Digital transformation has disrupted the way we work, helping businesses manage risk, reduce cost, and significantly increase efficiencies. Yet, in the environmental remediation space, traditional approaches to managing contaminated sites and portfolios have barely changed over the past 40-years.
Portfolios, a grouping of remediation sites that belong to the same organization, are generally managed project-by-project with analysis and decision-making based on data from a single site. They rely on the professional judgement and expertise of the project team. And, in many cases, little thought is given to how teams can apply the value of data across portfolios or utilize existing data from across the remediation industry.
In other words, the portfolio management process is very human-dependent.
In the current digital age, we can instead leverage data to enhance portfolio management and reduce the liability costs associated with contaminated sites. Coupled with the ever-evolving capabilities of Artificial Intelligence (AI) and Machine Learning (ML), we can solve the challenges of portfolio management in innovative and digitally-enabled ways.
The S3 Framework: a holistic data-driven strategy
When it comes to portfolio management, most organizations focus on reducing risk and minimizing overall liability with the ultimate goal of receiving regulatory confirmation of site closure or no further action (NFA).
Strategies to reach the end goal vary across a portfolio. And, organizations typically have a limited annual budget for portfolio management, which needs to be used wisely to achieve the greatest reduction of potential liabilities.
GHD Digital’s patent-pending advanced analytics model, the S3 Framework, enables organizations to optimize the management of their portfolios through AI and ML on big datasets. The S3 Framework outlines the process to determine the overall risk of each site by considering three different data sets:
- Site: Both the physical location and the location-specific contamination risks. These are typical site data sets like contaminant concentration, geology, groundwater gradient and presence of preferential pathways.
- Surroundings: The risk drivers surrounding the site. For example, sensitive receptors, offsite sources and possible co-mingled plumes, zoning, or the amount of development in the surrounding area.
- Setting: Data that are mostly non-physical constraints affecting budget, timeline, and the required pathway to liability reduction and closure. This data could include several factors such as the regulatory criteria set by the state or site-specific regulations to progress toward closure, for example. Setting might also consider any regulator backlog of sites, the historic performance of how a regulator approved a site closure, the potential for or existing legal challenges, and much more.
Figure 1: The three S’s – site, surroundings, and settings
This framework is only successful if we consider the relationship of each “S” together. By doing so, we group sites with similar S3 risk profiles into similar categories and we inject data, such as financial information, to create “what if” scenario models and estimations of time and cost to closure based on learnings from entire portfolios. By classifying the S3 risk, we quantitatively understand our sites and develop a data-driven strategy prioritizing sites and accurately forecasting schedule and spend.
Document mining unearths trapped data
To achieve the greatest benefit from the S3 Framework, we need large datasets with actionable data from the site, surrounding, and setting. However, much of the data required to create these datasets reside in documents not often readily available in centralized databases. By applying AI and ML through a practice called document mining, the process of structuring information from documents, we access and turn text and information into insightful data.
In many cases, site data are trapped in the form of reports, correspondences, and other visual outputs (e.g., Word and PDF documents). This data can include easy-to-recognize information, like tables or figures. However, there is also a need to gather and understand contextual data, such as the source of release, subsurface characteristics, and details of remedial alternatives.
Using document mining, we quickly understand distinct types of proposed or enacted remediation methods and potential options moving forward. Without analytics support, we would have to read thousands of individual documents and manually populate a vast table of attributes. Through AI and ML-enabled document mining, we can accomplish this task quickly and provide valuable insight and contextual details.
There is an extensive list of factors that might influence decisions around site remediation and overall portfolio management. Compounded by the significant number of datasets requiring comprehension – which are often disparate or disconnected – document mining increases our team’s efficiency in developing the most valuable datasets from the multitude of associated documentation.
Human plus machine (enhancement, not replacement)
By integrating the available digital technologies, we build efficiencies and make more informed data-driven decisions. It’s also incredibly important to understand the application of technology is only successful when augmenting or complementing human intelligence.
Digital technologies empower remediation and portfolio managers to be more informed, enabling more confidence and performance efficiency. So, instead of meeting the challenge of improving portfolio management with a ‘technology vs. human’ perspective, we use technology to amplify existing processes and outcomes.
Applying the S3 Framework enables remediation teams to strategically inject technology to create efficiencies, add new and innovative insights, achieve greater liability reduction with less money, and improve decision making.
The integration of AI and ML reduces the need for costly and time-consuming manual processes and creates and retains new knowledge. Each new insight is then applied across sites and portfolios, reducing long-term liabilities, and enabling better forecasting of future liability costs.
To illuminate the most valuable insights for portfolio management, we must converge artificial and human intelligence. After all, the main goal of remediation is to improve health, safety, and the environment, while increasing the levels of trust and transparency with all responsible parties, stakeholders, and regulators. And, the best way to ensure we achieve this goal is by bringing together our team’s remediation knowledge, applying both document mining to unearth trapped data as well as AI and ML within the S3 Framework.
We are only at the beginning of what’s to come. The remediation industry is already behind other industries in actualizing the benefits of digital technologies. The first step in this process is to illuminate what’s possible through new, digitally enabled ways of working. From there, we are only limited by our imaginations.
Meet the Authors
With extensive experience in applying advanced data analytics to the remediation space, Jonathan Eller, Brett Roberts, and Jayesh Srivastava bring new and unique opportunities to our clients through digital technologies and capabilities. Regardless of your immediate need, they can help you meet your remediation goals and digitally transform your portfolio management.
Jonathan Eller, Ph.D., Data Scientist, Digital Intelligence - GHD Digital | Jonathan.Eller@ghd.com
Brett Roberts, Global Market Leader, Digital Environment - GHD Digital | Brett.Roberts@ghd.com
Jayesh Srivastava, Ph.D., Head of Analytics and Data Science, Digital Intelligence - GHD Digital | Jayesh.Srivastava@ghd.com