Upload
ron-pearson
View
65
Download
3
Tags:
Embed Size (px)
Citation preview
Advantages and Limitations of Indoor Air Quality
Mathematical Models to Support Risk Assessments
Presented at: Northland Chapter, Society of Toxicology Meeting
March 10, 2015
Presented by: Ron Pearson, M.S., CIH
Environmental Health & Safety, Inc.
St. Paul, Minnesota
Today’s Presentation♦ Background information on IAQ Models –
pertaining to direct chemical use (not ‘environmental’) in industry or by consumers
♦ “Generalities and Realities”
♦ Tiered approach, types of IAQ models
♦ Model “validation”
♦ Model reliability – ART
♦ Model reliability – IHMOD
♦ Conclusions/future directions
3
(at Equilibrium)
Concentration(C)Generation Rate(G)
Ventilation(Q)=
Simple Box Model
Models are essentially mathematical equations that can estimate the concentration of a
contaminant in a space based on physical and chemical input parameters….
“All models are wrong, but some are useful…”
George. E. P. Box, British mathematician and pioneer in
Bayesian Statistical methods and quality control
4
EU REACh (Registration, Evaluation and Authorization of Chemicals)
♦ Probably the largest current driver of model use (along with California Prop65)
♦ Requires risk assessments for new and existing hazardous substances used in industry
5
IAQ Model Generalities / Realities♦ Unlike ‘environmental’ exposures (e.g.
groundwater, drinking water, food) the predominant exposure route for most industrial, commercial and consumer users of products (except for things like a topically applied cosmetic, or possibly certain industrial chemicals such as isocyanates) is likely to be inhalation
♦ While dermal maybe significant, it is also more complex than inhalation and less ‘validatable’
IAQ Model Generalities / Realities
♦ Modeling mixtures is problematic – they add complexity, both from the standpoint of multiple health endpoints and physicochemical parameters (e.g. generation rate) for estimating a ‘true’ representation of measured exposures
♦ IAQ models generally work much better for gases/vapors than particulate air contaminants
7
IAQ Model Generalities / Realities♦ Most users are not full-time users – their
model outputs inherently have weaknesses because of this basic lack of user familiarity
♦ Most users have no formal training in their use
♦ In many cases it is actually LESS expensive to get actual monitoring data than to hire a true “modeling expert” (BUT, we need to acquire an adequate number of samples to get statistically defensible data) – an exception may be very simplistic screening models
8
Common modeling mistakes♦ Initial model selection
– Use of a screening model used for exposure estimation
– Use of a complex model when a screening model will suffice
♦ Incorrect scenario identification
♦ Misinterpretation of input parameters
♦ Inaccurate input estimates
♦ Lack of understanding of the input parameters that have the greatest sensitivity
9
IAQ Model Generalities / Realities
♦ In general, there is a perception (mostly justified – especially with “Tier I” models) that the model results are biased ‘high’ and result in an overestimate of potential exposures
10
Understanding our goals / objectives for model use are critical
♦ Legal?
♦ Regulatory?
♦ Planning/prioritization?
11
Some commonly used IAQ Models
♦ EASE (Estimation and Assessment of Substance Exposure)
♦ CONSEXPO (Consumer Exposure Model)♦ MCCEM (Multi-Chamber Concentration and
Exposure Model)♦ ECETOC TRA♦ IH MOD (AIHA published)♦ COSHH Essentials (Control of Substances
Hazardous to Health)♦ Advanced Reach Tool (ART)♦ And many others….
12
IAQ Models should ideally be part of a “Tiered Approach” for
estimating exposures♦ Viewed by many environmental scientists as
the most legitimate way of realizing the utility of models
♦ We may utilize a ‘first tier’ model to conduct a gross ‘screening’ of an exposure situation
♦ Tier 1 Examples:
♦ ECETOC
♦ Stoffenmanager13
Tier 2 Models
♦ If we have more time/money/expertise, we may conduct a more refined analysis using a ‘second tier’ model:
♦ Tier 2 Examples:
♦ 2 Zone Model (IHMOD)
♦ Computational Fluid Dynamic Models
14
How do you “validate” an IAQ model?♦ We can evaluate them by internal and/or
external validation studies.
♦ Internal studies:– Look at underlying assumptions of models – do
they correlate well with established theories?
– Are parameters considered in their application valid?
♦ External studies:– Compare values predicted by a model with
quantitative data from air monitoring – how well do they agree and under what varying conditions? 15
So….how well do the models work? Let’s start with ART (Advanced
Reach Tool)!
♦ ART was designed to be used to estimate exposure levels for specific scenarios for groups of workers with common operational conditions and risk management measures (RMMs) in different workplaces (some refer to it as a ‘Tier 1.5’ model
♦ A study published in 2014 (Schinkel, et.al.) looked at ‘reliability’ (not ‘validity’)
16
ART
♦ Validation: compares the model estimates of exposure to measurement data (accuracy and bias)
♦ Reliability: is a measure of the consistency of assessments or of the ability of different assessors to reach the same conclusions about a specific case or scenario
17
ART Reliability Study examined 4 primary issues
1. Inter-assessor exposure estimates with ART
2. Inter-assessor agreement per model parameters
3. Individual assessor characteristics resulting in reliable estimates
4. Effect of training on assessor agreement
18
ART Reliability Study♦ Assessors were assigned to one of four
groups with comparable education, work experience, profession, age, and # years experience with other exposure assessment tools
♦ All had a minimum of a Master’s degree
♦ Half looked at solids/dust, half look at volatile liquids
19
ART Reliability Study
♦ Four scenarios (workplace conditions described by text and videos) – assessed by ~50 participants
– pre and post-training (3 hr. training session)
– their model outputs were compared with “gold standard” estimates compiled by the workshop instructors
20
ART Reliability Study
♦ Primary findings:
♦ No statistically significant difference in exposure estimates before and after training
♦ Large differences were observed between scenarios, e.g. ranging from 7% to 85% of assessors within a factor of three from the “gold standard” estimate
21
22
Conclusions♦ ART is an expert tool and extensive training
is recommended prior to use
♦ Improvements of the guidance documentation, consensus procedures, and improving the training methods could improve the reliability of ART
♦ Considerable variability can be expected
between assessors using ART to estimate exposure levels for a given scenario
23
IH MOD 2 Zone Model
♦ Tier II Model – published by AIHA Subcommittee on IAQ
♦ Estimates “near field” and “far field” air concentrations
♦ Assumes: the airborne concentration nearer an emissions “source” within a room is most often considerably greater than it is at farther points within the same room
24
IH MOD 2 Zone Model
♦ 2011 paper published by Jayjock, et.al.
♦ Looked at acceptability of the model for the purpose of the Daubert standard (governing admissibility of expert witness testimony during U.S. federal legal proceedings/many State proceedings)
♦ Model is based on simpler well-mixed box model
25
IH MOD 2 Zone Model♦ Application “bounding conditions”:1. Contaminant is instantaneously mixed
throughout the near-field and far-field work space (air concentration is uniform within each of the two spaces)
2. There is limited airflow between the two zones
3. The random air velocity between the two zones is uniformly distributed across the NF/FF interface surface
4. There are no significant cross drafts
26
Comparison of Modeled v. Measured Concentrations
27
IH MOD 2 Zone Model
Conclusion: The NF/FF model prediction was usually within the approximate range of 0.5-to 2-fold the measured concentration
28
Concluding Remarks
♦ If exposure estimates are judged to be acceptable (inaccurately), worker and/or end-user health could be compromised
♦ If exposure estimates are judged to be unacceptable, risk management measures (including engineering controls) could be in incorrectly applied (over-engineering) and time/money/resources wasted accordingly
♦ Ideally, exposure estimates will be both accurate and precise
29
Concluding Remarks
♦ Even sophisticated models have shortcomings for a variety of reasons
♦ IF modeling will be conducted, the most important thing is selecting a model that suits the needs of the assessment objectives
♦ Obtaining actual data (that is statistically sound) will generally provide a higher degree of confidence prior to implementing controls or pronouncing a given product and/or work environment as being ‘safe’
30
Future Research Needs
♦ Identification of which models work best for different scenarios
♦ Comparison of models “side-by-side” with actual air quality measurements under a wide variety of conditions in order to better characterize their uncertainty in predicting exposures
31
References♦ The Daubert Standard as Applied to Exposure Assessment Modeling Using the Two-
Zone (NF/FF) Model Estimation of Indoor Air Breathing Zone Concentration as an Example; Jayjock, M., et.al.; Journal of Occupational and Environmental Hygiene, October, 2011
♦ Reliability of the Advanced REACH Tool (ART); Schinkel, J., et.al.;Ann. Occup. Hyg., 2014, Vol. 58, No. 4, 450–468, January, 2014
♦ Global Net on “CONSUMER EXPOSURE MODELLING”; Report of the Workshop no. 2 on Source Characterization, Transport and Fate; European Commission Joint Research Centre, June, 2005
♦ Modeled Comparisons of Health Risks Posed by Fluorinated Solvents in a Workplace Spill Scenario; Jayjock, M., et.al.; Ann. Occup. Hyg., pp. 1–12, July, 2010
♦ Evaluation and Further Development of EASE Model 2.0, Kreely, K., et.al.; Annals of Occupational Hygiene 2005 49(2):135-145;
♦ Workshop Validating Exposure Risk Assessment Models, Logan, P., Hewett, P.; British Occupational Hygiene Society, April, 2012
♦ Evaluation of the Predictive Abilities of a Qualitative Exposure Assessment Model, Elliot, L., Oestenstad, K. ; Journal of Occupational and Environmental Hygiene, Nov., 2007
♦ Introduction to models and IH MOD, a new tool to aid industrial hygienists in exposure assessments; presentation/workshop by Armstrong, T.
32