3

Available Empirical Indicators of Offshore Industry Risk Profile

INTRODUCTION

As a way to understand the associated safety and environmental challenges relevant to the changing risk profile in the Gulf of Mexico (GoM), Chapter 3 provides an overview of publicly available data sources for occupational and process safety. The chapter begins by reviewing occupational safety data reported by the Bureau of Safety and Environmental Enforcement (BSEE). Next, both BSEE and industry-related process safety indicators are examined. The last section considers other publicly available BSEE data.

DATA SOURCES AND GAPS

In this section the committee presents an overview of available statistical indicators of offshore industry-wide risk and what they tell us about the offshore industry’s current and past risk profile. We begin with trend data on injuries, which are the longest time-series available, while also acknowledging that occupational injury rates are not indicators of the risk of disasters. We then review available data that could serve as indicators of major accidents, with a focus on incident data and trends that could address systemic risks as the committee defined it in Chapter 1. As is described, the industry and BSEE have made important improvements in collecting and aggregating indicator data since Macondo, but as yet, most trends are short and overall data reporting and sharing by companies are incomplete.

Occupational Safety

Reporting of fatalities, injuries, safety-related incidents, and hours worked by contractors and operators in U.S. waters to the U.S. Department of the Interior’s (DOI’s) BSEE covers nearly all companies operating on the Outer Continental Shelf (OCS) since 2011, and 99 percent or more since 2014 (BSEE, 2021). Data on fatalities and injuries were collected before 2010, but apparently, prior to this, reporting of work hours was inconsistent or incomplete. As a point of comparison of occupational risk, and based on Figure 1-1, the American Petroleum Institute (API) reports that occupational injuries and illnesses per 100 full-time workers in the oil and gas industry are considerably and consistently below that of private-sector employment in comparable industries (API, 2021). The occupational injury rate for the offshore industry is only one-quarter the rate of that in the mining industry. However, although the total reportable injury rate is commonly believed to be associated with the risk of fatalities, a reliable metric for comparing company safety records and predictive of major accidents, it has not been found to be a reliable indicator of any of these for the construction industry (Hollowell et al., 2020).

Fatalities

For OCS production, drilling, and decommissioning operations, almost all of which occur in the GoM, BSEE data report two fatalities in 2021, six fatalities in 2020, and six fatalities in 2019.¹ Between 2014 and 2018, fatalities averaged one per year, while between the years 2011 and 2013, fatalities averaged 2 to 3.² Although BSEE reports these fatalities, the investigation can be under the jurisdiction of other federal agencies, such as the U.S. Coast Guard or the Federal Aviation Administration, and can be reported in different databases.

Injuries

The lost-time injury rate for all injuries has declined considerably over the past 10 years. While 2020 and 2021 saw a slight uptick because of reported COVID-19 cases, total reportable injury rate would have decreased by 17 percent, near historical levels (see Figure 3-1). Trends in injuries separated by production, drilling, and decommissioning operations follow a similar pattern.

___________________

¹ See Offshore Incident Statistics, https://www.bsee.gov/stats-facts/offshore-incident-statistics.

² Ibid.

BSEE reports incidents aggregately and makes recent spreadsheets of incident data available online.³ Although the raw data list operator and facility name, the spreadsheets do not include structure identification or personally identifiable information. BSEE reports that the agency does track more detailed incident data internally but does not make this information public.⁴

Regardless of improving trends in occupational safety, major offshore safety failures are not predictable based on the frequency of such events. There was a spate of major disasters in the late 1970s and 1980s: the Ixtoc 1 well blowout and major spill in Mexico’s Bay of Campeche in 1970; the capsizing of the Alexander Kielland and the death of 123 people in the North Sea in 1980; the sinking of the Ocean Ranger off Newfoundland and the death of 84 people in 1982; the explosion and fire on the Piper Alpha and the deaths of 167 people in the North Sea in 1988; and the South Pass Block 60 (B platform) accident in the GoM with 7 fatalities in 1989 (NASEM, 2016). After these events, there was a gap of 20 years before the Montara blowout and major oil spill off Australia in 2009 and the 2010 blowout and explosion on the Deepwater Horizon. Occupational safety trends, as reviewed above, are not believed by experts to be indicative

___________________

³ See Offshore Incident Statistics, https://www.bsee.gov/stats-facts/offshore-incident-statistics.

⁴ J. Mathews, BSEE, presentation to the committee, June 2021.

of risks of disasters such as major blowouts, explosions, fires, and spills from offshore wells (Hudson, 2009). Therefore, in managing risks, a better approach is to focus on indicators of how well companies are managing processes to prevent major accidents.

Process Safety Indicators

Trends in process safety incidents (indicators of a barrier or control failure) should be better indicators of high-risk events and therefore may serve as components of systemic risk metrics. In the following subsections, we review available data from both governmental and industry sources.

BSEE Process Safety Indicators

In addition to safety-related incidents, BSEE collects and reports statistics on events that could have led to more significant impacts, such as fires and loss of well control (LOWC). Although generally considered lagging indicators, most of the incidents reported are minor and potentially indicative of major events. Review follows of illustrative examples that are drawn from recent BSEE reports. On what has been reported on an annual basis from 2009 to 2021, the United States averaged 3 fatalities, 211 injuries, and 20 spills of 1 or more barrels of oil from offshore incidents, including the Deepwater Horizon disaster in 2010 (see Table 3-1).

Trends in the fire incident rate do not suggest significant improvement over time (see Figure 3-2). BSEE, however, interprets the increase of fire incident rate since 2017 as driven by anticipated and controlled fires during hot work and living-quarter or galley electrical and grease fires (BSEE, 2020a, p. 4), or by evidence of a reportable fire that can include soot, charring, melting, and smell (as clarified by NTL 2019-05) (BSEE, 2022b, p. 18; Noem, 2019).

The rate of LOWC events fell markedly to zero between 2013 and 2017, but in 2021 has since risen to almost twice the previous highpoint rate in 2013 (see Figure 3-3).

In addition to keeping fairly complete reports on inspections and issuance of Incidents of Noncompliance (INCs), BSEE reports INCs per inspections performed for both well operations and production. Figure 3-4 shows an overall decrease in this rate from 2011 to 2021 for well operations.

For production operations (not shown), the trend in INCs per inspection had fallen by more than two-thirds in 2020 since reaching a peak in 2015, before increasing slightly in 2021 (BSEE, 2022b, p. 31). However, the issuance of INCs can shift with changes in policies and administrations.

TABLE 3-1 Number of Incident Types and Consequences on the U.S. OCS Reported by Year

NOTES: bbl = barrel of oil; OCS = Outer Continental Shelf. Data for 2009-2017 were reported by fiscal year. Beginning January 1, 2018, data were reported by calendar year. Data as of February 5, 2023.

^a For FY 2010, the number of fatalities includes the 11 people killed and the estimated total volume of oil spilled (4,928,100 bbls) from the Deepwater Horizon explosion.

^b 2017 4th quarter calendar year only; total volume of oil spills were not provided for 4th quarter.

^c 2018, 2019, and 2020 data reported by calendar year.

^d In accordance with BSEE regulations, all fatalities are required to be reported; however, not all fatalities are within BSEE’s jurisdiction.

^e Includes two non-occupational fatalities.

^f Numbers in the spills column include spills of oil, drilling mud, and other chemicals.

SOURCE: www.bsee.gov/stats-facts/offshore-incident-statistics.

Industry Process Safety Indicators

The Chemical Safety and Hazard Investigation Board (CSB), the American Bureau of Shipping (ABS), and other post-Macondo reports called for improved process safety indicators. Industry has responded with IOGP report 456 (IOGP, 2018) and API Recommended Practice (RP) 754.⁵ Both recommend data reporting of loss of primary containment (LOPC) events regardless of consequence. LOPC events are defined by IOGP in four tiers. Tier 1 and Tier 2 distinguish between events of greater or lesser significance.⁶ As shown in Figure 3-5, below, IOGP considers Tier 1 and Tier 2 process safety events (PSEs) to be primarily lagging indicators. It identifies Tier 3 and 4 PSEs as primarily leading indicators, but these distinctions are not exact.

IOGP Tier 3 and 4 PSEs “are primarily designed for monitoring and review of barriers and management systems that support their performance” IOGP (2018, p. 48) and thereby appear to be potential systemic risk indicators. Tier 3 events are defined as barrier failures that do not result in consequences that could be classified as Tier 1 or 2. Tier 4 measures focus on monitoring or maintaining barriers and take a variety of forms. Tier 3 and 4 PSE measures are not aggregated or benchmarked by IOGP, which views such measures as being derived from, and applicable to, individual

___________________

⁵ See https://www.api.org/products-and-services/standards/important-standards-announcements/754.

⁶ Tiers 1 and 2 have formal definitions involving multiple criteria that are not summarized here (IOGP, 2019, p. 12).

circumstances. Thus, aggregate data on these leading indicators, and their rates and trends, are not available from this source. However, IOGP has recently made a decision to begin reporting and sharing Tier 3 indicators among its members.⁷ IOGP Tier 1-4 PSE indicators could be reported for the U.S. offshore industry to SafeOCS, as described later. Such indicators, as well as other leading indicators discussed below, lack a direct relationship with the probability of system failure. This is due, in part, to the existence of layers of defense that prevent the failure of a single barrier from escalating into a major event. In time, and as indicators improve, they will assist in the development of probabilistic models that can guide design and operational decisions. In the near term, they can trigger deeper investigation and analysis by system managers into potential emerging issues.

Center for Offshore Safety

Since 2013, the Center for Offshore Safety (COS) has collected and reported safety incidents from its members. The COS Annual Performance Report (APR) for 2021 provides trend data for the 2017-2020 period. For 2021, the companies reporting to COS conducted a total of more than 45 million hours worked on the OCS, which is an increase from the 34 million hours

___________________

⁷ O. Skår, IOGP, presentation to the committee, June 2022.

in 2020. The hours worked in 2020 represent more than two-thirds of the hours worked on the OCS in that year; likewise, in 2019, companies reported 44 million hours worked on the OCS, representing about 60 percent of hours worked on the OCS.⁸ COS classifies safety performance indicators (SPIs) into 10 categories. These categories do not represent a severity score but, rather, simply different classifications, some of which include the same events. SPI 1 is defined to include one or more of the following: a fatality; five or more injuries in a single event; a Tier 1 Process Safety Event;⁹ a loss of well control event; $1 million or more in damage costs; or an oil spill in water of greater than or equal to 10,000 gallons. SPI definitions combine occupational injuries with process safety events, which compromises the potential for SPIs as indicators of systemic risk.

SPI 2 events are defined as ones that are less severe than SPI 1 events but still are quite significant. They include one or more of the following: a Tier 2 process safety event;¹⁰ a recordable injury; a collision resulting in greater than $25,000 in damage; a mechanical lifting incident; loss of station-keeping resulting in a drive off or drift off; lifeboat, life raft, or rescue boat event; or a loss of well control involving multiple barrier system failures and challenges. The number and rate of self-reported SPI 1 and SPI 2 incidents are shown in Figures 3-6 and 3-7.

Over the 2015-2021 period, both SPI 1 events and SPI 2 events decreased in 2021, after trending up over previous years. COS reports the number of events annually within each of the subcategories of SPI 1 and SPI 2, but does not estimate their rates. Notably SPI 1 and SPI 2 events that included equipment failures (SPI 3 events) increased to 35 percent in 2021 after declining to 16 percent in 2020 and reaching a high of almost 60 percent in 2017 (COS, 2021, Figure 4.13, p. 26). Incidents involving cranes or personnel or material handling (SPI 4) decreased in 2021 to a rate of 0.623 after reaching a peak rate of 0.945 events per 200,000 work hours in 2020 (COS, 2021, Figure 4.16, p. 29). Thus, SPI 3 and SPI 4 events appear to fluctuate considerably from year to year.

___________________

⁸ As an example, 34 million hours worked in 2020 (COS, 2021, p. 4) divided by 50.8 million total OCS hours worked as reported on BSEE Form 0131 and summarized in current BSEE’s OCS Performance Measures Incident Rate Charts, https://www.bsee.gov/sites/bsee.gov/files/performance-data-table-2010-2020-12-16-2021.pdf.

⁹ COS uses the API RP 754 definition of a Tier 1 event, which is one that involves an unplanned or uncontrolled loss of containment of any potentially hazardous material that results in death, injury, evacuation, fire or explosion, and discharge from pressure release device. See COS (2021, p. 51).

¹⁰ An API RP 754 Tier 2 event is an unplanned or controlled loss of any potentially hazardous material that is not a Tier 1 event and results in a recordable injury, a fire or explosion causing $25,000 in damage, a pressure release discharge resulting in several different consequences, and release of any material over certain defined thresholds. See COS (2021, p. 51, Appendix 1, SPI Definitions and Metrics).

COS uses self-reported completion of planned critical maintenance, inspection, and testing as a principal process safety leading indicator. For operators, the combined average in 2021 on this metric was 89.2 percent and ranged from 72.4 to 100 percent. The 89.2 percent is a decrease from averages between 2017 and 2021. For contractors, the combined average in 2021 was 98.4 percent and ranged from 95.2 to 100 percent, a slight increase from 2020 (COS, 2021, Figure 4.17, p. 30). As these averages suggest, wider variations are masked within the categories: between 75.9 and 98.6 percent of operators completed planned work on time in 2020 and between 95.3 and 100 percent for contractors in that same year. In 2018, however, the range for contractors was between 60.3 and 100 percent (COS, 2021, Figure 4.14, p. 25).

COS also has a Learning from Incident (LFI) activity that analyzes SPI 1 and SPI 2 events, along with events classified by member companies as “High Value Learning Events” (COS, 2021, pp. 36-40). The LFI process identifies issues as “Areas for Improvement” (AFIs). COS has classified and tracked these LFIs and AFIs since 2013, and reports the most recent 5 years of data in the 2021 annual performance report. The most frequently identified AFIs for 2021 include Operating Procedures or Safe Work Practices (37 percent); Process or Equipment Design or Layout (30 percent); and Quality of Task Planning and Preparation (26 percent) (COS, 2021, Figures 5.4 and 5.5, pp. 39-40). Notably, all of these are organizational management and human performance issues.

SafeOCS

BSEE is funding the development of an aggregate industry incident database through the SafeOCS. This activity is being managed by the Department of Transportation’s Bureau of Transportation Statistics (BTS), which is able, as a federally designated statistical agency, to guarantee the protection of data submitted from disclosure through litigation, state open record laws, or Freedom of Information Act requests. SafeOCS has three confidential data-reporting systems: (1) aggregated industry safety data (ISD), which are voluntarily provided company incident databases including near-miss and other safety data; (2) mandated well control equipment (WCE) data on blowout preventer (BOP) failure reports in drilling and nondrilling operations; and (3) mandated reporting of safety and pollution prevention equipment (SPPE) failures in production operations.

ISD Data

As of May 2022, the ISD program has commitments from 22 companies to participate by sharing their complete databases of incidents, including

16 operators that account for approximately 92 percent of GoM OCS production in 2020 and 87 percent in 2021. The framework for SafeOCS data presentation is well established and disaggregated beyond the SPI 1 and 2 levels used by COS, but there are as yet no risk exposure measures to allow estimation of rates over time. SafeOCS ISD is also notable as an effort to provide an aggregated database of near misses, which could serve as leading indicators. SafeOCS distinguishes near misses between personal and process safety, the latter of which could help in the development of systemic risk indicators. In February 2022, BTS released a public-facing dashboard that displays aggregated industry data submitted between 2018 and 2020.¹¹

In contrast to reporting by the IOGP, the SafeOCS ISD program is structured to collect and aggregate safety incident or indicator data that include Tier 3 and 4 process safety events (BTS, 2019). For SafeOCS, a Tier 3 event is defined based on the event in terms of its potential consequences, such as the opening of a pressure release valve or events exceeding the safe operating range of equipment (BTS, 2019, p. 40). A Tier 4 event is related to SEMS or other operating procedures, such as an audit finding that equipment was not tested on schedule. As noted earlier, the SafeOCS ISD program is still gaining in industry participation, but it does demonstrate the potential of aggregating and trending leading indicators of process safety. The public dashboard cited above shows total PSEs, but they are not yet classified in IOGP tiers. As only the first report, the process safety data summarized in the Phase 1 report is limited to classifications by consequence, type, and relative frequency. More years of data and more complete reporting might begin to reveal trends in systemic risk that could inform risk management practices. Unlike the voluntarily submitted SafeOCS ISD data, the WCE and SPPE data, described next, are required by regulation.

WCE Data

The Well Control Rule (WCR) finalized in 2016 includes requirements that offshore operators report failures of BOP equipment (BTS, 2021).¹² In issuing this regulation, BSEE responded to one of the recommendations from the National Commission on the BP Deepwater Horizon Oil Spill and Offshore Drilling, which recommended that BSEE require more reporting on leading indicators of offshore risk. Operators must report within 30 days and conduct an investigation and analysis of the cause of failure within 120 days. Reporting began in mid-2016, and, for companies exercising the option of reporting through SafeOCS, ensures and protects their confidentiality. (Note that in 2022, BSEE issued proposed changes to the WCR

___________________

¹¹ See https://www.safeocs.gov/sdp/dashboard/isd.

¹² Prior to 2019, the annual reports were titled Blowout Prevention System Safety Events.

that would eliminate the option for operators to report to SafeOCS.¹³) Design changes and changes in procedures following a failure must also be reported. The International Association of Drilling Contractors (IADC) and the IOGP have worked closely with BTS in the development of the reporting system through their Joint Committee on Blowout Prevention. BSEE provides data on well activity. To date, the annual report for 2021 is the most recent annual report (BTS, 2021). The 2021 period saw a decrease in event reporting and well activity (well starts declined 9.5 percent between 2020 and 2021). Highlights are as follows:

• In 2021, 389 WCE failure events were reported compared to 633 for 2020 and 995 for 2019.
• Between 2017 and 2021, SafeOCS received 4,633 failure events, an average of 927 events per year.
• The majority of events were minor; no leaks of wellbore fluids were reported in 2020 and 2021, and only one event was reported over a 5-year period.
• The majority of reports (more than 88 percent from 2017 to 2021) were about subsea WCE system events.
• The vast majority of events (more than 85 percent from 2017 to 2021) were discovered when equipment was not in operation, primarily during routine maintenance, inspection, or testing.
• External leaks were the primary source of discovery, and the most common root causes from 2017 to 2021 were wear and tear (48.1 percent), design issue (16.0 percent), and maintenance error (12.2 percent).

BTS provides a useful summary of total events in terms of escalating risk: Loss-of-containment (LOC) events are the highest risk; failures requiring stacked pulls are the second highest; failures during operation are the third highest; and failures during maintenance and testing are the lowest. BTS reports that operators with more stacked pulls while not in operation tend to have fewer stacked pulls while in operation (a riskier event).

A useful and important component of this reporting system is a requirement for reporting the result of root-cause failure analyses and recommended preventive actions, and BTS follow-up on these recommendations, because these can result in industry-wide changes in BOP performance and operating procedures.¹⁴

___________________

¹³ BSEE’s rationale for this change is that they would receive reports in a timelier way if they were only reported directly to the agency (BSEE, 2022a).

¹⁴ SafeOCS organized root causes into categories. Details about the 10 cases in 2021 that included a root cause failure analysis are reported in BTS (2021, Table 15).

SPPE Data

BSEE also requires OCS operators to report to SafeOCS any failures in fail-safe valves that serve as mechanical barriers to LOC during production. These valves take multiple forms listed in Box 3-1, per 30 CFR § 250.801. To date, four annual compilations of reports have been published, the latest of which is for 2020. BTS has also drawn on a variety of data sources provided by BSEE to construct measures of risk exposure.

Major findings from 2019 and 2020¹⁵ are:

• In 2019, 225 SPPE failures were reported on 184 of 6,029 active wells in the GoM; for 2020, 101 failures were reported on 90 of 5,715 active wells.
• For both 2019 and 2020, more than 92 percent of SPPE failures were on surface wells.
• None of the failures were characterized as health, safety, or environmental incidents.
• For both years, more than 79 percent of failures occurred on low-production wells located in 650 feet (200 meters) or less of water depth.
• In 2020, valve seat degradation was the most common source of failure (67.5 percent) and wear and tear the most common root cause (79.2 percent).
• Repair was the most common form of correction (59.3 percent of failures) in 2020.

An important caveat to the above reports is that BTS estimates from other sources, such as INCs, that almost 35 percent of SPPE failures were

___________________

¹⁵ See BTS (2020, Table 2, p. 12).

not reported to SafeOCS in 2019 and more than 41 percent of failures were not reported to SafeOCS in 2020 (BTS, 2020, Table 2, p. 12).

The 2020 annual report provides additional data on SPPE failure rates to inform risk management but are not summarized herein (BTS, 2020, Table 3, p. 18). BTS is engaged in multiple efforts to improve data definitions, exposure estimates, and reporting by operators. Both BOP and SPPE failure rates and trends over time could be part of measuring industry risk, but little trend data are available, and these metrics focus on heavily on equipment rather than organizational management and behavior. If root cause analyses for BOP and SPPE failures were to trace back to organizational and management issues rather than immediate causes, such reports could serve as useful indicators of systemic risk.

SEMS Audits

In 2010, following the Deepwater Horizon blowout, explosion, and oil spill, the federal government required operators on the OCS to develop, install, and maintain safety and environmental management systems (SEMS) based on API RP 75. API RP 75 had formerly been voluntary but had not been adopted by all OCS operators. The SEMS regulations also required a regular audit process by certified auditors. (COS subsequently established and maintains a process for auditor certification.) Audits are conducted on a three-year cycle and must include 15 percent of a company’s production and well operation assets. A regular and vigorous auditing program and oversight of follow up by companies on audit deficiencies would help ensure that SEMS becomes an effective management process rather than a paperwork exercise. Reports on audit deficiencies and corrective action plans can serve as indicators of the offshore industry’s commitment to reducing the risk of their operations, and BSEE oversight of this process can serve as an indicator of its role in enforcing safety management.

In 2020, BSEE issued a summary of the first three cycles of audits (BSEE, 2020b). Key findings from that summary are:

• Operators in deepwater have fewer SEMS deficiencies than those operating in water depths under 650 feet (200 meters) (average of 5 versus average of 11).
• Roughly half of deficiencies result from incomplete or inconsistent implementation of documented policy, work procedure, or practice.
• Good practices were noted in 40 of the 52 audits performed and represent one-third of audit findings.
• Deficiencies are shifting from those associated with designing and establishing SEMS to those associated with implementation, which

• BSEE takes as an indicator of industry maturity in the SEMS process.
• More than 60 percent of deficiencies were in five SEMS elements: safe work practices, mechanical integrity, hazard analysis, operating procedures, and management of change (MOC).
• Approximately 40 percent of corrective action plans in response to identified deficiencies could be considered simple corrections.
• Depending on the SEMS element, between 35 and 55 percent of mitigation elements were intended to improve future implementation, and between 5 and 20 percent sought to identify root causes of deficiencies (decision making, organizational workflow, oversight, and other human factors).
• The quality of SEMS audits themselves are improving over time (BSEE, 2020b, p. 16).

The first finding suggests a bifurcated risk profile across the industry. Operators in deepwater who have fewer SEMS deficiencies tend to be drilling for and producing oil and gas from high-pressure wells that can have much more significant consequences if blowouts, explosions, and large spills do occur. On the other hand, these wells are typically operated by large multinational firms with long-established safety programs using the latest and most sophisticated equipment. Rigs operating in water depths of 650 feet (200 meters) or less tend to be producing from older low-pressure wells with lesser consequences if blowouts occur, but are also operating on much older equipment, and by smaller operators that may have resource limitations compared to those in deepwater. Thus, process safety risks may be higher in deepwater and occupational risks higher in shallow water.

The BSEE document also identified several areas for improvement in BSEE oversight including of the audit process and thoroughness of corrective action plans. Audit results data are presented in aggregate, but the rigor and consistency of SEMS development and implementation varies across companies. BSEE’s report states that “there are large differences among operating companies in how effectively they establish, implement and maintain their SEMS” (BSEE, 2020b, p. 12). BSEE’s report also notes that additional rulemaking may be required to (a) enhance its oversight of how thoroughly and effectively it monitors and evaluates company follow-up to audit results and (b) ensure that all companies’ follow-through on corrective action plans includes examination of root causes (BSEE, 2020b, p. 16).

Additional BSEE Data

In overseeing offshore operations, BSEE manages and analyzes relevant data through electronic reporting systems such as the Technical Information

Management System (TIMS).¹⁶ Much of the data that are collected and posted regularly on a public-facing website and available for download¹⁷ pertain to information on topics such as operating companies, wells, platforms and rigs, oil and gas production, as well as INCs.¹⁸ For example, BSEE data show that the number of companies holding leases on the OCS and the number of contractors have fluctuated over the last 30 years; but since 2010, the number of companies holding leases has decreased by more than half. In fact, only about half of all operators who hold leases actively produce oil and gas, and that number has declined over the past 12 years, so that by the end of 2021 only about 39 operating companies had active leases with production (BSEE, 2022b).¹⁹

BSEE data provide details on production platforms in the GoM. More than half of all platforms in the GoM have been removed since 2010, the vast majority of which were in water under 500 feet (150 meters) deep. The average age of these remaining shelf platforms at the start of 2022 was more than 37 years, and 73 percent are at least 25 years old (see Table 3-2).

In addition to incident data discussed above, the BSEE website provides aggregate data for offshore activities, the number of inspections, and the total number of offshore work hours performed by operators and contractors between 2010 and 2021.²⁰ As an example of available data, the number of work hours that contractors have performed, on average, is more than 85 percent of all work over this period (see Figure 3-8).

As for the major phases of activity, more than 94 percent of the work related to exploratory and development drilling is performed by contracted workers, compared to 72 percent of production work.

In 2018, BSEE’s GoM region implemented a formal risk-based inspection (RBI) program that supplemented its current annual inspection program. BSEE’s RBI program is conducted through the GoM’s Office of Safety Management (OSM) and consists of two components—facility-based risk inspections and performance-based risk inspections (see NASEM, 2021, pp. 83-96). As part of its RBIs, OSM will include inspection data, findings and deficiencies from SEMS audits, and incident investigations and reports as part of an operator profile. OSM will use this information to monitor operator performance and hold annual reviews to discuss any recurring safety and environmental concerns. Production companies with recurring violations or indicators of unacceptable performance could be subjected to BSEE’s increased oversight list, which could place the company on a

___________________

¹⁶ See https://timsweb.bsee.gov.

¹⁷ See BSEE Data Center, https://www.data.bsee.gov/Main/Default.aspx.

¹⁸ See https://www.data.bsee.gov/Main/RawData.aspx.

¹⁹ See also https://www.data.bsee.gov and NASEM (2021, pp. 33-34).

²⁰ See https://www.bsee.gov/sites/bsee.gov/files/performance-data-table-2010-2021-10-26-2022.pdf.

TABLE 3-2 Total Annual Number of GoM Production Platforms by Age and Water Depth

NOTES: GoM = Gulf of Mexico. Discrepancies in the tables are due to inconsistencies in the raw data, such as missing or incorrect entries. Deepwater was historically defined as having a depth greater than 500 feet (150 meters), as in this table. Over the past couple of decades, deepwater has more often been defined as water depths greater than 1,000 feet (300 meters). The historical definition is used here to maintain consistency with other available statistical data.

SOURCE: https://www.data.bsee.gov/Main/Platform.aspx.

performance improvement plan in order to address systemic environmental and safety concerns. BSEE tracks much of these safety-related data internally and shares the information with the individual company.²¹ As noted above, BSEE does post much of these data (such as INCs by company and

___________________

²¹ J. Mathews, BSEE, presentation to the committee, June 2021.

some incident-related information) on its public-facing website; however, joining and analyzing all of these data by operator and facility are not always possible.

CONCLUSIONS

Conclusion 3-1: Voluntary and mandatory data reporting to SafeOCS demonstrate considerable progress in development of safety indicator data. Companies representing 92 percent of OCS hydrocarbon production are now providing company data voluntarily to the SafeOCS ISD program. Missing, as yet, are estimation of safety incident rates, such as by hours worked, that would normalize trends. Reporting of PSEs into IOGP’s four tiers would increase the availability and value of the data being reported, especially for tiers 3 and 4, which could serve as leading indicators of systemic risks tracing back to organizational management. Mandatory reporting to SafeOCS of well control and production safety system equipment performance indicates that such equipment failures resulting in LOC are infrequent, but there are apparent issues on nonreporting of SPPE data. Improvements in industry data collection, sharing, and reporting to SafeOCS offer promise for future trend analyses of the offshore industry risk profile, but more complete reporting of voluntary data, particularly process safety indicators, will be needed before such empirical information can inform judgments about the industry risk profile.

Conclusion 3-2: Available trends for safety indicators from COS offer a mixed picture. Safety Performance Indicator (SPI) 1 and 2 events both decreased in 2021, after trending up over previous years, but these appear to be driven by events related to occupational injury rather than systemic risk. COS selects and analyzes particular safety incidents to better understand why they occurred. COS classifies causes of Learning from Incident events into “Areas for Improvement” (AFIs). The most frequently cited causes identified in AFIs all point to management and human performance issues.

Conclusion 3-3: Trends in available BSEE process-safety incident data are inconclusive. The frequency of such events is generally low. Some trends appear to be improving, but others, such as loss of well control, are volatile, which makes it difficult to assess trends in systemic risk.

Conclusion 3-4: Reports based on SEMS audits and corrective action plans show progress in industry commitment to safety management and BSEE commitment to making SEMS more effective. However, BSEE reports that these audit results show large differences in the commitment of companies to SEMS. Reports on audit deficiencies and corrective action plans (CAPs) can also serve as indicators of the offshore industry’s commitment to reducing the risk of their operations. In addition, BSEE oversight of this process can serve as an indicator of its role in enforcing safety management. For example, operators are required to submit CAPs to BSEE in response to audit deficiencies, but the CAPs are not required to examine and address the root causes of the deficiencies. BSEE has indicated that it may need to correct this deficiency through rulemaking.

Conclusion 3-5: Although industry-wide safety performance indicator data have been developed and substantially improved since Macondo, the data are not sufficiently mature to estimate the industry’s systemic risk profile. Today, assessment of the industry risk profile will have to depend on qualitative indicators and expert judgment rather than good empirical indicators.

Conclusion 3-6: Whereas BSEE collects and posts much of its data on a public-facing website, safety data for operator and facility profiles are not publicly disclosed but are kept and tracked internally by BSEE. Many of these disaggregated data are not available external to BSEE, but are used in the agency’s RBIs and as part of its annual operator performance review. Benefits in understanding industry maturity and

performance could be derived from further availability and analysis of these data.

REFERENCES

This page intentionally left blank.