Main Content

EPA Revises Johnson and Ettinger Model for Vapor Intrusion

In 1991 Paul Johnson and Robert Ettinger, then of Shell Development Company, developed a complex mathematical model to predict vapor intrusion from subsurface contamination. As vapors migrate from source to receptor, their concentrations are attenuated (lessened) by varying degrees, depending on a number of factors, including: Building size, Soil-gas entry rate, Building air exchange rate, Soil type, Soil porosity, Soil moisture content, Depth to source, and Chemical-specific volatilization from groundwater.

The model did not take into account chemical breakdown, which is typically much more rapid for petroleum hydrocarbons (PHCs) than for chlorinated volatile organic compounds (CVOCs).

In 2003 EPA incorporated the J&E model into spreadsheets to calculate VI from soil, groundwater, soil gas, and Non-Aqueous Phase Liquids (NAPLs). The spreadsheets calculated not only indoor vapor concentrations, but also the risk to building occupants, based on EPA Regional Screening Levels (RSLs) for each chemical, and exposure factors, such as exposure time. There were some serious limitations with the spreadsheets. One was that the risk factors built into the spreadsheets were not easily updated. Users had to use a password to unlock the spreadsheets and change risk parameters, such as Reference Concentrations (RfCs), to match the latest Regional Screening Levels (RSLs). In 2004 EPA revised the spreadsheets with updated RSLs, but RSLs continue to change over time. In 2004, EPA also dropped the NAPL spreadsheets and soil spreadsheets, due to questions regarding their ability to estimate VI with any kind of accuracy.

There were other problems with EPA spreadsheets as well. Paul Johnson, coauthor of the J&E model, was critical of EPA’s spreadsheet version for various reasons, including the fact that the spreadsheets had built-in automatic checks for numerical errors, but there were no built-in reasonableness checks, making it possible to enter wildly inappropriate and conflicting data. Similarly, coauthor Robbie Ettinger made the point that while EPA’s spreadsheets incorporated the J&E algorithm, they were separate from the J&E model, and he did not agree with everything in them. Not surprisingly, after EPA released a report titled, “Uncertainty in the Johnson-Ettinger Model for Vapor Intrusion Calculations” in September 2005, states started to disallow the use of J&E modeling.

The latest version of EPA’s Spreadsheet for Modeling Subsurface Vapor Intrusion remedies many of the problems. Because the revised spreadsheet is linked directly to EPA’s RSLs, there is no longer any lag between them, and the complications of entering risk factors are avoided, providing that state or other jurisdiction accepts US EPA’s factors. (Some states, particularly California, often disagree with US EPA’s risk factors). The revised spreadsheet is also simpler in that groundwater and soil gas are entered into the same spreadsheet. The earlier versions had separate spreadsheets for each. The updated model also displays many of the calculations on the main page, such as Qsoil and Qbuilding, the rate at which soil gas and outdoor air, respectively, enter a building. Previous spreadsheets displayed the calculations on separate pages, and in a way that was harder to interpret. Additionally, the revised J&E model performs calculations for multiple chemicals on a single spreadsheet, instead of requiring a separate one for each chemical. This will save a lot of paper!

More importantly, like EPA’s earlier spreadsheets for J&E modeling, the updated spreadsheet is less conservative than default Vapor Intrusion Screening Levels (VISLs), making unnecessary vapor mitigation for low-risk sites less likely. For example, using Excess Lifetime Cancer Risk of 10E-5, Hazard Quotient (HQ) of 1, and groundwater temperature of 11 degrees Centigrade), the residential groundwater VISL for trichloroethene (TCE) is 9.9 micrograms per liter (ug/l). However, the revised J&E spreadsheet indicates that TCE in groundwater, using the same settings, has a target concentration of 19.4 ug/l. Additionally, using the J&E model, the acceptable concentrations of chemicals in soil gas or groundwater can be considerably higher than VISL levels, depending on soil type, depth to source, and other factors. Using the same default factors in the J&E model, but changing the soil type from sand to loam, (a mixture of sand, silt, and clay), raises the allowable level of TCE in groundwater to 194 ug/l. And unlike the revised VISL calculator, you can make changes to the J&E spreadsheets after downloading them to your computer and exiting EPA’s website.

A technical problem with the revised J&E spreadsheet is that the earlier-mentioned Qsoil/Qbuilding ratio is currently fixed at 0.003. This hairy looking factor boils down to the rate at which subslab soil gas (Qsoil) mixes with indoor air (Qbuilding) – “Dilution is the solution to pollution”, as they say. For now, changing the Residential setting to Commercial/Industrial automatically increases the ceiling height, air exchange rate, and slab thickness, which is helpful, but the dilution ratio, Qsoil/Qbuilding, stays at 0.003, and is taken from a lookup table, not calculated. According to Rich Kapuscinski, EPA’s website content manager, “We are and will be expeditiously addressing these problems, as we intend to have (and publish) a (revised) tool that is suitable for commercial, as well as, residential buildings.”

The other issue with the revised J&E model will be its acceptance by state and local agencies.

Author: Mort Schmidt is a Senior Scientist with Cox-Colvin & Associates, Inc. He received his BS and MS degrees in Geology and Mineralogy from The Ohio State University, and has been a Cox Colvin & Associates employee since 1997. His areas of expertise include vapor intrusion and contaminant investigation and analysis, and he currently serves as Cox Colvin’s Practice Leader – Vapor Intrusion Services. Mort is a Certified Professional Geologist with AIPG and is a registered Geologist in Indiana.