Project at a Glance
Title: Development of a model for assessing indoor exposure to air pollutants
Principal Investigator / Author(s): Koontz, Michael D
Contractor: Geomet Technologies
Contract Number: A933-157
Research Program Area: Health & Exposure
Topic Areas: Indoor Air Quality, Modeling
The primary mission of the ARB Indoor Program is to identify and reduce Californians' exposures to indoor pollutants:1ts. To fully consider indoor exposures in assessing risk, the ARB needs estimates of average and peak indoor exposures for the general California population as well as certain subgroups of that population such as individuals who may be highly sensitive to indoor pollutants. The model described in this document- the California Population Indoor Exposure Model (CPIEM) -- is a software product that has been designed to expedite the exposure-assessment process by providing a user interface and calculation tools for supplying and integrating all required information.
The primary function of the CPIEM software is to combine indoor air concentration distributions with Californians' location/activity profiles to produce exposure and dose distributions for different types of indoor environments. This function, referred to as Level S1-2 of the model, is achieved through a Monte Carlo simulation whereby a number of location/activity profiles that were collected in prior ARB-sponsored surveys are combined with airborne concentrations for specific types of environments (e.g. residences, office buildings). For many compounds, the concentration data are either limited or nonexistent. Consequently, a second function of the model (Level 3) is to estimate indoor air concentration distributions based on distributional information for mass-balance parameters such as indoor source emission rates, building volumes and air exchange rates.
The CPIEM software has several unique features:
* Ability to estimate distributions of Californians' inhalation exposure and potential inhaled dose, with accumulation across multiple indoor environments.
* Estimation of exposure/dose distributions using concentration distributions that are measured through field studies or simulated within the model itself.
* Dynamic modeling of hourly and daily indoor-air concentrations in an indoor environment, taking into account various types of indoor sources as well as outdoor concentrations, air exchange rates, and losses to indoor sinks.
* Capability within the model for saving inputs and for future expansion, allowing the repository of input data to grow, as new information becomes available.
CPIEM was subjected to various verification and evaluation efforts under this project. Verification steps for Levels 1-2 indicated that inputs (activity profiles, concentration distributions) are properly accessed and used by the model, and that exposures and doses are correctly computed and accumulated across locations. Model estimates were evaluated utilizing data from an NO2 study in Los Angeles. The average personal exposure estimated from the simulation agreed closely with that based on the field study, but the standard deviation of the modeled exposure distribution was lower than that for the field study. This downward bias was expected because of the limited information available for constructing concentration inputs for environments other than the residence.
Estimates from Level 3 of the model were evaluated for three pollutants--chloroform, benzo[a]pyrene (BaP) and nitrogen dioxide (NO2)--for which sufficient source-related information and field studies for comparison were both available. The modeled standard deviation for chloroform, relative to the mean, initially was lower than that for field measurements. This is believed to be due to initial description of each type of water use-- toilets, faucets, showering/bathing, dishwashing and clothes washing--as a separate source in the model. When the chloroform sources were treated in this manner, the model sampled a different emission rate for each type of water use in a given residence, whereas the emission rates are likely to be very similar. When these sources were combined in order to use a common emission rate within each residence, the modeled ratio of the standard deviation to the mean better reflected the ratio based on measurements. There was a similar finding for NO2--the mean of the modeled distribution matched the measurements well but the modeled standard deviation initially was low relative to measurements. When separate sources that were initially defined for cooking breakfast, lunch and dinner were subsequently combined into one source, the modeled concentration distribution matched the measurement data very well. In both cases--chloroform and nitrogen dioxide--the combined sources accounted for the same consumption (liters of water for chloroform, Btu of fuel for nitrogen dioxide) as the individual sources when aggregated. These modeling outcomes suggest that similar types of indoor sources should be combined whenever possible. The principles on which both components of the model are based are scientifically and mathematically sound, but the accuracy of the outputs is limited by that of the inputs. Data on concentration distributions, needed for Levels 1-2 of the model, are not yet available for many of the environments. Even in cases where measured concentrations are available, there can be inaccuracies due to biases inherent in monitoring devices or sampling strategies. For Level 3 of the model, there is a notable lack of information at present for many of the indoor sources as well as pollutant-specific penetration factors and decay rates.
For questions regarding research reports, contact: Heather Choi at (916) 322-3893
Stay involved, sign up with ARB's Research Email Listserver