SUGGESTIONS FOR REDUCTION
OF ANALYTICAL COSTS BY
ELIMINATION OF UNNECESSARY
DUPLICATE SAMPLES
by Douglas M. Chatham
Senior Chemist
SUMMARY
The field duplicates program should be eliminated from most
environmental projects. Additional sampling locations would provide
more information than duplicates. Field duplicates provide no
information about accuracy and questionable information about
precision. A much better evaluation of site precision could be
obtained by dividing the set of environmental samples into statistical
sub-sets. Elimination of field duplicates would reduce analytical
costs by at least 5%.
RECOMMENDATIONS
Any and all QC which contributes to the quality of the data
or are required for other reasons should be included regardless
of arguments presented in this paper. For each QC sample or analysis
proposed, Project Managers (PMs) and Quality Assurance Project
Officers (QAPOs) should ask what that determination contributes
to the quality of the data and whether it helps meet the project
DQOs. If a QC sample contributes nothing toward the DQOs, an argument
should be made against incurring the cost for that sample.
Collect and analyze field duplicates for Level IV (CLP) projects
only. Eliminate or greatly reduce the requirements for field duplicates
for Levels I, II, and III projects, unless it is necessary to
establish statistical measures of uncertainty in the definition
of extent of contamination.
DISCUSSION
The two types of Field Duplicates are split samples and co-located
samples. A split sample is a sample which has been thoroughly
blended and split between two containers. Often, the split samples
are sent to different laboratories. Split samples are intended
to measure the precision of the whole sampling and analysis procedure.
Most often, if they contain anything to measure, split samples
are a measure of how thoroughly the sample was blended before
being split. There is no way to determine an effect on the rest
of the samples at the site. Co-located samples are samples taken
in the same location but not blended. The intent of co-located
samples is to measure sampling precision or the variability of
the matrix.
";When designing experiments or procedures, it is important
to keep in mind that the overall objective is accuracy. It naturally
follows that those in charge of a project should ask whether additional
measurements really contribute to the accuracy of a method, or
simply to its precision.
In today's business world cost is very important, and each extra
measurement adds to the cost of a project. We all know that precision
is important, but we need to take a closer look at the costs and
benefits to the customer when expenses are increased for the sake
of improving precision without necessarily increasing accuracy."(1)
Often, the stated purpose of field duplicates is to measure the
precision of the complete process from sampling through analysis.
This is nice-sounding phraseology in a work plan, but what can
you do with the results? Due to the potentially large variability
inherent in the media being sampled particularly for soils and
sediments, one sample location out of twenty probably will not
represent the sampling or matrix variability. The result is that
these measurements are often reported as measures of ";precision";,
but they have no effect on the flagging or the use of the data.
As stated above, the source of the greatest variation in environmental
analytical results is the variability of the media. Comparable
results (<40% RPD) are seldom achieved from co-located duplicate
soil samples, even with the best efforts of the best sampling
technicians available. A statistical evaluation of all sample
results at a site should be used to measure the precision and
representativeness of the sampling program. These statistical
measures may provide confidence intervals for establishing extent
of contamination in a medium. For duplicates to adequately assess
sampling precision, they would have to be statistically selected
as a subset of the samples taken at the site and evaluated as
a subset to determine if the subset results in the same overall
values as the original set of samples.
Types of Investigations
The CERCLA preliminary assessment (PA) and the RCRA facility
assessment (RFA) are used to identify any releases or migration
from a facility based on the existing information and apparent
visual evidence. The PA or RFA are typically the first step in
the CERCLA or RCRA process. Samples are not usually taken during
these assessments.
The CERCLA Site Investigation (SI) and the RCRA Confirmation Sampling
Process (CS) are used to confirm or disprove suspected releases
to an environmental medium based on information obtained during
the PA or RFA. The SI or CS process is generally limited in scope;
the goal is to gather sufficient data to confirm the presence
or absence of contamination at Potential Release Locations (PRLs),
and to provide a basis for the scope of the CERCLA Remedial Investigation
(RI) or the RCRA Facility Investigation (RFI) process or a No
Further Action decision. The data quality levels for SIs or CSs
are generally required to be Level IV (CLP) or Definitive data
(2) since the data is required to be litigation quality. These
investigations are usually the first ones conducted at most sites
and historical data will be limited or non-existent. An expansion
of the sampling scope would provide more information than duplicates.
Additional sample points would provide more useful
data than field duplicates for all projects following the CERCLA
SI or RCRA CS studies. This includes the CERCLA Remedial Investigations
(RI), Feasibility Studies (FS), and Remedial Design/Remedial Action
Studies (RD/RA), the RCRA Facility Investigations (RFI), Corrective
Measures Studies (CMS), and Corrective Measures Implementations
(CMI), and all cleanup verifications and monitoring programs.
LIST OF REFERENCES
1. EPA, Risk Assessment Guidance for Superfund, Volume
I, Human Health Evaluation Manual (PartA), p. 5-5, EPA/540/1-89/002,
December 1989.
2. EPA, Test Methods for Evaluating Solid Waste, Physical/Chemical
Methods, SW-846, 3rd Edition, Final Update 1, November 1990.
