Jump to main content.


53 FR 37082-37247 Friday, Sept. 23, 1988 40 CFR Parts 280 and 281, Underground Storage Tanks; Technical Requirements and State Program Approval; Final Rules--Preamble Section IV. Analysis of Today's Rule--D. Release Detection

PREAMBLE
(37142-37169)

 


IV. Analysis of Today's Rule

D. Release Detection

1. Overview

2. Section-by-Section Analysis


IV. ANALYSIS OF TODAY'S RULE

D. Release Detection

This section of today's preamble provides a summary of the Agency's final approach to release detection, the proposed rule, and the major changes from the proposal. A section-by-section analysis of the final rule (IV.D.2.) discusses in detail the final release detection requirements, including highlights of major public comments received.

1. Overview

a. General Approach to Release Detection. As described earlier in this preamble, today's requirements that new and existing UST systems be properly installed, protected from corrosion, and equipped with spill and overfill protection will dramatically reduce UST system releases. Release detection is an essential backup measure to prevention, particularly for unprotected steel UST systems (prior to upgrading or replacement) and pressurized piping because they are more prone to releases. A variety of release detection methods have been successfully applied to USTs. These methods can be grouped into six general categories: tightness or precision tests, tank gauging systems, inventory control methods, ground-water monitoring, vapor monitoring, and interstitial monitoring. Each was discussed in the preamble to the proposal (52 FR 12714). State and local programs have chosen to rely on different combinations of these methods. They all appear to be successfully detecting releases when properly applied. To maintain flexibility in the selection of release detection methods, both for the implementing agencies and for the owners and operators, the proposal allowed the selection of release detection to be tailored to the characteristics of each site and, therefore, avoided unnecessary disruption of successful state and local programs. The most important features of the proposed and final rules are summarized in the sections below.

b. Highlights of the Proposed Rule. In the proposed rule, the release detection strategy relied on the use of either monthly detection methods or a combination of tightness testing (performed semiannually to every 5 years) and monthly inventory control. Frequent testing dramatically increases the probability of detecting a release and reduces the length of time a release can go undetected. A 30-day frequency was selected as a practical monitoring frequency that was sufficient to protect human health and the environment. The proposed rule required only one release detection method at each UST site, because frequent use of one monitoring method was sufficient to discover releases before they could cause significant damage to the environment.

Current industry practices generally do not include frequent release detection and most releases that are discovered are detected through impacts on the surrounding community or large inventory losses. Consequently, substantial time and effort will be required to reach the goal of monthly monitoring for all UST systems. The proposal phased in detection requirements over 5 years to allow the leak detection industry time to expand and to more evenly schedule the demand for detection equipment. The proposed rule also allowed less frequent use of tank testing (every 3 or 5 years when combined with monthly inventory controls) during the first 10 years of the program. The goal of the proposed approach was the installation of release detection as quickly as possible on the tanks most likely to leak.

To ensure flexibility, all proven methods of release detection were allowed in the proposed rule. In the absence of adequate data, the proposed rule did not set one performance standard for all release detection methods. Instead, each method was required to meet performance and design standards specific to that method. These standards were based on the experience of state programs that indicated these methods were effective under the specified conditions.

c. Major Changes in the Final Rule. Although the overall release detection strategy has not changed from the proposal, specific requirements on how and when release detection must be conducted have changed. The four most significant revisions to the proposed regulations include:

o More frequent monitoring of existing unprotected tanks during the 10-year upgrade period;

o Less frequent monitoring of new and upgraded tanks until age 10;

o Gradual phase-in of release detection based on tank age; and

o More stringent requirements for pressurized piping.

These changes, discussed generally in section III.B. of today's preamble, are presented in detail in the appropriate parts of the section-by-section analysis below. Additional revisions made to the release detection requirements are also discussed in the section-by-section analysis.

In addition, three important organizational changes were made in response to commenters' concerns. First, in the proposed rule, the requirements for hazardous substance USTs were in a subsection of the petroleum UST requirements, and several commenters noted that they had difficulty finding and understanding the requirements for these systems. In response to this concern, the release detection requirements for petroleum and hazardous substance USTs have been separated in the final rule and placed in §§ 280.41 and 280.42, respectively.

Second, tanks and associated piping were treated in the proposed rule as a single unit. Each method of release detection applied to the tank was required also to detect leaks from piping. In another section of the proposed rule, there were additional release detection requirements that could only be applied to the piping. Commenters noted that this structure limited flexibility in meeting the release detection requirements by forcing the same method to be used for tanks and piping. They noted that the same method may not work for both tanks and piping and several viable detection methods for piping were excluded. In addition, the information on causes of release indicates that piping is generally a greater release threat than tanks. Thus, today's final rule treats piping separately from and with equal importance to the tank. The release detection methods for tanks and piping have been separated in the final rule into §§ 280.43 and 280.44, respectively.

Third, in the final rule, the release detection requirements for each type of UST system, including allowed methods and required frequencies of testing, have been consolidated into two brief sections (§§ 280.41 and 280.42). The detailed performance standards for each method of detection are now contained in §§ 280.43 and 280.44. The section-by-section analysis of the preamble parallels this structure so that all discussion of the phase-in schedule, the methods and combinations of methods allowed for each type of system, and frequencies of testing are discussed first (sections IV.D.2.a-c.). Discussion of research and public comments on the technical details of each detection method for tanks and piping is reserved for later sections (IV.D.2.d-e.).

2. Section-by-Section Analysis

a. General Requirements (§ 280.40). (1) Use of One Release Detection Method. In the proposed rule, a single release detection method could be used to meet the requirement to detect releases from both the tank and connected piping. As discussed in more detail in the preamble to the proposed rule (52 FR 12718-12719), the use of redundant methods of release detection was not required at each UST site because the Agency was not convinced that the required use of these "backup" methods would provide significant environmental gains in comparison to the adverse impacts on program implementation. Some commenters opposed allowing only one method of release detection primarily because they believe all the methods are unreliable and insufficiently developed, particularly external methods. Other commenters, however, agreed with EPA's position on this issue and cited their own satisfactory experience with the various methods.

The final rule continues to allow the use of a single properly installed and operated release detection method for tanks when testing is performed monthly. When less frequent monitoring is used it must be backed up by use of monthly inventory control. Owners and operators remain free to use multiple methods if they desire, and state and local programs can require redundant systems.

EPA decided against requiring multiple methods because frequent use of a single detection method, when combined with the prevention measures contained in other sections of the rule, is sufficient to protect human health and the environment. The performance standards, design criteria, and limitations on the methods contained in the rule are intended to ensure that the optimum performance of each release detection method is achieved. Repeating the test monthly dramatically reduces the possibility of failing to detect a leak. Each test serves as a separate check of the integrity of the UST system. Field reports confirm the success of single methods in detecting releases from UST systems. For example, Dade County, Florida, has detected over 350 releases using ground-water monitoring wells. EPA's research on the best ways to use some of the different release detection methods is directed towards improving the field performance of various types of methods. For tanks and suction piping systems, one detection method, combined with prevention efforts, should virtually eliminate undetected releases.

The Agency chose not to rely on one method of detection for pressurized piping, however. Even with good efforts at prevention, these systems may still result in significant releases. Consequently, the final rule requires existing and new pressurized lines to use both automatic line leak detectors and another leak detection method (either monthly monitoring or annual line tightness tests).

(2) Scope of Release Detection (§ 280.40(a)(1)). The proposed rule provided a general requirement that the release detection method be "capable of detecting a release from any portion of the UST system." The purpose of this requirement was to ensure detection of both tank and piping leaks. A few commenters objected to the general nature of the wording of the requirement because it includes some portions of the UST system such as vent lines, fill pipes, and bungs on the top of the tank that do not normally contain regulated substances. Some tank tightness test methods do not test the top of the tank and, thus, do not detect the presence of holes in the vents and bungs. These portions of the tank only leak when the tank is overfilled.

EPA shares the concern of commenters that a strict interpretation of the wording in the proposed requirement could result in some release detection methods, particularly non-overfill tightness tests, not being allowed because they cannot detect releases from portions of the UST system that do not normally leak. The final rule's wording that the methods must be able to detect a release from "any portion of the tank and the connected underground piping that routinely contains product" is intended to make clear that tank tightness test methods that do not overfill the tank can be used, as long as they meet the other applicable performance standards and another acceptable method is used to test or monitor the piping.

Furthermore, releases from the top of the tank or vents occur during overfills, which are not a normal operating condition. Prevention of overfills is addressed in § 280.20(c), § 280.21 (d), and § 280.30, which together require that all new and upgraded tanks have overfill prevention equipment and spill catchment devices and that proper filling procedures be followed to prevent these "nonoperational" releases. These requirements are intended to prevent these types of releases; thus, EPA does not believe methods of release detection must be used that will detect them.

The additional information on causes of release that EPA has collected since the proposal (see section II.F. of this preamble) reinforces the fact that piping is a major source of releases. Therefore, the final rule continues to require that the methods of release detection that are used must be capable (either singly or in combination) of detecting a release from both the tank vessel and the piping that conveys product.

(3) Installation, Operation, Calibration, and Maintenance (§ 280.40(a)(2)). To ensure that the release detection method will reliably detect releases once in place, the proposed rule required installation, calibration, operation, and maintenance according to manufacturers' specifications. At proposal, the Agency decided against requiring certification of installers and servicers of release detection equipment because these programs are not currently developed and there are other effective approaches for ensuring proper installation (52 FR 12719). As discussed below, commenters stated that the rule should require certification of installers and servicers of release detection equipment. After consideration of these comments, EPA continues to believe that such a requirement would hinder rapid installation and flexibility in designing effective ways to regulate installers. Today's final rule will remain as proposed for the reasons discussed below.

The UST release detection field is a new and proliferating area of technology, and, because of this, some commenters suggested that a certification program should be required for installers and operators of release detection equipment to ensure high quality work and to assist owners and operators in selecting qualified personnel. Certification by EPA or state agencies was suggested. Other commenters felt that a state or federal certification program would limit the number of installers, would increase the cost of release detection provided by those installers who were already certified, and would delay widespread application of release detection methods. Some of these commenters suggested alternatives to state or federal installer certification.

By including the performance standard in the proposed rule, EPA acknowledged that ensuring proper installation and operation of equipment is important. Although a certification program is a viable approach to achieving this goal, EPA disagrees with commenters who felt that state or federal certification is the only way to ensure quality installations. Existing state programs that lack certification programs have been effective at discovering releases. As commenters noted, there are numerous possible approaches to ensure proper installation. Mandating certification would unnecessarily restrict states from designing alternative effective ways to regulate installers.

EPA agrees with commenters that the time required to conduct a certification program would seriously hinder rapid implementation of release detection. Installing leak detection quickly on existing tanks, which are primarily unprotected steel, will be of significant environmental benefit, even in the absence of certification. For these reasons, no additional requirements have been included in the final rule to ensure release detection installation and operation.

(4) Meeting the Performance Standards (§ 280.40(a)(3)). In the proposal's preamble (52 FR 12714-12718), the Agency described three possible approaches to ensure the quality of release detection equipment used to meet the regulations. EPA solicited comments on a general performance standard, certification of methods, and a method-specific performance standard. The Agency proposed the method-specific approach because it offered the greatest flexibility and facilitated rapid program implementation. Commenters generally agreed with the advantages and disadvantages ascribed to each approach, and many concluded that the method-specific approach is the best possible at present. The final rule thus retains method-specific standards by requiring each method used meet the standards in § 280.43 or § 280.44.

Commenters generally agreed with both the explanation and the conclusion in the proposal preamble concerning the use of method-specific release detection standards. Many commenters believe that the method-specific approach would be the most realistic approach given our current knowledge and that it would allow the widest range of choices among effective technologies. Some commenters, however, believed a general standard should be formulated based on the standard specified for tank tightness testing (0.1 gallon per hour). Because the method-specific standard would allow varying performance standards, these commenters felt that many owners would simply select the cheapest, least effective method. Other commenters recommended a certification approach, feeling that consistency was less important than ensuring that the regulated community knew exactly what devices met the standards.

The preamble to the proposal contained an extensive discussion of the merits of each of the three approaches to regulating release detection. One approach considered by the Agency was to specify a general standard for the leak rate or quantity which must be detected by a method and not specify individual methods or restrictions on their use. This approach was viewed as providing the most consistent level of performance and the one that best challenged manufacturers to develop defensible performance claims for their equipment. This approach is not used in the final rule for two reasons. First, the Agency does not have sufficient information to relate leak rates to the quantity of product detected by external methods under all possible site conditions. Second, eliminating the specific methods in the rule would slow program implementation by forcing owners and operators to wait for detailed, extensive performance information before conducting release detection. The final rule, however, incorporates the flexibility of a general performance standard by allowing, in addition to specific methods, any method which can detect a 0.2 gallon per hour leak rate with a probability of detection of 0.95 and a probability of false alarm of 0.05 within a month. For a discussion of this addition, see section IV.D.2.a.4. of today's preamble.

A certification approach applied at the federal level would provide the regulated community with the clearest direction concerning which release detection equipment was acceptable. This approach is not used in the final rule because it would slow program implementation and reduce the choices available to owners and operators over the next several years. Further, it was not viewed as necessary because comparable performance information for each method can be generated by private efforts without federal involvement.

Research results and data submitted by commenters after the proposal have reinforced the Agency's belief that all of the methods that were proposed are effective release detection techniques if used within the context of certain constraints (discussed in the section on individual methods). EPA believes that offering a broad selection of methods will make it easier for owners and operators to comply with the regulation. Also, a broad selection is consistent with the encouragement of existing industry trends and state programs, which have utilized a variety of release technologies and which have been proven effective at discovering leaks. Information gathered by EPA suggests, but does not conclusively prove, that all of the methods included in the rule can detect at least a 0.2 gallon per hour release within 30 days when used in accordance with the restrictions on that method. The Agency believes that all methods will eventually be able to prove they reliably detect 0.2 gallon per hour releases and has included that as a standard for approving new methods. This standard is discussed in section IV.D.2.a.4. Allowing a range of methods with specific standards does not mean the cheapest, least effective method will be selected, because cost and effectiveness are not necessarily related. For instance, under some site conditions (e.g., ground water within the excavation zone), manual ground-water sampling may be the cheapest method and will reliably detect releases substantially smaller than 0.2 gallon per hour.

(5) Delay in Detection Probabilities (§ 280.40(a)(3)). As discussed in the proposal preamble (52 FR 12719), a complete release detection performance standard includes not only the leak rate or quantity that a method must detect, but also the probabilities of detection (PD) and false alarm (PFA). In the proposed regulations, complete standards of this type were included for in-tank detection methods in the section on specific methods. Research results and commenters' concerns have caused the Agency to make several important changes in the final rule. The probabilities have been moved to the general requirements section (§ 280.40(a)(3)) and changed slightly, and the effective date has been delayed for two years. These changes are discussed below.

The proposal preamble emphasized the statistical nature of detecting leaks and the large number of variables that add to the uncertainty in declaring a leak. The probability of detecting a leak is dependent on its size. All methods are more likely to discover large leaks than small ones. With regard to detecting the smallest leaks, the Agency recognized that good methods properly operated may mistakenly declare a leak when none exists (false alarm) or fail to discover some leaks (missed detection). Consequently, the standards for tank tightness testing and automatic in-tank monitoring included the requirement that methods detect a specified leak rate with a PD of 0.99 and a PFA of 0.01. The Agency remains convinced that specifying probabilities in this manner better defines the performance standards and should help owners and operators make informed choices about leak detection.

The final rule retains the PD and PFA as part of the leak detection performance standard. The probabilities have been removed from the standards for specific methods and placed in the general requirements section. This change expands the coverage of the probabilities to include automatic line leak detectors and interstitial monitors as well as tank and line tightness tests and automatic in-tank monitors. The Agency made this change to clarify that all leak rates or quantities specified as part of a method-specific standard in § 280.43 or § 280.44 must be detected with the same level of reliability.

The proposed rule set the PD at 0.99 and the PFA at 0.01. In the final rule, the PD has been changed to 0.95, and the PFA has been changed to 0.05. The Agency made this change for several reasons.

First, the Agency is not convinced that the 0.99/0.01 specification was a realistic standard given the wide range of variables affecting leak detection results. EPA's study of tank testing methods has attempted to determine performance at this level of precision and only 2 methods of the 25 studied could meet the standard. No study of this sort has been undertaken for the other methods covered by the probabilities in the final rule. Further, EPA does not plan to conduct studies similar to the one for tank tightness testing for other leak detection methods. Rather, the Agency intends that manufacturers should evaluate their methods to prove they meet the standard in the rule. Thus, the change in the final rule will encourage manufacturers to undertake this research. EPA is developing procedures for testing release detection equipment in a common way to help manufacturers in evaluating their equipment. EPA chose the 0.95/0.05 specification in the final rule because it is a level of performance attained by a modified commercial tank test method in the National Motor Fuel Survey. In addition, several commenters felt that a probability of detection of 0.95 was more realistic and was adequate to protect human health and the environment.

Because the final rule requires frequent-to-continuous monitoring, the change in probabilities will have little environmental impact. For example, a test which detects 0.1 gallon per hour leaks 95 percent of the time in one test will discover 99.9 percent of 0.1 gallon per hour leaks in 3 consecutive tests. Further, tests that meet this standard are virtually certain to detect leaks larger than this threshold and will detect a significant number of leaks below the threshold. Under either standard, large leaks will be caught immediately and small leaks will be discovered before they cause environmental damage.

The final change in this section of the final rule is a delay in the effective date of the probabilities. As discussed above, few methods have been proven to meet the complete standard specified in the proposed rule. However, preliminary results from the EPA's tank testing study (Notice of Availability; 53 FR 10403) indicate that several methods could meet the standard with simple changes in procedures and equipment. The Agency also wanted to allow time for manufacturers of automatic tank gauging systems, automatic line leak detectors, and interstitial monitors to prove that their systems meet the complete standard. The Agency believes manufacturers can make the necessary changes to their methods and evaluate their performance in 2 years. Until the probabilities become effective, methods need only detect the leak rate or quantity specified for that method in § 280.43 and § 280.44. Methods installed or conducted during this 2-year phase-in will not need to be performed again or replaced after the probabilities become effective, but all methods used after that period of time will have to achieve the probability standard.

(6) Reporting of Positive Monitoring Results (§ 280.40(b)). The proposed rule contained a provision in Subpart E that the owner and operator must report all suspected releases indicated by the results of release detection monitoring. Perhaps because this release detection reporting requirement was in a different subpart from the release detection technical requirements, commenters stated that it was unclear when a suspected release needed to be reported. To clarify and strengthen the requirement to report all suspected releases, a provision has been added to Subpart D (§ 280.40(b)) in the final rule explicitly stating that any indication by the release detection method that a release has occurred must be reported in accordance with reporting procedures described in Subpart E.

It is intended that all release detection equipment be operated at least at the level of sensitivity indicated in the performance standard. For example, 2 years after the effective date of the rule, the tightness test threshold (i.e., test result that indicates a suspected release) should be set to detect 0.1 gallon per hour leak rates with a PD of 95 percent and a PFA of 5 percent; this threshold value may differ for various tightness testing devices. Manufacturers of the release detection equipment must determine what this threshold value must be to meet the performance standard and inform operators of their equipment. Until the manufacturer sets such a threshold, tank test operators should continue to use the current 0.05 gallon per hour threshold. For most tank testing methods, this will approximate the threshold level for detecting 0.1 gallon per hour leaks. The owner and operator would report a suspected release when a test result exceeds 0.05 gallons per hour or the threshold value provided by the manufacturer in accordance with § 280.50.

It is important to note that the performance standards such as 0.2 gallons per hour or 1/8 inch of product on top of the ground water are device performance standards set to exclude less effective equipment. The standards are not allowable contamination levels. Owners and operators are still responsible for correcting leaks and cleaning up any product released to the environment. It is in their interest to use the most effective release detection equipment and operate it so the device detects releases as quickly as possible to avoid potentially costly corrective action.

(7) Phase-in of Release Detection (§ 280.40(c)). The proposed rule required a 3- or 5-year phase-in of release detection, with the shorter phase-in period applied to USTs without protection from corrosion and the longer phase-in to those USTs with corrosion protection. These phase-in periods were based, in part, on the experiences of several state and local UST programs in initiating release detection under a phase-in schedule that was typically 3 to 5 years (52 FR 12677, 12703-12704). This phase-in was proposed to ensure that tanks with the greater risk of leaking (those unprotected from corrosion) had release detection installed first. As discussed in the proposal preamble, the total phase-in period covers 5 years to allow enough time for the release detection industry to respond to the demand, owners and operators of existing tanks to plan their needs, and implementing agencies to develop their programs. The proposed rule also required closure of existing USTs that could not meet the phase-in schedule.

Commenters recommended a variety of phase-in periods and generally recommended age as the most appropriate basis for the phase-in. Therefore, § 280.40(c) of the final rule phases in the implementation of release detection over 1 to 5 years based on the age of the system (oldest tanks first). Requiring the oldest tanks to phase in release detection sooner, within 1 year, ensures that those UST systems most likely to leak are addressed first. The final rule also retains the proposed requirement of closure of any USTs that cannot meet the release detection requirements by the phase-in date. A significant addition to the final rule is the requirement that existing systems with pressurized piping must retrofit line leak detectors within 2 years. Comments received regarding the phase-in of release detection are discussed in more detail below.

Commenters recommended a range of time periods from 3 to 10 years for completing the phase-in of release detection. Commenters supporting a longer period believed that the proposed phase-in periods would overburden the release detection industry, resulting in poor quality installations and late compliance, and cause economic hardship to owners and operators. Those commenters recommending shorter phase-in periods believed that a tighter schedule would prevent significant environmental damage. The Agency has decided in the final rule to retain the overall 5-year phase-in time period for the same reasons outlined in the proposal preamble (52 FR 12677, 12703, and 12704). Based on experience at the state and local level, EPA does not believe release detection can be installed and conducted at over 700,000 UST sites nationwide in less than 5 years. Not only will it be difficult to do in less than 5 years, but some of the release detection systems installed on a more expeditious timetable could be lower quality as a result. As discussed above, however, the greatest release potential will be during this interim period. Thus, lengthening the phase-in period would result in unacceptably greater environmental damage. Moreover, even accepting commenters' concerns about economic burdens, lengthening the phase-in would not provide a resolution because retrofitting would still be necessary ultimately and would be coupled with greater corrective action costs brought on by the delay in detection. For these reasons, today's final rule retains the 5-year phase-in period.

During the 5-year phase-in period, it is important to direct release detection efforts at the existing UST systems most likely to leak. Many commenters recommended achieving this goal by phasing in release detection based on the age of the UST system. They pointed out that this approach also has the advantage of stabilizing the demand for release detection, resulting in less burden for both the release detection industry and the owners and operators. It also has the advantage of addressing first the tanks that are most likely to leak. Several commenters pointed out their concerns that if the phase-in is not sequenced in its implementation, the regulated community will collectively wait until the last minute, and unavoidable further delays will ensue. A few commenters opposed an age-based schedule because it was too simplistic or ignored other important factors.

The Agency agrees with commenters suggesting a release detection phase-in schedule based on age. Although age is not the only factor in determining when a tank will leak, it is an important factor that is readily understood and determined and, therefore, easy to implement. This approach is already being used successfully in several state and local programs. The approach in the final rule is not a radical departure from the proposed phase-in schedule. The causes-of-release study indicates that most existing protected USTs are less than 10 years old. Under both the proposed and final phase-in schedules, these tanks must phase in release detection within 5 years after promulgation. EPA believes that the main impact of the revised schedule will be to spread out the phase-in of release detection on unprotected USTs, which represent over 75 percent of UST systems, over years 1 through 4 rather than require it all at year 3, resulting in fewer implementation bottlenecks.

Some commenters recommended a class approach (sensitive areas first) to phasing in release detection. Although the concept of retrofitting release detection in vulnerable areas first is appealing, EPA does not believe it is possible to identify sensitive classes in any meaningful way at the federal level (see section III.C. of this preamble). States can choose to phase in release detection based on a class approach without losing the ability to receive state program approval (see § 281.33) if they complete a phase-in of release detection at all existing tanks within 5 years and pressurized piping in 2 years.

EPA's information on the causes of release clearly indicates that pressurized piping represents a major source of uncontrolled releases. None of the requirements for existing systems in the proposed rule addressed the threat of catastrophic releases from pressurized piping. Consequently, in the supplemental notice (52 FR 48638), the Agency requested comment on the idea of requiring existing systems to retrofit line leak detectors on pressurized piping. Commenters generally felt that it was appropriate to require such a retrofit and recommended a variety of phase-in schedules. EPA agrees with commenters who recommended a short phase-in schedule because this piping is a significant environmental hazard, retrofitting line leak detectors is relatively easy and inexpensive, the devices are highly effective (see section IV.D.2.e.1. of the preamble), and many systems are already equipped with the devices. Consequently, the final rule requires that existing pressurized piping meet the same standards as new piping 2 years after the effective date of the rule (see section IV.D.2.b.2. of the preamble for piping requirements).

(8) Closure if Release Detection Is Not Installed (280.40(d)). In the proposed rule, EPA required closure of an existing UST system if a method of release detection was not installed by the end of the specified phase-in period. Most existing USTs are not protected from corrosion and, thus, are likely to corrode and eventually leak. The selected phase-in schedule discussed above is considered the maximum time that these systems should be allowed to operate without release detection. Therefore, the final rule continues to require that UST systems be closed if release detection cannot be retrofitted or applied by the phase-in date.

One commenter noted that the closure procedures of the proposed rule required a site assessment of the excavation zone before closure, the results of which might delay closure beyond the allowable time frame. Although the Agency recognizes that closure can be a time-consuming process, it should not require any more time than the selection and installation of release detection equipment. The final rule requires that, by the phase-in date, the owner must remove the tank or fill it with inert material and complete the site assessment. Should a release be discovered, responding to the findings of the site assessment is part of corrective action and need not be completed by the phase-in deadline. Owners and operators are expected to plan ahead to ensure that they complete installation of release detection or the closure procedures by the specified date. This will allow the implementing agency to ensure compliance with both requirements with a single inspection. For these reasons, EPA has retained in the final rule the provision to complete closure by the end of the phase-in period.

(9) Other Changes. One of the general requirements in the proposed rule required a site assessment prior to the installation of any external leak detection system to ensure compliance with the performance standards for the particular method used. To clarify that the site assessment is intended to include only an analysis of selected factors within or beneath the excavation zone, the general requirement has been deleted, and the only assessment requirements are contained in §§ 280.43(e)(6) and (f)(7) of the final rule. These changes are discussed below.

The importance of a site assessment in correctly selecting and applying an external method was discussed in the proposal preamble (52 FR 12720-12722). Although numerous factors were listed in the proposal preamble concerning a site assessment, EPA stressed that the assessment should, at a minimum, ensure compliance with the method-specific restrictions in the proposed rule. EPA requested comment on the proposal to include these or other site variables in the assessment requirement. Some commenters stated their belief that a complete site assessment is too extensive a technique to be required for demonstrating the performance of external release detection and does not provide much useful information because conditions at the site change constantly. Most commenters, however, agreed that a site assessment is appropriate before installation of external release detection systems. In fact, these commenters wanted to extend this provision by requiring site assessments for all release detection methods, thereby requiring a quick national survey of all UST site conditions. Others suggested at least requiring a site assessment periodically at all USTs.

The Agency continues to believe that site assessment of the excavation zone is necessary to ensure the reliability of external methods. The Agency also agrees with comments stating that the site assessment requires no more information beyond checking a site for compliance with the restrictions on the methods. Conditions in and below the excavation zone must be known before an external method is selected or installed because inappropriate excavation conditions can render some external methods ineffective. Internal methods are not typically affected by site conditions, and those methods that are affected (e.g., water table level can affect tightness tests) can account for these conditions without performing a site assessment. The major factors determining the effectiveness of ground-water and vapor monitoring were included in the method-specific performance requirements in the proposed rule, and, for most sites, an adequate assessment will require evaluation of only those factors.

The Agency decided against requiring a more extensive, more frequent, or a more widely applied site assessment because of the unnecessary burden it would place on implementing agencies and the possible delays in release detection compliance it would cause. The Agency believes that the greatest benefit for existing systems, short of upgrade or replacement, will be obtained by conducting release detection as quickly as possible. The site assessment for certain external methods is retained in the final rule in §§ 280.43(e)(6) and (f)(7) because EPA believes it is important to ensure that these methods work properly. The site assessment is not intended to be a general search for contamination at the site. Any contamination found, however, must be reported, and the owner and operator must comply with the corrective action requirements of Subpart F. A more detailed search for contamination is required when tanks close under Subpart G.

b. Requirements for Petroleum UST Systems (§ 280.41). (1) Requirements for Petroleum Tanks (§ 280.41(a)) -- (a) Overview. The proposed rule offered a variety of release detection methods for petroleum UST systems. New or existing UST systems could perform accurate monthly monitoring using automatic tank gauges, vapor monitors, ground-water monitors, interstitial monitors, or other methods approved by the implementing agency. The proposal allowed two exceptions to the monthly monitoring requirement. First, when combined with monthly inventory control, tank tightness tests could be performed semiannually at new UST systems. The proposal allowed semiannual tightness testing and inventory reconciliation for new tanks because the combination was believed to be as accurate as monthly monitoring. Second, when combined with monthly inventory control, tank tightness tests could be performed less frequently at existing USTs during the 10-year upgrade period (every 3 years for bare steel systems or every 5 years for protected tanks). Less infrequent tank tests for existing USTs were permitted during the phase-in period because the Agency believed the release detection industry lacked the capacity to perform monthly monitoring at all existing tanks in 3 to 5 years.

The use of monthly monitoring methods has been retained as an option for all petroleum UST systems in the final rule. The final rule also contains two exceptions for tightness testing similar to those in the proposal. An overview of the release detection requirements is presented in Figure 4. During the 10-year upgrade period at existing tanks that are not adequately protected from corrosion and lack spill and overfill equipment, the rule now requires either (1) annual tank tests and monthly inventory controls or (2) monthly monitoring. Tanks that meet the standards for new or upgraded tanks are required either (1) to conduct tank tests every 5 years combined with monthly inventory controls for a 10-year period following the date of installation or upgrade or until 1998, whichever is later, or (2) to conduct monthly monitoring. Also, in both cases, by the end of the 10-year period, these USTs must be using an approved monthly monitoring method.

BILLING CODE 6560-50-M

(b) Monthly monitoring. In the proposal preamble, EPA discussed a variety of possible release detection strategies ranging from continuous monitoring to an infrequent "check" of the tank system (52 FR 12676-12677). "Frequent-to-continuous" detection methods were proposed by the Agency because more frequent sampling improves the chances of finding leaks and limits the length of time over which leaks can progress unchecked. As a result, the proposal required tanks to be monitored at least monthly unless the owner or operator chose an option that included less frequent tank tightness testing in combination with monthly inventory control. (Monthly inventory control does not by itself meet the requirements for "monthly" monitoring and must be combined with periodic tightness testing.)

Commenters were divided over whether monthly monitoring provides adequate environmental protection and whether it was unduly burdensome. EPA continues to believe that the monthly monitoring frequency offers effective environmental protection. Moreover, checking the release detection equipment once a month is not difficult or expensive, according to several commenters and Agency research. More frequent monitoring would necessitate the use of continuous monitors, which are not needed at all sites and which may be less effective at some sites. Thus, the final rule (§ 280.41(a)) requires monthly monitoring as a baseline for all new and existing petroleum UST systems.

(c) Tank Tightness Testing and Inventory Control (§ 280.41(a)(2)). In the preamble to the proposal, EPA noted practical problems with conducting monthly monitoring at all existing tanks over the proposed phase-in period. These methods have not been practiced on a mass scale in the past, and the Agency expressed the concern that the industry did not have sufficient capacity to capably install monthly monitoring at 1.4 million USTs in 5 years. Because tank tightness testing and inventory control are commonly used as effective release detection methods not requiring the installation of permanent equipment, EPA allowed these methods as options for existing USTs during the 10-year phase-in period for upgrading (see section IV.D.2.b.1).

The large number of existing systems to be tested and the limited industry capacity caused the Agency to propose less frequent tank testing for existing systems than for new systems: every 3 years at unprotected existing systems and every 5 years at protected existing systems. The proposed frequencies were selected in recognition of the differing probability of releases at unprotected and protected UST systems. For new tanks, the proposal allowed monthly inventory control combined with semiannual tightness testing because the combination was believed to be as effective as monthly monitoring using the other approved methods.

Many commenters were concerned about the use of tank tightness testing and inventory control as an alternative to the monthly monitoring requirements. Commenters particularly questioned the appropriateness of more frequent monitoring being proposed for new tanks than for existing unprotected tanks. Although there was a variety of opinions on what the proper frequencies were, commenters uniformly felt that new tanks were less likely to leak and should be monitored less often than existing tanks. The new information presented in the "Causes of Release Study" corroborates these concerns that the most serious environmental threat is posed by older, unprotected steel tanks (see section II. F. of the preamble). Available evidence demonstrates that new or upgraded tanks are extremely unlikely to leak over their normal operational lifetimes especially within 10 years of installation or upgrade. EPA solicited comments on this issue and this new information in the supplemental notice published December 23, 1987 (52 FR 48641-48642). Most commenters agreed with the Agency's conclusions on these matters, which are reflected in the final rule.

The Agency continues to believe that monthly monitoring cannot be installed on all UST systems within 5 years and that allowing tightness testing combined with inventory control will ensure that release detection can be provided to all existing USTs as soon as possible. During the phase-in period, EPA believes that priority should be given to requiring application of available release detection resources to older bare steel systems. Accordingly, in § 280.41(a)(2), the final rule requires that if owners and operators of existing unprotected UST systems choose tightness testing, it must be performed yearly rather than every 3 years as proposed. Existing protected systems (with spill and overfill prevention equipment) are required to be tested every 5 years during the 10-year upgrading period, the same as proposed. Because extremely few new or upgraded tanks are expected to leak during the first 10 years of their operational life, under the final rule (§ 280.41(a)(1)), these tanks may also conduct tightness testing every 5 years. This approach has the advantage of encouraging upgrade or replacement of unprotected tanks before the end of the phase-in period, resulting in improved environmental protection. At the end of the 10-year upgrading period or at the end of the 10-year operational life of new or upgraded systems, these tanks must be equipped with a monthly monitoring method.

The proposed rule required tank tightness testing to be combined with inventory control (or another method of equivalent performance) for several reasons. Frequent tank tightness testing is not practical because it requires extensive preparation, including a shutdown of operations. It is, however, a sensitive method that provides very accurate results. Manual inventory control is less sensitive but can provide nearly continuous (daily) release detection that can reliably detect larger releases. The rule proposed the combination of the two techniques to compensate for each component's disadvantages. Several commenters on the proposed rule viewed the combination of techniques as redundant and stated that each method is adequate on its own. Other commenters agreed that, separately, each of these techniques would be an inadequate release detection method.

The Agency evaluated different approaches to tank tightness testing and inventory control since the proposal (Notice of Availability; 53 FR 10403). The results of the studies, which are discussed in more detail in the preamble section on these methods, confirm that monthly inventory control is effective at reliably detecting larger leaks (about 1 gallon per hour) and that tank testing can reliably detect much smaller leaks (0.1 gallon per hour). This research and information submitted by commenters convinced EPA to retain the combination of infrequent tank testing and monthly inventory control in the final rule as an exception to monthly monitoring in certain situations.

The proposed semiannual tightness testing and inventory control for new USTs has been deleted from the final rule, because the Agency does not believe that the combination is as effective as the other monitoring methods. The Agency continues to believe that monthly monitoring is necessary to protect human health and the environment; less frequent monitoring is allowed only as an interim measure. Currently, conducting monthly tank tightness testing is not a practical or economical method. Tank testing methods may be developed in the future, however, that can be performed on a monthly basis to detect leaks of 0.2 gallon per hour. The final rule allows the use of this method without inventory control once the method is proven to meet the performance standard in the section on other methods (§ 280.43(h)).

(d) Manual Tank Gauging (§ 280.41(a)(3)). In addition to the other release detection methods in the proposed rule, the final rule also includes manual tank gauging. The Agency requested comment on the use of this method in the supplement to the proposed rule (52 FR 48641), citing a study submitted by a commenter on the proposal showing that this method was effective for used oil tanks. EPA conducted an analysis of this study (Notice of Availability; 53 FR 10403) and found that weekly tank gauging can detect 0.2 gallon per hour leaks with a PD of 95 percent and a PFA of 5 percent for tanks smaller than 550 gallons.

Because it provides the same level of protection as other monthly monitoring methods, the final rule allows use of this method for any tank with nominal capacity of 550 gallons or less. Detailed discussion of research and comments on the inclusion of this method is provided in Section IV.D.2.d.(2). of today's preamble.

(2) Requirements for Petroleum Piping (§ 280.41(b)) -- (a) Overview. The proposed regulation required that each release detection method chosen for the tank also detect releases from the piping. In addition, new pressurized piping was required to have equipment capable of detecting and shutting off a release of at least 2 gallons per hour unless the piping had continuous or interstitial monitoring. Suction piping that meets certain minimum design specifications was exempt from these release detection requirements.

Commenters agreed that pressurized piping was allowable but that additional release detection requirements were necessary. Some commenters had reservations about automatic shutoff devices and flow restrictors and recommended backup release detection or double-walled piping. New information acquired by the Agency since proposal on causes of release indicates that pressurized piping, along with spills and overfills, is the major source of releases, particularly large-volume catastrophic releases (see section II.F. of this preamble). Because of this new information, the Agency requested additional comment on release detection issues related to underground piping in the supplemental Federal Register notice published December 23, 1987 (52 FR 48641-48642).

The final rule has been revised to reflect the importance of preventing and rapidly detecting piping releases by including additional release detection requirements for pressurized piping and further encouraging the use of suction systems. Figure 5 summarizes the requirements for petroleum piping. Pressurized piping must have a release detection device that monitors the line at least hourly and automatically shuts off or restricts product flow or sounds an alarm when there is an indication of a leak. The owner and operator must also conduct either monthly monitoring or an annual line tightness test. The monthly monitoring may include vapor monitoring, ground-water monitoring, interstitial monitoring, or other methods that meet the performance standard or are approved by the implementing agency. The performance standards for the piping release detection methods are contained in a separate section of the rule (§ 280.44) and are discussed in section IV.D.2.e. below.

BILLING CODE 6560-50-M

The Agency notes that one release detection method can be used as the sole method if it can meet both the hourly release detection requirement and the annual or monthly release detection requirements. For example, double-walled piping with continuous interstitial monitoring that meets the performance standard continues to be an acceptable option for pressurized piping and would not require shutoffs, restrictors, or tightness tests. The system must be equipped, however, with an alarm that will indicate when a release into the interstitial space has begun.

Suction piping systems meeting the "no leak" criteria described below in subsection (c) continue to be exempt from release detection. Other suction systems must operate with monthly release detection or a line tightness test every 3 years.

(b) Requirements for Pressurized Piping (§ 280.41(b)(1)). As discussed in the proposal preamble (52 FR 12743-12745), the Agency was concerned that pressurized piping, which uses a pump in the tank to force product through the line to the dispenser, could result in large volumes of product being quickly released to the environment through a hole or crack. In contrast, suction piping appears to be intrinsically much safer because product is transferred at less than atmospheric pressure by a pump near the dispenser drawing product from the tank by suction, and failures will result in air or ground water flowing into the pipe rather than product being released during operation. The Agency considered not allowing pressurized piping at new installations but its widespread use, the availability of release detection technology (see discussion below), and some of the operational limitations of suction piping persuaded EPA to include it in the proposed rule.

Commenters agreed with EPA's position that pressurized piping could result in large-volume releases in the absence of release detection but felt that pressurized piping with release detection should be allowed in the final rule. They noted that pressurized piping is widely used, economical, efficient, dependable, and not susceptible to vapor lock. Commenters also noted that suction piping is more expensive to operate and requires longer times to dispense product and that its use is restricted by elevation above sea level, height differences between storage and delivery, flow rate, temperature, and length of horizontal piping. The Agency has continued to allow the use of both suction and pressurized piping in the final rule subject to the requirements discussed below.

o Immediate Detection of Large Leaks (§ 280.41(b)(1)(i))

At proposal, data received from state agencies indicated that piping was involved in 20 to 35 percent of all releases. Pressurized piping was also identified as the most common petroleum dispensing system at new installations. Documented cases raised the possibility of sudden large releases from these systems. These factors led the Agency to conclude that additional release detection was required for pressurized piping so that large-volume releases could be stopped as quickly as possible. The monthly monitoring frequency required for the tank was considered inadequate given the potential environmental damage due to a release from pressurized piping. At proposal, piping pressure sensing methods were commercially available that manufacturers claimed could detect and shut off a release of at least 2 gal/hr from pressurized piping. Although the performance of these devices had not been evaluated, the magnitude of the problem and the availability of control technology caused the Agency to require shutoff devices in the proposed rule. The proposal preamble requested information on the field performance of pressurized piping release detectors (52 FR 12744).

A variety of approaches are available to rapidly detect large leaks in pressurized lines. Two main types of continuous in-line release detection devices are commercially available for pressurized piping: flow restrictors and flow shutoff devices. Both devices react to pressure behavior in the line. A flow restrictor monitors the increase in line pressure after the dispenser is turned on. If there is a leak in the line preventing the line from reaching operating pressure, the restrictor allows a limited flow of product through the line to the dispenser, and thereby signals a leak to the operator. An automatic shutoff device monitors pressure changes during periods when the dispenser is off. If the line pressure drops enough to indicate a leak, the pump turbine is shut off, and no product can be dispensed. In addition to in-line detectors, continuous interstitial, vapor or ground-water monitors may also give rapid reliable warning when a leak occurs.

Commenters generally agreed with the need for pressurized piping release detection that was conducted more frequently than monthly. Many felt that only flow restrictors should be required because they are widely available and have proven performance. Based on information from UST installers and on causes-of-release data, flow restrictors operating at a 3 gallon per hour detection limit will eliminate 80 to 95 percent of the volume of releases occurring from piping. These devices have been in use for years and can be immediately installed at a large number of existing or new UST installations.

The Agency recognizes that other methods of frequent piping detection can achieve the same level of performance as flow restrictors, however. The proposed rule allowed interstitial monitoring or continuous ground-water or vapor monitoring instead of automatic shutoff devices but did not include a performance standard for these methods. Research conducted since the proposal (Notice of Availability; 53 FR 10403) shows that under certain circumstances ground-water monitoring will not allow immediate detection of a release. To ensure an equivalent minimum level of protection against catastrophic pressurized piping releases, the final rule includes the requirement that all pressurized piping have automatic line leak detectors that provide warning of 3 gallon per hour releases within an hour. EPA expects that this requirement can be met using flow restrictors, automatic shutoff devices, continuous interstitial monitors, and some continuous vapor monitors. Continuous ground-water monitoring may also meet this requirement under certain site conditions (e.g., shallow water table). EPA expects that some of these devices will be sufficiently sensitive to meet the additional monitoring requirements discussed below. If a device meets both standards, no additional monitoring is required. A detailed discussion of the performance of these systems is contained in section IV.D.2.c. of today's preamble.

o Additional Monitoring for Smaller Leaks (§ 280.41(b)(1)(ii))

Commenters suggested the use of inventory control or line tightness testing in conjunction with flow restrictors, shutoff devices, or continuous monitors. The new causes-of-release information acquired since proposal shows even more strongly that piping is a major source of leaks from UST systems. Because of the importance of controlling pressurized piping releases, EPA agrees that additional release detection beyond the immediate detection discussed above is necessary. Flow restrictors may not detect small releases, so additional monitoring is necessary to detect these releases. Therefore, the final rule has been revised to require an annual line tightness test or monthly monitoring using one of the accepted methods for tank monitoring. Line tightness testing is more sensitive than the other methods (see section IV.D.2.e. below) and so can be conducted less frequently for equivalent protection. Further, because of the problems cited by commenters with the reliability of flow restrictors and the ability of operators to override them, the final rule also requires that the operation of the line leak detector be checked annually in accordance with manufacturer's requirements. This system of checks will limit the likelihood that pressurized piping will release large volumes suddenly. This approach, coupled with requirements for corrosion protection and careful installation and testing of piping (which will significantly lower the incidence of piping releases over time), will protect human health and the environment.

o Secondary Containment for Piping

Some commenters recommended that secondary containment with interstitial monitoring be required for all pressurized lines. The Agency feels, however, that the release detection options described above for single-walled pipes are protective of human health and the environment and that the additional stringency gained by double-walled pipes does not need to be required. EPA also notes that secondarily contained piping is a relatively new technology that would require significant new training of installers to "de-bug" the applications of this technology for widespread use. In any event, the monthly or annual tests will detect the small slow leaks from piping while the shutoffs or restrictors will detect the large releases. EPA agrees that secondary containment for piping is environmentally protective, and owners and operators may choose to install continuous interstitial monitors which meet the above standards.

(c) Requirements for Suction Piping (§ 280.41(b)(2)). In the proposed rule, all release detection requirements were waived for suction piping that meets six design and operating standards concerning pressure, slope, run of the piping system, and use of properly located check valves. This piping design, common in Europe, ensures that little, if any, product will be released if a break in the line occurs. The Agency also felt that this exemption would encourage the use of suction piping over pressurized piping wherever possible, which is desirable because suction systems are less likely to leak than pressurized systems. Suction systems which do not meet the criteria for exemption were required to be monitored in the same manner as the tank, because the Agency believed small, continuing releases were possible from these systems. Changes in the requirements for both exempt and non-exempt systems are discussed below.

o Design Criteria Revised

The proposed rule contained design criteria for exempt suction systems concerning the number and placement of check valves and the slope of the piping because small releases can occur in an improperly designed or installed system. Two of these criteria have been deleted and one criterion has been added in the final rule. First, commenters noted that the 2-percent slope mandated in the proposed rule is difficult to meet using current design practices, and the API and PEI codes of practice recommend a 2-percent slope. Commenters also noted that a 2-percent slope decreases the maximum distance between the tank and the pump, thus limiting station design. In the final rule, therefore, the specific value for the slope has been eliminated, and only the general requirement has been retained that the slope be such that product will drain back into the tank when suction is released.

Second, the proposed rule also required that suction systems have only one check valve as close to the pump as possible and not have a foot valve. These criteria overlap, however, and the foot valve restriction has been deleted from the final rule. This change does not weaken the final rule because the requirement for only one check valve near the pump will prevent a large volume of product being held in the portion of the pipe from which a leak could occur.

Finally, today's rule contains an additional provision that exempt suction systems must have a means of verifying compliance with the design requirements. In the preamble to the proposal, the Agency noted that suction systems in West Germany are equipped with a means of ensuring that the contents of the line are under less than atmospheric pressure (52 FR 12745). This serves to ensure that if a hole in the line develops the liquid drains back into the tank and is not released to the environment. Although the Agency received no comments directly related to this issue, commenters did note that service contractors frequently correct operational problems with suction systems by adding additional check valves to the line. An inspector could not easily discover this alteration and the system could begin to leak. Consequently, the Agency decided to require that exempt suction systems be equipped with a means for an inspector to readily determine that the system continues to meet the design requirements.

o Non-Exempt Suction Systems

Under the proposed rule, American-design suction systems, which commonly have a check valve near the tank or at piping unions, were required to have release detection conducted as frequently as for the tank. Commenters on the proposal noted that eliminating check valves at piping unions or near the tank is not practical in some applications and requested that these systems also be exempt from release detection requirements. The causes of release information acquired by EPA clearly indicates that both types of suction piping are dramatically less likely to have large leaks than pressurized systems. Based on this information, EPA requested comment on the use of less frequent testing (every 1 or 3 years) for suction systems in the supplement to the proposed rules (52 FR 48642). Commenters disagreed on an acceptable testing frequency but generally agreed that all forms of suction line pose a limited environmental threat. EPA agrees with commenters who noted that non-exempt suction systems can have small continuing releases under some conditions and, therefore, should not be exempt from detection requirements. Because the leaks from these systems are limited and will usually be signaled by operating problems, EPA believes that a 3-year frequency, in conjunction with required prevention measures, is protective of human health and the environment. The reduced frequency of testing for these systems should further encourage the use of suction systems should further encourage the use of suction systems over pressurized systems, even for applications where the exempt design requirements are impractical. These incentives add to the environmental protection provided by the piping release detection standards.

c. Requirements for Hazardous Substance UST Systems (§ 280.42). The proposed rule required the use of secondary barriers with interstitial monitoring at all new or upgraded UST systems storing hazardous substances, unless the owner and operator: (1) demonstrated that an alternative method of release detection would accurately and reliably detect a release of the hazardous substance from the UST system and (2) obtained approval to use another method from the implementing agency. Also, in the proposal, owners and operators of existing hazardous substance USTs were allowed during the 10-year upgrade period to use any of the methods of detection allowed for petroleum USTs, if the performance requirements for that method could be met. At the end of the proposed 10-year upgrade period, owners and operators would have to upgrade or replace this release detection with secondary containment and interstitial monitoring unless a variance was approved.

After consideration of numerous public comments submitted on these issues and additional research, EPA continues to believe that release detection and corrective action technologies are not as readily understood or widely used for the broad range of hazardous substances as they are for petroleum. EPA believes that secondary containment continues to be the most effective demonstrated method of release detection for new UST systems containing hazardous substances. Thus, EPA has retained the proposed approach in the final rule but has revised (1) the performance standards for the secondary containment and (2) the procedures for receiving a variance. In the final rule, in applying for variance, owners and operators of new or upgraded UST systems must provide (in addition to the demonstration of the effectiveness of a release detection method) information about applicable corrective action technologies, the inherent health risks, the chemical and physical properties of the stored substance, and any relevant characteristics of the particular UST site that would impact a future clean-up. These factors will be used by the implementing agency to guide its decision on whether to allow the alternate release detection method for the hazardous substance.

(1) Release Detection Requirements (§ 280.42(a) and (b)). Commenters took a variety of positions regarding the necessity of secondary containment for hazardous substance USTs. Many felt that the approach in the proposal was appropriate while others thought that specific chemicals, or all regulated substances, should be treated in a manner similar to petroleum. Many of these arguments centered on the similarities or differences between certain hazardous substances and components of petroleum. Commenters also disagreed about the most effective form of secondary containment, some favoring double-walled tanks, and others advocating excavation liners. Commenters generally were opposed to allowing soils as an excavation liner due to the difficulty of ensuring that the soils were properly treated and compacted.

The final rule has retained the proposed secondary containment requirement for new and upgraded hazardous substance UST systems with some significant revisions. The final secondary containment requirement is based on both technical and implementation considerations. From the technical perspective, secondary containment is believed desirable because it ensures that all hazardous substance USTs will be provided with effective detection methods, and, if a leak occurs from the primary containment structure into the interstitial space, corrective action will be simplified because it is very unlikely to impact the surrounding environment. As discussed in later sections of today's preamble, EPA has extensive information on the performance of various release detection methods and corrective action technologies already being applied to petroleum tanks. (The Agency allowed single-walled tanks and release detection for storage of petroleum substances because of the detectability of these substances and the belief that small releases could be cleaned up relatively easily.) Information about the performance of release detection and corrective action methods for hazardous substances, however, is not as readily available. Most importantly, there is very limited field experience with detection methods for hazardous substance tanks. This is made more significant by the fact that numerous hazardous substances are more toxic than petroleum and are less likely than petroleum to be detected by smell or taste. It is also important to note that, when replacing these hazardous substance tanks, industry has generally chosen to put them aboveground, in vaults, or in double-walled tanks. This lack of information and experience with effective single-walled tank strategies for hazardous substance USTs has caused the Agency to conclude that secondary containment is the most technically prudent approach to protecting human health and the environment.

In today's final rule, the Agency has given important consideration to both technical and implementation concerns. From an implementation standpoint, the secondary containment requirement is considered feasible by EPA because there are significantly fewer hazardous substance systems subject to regulation than there are petroleum tanks. In addition, secondary containment appears to be consistent with existing industry practices and state regulations for storage of these substances. As noted in the preamble to the proposal, state and local programs have adopted a variety of requirements for hazardous substance USTs that tend to emphasize secondary containment. In addition, these regulatory programs are generally not as widespread or well established as petroleum tank programs, which makes adoption of the secondary containment standard less disruptive to ongoing implementation. Although EPA recognizes that some hazardous substances can be both detected and cleaned up as easily as petroleum, there are many chemicals that cannot, and specific criteria for distinguishing them for UST release detection purposes are difficult to establish, especially in a rule to be applied nationwide. Commenters did not provide any workable approaches on new information upon which EPA could develop such criteria. Owners and operators are eligible for a variance if they can demonstrate that effective detection and cleanup technologies exist for the specific hazardous substances being stored (see the next section of the preamble).

The specific requirements for secondary containment of hazardous substance UST systems have been revised based on public comment and new information. Many commenters assumed that the proposal's alternative for secondary containment with interstitial monitoring required full secondary containment that would prevent the release of chemicals to the environment even in the event of a catastrophic tank failure. The proposal, however, was intended only to ensure detection of releases and not necessarily to contain them. For example, it allowed both excavation zone liners and double-walled tanks that did not have 360E coverage of the inner wall. Based on commenters' concerns and the preceding discussion, the final rule requires for new or upgraded UST systems storing hazardous substances that the outer barrier be capable of containing a release until it is detected and cleaned up. The effect of this requirement is to require both double-walled tanks and liners to completely surround the inner tank and be checked for releases as frequently as necessary to prevent the release of hazardous substances to the environment should a leak occur. This monitoring would, at a minimum, need to be conducted at least monthly.

In the final rule, the Agency has taken an approach that is based on the one followed for hazardous waste tanks under Subtitle C of RCRA (40 CFR 264, 265; Subpart J). It is the Agency's belief that many of the hazardous substances covered in today's rule have properties that are very similar to those of hazardous waste. Consequently, both rules require that a leak at any time during the operational life of the facility will be contained until it is detected and removed from the containment system. Because there are significantly more hazardous substance UST systems than hazardous waste tank systems, and because Subtitle I UST systems are not part of a permitting program, the Agency has retained the performance-oriented approach that was proposed. (It is the same general approach used for petroleum tanks.) This performance-oriented monitoring approach is intended to provide enough flexibility to control the greater number and variety of hazardous substance tanks, without the use of permits, while at the same time providing the same level of protection as provided by the tank requirements under Subtitle C of RCRA.

(2) Variance to Release Detection (§ 280.42(b)). As stated in the proposal preamble, EPA recognizes that secondary barriers with interstitial monitoring may not be necessary for all hazardous substance UST systems because there may be alternate release detection methods that, based on the tank system characteristics, location, and the nature of the substance stored (52 FR 12741), adequately protect human health and the environment. The Agency, therefore, proposed to allow the use of other release detection methods if the owner or operator demonstrated that an alternate method reliably detects the release of the hazardous substance from the UST system and obtains the approval of the implementing agency (see proposed § 280.41(b)).

A number of commenters supported the variance provision, while others opposed allowing a variance in any situation. The Agency has concluded that the risks posed by the use of an alternate method will be minimized by requiring the owner and operator to apply for a variance, and by stipulating that an alternate release detection method can be substituted only if it is as reliable as those used for petroleum and petroleum-based substances, and all performance criteria for the release detection method are satisfied at the site. In addition, information about available corrective action technologies, and the inherent health risks and chemical properties of the substance stored, will also need to be provided by the owner and operator to the implementing agency for review. These other factors may override the detectability criteria. For example, if a given vapor monitoring device detects a hazardous substance with better sensitivity, accuracy, and response time than it detects petroleum, and if all the other criteria for the operation of the method are met at the site, then the owner and operator is eligible to apply for a variance from the secondary barrier requirement. However, the implementing agency will approve the alternate method only after considering the additional information about the characteristics of the substance stored and the availability of corrective action methods for that substance if a release occurs at the site. Thus, although an effective and accurate vapor monitoring method is available, the implementing agency may decide not to approve the variance request if the UST is located near a drinking water well or there is no corrective action technology that can be applied to remove the release before this resource is adversely impacted.

For existing hazardous substance USTs, the Agency allows the use of alternate release detection for up to 10 years, at which time owners and operators must have applied for and received a variance and must have upgraded the existing system to new tank standards. Although the application for a variance can be made any time within the 10-year upgrade period for existing USTs, the owner and operator must meet the appropriate performance standards in §§ 280.43 and 280.44 for an alternate release detection methods at all times during the 10-year interim period.

(a) Application Process. In the preamble to the proposal, the Agency requested comments on procedures that could be used by owners and operators to apply for a variance (52 FR 12743). Commenters provided a number of suggestions on how the variance application process could be implemented. These suggestions included a variance application procedure similar to the one promulgated in the final rules for hazardous waste tanks (40 CFR Parts 264 and 265), a "nationwide" variance application procedure for single companies storing similar products in tank systems located in several states, or a joint petitioning procedure for different companies in the same state that have similar characteristics of tanks, substances, and release detection methods.

The Agency evaluated all the suggested application procedures and decided to defer the specific implementation details to the implementing agencies. Numerous factors, such as differences among the traditional procedures already followed by implementing agencies, variability of site conditions, and the number and configuration of tanks at each site, suggest that administrative details for dealing with this variation should be left to the specific agency reviewing the request. The Agency plans to develop information about a number of variance application options that can be used by the implementing agencies in designing a variance process in their jurisdiction. The final rule contains only the baseline requirements that the owner and operator must satisfy in order to use a substitute method of detection at new or upgraded hazardous substances USTs; these requirements are to: (1) demonstrate the detectability of the substance using the alternative release detection method; (2) provide data about the corrective action technology that will clean up a release within the constraints of the UST site characteristics and the inherent chemical, physical and health-risk properties of the substance; and (3) obtain variance approval before the installation and operation of the alternate release detection method.

(b) Evaluation Criteria. In response to EPA's requests for public comment on the type of information and criteria that could be used by the implementing agencies to grant variances (52 FR 12742), commenters recommended that EPA develop a list of hazardous substances that would qualify for variances on a class basis. They also recommended that EPA develop a list of hazardous substances that are petroleum-like substances as well as identify physical properties for hazardous substances that could be considered for variances. No commenters provided a workable set of criteria for doing this, however. Significant new information was not provided by commenters concerning the characteristics of hazardous substances. The Agency proposed to compile a similar list of the specific petroleum substances but rejected this approach in the final rule. (See section IV.A.4. for a discussion of the definition of "petroleum" and "petroleum UST systems.") After considering the different evaluation criteria in a variance program, the Agency decided that a specific set of national evaluation criteria is not workable given the diversity of the of the chemical properties of the hazardous substances, the USTs in which they are stored, and the variability of site characteristics. The Agency believes that specific variance program implementation details (for example, the use of "class" variances) are best left to the implementing agencies. The Agency does, however, intend to assist the implementing agencies in their variance program development by providing alternative procedures for simplifying the application and approval process and by providing more information about the general criteria that must be satisfied at a minimum nationally in each variance application.

(c) Specific Changes to the Rule. The final variance requirements only differ in minor ways from those in the proposed rule. The requirement to notify the implementing agency of the intent to conduct a demonstration for a variance has been removed. Instead, owners and operators will be required "to demonstrate that an alternate method can detect a release" before it is used at a new or upgraded UST system. Thus, prior notification is simply an item of administrative convenience that can be required by implementing agencies at their discretion. The final requirements make clear that it is advantageous for the owner and operator to apply as early as possible, because an alternative method cannot be used to meet the new or upgraded hazardous substance UST system release detection requirements without approval of a variance by the implementing agency.

In the final rule, owners and operators are also required to submit to the implementing agency information about the availability of corrective action technologies that could be used should there be a release of the stored hazardous substance at that site, including a consideration of any unusual health risks that are posed by a release of the stored substance. The Agency believes that this additional type of information is needed by the implementing agency to be able to evaluate variance applications. This change is in response to those commenters who believed that the proposed approach was not stringent enough and could result in adverse impacts to human health and the environment. Also, these new criteria provide the further clarification of evaluation criteria that was suggested as needed by other commenters. EPA believes these general criteria will more adequately (than the proposed approach) ensure that only variances will be applied for that, if approved, will protect human health and the environment.

d. Methods of Release Detection for Tanks (§ 280.43). The proposed rule presented in one section the release detection options that addressed the complete UST system (both tanks and piping). It also specified in one section the methods that need to be combined, the frequency of testing, and the applicable performance standards. The final rule has been reorganized in response to several commenters who stated that this format was unclear and that it was confusing in the way it combined tank and piping detection methods and petroleum and hazardous substance requirements. All of the general requirements, such as combinations and testing frequencies, are now contained in § 280.41 for petroleum and § 280.42 for hazardous substances. Any substantive changes in these areas are discussed in previous sections of the preamble (IV.D.2.b. and c.). The final rule in § 280.43 now simply lists the methods that can be used for detecting leaks in tanks and the conditions under which they can be used to meet the requirements of § 280.41 and § 280.42. The methods applicable to underground piping are listed separately in § 280.44 and are discussed in a later section of this preamble (IV.D.2.e.)

(1) Inventory Control (§ 280.43(a)). The proposed rule allowed the use of monthly inventory control in combination with periodic tank tightness testing as a method of release detection. Although listed as a separate method in the final rule, inventory controls still must be combined with another method of detection (see section IV.D.2.b.1. above). The Agency acknowledged that inventory control is affected by many variables but maintained that the technique is effective when correctly performed by trained personnel. The proposed rule contained weekly and monthly performance standards and procedural requirements to optimize the effectiveness of inventory control.

Commenters generally felt that the performance standards and the procedures were too stringent, and that the method was unreliable because the variables were essentially uncontrollable. EPA's research since proposal demonstrates that the proposed rule would in fact result in an unacceptable rate of false alarms; consequently, the final rule has been revised to eliminate the proposed weekly performance standard and to change the monthly performance standard to 1 percent of monthly product throughput plus 130 gallons. In addition, the proposed requirement that only tanks partially within the ground water perform a monthly test for water within the tank has been expanded to apply to all tanks. Also, the proposed limitation on the accuracy of the dispensing meter has been revised to make it consistent with local weights and measures standards. These changes are discussed in more detail below.

(a) Performance Standard. EPA proposed that inventory control be conducted to detect a release of at least 5 percent of flow-through on a weekly basis and 0.5 percent of flow-through on a monthly basis; the latter standard was based on an API recommended practice. The Agency was concerned that having only a monthly inventory control requirement would lead to large releases occurring for a month before detection. Therefore, the proposal included a weekly loss standard that was intended to provide early warning of major losses.

Several commenters stated that the proposed 5 percent weekly and 0.5 percent monthly flow-through requirements were too stringent and too difficult to achieve, particularly the weekly requirement. They also felt that the weekly requirement would result in many false alarms, thus unnecessarily burdening owners and operators and the implementing agencies. They suggested that detection of such small leaks from tanks with low flow-through of product was difficult, that the temperature influences on volume were too great to obtain accurate data, and that the technology was not sufficiently developed to detect small releases and, at the same time, ensure against false alarms. Other commenters supported the monthly flow-through requirement.

In response to commenters' concerns, the Agency has since proposal evaluated several approaches to inventory control to determine the minimum leak rate that could be reliably detected. The evaluation considered basing inventory leak determinations on the percent of throughput, percent of throughput plus a constant volume, or number of days exceeding a given loss. A data base of actual inventory records consisting of over 20,000 measurements at nearly 600 tanks nationwide was used in a computer simulation of leak rates. The study indicated that the proposed monthly standard would result in a large number of false alarms (approximately 30 percent on a monthly basis). The most sensitive of these inventory control methods would detect a loss of about 1 gallon per hour at a PD of 95 percent and a PFA of 5 percent. This investigation revealed that using sophisticated statistical inventory control or pooling the inventory data for several months would improve the sensitivity of all the methods. Notice of this study was published for comment in the Federal Register (53 FR 10403). A commenter submitted a study that confirmed the high false alarm rate of the monthly standard and indicated that the weekly requirement also had an unacceptably high false alarm rate (over 50 percent).

In response to these comments and the studies discussed above, EPA has modified the final rule. The Agency acknowledges that other methods can detect smaller releases, but the final rule allows inventory control combined with tank testing for use during the phase-in period or with new, protected tanks because of its general effectiveness, low implementation costs, and ease of application. As discussed above, however, the proposed 5 percent weekly detection rate produces an unacceptably high false alarm rate, as does the proposed 0.5 percent monthly standard. Consequently, the requirement for weekly inventory control has been omitted from the final rule and the monthly requirement has been changed to 1.0 percent of flow-through plus 130 gallons, which has a false alarm rate of about 5 percent. The Agency believes that owners and operators will take inventory results more seriously when the number of false alarms are reduced. EPA has further concluded that inventory control offers a practical and effective means of detecting releases only when combined with tank testing or automatic tank gauging.

In addition to the performance standard, the proposed rule contained six restrictions that had to be followed when conducting inventory control. The restrictions that received the most comment involved measurement accuracy (to 1/8 inch), the accuracy of the dispensing meters, and the requirement to measure water only when some portion of the tank is below ground water. These restrictions are discussed below.

(b) Dipstick Measurement Accuracy. At proposal, the Agency believed that reading dipsticks to the nearest 1/8 inch could be performed and that reading to this level would reduce major errors involved with the measurement of the stored product's height (52 FR 12727). A number of commenters felt that many factors interfere with precise dipstick measurements, and, therefore, it is not reasonable to require accurate measurements for an imprecise method. Commenters also felt that the 1/8-inch requirement was inconsistent with existing conversion tables, which are usually marked in whole inches. Other commenters suggested reasons why dipstick methods of measurement were inappropriate. The EPA-sponsored research discussed in the previous section indicates that increased stick-reading accuracy down to and including 1/8-inch does improve the performance of inventory control. The Agency also has concluded that existing dipsticks marked to 1 inch can be successfully read to the nearest 1/8 inch to improve accuracy, or that conversion tables can be modified.

(c) Dispensing Meter Accuracy. Accurate metering is essential to inventory control. Therefore, based on the National Conference on Weights and Measures Standard, the Agency proposed that dispensing should be metered to within 5 cubic inches for every 5 gallons of product withdrawn, even if it meant that certain owners and operators would have to install or calibrate dispensing meters. Several commenters preferred the requirement to be 6 cubic inches per 5 gallons, in accordance with the National Bureau of Standards (NBS) Handbook #44, or that it be the same as the local weights and measures standards. EPA agrees with commenters that local weights and measures calibration standards will be adequate to ensure accurate inventory reconciliation where these standards are applicable. At dispensing meters that are not covered by local weights and measures standards, the nationally recognized NBS standard is appropriate. The final rule has been revised accordingly to allow both calibration standards.

(d) Monthly Water Check. The proposal required the measurement of any water in the tank at least monthly if any portion of the tank is within the water table. Water in a tank could result from ground-water intrusion, indicating that there is a hole in the tank. One commenter suggested more frequent measurement of water under all situations rather than only in those cases where the tank is partially within the water table. The Agency agrees that water could enter the tank even when the water table is deep (e.g., when rain water temporarily collects in the excavation pit). The presence of water in a tank can indicate a hole in the tank, whether the water comes from ground water or rain water. The provision now requires the water measurement monthly at all tanks conducting inventory control. The monthly frequency was retained because it agrees with the frequency of other release detection methods, which was selected as the frequency that is protective of human health and the environment. The measurement of water in the tank is a simple procedure routinely conducted by many operators already, so this requirement should not significantly increase the operator's burden.

(2) Manual Tank Gauging (§ 280.43(b)). In the supplement to the proposed rule (52 FR 48635), EPA requested comments on the use of static inventory control (or manual tank gauging) as a release detection technique for used oil (or other types of) UST systems. Commenters generally supported the use of this detection method, although they differed widely on what tanks should be allowed to use this method, how long the test should be, and how frequently it should be performed. Research submitted by one commenter, and EPA's own analysis, indicates that this method is as effective as inventory control at smaller tanks (below 2,000 gallons) and can reliably detect leaks of 0.2 gallon per hour for tanks of 550 gallons or smaller. Consequently, final rule allows its use as the sole release detection method for tanks 550 gallons or less in capacity. Also, manual tank gauging can be used in the final rule for tanks of 551 to 2,000 gallons as a substitute for the inventory control part of the method (that combines monthly inventory control with periodic tank tightness testing). Manual tank gauging cannot be used to meet the release detection requirements at tanks larger than 2,000 gallons.

Commenters were divided over the applicability and effectiveness of manual tank gauging. Some commenters recommended that manual tank gauging be approved because it is effective and inexpensive and because used oil poses less environmental risk than other regulated substances. They also noted that the lack of piping and small deliveries at these tanks made releases less likely. Some commenters opposed the use of manual tank gauging because they believed the method is not sensitive enough and includes too much room for human error. These commenters also frequently recommended requiring secondary containment for used oil because it contains hazardous constituents and no other detection methods are as effective.

As discussed earlier in today's preamble (section IV.A.3.g.), the Agency believes that used oil does not differ substantially from petroleum products and should be regulated in the same manner. Consequently, today's final rule requires that these tanks conduct accurate monthly monitoring using the methods applicable to other petroleum tanks. The Agency requested comments in the supplement to the proposed rule on the possibility of including manual tank gauging as a monitoring alternative; specifically, for used oil tanks or other types of tanks. Commenters recommended a wide range of specific requirements. Suggested frequencies ranged from daily to every 5 years and suggested test lengths ranged from 6 hours to 72 hours or longer. Some commenters recommended this test should apply to other regulated substances, especially products similar to used oil. Commenters disagreed on the appropriate size limit for tanks allowed to use this method.

In order to determine more precisely the performance of this method, EPA analyzed a report submitted by a commenter. EPA's evaluation (Notice of Availability; 53 FR 10403) indicates that for tanks smaller than 550 gallons, manual tank gauging can detect 0.2 gallon per hour leaks with a PD of 0.95 and a PFA of 0.05 when performed in accordance with today's rule. For tanks between 550 gallons and 2,000 gallons the method achieves the performance of inventory control. For those tanks, the method is capable of detecting leaks of 1 gallon per hour or less with a PD of 95 percent and a PFA of 5 percent, even under extreme temperature conditions. Based on commenters' suggestions and these study results, the final rule includes manual tank gauging as an allowed leak detection method. In today's rule, only tanks of 550 gallons or smaller may use this as the sole method of release detection. Larger tanks (551-2,000 gallons) may use it in place of the inventory control part of the release detection method that combines inventory control with periodic tank tightness testing.

According to the Agency's evaluation, the performance described above can be achieved if the following limitations are met: the test period is at least 36 hours, depth measurements are taken twice and averaged, and the depth is read to the nearest one-eighth of an inch. These criteria have been included in today's final rule.

(3) Tank Tightness Testing (§ 280.43(c)). The proposed rule allowed periodic tightness testing in conjunction with monthly inventory control as a release detection method for all tanks. While acknowledging the uncertain performance of tank testing, the Agency believed that it was a demonstrated and effective method that would be available to meet the large demand for release detection following promulgation. To maximize the performance of tank tightness tests, the proposed rule included a performance standard of 0.1 gallon per hour at a PD of 99 percent and a PFA of 1 percent. Commenters disagreed about whether tightness testing should be included in the rule and what level of performance it should be required to meet. Additional evaluation by EPA indicates that, although few methods can now meet the proposed standard, several methods could make a few changes to equipment and protocol and meet the proposed standards. For these reasons, the final rule retains the proposed tank testing standard of 0.1 gallon per hour.

(a) Performance Standard. At proposal, EPA recognized that many factors such as temperature changes, tank end deflection, and vapor pockets affect the accuracy of tank tightness tests (52 FR 12724-12725). To limit the allowable methods to those that most effectively compensate for these problems, the Agency included a performance standard in the proposed rule. The level of this standard was selected based on the results of the national survey of underground motor fuel storage tanks and the experience of tightness testing practitioners; little evidence was available to support the NFPA 329 criterion of 0.05 gallon per hour. The proposed standard of 0.1 gallon per hour at a PD of 99 percent and a PFA of 1 percent was believed to be the maximum performance achievable on the typical sizes of tanks in use. A high PD was included to adequately protect human health and the environment, and a low PFA was included to prevent a heavy burden on owners and operators and the implementing agencies that is caused by investigating a large number of false alarms.

In the proposal preamble, EPA acknowledged that there was insufficient information on the performance of available tank tightness tests, particularly their ability to meet the NFPA 329 standard of 0.05 gallons per hour with the proposed probabilities of detection and false alarm. It was further stated that the Agency was beginning an investigation to acquire this data (52 FR 12724-12725). Since that time, EPA's research laboratory at Edison, New Jersey, has evaluated 25 tank tightness testing methods representing a wide range of approaches. During each evaluation, blind tests were conducted by company operators, the data were evaluated by EPA, and computer models were used to simulate leak rates and probabilities and compare them to the operational data. The results of these evaluations indicate that, while most of these currently available methods of tank tightness testing are capable of detecting leaks of 0.1 gallon per hour, only a few could presently detect this leak rate at the specified probabilities. Even with the oftentimes simple equipment and procedural changes that are necessary to achieve the needed level of improved performance, some methods probably would not be able to meet the proposed PD and PFA. Notice of the results of this study was published in the FEDERAL REGISTER on March 31, 1988 (53 FR 10403).

Some commenters felt that tightness testing is too unreliable to be an allowable release detection method, at least without further evaluation. Several commenters stated that the proposed standard was too stringent to be met with existing technology, particularly for larger tanks. Others recommended a more stringent standard for the release rate, usually citing the NFPA 329 criterion of 0.05 gallon per hour as the way to spur development of the technology. One commenter suggested that the performance standard should become more stringent over time as the methods develop. The proposed standard was supported by some commenters.

Based on the Agency's evaluation of tank tightness test methods and the concerns raised by these commenters, the final performance standard has been revised. Many of the methods evaluated by EPA at its Edison, New Jersey laboratory were able to detect a release of 0.1 gallon per hour so this value was retained in the final rule. The detection probabilities associated with the standard have been moved to the general requirements section (and are discussed earlier in this preamble in section IV.D.2.a.). During the two years before these probabilities are effective, the Agency believes additional standards are needed. Thus, the final rule language has been changed to include the requirement that tank test methods must account for the effects of thermal expansion or contraction of the product, vapor pockets, tank deformation, evaporation or condensation, and the location of the water table. These variables have been identified as important in EPA research (Notice of Availability; 53 FR 10403) and in the NFPA 329 recommended practice.

The Agency's evaluation indicates that many tightness test methods can meet the final performance standard with relatively minor procedural or equipment changes. Such changes, however, have not been developed and instituted by the manufacturers of many of the different devices, tried in the field, or evaluated in laboratory conditions. In addition, manufacturers have not described their performance in terms of probability of detection and probability of false alarm in the past. There will be a large demand for tightness testing for release detection and confirmation following promulgation of the final rule. Tightness testing is the most widely available release detection method and the method that is likely to be used by many UST owners and operators over the next 5 to 10 years. EPA was concerned that, if few or none of the tightness test methods could meet the performance standard, there would be insufficient industry capacity to meet demand for release detection. For these reasons, the final rule has been revised to delay the effective date of the probabilities of detection and false alarm (see discussion of this delay in section IV.D.2.a.5. above).

The Agency decided to give manufacturers 2 years in which to develop and put into operation the necessary procedural and equipment changes or to develop new methods that meet the entire performance standard. Such a delay was recommended by some commenters and will ensure that necessary equipment and procedural changes can be made by method manufacturers. As discussed previously in the preamble, the Agency believes significant environmental benefit can be obtained by conducting release detection quickly. The delay in performance standards should allow more wide spread and rapid testing of tanks.

(b) Relationship of the Final Performance Standard to NFPA 329. As noted above, some commenters felt that the proposed performance standard of 0.1 gallon per hour at a PD of 99 percent and a PFA of 1 percent was too lenient and was a relaxation of the standard currently followed by the industry, which is 0.05 gallon per hour recommended in NFPA 329. However, the Agency has concluded that the final performance standard is not less stringent than NFPA 329. First, the standard in the final rule still requires testers to declare a leak at a threshold value of 0.05 gallon per hour (see section IV.D.2.a.4. above). Second, the NFPA guideline specifies a detectable leak rate but does not specify the probability with which this leak rate must be detected. Most existing volumetric methods will detect 0.05 gallon per hour leaks at least a portion of the time. A relatively poor method that detects a leak of 0.1 gallon per hour with a PD of 50 percent and a PFA of 1 percent could still claim to meet the NFPA criterion because the probability of detecting a leak is not specified. Direct comparison to the Agency's performance standard shows that the NFPA 329 criterion would allow more leaks to go undetected and also cause more false alarms. EPA's final performance standard intended to eliminate the use of poor tightness test methods, ensure that more leaks are detected, and cause fewer false alarms. Thus, it provides for better protection of human health and the environment while ensuring unnecessary or counterproductive burdens on owners, operators, and implementing agencies are minimized.

(c) Large Tanks. In the preamble to the proposed rule, EPA stated its doubts that the performance standard was attainable for large tanks (52 FR 12725). Some commenters agreed and some disagreed. Research at EPA's Edison facility suggests that the 0.1 gallon per hour standard cannot be met for large tanks using current test methods. Rather than allowing less stringent test methods for large tanks, which would create the potential for ignoring large releases, EPA believes that owners and operators of large tanks will have to select other release detection methods besides tank tightness testing unless improved or new methods are developed that allow the standard to be met for the larger tanks. (See section IV.A.3. of this preamble on bulk underground storage tanks.)

(4) Automatic Tank Gauging Systems (§ 280.43(d)). Automatic tank gauging systems (ATGS) were included in the proposed rule as one of the options for release detection (although the name of this option as proposed was "automatic monitoring of product level and inventory control"). These monitors generally have two modes of operation: leak detection mode (product level monitoring) and inventory control mode. The proposed rule required that the leak detection mode be used at a minimum once a month and meet a performance standard of 0.2 gallon per hour with a PD of 99 percent and a PFA of 1 percent and that the automatic inventory control be conducted to meet the performance requirements for manual inventory control. Some commenters stated their belief that ATGS are unreliable, unproven, and too costly. Other commenters felt that the proposed performance standards were too stringent. Many commenters favored the use of ATGS but suggested changes to the requirements in the proposed rule. As discussed below, the Agency has retained ATGS in § 280.43(d) of the final rule. The proposed leak rate standard has been retained but the probabilities have been revised and their effective date delayed 2 years. The proposed requirement that the tank be 80 percent full at the time of the test has been deleted. These changes and comments are discussed in more detail below.

(a) Performance Standard. At proposal, the Agency recognized there were limited performance data on ATGS, and that there was still significant opportunity for human error. However, ATGS were a relatively commonly used release detection method with the potential to be very sensitive in detecting leaks, and EPA wanted to include as many effective release detection options as possible in the rule to provide flexibility (52 FR 12736-12737). To limit the allowable ATGS to those that effectively control the possible sources of error, the proposed rule included a performance standard of 0.2 gallon per hour with a PD of 99 percent and a PFA of 1 percent. The 0.2 gallon per hour value was based on the equipment manufacturers' claims. This performance standard was intended to challenge manufacturers to prove that their systems could meet the standard.

Some commenters felt that ATGS were unproven and unreliable as a release detection method primarily because of their sophisticated electronic components. Other commenters felt that the PD was too stringent, while others agreed with this part of the proposed performance standard. A commenter provided a study conducted over several months at an operating service station equipped with several ATGS. One of the devices already clearly achieved the proposed performance standard. This study confirmed the Agency's conclusion that ATGS can be effective in field situations; thus, this release detection method has been retained in the final rule.

EPA's own research and the results of the field evaluation of ATGS demonstrate that 0.2 gallon per hour leaks can be reliably detected by several of the difficult types of available devices, and the proposed leak rate of 0.2 gallon per hour has been retained in the final rule, but the PD has been changed to 95 percent and the PFA to 5 percent. In addition, the final rule schedules the PD and the PFA standards to take effect 2 years after the effective date of today's rules (see discussion of this delay in section IV.D.2.a.5. above). Any owner or operator installing an ATGS in the intervening 2 years will not have to replace it after 2 years, unless it does not meet the performance standard. EPA's review of the systems now available indicates that they should be able to meet the standards with only minor adjustments.

Some commenters felt that the Agency should set a detectable leak rate for ATGS of 0.1 gallon per hour rather than the 0.2 gallon per hour requirement in the proposed rule. EPA, however, does not agree because current equipment as presently used cannot meet such a standard, and the suggested approach would effectively eliminate ATGS as a release detection method. Furthermore, since the product level test is conducted monthly, the performance of this method can statistically equal or exceed the sensitivity achieved by periodic tank testing, even though the monthly performance standard is less stringent. The Agency intends to keep as many effective methods available to the owners and operators so that they will have flexibility in selecting a release detection method that best suits their needs.

(b) 80-Percent-Full Requirement. The proposed rule required that the tank be 80 percent full during the monthly test. This requirement was intended to ensure that a large portion of the tank's surface was checked for releases (52 FR 12737).

Many commenters opposed this requirement because it would be difficult for tanks with low monthly flow-through to implement, and it would also be an impediment to businesses with UST systems that are in use continuously. New information now indicates that this limit on conducting the test is unwarranted. Information from experienced UST field personnel, a study in Suffolk County, NY, and comments received by EPA demonstrate that most corrosion holes occur in the sides and bottom of tanks, not the top, so there is little need to always test the top of the tank for releases.

In the absence of a specific requirement, release detection tests conducted every 30 days are expected to include tests over the range of levels to which the tank is typically filled, so those areas of the tank routinely in contact with product will be tested. The inventory control requirement will also ensure that releases from the top of the tank are detected if they occur. Also, many owners and operators do not keep their tanks 80 percent full during normal operation; if tanks are rarely filled above a certain point, testing above that point is simply unnecessary. Also, the overfill prevention requirements in § 280.20 and § 280.30 of today's rule will prevent product from reaching the very top of the tank. Therefore, the Agency has deleted this requirement from today's final rule.

Although testing at 80 percent of capacity is no longer required, the testing procedure should be conducted when the tank is near its highest level and the testing should be conducted after waiting a sufficient time after the delivery of the product. This waiting period is necessary because, after product is added to the tank, it takes time for the product to mix and achieve the nearly static condition necessary to conduct meaningful product level monitoring (52 FR 12725). EPA anticipates that this waiting period will become a part of the manufacturer's recommended procedures as they attempt to meet today's standard concerning detection probabilities.

(c) Combination with Inventory Control. In combination with ATGS, the proposed rule required inventory control that meets the same performance standards required for manual inventory control. Inventory control was required in addition to product level monitoring because it was believed that product level monitoring was not as accurate as tank tightness testing or external methods and, therefore, a backup release detection method was needed. This requirement is not burdensome because ATGS routinely collect the information needed to conduct inventory control. No comments were received on this issue. Therefore, inventory control is still required in conjunction with ATGS in the final rule. Some of the performance standards for inventory control have been revised in today's final rule and are discussed in section IV.D.2.d.1. above.

(d) Effectiveness with Piping. A commenter was concerned about the effectiveness of ATGS in detecting releases from piping. EPA agrees that, in the leak detect mode, this method will not detect piping leaks, and the inventory mode will detect piping leaks but not with sufficient sensitivity to detect small leaks. The causes-of-release information collected by EPA since the proposal (see section II.F. of this preamble) indicates that piping is a major source of releases. Therefore, the final rule has added several requirements for piping (see section IV.D.2.b.2. of this preamble). The owners and operators of UST systems using ATGS as a release detection method for the tank will have to select an additional means of release detection for the piping, such as line tightness testing.

(5) Vapor Monitoring (§ 280.43(e)). The proposed rule allowed the use of vapor monitoring in the excavation zone as a method of release detection as long as certain conditions and limitations were met that maximize performance. The information available to the Agency at the time of proposal indicated that vapor monitoring was, under optimum conditions, a sensitive and reliable release detection method. Research conducted by the Agency since the proposal, as well as information submitted by commenters, has confirmed that vapor monitoring is capable of detecting extremely small leaks (0.003 gallon per hour) under certain conditions. Some of the proposed conditions concerning the use of this method, however, now appear unnecessary. In summary, the final rule includes vapor monitoring, omits the 500-ppm background restriction, and allows monitoring for selective components of the stored regulated substance, including tracers mixed in the substance.

(a) Effectiveness of Method. Some commenters expressed concern that vapor monitoring was not developed enough to be relied upon as a sole method of release detection. The Agency solicited comments and additional data on the effectiveness of vapor monitoring in the proposal preamble (52 FR 12728). EPA included it as a method of release detection in the proposal because existing data indicated that it should be an extremely sensitive monitoring tool and it has already been used successfully in some state and local UST programs. The Agency has received no research results or data from commenters that would cause it to alter its earlier conclusions. In fact, several commenters submitted detailed field and experimental data verifying the effectiveness of vapor monitoring. A theoretical computer modeling study done by EPA confirms these results (Notice of Availability; 53 FR 10403). Although there are still not enough data to specify a more complete performance standard for this method (in terms of probability of detection, false alarm, and leak rate), the Agency believes this method will provide effective release detection, and has retained it as an option in the final rule, as long as it meets the limitations discussed below.

In response to concerns raised by some commenters, the Agency wishes to note that in the final rule, a vapor monitoring well does not necessarily mean a typical ground-water well. Instead, a vapor monitoring well means any sampling point from which vapors are collected and brought to the monitor by any means.

(b) Background Concentration. In the proposed rule, EPA limited the use of vapor monitoring to those areas with less than 500-ppm background concentration of total organic hydrocarbons in the soil gas of the excavation zone. Many commenters felt that the 500-ppm restriction was too stringent, stating that natural background concentrations of methane, for example, are higher than 500-ppm in some areas of the country. Other commenters wanted a restriction based on site-by-site evaluations. The 500-ppm background level was originally set as an attempt to recognize that background effects can interfere with effective vapor monitoring (52 FR 12729-12731). Commenters described research and field experience, however, that indicate vapor monitoring should work well at several thousand ppm levels in the soil gas if a volatile substance (such as gasoline) is being stored in the tank. In addition, EPA research has shown that at many sites having no recorded release from the UST system, the total background concentration exceeds the 500-ppm level. This type of background level is probably due to an accumulation of spills and overfills.

To avoid precluding the use of vapor monitoring at many sites where it is potentially applicable, no specific numerical restrictions on background levels is included in today's final rule. Instead, there is now a more general performance requirement that the background concentration must not interfere with the ability of the method used to detect releases. For example, at a site with too high a background level (e.g., over 10,000 ppm for gasoline), leaks may not produce a detectable concentration increase over background. Determining if the background levels are too high depends on the resolution of the sensor, the volatility of the product being monitored, and other site conditions.

(c) Measurable Component. Several commenters expressed concern that the proposal did not allow monitoring for specific components of the stored substance and, instead, appeared to be focusing only on total organic hydrocarbons as the detection criterion. EPA's research indicates that monitoring for specific components or tracer compounds may be very advantageous under certain site conditions because it eliminates some of the difficulties with background interference levels and false alarms. For example, sensors that detect only the lightest components of gasoline (e.g., butanes and pentanes) may suffer fewer problems with high background levels because these constituents vaporize so rapidly that the potentially confounding effects of past spills at the site are minimized.

Monitoring for these components might also allow spills to be more easily distinguished from equipment leaks, because the level of the light chemical components would return to background levels more quickly if it was suddenly elevated due to an episodic event like a spill. Similarly, if a carefully selected tracer compound that is not already present at the site is placed in the tank, background interference problems can be eliminated. Some of the advantages of tracer methods were also discussed in the proposal preamble (52 FR 12730).

For the above reasons, EPA believes that effective vapor monitoring systems can be designed to monitor for specific components, tracer compounds, or total organic hydrocarbons. Consequently, the Agency explicitly allows these additional methods in § 280.43(e) of the final rule.

(d) Sensitivity of the Vapor Monitor. The proposed rule stated that the threshold of the vapor monitor must be preset specifically for the type of regulated substance stored in the UST. This requirement was included in an attempt to increase the sensitivity of the monitor to detect releases in areas with background hydrocarbon levels. Since proposal, however, changes have been made to the vapor monitoring requirements based on public comment and new EPA analysis that make this requirement unnecessary or even undesirable. Tracer compounds may now be used for vapor monitoring, so specificity of the monitor to the regulated substance is undesirable. The EPA-sponsored computer modeling that was conducted on vapor monitoring performance indicated that, in some circumstances, a monitor would be more effective if it monitored for a single component of the stored substance than the complete stored substance because it would be able to differentiate more distinctly between a leak of that component and existing background hydrocarbon levels. Again, specificity of the monitor to the complete regulated substance would be undesirable. The final rule now simply requires that the monitor be capable of detecting a significant increase above background of the regulated substance, a component or components of the substance, or a tracer compound. This change will allow the use of existing sensitive monitors such as those measuring BTX (benzene, toluene, xylene) or "total hydrocarbons" because they can detect components of many regulated substances.

(6) Ground-Water Monitoring (§ 280.43(f)). The proposed rule allowed the use of monitoring for free product on top of the ground-water table to determine the presence of a release from an UST system. Many commenters agreed with this approach to release detection but wanted EPA to ease some of the limitations placed on the use of the method. The final rule still allows monitoring on top of the water table for free product but with several changes: well placement is no longer limited to the excavation zone; the well screen must be designed to prevent clogging and intercept the water table at both high and low ground-water conditions; and the well must be sealed from the ground to the top of the filter pack.

(a) Effectiveness. In the proposal the Agency recognized that there are several concerns about the use of ground-water monitoring, primarily the fact that the resource being protected (i.e., ground water) is the medium in which the release is detected. The Agency included this release detection method in the proposed rule, however, because this method has been demonstrated to successfully detect small petroleum releases and it is currently in widespread use in several state UST programs, such as that in Florida. EPA proposed limiting ground-water monitoring only to floating free product because its presence can be detected more quickly and reliably following a release than can dissolved product. The proposed rule also contained several limitations on well design, well placement, and equipment performance that together were intended to limit the use of this method to those conditions under which rapid detection could be ensured.

Most commenters supported ground-water monitoring as a release detection method in common use that has successfully detected leaks. One commenter expressed reservations about the method, saying that released product can migrate away from monitoring wells around an UST system and claimed some problems with Florida's program. EPA continues to acknowledge that the method is not completely risk-free and also believes that it provides a level of protection equivalent to the other allowed release detection methods. In addition, not allowing ground-water monitoring in the rule would force states like Florida that depend primarily on ground-water monitoring to completely revamp their programs, thus disrupting established and effective programs, delaying implementation, and unnecessarily increasing expenditures because of the replacement of all the existing wells with other equipment. Therefore, in consideration of the limitations discussed below, EPA has retained ground-water monitoring as an option in the final rule.

(b) Limitations. The limits on well placement in the proposal restricted this method to areas with the water table 20 feet or less below the surface and with soils having a hydraulic conductivity of at least 0.01 cm/sec. Also, the wells had to be placed within the excavation zone. These restrictions were intended to ensure rapid detection by minimizing the distance that a release must move between the UST and the monitoring point and minimizing the time taken to move that distance. Ground-water monitoring was also limited to use with products that are immiscible in water and lighter than water so the product can be detected by the monitors.

o Depth to Ground Water

Many commenters believed that the proposed 20-foot maximum depth-to-ground-water restriction was too stringent and requested that this maximum value be increased or even deleted entirely from the final rule. Others agreed with the proposed rule. The Agency still believes, however, that increasing the allowable maximum depth would increase the volume of the release that could occur before detection. Detection of releases is also slower and less certain with deeper wells because subsurface geology can inadvertently direct product away from the monitoring wells, even if they are located close to the UST system. EPA research conducted since the proposal suggests that 20 feet is the maximum depth to ground water that will permit detection in 30 days when the hydraulic conductivity is 0.01 cm/sec (Notice of Availability; 53 FR 10403). Also, once a release is detected by a deeper well, corrective action will be more difficult and costly because more product has been released. Therefore, EPA has retained in the final rule the 20-foot depth-to-ground-water restriction when ground-water monitoring is used as the sole release detection method. If this method is used as a supplemental device to another approved form of leak detection or as a release investigation method, EPA encourages the use of these wells at greater depths or in less permeable soils.

o Placement of Monitoring Wells

Commenters also objected to the proposed requirement that the monitoring wells intercept the excavation zone. One commenter noted that this requirement could result in existing tanks being punctured during drilling of the well. Another commenter said such a requirement violates some existing state laws intended to prevent contamination. The Agency shares this concern as one expressed by numerous regulators in the field. Apparently, not all owners and operators of existing USTs know the orientation and dimensions of their UST system, not all well drillers have the equipment to find the exact location of their UST, and errors in well installation can occur even when the placement of the UST is known.

Although a slight increase in the allowable distance between the UST and the well may result in slightly larger releases before detection, EPA decided that this was preferable to the catastrophic release that would occur if a tank were punctured. Today's final rule has been revised to require that the monitoring wells or devices be placed within the excavation zone or as close to the excavation zone as is technically feasible. This change also applies to new tanks although placing the wells in the excavation zone should rarely be technically infeasible.

The soil hydraulic conductivity limitation is retained in the final rule and will have to be met whether the well is placed outside or inside the excavation zone. Thus, the revision on well placement is not a major change because wells are allowed outside the excavation zone only when the well is in close proximity to the UST system, the soil is very porous, and released product can move quickly to the well. If the soil outside the excavation zone cannot meet the conductivity requirement, the well must be placed inside the zone.

o Immiscible in Water

The proposed rule required that the regulated substance be immiscible in water and have a specific gravity less than one in order to use ground-water monitoring. These requirements were needed to ensure that released product would float on top of the water table, where it could be detected by the monitors. The Agency has retained these requirements in the final rule but notes that ground-water monitoring is intended for use with gasoline and other substances that are, in fact, slightly soluble in water. Thus, the immiscibility requirement does not exclude substances which are in fact slightly soluble. The slight solubility will not interfere with rapid detection because most of the product is still floating on top of the water table where the monitor can sense it. For example, gasoline has been successfully detected by ground-water monitoring in state programs such as Florida's.

(c) Design Specifications. In the proposed rule, EPA did not include any design specifications for the monitoring wells and networks because site conditions vary widely throughout the U.S., and the Agency wanted to allow the implementing agencies and owners and operators as much flexibility as possible in designing the wells and monitoring network to fit the site. Commenters suggested that the rule should contain more specific requirements and restrictions. EPA continues to believe that tailoring the wells and the network to the specific conditions will result in better release detection, so few specific requirements can (or should) be included at the national level. In addition, most states, such as Florida, that are relying on ground-water monitoring as the preferred release detection method have already included detailed design specifications in their UST programs.

However, upon further review of the proposed limitations and capabilities on ground-water monitoring and the experience using this method in Florida, and other areas, the Agency did decide that three well-design criteria should be added to the final rule to prevent common problems. First, if the top of the water table is above or below the screened interval of the well, then the free product floating on top of the water table will not be able to enter the well and be detected by the monitor. For this reason, today's final rule requires that the monitoring well screen must allow entry of regulated substance into the well under both high and low ground-water conditions. Second, the final rule requires that screening be designed to prevent migration of soil or filter pack into the well, which would clog the screen and prevent product from entering and being detected. The third criterion added to the final rule is the requirement that the wells be sealed from the surface to the top of the filter pack, which will prevent possible contamination by hydrocarbons washed from the surface by rain water that might cause a false alarm or mask a future release.

(d) Sensitivity of Monitor. The last limitation included in the proposed rule to ensure rapid detection using ground-water monitoring was a performance standard requiring that the monitoring equipment be capable of detecting the presence of at least one-eighth of an inch of free product on top of the ground water. This value was selected because it is the maximum performance that manufacturers continue to claim can be achieved by existing automated monitoring equipment. This requirement was intended to apply both to automated and manual monitoring techniques. Some commenters wanted to make the standard more stringent by reducing the criterion of one-eighth of an inch of free product in the well to one-sixteenth of an inch or by replacing it with nonquantitative terms such as detection by "sheen" or by human sight or smell. The commenters felt that the Agency did not give enough justification for selecting the 1/8-inch value and that it is essentially defining the volume of an acceptable release.

The Agency reiterates that the 1/8-inch requirement was selected as the performance standard because it is the sensitivity of existing automated equipment, not because it is an acceptable release. A preliminary EPA analysis indicates that several commercial devices can detect 1/8 inch of product on top of a water table (Notice of Availability; 53 FR 10403). To set a performance standard that is more stringent than can be met by existing technology would eliminate use of this method, which has proven effective in several local UST programs. The Agency considered allowing only manual methods of collecting and analyzing ground-water samples, which may be more sensitive than automated monitors; however, manual methods are very subjective and can only be conducted intermittently, whereas automated methods can be continuous and are less subjective. Therefore, today's final rule retains the 1/8-inch performance standard, and both manual and automated monitoring are acceptable.

(7) Interstitial Monitoring (§ 280.43(g)). Interception barriers and interstitial monitoring were two methods of release detection allowed in the proposed rule. Because they are two distinct methods used to detect releases, they were treated separately. These methods and the requirements that were proposed for them were discussed in detail in the preamble to the proposed rule (52 FR 12735-12739). Commenters were in favor of allowing these methods but suggested changes to some of the requirements. In response to comments on several technical issues raised in the proposal concerning both methods, EPA has changed some of the technical requirements. Based on these comments, the final rule has consolidated the requirements for both methods into one section, eliminated the use of soil/clay liners, and added a requirement to prevent interference with effective cathodic protection.

Section 280.41 of the proposed rule allowed monitoring between an UST and two types of impermeable barriers as two separate release detection methods. The first method, proposed in § 280.41(f), allowed monitoring for liquids in the unsaturated zone between an UST and an interception barrier immediately below it. Interception barriers are basically partial excavation zone liners: they are located immediately below the UST and come only part of the way up the sides of the excavation pit. The second method, proposed in § 280.41(h), allowed interstitial monitoring between an UST and a secondary barrier that surrounds the entire UST system. These barriers are not the partial, catch basin-type of barriers allowed under the first method. These barriers are either integral to the tank system design itself (e.g., double-walled tanks or pipes) or they are located within the UST excavation area along the bottom and sides of the pit and present a barrier between all parts of the UST system and the environment (e.g., flexible membrane pit liners).

Using either of the above barrier-type methods, the interstitial space between the tanks and the barrier can be monitored by a variety of devices designed to detect a variety of changes in operating conditions (e.g., pressure changes with double-walled tanks or presence of liquid or gaseous product in the interstitial space between the barrier and the UST system). Because improper design can make these systems ineffective, the proposed rule included a number of design criteria to ensure effectiveness as a release detection method.

(a) Consolidation of Sections. The proposed rule included a set of general performance standards and design limitations for each method to ensure effective detection of released product. The requirements for the two methods were essentially the same. Despite their similarity, they were included separately in the proposed rule to make it clear that they are two distinct proposed methods and that both are acceptable means of detecting releases. These two methods were not intended to prevent releases, but were intended to contain releases long enough to direct the regulated substance to a monitor for detection; EPA intended that release prevention be covered in the UST design and installation sections of the rule (52 FR 12735-12739).

Today's final rule has been reorganized so that the performance requirements for both of these methods are discussed in a single section (§ 280.43(g)). The Agency has decided to eliminate the separate and duplicative sections in the proposed rule on interception barriers and interstitial monitoring because this appeared to be a source of confusion to some commenters. This is only an organizational change, not a deletion of a possible release detection method, and does not change the substantive intent of the proposal. The consolidated design and performance limitations for both methods remain as proposed with the exception of the changes noted below.

(b) Performance Standard. The proposed rule required that the interstitial monitor between an UST and a secondary barrier be capable of detecting any release from the UST into the interstitial area. One commenter objected to this wording because it requires that the interstitial monitor would be capable of detecting any release, no matter how small. The Agency disagrees with the commenter and believes that interstitial monitors should be capable of detecting a release into the interstitial area. The available data on monitor performance indicates that they are very sensitive and will, in fact, be able to detect the type and size of release that is likely to occur from a secondarily contained UST system. Ideally, EPA agrees with the commenter that a performance standard should be included in the rule to define the sensitivity of interstitial monitors. The data are insufficient, however, to determine a performance standard. Secondary containment with interstitial monitoring is a very sensitive release detection method and is believed to provide maximum protection of human health and the environment, and the Agency did not want to eliminate it from the rule for lack of a specific performance standard. Therefore, the final rule continues to require that interstitial monitors be capable of detecting a leak from any portion of the tank that routinely contains product.

(c) Soil/Clay Liners. The Agency solicited comments on the performance of barriers for purposes of UST release detection (52 FR 12736, 12739). The subject the Agency received the most comment on was the question of the use of soil/clay liners. Some commenters approved of the use of soil/clay liners but suggested that these liners needed more stringent limitations. Other commenters recommended that these liners not be allowed in the final rule because they are not impermeable to all gasoline constituents. Recently completed studies by EPA's Office of Solid Waste on the effectiveness of soil/clay liners compared to synthetic liners (Notice of Availability, 53 FR 10403), as well as information submitted by commenters on the proposal, have caused the Agency to delete this proposed technical option from today's final rule. This means that barriers constructed from native soils or artificially treated soils (for example, bentonite-sealed soils) are excluded from use under the revised performance requirements for barriers. In general, soil/clay barriers are not being allowed because there is enough evidence about the inadequate performance of these materials as reliable barriers to question their reliability for release detection purposes.

(d) Interference with Cathodic Protection. Several other commenters noted that barriers that completely line the excavation might interfere with the cathodic protection system. For example, flexible membrane barriers are usually non-conductive and could electrically isolate the anodes from the tank system, preventing the flow of protective current. The Agency agrees that this could be a problem although no failures of this type have been reported. A general requirement has, therefore, been added to the rule, stipulating that barriers must not interfere with cathodic protection (§ 280.43(g)). EPA believes this can be met in most cases simply by ensuring that the components of the protection system are placed inside the barrier system.

(e) Compatibility of Liner with Product. The proposed rule contained a requirement that the secondary barrier be compatible with the regulated substance to prevent the product from eroding the integrity of the liner over time, causing holes and possible releases to the environment (52 FR 12735, 12736, 12739). The Agency agrees with commenters who noted that a small amount of liner deterioration is inevitable. Accordingly, the wording of the proposed compatibility performance requirement in § 280.43(g) has been changed to indicate that some deterioration is permissible as long as it does not prevent the detection of a release. This requirement was revised to ensure that a basic level of compatibility is achieved and to make the owners and operators responsible for ensuring barrier materials (e.g., flexible membrane liners) are compatible with the product stored (see also § 280.32).

(8) Other Methods of Release Detection (§ 280.43(h)). As discussed in the preamble to the proposal (52 FR 12739-12740), EPA has identified over 250 commercially available release detection devices. The Agency continues to believe that methods of detection other than the seven general methods listed in the rule may also be able to successfully detect releases under certain circumstances. Thus, the proposed rule allowed the use of other methods of release detection if they were approved by the implementing agency as no less stringent than one of the other methods listed in the rule. Commenters generally preferred that any mechanism for approval of a new method be at the federal level (discussed in more detail below). The final rule, however, retains the approval mechanism as proposed but provides an additional mechanism for allowance of other methods. A new method may be used if it can detect a release of 0.2 gallon per hour or 150 gallons within a month with a PD of 95 percent and a PFA of 5 percent. The Agency felt that adding the second alternative mechanism, one which is self-implementing, will provide consistency among methods, offer additional flexibility for owners and operators to choose new or improved technologies of equivalent protection to those specifically allowed in the rule, and spur innovation.

(a) Other Methods Approved by Implementing Agencies. The specific methods EPA proposed have demonstrated effectiveness in the field and are already in extensive use. An important purpose for including these methods was to make it clear that their use was allowed for meeting the proposed release detection requirements and under what conditions they could be used. The Agency intends to continue to develop and provide information helpful to the implementing agencies in evaluating new methods. EPA will also continue to foster identification and development of new methods. Although the Agency wants to allow new methods, it was also concerned that, to protect human health and the environment, they be limited to the methods that are at least as stringent as the methods proposed. Therefore, the proposed rule included a mechanism to allow the use of a new release detection method if the owner or operator could demonstrate to the implementing agency that the method could detect releases before they migrated beyond the excavation zone as effectively as one of the methods already in the rule.

Generally, commenters agreed with the need for allowing new methods but most felt it should be a federal approval process conducted by EPA, not by the implementing agency as proposed. The commenters were opposed to delegating approval authority to the implementing agencies because the lengthy and repeated (for each state or local agency) approvals would discourage method development and because state and local officials do not have the knowledge to make these evaluations. All of the commenters felt that the approval mechanism would be more efficient at the federal level, where approval could be granted one time, rather than 50 times, by means of an approved list or a revised regulation.

As stated in the proposal, the Agency wants to foster innovation and development of new release detection methods and to allow them to be implemented quickly (52 FR 12739). Therefore, the Agency is concerned that the development and publication of a federal list of approved methods or a revised regulation, as suggested by some commenters, would take too long. The Agency could decide, at some future time, to revise the final regulation to add new general methods. The Agency is convinced, however, that allowing approval by the implementing agency, including those at the state and local level, will enable a new method to be used more quickly because the implementing agencies would not have to wait for a federal approval before a method could be implemented. In addition, the precedent set when a new method passes an evaluation in one implementing agency should facilitate succeeding reviews by other agencies.

Implementing agencies are developing UST programs quickly, and the Agency's primary concern is to meet their needs as rapidly as possible. The Agency's research on release detection methods will provide important information to state and local agencies for use in their decisions on which release detection methods to allow. The Agency will continue to encourage private sector evaluation of new release detection methods and the exchange of this information with the implementing agencies. In addition, as discussed below, the Agency is providing another, self-implementing alternative for use of methods not explicitly included in the rule.

In addition to the reasons given above, a federal approval listing process or a revised regulation would not ease some of the problems that commenters foresee with delegating approval to the implementing agencies. EPA's inclusion of methods in the final rule, or its subsequent endorsement of a new method, does not automatically make the method acceptable in a state because states or local governments may, under their own authority, impose release detection requirements more stringent than EPA's. Each state can review each method and decide whether or not to allow it. In fact, a number of state and local agencies are already implementing their own UST programs, and some of these programs have more restricted lists of approved methods than the federal rule. Thus, the final rule retains the proposed option of approval of other methods by implementing agencies.

The standard for implementing agency approval has been changed in the final rule to make it consistent with other changes in the rule. The revised ground-water monitoring standard no longer requires that the monitoring wells pass through the excavation zone. As discussed in section IV.D.2.d.6. above, the Agency did not believe that this requirement was necessary to protect human health and the environment. The revised method can no longer detect a release before it migrates beyond the excavation zone, making the standard for implementing agency approval meaningless. To retain the consistency between this section and the method requirements, the standard for comparing new and existing methods in the final rule has been revised. The standard now specifies that methods approved by the implementing agency must be as effective as one of the other methods allowed in the rule. Methods are considered to be equivalently protective if they can detect a small release as quickly and reliably as other methods included in the rule. This change is consistent with changes in § 280.42(b)(5)(i) (see section IV.D.2.c.2. of the preamble).

(b) Other Methods That Meet a Performance Standard. The Agency included in the final rule a second mechanism by which a new release detection method can become approved. A new method may be used to meet the release detection requirements if it can be demonstrated to detect a leak rate of 0.2 gallon per hour or 150 gallons within a month with a PD of 95 percent and a PFA of 5 percent. This performance standard for alternative release detection methods contains two equivalent leak rates, and the owner or operator may demonstrate compliance with either format. Although external monitoring methods are capable of detecting very small releases, it is more difficult to demonstrate that they meet a small hourly release rate than a larger, though equivalent, volume. The Agency was concerned that, if only the 0.2 gallon per hour release rate was included in this performance standard, manufacturers of new and effective external monitoring equipment or experimental methods would be discouraged from developing the methods or would be unable to demonstrate compliance to the satisfaction of the implementing agency. As discussed previously, EPA wishes to encourage development of new release detection methods.

Unlike the performance standards for the specific allowable release detection methods, the PD and PFA values for the performance standard for new alternative methods are effective immediately. As discussed above in section IV.D.2.a.5., compliance with the probabilities applicable to all methods is delayed for two years. This delay was included in the final rule to allow manufacturers time to modify existing methods, which are already in wide use, and develop the required documentation of performance while still providing the immediate release detection needs required in the rule. These allowable methods were identified in the rule because they are widely used and expected to work well, often without significant improvements. New alternative methods, however, should be developed from the beginning to meet the most stringent performance requirements. In addition, the Agency was concerned that only requiring a leak detection capability of 0.2 gallon per hour could be interpreted to allow imprecise methods such as inventory control to be used alone for the first 2 years of the program. Methods such as ATGS can already operate almost to the required probabilities of detection and false alarm, whereas inventory control cannot come close to these levels, and to allow its use alone even for 2 years would be harmful to human health and the environment.

The evidence gathered by EPA from laboratory evaluations and field experience indicates that the methods specifically proposed (except inventory control) should be able to meet this performance standard now or in the near future (see discussions above for each method). As new methods use this mechanism to become approved, this will ensure consistency of performance among new methods. The net effect of including this alternative in the final rule is to move closer to the general performance standard for all methods considered desirable by many commenters (see discussion in section IV.D.2.a.4. above).

The addition of this alternative will have the effect desired by commenters and EPA of spurring innovation and development of new technology for release detection because there is now a specific and measurable goal for manufacturers to work towards. In particular, this approach will provide flexibility to develop new release detection methods for unusual UST systems such as bulk tanks, for which current methods are inappropriate or expensive. This approval mechanism will have the additional advantage of allowing a new method proven to meet the standard to be used without any approvals in states which allow this approval mechanism.

This approach also clarifies what minimum equivalent performance must be demonstrated to the implementing agency under the other approval procedure for new methods (see preceding section). EPA did not, however, want to make this performance standard the only means by which a new method could become approved because it may not be possible to easily determine a leak rate for some methods within the next few years, particularly external ones. Such methods can still be very effective at detecting releases and the Agency wants to encourage the development of sensitive methods. If method developers can demonstrate to the implementing agencies the sensitivity of their methods in ways other than leak rates, then they should be able to do so. For these reasons, the performance standard approach to approving new methods is included in the final rule in addition to, not in place of, the proposed mechanism requiring review by the implementing agency.

e. Methods of Release Detection for Piping (§ 280.44). The general release detection requirements for pressure and suction lines are discussed in section IV.D.2.b.2. of this preamble. This section discusses the performance standards for those required release detection methods.

One commenter noted that piping and tank release detection methods should be separated because not all tank methods apply to piping and vice versa. A separate section addressing piping release detection methods has been added to the final rule to address these concerns. Separating the methods for tanks and piping allows owners and operators greater flexibility in designing a system. For example, at a station with extensive piping, installing flow restrictors and conducting an annual tightness test for the piping and using vapor monitoring for the tanks may better protect the environment and cost less than installing vapor monitoring for both the tanks and piping.

The proposed rule required either continuous monitoring devices or automatic shutoff devices on all pressurized lines as well as a line tightness test in conjunction with scheduled tank tightness tests. The proposed rule established a leak rate for the automatic shutoff device and indirectly required line tests to meet the tank test standard (0.1 gallon per hour with a PD of 99 percent and a PFA of 1 percent) but set no other performance standards. The Agency requested comment regarding the field performance of pressurized piping release detection methods (52 FR 12744). Commenters noted that additional performance parameters should be provided in the rule, such as detection limits and line operating characteristics. The Agency agrees and has accordingly added further specifications to the piping release detection methods allowed in the final rule to ensure that they meet these minimum performance standards. Thus, probabilities of detection and false alarm have been added for the automatic line leak detectors and line tightness tests; and (for reasons explained earlier) the effective date of these probabilities is delayed for 2 years. Because leak rates depend on the pressure in the line, the Agency agreed with commenters suggesting that the minimum performance standards for line leak detectors, which operate by detecting changes in line pressure, should be specified in terms of the line operating pressure. Each of the piping release detection methods is discussed below.

(1) Automatic Line Leak Detector (§ 280.44(a)). The proposed rule required that the automatic shutoff device be capable of detecting and shutting off a release of at least 2 gallons per hour. This value was selected based on manufacturers' claims. Most commenters felt that the performance standard suggested in the supplemental notice (52 FR 48638) of 0.1 gallon per hour with a PD of 99 percent and a PFA of 1 percent was too stringent, and that 2 gallons per hour was below the detection level of flow restrictors.

The performance standard in the final rule for automatic piping release detection methods (including flow restrictors, shutoff devices, and interstitial or external monitors) has been set at 3 gallons per hour at 10 psi with a PD of 95 percent and a PFA of 5 percent. The 3 gallons per hour value and the probabilities were selected based on a study conducted by EPA's Office of Research and Development of the behavior of pressurized lines, an evaluation performed and submitted by a commenter, and manufacturers' written claims. The value of 10 psi was also selected because it is the pressure at which a typical line leak detector operates. A manufacturer can test a device at any convenient operating pressure and mathematically convert the results to 10 psi to determine if the device meets the performance standard. As discussed elsewhere in today's preamble (section IV.D.2.a.5.), the effective date of the PD and PFA is delayed for 2 years following promulgation.

The final rule also requires that an automatic line leak detector be capable of checking for releases hourly and either restrict or shut off flow of product or be equipped with an audible or visual alarm. The Agency intends the term automatic line leak detector to include a wide variety of devices that meet the standard including automatic shutoff devices, automatic flow restrictors, continuous interstitial monitors, continuous vapor monitors, or continuous ground-water monitors. The hourly detection frequency was selected because pressurized lines can release large volumes of product quickly, so very frequent monitoring is necessary during operation to protect human health and the environment. The equipment currently on the market either operates continuously or conducts a test each time the pump is turned on to dispense product, provided several minutes have elapsed since the previous dispensing, so meeting this requirement should not be difficult. The Agency believes the operators must be alerted immediately to the presence of leaks in pressurized lines. To do this, a clear indication such as flow restriction or shutoff or an alarm is considered necessary.

The final rule also contains the requirement that all automatic line leak detectors be checked annually according to manufacturer's requirements. This requirement was added in response to commenters' concern that line leak detectors can malfunction or be overridden by unwise operators. The possible burden of an annual maintenance check is outweighed by the importance of detecting and stopping pressurized releases.

(2) Line Tightness Test (§ 280.44(b)). The line tightness test that is required in the final rule annually for pressurized piping and every 3 years for American-style suction systems is part of the tightness test option that was proposed for the entire UST system. Work conducted at EPA's test laboratory in Edison, NJ, has demonstrated that line tightness test methods should be able to meet a performance standard of 0.1 gallon per hour with a PD of 95 percent and a PFA of 5 percent with, perhaps, some minor modifications in procedure and equipment (see section IV.D.2.b.2.). Therefore, this performance standard has been adopted in the final rule. As discussed above and in section IV.D.2.a.5., application of the performance standard for line tightness testing has been delayed for 2 years.

As discussed above, the performance standards for line release detection must be stated in terms of the line operating pressure. The value of 1.5 times the operating pressure was selected for the line tightness test because most operators are currently performing tightness tests at this pressure, it is the procedure recommended by NFPA 329 for hydrostatic testing, and it covers the range of line operating pressures, including suction lines. It should be noted that, for safety reasons, all line tightness tests should be performed at positive pressure, not a vacuum, even for suction lines. For example, most suction lines operate at 3 to 5 psi negative pressure; therefore, tightness tests should be conducted at about 7 psi positive pressure.

(3) Applicable Tank Methods (§ 280.44(c)). In the proposed rule, six categories of tank release detection were allowed to meet the monitoring requirement for the "UST system," which included the associated underground piping. As discussed in section IV.D.1.a. above, the final rule now separates the release detection methods for tanks and piping because not all tank release detection methods can be used for piping and vice versa. As noted by commenters, some of the tank monitoring methods are in fact applicable to piping, such as vapor monitoring, groundwater monitoring, and interstitial monitoring, and the Agency wanted to include their use as an option for piping release detection. Therefore, the final rule allows monthly monitoring with one of the applicable tank monitoring methods if it is capable of detecting a release from the portion of the underground piping routinely containing product and meets the restrictions applicable to the use of those methods. This is one of the options for the monthly release detection requirement, in addition to the automatic line leak detector requirement.

f. Recordkeeping (§ 280.45). The proposed rule required that all UST system owners and operators maintain records on the release detection systems required in the rule. The requirement to keep records of performance claims, test results, and equipment maintenance was included because of the importance of each of these activities in the successful detection of releases and in demonstrating compliance to the implementing agency. Commenters generally felt that the requirements were too burdensome and would be particularly difficult to achieve if the testing was done by a service company. The main areas of concern were the requirement to keep performance claims and the components of an adequate performance claim.

Today's final rule retains the recordkeeping requirements as proposed with two revisions. First, only release detection equipment permanently located on-site must have written documentation of calibration, maintenance, and repair on file. Second, manufacturers' schedules of calibration and maintenance for release detection equipment must be retained for 5 years from the date of installation.

The Agency required in the proposed rule that all UST system owners and operators maintain three types of records demonstrating compliance with the applicable release detection requirements: documentation of method performance monitoring results; and general operation, maintenance, and repair. It was felt that these records demonstrate that certain past events important to effective release detection using that method actually took place and could be used by implementing agencies to determine compliance (52 FR 12747). In general, commenters felt that the proposed recordkeeping requirements as a whole were too burdensome. The Agency believes, however, that the requirements are not particularly burdensome because many of the records will be supplied by manufacturers, sales personnel, or service people; not much paperwork is involved and the required records would generally be kept on file anyway; and paperwork will be added infrequently. In general, some records are needed to remind the owners and operators when maintenance is scheduled and to help them keep the equipment under warranty. Finally, properly maintained records are necessary to allow later analysis of the tank systems by either the owners and operators or the implementing agency in the event of a release investigation or system closure. Each of the individual requirements receiving comment or revised in the final rule is discussed below.

The requirement to maintain manufacturers' performance claims and justifications for 5 years was intended to encourage manufacturers to evaluate their equipment and develop documentation of the proof of performance and to cause owners and operators to review this information while selecting an appropriate release detection method. Many commenters expressed confusion over what these performance claims would be and were concerned that the owners and operators were being required to substantiate the claims. It was not the intent of the proposed or final rule to require the owners and operators to provide the proof of performance claims, only for them to ask for and acquire information from the manufacturer. In order to compete and successfully market equipment, the manufacturer will have to develop convincing documentation demonstrating that the release detection method meets the minimum performance requirements.

As discussed in the proposal preamble (52 FR 12719), there are several types of information that the manufacturer should include in this documentation and that the owners and operators should look for. The final rule does not require any specific information, however. EPA recognizes that the level of detail will vary by type of method and that, over time, manufacturers will develop standardized claims that will help guide the owners and operators. This requirement will place owners and operators in the position of having to review the claims and select a system that will meet regulatory requirements. Thus, owners and operators are responsible for achieving the goal of effective release detection and demonstrating to the implementing agency that the owners and operators have made an effort to comply with the regulation. For these reasons, this section of the final rule will remain unchanged from the proposal.

The proposed rule required that all records of calibration, maintenance, and repair be maintained by all owners and operators for at least one year. These are important procedures for the proper functioning of release detection equipment, particularly automated systems, and must be available for inspection to demonstrate that the system is working as well as it can. One commenter noted that the recordkeeping requirements would be difficult to meet for owners and operators who hire release detection services to do the monitoring. EPA agrees with this, and today's final rule states that this requirement applies only to owners and operators of equipment permanently located on-site. Owners and operators who hire release detection services need not retain the servicing records of their contractors. Owners and operators, however, must ensure that the service is performed well and according to specifications and will be responsible for the cleanup of any undetected releases.

Another addition to today's final rule that was not in the proposed rule is the requirement that any schedule of required calibration and maintenance provided by the release detection equipment manufacturer must be retained for 5 years from the date of installation. This information should be provided by most manufacturers in the same brochures as performance claims; thus, this new requirement should not result in much additional paperwork. This requirement was added as a clarification to the final rule to remind owners and operators of the importance of maintaining these data. With this requirement, implementing agencies can verify that required maintenance and calibration were performed. Also, if ownership of a site is transferred, the new owner or operator will be able to understand and ensure that the proper schedule is maintained.

g. Other Release Detection Issues -- (1) Release Detection Variances for Low-Risk Sites. In the Supplemental Notice (52 FR 48641), EPA noted that protected tanks in some areas of the country may not require frequent-to-continuous release detection. EPA requested comments on whether, within these less sensitive areas, variances should be allowed for protected tanks to conduct internal inspections or less frequent monitoring. EPA requested comment on the appropriateness of these alternative release detection approaches and under what conditions they might be used. There was little agreement among commenters on either issue, and no specific recommendations on how low-risk areas could be defined or identified at the federal level. The final rule does not include a release detection variance for low-risk areas because of the difficulty in operationally defining and implementing a variance procedure which adequately protects human health and the environment.

Commenters who supported the use of the variance felt that it would improve environmental protection by focusing resources on UST systems in higher risk areas. Commenters opposed to the variance noted that leaks can pose environmental problems even in low-risk areas. Their experience suggested that impermeable formations may actually have hidden fractures that allow product to reach deep ground water and necessitating expensive cleanup efforts. They further suggested that, under any circumstance, product leaks may pose an explosion hazard. These commenters also noted the difficulty of managing a variance process.

The Agency agrees with commenters who noted significant problems with a variance procedure. No commenters were able to suggest a manageable variance procedure given the size of the UST universe. As noted earlier in the preamble, defining sensitive or low-risk areas at the federal level is problematic (see section III. of today's preamble). Because the Agency could not develop reasonable criteria for granting variances on a case-by-case or prospective basis, no release detection variance is included in the final rule. Consequently, less protective detection options such as infrequent monitoring for protected tanks over 10 years of age will not be permitted in any areas regardless of risk.

(2) Internal Inspection as an Alternative to Release Detection. The Agency also considered the use of internal inspections as a substitute for release detection. EPA solicited comment in the December 23, 1987, supplement to the proposal on the use of internal inspections at low-risk UST sites. The Agency received many comments both for and against this option. In general, supporters of internal inspections believe that this would provide important information on the internal and external corrosion and the structural integrity of tanks. Some commenters specifically recommended allowing internal inspection as a release detection alternative at low-risk sites (e.g., sites that are not in vulnerable ground-water supply areas or in close proximity to surface waters and residential areas) while others supported its use at all UST sites. Schedules for conducting internal inspections, such as at periodic intervals (e.g., every 3 to 5 years; or ever 10 to 15 years) or based on tank age, were recommended by several commenters.

Reasons given by commenters not supporting this option included: (1) not all tanks are constructed with manways, (2) internal inspections are time consuming and would be cost effective only for larger (bulk) tanks, (3) internal inspections do not provide sufficient information on tank integrity, and (4) inspections are a safety risk to the inspector.

The Agency does not have sufficient data on the performance of tanks subject to internal inspection programs that would allow it to determine that such an approach without release detection would be protective of human health and the environment. The Agency is aware that internal inspections are widely used by owners and operators of bulk tanks for evaluating tank integrity. Based on insufficient information on the use of internal inspections with all USTs and the lack of industry consensus codes, however, the Agency has decided not to include this as a release detection alternative in the final rule.

[ Back to Section IV.C. ][ Ahead to Section IV.E. ]

Top of page


Local Navigation


Jump to main content.