Mercury Work Group
Phase II Reports >> Facilities Loadings

Facilities Loadings | Pretreatment Manual | Tech. Identification
Hg Management Guidebook | Mercury Products Database

For more information, contact David Eppstein by email at
deppstein@masco.harvard.edu, or by calling 617-632-2860.

 
5.0 DISCHARGE MERCURY CONCENTRATION DATA

We extracted mercury concentration test data for all permitted discharges from the MWRA Laboratory Information Management System for MWRA Fiscal Years (FY) 1995 and 1996. FY 1995 began on July 1, 1994 and ended on June 30, 1995. Similarly, FY 1996 began on July 1, 1995 and ended on June 30, 1996. We exported the test data into a personal computer spreadsheet program where we sorted the data and matched it to the facilities selected for the study.

For each sampling location, we then used the spreadsheet program to calculate average mercury concentrations for each half-year period of the two fiscal years under study. That is, we calculated average mercury concentrations for four six-month periods beginning June 1994, January 1995, June 1995, and January 1996. In the calculations, we handled all non-detected sample results (also known as below detection limit (BDL) results) in two distinct ways:

• We considered each calculated average mercury concentration to be at a minimum value if all individual non-detects were taken to be 0.0 µg/L (ppb).

• We considered each calculated average mercury concentration to be at a maximum value if all individual non-detects were taken at the stated detection limits (typically, 0.2 µg/L to 1.0 µg/L (ppb)).

In this way, a possible range of calculated mercury concentration averages for each sampling location could be determined and could possibly provide insight into expected accuracy of the calculations. We also showed in the spreadsheet the number of sample data points contributing to each average. As shown, mercury concentration averages were often derived from only one data point but could be derived from as many as 28 data points.

We sorted the individual sampling locations and their mercury concentration averages into the five groups: Clinical, Incinerator, Laundry, Other, and Research. For each of the five groups, we calculated overall average concentrations for each half-year interval. These overall averages, hereinafter called Group Concentration Averages (GCA's), were calculated after two rounds of data censoring.

We used a commonly-accepted statistical analysis technique as follows:

• In the first censoring round, we examined individual sampling locations and half-year intervals having multiple data points. We assumed that the concentration data had a Normal Distribution and tested the data for outliers. Outliers are unusually high or low values that sometimes result from one-time events at a facility or from sampling or testing errors. We used a standard test for outliers by applying an upper limit as follows:

Upper Limit = Average + 3 x Standard Deviation.

This limit would include about 99.7 percent of all normally distributed data. We believed that this conservative upper limit would eliminate the dominant effects of very high discharge concentration measurements in the individual averages.

• In the second censoring round, we similarly looked for outliers in the assembly of individual average mercury concentrations within each group for each half-year interval. We calculated the initial group half-year averages and applied the Normal Distribution upper limit as above to remove outliers from the individual mercury concentration averages. In this way, we reduced the dominant effects of any high individual averages. Thus, our GCA’s were the post-outlier group averages of individual mercury concentration averages.

It is important to note that we censored the mercury concentration data only for purposes of calculating the Group Concentration Averages (GCA’s). We calculated all individual facility loadings estimates using mercury concentration averages derived from totally uncensored data.

The mercury concentration data and averages for Fiscal Year 1995 are presented in Table B-1 of Appendix B. Mercury concentration data and averages for Fiscal Year 1996 are presented in Table B-2 of Appendix B. To produce these tables, we first sorted the data into the five study groups and then by MWRA Facility Identification Numbers and Sampling Location Numbers. The tables show that we averaged any duplicate concentration test results for a single date before using them in our calculations of averages for the respective half-year interval. The tables also show the number of samples test results that made up each concentration average.

At the end of each group listing in the tables, we presented statistical summaries that show overall mercury concentration averages, numbers of samples and tested sample locations, the normal distribution 99.7 percent upper limits, and the mercury concentration averages after removal of outliers. These post-outlier averages are the Group Concentration Averages (GCA’s). We also provided the total number of samples tested for each group. In addition, the average number of samples per sampling location is shown for each half-year interval.

As an example, page 9 of Table B-1 shows that 22 sampling locations of our 64 Clinical facilities had mercury concentration data in the first half-year of FY 1995. There were 101 samples tested for mercury, averaging a respectable 4.6 samples per sampling location. The "minimum" and "maximum" Clinical concentration averages were 3.22 µg/L and 3.44 µg/L (ppb). After removal of outliers, the Clinical GCA’s became 1.35 µg/L and 1.58 µg/L (ppb). Thus, the overall average mercury concentrations for Clinical discharges were lowered by nearly 60 percent through the censoring of outliers in both raw concentration data and individual averages.

The following tabulation summarizes the numbers of sample test results used to calculate the respective GCA’s:

We believe that the censoring of outliers was an important part of our methodology for estimating the group mercury loadings. We used the resulting GCA’s within each study group to calculate mercury discharge loadings for sampling locations that had no mercury concentration test data. We believe that, with this approach, we have made reasonable total loadings estimates for the identified groups of clinical, incinerator, laundry, other, and research dischargers. We have summarized the GCA’s in Table 1 for FY 1995 and Table 2 for FY 1996.

The two tables also show the numbers of sampling locations having mercury concentration data in each of our five study groups for each half-year interval. Adding these numbers, we find that the numbers of sampling locations having mercury concentration data in our five study groups for FY 1995 were 92 in the first half-year and 117 in the second half-year. For FY 1996, the corresponding numbers were 116 in the first half-year and 124 in the second half-year. Dividing these numbers into the associated total number of tested samples from the above tabulation, we can see that about 3.5 samples were tested per sampling location for each half-year period except for the second half-year period of FY 1996 when only 1.8 samples were tested per sampling location. The reason for this falloff in sampling is beyond the scope of this study.

From another perspective, we had actual mercury concentration data for 35 to 51 percent of Clinical sampling locations, 19 to 23 percent of Other sampling locations, and 23 to 38 percent of Research sampling locations. Incinerator and Laundry sampling locations had higher percentages. Thus, the GCA’s for each study group allowed us to estimate mercury discharge loadings for the majority of sampling locations that did not have concentration data.

In addition, for many locations that had concentration data, there were only one or two available sample test results with which to calculate mercury concentration averages. Clearly, our study would have benefited from the availability of more concentration data.

TABLE 1
FY1995
Group Concentration Averages (GCA's)
ug/L (ppb)

TABLE 2
FY1996
Group Concentration Averages (GCA's)
ug/L (ppb)

RETURN TO FACILITIES LOADINGS REPORT
TABLE OF CONTENTS
 

Mercury Work Group Site Guide
PHASE I REPORTS	PHASE II REPORTS	OTHER REPORTS
Executive Summary End-of-Pipe Report Operations Report Infrastructure Report	Facilities Loadings Pretreatment Manual Tech. Identification Hg Management Guidebook Mercury Products Database	RFP Guidelines
RELATED LINKS	CONTACT US	MWG HOME

 
ABOUT MASCO, INC. :: AREA PLANNING & DEVELOPMENT
PARKING & TRANSPORTATION SERVICES :: COMMUTEWORKS ::  MASCO SERVICES, INC.
375 LONGWOOD AVE. :: COLLEGES OF THE FENWAY :: CHILD CARE CENTER
MERCURY WORK GROUP :: MASCO MEMBER INSTITUTIONS :: QUARTERLY UPDATE
DIRECTIONS :: THE LONGWOOD MEDICAL AND ACADEMIC AREA
SITE GUIDE :: CONTACT MASCO, INC. :: RETURN HOME

08/16/2006

Copyright © 2000 MASCO, Inc.
375 Longwood Avenue
Boston, MA 02215
Phone: 617-632-2310
Fax: 617-632-2759