mechanical-joint failure from corrosion of the bolts, although some of the bolts uncovered have shown evidence of gal vanic action. Also excluded from the record are instances of rubber gasket joint failures on hydrants, bends, and other fittings if thrust blocking has shifted, permitting a joint to slip. There is a need for some simple and effective means to provide positive Fie. i- Flpcliflt FUturei by Mouth Of Bepair, 1950-49 The solid oud dashed cur-ees represent circumferential pipe breaks and failures in sulfur compound joints, respectively. As can be seen, circumferential breaks appear to occur during periods of low temperature, whereas joint failures occur during periods of rapid. temperature change. Joint failures have outnumbered pipe breaks in Indianapolis two to one in the past 10 years. Sulfur compound has been responsible for 75 per cent of joint'failures, although only 35 per cent of all joints are of this material. It is interesting to observe from Fig. 3 that lead joints, which have the greatest age, are showing a decreasing CAPCO JEN 0019611 1056 110WARP W. N1EMF.YF.R J.mr.AWWA incidence of failure, whereas sulfur compound failures have licen increas ing. The seasonal pattern of both sulfur compound failures and circum ferential breaks indicates a relationship of failure to water temperature. By plotting such failures by month of re pair and superimposing water tem perature curves on the graph as shown in Fig. 4, one can sec a rather fixed relationship. A similar plot of other types of pipe breaks failed to produce a similar relation. Lead joint fail ures so plotted did show a somewhat frequency curve would' probably lie much more pronounced if the months 0 of failure were known and used for this record rather than months of re pair. Nevertheless, the relationship of failures to temperature is sufficiently clear to establish that the longitudinal stresses in pipelines imposed by re straint of sulfur compound are a cause of its poor performance. It is felt that material fails under -fatigue of repeated stressing. Again; in Fig. 4 the circumferential breaks are shown to start occurring TABLE 4 Indianapolis Main and Joint Failures by Size and MalertaU 1950-59 ! No. ot Circumferential Break* No.'of Joint Failure* | I Sue i*. Pit.Cut Pipe Centrifucally Cast Pij* Lead Sulfur Lead Sulfur Screwed Sulfur Rubber Pit-Cast Pit-Cat Centrifutally 1 Joint Joint Joint Joint Joint Pm* Pipe Cast Pit*e i 1 21 4u 6 jt 20 8 148 17 24 25 43 28 8 10 S IS 1 9 37 10 10 7 1 ll 12 2 3 16 94 24 ' IS 3 15 71 15 20 24-36 3 1 9 52 23 9 6 AH sizes 62 13 20 169 18 128 307 83 * Comtuirtson hould be made with Table 5. higher incidence when water tempera ture was approaching its minimum, although not in a pronounced pattern. From Fig. 4, it can be seen that sulfur comi>ound failures have two peak periods of occurrence, in spring and fall when the water temperature is approximately at the annual mean. Minimum frequency is first at the maximum water temperature and then again after the minimum level has been passed. The peaks and valleys of the when water temperature falls to 40 F and to reach peak frequency when it is at 34F or less. On several occa sions when relief from high frequency has been desired, breaks have been stopped completely by the addition of well water to the surface supply, thus raising the water temperature by 4 or 5F. This stoppage has been achieved regardless of the depth of frost pene tration. It must, therefore, be con cluded that longitudinal stress created CAPCO JEN 0019612 Aug. I960 MAIN" BREAKS--INDIANAPOLIS 10S7 iv temperature change has been a pri mary cause of circumferential pipe breaks during the past 10 years. As there is no evidence of cold-weather effects on the incidence of pi|>e breaks prior to 1941, it is necessary to analyze the records further to find factors that were not present at that time. Comparison of Materials From a classification of circumfer ential breaks and joint failures by sizes and types of materials (Tables 4 and 5), a difference in performance of materials in pipelines can be noted. Tt is significant that 85 per cent of the breaks has been in the 6- and 8-in. sizes and that 70 per cent of these has the same percentage of total mains as the centrifugally cast pijie. The dif ference in the performance of the two ^ypes of pipe with the same jointing material must be attributed cither to the difference in wall thickness nr to the longer joint spacing of the centrifu gally cast pipe, or to l>oth. In compar ing the performance of the sulfur com pound installed in these two types of pipe material, it is found that 32 per cent of its failures in 6- and 8-in. si/.es occurred in the centrifugally cast pipe and 68 per cent occurred in the pitcast pipe, even though the shorter length of the latter reduces the longi tudinal stress to be ahsorlicd by each joint by one-third. TABLE S Pipe and Joint Failures by Material, IVS0-S9 1 y\< of Pipe Type of Joint Pit cast Centrifugal!)' cast lead sulfur screwed sulfur rubber Avg Length of Mams in SeMrv(ice 482 187 5 PH 159 No. of Brraki lr 100 mi. 12.9 7.0 400.0 88.5 11.5 Avg So. of Joints tn Service t.OOffi 241 94 74 No. of Future* LOW Joints 0.55 5.27 1.12 occurred in the unlincd centrifugally cast pipe laid with sulfur compound joints, although the combination of materials represents less than 19 per cent of the total mains in the system. There was a high incidence of sulfur compound failures in the larger sizes of pipe of this construction, although there was a complete absence of pipe hreaks. It is also significant that only 5.5 ocr cent of the hreaks have l>ecn in the pit-cast pipe laid wi;h sulfur compound joints, es'en though the pi|>e- lincs made of this combination of mate rials have been in service an average of 13 years more and represent about Boom Action It is recognized that beam action or cantilever action stresses of varying degree have attended many of the cir cumferential breaks and have contrib uted to their.incidence. On the other hand, in a great many of these breaks the ruptured ends of the pipe have opened up as much as $-in. with the ends remaining in perfect alignment, indicating absence of deflection and the presence of contraction only. Aside from the influence of beam ac tion, the record does indicate that the centrifugally cast p'ipe laid with sulfur compound joints has not satisfactorily CAPCO JEN 10S8 HOWARD W. NIEMEVKR Jour. A WIVA withstood the longitudinal stress from tial breaks in centrifugally cast pipe temperature change and has contrib- with rubber gasket joints has licen uted heavily to the Indianapolis rec shown (Fig. 3) to be much lower than ord of failures. The rising incidence in the same pipe with sulfur compound of failures would indicate cither a pro joints, but not so low as in pit-cast gressive loss of pi(>e strength from cor pipe with sulfur compound joints. It rosion or failures from metal fatigue. must lie recognized that an 18-ft length The pit-cast pipe ap|>ears to have failed of 6- or 8-in. pipe is a reativcly poor from beam or cantilever action, and !>eam and will fail under abnormal beam its joints have failed (roin longitudinal action stresses. It must also he recog stress. nized that such abnormal stresses will Ago of Pipe naturally develop from changing condi tions in a system and are beyond con With regard to failures in the pit- trol of the utility. A trouble-free sys cast pipe laid with lead joints, it should tem can, therefore, lie attained only by be pointed out that these mains range providing allowances for such stresses. in age from .15 to 80 years. This early Heavier wall thickness, shorter lengths, pipe was not manufactured with the or a more ductile iron, at least for the quality control that has been used in smaller diameter pipe, may provide a mote, recent years. It has been subject desirable improvement. Ductile iron to tuheretdation and whatever exter- appears to he more desirable, as it not nul corrosive conditions have existed, only provides the needed elasticity and especially some known electrolysis in deflection limits, but also resists impact the days of streetcars. This pipe is damage in pipe handling and from all contained in the central and oldest accidents. part of the city where many of the streets arc crowded with every possible type of utility structure. Many of the other utility lines lay over or tinder the water mains and they have not always liecn installed with proper clearances. Under such conditions, an incidence of slightly more than one circumferential break per year in each 100 mi of old mains is not surprising. Other Factors Economic Considerations Economic considerations have obvi ously had much hearing on the design of pipelines for distribution systems. Installation savings, however, are only temporary economies, if they result in a high incidence of failures. The prop erly damage that frequently results, the unscheduled disruptions to cus tomer and fire protection services, the The tisc of the rubber gasket joint has eliminated hori/.outal stressing from tcmjicraturc variations, and will help to avoid failures. The use of cost of rcjiairs, and the hardship im posed on maintenance |>ersoniiel who must frequently work lung hours under most adverse weather conditions arc cement lining will eliminate the loss of adverse consequences of^water main pipe strength by internal corrosion. failures. The distribution engineers These improvements, however, will must decide if the failure cxjicneiiccs not prevent failures caused by other in any one system justify the cost of factors. The incidence of circumferen improved materials. CAPCO JEN 0019614