Alaska mining spill retrospective

Completed November 2021, released April 2022

Clients: Brooks Range Council, Earthworks, National Parks Conservation Association, and the Tanana Chiefs Conference

Alaska oil and gas industry spill impacts on polar bears and walruses

Completed June 2021

Client: Defenders of Wildlife

The Stibnite Gold Project (SGP) will require large quantities of more than 30 different hazardous materials being transported to and from the mine site in thousands of heavy truckloads each year. Those hazardous materials include fuels, explosives, acids, and toxic materials, but the dangers posed by the reagents are not discussed.

• The SGP DEIS had inconsistent descriptions of the reagents to be used and the number of trips of hazardous materials trucks would require and multiple arithmetic errors.

• The transportation corridor analysis area did not consider any risks beyond Cascade, Idaho.

• EISs for other mines include spill risk rates that are on the order of 2.0 x 10-7 spills per truck-mile, but the SGP DEIS estimated a spill rate ranging from 1.4-1.9 x 10-9 spills per truck-mile, which is two orders of magnitude lower than rates published in multiple sources.

• Using a spill risk rate of 1.6 x 10-7 spills per truck-mile, I found that probabilities of spills and accidents range from possible to probable (if not certain) for all Action Alternatives for the analysis area considered in the SGP DEIS and the full length of the transportation corridor (Figure).

• Without an accurate characterization of the true exposure along the transportation corridor and the spill rate per truck-mile, it is impossible to make informed statements about spill likelihood and the potential consequences to the environment and to public safety.

Overall, the analysis of the potential impacts from hazardous materials in the SGP DEIS is inadequate to make an informed decision.

The fundamental question underlying the fish habitat assessment for the proposed Pebble Mine is:

How will fish (specifically seven species of interest at defined life stages) respond if Pebble Mine permanently changes the flows in the streams around the mine site?

The US Army Corps of Engineers (USACE) and Pebble Limited Partnership (PLP) ask the reader to believe that the streams in the vicinity of the Pebble deposit will be better suited to producing fish if they had less water and less variability in flow magnitudes than they do without the mine. The USACE and PLP used a complicated and interconnected set of models (surface water and groundwater models, PHABSIM, and HABSYN) to address this, each with serious methodological weaknesses or flaws in execution.

• The surface water model that is the linchpin of all further flow modeling does not predict flows based on precipitation precisely. There are competing and fundamentally different groundwater models in different portions of the overall water modeling effort that even the authors of the EIS do not understand.

• PHABSIM was developed by scientists and managers with no background in ecological or environmental modeling 40 years ago. It is based on outdated concepts and techniques, and was never subjected to fundamental, rigorous peer review during its development.

• Even if PHABSIM were a defensible model, the requirements for getting the most accurate results from it were not met by the sampling conducted on the three streams, either in transect placement or in the range of flows sampled.

• PHABSIM was modified into the proprietary and unvetted model HABSYN, which theoretically gave a greater degree of spatial resolution by extrapolating from 143 transects with depth, velocity, and substrate observations to 1,977 habitat units where no observations had been taken.

• HABSYN was used at two different time steps (monthly and daily) with different hydrology periods (1942-2017 and 2005-2014). The results of the model were not consistent across the two runs.

The end result of PLP’s PHABSIM/HABSYN modeling effort is an attempt to use a tier of poorly designed models that do not work well together to show that permanent and often significant reductions in streamflow will not decrease the quantity or quality of fish habitat in the NFK, SFK, and UTC. In short, there are flaws and limitations of each component in that analysis and their characterization of the effects of the proposed Pebble Mine on fish habitat are scientifically and statistically unsound.

This analysis was a review of the spill risks associated with the transportation corridor and potential releases from tailings storage facilities. The release scenarios considered in Pebble Limited Partnership’s (PLP’s) Final EIS were limited to a small subset of spill sizes, substances, and sources.

My review of the transportation corridor considered the quantities of diesel, ore concentrate, ore concentrate slurry, and reagent movement via trucks, barges, and pipelines in light of historical spill data from Alaska. (The image is the fault tree for potential diesel spills. Similar diagrams can be made for ore concentrate and mine reagents. See Case Study #2 in Lubetkin (2020) for details.)

I also applied basic rules of probability to the results of PLP’s tailings storage failure modes and effects analysis.

• The annual risk probabilities and consequences for the potential failure modes analyzed at a Failure Modes and Effects Analysis (FMEA) workshop were “semi-quantitative” and “subjective”, rather than analytical or based on calculations or statistics. Those ratings did not include confidence levels about the workshop’s assessments.

• There are 90 potential failure modes considered across the water management pond, the north and south embankments of the pyritic tailings storage facility, the bulk tailings storage facility, and the bulk tailings storage facility post-closure that could result in a partial or full breach. To avoid a breach, none of those 90 failure modes can occur over the project lifetime.

• Breaches from any of the structures could release anywhere from less than 1% of the volume to 100% of the volume. The three release scenarios modeled – one of which was not a breach but a pipeline rupture – only considered the environmental impacts of releases of less than 1% of the potential volume. Actual partial and full breaches could release much higher quantities than those described in the FEIS with much more harmful consequences.

• The FMEA workshop resulted in estimated annualized failure rates for each potential failure mode. Using basic rules of probability, the probability of success (that is, not one of the potential failures modes occur) can be calculated for each structure for a year. Those annual success probabilities can then be combined to find the probability that none of the structures have a breach in one year or over any specified length of time.

The Sea Port Oil Terminal (SPOT) Deepwater Port Project draft EIS used two models to analyze oil spill size, frequency, and fate. I analyzed both models for consistency of their inputs, assumptions, and results, and for technical flaws. The expected spill frequencies at various hole diameters from the two models were substantially different from one another (Figure).

The SPOT DEIS fate modeling exercises used an offshore spill volume based on a Bureau of Ocean Energy Management (BOEM) (2012) estimate of 2,200 bbl as a median large offshore spill volume. This value is unjustified because it only used data from 1996-2010, which is both too short a time frame for characterizing the size and frequency of large spills and does not use current spill data. When data from 1972-2017 are used and volumes from spills caused by the same hurricane are aggregated, the median spill size (n = 31 spills >1,000 bbl) is 3,489 bbl.

I coupled spill data and estimates of spill rate from BOEM with estimates of the volume of oil that would be handled at the terminal to find the number of spills expected in four different size classes.

The worst case scenario presented in the DEIS used 30 minutes for the shut down time in the event of a subsea release. I calculated volumes of oil released for various shut down times from 0.5 to 48 hours.

The EIS for Lease Sale 193 included an estimate that the frequency of large oil spills per billion barrels of oil produced would be lower in the Arctic than in the Gulf of Mexico. The estimate used by the Bureau of Ocean Energy Management (BOEM) was based on a fault tree analysis that included spill causes shared in the Gulf of Mexico and the Arctic and those unique to the Arctic. I examined BOEM’s spill records, the cause attributions of large spills, the model assumptions, and math used in the fault tree and concluded there were significant errors leading to an underestimated spill rate. (See Case Study #3 in Lubetkin (2020) for details.)