Mini-SOW-OilSpillRates2.docx

SECTION B

STATEMENT OF OBJECTIVES AND PROJECT MANAGEMENT PLAN

B.1 INTRODUCTION

The mission of the Bureau of Ocean Energy Management (BOEM) is to manage the development of U.S. Outer Continental Shelf (OSC) energy and mineral resources in an environmentally and economically responsible way. Management of the oil and gas resources of the Outer Continental Shelf (OCS) is governed by the OCS Lands Act (43 U.S. Code [U.S.C.] 1331 et seq.), which sets forth procedures for leasing, exploration, and development and production of those resources. Section 18 of the OCS Lands Act calls for the preparation of a nationwide offshore oil and gas leasing program, setting forth a five-year schedule of lease sales designed to best meet the Nation’s energy needs. BOEM and the Bureau of Safety and Environmental Enforcement (BSEE) within the U.S. Department of the Interior (USDOI) are responsible for implementing the requirements of the OCS Lands Act related to OCS oil and gas exploration and development.

Management of OCS oil and gas includes offshore leasing, resource evaluation, review and administration of plans for oil and gas exploration, development, production and decommissioning, National Environmental Policy Act (NEPA) analyses, and environmental studies (http://www.boem.gov/). One aspect of these responsibilities is assessment of potential environmental effects of an OCS oil and gas program, including accidental events. In order to complete such assessments, BOEM requires an understanding of oil spill probabilities, including spill sizes, frequency, and location. BOEM also conducts modeling of oil-spill trajectories to identify areas that may be impacted by a spill, should a spill occur in a given area (e.g., Ji and Schiff 2023). The oil-spill modeling outputs are used to help inform analyses of potential effects resulting from a proposed action for NEPA and consultations with other federal agencies.

National spill occurrence rates were last updated in 2016 (ABS 2016). The 2016 data compilation and analysis compare OCS spill data from1964 through 2015. Different types of spills (i.e. platform and pipeline; petroleum-based and crude only), spill size classes, intervals of time, and casual factors are considered. Spill rates were determined in terms of the volume of oil produced or handled. Additional OCS oil spill data were collated for fault tree analyses in the Arctic and the Atlantic (Myers et al. 2018). Stalfort et al. (2021) generated statistics for oil spills in the US GOM OCS from 1974 to 2019, quantified spill occurrence rates for an extreme case using the extreme value analysis and power law analysis, and developed a spill rate methodology to compute occurrence rates for the Atlantic OCS region. BOEM is proposing to update the OCS spill data and spill rate analyses through calendar year 2024 (1964–2024). Additionally, BOEM is interested in looking at what factors have caused oil spills over time and identifying trends within the oil spill data.

Artificial intelligence (AI) has been increasingly applied to statistics in risk analysis (e.g., Norvig and Russell 2021; Aziz and Dowling 2019) and can be a powerful tool in the domain of oil spill management (De Kerf et al. 2020; Mohammadiun 2023). AI algorithms can be used to analyze historical data on accidental oil spills; environmental conditions; vessel traffic patterns; exploration, development, and production trends; and other relevant factors to estimate the likelihood and potential impact of oil spills. Machine learning models such as random forests, support vector machines, or neural networks can be trained on large datasets to identify potential exposure variables, casual factors, and patterns of oil spill risks (Haupt 2008). Deep neural networks can be used to process complex oil spill data to identify oil spill patterns with greater accuracy. AI has the potential to revolutionize oil spill risk analysis and offshore environmental cost analysis, leading to improved preparedness, faster and more effective oil spill response, and minimized environmental and financial damage.

B.2 STUDY OBJECTIVES

The overarching goals of this study are 1) to update oil spill rates of OCS platforms, OCS pipelines, and U.S. barges, 2) to obtain oil spill information for BOEM’s Offshore Environmental Cost Model (OECM), 3) to analyze oil spill trend and casual factors. Having updated oil spill and oil spill occurrence rate data is critically important to analyze the potential risk and consequence of OCS oil spills, investigate causal factors contributing to the occurrence, size or frequency of oil spills, enhance oil-spill response planning, and target future regulatory reform to better manage risk.

Specific objectives are:

· Compilation of oil spill data and QA/QC of the data

· Examination of historical spill occurrences and of volume of oil handled

· Analysis of other potential exposure variables

· Estimate spill occurrence rates and normalize these rates based on number of spills per volume handled and other potential exposure variables

· Analysis of oil spill trend and casual factors

· Provide information for OECM model

· Prepare a 508 compliant report according to BOEM’s specifications.

B.3 SCOPE OF WORK

B.3.1 Task 1: Examination of historical spill occurrences and of volume of oil handled Data on spills from platforms and pipeline on the OCS are collected and maintained by BSEE. All OCS spills ≥ 1 bbl are required to be reported to BSEE, including the data for the 1964-2024 period for OCS platform and pipeline spills. Oil spills of any size are also required to be reported to the National Response Center. Data for the additional spill sources (tankers and barges) should be obtained by the Recipient from sources such as the National Response Center, the US Army Corps of Engineers (Waterborne Commerce), British Petroleum, and other appropriate sources. All oil spill data compilated should go through QA/QC.

For the purposes of oil spill risk analysis, BOEM defines oil spill in multiple categories, including: 1) small spills (< 1,000 bbl), 2) large spills (≥ 1,000 bbl), 3) very large spills (≥ 10,000 bbl), and 4) extra large spills (≥ 100,000 bbl). Spills are also classified by their sizes in detail:

Spill Class Sizes

≥ 1 to < 5 bbl

≥ 5 to < 10 bbl

≥ 10 to < 20 bbl

≥ 20 to < 50 bbl

≥ 50 to < 100 bbl

≥ 100 to < 500 bbl

≥ 500 to < 1,000 bbl

≥1,000 to <10,000 bbl

≥1,000 bbl

≥10,000 bbl

≥100,000 bbl

Number, mean spill size, and median spill size should be determined for each spill class size using relevant formulas from the appropriate probability distribution. Error analysis should be conducted for each of these spill class sizes. BOEM will provide the Recipient with exposure data, such as annualized volume of oil handled. An exposure variable is some quantity or measure related to oil operations, production or transportation which has a precise statistical relationship to spill occurrence.

B.3.2 Task 2: Consideration of alternative exposure variables Historically the estimated mean number of spills, in conjunction with the Poisson distribution, has been used to estimate the probability of spill occurrence. This estimated mean number of spills is normalized as a function of the oil handled. “Handled” has been defined for platforms and pipelines as oil produced and for tankers and barges as oil transported. This exposure variable was previously selected based on the fact that it was simple to define, and it was both a measured and predictable quantity. There is existing historic data on volume of oil handled and associated spills. Further, BOEM estimates the future OCS oil production. However, there is the potential for a better suited exposure variable. Lanfear and Amstutz (1983) examined the cumulative frequency distributions of oil spills, tested pipeline miles as an alternative exposure variable for pipeline spills, and discussed the trend analysis of offshore spills performed by Nakassis (1982). Eschenbach and Harper (2006) examined other alternative exposure variables including time, oil production, pipeline mile-years, and platform-years. Additional alternative exposure variables have and could be considered such as production years, age of infrastructure, length of pipeline transport, etc. (Eschenbach and Harper 2010; Robertson et. al 2013, 2020a, 2020b).

AI methods can be used to conduct probabilistic risk assessments and sensitivity analyses to quantify the uncertainties associated with various input parameters and assumptions in oil spill risk analysis, such as climate change, hurricane frequency, age of oil facility, water depth (shallow water area vs. deep water area), size of spills (small spills vs. large spills), etc. An AI-based decision support systems could integrate data from multiple sources to assist risk estimation. These methods may employ optimization algorithms, expert systems, or reinforcement learning techniques to generate more accurate and reliable risk assessments. Any alternatives should be presented for BOEM review and approval.

B.3.3 Task 3: Estimate spill occurrence rates and normalize these rates based on number of spills per volume handled and other relevant exposure variables Oil spill rates will be calculated for OCS platforms and OCS pipelines for relevant spill size classes, U.S. and Alaska North Slope tankers, and U.S. barges using the produced volume exposure variable along with any alternative exposure variables determined to be appropriately suited. The analysis should evaluate and analyze spill and spill rate data to confirm that observations are random and independent. Confidence intervals for spill rates should be calculated using relevant formulas.

The oil spill rates will be analyzed for two periods: the entire period of available data (1964 – 2024) and the latest 15 years (2010–2024). If calendar year 2025 data becomes available for this study in time, the latest 15 years will be 2011–2025. Additionally, the oil spill rates will be analyzed in the Gulf of Mexico region.

B.3.4 Task 4: Analyses on oil spill trend and casual factors Analyze spill rates for relevant spill size classes over the entire record (1964 through 2024) for OCS platform and pipeline spills. Statistically evaluate and determine other appropriate timeframes for rate analyses within the available record. The traditional methods (e.g., Anderson et al. 2012; ABS 2016) will be used to identify oil spill trend, patterns, and casual factors.

While AI applications in market risk analysis are well-established, their use in oil spill risk analysis is still emerging. It is expected that machine learning algorithms should be applied to identify oil spill trend and patterns, assess casual factors, and optimize risk mitigation strategies. Alternatives should be presented for BOEM review and approval.

B.3.5 Task 5: Oil spill information for OECM

BOEM’s Offshore Environmental Cost Model (OECM) (Industrial Economics Inc. 2023) uses estimates of mean spill rate (bbl/bbl produced or transported), mean spill size, and spill class as a percentage of total spills for five different categories: platforms, OCS pipelines, OCS vessels, crude oil and gas tankers, and refined oil and gas tankers. The two different oil types are crude and diesel. These data are important inputs to the OECM, which is integral to the Net Benefits analysis that supports the development of the National Outer Continental Shelf Oil and Gas Leasing Program. The analysis will be conducted using two periods of data sets: extended 1964-2024 (60 years) and recent 2004-2024 (20 years). The area of the oil spill data for this analysis will be all regions in the OCS. The data used in the analysis will also be provided to BOEM for future analysis.

An example screenshot from the OECM illustrates the required information for OCS platforms (oil source) on the spill rate (bbl spilled per bbl produced or transported), mean spill size, and the distribution of spills across spill size classes for the crude oil type. More specifically, the oil sources include five categories: platforms, OCS pipelines, OCS vessels, crude oil tankers, and gas tankers. The oil type includes two types: crude and diesel.

The oil spill size classes for different oil sources and oil types include: platforms, OCS pipelines, and OCS Vessels are subdivided in to 5 spill sizes (very small, small, medium, large, and very large) defined as: (a) 1 to 10 bbls; 11 to 100 bbls; 101 to 1,000 bbls; 1,001 to 10,000 bbls; and 10,001 to 100,000 bbls for crude oil; and (b) 1 bbl; 2 to 10 bbls; 11 to 100 bbls; 101 to 1,000 bbls; and 1,001 to 3,600 bbls for diesel. Because of OECM’s unique application, the definitions of oil spill sizes in Task 5 are different from the ones in oil spill risk analyses (e.g., Anderson et al. 2012; ABS 2016; Ji and Shiff 2023).

Crude and refined oil and gas tankers are subdivided in to 6 spill sizes (very small, small, medium, large, very large, and extra large) defined as: 0.1 to 10 bbls; 10.1 to 100 bbls; 101 to 1,001 bbls; 1,001 to 10,000 bbls; 10,001 to 100,000 bbls; and over 100,000 bbls for both diesel and crude oil.

Table 1 Example of Data Categories and Table used for Platforms, Pipelines, and Vessels Oil Source:

Oil Type:

(bbl) Mean Sill Rate (bbl/bbl produced or transported)

% of Total Spills

Table 2Example of Data Categories and Table used for Tankers (Note the Extra Large Spill Size row unique to Tankers) Oil Source:

Oil Type:

(bbl) Mean Sill Rate (bbl/bbl produced or transported)

% of Total Spills

B.3.6 Task 6: Prepare a 508 compliant report according to BOEM’s specifications.

This report presents the data and methods used, data analyses, and significance of findings. The report will present results from platform and pipelines independently. Further, the results will be presented with the aforementioned timeframe determination. This report will also include data used in this study. The report should be 508 compliant and follow BOEM’s specifications (ESP Report and Footprint Specifications https://www.boem.gov/environment/environmental-studies/report-specifications/esp-report-and-footprint-specifications).

The data used for the analysis will be delivered with the draft report. The Recipient shall prepare and deliver a database containing all data used to support the analyses to BOEM with a format and structure agreed upon by BOEM. The data format could be, but are not limited to, Excel, csv, Python, R, and/or text, with the final dataset formatted and ready to be analyzed so that the data can be double checked or analyzed later to address new questions that might come up. This database shall be delivered six months before the end of the contract or at the same time as the draft report delivered.

B.4 PROJECT MANAGEMENT

The BOEM Project Officer (PO) will work with the Principal Investigator throughout all project phases. The BOEM PO will participate as a co-principal investigator in the planning of the overall study design, coordination, and decisions concerning the direction of the research. The BOEM PO will identify, process, and apply relevant agency information throughout the project. BOEM scientists will provide assistance in gathering and analyzing in-situ observations that were collected by BOEM and in verifying the results. In particular, BOEM scientists will conduct oil spill data analysis and QA/QC of the data in house. The participation of BOEM scientists will improve the final product of this project. Additionally, the BOEM PO will contribute to the final data analysis and preparation of publications arising from the project.

B.4.1 Project Management Plan (PMP)

The Recipient shall operate under the negotiated PMP. This plan shall include a Project Manager (PM) with sufficient authority to serve as the Contract Administrator, the point of contact with the BOEM, and provide leadership to the study team. The PM will ultimately be responsible for all deliverables, budget control, quality assurance for all products, compliance and adherence with schedule of the study, and that all personnel work synergistically and cooperatively for the common goal of the study. The PMP shall also include description of a Data Management Plan (DMP).

BOEM envisions the project period of 24 months to review the historical data, select the exposure variable, estimate and normalize rates, and complete the final report. The project period shall encompass all tasks from initial planning, through and including the BOEM’s final acceptance of all deliverables. The Recipient shall deliver all required deliverables in accordance with Sections C, E, and F.

B.4.2 Data Management Plan (DMP)

The Recipient shall develop a Data Management Plan (DMP) for this study that shall include procedures to handle data collection/acquisition, data storage and backup, dataset selection (external sources), Quality Assurance/Quality Control (QA/QC) procedures, dissemination and distribution protocols, and archival procedures. The DMP shall be submitted to the BOEM PO for review and approval prior to implementation. A designated Data Manager shall implement the plan to ensure compliance by all Principal Investigators (PIs) involved with the study. After all QA/QC procedures have been completed, results shall be submitted to the BOEM PO.

No originals shall be mailed without prior backups. These files are proprietary and shall not be used by the Recipient for any purpose other than specified by the Contract, unless authorized in writing by the CO, and must be secured at all times. When distributed, the Recipient shall include language to the data Recipient that ensures that the data are not distributed further, and that no publication is done without giving acknowledgement to BOEM.

B.4.3 Data Administration

The Recipient shall implement a data administration system to be used to ensure continuous evidence of data possession and control with signatures, dates, times, and location of the data being noted. The Recipient shall ensure the proper formatting and reporting of all data and the distribution of data, as required, to study participants and PIs of other ongoing BOEM studies.

B.4.4 Data Control

The Recipient shall utilize a data inventory control system to monitor study progress, identify gaps in the information acquired, and suggest additional processing requirements. The inventory control procedures shall enable the Recipient to document data availability, data reduction, and data analysis at each reporting period.

B.4.5 Data Utilization

The Recipient's DMP shall ensure that all data are processed, validated, and made available as needed to study participants and other ongoing BOEM studies. The data must be retrievable, as necessary or desirable, for any future analysis.

B.4.6 Archiving

The Recipient shall provide a method for the safe storage and easy retrieval of all measurement results, acquired data, photographs, and video. The Recipient shall ensure that no loss of information occurs in any processing step taking place prior to archiving. The PM shall inventory the data and verify the presence of any and all necessary documentation prior to archiving. Original data shall be copied as early as possible and original copies shall never be mailed until at least one verified copy has been made. The Recipient shall retain a copy of the information collection logs and information delivered to BOEM for a period of one year from the Contract completion date.

B.4.7 BOEM Data

The Recipient shall submit digital copies of all data used in this study. This deliverable shall be completed after all QA/QC procedures are finalized. The preferred format of the files shall be determined based on an agreement between the PO and the Recipient at the Post-Award meeting. Any changes in data formats during the contract period of performance must be agreed upon in advance by the Government and the Recipient. The Recipient shall be responsible for the submission of the Contract data to the BOEM in digital format according to the schedule in Section F.5.

B.4.8 References

ABS Consulting, Inc. 2016. 2016 Update of Occurrence Rates for Offshore Oil Spills. Prepared by ABS Consulting Inc. for USDOI, BOEM/BSEE. Sterling, VA: USDOI, BOEM/BSEE. 95 pp.

ABS Consulting, Inc. 2018. US Outer Continental Shelf Oil Spill Statistics OCS Study BOEM 2018-006. Anchorage, AK: USDOI, BOEM, Alaska OCS Region. 44 pp.

Anderson, C.M. Mayes, M., and LaBelle, R.P. 2012. Update of Occurrence Rates for Offshore Oil Spills. Bureau of Ocean Energy Management OCS Report 2012-069. Bureau of Ocean Energy Management, Herndon, VA Aziz, S. and Dowling, M., 2019. Machine learning and AI for risk management. Disrupting finance: FinTech and strategy in the 21st century, pp.33-50.

De Kerf, T., Gladines, J., Sels, S. and Vanlanduit, S., 2020. Oil spill detection using machine learning and infrared images. Remote sensing, 12(24), p.4090.

Eschenbach, T. G. and Harper, W. V. 2006. Estimators for the Beaufort/Chukchi Sea OCS (Statistical Approach). Anchorage, AK: USDOI, MMS, Alaska OCS Region; 2006; OCS Study MMS 2006-059.

Haupt, S.E., Pasini, A. and Marzban, C. eds., 2008. Artificial intelligence methods in the environmental sciences. Springer Science & Business Media.

Industrial Economics Inc. 2023. Forecasting environmental and social externalities associated with Outer Continental Shelf (OCS) oil and gas development, volume 1: 2023 revised Offshore Environmental Cost Model (OECM). Sterling (VA): U.S. Department of the Interior, Bureau of Ocean Energy Management. Report No.: OCS Study BOEM 2023-056. https://espis.boem.gov/Final%20Reports/BOEM_2023-056.pdf Ji, Z.-G. and Shiff, J. 2023. Oil spill risk analysis: Gulf of Mexico Outer Continental Shelf (OCS) Lease Sales in the Eastern Planning Area, Central Planning Area, and Western Planning Area and Gulf-wide OCS Program, OCS Report 2023-026. Bureau of Ocean Energy Management, Sterling, VA.

Lanfear, K.J. and Amstutz, D.E. 1983. A Reexamination of Occurrence Rates for Accidental Oil Spills on the U.S. Outer Continental Shelf, 1983 Oil Spill Conference, American Petroleum Institute, Washington, DC Myers J, Roberts B, Lee B, Hill S (ABS Consulting, Inc., Arlington, VA). 2018. US outer continental shelf oil spill causal factors report. Anchorage (AK): U.S. Department of the Interior, Bureau of Ocean Energy Management, Alaska OCS Region. 36 p. Report No.: OCS Study BOEM 2018-032. https://espis.boem.gov/final%20reports/BOEM_2018-032.pdf Mohammadiun, S., 2023. Selection and optimization of marine oil spill response operations using artificial intelligence and soft computing techniques (Doctoral dissertation, University of British Columbia).

Nakassis, Anastase. 1982. Has Offshore Oil Production Become Safer?, U.S. Geological Survey Open-File Report 82-232.

Norvig, P. and Russell, S., 2021. Artificial intelligence: a modern approach, Global Edition. Pearson, Harlow, 1, pp.1239-1269.

Robertson TL, Campbell LK, Pearson L, Higman B. 2013. Oil spill occurrence rates for Alaska North Slope crude & refined oil Spills. Report to the Bureau of Ocean and Energy Management. https://www.boem.gov/sites/default/files/uploadedFiles/BOEM/BOEM_Newsroom/Library/Publications/2013-0205.pdf Robertson T, Campbell LK, Fletcher S (Nuka Research and Planning Group, LLC, Seldovia, AK). 2020a. Oil spill occurrence rates from Alaska North Slope oil and gas exploration, development, and production. Anchorage (AK): U.S. Department of the Interior, Bureau of Ocean Energy Management, Alaska OCS Region. 84 p. Report No.: OCS Study BOEM 2020-050. https://espis.boem.gov/final%20reports/BOEM_2020-050.pdf Robertson T, Campbell LK, Fletcher S (Nuka Research and Planning Group, LLC, Seldovia, AK). 2020b. Oil spill occurrence rates for Cook Inlet Alaska oil and gas exploration, development, and production. Anchorage (AK): U.S. Department of the Interior, Bureau of Ocean Energy Management, Alaska OCS Region. 81 p. Report No.: OCS Study BOEM 2020-051. https://espis.boem.gov/final%20reports/BOEM_2020-051.pdf Stalfort D, Roberts B, Culvern C (ABSG Consulting, Inc., Spring, TX). 2021. Alternative Oil Spill Occurrence Estimators for Determining Rates for the Atlantic Outer Continental Shelf. Sterling (VA): US Department of the Interior, Bureau of Ocean Energy Management. 97 p. Report No.: OCS Study BOEM 021-065. https://espis.boem.gov/final%20reports/BOEM_2021-065.pdf