Stress simulations rely on single borehole while contaminant transport studies lack field validation, annual report finds

Independent scientists monitoring Canada’s deep geological repository program say key models predicting underground water flow and rock stability remain too immature for regulatory use, with one fracture-network simulation judged “insufficiently mature to be used with confidence at this stage.”

The warnings come in the December 2025 annual report of the Adaptive Phased Management Geoscientific Review Group (GRG), a five-member international panel established in 2012 to review work by the Nuclear Waste Management Organization (NWMO) on the proposed repository for used nuclear fuel at the Revell Site in the Wabigoon Lake Ojibway Nation and Ignace area of northwestern Ontario. The report was posted to the NWMO web site on July 22.

The site was selected in November 2024. Since then, the GRG says, the geoscience team has produced “high quality work” and achieved an “excellent final product” with its updated three-dimensional geological model integrating data from all six deep boreholes. But the reviewers flagged several shortcomings in the downstream models that will underpin safety claims.

FRACTURE NETWORK MODEL REJECTED FOR DOWNSTREAM USE

A preliminary hydromechanical Discrete Fracture Network model drew sharp criticism. The GRG reviewed the model in September 2025 and found it misaligned with the site’s geology.

“The GRG noted a lack of alignment with the current geological and hydrogeological conceptual understanding of the site, including an overreliance on a shear-dilation hypothesis as the primary mechanism controlling elevated transmissivity, and an under-appreciation of the role of sub-horizontal structures in controlling flow dynamics,” the report states. “The GRG suggests that improvements will need to be made to the hydro-DFN model before it can be implemented for use in downstream work, such as groundwater flow and contaminant transport.”

Following a web-meeting in October, the panel hardened its view. “The GRG is of the opinion that the hydroDFN approach is insufficiently mature to be used with confidence at this stage,” the reviewers wrote. “The GRG considered that the modelling work should be based on the current conceptual understanding of the hydrogeological system at the Revell Site, as developed by the Geoscience team.”

In plain terms, the hydroDFN is a computer simulation of the cracks and fractures in the bedrock that would carry groundwater through the repository zone. The reviewers found the model relies too heavily on one theory about how cracks open and fails to account for flat-lying geological structures that may actually dominate underground water movement. Because this model feeds directly into safety calculations about how quickly radioactive material could travel if containers failed, the reviewers say it cannot yet be trusted for regulatory submissions.

STRESS MODEL BUILT ON SINGLE BOREHOLE

Underground stress conditions are being modelled with two separate approaches, one using Rock Mass Classification and another using Discrete Fracture Networks. But the GRG notes both efforts currently rest on thin data.

“The GRG has acknowledged that these initial 3D stress models, anchored on a single stress profile using DCDA measurements from IG_BH01, are a useful starting point for stress modelling at the site,” the report says. The panel adds that additional data to be collected in the forthcoming detailed phase of site characterization, such as further DCDA measurements and stress measurements by over-coring, will provide a basis for future iterations to arrive at suitable 3D stress models.

Rock stress—the pressure bedrock exerts in all directions—determines how stable underground caverns will remain over centuries and how the tunnels must be engineered. The reviewers note that NWMO is currently extrapolating site-wide conditions from a single deep borehole, equivalent to predicting weather for an entire province using one weather station. Without additional borehole measurements, the reviewers say the models remain preliminary.

GEOMECHANICAL REPORTS CRITICIZED FOR VAGUE LANGUAGE

Long-term geomechanical stability analyses for the proposed repository were found lacking in precision. The GRG reviewed an initial stability report in early 2025 and concluded that “while the report is overall well-structured, many improvements would be required in the second phase of work, both with respect to the modelling approach as well as in precision of reporting.”

The reviewers specifically highlighted “the need to clearly describe the scope of the modelling work, and to remove qualitative and vague language throughout.”

In October, the GRG reviewed an updated work plan and cautioned that “the stress model employed as input to the modelling must reflect the geological framework.” The panel also warned that “a simple comparison of results is insufficient, and any differences in results based on these approaches must be reconciled.” The reviewers further stressed that “vague and generalized interpretations must be avoided and, overall, the graphical presentation needs to be improved to assist the non-specialist reader in understanding the essence of the report.”

The reviewers are essentially saying the safety calculations must be presented with enough clarity that regulators, community members, and other non-specialists can understand exactly what is proven and what remains uncertain. Imprecise or generalized language in documents meant to demonstrate long-term safety undermines public and regulatory confidence, the panel indicates.

DATA GAPS ON CONTAMINANT RETENTION

The GRG identified missing field data that is fundamental to proving radionuclides would be trapped by the rock rather than carried away by groundwater.

“The GRG pointed out that more data will need to be collected during the next phase of site characterization to help reduce uncertainty in understanding the role of diffusion and sorption close to and immediately surrounding groundwater-conducting features,” the report states. “The GRG also identified in their feedback critical aspects needing attention, for example hydraulic interference tests, acquisition of more hydrogeochemical data across fractures, and matrix diffusion and sorption during radionuclide transport.”

The panel also said mineralogical information concerning the rock matrix adjacent to groundwater-conducting features will be needed to assess sorption capacity.

Diffusion and sorption are natural safety barriers: diffusion pulls radioactive particles from water-filled cracks into the dense rock matrix where they move slowly, while sorption binds them chemically to minerals. The reviewers say NWMO has not yet collected enough mineral and water chemistry data from the fractures themselves to prove these barriers will work as predicted. Without that proof, the safety case rests on assumptions rather than measured site properties.

UNCERTAINTY DOCUMENTATION MANDATED

As NWMO prepares regulatory submissions, the GRG is insisting that remaining unknowns be disclosed plainly. “This important step must include clear documentation of remaining assumptions, uncertainties and limitations within adopted models, with associated reports clearly documenting the current site understanding,” the panel wrote.

The reviewers want every model submitted to regulators to contain an honest accounting of what scientists do not yet know. This transparency is critical because regulators and the public must be able to distinguish between established facts and informed estimates when evaluating whether the repository will safely isolate waste for hundreds of thousands of years.

SOUTH BRUCE DATA SHOULD BE RELEASED

The GRG continued to stress that transparency around the unselected South Bruce Site in the Saugeen Ojibway Nation and South Bruce area would bolster public trust.

“The GRG continues to stress that, for the purpose of building public confidence in site understanding at the Revell Site, the geoscientific information acquired at the South Bruce Site in the Saugeen Ojibway Nation (SON) and South Bruce area should be made available, so that the public are aware of the quality of the work completed at the site not selected,” the report states.

The panel suggests that publishing the scientific work completed at the location that was not chosen would demonstrate the rigor and quality of the selection process, helping communities verify that the final decision was based on thorough comparative analysis.

REVIEW SCOPE

During 2025, the GRG held eight virtual meetings and one four-day in-person meeting in Toronto, including a one-day joint session with NWMO’s Safety Assessment Review Group. The panel formally reviewed fourteen technical documents—twelve for the Revell Site and two for the South Bruce Site—and received four additional reports for information only.

The GRG’s membership in 2025 comprised Dr. Peter Kaiser (Chair) of Laurentian University, Dr. Alexander Cruden of Monash University, Dr. Sven Follin, Dr. Andreas Gautschi, and Dr. Michael Stephens.

Click here to read the full report.