Data Analysis
Material is deposited on each stage of a Cascade Impactor. Each stage has an efficiency curve. It is not analogous to a screen filter or a go/no-go gage. Each stage collects a range of sizes. For this reason it is futile to attempt to garner any information from a single stage in the cascade.
This can be best illustrated by example. Consider a near perfect Cascade Impactor wherein the stage D50 points are close to the popular numbers of 10, 4, 2.5 and 1 micrometers. We have set up a realistic data set for the unit running at its rated flow rate of 10 lpm, for 180 minutes.
The initial data appears as:
Stage
# |
Cut Dia.
μm |
Net Weight
mg. |
Concentration
mg/m3 |
| 1 |
10.6 |
1.019 |
0.57 |
| 2 |
4.1 |
1.252 |
0.70 |
| 3 |
2.5 |
2.046 |
1.14 |
| 4 |
1 |
1.704 |
0.95 |
| Filter |
|
0.666 |
0.37 |
| Total |
|
6.687 |
3.72 |
Alas, the concentrations are false. The weights on the stage do not correspond to the defined “Numbers”. To ascertain the truth, the data must be analyzed, plotted on a log probability graph and extracted as shown in Four Stage LORI Impactor.
The true data is:
Stage
#
|
Cut Dia.
μm
|
Net Weight
mg.
|
Concentration
mg/m3 |
True
Concentration
|
Error
%
|
| 1 |
10.6 |
1.019 |
0.57 |
3.22 |
82.4 |
| 2 |
4.1 |
1.252 |
0.70 |
2.1 |
66.9 |
| 3 |
2.5 |
2.046 |
1.14 |
1.4 |
18.8 |
| 4 |
1 |
1.704 |
0.95 |
0.39 |
-142.7 |
| Filter |
|
0.666 |
0.37 |
|
|
| Total |
|
6.687 |
3.72 |
|
|
The error in not analyzing the data is seen to be quite substantial |
Copyright © 2006 by BGI / Modified:
Wednesday, January 17, 2007
|
