Representative sampling

1. Representative Sampling
and Laboratory QA/QC

2. Agenda
Language of Sampling
Types of Samples
The Unspoken Assumptions
Ideal Sampling Locations
Data Quality Objectives
Examples:
The Good, The Bad, and The Ugly

3. 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.

4. 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.

5. Representative Samples
What is representative?
Sample should represent or be typical of
the wastewater it is collected from.
If the true value of BOD
5
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

6. Representative 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.

7. Population Characteristics
68%
95%
99%
Mean or Average

8. Standard 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.

9. 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

10. 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.
Also applicable for estimating
performance under a given set of
conditions.

11. Composite Samples
Frequently used to estimate average
values over a 24-hour period.
BOD
5
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.

12. Composite 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.

13. 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.

14. Simple vs Flow Proportional
Time Flow (MGD) NH3-NSimpleFlow 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 18.5MGD
Total Lbs - Simple 3549
Total Lbs - Proportional4146
Difference of 15%

15. Manual vs. 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.

16. Examples of Autosampler Misuse
The BOD
5
Producing Equalization
Basin.
The No-Flow at Low-Flow Problem.
The 1 Day MCRT Nitrifying Basin.

17. 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.

18. 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.

19. 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.
When compositing or aliquoting, mix
samples well before pouring.

20. Guidelines for
Representative Sampling
Wiers are not good sampling points.
Solids settle upstream of weirs.
Oils and greases build-up downstream.
Materials tend to collect on the sides
and bottoms of channels. Avoid edges.
Before collecting the sample, rinse the
sampler and sample container several
times.

21. Agenda
Language of Sampling
Types of Samples
The Unspoken Assumptions
Ideal Sampling Locations
Data Quality Objectives
Examples:
The Good, The Bad, and The Ugly

22. 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.

23. 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?

24. Why Collect the 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.

25. 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.

26. 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.

27. What level of QA/QC
is Needed?
Field and laboratory?
Frequency of QA/QC Samples?
Permit required analysis – every time.
Process control – weekly perhaps.
Certified standards needed?
Outside laboratory involved?

28. Who Will Take the 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.

29. Example Problems

30. Plan of Attack
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?
Sampling location.
Grab or composite.
Frequency of
sampling.
Analytes needed.
QA/QC required.

31. Calculate MCRT
Where to collect
samples?
Type of samples to
collect?
If multiple basins are
in use?
If basins are
independent?

32. Estimate TF Performance
For average
performance.
For peak
performance.
Develop DQOs.
What if there is
recycle?

33. Quality
Assurance/Control

34. 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!

35. 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, BOD
5

36. Precision versus Accuracy
Neither precise nor accurate.
Precise, but not accurate.
Accurate, but not precise.
Accurate and Precise. BOTH
ARE NEEDED.

37. Precision versus Accuracy
Neither
Precise
Accurate
Both

38. Interferences
Substances in a sample that
cause
False Positives
False Negatives
Look for Interferences at the time
of analysis.

39. How do we know if we are:
Free of contamination?
Accurate and precise?

Lacking Interferences?

40. Checking for
Contamination
BLANKS

41. Filter Blank
Only needed when analyzing for
dissolved substances.
Total Suspended Solids (TSS)
Ortho-phosphorus

42. 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.

43. Reagent Blank
Ultra-pure water 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.

44. Limits 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.

45. 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?

46. Checking for Precision
DUPLICATES

47. 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.

48. Relative 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.

49. Relative Standard Difference
Original Sample Result – 300 mg/L BOD
5
Duplicate Sample Result – 350 mg/L BOD
5
Calculate the RPD
100*(300 – 350)
((300 + 350)/2) = (50/325)*100 = 15.4%
Within Limits?

50. 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.

51. Relative Standard Difference
Original Sample 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?

52. Checking for Accuracy
STANDARDS

53. 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

54. 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: BOD
5
standard should be
within + 15% of the true value.

55. 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%

56. Additional Quality Control
Spike Calculations
Sample Hold Times and Preservation
Instrument Calibration
Instrument Logs and Performance
Checks
Accurate Record Keeping
Secondary Review of Calculations

57. Hold Times and Preservation
Parameter Preservative Hold time
Alkalinity 4
o
C 14 Days
Ammonia Nitrogen H
2
SO
4
to pH<2, 4
o
C 28 Days
BOD
5
/CBOD
5
4
o
C 48 Hours
COD H
2
SO
4
to pH<2, 4
o
C 28 Days
Conductivity 4
o
C 28 Days
Fecal coliforms 4
o
C 24 Hours
Hardness HNO
3
to pH<2, 4
o
C 28 Days
Nitrate 4
o
C 48 Hours
Nitrite 4
o
C 48 Hours
Total Suspended Solids 4
o
C 7 Days
Total Dissolved Solids 4
o
C 7 Days
Total Solids 4
o
C 7 Days
Trace Metals HNO
3
to pH<2, 4
o
C 180 Days

58. 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.

59. 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.

60. We Do Analytical Work
GOOD
FAST
and CHEAP
Pick Any Two.

61. 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
Not all QA/QC applies to every
analysis.

62. Questions?