Best Practices for Filter Media

Analysis: Using ANSI/AWWA B100

for improved specification & filter
maintenance development

Rich Reavis, E.I.T.

Process Engineer
 Background
Research Assistant BSU Advanced Materials Lab
Surrogate fuels for nuclear reactors, ceramic nano-
powders
Particle Size Analyzer (PSA), sampling & analyzing
methods
New Product Development Engineer
$20+ million job, media specification issues
Worked with manufacturer, engineer, contractor &
owner
 Learning Objectives
As a result of this presentation, you will be able to:

1. Understand the key components of the often overlooked

process of sieve analysis.

2. Identify the most influential factors that affect particle size

distribution results.

3. Implement recommended procedures from AWWA B100 and

various ASTM Standards.

4. Provide an outline for filter media sieve analysis procedures

to ensure accurate and repeatable results, improving overall
filter maintenance programs.
 AWWA B100 Background
Granular filter media standard since 1950

References NSF, ANSI, and ASTM standards

B100 procedures take precedence to referenced standards

Not valid for Granular Activated Carbon (GAC)

See ANSI/AWWA B604

Open for interpretation in many statements and recommended

procedures
 Definitions

Particle Size Distribution Various size and dimension range

description of a filter material determined by standard sieve analysis
procedures.

Image courtesy of http://www.analizator.su/ Image courtesy of http://www.cobalt-nickel.com

 Definitions
Sieve Analysis The method by which a dry aggregate of known
mass is separated through a series of sieves of progressively
smaller openings for determination of particle size distribution.
In accordance with ASTM C136, as modified and supplemented by
AWWA B100.

Sieve Calibration A process of verifying the sieve mesh cloth is

within allowable tolerances of nominal sizing, either by
manufacturer certification or using standard reference material.
In accordance with ASTM E11, as modified and supplemented by
AWWA B100
 What is Effective Size?
Effective Size (ES) - The screen aperture size which 10 percent of a sample will
pass (by dry weight).

Typical specification sand ES = 0.55 0.65 mm, anthracite ES = 1.0 1.1 mm

Sieve
Stack

90% cumulative
Retained Media
weight

ES 10% cumulative
weight

ES = D10
 What is Uniformity Coefficient?
Uniformity Coefficient (UC) - A ratio of the 60 percent and 10 percent passing
aperture size (by dry weight). A relative measure of size between largest and
smallest particles 1.0 would be completely uniform and nearly impossible to
achieve.
Typical specification sand & anthracite UC < 1.5

Largest to Smallest Particle Size in Sample

UC Close
to 1 Roughly
Same ES
Higher
UC

UC = D60/D10
 Sample Method
Key Sample Size
Parameters Sieve Calibration
Analysis Method
Sample Method
 Sampling Factors
Sample directly off mesh screens at manufacturers site (best practice).
Cross-sectional sampling while loaded into bags/cars

Transit to jobsite leads to particle segregation

Fines tend to settle at bottom.
Do not scoop samples from top or collect exclusively from one area.

Anthracite grains more brittle and susceptible to size degradation

Different media probes used for sand and anthracite.

Make sure media does not fall back into hole which media probe is
sampling (Anthracite)
Broken grains produce artificial results (smaller)

Avoid splitting sacks between filter cells.

Unload entire sack into cell

Image courtesy of Wikipedia.com

 Bag Sampling Method
Supplies Needed
Permanent Marker
Plastic Sandwich bags
Grain Thief (sand) / 3 thin-walled
steel pipe (anthracite)
Rubber Mallet
Duct Tape
Running water (anthracite) Sand Super-Sack

Sample from each third of bag, Sample from three directions in each Sample three vertical angles
away from middle of sack hole
SIDE VIEW
FRONT VIEW TOP VIEW
Anthracite Super-Sack
 Filter Sampling Method
Supplies Needed
4 x 4 Plywood
Golf Tube or thin PVC 1.5 D
Sand paper rough up inside of tube
Individual plastic sandwich bags
Permanent marker

Procedure 1
Mark each bag in 6 increments, starting at 0- 6
Mark the tube in 6 increments (use Whiteout)
Place plywood in filter as sampling base
Use tube to sample first 6 (lightly twist and push down)
2
Place sample in bag, continue to media interface
Sample from 3 locations (combine like-depth samples) from
3 areas within the filter
3
 Sample Method
Key Sample Size
Parameters Sieve Calibration
Analysis Method
Sample Size
 B100 Sample Size Criteria
Lot Size Determination - Supplier and contractor must agree upon lot size.
More samples to analyze = more $$$.
Large lots, risk rejecting a lot of good material.

Random Bags - Number of bags to be sampled in composite sample based upon

number of bags in lot (Table 4, Section 5).
Lot Size = 26 50 bags Sample at least 5 bags from each lot

Composite sample - Maximum particle size determines minimum sample size, by

weight (Table 3, Section 5).
Particles 3/8 and smaller at least 10 lb composite sample

Sieve Sample - Total weight for sieve analysis (Table 6, Section 5). Use mechanical
splitter to divide composite sample into sieve weight required.
Particles 2.36 mm or less 0.17 0.3 lb (75-150 g)
 Sample Size
RANDOM
LOT
SACKS

SIEVE SPLIT COMPOSITE

ANALYSIS SAMPLE SAMPLE

75-150g
~10 lbs
 Sample Method
Key Sample Size
Parameters Sieve Calibration
Analysis Method
Sieve Calibration
 AWWA B100 Statements
For precise work, sieves should be calibrated annually according to
procedures in ASTM E11 1

If questions of compliance arise when nominal standard sieve openings

are used, standard reference materials (glass spheres) certified by the
National Institute of Standards and Technology should be used in
accordance with their calibration procedure 1

If standard reference material for calibration is not used, the data shall be
replotted using both the maximum and minimum permissible variation of
the average opening from the standard sieve designation as shown in
Table 1, column 4 of ASTM E11The material shall be in compliance if
either of the plots agree with the purchasers specifications. 1
AWWA B100 values do not match ASTM E11 values for permissible variation.
These values were given as percent variation prior to latest revision in 2010.
Use B100 table values.
 Calibration Methods
Glass spheres - Standard reference material provided by National Institute
of Standards and Technology (NIST).
Accuracy 1 m

Optically Measure Independently measure or have a certified lab

measure.

Master Calibrated Sieve Stack Run analysis and check against.

Certified Calibrated Reference Sample Run analysis and check against.

Purchase new certified sieves

Test Sieve Grades 4
Compliance (66%) manufactured and inspected to be in compliance with
Specification E11, standard test sieve.
Inspection (99%) average aperture size, certificate requires inspection data.
Calibration (99.73%) number of aperture and wire diameters measured,
average aperture size, standard deviation and average wire diameter,
certificate requires inspection data.
 Aperture Sizing
Aperture the dimension defining an opening in a screening surface. 4

Compliance Sieve
Standard, mesh measured before mounted in frame X

Y
Inspection Sieve
Mesh measured after mounted in frame
+
=
Calibrated 2
Includes certificate listing the actual aperture size to 0.1 m
AWWA Mesh Size Standards
Nominal Minimum Maximum Variation

__Nominal ES = 1.10
Minimum ES = 1.066
Maximum ES = 1.153

__Nominal UC = 1.314
Minimum UC = 1.319
Maximum UC = 1.310
 Common Sieve Stacks
Adjacent Sieves shall have openings such that the ratio between them is
the fourth root of 2 (1.1892). 1

Must have a minimum of 6 sieves with the largest not retaining more
than 5% of the total sample weight. 1

Total weight allowed per sieve after shake test given in Table 7, Section
5. 1

Recommended Shake Times

Sand about 10 minutes
Anthracite about 5 minutes

All media must be dried entirely, in an oven around 240 F, before

entering sieve stack. 1
 Sample Method
Key Sample Size
Parameters Sieve Calibration
Analysis Method
Analysis Method
 AWWA Standards

B100-09 Standards (Latest Revision 1/09)

Sieve data plotted on log-probability, graph paper,
or comparable computer programs. 1
No calibration, then replot using maximum and
minimum permissible variation of the average
opening from the nominal sieve dimension. 1
A smooth curve is to be drawn through plotted
points. 1
Standard Particle Distribution Curve
 Interpolation Methods
Mesh Sieve, mm Retained, g % Passing
8 2.36 100.0
10 2.047 0.70 99.3
12 1.73 13.10 86.2
14 1.374 47.80 38.5
16 1.199 27.50 11.0
18 1.01 7.50 3.5 ES found
20
25
0.868
0.696
1.90
0.80
1.6
0.8
through
30 0.601 0.80 0.0 interpolation

Linear Curve
y2
y2

y
y
y1
y1

x1 x x2
x1 x x2

1 2 1 2 4
y = mx + b = = 2 + + =
1 2 1 2
Linear Polynomial Variation

__Poly ES = 1.110
__Linear ES = 1.084

__Poly UC = 1.314
__Linear UC = 1.362
Calibrated Nominal Variation

__Calibrated UC = 1.264 __Calibrated ES = 1.141

__Nominal UC = 1.314 __Nominal ES = 1.110
Nominal Minimum Maximum Variation

__Nominal ES = 1.10
Minimum ES = 1.066
Maximum ES = 1.153

__Nominal UC = 1.314
Minimum UC = 1.319
Maximum UC = 1.310
 Whats The Big Deal?

Methods Size Designation

Linear Nominal
Curve Calibrated
Minimum/Maximum

6 results from one sample

ES range usually 0.1 mm in
project specifications
 Sand vs. Anthracite

Size, mm
Sieve step size larger for 4.750
anthracite 4.000
3.350
2.800
2.360
Interpolation differences more 2.000 Typical
1.700
noticeable with larger size range 1.400
Anthracite
1.180 Range
1.000
Sand sieve results generally more 0.850
0.710
precise no matter what method Typical Sand
0.600
Range
used 0.500
0.425
0.355
0.300
0.250
0.212
 Why Curve Fit?

Proportionality Parabolic Why not this?

y = mx + b y = x2 Derivatives (rate of
change)
y = m (slope) y = 2x

y y y
4

1 1

-2 -1 0 1 2 x -2 -1 0 1 2 x 0 x

y = |x| y = x2 y = x2 ??
 Calculus Overview
Curve fitting assumes continuous function for all data points (percent passing)
in interval (sieve size designations), assuming infinite number of sieve sizes in
stack to create a smaller step-size, producing a smooth curve.

Limited data points means interpolating to project results, assuming data

points do not fall exactly on known values.

Curve characteristics between data points:

Tangent lines (slope)
First Derivative Test x = (-,0) x = (0, )
y
Increasing Decreasing
y < 0 y > 0 4
y = x2 + +
y = 2x - +
Local maximum/minimum
y = 0 1 y = 2 + +

Second Derivative Test -2 -1 0 1 2 x

Concave up Concave down
y > 0 y < 0
 Mean Value Theorem
First introduced by Joseph-Louis Lagrange in the late 1700s. 3
Guarantees there is a point where the tangent line to the
curve is parallel to the chord connecting two given data
points. 3
()
=

Percent Passing

Linear Interpolation
()
() = () () () = ()

10 Polynomial Curve
= 2
a b
Particle Size, mm
 The Power of R2

R2 is a statistical measurement to Trendlines

determine how well a trendline predicts
real data points. R2 values range from 0 Polynomial
(worst) to 1 (best). Power
Exponential
In sieve analysis, certain ranges of data
are of interest (D10, D50, D60, D90). By Root
fitting and interpolating specific sections Logarithmic
along the distribution curve, a better R2 Linear
value can be produced.

Polynomial fitting has proven to be best

for sieve analysis producing an R2 value of
1 nearly every time.
 Normal Distribution Curve

Mean

__Low UC
__Medium UC
__High UC
Percent Retained

-2 -1 +1 +2
Particle Size
High UC Nominal vs Calibrated

__Nominal UC = 1.526
__Calibrated UC = 1.510

__Nominal ES = 0.999
__Calibrated ES = 1.009
 High UC - Nominal

__Poly UC = 1.526
__Linear UC = 1.542

__Poly ES = 0.999
__Linear ES = 0.999
Low UC Nominal vs Calibrated

__Nominal UC = 1.221
__Calibrated UC = 1.166

__Nominal ES = 1.072
__Calibrated ES = 1.119
 Low UC - Nominal

__Poly UC = 1.221
__Linear UC = 1.291

__Poly ES = 1.072
__Linear ES = 1.028
 Low UC - Calibrated

__Poly UC = 1.166
__Linear UC = 1.266

__Poly ES = 1.119
__Linear ES = 1.039
 High vs Low UC

UC = D60/D10

__UC = 1.510 Close to

__UC = 1.266 Linear

__ES = 1.009
__ES = 1.039
 Key Graph Attributes
Data point location versus major gridline

Slope of graph
Steep (Lower UC)
Flat (Higher UC)

Data range to interpolate

Anthracite vs Sand
Interpolate correct range

Method
Linear vs Curve

Sieve Size Designation

Calibrated, Nominal, Max/Min

Number of sieves = number of data points for analysis

Significant Figures
 Sample Method
Key Sample Size
Parameters Sieve Calibration
Analysis Method
 Conclusions
Lower UC material will have major fluctuations between interpolation and
calibration methods.

Methods less important for higher UC material.

Variance between methods not as noticeable with higher UC samples.

Sampling methods can vary results.

Sample size is important to identify and can lead to varied results.

Equipment calibration can lead to varied results.

Different analyzing methods can produce varied results.

 What does it all mean?
Tighten quality assurance and enhance
performance of filter.

Increase communication and trust between

customer, engineer, contractor, and supplier.

Saving money and time for all involved with

project.
 Who Benefits?
Customer Contractor
Receive verified product Upfront instructions on sampling
Future maintenance methods
Less impact on schedule Less impact on schedule
Avoid multiple interpretations of
ambiguous testing criteria

Engineer Supplier
Provide more concise Product is fairly and consistently
specifications that protect tested
everyone
More comfortable to bid job
Less impact on schedule
Less impact on schedule
 Recommendations
Consistent method for all parties involved with
sieve analysis on project
Sampling Techniques
Sample Sizing
Sieve Sizing or Calibration
Analysis Interpolation Method
Implement maintenance program that contains
proper sampling and analysis techniques
Ask for calibration certificates and analysis
methods from lab
 References
1AWWA Standard for Granular Filter Material, ANSI/AWWA
B100-09, American Waterworks Association, 2009.
2AWI Standard for Filter Media Sieve Analysis Procedures,
Anthratech U.S. (AWI), 2012.
3Hass, J., Weir, M.D., Thomas, Jr., G.B. (2007). University
Calculus: Part One Single Variable. Boston, MA: Pearson
Education, Inc.
4ASTM Standard E11, 2009, Standard Specification for Woven
Wire Test Sieve Cloth and Test Sieves, ASTM International,
West Conshohocken, PA, 2009, DOI: 10.1520/E0011-09E01,
www.astm.org.
Questions?