Representative Sampling
and Laboratory QA/QC
Agenda
Language of Sampling
Types of Samples
The Unspoken Assumptions
Ideal Sampling Locations
Data Quality Objectives
Examples:
The Good, The Bad, and The Ugly
Introduction
Sampling usually given little thought
relative to analysis, even though critical.
Largest errors in results from sampling.
Samples are taken:
At the wrong place.
At the wrong time.
And of the wrong type.
Result: Non-representative results.
Samples vs. Populations
The population is the
total or all of the
possible answers we
might get by sampling.
All of the individuals in
this room.
Every 100 mL aliquot in 2
MG of influent.
We sample because we
can’t count the whole
population.
Representative Samples
Samples
What is representative?
Sample should represent or be typical of
the wastewater it is collected from.
If the true value of BOD5 in the wastewater
is 280 mg/L, then the sample should be
close to this value.
How do we know the sample is
representative?
Answer: Statistics
Representative Samples
Samples
Representative
samples should be
very close to the
mean value of the
population.
How do we know we
are close to the
mean?
Look at the sample
standard deviation.
Population Characteristics
Mean or Average
68%
95%
99%
Standard
Standard Deviation
Deviation
The standard
deviation tells us
how spread out the
data are.
If the mean is 20
and stddev is 2,
then 68% of all
measurements are
between 18 and 22.
Types of Samples
Grab Samples
Exactly what it sounds like. One sample
collected at a particular point and time.
Composite Samples
Multiple samples collected and added
together to make one sample.
Time Composite.
Space Composite.
Flow Proportional Composite.
Manual versus Automatic
Grab Samples
May be used where population is not
changing suddenly or changing a great
deal over time.
Must be used for particular analyses:
Residual chlorine.
Fecal coliform.
Alsoapplicable for estimating
performance under a given set of
conditions.
Composite
Composite Samples
Samples
Frequentlyused to estimate average
values over a 24-hour period.
BOD5 loading to aeration tanks.
TSS leaving the WWTP in the effluent.
Gives information over a longer period
of time or space.
Permit samples are often flow
proportional composites.
Composite
Composite Samples
Samples
Consideration must be given to
sample handling and storage during
compositing.
We don’t want the sample
characteristics to change while we
are sampling.
Refrigeration often used to slow
biological activity.
Chemicals may also be added as
preservatives.
How to Composite
Simple Composite – Add equal volumes
of samples collected from different
times or locations. Mix thoroughly.
Flow Proportional Composite – Volume
of each subsample based on flow.
Estimate total volume of sample required.
Estimate total flow over sampling period.
Calculate sample volume per flow.
Simple vs Flow Proportional
Time Flow (MGD) NH3 -N Simple Flow Prop.
Midnight 10 12 12 120
4 a.m. 15 15 15 225
8 a.m. 18 20 20 360
Noon 28 40 40 1120
4 p.m. 26 37 37 962
8 p.m. 14 14 14 196
NH3 -N Conc, (mg/L) in Sample = 23.0 26.9
Avg Daily Flow 18.5 MGD
Total Lbs - Simple 3549
Total Lbs - Proportional 4146
Difference of 15%
Manual
Manual vs.
vs. Automatic
Automatic
Manual samples are collected by hand.
Automatic samples are collected by
machine.
Cautions for automatic samplers:
Not necessarily better.
Not accurate when collecting <20 mL.
Clean frequently; clogging.
Variable flows and intake location.
Examples of Autosampler Misuse
TheBOD5 Producing Equalization
Basin.
The No-Flow at Low-Flow Problem.
The 1 Day MCRT Nitrifying Basin.
Unstated Assumptions
For
simplicity, we ASSUME that the
population we are sampling from is:
Normally distributed.
Completely mixed.
We also ASSUME that our sample
value approximates the population
mean.
These assumptions are not always true.
Guidelines for
Representative Sampling
Samples should be collected:
Only where wastewater is well-mixed.
In the center of the flow channel.
Horizontally and Vertically.
Avoids floating scum and settled solids.
Ensure that samplers and sample
containers are clean, uncontaminated,
and suitable for the planned analysis.
Guidelines for
Representative Sampling
Recommended Sample Containers:
HDPE appropriate for most analyses.
Trace metals, oil and grease, volatiles
should be sampled in glass containers.
Pre-cleaned or sterilized containers for
phosphorus testing and fecal coliforms.
Whencompositing or aliquoting, mix
samples well before pouring.
Guidelines for
Representative Sampling
Wiers are not good sampling points.
Solids settle upstream of weirs.
Oils and greases build-up downstream.
Materialstend to collect on the sides
and bottoms of channels. Avoid edges.
Before collecting the sample, rinse the
sampler and sample container several
times.
Agenda
Language of Sampling
Types of Samples
The Unspoken Assumptions
Ideal Sampling Locations
Data Quality Objectives
Examples:
The Good, The Bad, and The Ugly
DQOs
There are a lot of choices in sampling.
What type of sample to take.
Where to collect the sample.
What time of day to collect the sample.
How do we know what we need?
What data quality objectives (DQOs)
are all about.
DQOs
A sampling and analysis plan of attack.
Plan of attack is determined by
answering these questions:
Why are we taking the sample?
What do we want to know?
How will the data be used?
What level of QA/QC is needed?
Who will take the samples?
Why
Why Collect
Collect the
the Sample?
Sample?
For process control:
Wasting calculations.
Calculation of unit process efficiency.
Estimating plant capacity.
For permitting:
Required analyses for DMR.
Required analyses for biosolids disposal.
Quantifying receiving water quality.
What Do We Want to Know?
Seems like a simple question……
Often neglected in sampling and
analysis plans.
Are we interested in:
Average performance?
Performance at peak load?
Dictates type of sample AND time of day.
Dictates sampling location.
How Will the Data be Used?
Internally or externally?
Public access?
Will results prompt capital
expenditures?
Does data need to be legally
defensible?
Dictates total number of samples,
analysis method, and QA/QC needed.
What level of QA/QC
is Needed?
Fieldand laboratory?
Frequency of QA/QC Samples?
Permit required analysis – every time.
Process control – weekly perhaps.
Certified standards needed?
Outside laboratory involved?
Who
Who Will
Will Take
Take the
the Samples?
Samples?
Daytime sampling and analysis not
usually a problem, but….
Nights and graveyards?
Week-ends?
May limit types of sampling to be done.
Autosamplers eliminate this problem,
but still need to be checked.
Example Problems
Plan of Attack
Why are we taking Sampling location.
the sample? Grab or composite.
What do we want to
know?
Frequency of
sampling.
How will the data be
used? Analytes needed.
What level of QA/QC required.
QA/QC is needed?
Who will take the
samples?
Calculate MCRT
Where to collect
samples?
Type of samples to
collect?
If multiple basins are
in use?
If basins are
independent?
Estimate TF Performance
For average
performance.
For peak
performance.
Develop DQOs.
What if there is
recycle?
Quality
Assurance/Control
Why Do We Do It?
To Check for Contamination
To Verify
Precision
Accuracy
To Determine if Interferences are
Present
ENSURES DATA QUALITY and
GIVES CONFIDENCE!
Contamination
Results in a false positive.
Caused by dirty glassware and improper
sampling or handling techniques
Can happen at any stage of sampling or
analysis
Happens when we add something to the
sample
Examples: Phosphorus, Fecal Coliforms, BOD5
Precision
Precision versus Accuracy
Neither precise nor accurate.
Precise, but not accurate.
Accurate, but not precise.
Accurate and Precise. BOTH
ARE NEEDED.
Precision
Precision versus Accuracy
Neither Accurate
Precise Both
Interferences
Substances in a sample that
cause
False Positives
False Negatives
Lookfor Interferences at the time
of analysis.
How do we know if we are:
Freeof contamination?
Accurate and precise?
Lacking Interferences?
Checking for
Contamination
BLANKS
Filter Blank
Only needed when analyzing for
dissolved substances.
Total Suspended Solids (TSS)
Ortho-phosphorus
Filter Blank
Checks for contamination during filtering.
Set up and clean filtration apparatus. Special
cleaning should not be done for blanks.
Filter a volume of ultra-pure water.
The filtrate is the filter blank.
The filter blank should be treated like any
other sample.
Reagent Blank
Ultra-purewater analyzed as a sample.
Accounts for differences in reagents
between lot numbers or batches.
Often used to “auto-zero” and
instrument.
Subtracts out background.
Can be a check for contamination.
Limits
Limits for
for Blanks
Blank values should be less than the
MDL.
MDL = Method Detection Limit.
Lowest concentration used for reporting.
Calculated value that may be different
for different laboratories and analysts.
See EPA method for how to calculate.
Acceptable or Unacceptable?
MDL for Nitrate test is 0.5 mg/L
Field blank reads 0.2 mg/L
Filter blank reads 1.7 mg/L
Reagent blank reads 0.1 mg/L
Sample results are higher than normal.
What happened?
Checking for Precision
DUPLICATES
Field Duplicate
A second sample taken at the same
time and place as the original sample.
Placed into a separate sample bottle.
Checks whether or not the sample is
representative.
Tells us how heterogeneous the
population is.
Relative
Relative Standard
Standard Difference
RSD = ( A - B) * 100
((A+B)/2)
Where
A = Original Sample Result
B = Duplicate Result
Results from a field duplicate should agree
within +/- 20% RPD of original sample.
Relative
Relative Standard
Standard Difference
Original Sample Result – 300 mg/L BOD5
Duplicate Sample Result – 350 mg/L BOD5
Calculate the RPD
100*(300 – 350)
((300 + 350)/2) = (50/325)*100 = 15.4%
Within Limits?
Lab Duplicate
Tests analyst’s ability to take a representative
sample from the field sample.
Two aliquots are taken from the same sample
bottle and subjected to the same sample
preparation and analysis steps.
Don’t confuse a duplicate with a replicate. A
replicate is a second reading from the same
aliquot.
Relative
Relative Standard
Standard Difference
OriginalSample Result – 300 mg/L TSS
Duplicate Sample Result – 180 mg/L
TSS
Calculate the RPD
100*(300 – 180)
((300 + 180)/2) = (120/240)*100 = 50.0%
WHAT HAPPENED?
Checking for Accuracy
STANDARDS
Standards
Standards
Contain a known concentration of analyte.
Should be within the same range as the
sample concentrations.
Standard Methods recommends 5 to 50 times
the MDL.
May be purchased “certified” from outside
vendors.
Environmental Resource Associates
Hach, SPEX, VWR Scientific Products, and others
Standards
Standards
Standards should be analyzed
Each time an instrument is calibrated.
Once per sample batch.
Once per lot of reagents.
Standard percent recoveries should be
within + 10% of the true value.
Exception: BOD5 standard should be
within + 15% of the true value.
Percent Recovery Calculation
Certified Standard Concentration = 45.0 mg/L
Measured Standard Concentration = 42.0
mg/L
Percent Recovery = (Measured
Concentration / Certified Value)*100
% R = (42.0 / 45.0)*100 = 93.3%
Additional Quality Control
Spike Calculations
Sample Hold Times and Preservation
Instrument Calibration
Instrument Logs and Performance
Checks
Accurate Record Keeping
Secondary Review of Calculations
Hold Times and Preservation
Parameter Preservative Hold time
Alkalinity 4o C 14 Days
Ammonia Nitrogen H2SO4 to pH<2, 4oC 28 Days
BOD5/CBOD5 4o C 48 Hours
COD H2SO4 to pH<2, 4oC 28 Days
Conductivity 4o C 28 Days
Fecal coliforms 4o C 24 Hours
Hardness HNO3 to pH<2, 4oC 28 Days
Nitrate 4o C 48 Hours
Nitrite 4o C 48 Hours
Total Suspended Solids 4o C 7 Days
Total Dissolved Solids 4o C 7 Days
Total Solids 4o C 7 Days
Trace Metals HNO3 to pH<2, 4oC 180 Days
Instrument Calibration
Minimum of a blank and one standard.
Standard Methods recommends a blank
and THREE standards.
EVERY time the instrument is used or
once per day.
Quality Assurance and Quality
Control: Is it All Really
Necessary?
Permitted Analyses vs. Analyses for
Process Control
Remember! Process control decisions
are only as good as the data they are
based on.
We Do Analytical Work
GOOD
FAST
and CHEAP
Pick Any Two.
QA/QC by Standard Methods
Reagent Blanks – One per 20 Samples
Duplicates – One per 20 Samples
Spikes – One per 20 Samples
Instrument Calibration – Every Time
Used
Calibration Blank and Three Standards
Notall QA/QC applies to every
analysis.
Questions?