managing uncertainty through better upfront planning and flexible workplans
DESCRIPTION
Managing Uncertainty through Better Upfront Planning and Flexible Workplans. Northeast States’ Improving the Quality of Site Characterization. Albert Robbat, PhD Tufts University, Chemistry department Center for Field Analytical Studies and Technology Medford, Massachusetts 02155 - PowerPoint PPT PresentationTRANSCRIPT
Managing Uncertainty through Better Upfront Planning and Flexible Workplans
Albert Robbat, PhD
Tufts University, Chemistry department
Center for Field Analytical Studies and Technology
Medford, Massachusetts 02155
tel 617-627-3474; [email protected]
Northeast States’ Improving the Quality of Site Characterization
• Hazardous waste site characterization and cleanup is expensive and time-consuming.
• Small sites to extremely large sites require the same systematic planning and scientific assessment.
• Money is tight, yet data volume is needed to make sound scientific decisions as to the nature and extent, if any, of contamination.
• New sampling and analytical measurement technologies have been that have the potential to greatly reduce cost and time.
• New processes have been developed and promoted by state and federal agencies, but are rarely used.
• Why??
What’s the Problem
What’s The Opportunity
• Systematic Planning, Dynamic Workplans, Field Analytics and On-site Decision Making together can:
Provide more information at less cost and over shorter time periods.
Provide screening to quantitative “risk” quality data, when and where needed.
Increase field personnel efficiency and on-site decision making confidence.
Increase site-specific information and final site characterization decision confidence.
Rapid in situSample Collection and Analysis
What is Decision Making Uncertainty?
What is Sampling and Analysis Uncertainty?
How Many Soil Samples?
What Role Does Heterogeneity Play in the Sample Collection, Analysis, and Decision Making Process?
Sample ID Compounds Field (ppb)Laboratory
(ppb)
S2-B2-(20-22) 1,1-dichloroethene 30 < 501,1-dichloroethane 41 < 50cis-1,2-dichloroethene 560 < 501,1,1-trichloroethene 300 250toluene 37,000 2,000tetrachloroethane 120 < 50ethylbenzene 990 240m/p-xylene 7,400 1,200o-xylene 2,200 480
S3-B1-(13-15) toluene 280,000 58,200ethylbenzene 3,000 14,500m/p-xylene 320,000 58,700o-xylene 83,000 25,500
S3-B23-(13-15) 1,1-dichloroethene 15 < 10carbon tetrachloride 6 < 10tetrachloroethane 23 < 10ethylbenzene 7 < 10o-xylene 17 < 10
Field versus Laboratory VOC Data Comparison
No degradation or loss of analyte due to time delays
T y pe o f A na ly s isS ite 1
Sa m ple sS ite 2
S a m ple sSite 3
S a m p le sT o ta l Sa m ple s
A na ly ze d
P r o je cte d A ctu a l P r o je c te d A c tu a l P ro je c te d A c tua l P ro je c te d A c tua l
VO C S am p le sSc re e n ed
1 62 21 0 13 5 17 7 28 8 2 1 4 5 8 5 6 0 1
VO C S am p le sQ u a nt ifie d
42 51 36 5 8 5 9 4 9 1 3 7 1 5 8
P C B/P A H S am p le sQ u a nt ifie d
42 46 0 1 2 0 1 0 4 2 6 8
M e ta ls Sa m p le sq ua n tif ied
51 22 44 5 4 3 6 4 5 1 3 1 1 2 1
S a m p l e n u m b e r i n c l u d e s f i e l d d u p l i c a t e s .
D y n a m i c W o r k p l a n P r o j e c t e d a n d A c t u a l N u m b e r o f S a m p l e s A n a l y z e d
Projected vs Actual Number of Samples Analyzed
Traditional Approach
Off-SiteSamples Results
1. Planning Phase
2. Sample Collection 6. Decisions Made
3. Transportation
4. Lab Analysis
5. Results Returned
Characteristics- pre-planned sampling grids- off-site lab analysis- static work plans
Problems- high cost per sample- surprise results- pressure to oversample- multiple trips to field
Dynamic Workplan Approach
Planning Phase
Sample Collection Decisions Made
Field Analysis
Characteristics- Real time sample analysis- Rapid field decision making- Dynamic workplans
Advantages- Reduce cost per sample- Increase # of samples- Reduce # of field visits- Faster, better, cheaper
Requirements- Field analytical methods- Decision support in the field
Select Core Technical Team• Designate one member with authority to make final field decisions• Develop workplan “thought process and rules-to-follow” in the
field• Although in Massachusetts and Connecticut upfront buy-in is not
needed, adherence to the documented “thought process” will help insure acceptance of field data results
Develop Conceptual Model & Decision Making Framework
• Produce map depicting vadose zone and groundwater flow systems that can influence contaminant movement
• Establish DQO’s to ensure type, quantity, and quality of field data
Develop Standard Operating Procedures • Produce performance methods that support the DQO process • Document MDL’s prior to field mobilization
Systematic Planning and Dynamic Workplan
Develop Data Management Plan • Integrate chemical, physical, geological, and hydrogeological data
Develop Quality Assurance Project Plan • Define technical team/regulators responsibilities consistent with
EPA/state policy
Prepare Health and Safety Plan • Establish DQO’s to monitor worker/community safety
Systematic Planning and Dynamic Workplan
Field Requirements• Collect samples quickly
• Analyze samples quickly
• Review and report results quickly
Performance-based
PBMS/Keys to Success
Experienced, trained personnel
Data produced must provide level of assurance that it meets sufficient accuracy, precision, selectivity, sensitivity, and representativeness to meet project-specific DQO’s
Legal Defensibility, Rule 702, Determination of Reliability• technique tested, subject to peer review, accepted by scientific
community• method reproducible, with potential rate of error known
Visible & well-documented practices and procedures manuals for effective quality system
High Performance/Quality Control
Blanks, LCS, SRMs, MS, MSDs
Calibration & Continuing Calibration
Peak Integration
MDL’s
DQO’s
Data Useability
Reporting
In situ or Hand-held Vapor Analyzers• ECD, FID, PID provides signal response in seconds
Portable GC’s with Selective Detection• ECD, FID, PID provides screening data in seconds to 10’s
minutes
Field GC’s with Selective Detection• ECD, FID, PID provides semiquantitative data in 10’s of minutes
Field GC’s with Mass Spectrometry Detection• Provides semiquantitative to quantitative data in seconds to 10’s
of minutes
In situ Mass Spectrometry• Provides semiquantitative data in seconds
Immunoassay or colorimetric Kits• Provides screening data in 2-15 min
Field Analysis of Organics
x-ray Fluorescence Spectroscopy• Provides screening to quantitative data in seconds to 10’s of minutes
Inductively Coupled Plasma/Optical Emission Spectroscopy• Provides quantitative data in minutes
Anodic Stripping Voltammetry• Provides quantitative data in minutes
Immunoassay or colorimetric Kits• Provides screening data in 2-15 min
Field Analysis of Metals
Selected Projects! Hanscom Air Force Base, Bedford, MA (Volatiles, Semi-volatiles, Metals)
! Joliet Army Amunition Plant, Joliet, IL (Explosives)
! NJ Superfund Site (Metals)
! MCAS, Yuma, AZ (Volatiles, Semi-volatiles, Metals)
! Fort Devens, Ayer, MA (Volatiles, Semi-volatiles)
! KY Pipeline Company (PCBs)
! Landfills, MA & VA (Volatiles, Semi-volatiles, Metals)
! Naval Security Station, Washington, DC (PCBs)
! New England Coal Gasification Plants (PAHs)
! Midwestern Manufacturing Co. (Volatiles, Semi-volatiles)
Comparison of Field Technologies for PCBs and PAHs
Polycyclic Aromatic Hydrocarbons Polychlorinated Biphenyls
Site-specificDQO’s and
Action LevelAttributes GC/FID TD GC/ MS
EnzymeKits
GC/ECD TD GC/MSEnzyme
Kits
Ye s Selectivity N o S p ec ia t e c la ss -sp ec ific Ye s S p ec iat e c la ss -sp ec ific
1 -pp m /P A H0.5 -p pm t ot a l P C B Sensitivity 0.5 -p pm 0 .3 -pp m
M FG . a ndC om p o un dD e p e nd e nt
0 .03 -p p m 0.2 -p pmAroc lor
D e p e nd e n t0 .5 to 1 -pp m
4 0% Precision 4 0% 40 %M FG .
D e p e nd e nt 4 0%
3 0 % 4 0%M FG .
D e p e nd e n t 4 0 %
N oN o
Accuracy biase d to ward:
false positivefalse negative
YesN o
N oN o
YesN o
Ye sN o
N oN o
Ye sN o
AnalysisRate/Sample 2 0 -m in 1 0- m in 1 0 -m in 20 -m in 1 0 -m in 1 0 -m in
Total Number ofSam ples
Analyzed per10-hr Work Day
1 9 32 32 1 9 32 3 2
Comparison of Method and Data Quality Attributes
QC ParametersField PBMS
SW-846 Modified Method 8260ALaboratory Analysis
SW-846 Method 8260A
Instrument PerformanceTests MS Tuning
perform instrument check, minimumrequirement once to initiate shift
perform instrument check, minimumrequirement once to initiate 12-hr shift
Initial Calibration 5-point
DQO dependent; match SW 846 or all RF%RSDs 40% with no more than a > 30%
or all RF %RSDs 30%
calibration check compounds (CCC)%RSD’s < 30%, if all RF %RSD 15%
then use Ave. RF else use linear regression
Laboratory ControlStandard
sample throughput dependent, can match SW 846
after each initial calibration;percent accuracy within 80% to 120%
Continuing CalibrationVerification
DQO dependent; match SW 846 or begin &end of day, % Diff for all compounds
40% and no more than a > 30%
one per 12-hr shift; (calibration checkcompound) CCCs < 20%. All analytes
within ± 25% of expected value
Method Blankonce per day and after highly contaminatedsample; all target compound conc. < PQL
one per analytical batch;all target compound conc. < PQL
Surrogate Spike AnalysisDQO throughput rate dependent; for eachsample, blank, standard or other QC run
for each sample, blank, standard or otherQC run, laboratory established recovery
limits (e.g. 80-130 %)
Sensitivity 5-2500 ppb levels, matrix dependent 5-2500 ppb levels, matrix dependent
Selectivitycan do up to 97 VOCs 2-6 ions per analyte;
minimal chromatographic separation, selectivity achieved by IFD software
can do up to 97 VOCs with 1-6 ions percompound; adjust chromatography to
separate VOCs of interest
Precisionreplicate analysis
QC acceptance criteriareplicate analysis
QC acceptance criteria
Accuracysample throughput dependent; can matchSW 846; laboratory control check sample
(LCS) once per day
surrogate dependent recovery within 70-120%; laboratory control check sample
(LCS) once per 12-hr shift
Othercarryover monitored by analysis of blanks,
watch baseline on chromatogramscarryover monitored by analysis of blanks,
watch baseline on chromatograms
VOC Analysis of Soil by Purge and Trap GC/MS
QC ParametersField PBMS
SW-846 Modified Method 8270CLaboratory Method
SW-846 Method 8270B
Instrument PerformanceTests MS Tuning
perform instrument check, minimumrequirement once to initiate shift
perform instrument check, minimumrequirement once to initiate 12-hr shift
Initial Calibration 5-point
DQO dependent; SW 846 or all RF %RSDs 40% and no more than a > 30%
calibration check compounds (CCC)%RSD’s < 30%, if all RF %RSD 15% then
use Ave. RF else use linear regression
Laboratory ControlStandard
DQO throughput dependent; after each initialcalibration, percent accuracy 80% to 120%
after each initial calibration; percentaccuracy
within 80% to 120%
Continuing CalibrationVerification
DQO dependent; can match SW 846 or begin& end of day, % Diff for all compounds
40% with no more than a > 30%
one per 12-hr shift;%D for all compounds 20%
Method Blankonce per extraction batch; all targetcompound concentrations < PQL
one per extraction batch; all target compound concentrations < PQL
Surrogate Spike Analysissample throughput dependent; for eachsample, blank, standard or other QC run
for each sample, blank, standard or other QCrun, laboratory established recovery limits
(e.g. 20-130 %)
Sensitivity 100-ppb to 1000-ppb 660-ppb to 3300-ppb
Selectivitycan do up to 350 SVOC 2-6 ions per analyte;
minimal chromatographic separation,selectivity achieved by IFD software
can do up to 350 SVOC with 2-5 ions peranalyte; adjust chromatography to separate
SVOC of interest
Precision replicate analysis QC acceptance criteria replicate analysis QC acceptance criteria
Accuracysample throughput dependent; can match SW
846 for surrogate and MS/MSD recoveriessurrogate recovery compound dependent;
MS/MSD per extraction batch
SVOC Analysis of Soil by Thermal Desorption GC/MS
Cost Comparison and Data Turnaround Times
Analyte Current Laboratory Approach
Data Turnaround: 14 to 30 daysFaster Turnaround: 50-150% surcharge
PBMS TDGC/MS with IFD
Data Turnaround: < 7 days
VOCs$125/sample
SW 846 method 8240/826025-min/sample analysis
$75/samplemodified 826020-min/sample
PCBs
$100/sampleSW 846 method 8080
20-min/sample analysis;sample preparation
2-hr/batch of 20 samples
$100/samplemodified 8270
10-min per analysis;sample preparation
1-hr/batch of 20 samples
PAHs
$145/sampleSW 846 method 8100/8310;
20-min/sample analysis,sample preparation
2-hr/batch of 20 samples
Explosives
$180/sampleSW 846 8330/USAED 3020-min/sample analysis;
sample preparation18-hr/batch of 20 samples
Semi-VOCs
$375/sampleSW 846 method 8270
40-min/sample analysis;sample preparation
4-hr/batch of 20 samples
$100/samplemodified 8270
20-min per analysis;sample preparation
1-hr/batch of 20 samples
Analyte Current Laboratory Approach
Data Turnaround: 14 to 30 daysFaster Turnaround: 50-150% surcharge
PBMS TDGC/MS with IFD
Data Turnaround: < 7 days
VOCs$125/sample
SW 846 method 8240/826025-min/sample analysis
$75/samplemodified 826020-min/sample
PCBs
$100/sampleSW 846 method 8080
20-min/sample analysis;sample preparation
2-hr/batch of 20 samples
$100/samplemodified 8270
10-min per analysis;sample preparation
1-hr/batch of 20 samples
PAHs
$145/sampleSW 846 method 8100/8310;
20-min/sample analysis,sample preparation
2-hr/batch of 20 samples
Explosives
$180/sampleSW 846 8330/USAED 3020-min/sample analysis;
sample preparation18-hr/batch of 20 samples
Semi-VOCs
$375/sampleSW 846 method 8270
40-min/sample analysis;sample preparation
4-hr/batch of 20 samples
$100/samplemodified 8270
20-min per analysis;sample preparation
1-hr/batch of 20 samples
Contaminated ?
Delineate zone of contamination
1) Focus sampling and analysis targeting only those contaminants found in step 1.
Step 3Locate vertical and horizontal boundaries by stepping-out at 4 locations and 2 depth.
Step 2
2) Employ geostatistical sampling tools and adapt strategy based on the data obtained.
Verify "Clean Sites"Analysis of 4 depth samples from 5 random locations.
Field quantitive analysis of all contaminants. Total 20 samples.
Verify non-detect at boundary
contaminated zone. Total 8 samples.Analyze 4 samples at 2 depths from outside
Sampling Based on Initial Conceptual ModelStep 1
2) Conduct soil and soil gas survey
for all CLP TCL contaminants.
1) Conduct geophysics survey based on areas known to contain contaminants.
Verify the TARGET list Analyze inside contaminated area for all CLP TCL analytes at 4 random locations and at 4 depths.
Quantitate full suite of contaminants. Total 16 samples.
Send 5 split samples from random locations to
Total 5 samples.
Send 4 split samplesto laboratory .
To tal 4 sam ples.
Send 4 split samples to laboratory. Total 4 samples.
Phase 1Site Screening
Phase 2On-Site Verification
Phase 3 Lab Verification
Yes
No
Sampling and Analysis Flow Chart
Why is the Same Technology Readily Accepted in Other Regulated Markets?
Answer
They have learned how to deal with measurement and decision making
uncertainties!!
Research Funding & Logistical Support
• U.S. Environmental Protection Agency
• Army Environmental Center, Joliet Ammunition Plant
• Hanscom Air Force Base
• Department of Energy
• Agilent Technologies
• State Regulatory Agencies
• OHM, CH2MHill, Jacobs Engineering, Bechtel
• Charles River Laboratories, Pharmacopeia