Long-term monitoring and operations of remediation systems over the life of groundwater cleanup projects may cost as much as discovery of the problem and construction of cleanup facilities. When driven by measurement data, long-term monitoring optimization integrated with remediation process optimization offers opportunities to reduce costs without significant changes in risk or uncertainty. These savings result from adjustments to monitoring locations and schedules, plus adjustments to the remediation process, that are based on better use of sampling information. A data-centered, software-based approach is proposed to accomplish cost-effective long-term monitoring of contaminated groundwater sites. Well-known statistical process control and standard groundwater simulation modeling methods will be integrated with optimization and data-assimilation tools. This PC-based software system will optimize long-term monitoring while allowing site managers and stakeholders to (1) provide time-varying management goals and priorities, and (2) select (or vary) how much uncertainty/risk can be tolerated. Propagation of uncertainty, treatment of nonstationary and nonseparable plume statistics, and multi-stakeholder/objective optimization are key areas of innovation. During Phase I, research will be performed for a method to zone plume data so that well-established statistical methods can be applied. A data-centric Bayesian method that embodies the worth of data will be developed, with special attention to computational efficiency and robustness. Proof-of-concept will be performed using data from a real site.
Commercial Applications and Other Benefits as described by the awardee: Recent GAO and EPA reports estimate that between $32B and $37B will be spent at Superfund sites alone in the next 40 years for monitoring and operating groundwater cleanup systems. A modest, 10 percent increase in productivity over that period yields about $3.5B that can be used for other purposes. This productivity is accomplishedleveraging sunk costs for characterization, modeling, and facilitiesby adapting the number of samples and the operation of the remediation system to time-varying measurement data.
