Setting
This intervention was conducted in an elementary school occupied by approximately 455 students in kindergarten through 5th grade and staff. The 6,000 square meter (m2) single-story school building contains 24 classrooms and common areas constructed circa 1961 and two masonry and three modular classrooms built after 1978. The envelope of the original structure consists of alternating sections of brick over concrete block and curtain wall. The curtain wall is composed of aluminum framing, single pane glass windows, and mineral fiber (transite) panels. Classrooms, the library, gymnasium, and kitchen are located along the perimeter of the building. Hallways, specialty rooms (e.g., music and art), staff offices and restrooms comprise the remaining interior space.
Unit ventilators with internal fans and steam radiators are used to ventilate and heat the classrooms. The building is not air conditioned. Each classroom has a thermostat that establishes the temperature set point for the ventilation and heating systems. During the heating season, the steam boiler is powered and thermostat set points are set to 20 degrees Celsius (°C). At other times, thermostats are set to 17°C. When set points are met, the outdoor air dampers in the unit ventilators supply an approximate 50:50 mix of recirculated air and outdoor air to classrooms. When room temperatures exceed thermostat set points, outdoor air dampers open fully and supply air is up to 90% outdoor air. Central exhaust systems operate continuously on days when school is in session and remove an average of 5.1 cubic meters of air per minute (m3/min) from each classroom.
Transite panels in the curtain wall are set in PCB-containing caulk that fills the channels of aluminum framing. Bulk samples of the caulk were analyzed by EPA Method 8082 and found to contain 1,830 - 29,400 parts per million (ppm) of total PCBs, primarily as Aroclor 1260. The number of transite panels in each classroom ranged from 3 to 9, depending upon the size and configuration of the room. With each transite panel having an area of 0.8 m2, there was 10 to 30 linear meters of caulk on the interior face of the exterior wall in each room. Beads of caulk were approximately 5 millimeters (mm) wide, totaling 0.05 to 0.16 m2 of PCB caulk on the interior of each room. Fluorescent light ballasts in the building did not contain PCBs. PCBs were found in ceiling tiles, plastic cove base, and floor tile mastic as well, but at concentrations less than 150 ppm and generally two orders of magnitude lower than the levels present in caulk.
Design
Three interventions intended to mitigate concentrations of PCBs in indoor air were evaluated in series: ventilation; contact encapsulation; and physical barriers. Indoor air samples from multiple treatment and control classrooms were collected before and after each intervention and analyzed for PCBs. Prior to the first intervention, air sampling was conducted on August 25 - 27, 2010 during which air exchange rates in individual classrooms ranged from 0.4 to 5.9 per hour (h-1) and the sampling period-average ambient temperature ranged between 18 - 25°C.
Ventilation
Following the pre-intervention sampling, ventilation rates in classrooms were increased by replacing filters, repairing fans in the unit ventilators (Figure 1, Panel A), and adding supplemental ventilation with high efficiency particulate air (HEPA)-filtered outdoor air in one room. Post-intervention sampling for ventilation was conducted on September 6, 2010. During the sampling, air exchange rates (AER) in classrooms ranged from 2.7 to 12 h-1 and ambient temperature was 24°C.
Contact encapsulation
After adjusting ventilation rates to within design levels in classrooms throughout the school, we evaluated whether encapsulating PCB caulk would influence PCB levels in indoor air. For this objective, interior-facing beads of caulk were sealed with adhesive-backed polyethylene tape which in turn was covered with a bead of silicone caulk (Figure 1, Panel B). Encapsulation was completed and evaluated in two stages. First, PCB caulk directly accessible to classrooms (approximately 75% of interior caulk in each classroom) was encapsulated (left side of Panel C, Figure 1). Afterwards, indoor air samples were collected on September 19, 2011 when the ambient temperature was 21°C. Next, PCB caulk located behind unit ventilators and convective heaters was encapsulated (right side of Panel C, Figure 1) and indoor air samples were collected on September 29, 2011 at an ambient temperature of 25°C.
Physical barrier
For the third intervention, a false wall was constructed over the encapsulated caulk (Figure 1, Panel D). In addition to the potential to reduce indoor air PCB concentrations further, the intent of the false wall was also to limit direct contact with and disturbance of encapsulated PCB caulk. To construct a false wall, aluminum-backed fiberglass insulation board was inserted over the interior face of a transite panel and sealed with silicone caulk. Gypsum board was affixed over the insulation board and the surrounding aluminum frame. The gypsum board was sealed with silicone caulk and then coated with latex paint. False walls were constructed in classrooms during November 2010. Post-intervention sampling was conducted after construction of the false walls was completed. Indoor air samples were collected on November 4, 11, 20, and 24, 2010, and December 2, 2010. Corresponding ambient temperatures were 9.3, 8.7, 9.7, 6.7, and 1.7°C, respectively.
Longitudinal assessment
To evaluate whether the mitigation measures were effective over time, we conducted longitudinal follow-up sampling in the building. Nine indoor sampling events were carried out from February 23, 2011 through December 29, 2011; ambient temperatures ranged from -2.1 to 28°C. Three to ten classrooms were sampled during each event.
Data collection
Ninety-six (96) samples of school-day average PCBs in indoor air of the elementary school were collected during 18 sampling events from August 25, 2010 - December 29, 2012. The samples were collected in accordance with EPA Method TO-10A. School-day in this context corresponds to 6.5 hour sampling durations that typically began between 9:00 a.m. and 10:00 a.m. An outdoor air sample was collected on each day of indoor air sampling. Sampling pumps and cartridges were suspended from aluminum tripods 1 meter above floor level. Pump flows were set for 4 liters per minute and verified by a calibrated flow meter [Bios Drycal] at the start and end of the sampling period. The beginning and ending flow rates were averaged to calculate the total volume of air sampled. Samples were assayed by gas chromatography-mass spectrometry for PCB homologs following EPA Method 8270 C-SIM [Alpha Analytical, Mansfield, MA]. Results were reported in nanograms per cubic meter (ng/m3) for each homolog and for total PCBs as the sum of the individual homologs. Homologs reported by the laboratory as non-detect were treated as zero when computing total PCB concentrations in air.
Windows and doors were closed and unit ventilators were operated at a fixed fan speed during all sampling events. Outdoor air delivery rates through the unit ventilator in each room was measured with a calibrated balometer and used to estimate AER. The accuracy of the AER estimates was confirmed by tracer gas following the American Society for Testing and Materials (ASTM) Standard E741-00, Standard Test Method for Determining Air Change Rate in a Single Zone by Means of a Tracer Gas Dilution Method conducted in a subset of the classrooms. The average ambient temperature during each sampling event was obtained from an Automated Surface Observing Station (ASOS) weather station located 4 km west of the school.
Data analysis
Indoor air samples from multiple classrooms were collected before and after each intervention and analyzed for total PCBs. All total PCB concentrations were greater than the limit of detection. To ensure comparability of data among monitoring periods, the analysis was based on rooms in the original building with operating unit ventilators and transite panels set in aluminum framing with PCB caulk.
Ventilation rate and temperature have been reported elsewhere to have an influence on PCB levels in outdoor and indoor air [17–19]. To account for this effect when evaluating efficacy of interventions, measured concentrations were normalized from the measured AER to an AER of 11.3 m3/min and for vapor pressure from indoor temperature to a temperature of 20°C. On sampling days that ambient temperature exceeded 20°C (the thermostat set point) indoor temperature was assumed to be equal to ambient temperature. On other days, indoor temperature was assumed to be 20°C. PCB vapor pressure was estimated with the Clausius-Clapeyron equation from temperature-specific vapor pressures compiled by Li [20] and a homolog mixture for Aroclor 1260 reported by ATSDR [21]. The heat of evaporation estimated for Aroclor 1260 was 82 kilojoule per mole (kJ/mol).
Pre- and post-intervention measurements were not always made in the same room. Therefore, two analyses were used to evaluate pre-intervention and post-intervention concentrations of PCBs in indoor air. In one analysis, data from all rooms sampled pre- and post-intervention were used to test the null hypothesis that median concentrations of PCBs in indoor air were equal before and after an intervention (Wilcoxon Rank Sum test). In the second analysis, paired pre- and post-intervention measurements were analyzed using the sign test which allowed for control of room-specific factors that might influence PCB concentrations (n = 3-8 pairs depending on the intervention).
The analysis of variability over time was performed by first compiling summary statistics for airborne PCB levels, AER, surface area of PCB caulk in a classroom, and ambient temperature. A linear mixed effects model controlling for repeated measures from individual classrooms was used to test for significant variability of indoor air PCB concentration with ambient temperature, AER, and surface area of caulk per classroom. The relationship between ambient temperature and PCB levels in indoor air was evaluated through additional analysis with regression models.
Quality assurance
To ensure traceability and accuracy of the data, a series of quality assurance steps was performed. A chain of custody (COC) form followed each air sample from the field, to the laboratory, and finally to the database manager. Sample collection and analysis procedures were performed in accordance with quality assurance measures prescribed by EPA Method TO-10A. The DL ranged from 5-10 ng per homolog (nominally 10 ng/m3 in equivalent air concentration) over the course of investigation. Recovery efficiency was measured using fortified samples. Two surrogates, Cl3-BZ#19-C13 and Cl8-BZ#202-C13, were added to each sample and their average recoveries were 93% (SD = 22%) and 86% (SD = 18%), respectively (N = 142). PCBs were not detected in any field blank samples (N = 18). The precision of samples was expressed as the root mean square error (RMSE) between paired primary samples and duplicates. Paired samples showed good agreement with a slope of 0.97, RMSE of 36.5 ng/m3, and coefficient of variation of 14.6%. An outdoor air sample was collected during each sampling event and total PCB concentrations ranged from 3.8 to 10.4 ng/m3. To ensure the encapsulant was effectively blocking the release of PCBs into classrooms, wipe samples were collected on the encapsulated surfaces periodically during the longitudinal study. A total of 70 of these wipe samples were analyzed by the EPA Method 8082 and all of the samples were below 1 microgram per 100 square centimeters (μg/100 cm2), which was the detection limit.