The Department of Energy needs to better oversee the quality assurance of scientific computer modeling at Hanford and other environmental cleanup sites, according to a Government Accountability Office report to Congress.
The modeling is used for significant decisions on how to clean up radioactive and hazardous chemical waste, but the "oversight of the quality of the models has not always been commensurate with the models' importance," the report said.
"Because the decisions (DOE) makes must protect human health and the environment for thousands of years into the future, it is critical that the models on which (DOE) bases its decisions are of the highest quality possible," the report said.
Accurate models also are needed to ensure that DOE and Congress have solid data for decisions on cleanup efforts that will take decades and cost billions of dollars, the report said.
The computer simulation models use mathematical formulas to describe or predict physical and chemical environmental processes.
They might be used to analyze the effectiveness of different cleanup proposals, to assess the performance of strategies that have been selected or to simulate cleanup operations to provide help in planning and budgeting.
But the models have had some quality issues, the review said.
Last year the Washington State Department of Ecology told DOE it would no longer accept the use of a DOE model to analyze soil and ground water contamination at Hanford, the report said. The model is inadequate to capture complexities of the movement of contamination through the soil, according to the state.
An independent review last year of a draft environmental study with alternatives for closing Hanford's underground tanks holding radioactive waste raised concerns about the modeling done in the study. The review concluded that uncertainties in the modeling were not adequately addressed and the draft study was insufficiently precise to be used to make a cleanup decision.
In 2009, another review concluded that a model used to predict the composition of radioactive waste as it is prepared for treatment at the Hanford vitrification plant had limited capabilities. The result could be waste that the vit plant could not accept for treatment, the report said.
DOE does have general quality policies for its computer models, but it has not regularly assessed how its contractors implement them, the report said.
It reviewed eight cleanup decisions at Hanford and the Savannah River Site in South Carolina, and found that DOE conducted only three quality assurance assessments that addressed quality standards for the models used in those decisions.
In fact, one quality assurance assessment was done only after a contractor discovered data quality errors in a 2005 computer model used to support a prior Hanford environmental impact statement, the report said.
A Hanford official said quality assurance assessments are conducted primarily on models and their associated software if their failure would have significant safety consequences to workers, the public or the environment, according to the report.
DOE has begun efforts to promote consistency in the use of models across the environmental cleanup complex nationwide.
"However, some of these efforts are still in their infancy, and it remains to be seen whether any improvements in (DOE's) management of its models will result," the report said.
DOE agreed with the report's recommendations for improvements. They include clarifying quality assurance requirements, ensuring models are assessed for compliance with those requirements and developing a comprehensive strategy to manage models. The strategy would promote consistency, reduce duplication and ensure the sharing of information, the report said.