United States Solid Waste And EPA 510-R-92-702 Environmental Protection Emergency Response September 1987 Agency 5403W Causes Of Release From UST Systems ------- FINAL REPORT TO U.S. EPA/OUST CAUSES OF RELEASE FROM UST SYSTEMS EPA CONTRACT: 68-01-7053 SUBCONTRACT: 939-5 WORK ASSIGNMENT: 24 September 30,1987 ------- TABLE OF CONTENTS INTRODUCTION 1 SUMMARY 2 I. GENERAL CAUSES 4 II. TANKAGE 5 A. BARE STEEL 5 B. NEW GENERATION TANKS 7 C. INTERIOR CORROSION 10 D. INTERIOR LINING 11 E. RECERTIFICATION 12 III. PIPING 12 A. SUCTION DELIVERY SYSTEMS 15 B. PRESSURIZED DELIVERY SYSTEMS 18 IV. NON OPERATIONAL COMPONENTS 22 V. SURFACE RELEASES - SPILLS & OVERFILLS 23 APPENDIX 27 REFERENCES 28 TABLE I - OVERALL SUMMARY OF UST RELEASES 33 ATTACHMENT A - QUESTIONS FOR MAJOR COMPANIES 34 ATTACHMENT B - QUESTIONS FOR LOCAL REGULATORS 36 SUMMARY OF CAUSES OF RELEASE 38 TRIP REPORT SUMMARIES 68 SERVICE STATION TESTING COMPANY INC. NOTES OF JULY 8TH, 1987 MEETING - PEI INSTALLERS ¦\ ------- INTRODUCTION This report summarizes EPA's efforts to clarify and gain field observations for the Causes of Release from Underground Storage Tank (UST) Systems. EPA's contractor gathered data from knowledgeable companies and individuals to assess the regulatory positions taken in the proposed underground storage tank regulation. Major oil corporation - Amoco, Ashland, Murphy; large convenience store chains - Southland (7-Eleven), Circle K; local regulatory bodies in California, Florida, Texas and New York; national special interest associations - API, PEI; national trade associations - FRPTI, STI, ACT; and numerous contractors and individuals have shared their opinions, experiences and data to assist in this effort. Sources of information are listed in the Appendix under "References"; each has been assigned a number and where information is cited in the report, the source number appears in parenthesis beside the text. Full reports for the "References" may be found in Volume II where full data and further findings can be gleaned. The Appendix of Volume I also contains sheets summarizing data collection from the sources entitled "Summary of Causes of Release", short summaries of site visits and two of the sources for the report. Data from Service Station Testing and notes from a meeting with PEI installers. 1 ------- SUMMARY In 1987, Jacobs Engineering assisted USEPA in gathering data relating to causes of release from UST systems. Data was sought from major corporations, regulatory agencies, vendors of equipment, and installers to substantiate or question positions taken in the proposed regulation. Individuals were contacted by telephone or site visits. Data was often obtained based on the recollection of an individual's personal experience or from their data files. The data collected has verified most original positions taken by EPA in the proposed regulation. Some new concerns have been raised. The investigations have focused on the four general components of the typical UST system: X. Tankage II. Piping III. Non-operational Components IV. Surface Releases-Spill and Overfill The most significant findings developed to date are: 1. That product delivery piping releases and spills/overfills are the most numerous sources of releases, and not the tankage as originally believed; 2. Numerous tank fittings, vent lines, fill pipes and blind bungs at the top of USTs are loose and leak in the event of overfills even more frequently than tanks and delivery piping; 3. The older "bare" steel tanks do fail primarily by corrosion but the "new generation" USTs of FRP, FRP coated steel with cathodic protection or clad/composite tanks appear to have virtually eliminated failure induced by external corrosion; 4. While corrosion is clearly the major failure mode for existing tankage, corrosion, poor installation techniques and workmanship, accidents and natural events (e.g. frost heaves) appear to be the four major failure modes for [piping; 5. Pressure piping appears to pose a significant threat of "run away" releases and future piping release volumes could be drastically (at least 70%) reduced by a simple and inexpensive retrofit of a continuous line-pressure 4 monitoring device;, the retrofit could be accomplished in the next couple of years; 2 ------- 6. Major upgrade, retrofit and replacement programs have been previously initiated voluntarily by both major oil corporations and large convenience store chains that reinforce the major points of the pending regulation as necessary; 7. Education and enforcement are at the head of the list of requirements for successful implementation of the regulation; and 8. Overall, current causes of tank leakage are definitely controllable; but piping leaks are controllable to a somewhat lesser extent, due to the high probability of human errors during installation and the more vulnerable position/location of the piping near the surface of the ground. 3 ------- I. GENERAL CAUSES The preamble to the proposed regulation reported information that indicates the primary source of UST releases was from tankage. Current additional data tends to indicate that earlier information is somewhat misleading. In fact, current data available to EPA that indicates only 15-20% of the incidents reported are the tanks and result in an average release of 600- 700 gallons of product. The field observations and data have been gathered from people responsible for approximately 175,000 tanks nationwide. Interviewees1 data and comments were summarized and are presented in Table I of the Appendix. Collectively, piping was ranked as the first or second most important (and frequent) cause of release, spills and overfills ranked first or third, and tanks were rated either second or third. Using a ranking scheme of one to three for occurrences of UST releases (pipe, tank or spill/ overfill), the average ranking gathered from these numerous field contacts is: What's happening in the existing UST world? Service Station Testing Company, Inc. of Sari Antonio, Texas (64), has kept very accurate data on their testing work (which is their sole business) over a period from 1981 to mid-1987. The results of their overall data are represented in the graph on page 5. If the available data is examined, which does not encompass spills and overfills, initial testing of existing UST systems would find that 5% of the tanks are leaking, 10% of the UST systems have leaking product delivery lines, and 15% of the UST systems have vents or tank fittings that are not tight under overfill conditions; thus for the non-tight systems, 16% of the total was due to tanks leaking and 84% was due to faulty vents, fittings or delivery piping problems. (Service Station Testing performed 3,746 tightness tests and of those that were non-tight, 92 were due to tank incidents, 176 were due to the product delivery lines, 272 were due to loose tank fitting and vents or fill pipes on top of the tanks.) Of a total of 1,921 tanks that were tested: 228 were FRP, 57 were FRP steel composite and 1,636 were bare steel. All 92 leaking tanks were made of bare steel and all but 4 were over 10 years old when found to leak. TABLE A RANKING CAUSE OF RELEASE POSITION 1 2 3 SOURCE Pipe(Pressure) Spill/Overfill Tank 4 ------- TOTAL SYSTQC TESTS) = 1921 25% OF SYSTQ6 FAIL TImms TEST DELIVERY PIPES I QPmATICNftL | LEAKS TW< FITTIMSS vacs, 1 leak my FILLPIPES, I MING BUNG HOLES I J OVffiFILLK GRAPHICAL reTCSeHATION OF DATA FROM SERVICE STATION TESTING COMPANY, INC. OF SAN ANTONIO, THAS ------- II. TANKAGE A. BARE STEEL Estimates provided by very experienced installation contractors (22) were that 50% of tanks in existence could not pass tightness testing five years ago, and their experience suggests that this figure has been reduced by increased awareness, use of new equipment, and contractor education to the point where these contractors presently believe the figure to be less than 20%. About 75% of the existing tank population is of the "bare" steel type, and the majority have been in the ground for at least 10-15 years—the critical time period for their failure by corrosion. The ticking time bomb analogy that has been used in the past concerning these tanks is significantly mollified however, by numerous reported field observations that many existing tanks have at closure been seen to have "plugged" corrosion holes that do not show any evidence of leaking when unearthed. Also, field observations, including several local communities that were visited (for example Austin, Texas) (11), indicate that numerous old tanks of bare steel are being,closed which are in excellent shape with no holes. Another example is Suffolk County New York's investigation (16) for EPA which is showing about one third of the older closing tanks have corrosion perforations, and half (or 1/6 of the total) of these show signs of leakage about half of those studied did not have significant corrosion. Tank testing programs (based on about 10,0,00 tank system tests) indicate that about 5 to 7% .of tanks actually leak when they are tested for the first time. Very few of the tanks less than 12 years old are ever found to have holes. Generally, most tankage is presently of the "bare" steel vintage; of the total tank population some 70-80% are "bare" steel. This type of tankage is gradually decreasing due to voluntary upgrade programs, local regulation, and the federal interim prohibition. Externally coated and cathodically protected steel tankage, such as STI-P3, account for about 8% of the existing population. Their usage has recently experienced a very sharp increase (since their introduction some twenty years ago). Another 12-15% of the existing tankage is made of fiber- glass reinforced plastic (FRP) construction. Another 8% of the population is a mixture of clad, composite and corrosion resistant metals. The existing UST world is presently estimated to be as follows: 6 ------- TABLE B THE EXISTING IJST WORLD Present Share Estimated Future Growth Tvne of Tank of Population Number in Existence1 Trend2 "Bare" Steel 70-80% 900,000 to 1,000,000 Rapid Decrease Coated with CP 8% 100,000 Rapid Increase FRP 12-15% 156,000 to 195,000 Moderate Increase Composite, 5-8% Corrosion Resistant 65,000 to 100,000 Moderate Increase 1Based on EPA's estimate of 1,318,000 UST systems in existence - See Table I of preamble. ®Based on PEI meeting, also see Table F. Numerous tank failure histories indicate that, when failure occurs, 95% of "bare" steel tankage fails from corrosion. There is a wide disparity of opinion about how to assign causes of release due to external, internal, or a combination of both types of corrosion. Accurate data or studies which convincingly differentiate among corrosion causes are very few, and internal tank inspections are not common. Based on opinions of major corporate owners, tank lining companies and independent consul- tants studies, the estimated spread (Table C) provide a rough approximation of the cause of corrosion holes (about 50% of which are probably rust plugged and don't leak) in "bare" steel tanks: TABLE C CAUSE OF CORROSION PERFORATIONS TYPE OF AVERAGE AGE % OF TOTAL CORROSION AT FAILURE CORROSION FATLITRF Internal 10-20 yrs 6-10 External 10-20 yrs 70-80 Combination 10-20 yrs 15-19 Tabulation of testing data from Service Station Testing (64) (Table D) reinforces the data in Table C. 7 ------- TABLE D RESULTS OF 980 STEEL TANKS TESTED (WHERE AGE WAS SPECIFIED) TANK AGE NO. OF TANKS NO. OF TANKS LEAKING 6 years 190 2 6-11 years 145 4 — — — — — — — — — — — — — — — - -BREAKTHROUGH 12 years 38 5 OF CORROSION 13 years 30 3 BEGINS 14 years 55 1 15 years 80 5 16 to 20 years 252 11 20 years 190 11 Data submitted by the Internal Tank Lining Industry (24) supports and substantiates the above results (Table F): TABLE E TANK AGE 0-5 5-10 10-15 15-20 20-30 30+ Years Years Years Years Years Years AGE RELATION TO FAILURE NUMBER %_ 232 0. 1,204 4, 7,391 30, 10,336 42, 4,478 18, 811 3.4 BASIS: 24,452 Tanks found to be Leaking and subsequently repaired and lined. All tanks are bare steel. The clarion message from the field on over 90% of tank failures (17, 18, 22, 39) to date is that the primary cause is due to improper backfill: it is not select (clean sand or pea gravel); if select, it is contaminated with rubbish, wood or other soils; or it is improperly placed and compacted. Of all the current failure modes, corrosion of "bare" steel is by far of greatest importance; and the tank manufacturers have responded with exterior coated and protected steel tanks and tanks of corrosion-resistant materials such as FRP. 8 ------- B. NEW GENERATION TANKS As early as twenty years ago, manufacturers began to respond with innovations to attack the number one cause of tank failure - exterior corrosion. Tanks began to appear that were all fiberglass reinforced plastic (FRP), steel coated with a non- corrodible resin or plastic and having sacrificial anodes and clad or composite construction. Initial acceptance by owners and operators was slow due to higher initial costs. However, as environmental awareness increased, sales began to rise, slowly at first, but a dramatic acceleration in utilization of new generation tanks occurred with the introduction of the Interim Prohibition. Representatives of the various trade associations for the individual types of new generation tanks have provided sales data for the period from 198 0 through 1986 and estimates for 1987 - see Table F. TABLE F PRODUCTION OF NEW GENERATION USTs Year FRP1 Comoosite2 STIP33 1980 9,000 N.A. N.A.4 1981 10,000 N.A. N.A. 1982 11,000 N.A. N.A. 1983 12,000 3,000 N.A. 1984 13,000 6,500 7,000 1985 14,000 8, 000 14,000 1986 15,000 10,000 28,000 1987(est) 16,000 12,500 45,000 JEd Neshoff, Data from FRPTI. 2Bob Holland, Data from Association of Clad Tankers sWayne Geyer, Data from Steel Tank Institute 4N.A. - Not Available Most existing steel tankage that is coated or FRP-clad on the exterior, or fitted with cathodic protection, is less than five years old. However, some tank systems of this type are at 9 ------- least 13 to 20 years old. So far, reported failures observed in the field due to corrosion (or other reasons) from such tanks are very rare, if any. Clad tankage is very popular in Sweden and Denmark (70) where officials report their tank problem "has gone away" since such tankage was required in 1972. Clad and composite tankage has been produced in this country for 15 to 20 years in the U.S. There is no known case of a clad tank's failing from corrosion; in fact, manufacturers report today's clad tanks are even better than 10 years ago. One group of tank manufacturers who have formed the Steel Tank Institute, produce a protected new generation tank, STIP3, of steel coated with a non-corrodible resin or plastic material and have sacrificial anodes for additional corrosion protection should the non-corrodible coating be damaged and the bare steel exposed. Installation contractors (22) in the field report if we used this type years ago, the exterior corrosion problem would not exist today. The STIP3 tank is a favorite of corrosion engineers. Very few failures have been reported and those failures are due to installation damage or improper maintenance, not design (21,22). In the Province of Ontario, Canada, STIP3 tanks have been widely used and the tank releases from corrosion are going away. FRP tankage appears to rarely fail due to corrosion (e.g., because unanticipated solvents are encountered which are incompatible with the tank resin and dissolve it). Overall, annual failures of all existing FRP tankage appear to have occurred at less than a rate of 0.25% per year of the total of FRP tanks installed nationwide (21) (conservatively computed based on the number of failures in one year—in a total population of 200,000 divided by the number of tanks manufactured in one year). Numerous sources appear to support the field estimates collected by EPA that less than 0.5% of the existing FRP tanks have ever had a problem. Even these small failure rates represent a decline of 50% between 1976 and 1986 as reported by Owens Corning Fiberglass. Failures in FRP tanks have happened very early in the tank's life due to cracking, however most of this type of failure occurred over 10 years ago and appears to be rare today. The tank manufacturers, several tank owners, as well as installation contractors claim these FRP failures were primarily caused by very poor installation practices or, on very rare occasions, by a defective tank. For example, a group of 8 installers (22) from around the country identified 8 failures in 1500 to 2000 installations, Ashland Oil (48) has recorded only one failure in 107 FRP installations, CAE Fiberglass and the Ontario Government's Fuel Safety Branch (45) reported one failure in 7,000 FRP tanks; Circle K Convenience Stores (39) have 10 ------- recorded one failure in 2000 installations, and The Southland Corporation (12) has recorded 19 failures in 3000 installations. Circle K and Murphy Oil (15) have totally based their new and retrofit programs based on FRP tanlcs (as have many other major oil companies in the U.S.)* Heightened installer awareness of proper practices and techniques appropriate to FRP technology, manufacturer-sponsored contractor education programs, and production quality assurance appear to be responsible for the present low failure rate of FRP tanks (21). It appears that many of the reported FRP installation failures occurred over 10 years ago (22). Double wall steel and FRP tankage has been introduced to provide secondary containment for UST releases. Present usage appears to be concentrated in jurisdictions (3,4,5,6,7,8) with sensitive environmental areas. The cost of this type of tankage has decreased since introduction to the market place. One contractor group (22) felt, double walled tankage to be one the better potential solutions for tank releases but, they noted lack of operating histories and costs have held voluntary usage at a low level. C. INTERIOR CORROSION Interior corrosion of steel tanks appears to be another failure mode with steel tanks (21, 24, 31, 40, 70), but thus far has been largely ignored. New tank designs have addressed and greatly reduced the exterior corrosion failure potential. As exterior corrosion recedes through more preventive measures, it is possible that interior corrosion will eventually become, over the long term, the primary steel tank failure mode. However, the incidence of corrosion induced tank failures is expected to then be significantly less than today and take longer to manifest itself after external corrosion is prevented through new tank designs. Studies in Sweden and Denmark (58, 70) indicate internal corrosion to be a significant cause of release when storing gasoline and the main'cause of release if storing fuel oil. In Switzerland, internal corrosion was found to be the cause of release in 5% of the investigated incidents. In Denmark (18) and Sweden (17) it is considered so severe that internal sacrificial anodes are required and internal inspections are required every 10 years to examine the internal tank structural condition (anode weight is designed to provide protection for a 10 year period). Numerous contacts in private industry (13, 14, 15, 25, 35, 36, 38, 39, 40) have reported problems with pitting and perforations inside of steel tanks under the drop tube. The tank liners data confirms these reports and the tank industry has voluntarily responded by providing "striker plates" under all openings. (They are required by UL in Canada.) Where internal 11 ------- corrosion is identified generally, the breakdown by location is given in Table G. TABLE G LOCATION OF INTERNAL CORROSION TANK NO. At the Sludge Line Upper Tank Pitting Pitting Under Drop Tube Pitting in Bottom of Tank Holes Under Drop Tube Other 12,291 8,283 1,228 2,296 1,652 259 58 9 16 86 12 2 percentages add to more than 100% as more than one location was reported for a single tank. D. INTERIOR LINING Tank interior lining has been identified as a world-wide technology. In the U.S. it is a widely used technique that has been employed by major corporations (e.g., Amoco (14), Ashland Oil (35), as well as by small owner/operators) as a short term, but effective, solution for both older or perforated and repaired tanks, or as preventive maintenance measure for sound non-leaking tanks. Data received from Ashland Oil (35), Shell Oil of Canada, the Ontario Fuel Safety Branch (10) and numerous data from the tank liners themselves, indicates this to be a successful procedure for extending an existing tank's non-leaking life. Even when employed in the absence of external cathodic protection—failure rates are reported to be very low. This technology is reported to be used widely in Europe (70). Two tank lining companies (24) have submitted data to EPA that was collected from their installers in the field, this data covers 35,349 motor fuel tanks which have been lined; 26,000 of the tanks were leaking at the time of repair. The tanks were lined with a 120 mils thickness (about 1/8 inch) of coating after the interior tank shells were sandblasted and perforations were repaired. Only 197 tanks have been reported as failed since lining (0.5% of the tanks lined). The tank liner installers also indicated that internal corrosion was a major cause of failure, either alone or in conjunction with external corrosion. Their data further indicates internal corrosion has caused failure in 45% of the repaired USTs. Cathodic protection was not 12 ------- retrofitted on the repaired USTs and, in fact, about 1100 tanks had cathodic protection prior to repair. 13 ------- E. RECERTIFICATION A new area has been identified through the investigation: recertification and reuse of tankage (21). It is apparently not uncommon for some people to dig up and reuse a protected tank elsewhere. Presently this practice has been used with some FRP tankage and FRP tank manufacturers offer recertification, warranty continuance and even warranty transfer to third parties. Additional information on procedures, criteria for acceptance and the possible extension to steel tankage maybe necessary in the future as more of the long-lasting new tank varieties are placed into service in one location and then later moved. The Steel Tank Institute and API have reported to EPA that they do not foresee this as an area of immediate concern. III. PIPING The preamble to the proposed EPA regulation cited reports that indicate the contribution of product delivery piping as a cause of release to be less than that of tanks. However, virtually all field contacts made over the last several months rate delivery piping or fittings on top of the tank as the primary cause of release and estimate that it was responsible for 80 to 85% of all releases. Actual files and written databases on this subject appear to be few and imprecise. Most local regulatory release incidents reports did not distinguish between piping or tank releases. Where they do exist they are usually the assumptions of inspectors in the field who see only the disinterred tanks, because the piping is often left in the ground. The primary cause of piping failure is cited to be installation practices and techniques. The complexity of a typical piping system may be appreciated by examining Figure 1 which schematically shows the amount of pipe, numbers of fittings and changes of direction in a typical retail motor fuel outlet. Each joint is a potential leak source. Two types of piping (delivery) systems are now employed in dispensing product from USTs: suction and pressure. Presently several experienced contractors estimate a roughly equivalent use of both systems in the retail motor fuel sector; however, 95% of the new UST systems in high volume retail applications are reported as installing the pressurized type while 90% of the new and existing non-retail motor fuel installations are still of the suction type system. 14 ------- Tank Truck 8utMMrg«d Pump. Assembly 8ut>m«rg«t Pump Assembly Product Dispensers Typical Four-Tank Station ------- TABLE H ESTIMATED USAGE OF PRESSURE AND SUCTION DELIVERY SYSTEMS Existing non-retail motor fuel sector; (Approximately 705,000 tanks) (New tanks) Existing retail motor fuel sector; (Approximately 676,000 tanks) (New tanks) % SUCTION 90% (90%) 40% (10%) o/n PRESSURE 10% (10%) 60% (90%) There also is reported to be a wide variation in the potential size of releases from the two types of piping systems. Service Station Testing (64) found 9.2% of pressurized piping systems (of 1351 total tests) and 6.8% of suction piping systems (of 474 total tests) non-tight. In the absence of large databases, several experienced contractors (22, 24, 47, 64, 69) have been consulted. Contractors repair and remove systems as well as install them and have continuing exposure to the primary causes of line failures. Their consensus was that piping systems do not enjoy the same longevity as tanks. Frequent modifications and routine alterations at the tank site tend to reduce the undisturbed life span of piping. Their field experience indicates failures can be attributable to two factors: corrosion and leaking joints - which are commonly induced by poor installation practices. If line systems were left in place for 30 years, contractors believe failure from corrosion would account for a 20% failure rate and damaged or loose fittings for another 40%. Corrosion is precipitated by non-select backfill and contaminated backfill; therefore clean (select and uncontaminated) backfill should greatly reduce the corrosion problem, but some type of cathodic protection is still required. Presently no pre-engineered cathodic protection is available for piping, most steel piping is currently protected by galvanizing, coating and wrapping, or coating alone, and the threaded portions at joints is the most common failure point because the protection is removed while threading and never replaced. If threaded steel pipe is used, some type of sacrificial anode system for cathodic protection would eliminate some fitting failures due to installation errors. Fitting failure is from either corrosion, untightened joints, cross- threaded joints or improperly made joints. Contractor education and skills in the complex pipe installation task need to be improved. 16 ------- Piping systems are of two materials of construction: metal or FRP. The contractors (22) suggest that they both have unique advantages and disadvantages. Both installer/contractors (22) and owners (12, 39) have estimated that piping is damaged 10% of the time at new installations sometime between the completion of installation of equipment and completion of paving. Therefore, they clearly recommend that some type of pre-start-up function test is essential as a sound practice, particularly with pressurized piping. TABLE I COMPARISON OF THE COMMON MATERIALS USED IN UST PIPING SYSTEMS METAL PIPING SYSTEM 1. SUBJECT TO CORROSION 2. HEAVY 3 . HIGH RESISTANCE TO CRUSHING/FRACTURE 4. JOINTS FAILURE BY TENSION-LOWEST POTEN- TIAL 5. LITTLE FROST HEAVE FAILURE 6- HIGH PUNCTURE RESISTANCE 7. SPECIAL SKILLS REQUIRED FOR ASSEMBLY 8. FABRICATION TOOLS REQUIRE CONSTANT CARE AND ATTENTION 9- COLD DOES NOT AFFECT FABRI- CATION FRP PIPING SYSTEMS NON CORROSIVE LIGHTWEIGHT LOWER RESISTANCE TO CRUSHING FAILURE JOINT FAILURE BY TENSION-HIGHEST POTEN- TIAL HIGH FROST HEAVE FAILURE LOW PUNCTURE RESISTANCE SPECIAL SKILLS REQUIRED FOR ASSEMBLY FABRICATION TOOLS INEXPENSIVE THROW AWAY TYPE CATALYZED JOINT CEMENTS REQUIRE 60°F FOR PROPER CURE 17 ------- A. SUCTION DELIVERY SYSTEMS Suction dispensing lines are considered much more intrinsically safe than pressurized lines because they operate at less than atmospheric pressure between the tank and the dispenser; thus, during operation fluids outside the pipe will leak in while the conveyed fluid will not leak out. This simplistic approach generally leads some to a conviction that a suction system should be used in all cases and pressure systems not employed. However, a closer comparison of the two systems indicates that the suction type is not always the most ideal operating type of system (See Table J). While suction systems offer the least expensive approach to reduce the threat of piping-related releases, they do not work well at high altitudes, in hot climates or in high-volume delivery situations. TABLE J COMPARISON OF PIPING SYSTEMS SUCTION TYPE PRESSURE TYPE 1. NEGATIVE DELIVERY TO DISPENSER POSITIVE DELIVERY TO DISPENSER 2. LIFT INCREASES PUMP WEAR FLOODED SUCTION-NO CONTRIBUTION TO PUMP WEAR 3. VAPOR LOCK FROM ALTITUDE OR HEAT NO VAPOR LOCK 4. MAXIMUM LIFT IS 15 FEET (LIMITS BURIAL DEPTH OF TANK) NO LIFT PROBLEM- fUNLIMITED BURIAL DEPTHS 5. LITTLE OR NO RELEASE TO ENVIRONMENT POTENTIAL FOR LARGE RELEASES TO ENVIRONMENT 6. PIPING DESIGN, LAYOUT VERY CRITICAL PIPING DESIGN. LAYOUT LESS CRITICAL 7, INHERENT RELEASE PREVENTION ADD-ON RELEASE PREVENTION 18 ------- Review of suction systems with contractors (22), owners and equipment manufacturers indicate that suction systems cannot be utilized in all situations. The maximum lift capability of a suction pump is reported as fifteen (15) feet. Due to the lift restrictions of the pump, a nominal tank of 10 foot diameter with 2 feet of cover, the tank would have to be located within 50 feet of the dispenser as the lift is consumed by line friction losses. Additionally, manifolding of suction delivery lines cannot be practiced which requires additional lines per site, increasing installation costs and increasing the potential release sites. Ideally, the tank also should be located directly below the suction pump and the lift requirement held to a minimum to reduce wear on the pump. The location of the check valve in a suction piping system has been of concern. In Europe (70), the check valves are located just below the pump; in the United States, most check valves are located at the beginning of the suction line near the bottom of the UST, which maintains the product delivery line full of free product at all times. Placement of the check valve at the top of the tank is also practiced, utilization of a foot valve is beneficial in reducing a pump's power consumption and the wear and strain on the pump. However, placement of the valve near the dispenser is beneficial in reducing the volume of a potential release, as the product will drain back into the tank in preference to through a hole in the pipe and into the environment. B. PRESSURIZED DELIVERY SYSTEMS Pressurized piping systems are reportedly on the increase in the retail motor fuel sector, representing about 95% of new retail motor fuel systems installations (22). The turbine pump is submerged in the product in the tank? the piping from the pump discharging to the dispenser is normally at operating pressures of 3 0 pounds per square inch. A check valve next to the submerged pumps discharge point is used to maintain the fluid in the line at operating pressure during product delivery, the pressure is reduced to 8 - 12 PSI and held even while the pump is not operating. Should the delivery line be breached, free product will be released until the pressure in the pipe is reduced to the pressure outside the pipe and equilibrium is established. Without add-on instrumentation or devices, this pump can rapidly push large volumes of product out of breaches in the line during operation when product is called for (at the dispenser). However, in a leaking line product will generally not only be forwarded through the dispenser to a customer, but also through the hole into the environment at the same time. The pump simply pushes more volume to meet this dual increase in demand. 19 ------- The consensus from the field was that releases from pressurized piping systems clearly can be catastrophic in the absence of monitoring and automated pump flow restriction devices - one incident of a release of 20,000 gallons in one day was reported (22). While such catastrophic high volume releases are the exception, the field experiences of nine contractors cited their ability to recall easily over one hundred and fifty large volume pressurized releases. One contractor's field observations included estimates of a typical size range of between 600 to 6,000 gallons withput the use of automatic detector/flow restriction devices. However, even with the use of these commonly available devices, the expected high number of release incidents from piping at rates of 3 gallons per hour or less would still indicate a substantially larger volume of product being released from pressurized piping than from tanks. For example, in Dade County, Florida (9), piping releases account for 21% of all written data on releases (215 incidents from 1984 to April 1987). Line losses by volume are tabulated from Dade County files as: TABLE K DELIVERY LINE PRODUCT LOSSES DADE COUNTY. FLORIDA 1984-1987 No of Incidents Volume of Release 2 10-99 Gallons 3 100-499 Gallons 3 500-999 Gallons 7 1000-9999 Gallons 2 10,000+ Gallons As previously mentioned, one very experienced contractor/line tester reported pressurized line leaks as commonly falling into the 600 gallon to 6,000 gallon range. The most common and readily available automated, in-line pressure device reduces the release rate, but does not stop the release; however, if it is carefully monitored or maintained, it is reported by several experienced contractors/installers to have significant mitigating value. Unfortunately, about half of all owner/operators with pressurized lines were reported to have not installed these devices in an effort to reduce their initial investment capital outlays. If installed and properly monitored and maintained, one experienced ad hoc workgroup (22) of installation contractors estimates that 70% of the volume of product lost through pressure pipe releases from existing UST systems could be avoided (within two to three years) by retrofitting each line with a simple, inexpensive continuous in- 20 ------- line pressure monitor that automatically restricts product flow in the presence of a significant line leak. Present models of these devices are commonly reported by installers to be more dependable and fail safe,. (A maximum retrofit cost of $1,000 has been indicated, but with a typical total cost of $500 per pump in 80% of the cases.) Several companies are now performing simple pressure tests on piping on an annual basis. Pressure is applied from the impact valve back to the pump's check valve, the pressure is observed over a 30 minute to 1 hour period for decay. Loss of pressure instigates more detailed investigation which has located faulty line leak detectors, loose fittings, faulty check valves and line corrosion failures. The cost of an annual test of this nature is from $300-$500 per site. (This type of test could be utilized to test suction systems also.) A potential method of continuous monitoring of pressurized vljfcies has been identified. A pressure gauge could be installed at or near 1:he dispenser and the gauge observed during periods of dispenser inactivity. A loss of pressure to less than 5 psi in thirty minutes would indicate potential loss of system integrity. The additional cost for this check at new installations would be in the $25 to $35 range; however, care to bleed all air from the line prior to gauge installation is necessary. Sophisticated remote monitoring using pressure transducers would raise the cost into the range of $500 to $600 per dispenser. 21 ------- w m p o i a OJ Typical Tank System Assembly Tan* Product Lin® Piping Unions Swing Joint Flex Hom Submerged Pump V«nt Line 8. Fill Pipe Riser 8, 4" Plug Into Tank Bung 10. Dlsp«nalng Unit ------- IV. NON OPERATIONAL COMPONENTS Numerous data has been obtained primarily from several commercial tank testing surveys (64) concerning the field performance of non-operational components of tank systems. The testing was most often due to local government testing programs, and the data available to EPA corroborates a widespread failure of non-operational components of the tank system. These components provide the most common source of system non-tightness under conditions of a standpipe tightness test. These non- operational components consist of: (See Figure 2) A. Tank bung holes B. Tank manholes C. Vent and fill lines D. Vapor recovery lines E. Manifold piping (connects tanks together) These components are called non-operational because releases from these sources are episodic and of small volume when they occur because they only occur when an UST is overfilled or manifolded tanks are filled through one of the connected tanks* drop tube. In other words, they do not leak under normal operating conditions because they are located above the top of the tank. Releases from the following common sources are reported (22) as the result of improper installation practices: 1. Tank bung hole protectors are not replaced with screw- in plugs at installation. 2. These bung plugs are not tightened at installation. 3. Vent lines are fabricated of the wrong material, e.g., PVC. 4. Vent line and vapor line joints are not tightened or cemented because they only contain "air". 5. Poor backfill or site selection give rise to tank settling. 6. Vehicular traffic can damage vent line and fill pipe connections to the tank. 7. Improper cover or pavement thickness can lead to damage from normal traffic. 23 ------- Service Station Testing Company (64) in San Antonio, Texas, has performed in excess of 3700 tank and system tightness tests. Of the systems tested, 364 were found to be non-tight and 272 (74.7%) of the test failures were the result of non-tight tank . fittings or vent lines. In the "Summary of City/County Reports" (67) it is noted that 13% of the identifiable causes of release are directly attributable to loose tank fittings. A 1986 draft EPA report (68), in 1986, which investigated 158 release incidents found that 15.5% of the releases are attributable to fill pipes and vent pipes. Numerous unreported incidents are believed to have also occurred to date. Preliminary results from an on going EPA sponsored investigation in Suffolk County (63), N.Y., that has been corroborated by numerous installation contractors nationwide, report that exhumed bare steel tanks show evidence of non-operational sources of leakage which has been seen to deteriorate the exterior bitumen or asphaltic coating on the tank shell. The deterioration is traceable to leaks at fill pipes, vent lines and bungs from the pattern of deterioration and the discoloration o£ surrounding soils. Additionally, recently released free product was sometimes in evidence in the soil surrounding the UST. Releases from these non-operational components are difficult to detect without the use of precision tightness tests or exhumation of the top of the tank system, because the release occurs only when filling a tank or overfilling occurs, these releases are too small to be detected by any inventory monitoring system. Two avenues are obviously available to stop this type of release: ensure proper installation or eliminate overfills. Elimination of overfills is believed to be the most fail-safe remedy and probably the easiest to implement. For example, a recent EPA visit to a prominent tank manufacturer revealed they are still having significant problems in getting tight bong hole covers applied at the factory. If the stored product is never allowed to reach these system weak points, above or on top of the tank then it can never be released. This appears to be the widespread approach to addressing the problem in several European countries. V. SURFACE RELEASES - SPILLS AND OVERFILLS Spills and overfills (along with the ensuing releases from non-operational components) are probably the most common type of UST related release to the environment. It is believed that most incidents go unreported due to the typically small volume of product lost (less than 20 gallons). Most excavated "bare" steel 24 ------- tanks show evidence of spilled material, e.g., asphaltic coating near the drop tube bung has been dissolved, discolored soil is present, etc. Regulatory officials in Dade County, Florida (7), cite spills/overfills as the primary cause of release— 45% of incidents reported—and twice the tank or piping problem. TABLE L SPILLS AND OVERFILL LOSES DADE COUNTY. FLORIDA m 1984-1987 OF SPILLS VOLUME OF SPILLS 9 10-99 Gallons 5 100-499 Gallons 3 500-999 Gallons 3 1000-9999 Gallons 0 10,000 + Gallons Data from Virginia's State Water Control Board (23) documents spills and overfills being responsible for 12% of all UST related releases. Documentation of European (70) experience cites 63% of releases due to overfilling and 65% of these overfill releases were less than 2 65 gallons. Experienced installation contractors (22) carefully and repeatedly suggest that spills and overfills should not be lumped together, they point out that attempts to control one may not control the other. Spills are reported to usually occur at the time delivery hoses are disconnected from the tank fill tubes, because the delivery hose either was not drained or the disconnect stop valve (on the truck's fill tube) was not completely closed. Overfills are primarily a result of the failure to gauge a tank's available capacity against the quantity being delivered. Informal discussions conducted by EPA with an ad hoc installation contractor group (22) pointed out that deliveries were often made at night, and drivers are in a hurry because they are paid by the loads delivered, not by the hour. Two former delivery truck drivers in the group estimated the following frequency and size based on their own experiences with the industry. 25 ------- The spilling or dumping, of small amounts of product, as cited by these former transporters, hasn't been previously seen as an environmental problem in the industry. Its curtailment was only governed by the ethic of not wanting to throw away valuable product. However, in the middle of the night with no one else around, a delivery route only partially completed, and nowhere else to put excess product, circumstances dictated throwing it away "down the hole". Several corrective steps have been suggested to stop this bad practice. (Table N) Numerous European countries appear to have been requiring the use of overfill protection devices. Switzerland, West Germany, France and Sweden (70) require automatic shut-off overfill devices. Automatic sensor shut-offs in addition to other automatic shut off devices are utilized in Europe. Ball float valves have been employed in the United States but operating difficulties have arisen in conjunction with coaxial vent and vapor recovery systems (ball float rises and stops delivery flow due to the reduced relief capacity of the vent line). Catchment Basins are also available and sometimes used, in the U.S. to contain small spills from hoses during the delivery process. They are positioned to surround the top of the fill tube and (depending on design) hold from 5 to 45 gallons of product. Generally, they must be manually drained into the tank after the product level in the tank drops, through dispensing of product. Numerous contacts cited reservations/operational problems concerning the use of catchment basins. 1. Water accumulation (due to rainfall) which is erroneously dropped into the tank and can facilitate internal corrosion especially if salt (in the air) is present (as in Northern and Coastal Regions). 2.. Failure to drop the contained fuel into the tank can allow a safety hazard to develop because fuel in the basin will foster vaporized gasoline and air to combine and make a potentially explosive mixture. 3. Crossing vehicular traffic can cause friction between the metal cover and lid over the basin cover creating sparks that fall into the reservoir. 4. Transporter failure to inform the owner/operator that material has been spilled into the basin which exacerbates the above cited problems. Elimination and containment of spills and overfills is an area where new and improved equipment are fast becoming available. Numerous contacts with the field suggested they should be encouraged. 26 ------- TABLE M TRANSPORTER ESTIMATES OF SPILLS AND OVERFILLS (22) Frequency 1 of every 25 deliveries 1 of every 100 deliveries Size of Spill/Overfill spill 3-5 gallons overfill and release 20-30 gallons TABLE N POTENTIAL CORRECTIVE ACTION FOR SPILLS AND OVERFILLS CORRECTIVE ACTION Tank manual dipping at delivery Automatic level indication Ball float check Valve ADDRESSES Overfill Overfill Overfill ADVANTAGES Fast, inexpensive DISADVANTAGES Degree of accuracy, human error Degree of accuracy Expensive to install Simple, automatic Problems with coax vapor recover & vent systems positioning at installation (22) Leak Tight Disconnect on Hoses Catchment basins (14, 15, 22, 24) Driver (20) Education & Certification (Maryland) Spills Spills/ Overfills Spills/ Overfills Fast, Inexpensive Maintenance Civil Fines (2, Spills/ 3) (San Diego) Overfills Contains small quantities (up to 40 gallons) Inexpensive (To Owner) Manual draining, explosion hazard, water contamination of product Human error Failure to Report 27 ------- APPENDIX 28 ------- REFERENCES TRIP REPORTS 1. PIECO (Petroleum & Industrial Equipment) Hialeah, Florida - May 20, 1987 - Frank Hicks (JEG) 2. SAN DIEGO FIRE MARSHALL San Diego, California - May 19, 1987 - Tom Willard (VERSAR) 3. SAN DIEGO DEPARTMENT OF HEALTH SERVICES San Diego, California - June 1, 1987 - (VERSAR) A. SAN FRANCISCO WQCB San Francisco, California - March 27, 1987 - Tom Schruben (Jacobs) 5. SUNNYVALE TRIP I Sunnyvale, California - March 27, 1987 - Tom Schruben (Jacobs) 6. SUNNYVALE TRIP II Sunnyvale, California - April 26, 1987 through May 6, 1987 - Bill Meyers (VERSAR) 7. BROWARD COUNTY Broward County, Florida -- April 2, 1987 - Tom Schruben (Jacobs) 8. DADE COUNTY Dade County, Florida - April 3, 1987 - Tom Schruben (Jacobs) 9. DADE COUNTY FILES Dade County, Florida - May 13, 1987 through May 14, 1987 - Frank Hicks (Jacobs) 10. ONTARIO PROVINCE - FUEL SAFETY BRANCH Ontario, Canada - May 20, 1987 through May 22, 1987 - Robin Parker (Jacobs) 11. AUSTIN Austin, Texas Trip I - March 9, 1987 - VERSAR Trip II - April 27, 1987 - VERSAR 12. SOUTHLAND CORPORATION Dallas, Texas - July 9, 1987 - Robin Parker (Jacobs) 13. RYDER TRUCK RENTAL INC. Miami, Florida - May 18, 1987 - Frank Hicks (Jacobs) 14. AMOCO OIL COMPANY Chicago, Illinois - June 18, 1987 - Robin Parker (Jacobs) 29 ------- TRIP REPORT (Continued) 15. MURPHY OIL COMPANY El Dorado, Arkansas - May 27, 1987 - Robin Parker (Jacobs) 16. SUFFOLK COUNTY Long Island, New York - March 4, 1987 17. SWEDEN - April - Phil Stapleton (Dames & Moore) 18. DENMARK - April 2, 1987 through April 3, 1987 - Phil Stapleton (Dames & Moore) 19. SAN JOSE FIRE PREVENTION BUREAU San Jose, California - April 3, 1987 - Tom Schruben (Jacobs) 20. MARYLAND Annapolis, Maryland - April 20, 1987 - Tom Schruben (Jacobs) 21. FIBERGLASS REINFORCED PIPE & TANK INSTITUTE McLean, Virginia - July 14, 1987 - Robin Parker (Jacobs) 22. PEI INSTALLERS Dallas/Ft. Worth Airport - July 8, 1987 - Robin Parker (Jacobs) 23. VIRGINIA STATE WATER CONTROL BOARD Richmond, Virginia - August 5, 1987 - Elaine Strass (Jacobs) 24. ARMOR SHIELD Cincinnati, Ohio - June 19, 1987 - Robin Parker (Jacobs) PHONE CALLS 25. PIECO (Petroleum & Industrial Equipment) Hialeah, Florida - May 8, 1987 - Frank Hicks (Jacobs) 26. STATE OF CALIFORNIA - May 21, 1987 - Tom Schruben (Jacobs) 27. STATE OF NEW YORK (Paul Soss) - May 20, 1987 - Jacobs 28. HERTZ - May 13, 1987 - Robin Parker (Jacobs) 29. MURPHY OIL COMPANY El Dorado, Arkansas - May 13, 1987 - Robin Parker (Jacobs) 30 ------- PHONE CALLS (Continued) 30. MURPHY OIL COMPANY El Dorado, Arkansas - July 22, 1987 - Robin Parker (Jacobs) 31. ARKANSAS BEST FREIGHT SYSTEM Ft. Smith, Arkansas - June 4, 1987 - Robin Parker (Jacobs) 32. SOUTHLAND CORPORATION Dallas, Texas - May 11, 1987 - Robin Parker (Jacobs) 33. MOBIL OIL Fairfax, Virginia - May 14, 1987 - Robin Parker (Jacobs) 34. NEW YORK SPILL PREVENTION & RESPONSE - May 20, 1987 - Tom Schruben (Jacobs) 35. ASHLAND OIL Ashland, Kentucky - June 16, 1987 - Robin Parker (Jacobs) 36. BOEING Seattle, Washington - June 11, 1987 - Robin Parker (Jacobs) 37. SOUTHLAND CORPORATION Dallas, Texas - March 26, 1987 - Dave O'Brien (Jacobs) 38. SOUTHLAND CORPORATION Dallas, Texas - April 7, 1987 - Dave O'Brien (Jacobs) 39. CIRCLE K STORES Phoenix, Arizona - July 17, 1987 - Robin Parker (Jacobs) 40. CONSOLIDATED FREIGHTWAYS Menlo Park, California - August 7, 1987 - Mary Ann Parker (Jacobs) 41. RICK BRODIE Phoenix, Arizona - July 30, 1987 - Tom Schruben (EPA) 42. OWENS CORNING FIBERGLASS Conroe, Texas - July 29, 1987 - Robin Parker (Jacobs) 43. ONTARIO CANADA MINISTRY OF ENVIRONMENT Ontario, Canada - July 28, 1987 - Ramesh Maraj (Jacobs) 44. ARMOR SHIELD COMPANY Cincinnati, Ohio - July 29, 1987 - Tom Schruben (Jacobs) 45. ONTARIO CANADA (John Gerders) - FUEL SAFETY BRANCH Ontario, Canada - July 25, 1987 - Ramesh Maraj (Jacobs) 31 ------- PHONE CALLS (Continued) 46. PROVINCE OF MANITOBA - July 29, 1987 - Tom Schruben (Jacobs) 47. TANK LINERS, INC. - August 4, 1987 - Tom Schruben (Jacobs) 48. ASHLAND OIL Ashland, Kentucky - July 25, 1987 - Ramesh Maraj (Jacobs) 49. VIRGINIA STATE WATER CONTROL BOARD Richmond, Virginia - July 23, 1987 - Kelly Munyon (Jacobs) 50. FLORIDA DEPARTMENT OF ENVIRONMENTAL REGULATION - April 1, 1987 - Tom Schruben (Jacobs) 51. SWEDEN & DENMARK - June 23, 1987 - David O'Brien (EPA) 52. DATA GATHERING ON CAUSES OF RELEASE - May 18, 1987 - A1 Nugent (Hart) STUDIES 53. AMERICAN PETROLEUM INSTITUTE - TANK & February 5, 1987 PIPING LEAK SURVEY 54. AMERICAN PETROLEUM INSTITUTE - PRECISION May 12, 1987 TESTING OF UNDERGROUND STORAGE TANKS OWNER BY MAJOR PETROLEUM COMPANIES: A Look at 5767 Underground Storage Tanks 55. CALIFORNIA STATE WATER RESOURCES CONTROL BOARD 56. DENMARK "DISMANTLING OIL TANKS" 1985 57. HART ASSOCIATES, INC. May 18, 1987 "PERCEPTIONS OF THE CAUSES OF RELEASE" 58. "UST CAUSES OF RELEASES - EUROPEAN REPORT June 1, 1987 FINDINGS" 59. SURVEY OF FIRE SERVICE POSITION REGARDING June 5, 1987 REPAIRS TO UNDERGROUND STORAGE TANK SYSTEMS 60. ICF TANK FAILURE ANALYSIS 32 ------- STUDIES (Continued) 61. MIDWEST RESEARCH INSTITUTE "PROCEEDINGS & RECOMMENDATIONS OF THE EXPERT PANEL ON CORROSION" 62. NEW MEXICO HEALTH & ENVIRONMENT DEPARTMENT 63. SUFFOLK COUNTY DEPARTMENT HEALTH SERVICE "INTERIM REPORT I: TANK CORROSION STUDY" 64. "SERVICE STATION TESTING COMPANY" TANK SYSTEM STATUS (MOSTLY TEXAS) 65. NACE STEERING COMMITTEE MEETING 66. STATE FIELD DATA UST TESTING PROGRAMS - Sammy Ng (USEPA) 67. VERSAR, INC. "SUMMARY OF COUNTY/CITY REPORTS ON RELEASES FROM UST" 68. "ANALYSIS OF DOCUMENTED CAUSES OF SUBTITLE I UNDERGROUND STORAGE TANK RELEASE INCIDENCE" 69. SAN DIEGO, CALIFORNIA, TANK AUDIT, INC. 70. ANALYSIS OF EUROPEAN UNDERGROUND STORAGE TANK PROGRAMS Dames & Moore July 7, 1987 through July 8, 1987 August; 11, 1987 July 31, 1987 July 31, 1987 June 22, 1987 July 21, 1987 February 20, 1987 March 7, 1987 May, 1986 June 15, 1987 33 ------- TABLE I OVERALL SUMMARY OF UST RELEASES RANKING OF MAJOR CAUSE ORGANIZATION PIPE FAILURE TANK FAILURE SPILL/ OVERFILL PEICO Sunnyvale, CA (Trip 1) Sunnyvale, CA (Trip 2) San Jose, CA Broward County, FL Dade County, FL Dade County, FL FILES Ontario Province Southland Corporation Ryder Truck Rental, Inc. Arkansas Best Freight Sys. Yellow Freight System Roadway Express Commercial Journal Carrier Los Angeles City Fire Dept. Fuel Quality Services American Trucking Association Austin Undergrnd Stor. Program Amoco Oil Company Murphy Oil Circle K Stores San Diego, CA Suffolk County Denmark PEI State of Virginia Southland Corporation (FRP) Ashland Oil Consolidated Freightways Boeing 2 2 1 2 2 2 1 1 2 2 1 1 1 1 1 1 1 1 2 1 1 1 2 1 2 1 1 2 1 3 2 2 1 3 3 2 2 3 3 2 3 3 2 2 2 2 2 1 2 2 3 2 1 2 2 1 2 1 1 3 3 1 1 1 1 1 3 2 2 3 3 3 3 1 3 3 2 3 1 Average Totals 1 2 3 1.37 18 11 0 2.14 4 16 8 2.0 10 3 10 34 ------- |