United States
Environmental Protection
Agency
National Risk Management
Research Laboratory
Research Triangle Park, NC 27711
Research and Development
EPA/600/SR-96/028
May 1996
EPA Project Summary
Development of a Radon
Protection Map for Large
Buildings in Florida
Kirk K. Nielson, Rodger B. Holt, and Vern C. Rogers
A radon protection map was devel-
oped to show from soil and geologic
features the areas of Florida that re-
quire different levels of radon protec-
tion for large building construction. The
map was proposed as a basis for imple-
menting radon-protective construction
standards in areas of high radon risk
and avoiding unnecessary regulations
in areas of low radon risk.
The map utilized 3,919 geographic
regions defined by digital intersection
of soil maps with surface geology
maps. Regional radon distributions
were modeled from radon source and
transport properties. Aeroradiometric
measurements from the National Ura-
nium Resource Evaluation (NURE) pro-
gram were digitally overlaid on the map
regions to estimate surface radium con-
centrations. Geologic classifications
and radium and emanation measure-
ments characterized deeper soils. Ra-
don transport properties (moisture, dif-
fusion coefficient, and air permeability)
were calculated from data in soil sur-
vey data bases. Indoor radon was mod-
eled as annual average concentrations
in a reference large building. The refer-
ence building was modeled on the soil
and moisture profiles of each region to
determine regional radon potentials.
Confidence limits for the regional ra-
don distributions were calculated from
variations in the regional radon source
and transport properties.
Separate model analyses estimated
the effectiveness of different building
construction features in reducing ra-
don entry. Radon resistance factors for
each feature were ranked and ordered
to select a group of passive features
having a combined radon reduction of
a factor of 3.3. A cost-benefit analysis
used the feature effectiveness with re-
gional radon variations to estimate a
95% confidence limit for optimum use
of radon-protective features. The map
was divided into green, yellow, and red
tiers corresponding to regions with low
(<4 pCi L1), intermediate (4 to 13 pCi L1),
and elevated (>13 pCi L1) potential for
annual average radon concentrations
at the 95% confidence limit: the green
tiers comprised 3,650 of the 3,919 re-
gions, including 92.9% of the state area;
yellow tiers comprised 223 regions, in-
cluding 6.1% of the state area; and red
tiers, 46 regions, including 1.0% of the
state area.
The map was compared with over
275,000 measurements in 20,156 large
buildings. A statewide bias of only
-0.004 + 1.067 standard deviations sug-
gests excellent average agreement. Ob-
servations of 306 buildings with the
greatest bias showed that, with crawl
spaces, 89% measured low and only
11% measured high. For slab-on-grade
buildings, 48% measured low and 52%
measured high. The number of outly-
ing comparisons was consistent with
the number expected in the extremes
of the bias distribution.
This Project Summary was developed
by EPA's National Risk Management
Research Laboratory's Pollution Pre-
vention and Control Division, Research
Triangle Park, NC, to announce key
findings of the research project that is
fully documented in a separate report
-------
of the same title (see Project Report
ordering information at back).
Introduction
Radon (222Rn) gas from the decay of
naturally occurring radium (226Ra) in soils
can enter indoors through building foun-
dations. With elevated entry and inad-
equate ventilation, radon can accumulate
to levels that pose significant risks of lung
cancer with chronic exposure. The Florida
Department of Community Affairs (DCA)
is developing radon-protective building
standards to reduce radon-related health
risks. Statewide radon maps are also be-
ing developed to target the standards to
regions of greatest need. This report de-
scribes a large building radon protection
map that was developed to show where
radon-protective building standards are
needed and to avoid unnecessary stan-
dards where they are not needed.
Several institutional and scientific crite-
ria led to the present technical approach.
The maps had to identify as precisely as
possible the regions needing radon-pro-
tective building features. The maps also
had to avoid political and institutional
boundaries (city, county, etc.) that are not
radon-related. The maps were based on
radon entry into a reference large building
to accurately reflect regional soil and mois-
ture effects. The approach used to achieve
these objectives is similar to that used
previously for the residential radon pro-
tection map. It therefore capitalizes on
existing region, soil, geology, and radio-
logical definitions, and involves:
a. Using the previously defined map
polygons (defined from existing soil
and geologic maps).
b. Using the previously defined soil
profiles associated with each radon
map polygon, and their associated
radon generation and transport
properties.
c. Calculating numeric indoor radon
concentrations for individual soil
profiles and a time- and area-
weighted average to represent each
radon map polygon.
d. Using cost-benefit analyses and
feature effectiveness rankings to de-
termine cut points for different lev-
els of radon protection.
e. Plotting the large building radon pro-
tection map by color-coded radon
protection tiers.
Indoor radon levels for large-buildings
were calculated using new model analy-
ses for a reference large building along
with previous statewide calculations for
houses. The new analyses were performed
by the Florida Solar Energy Center (FSEC)
using a representative subset of soil ra-
don generation and transport properties.
The analyses modeled the same refer-
ence large building on the soil and mois-
ture profiles of each geographic region to
reflect the regional differences in radon
potential without interference from build-
ing differences. The radon calculations uti-
lize data developed cooperatively by the
University of Florida (UF), the Florida Geo-
logical Survey (FGS), the FSEC, the U.S.
Geological Survey (USGS), and Rogers
and Associates Engineering Corp. (RAE)
working together in the Florida Radon Re-
search Program (FRRP).
This report presents the approach and
data used for the large-building radon pro-
tection map; the radon calculations from
multi-zone large-building modeling and ex-
isting residential modeling; the radon re-
sistant construction features and their ef-
fectiveness and ranking; the cost-benefit
analysis; the map technical definition; and
the map validation.
Radon Distributions in the
Reference Large Building
The FSEC 3.0 model was used for the
multi-zone, large-building model calcula-
tions. For computational economy, the cal-
culations were performed on only a repre-
sentative subset of 11 soil series under
different seasonal moisture conditions and
different radon source combinations (a to-
tal of 124 model analyses). The complete
soil and moisture details were then inter-
polated by fitting the large-building results
to the previous Radon Emanation and
Transport into Dwellings (RAETRAD) resi-
dential analyses. Regression coefficients
for computing the statewide distributions
from the large-building analyses averaged
0.99, and comparisons of individual and
fitted calculations had a correlation coeffi-
cient of 0.998.
The FSEC 3.0 calculations utilized ac-
tual soil properties for the top 2.0 to 2.5 m
and extrapolated soil properties for the
remaining profile to a depth of 5.0 m.
Different radon source strengths were al-
lowed for the upper and lower zones. Soil
moistures throughout the entire 5-m pro-
file were defined for a high water table
season, a low water table season (2 m
lower), and 1-month intermediate-depth
transition periods between seasonal lev-
els. The soil moisture contents were cal-
culated for each soil series from water
drainage data.
The model analyses represented the
reference large building as if it were lo-
cated on each soil profile under each
specified moisture and radon source con-
dition. The reference large building was
defined and analyzed to have the approxi-
mate characteristics of the Polk Life and
Learning Center located in Bartow, Florida.
The reference building consists of a single-
story structure with a floor area of 1,808
m2 (19,456 ft2), and normalized dimen-
sions of 1.6 x 1 x 0.12. The building has a
floating-slab floor design with a 0.15 mm
(6-mil) polyethylene vapor barrier between
the slab and soil. Slab openings included
a perimeter shrinkage crack (0.0007 m2
crack area per m2 floor area), and cen-
trally located slab openings totaling 0.0004
m2 open area per m2 of floor area. A
single air handler circulated internal and
outdoor air through six zones. Two occu-
pied zones were at -2 Pa and +2 Pa
pressure, and a mechanical room was at -5
Pa and contained 67% of the floor leak
area. The FSEC model calculations were
compared previously with indoor radon
measurements in the same building.
Soil radium concentrations were defined
as the geometric means of NURE mea-
surements for estimating median radon
potentials in all map regions containing
NURE data. Estimates for regions without
NURE data were defined from the geo-
metric mean for the geology unit contain-
ing the map region. In the few cases where
NURE data missed an entire geology unit,
the geometric mean for the regional soil
unit was used. The geologic definitions for
deep-soil radium classifications were de-
veloped by the USGS.
Radon Resistant Construction
Features
The calculated statewide radon distri-
butions readily show regional geographic
differences, but their absolute values ap-
ply only to the reference large building.
Actual radon levels in new buildings will
differ because of design and operational
differences, and they can be significantly
reduced by incorporating radon-resistant
features.
Construction features for reducing in-
door radon levels are categorized as ei-
ther passive or active. Passive features
act as barriers to radon entry into the
building. They generally have no mechani-
cally operating parts that can be disabled,
and they require no energy costs or user
attention (except normal maintenance,
such as re-caulking). An example of a
passive radon control feature is the caulk
seal around pipe penetrations in a floor
slab. Active radon controls, on the other
hand, generally require mechanical op-
eration and incur energy costs for electric
fans or supplemental heating and cooling.
An example of an active radon control
-------
feature is a fan-driven sub-slab depres-
surization system.
Different passive and active radon con-
trol features reduce radon entry to differ-
ent extents. Furthermore, certain features
such as sealed floor cracks are unlikely to
be implemented completely, but are still
effective to the extent that they are uti-
lized. For the large building analyses, five
passive and three active radon control
features were analyzed. The passive fea-
tures include (a) sealed return ducts in
the air handling system, (b) sealed slab
penetrations, (c) sealing of most (80%)
floor slab cracks, (d) monolithic design of
the floor slab and foundation stem wall,
and (e) use of low-slump (10-cm) con-
crete for the floor slab. The active fea-
tures include (a) an active sub-slab de-
pressurization system, (b) increased build-
ing ventilation by outdoor air, and (c) pres-
surization of the building interior.
The effectiveness of the radon control
features was analyzed with the FSEC 3.0
model. In each simulation, the reference
large building was placed on a soil profile
with the desired water table depth, and
indoor radon was averaged for the occu-
pied building areas (zones 1 and 2). The
effectiveness of each feature in reducing
indoor radon was defined as a radon re-
sistance factor, which was the ratio of
radon concentrations in the reference large
building without and with the feature:
RRF,.= Co/C,. (1)
where:
RRF; = radon resistance factor
for feature / (dimension
less),
Co = radon concentration
without feature /(pCi L~1),
and
Cj = radon concentration with
feature / (pCi L1).
Significant synergism between features
involved interactions with the foundation
design. Analyses for only floating slab de-
signs gave conservatively lower effective-
ness factors than if other slab designs
were considered. Therefore, analyses rela-
tive to floating slab floors were used in
estimating and ranking the effectiveness
of each radon resistance factor. The RRF
factors averaged 1.98 for sealed return
ducts, 1.75 for sealed slab penetrations,
1.64 for mostly sealed perimeter floor
cracks, 1.64 for monolithic slab and stem
wall, and 1.16 for reduced-slump concrete.
RRF factors for active features were 4.5
for active sub-slab depressurization, 2.33
for increased ventilation by outdoor air,
and 2.31 for building pressurization.
The return duct sealing feature was re-
moved from the passive feature group for
the radon standard because duct sealing
is already required by the Florida Energy
Code. The RRF for perimeter crack seal-
ing was combined with that for monolithic
slab design because of their identical mag-
nitude and their interaction when com-
bined. The overall RRF for the passive
feature group was therefore defined as
3.3 from the product of 1.75x1.64x1.16.
This factor is used in the cost-benefit analy-
sis to estimate benefits from passive ra-
don controls. The passive features can
therefore reduce radon levels in large build-
ings from as much as 13 to less than 4
pCi L1. Similar grouping and selection of
active radon control features is not re-
quired for the large-building radon protec-
tion map.
Cost-Benefit Analysis
The radon distributions calculated for
each geographic region are compared with
the 4-pCi L1 radon standard to determine
the potential need for radon control fea-
tures. Although the distribution medians
could be used for this comparison, medi-
ans just below the 4-pCi L1 level would
permit up to half of the polygon's land
area to exceed the standard without re-
quiring radon controls. A more conserva-
tive approach is desirable; however, the
costs of requiring radon controls in more
buildings are less effective because they
would include increasing proportions of
buildings in low-radon-potential areas.
Therefore, a cost-benefit analysis was per-
formed to determine the appropriate con-
fidence limit to apply to the regional radon
distributions for comparing them to the
standard. The confidence limit constitutes
a safety factor that is based on the geo-
graphic variation in regional radon poten-
tial.
The cost-benefit analysis optimizes the
trade-off between added costs for radon
controls and health benefits from reduced
radon levels. It assumes complete compli-
ance in regions where radon controls are
required. A cost-benefit analysis previously
helped determine the indoor radon reme-
dial action limit of 4 pCi L1. Given this
limit, an upper confidence limit was de-
fined for regional radon levels that bal-
anced the benefits from reduced radon
exposure with the costs of the radon con-
trols in an entire region. The confidence
limit was determined as for the residential
map using the following iterative approach:
Step 1. Choose an initial confidence
limit.
Step 2. Eliminate those regions (map
polygons) of the state identi-
fied by the statewide radon
distribution calculations to
have indoor radon concen-
trations of less than 4 pCi L1
at the selected confidence
limit.
Step 3. Determine the confidence
limit for the remaining areas
for which the health benefit
from implementing radon con-
trols equals the cost for imple-
menting the controls.
Step 4. Replace the initial confidence
limit chosen in step 1 with
the final confidence limit cal-
culated in step 3.
Step 5. Repeat steps 2 through 4 un-
til the assumed initial confi-
dence limit equals the calcu-
lated final confidence limit.
A cost of $90 per 93 m2 (1,000 ft2) of
building footprint was estimated from UF
data for sealing cracks and penetrations.
An additional cost of $35 per 93 m2 of
building footprint for improved concrete
brings the passive-group total to $125 per
93 m2 The cost of installing an active
sub-slab depressurization system is ap-
proximately $50 per 93 m2 of building foot-
print.
A statewide distribution of soil-related
radon concentrations in buildings was de-
termined by combining individual distribu-
tions from each region. The resulting state-
wide distribution was integrated to show
cumulative totals. For example almost 90%
of the state would cause radon levels of
less than 1.75 pCi L1 in the reference
building. Based on the EPA action level,
potential radon levels exceeding 4 pCi L1
require at least passive radon controls.
Multiplying by the passive effectiveness of
3.3 indicates that radon levels exceeding
13 pCi L1 require the additional control of
active radon controls. Other parameters
used in the cost-benefit analysis include
an average occupancy factor (3.33 per-
sons per 93 m2), an effective annual ben-
efit from radon reduction (4.32x10~5 lives
saved per person pCi L1 year of exposure
reduction), and an average cost for each
life saved ($700,000 per life).
The cost-benefit analysis gave a confi-
dence limit of 94.3%, indicating that if at
least 5.7% of a region would exceed 4 pCi
L1, it is cost-effective to institute passive
radon control features throughout the re-
gion. For convenience in preparing the
radon protection map, a conservative con-
fidence limit of 95% is recommended and
-------
used. The EPA's traditional recommenda-
tions of a 95% upper confidence limit for
data used in risk assessment calculations
are consistent with this value.
Large Building Radon
Protection Map
The large-building radon protection map
divided the 3,919 map regions defined
previously into three radon protection cat-
egories, based on their numerical radon
concentrations at the 95% confidence limit.
Regions with limiting radon levels below
the 4 pCi L"1 limit were colored green to
designate low radon potential. Limiting ra-
don levels in the remaining polygons were
then colored yellow to designate interme-
diate radon potential if they were below
the 13 pCi L1 threshold, or red to desig-
nate elevated radon potential.
The green category, which requires no
radon controls beyond adherence to ex-
isting building codes, consists of 3,650
regions (92.9% of the state area). The
yellow category, which requires passive
radon controls, consists of 223 polygons
(6.1% of the state area). The red cat-
egory, which requires both passive and
active radon controls, consists of 46 poly-
gons (1.0% of the state area).
The large-building radon protection map
includes slightly less area in the red cat-
egory, but more than twice as much area
in the yellow category as the previous
residential radon protection map, because
of two competing differences. The larger
yellow category resulted from consistently
higher average radon levels in the large-
building simulations than in the previous
residential radon simulations. The slightly
smaller red category resulted from a
greater effectiveness of passive radon con-
trols in large buildings than in residences
(3.3 versus 2.1), thereby making passive
controls adequate in a few regions where
active controls were needed for resi-
dences.
Statewide Validation of the
Large-Building Radon
Protection Map
The large-building radon protection map
was validated by comparing radon mea-
surements with the mapped radon distri-
butions. Despite construction and occu-
pancy differences between the reference
and measurement buildings, indoor radon
data from the Florida Health and Rehabili-
tative Services (MRS) large-building data
base provided the most direct compari-
sons. The data base contained over
300,000 individual measurements in nearly
22,000 buildings throughout Florida in the
November 1994 comparison set. Although
dominated by population centers (49% of
the buildings in 6 of the 67 counties), data
were sufficient to represent parts of most
counties.
The radon comparisons utilized a bias
statistic to normalize each comparison to
a common basis. The normalization was
required because radon distributions in
some polygons had much greater vari-
ance than in others. The bias statistic
used for the comparisons had the form:
Z = [In(Meas) - In(Map)]/
(2)
where:
Z
Meas
Map
measurement-map bias
statistic (standard deviations),
value of the measured pa-
rameter (point in time),
median value of the mapped
distribution (annual average
basis),
uncertainty [geometric stan-
dard deviation (GSD)] in rep-
resenting the annual average
by a measured value, and
GSD of the mapped distribu-
tion.
The estimate of GMeas was calculated
from previous comparisons of short-term
and annual-average indoor radon esti-
mates in houses, and was adjusted for
the difference between house and large-
building occupancy using data for schools.
The resulting estimate of GMeas = 2.2 was
used statewide for estimating the uncer-
tainty in annual average radon concentra-
tions from a single measurement in a large
building. The statewide distribution of bias
statistics averaged -0.004, showing that,
on average, the measurements and cal-
culated values were equivalent. The stan-
dard deviation of the Z distribution was
1.067, indicating that the variations be-
tween the measured and calculated val-
ues were essentially equivalent to the
variations predicted from the map and
measurement uncertainties.
A search for potential radon anomalies
used the distribution of comparison statis-
tics. Buildings with very large positive or
negative Z statistics were identified, and
their locations and attributes were com-
piled from MRS data to identify possible
trends or regional clusters. Nine counties
in three parts of the state contained 313
of the 422 identified buildings. Of these,
168 were potential negative anomalies
(measurements lower than calculated val-
ues) and 145 were potential positive
anomalies (measurements higher than cal-
culated values).
Non-intrusive field observations were
made at 306 of the buildings. The con-
struction type, adjacent pavement cover-
age, type of entry doors, building purpose
(for occupancy estimates), and the eleva-
tion difference between grade and the first
floor were recorded. Gamma radiation and
latitude/longitude coordinates were also
measured. Several significant trends were
observed. Buildings constructed with crawl
spaces accounted for only 15% of the
observed buildings (47 total), but of these
buildings, 89% were potential negative
anomalies. Slab-on-grade buildings ac-
counted for 257 buildings, or 84% of the
observed buildings, with almost equal num-
bers of potential positive and negative
anomalies. A total of 10 observed build-
ings were frame construction, and all 10
were potential negative anomalies. Simi-
larly, nine buildings were observed that
were of poured-concrete construction, and
eight of these were potential negative
anomalies. With exposed (no veneer) con-
crete block construction, 85% of the build-
ings were potential positive anomalies, and
only 15% were potential negative anoma-
lies.
The overall comparison of the measured
and mapped radon levels is excellent. Ap-
proximately 69% of the 20,156 points sur-
veyed lie within 1 standard deviation of
their mapped value, compared to 68%
expected from map and measurement un-
certainties. Similarly, 93% were within 2
standard deviations of mapped values,
compared to approximately 95% expected
from map and measurement uncertain-
ties. Partitioning the field observation cases
into potential negative or positive anoma-
lies gives further insight.
-------
Kirk, K. Me/son, Rodger B. Holt, and Vern C. Rogers are with Rogers and
Associates Engineering Corp., Salt Lake City, UT 84110.
David C. Sanchez is the EPA Project Officer (see below).
The complete report, entitled "Development of a Radon Protection Map for Large
Buildings in Florida," (Order No. PB96-168 216; Cost: $31.00, subject to
change) will be available only from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Air Pollution Prevention and Control Division
National Risk Management Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
United States
Environmental Protection Agency
National Risk Management Research Laboratory (G-72)
Cincinnati, OH 45268
Official Business
Penalty for Private Use $300
BULK RATE
POSTAGE & FEES PAID
EPA
PERMIT No. G-35
EPA/600/SR-96/028
------- |