Paul Johnson, Ph.D. Lilian Abreu Ph.D. Candidate Department of Civil and Environmental Engineering

1. Site-Specific Modeling in the Context of the OSWER Guidance? OSWER Guidance Paul Johnson, Ph.D. Lilian Abreu Ph.D. Candidate Department of Civil and Environmental Engineering Ira A. Fulton School of Engineering

2. OSWER Guidance (11/29/02) Tier 3: Site-Specific Pathway Assessment “Modeling is considered to be useful for determining which combination of complex factors (e.g., soil type, depth to groundwater, building characteristics, etc.) lead to the greatest impact and, consequently, aid in the selection of buildings to be sampled. It is recommended that sampling of sub-slab or crawlspace vapor concentrations and/or sampling of indoor air concentrations be conducted before a regulator makes a final decision…”

3. OSWER Guidance (11/29/02) • Tier 3: Site-Specific Pathway Assessment - Issues • Why are you limited to near-foundation (e.g., sub-slab) soil gas data in Tier 3, when you can use soil gas data at any depth or groundwater data in Tier 2? • Why is semi-site-specific J&E modeling used in Tier 2 to assess impacts, but site-specific J&E modeling is not allowed in Tier 3? • If you allow site-specific modeling to decide on a subset of buildings does not need to be monitored - aren’t you using it to screen out sites? • If you could do site-specific modeling in Tier 3 - who is qualified to perform it and who is qualified to review the output? • Future use scenarios/no building currently present?

4. Site-Specific Modeling Options… [What might we do if we ignored the current language in the OSWER guidance?] 1. Site-specific a-value determined from “tracer” input Use of J&E model with site-specific inputs Multi-dimensional numerical codes

5. Option #1 Determination of, and use of, a site-specific a-value • measure soil gas and indoor concentration of tracer (a conservative chemical not expected to be confounded by ambient or indoor air background sources; radon, 1,1 DCE, etc.) • Derive site-specific a-value • Estimation indoor air concentrations for chemicals of concern using that site-specific a-value

6. enclosed space Option #2 Use of J&E (1991) model for site-specific assessment Site-specific a = Qsoil/Qb Layered settings? Measured Deff? Perched water? Fresh-water lens? Well-Mixed Diffusion [pseudo-steady state] vapor migration Source [steady or transient] Vapor Source

7. Generalized Sensitivity Assessment of the J&E (1991) Model* • The output only depends on three parameters (A, B, C) • If you understand sensitivity to those three parameters, you can quickly assess the sensitivity to any specific input. P.C. Johnson. 2002. Sensitivity Analysis and Identification of Critical and Non-Critical Parameters for the Johnson and Ettinger (1991) Vapor Intrusion Model. API Technical Bulletin.

8. Generalized Sensitivity Assessment of the J&E (1991) Model

9. Generalized Sensitivity Assessment of the J&E (1991) Model Most of the time critical*, but pretty well-constrained: [(VB/AB), LT, DTeff, EB] Sometimes critical, but data hints at their reasonable values: [(Qs/QB)] Rarely critical, and any reasonable value works: [h, Lcrack, Dcrackeff] • • If your analysis suggests “high” sensitivity to any inputs…you are probably using: • an inconsistent set of input values, or • - an unreasonable set or unreasonable range of input values

10. Needed Improvements… 1) Reformat the calculations in terms of: [(VB/AB), LT, DTeff, EB, h, (Qs/QB), Lcrack, and Dcrackeff] Eliminate the Qs calculation and input (Qs/QB) values based on empirical analysis. Input moisture saturations instead of individual moisture contents and total porosities Integrate the spreadsheet with the graphical flowchart for identifying critical parameters Constrain users to reasonable ranges and combinations of inputs… • Confusion stemming from (improper) use of the EPA spreadsheets could be minimized with the following changes:

11. Johnson et al. ES&T 1998 syringe small diameter tubing tracer gas tracer gas Time = 0 Time = t1 In Situ Diffusion Coeff. Measurements 1-L tedlar bag sample volume (≈ 9 cm radius) % Mass Recovered tracer gas Time = t2 Time Soil Gas Withdrawn

12. Effect of changing atmospheric conditions and occupant habits? Future use scenarios? Option #3 Multi-dimensional multi-component numerical code Effect of building construction (slab vs. basement)? Effect of aerobic biodegradation on a? Near foundation soil characteristics Effect of lateral separation between building and vapor source? Variation in a withconcentration, depth and soil type? Sub-foundation vs. near-foundation soil gas sampling?

13. Sample details for simulations 10 m x 10m footprint 5 Pa constant building under-pressurization 1 mm wide full perimeter crack 12/d exchange rate Fine to medium sand Grid spacing is variable - finer detail near cracks, source boundaries, and domain boundaries 30 m x 30 m constant source (200 mg/L-vapor)

14. A Sample Pressure Field… Symmetrical Simulation - cross-section through plane of symmetry

15. Alpha=1.2e-3 , Qs=4.1 L/min Changes in a with Source Position and Depth… No biodegradation

16. Alpha=9.3e-6 , Qs=4.1 L/min Changes in a with Source Position and Depth… No biodegradation

17. Changes in a with Source Position and Depth… Source no longer under building No biodegradation

18. alpha=1.2e-3; Qs=4.0 L/min Changes in a with Building Cons-truction alpha=6.1e-4 , Qs=5.1 L/min No biodegradation

19. alpha=1.4e-4 (w/biodegradation) alpha=1.2e-3 (no bio) Near-Foundation vs. Through-the-Foundation Sampling?

20. Changes in a with Depth with Bio-decay alpha=1.3e-18 (w/biodegradation) alpha=5.7e-4 (no bio)

21. In Progress… • Manuscript #1 – Model development and application to study of lateral distance and depths vs. impacts. • Manuscript #2 – Effects of aerobic biodegradation on impacts (source strength, depth, distance) • Study of role of sub-slab characteristics, pressure fluctuations, wind effects, etc. • Use of model to develop nomograph identifying sites where impacts may not be significant, based on • Building footprint • Depth to vapor source • Vapor source strength

22. Final Thoughts… • 1. Draft OSWER Guidance is inconsistent with respect to the role of modeling for site-specific pathway assessment (and the role of modeling in general..) • If the role of site-specific modeling is expanded, then we need to be prepared to address: • What options are allowed? • What data is required? • How to ensure that the use of site-specific modeling is technically credible?

23. Discussion

24. Groundwater Data Interpretation Issues With samples collected across conventional well screen intervals, there are multiple realizations that would correspond to the same depth-averaged groundwater concentration (in other words, the measured concentrations do not correspond to a unique vapor transport scenario) C1 > C2 C2 > C3

25. Model Inputs - what do we know? Input Thoughts Reasonable* H, Dair Tabulated Chemical Properties Actual Value E 10 - 20 d-1 (energy efficiency studies) 12 d-1 (VB/AB) 2 - 3 m (= ceiling height) 2.5 m (Qsoil /QB) <0.01 (radon studies and Colorado field data) 0.0001 - 0.01 LT 0.5 - 50 m actual value (qm/qT) 10% - 50% (vadose zone + crack) 0.10 Lcapillary 1 - 100 cm 5 cm (qm/qT) 90% - 98% (capillary fringe) 0.95 qT 0.25 - 0.50 0.35 Lcrack 15 - 60 cm (6 - 24 inches) 15 h 0.0001 - 0.01 (1=bare dirt floor) 0.001 * reasonable conservative value

26. Sensitivity of DTeff to Moisture Content.. • DTeff not very sensitive to reasonable variation in moisture content for a given soil type. • DTeff more sensitive to variation across gross changes in soil types (i.e. sands -> clay about 5X change). • The most significant change occurs between vadose zone and capillary fringe soils **however** the magnitude depends on H (beware at small H!) qT = 0.50 capillary zone qT = 0.35 Vadose zone soils Sands/ Gravels Silts/ Clays