This document describes a design of experiments study to optimize the regeneration of spent activated carbon cloth used to adsorb volatile organic compounds in automobile painting processes. A full factorial design was used initially to screen important factors. A central composite design was then employed to develop regression models and identify optimum regeneration conditions that minimize residual compounds ("heel") and maximize pore volume for reuse. The optimum conditions were identified as a molecular weight of 79.2 g/mol, kinetic diameter of 0.28 nm, and heating temperature of 259°C, achieving 4.1% heel and 0.82 cm3/g pore volume.

1. MEC E 668: Design of Experiment
Instructor: Professor Kajsa Duke
Final Project
by
Monisha Alam
Supervisor: Dr. Zaher Hashisho
Design of Experiments to Optimize the Regeneration
Process of Spent Activated Carbon Cloth by Resistive
Heating Method

2. Introduction
• Volatile organic compounds
(VOCs) emitted from car painting
solvents in automobile industries
cause indoor air pollution
• Activated carbon cloth (ACC)
• highly porous material, used as
adsorbent
• adsorbs (VOCs) on surface &
inside pores
2
http://northharfordcollision.net/wp-content/uploads/2013/06/car-painting.jpg

3. Introduction
• ACC used for 1 adsorption cycle: spent ACC
• Spent ACC reused for economic purpose
• Regeneration (VOCs are removed from pores of ACC) for
economical reuse
• Resistive heating: higher heating rate, fast desorption
• To find optimized regeneration conditions:
• “Factorial Design”: deals with several factors at a time
• “Best-guess” : inefficient due to inadequate previous study
• “One-factor-at-a-time”: lengthy, costly, no interaction
3

4. Introduction
Successful regeneration indicates:
• Minimum Heel (residual amount of strongly adsorbed VOCs
on ACC)
• Maximum Pore Volume available for adsorption
Objective:
To identify optimum conditions to obtain regenerated ACC that
contains:
• heel (< 5 wt% of the virgin ACC)
• pore volume (≥ 0.8 cm3/g)
4

5. • Spent ACC samples wrapped in hollow cylinder shape
(1.65 cm inner diameter, 10 cm length)
• Two stainless steel electrode tubes, with heating elements
Materials and Methods
5
Regeneration CartridgeSpent ACC

6. Experimental Setup
6
Thermocouple
Quartz
Reactor
3 Layered
ACC
Electrode
Tube
Application of
Heat
Flow of N2

7. Measurement of Heel & Pore Volume
7
Measuring Heel
Measuring
Pore Volume
% of Heel =
Mreg = Mass of regenerated ACC
MV = Mass of virgin ACC
(MReg – MV ) / MV x 100%

8. Stage 1: Full Factorial Design
Screening Test with experimental &
made-up data
• Responding Variables
1. Amount of heel: < 5%
2. Pore volume: ≥ 0.8 cm3/g
• Controlled Variables
1. Heating rate: 10 °C/min
2. Electrode Tubes: 1.65 cm dia
• Nuisance Factors
1. Batch of ACC
2. Voltage generator
8
6.8
6.82
6.84
6.86
6.88
6.9
0 1 2 3 4 5 6
PercentageofHeel
Batch of ACC
Percentage of Heel vs. batches
0.722
0.724
0.726
0.728
0.73
0 1 2 3 4 5 6
PoreVolume
Batch of ACC
Pore Volume vs. batches
Effects of Nuisance Variables

9. Stage 1: Full Factorial Design
Low (-) High (+)
1. Kinetic diameter (nm) 0.3 0.8
2. Molecular weight (g/mole) 80 140
3. Heating temperature (°C) 60 260
4. Heating duration (h) 1 3
5. Nitrogen flow rate (L/min) 1 3
No. of runs = 25 = 32
9
Manipulated Variables

10. Stage 1: Full Factorial Design-Results
10
Pareto Chart of Standardized Effects: % of Heel - H
p=.05
(5)N2 Flow Rate-E
3by4
2by5
2by4
3by5
2by3
(1)Kinetic Diameter-A
(2)Molecular Weight-B
(3) Heating Temperature-C
(4) Heating Duration-D

11. Stage 1: Full Factorial Design-Results
11
Pareto Chart of Standardized Effects: Pore Volume - V
p=.05
4by5
3by5
1by5
2by4
1by2
1by4
(3)Heating Temp.-C
(2)Molecular Weight-B
(1) Kinetic Diameter- A
1 by 3

12. Stage 2: Central Composite Design
12
Responding Variables Objective Acceptable Range
i Percentage of Heel (%) To minimize < 5
ii Pore Volume Recovered (cm3/g) To maximize > 0.8
Manipulated Variables -1.682 -1 0 +1 +1.682
A Adsorbates Kinetic Diameter (nm) 0.22 0.25 0.30 0.35 0.38
B
Adsorbates Molecular Weight
(g/mole)
46.4 60 80 100 113.6
C Heating Temperature (°C) 226.4 240 260 280 293.6
Controlled Variables Set Conditions
1 Heating Duration (h) 1
2 Nitrogen Flow Rate (L/min) 1
3 Heating Rate (C/min) 10
4 Electrode Tube 1.65 cm dia
Curvature effects, surface response, 5 levels, No. of runs: 16

13. Central Composite Design: Results
13
Pareto Chart of Standardized Effects: % of Heel-H
p=.05
1Lby3L
1Lby2L
2Lby3L
Molecular Weight-B(Q)
Heating Temp.-C(Q)
Kinetic Diameter-A(Q)
(3)Heating Temp.-C(L)
(1)Kinetic Diameter-A(L)
(2)Molecular Weight-B(L)

14. Central Composite Design: Results
14
Pareto Chart of Standardized Effects: Pore Volume-V
p=.05
2Lby3L
1Lby2L
Kinetic Diameter-A(Q)
1Lby3L
Molecular Weight-B(Q)
Heating Temp.-C(Q)
(3)Heating Temp.-C(L)
(1)Kinetic Diameter-A(L)
(2)Molecular Weight-B(L)

15. Central Composite Design: Surface Plots – % of Heel
15

16. Central Composite Design: Surface Plots–Pore Volume
16

17. Central Composite Design: Regression Model
For Engineering Values
• Percentage of Heel, H = 68.135 – 71.840 Ae + 169.60 Ae
2 + 0.091 Be –
0.456 Ce + 0.0008 Ce
2
• Pore Volume, V = – 4.3724 – 44.7958 Ae + 92.9736 Ae
2 + 0.0311 Be –
0.0009 Be
2 + 0.0818 Ce – 0.000155 Ce
2 – 0.020039 Ae Be – 0.0241919 Ae Ce
17
A = kinetic diameter, B = molecular weight, C = heating temperature.
Subscript “c” : coded value, “e” : engineering value.
Factors Relationships (coded to engineering)
Kinetic Diameter Ae = 0.0482 Ac + 0.3
Molecular Weight Be = 19.982Bc + 80
Heating Temperature Ce = 19.982Cc + 260

18. Fitted Surface: Pore Volume-V
> 1.1
< 1.1
< 1
< 0.9
< 0.8
< 0.7
< 0.6
< 0.5
-2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0
Kinetic Diameter-A
-2.0
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
2.0
MolecularWeight-B
Central Composite Design: Optimum Results
18
Fitted Surface: % of Heel-H
> 8
< 8
< 7
< 6
< 5
< 4
< 3
-2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0
Kinetic Diameter-A
-2.0
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
2.0
MolecularWeight-B
Heel < 5%
Favorable Regions
MATLAB optimum results (coded values):
Kinetic Diameter = -0.44
Molecular Weight = -0.03
Heel = 4.07%, Pore Volume = 0.82 cm3/g
Pore Volume ≥ 0.8 cm3/g

19. Central Composite Design: Optimum Results
19
Favorable Regions
MATLAB optimum results (coded values):
Molecular Weight = -0.03,
Heating Temperature= -0.07
Heel = 4.07%, Pore Volume = 0.82 cm3/g
Fitted Surface: Pore Volume-V
> 1
< 1
< 0.9
< 0.8
< 0.7
< 0.6
< 0.5
< 0.4
-2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0
Molecular Weight-B
-2.0
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
2.0
HeatingTemp.-C
Fitted Surface: % of Heel-H
> 8
< 8
< 7
< 6
< 5
< 4
< 3
-2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0
Molecular Weight-B
-2.0
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
2.0
HeatingTemp.-C
Pore Volume ≥ 0.8 cm3/g
Heel < 5%

20. Central Composite Design: Optimum Results
20
Favorable Regions
MATLAB optimum results (coded values):
Kinetic Diameter = -0.44
Heating Temperature= -0.07
Heel = 4.07%, Pore Volume = 0.82 cm3/g
Fitted Surface: % of Heel-H
> 8
< 8
< 7
< 6
< 5
< 4
-2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0
Kinetic Diameter-A
-2.0
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
2.0
HeatingTemp.-C
Fitted Surface: Pore Volume-V
> 1.1
< 1.1
< 1
< 0.9
< 0.8
< 0.7
< 0.6
-2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0
Kinetic Diameter-A
-2.0
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
2.0
HeatingTemp.-C
s
Heel < 5%
Pore Volume ≥ 0.8 cm3/g

21. Discussions
• VOCs properties (molecular weight & kinetic diameter) more
significant than process parameters (heating duration etc.)
• Obtained results in well agreement with literature
• Models were verified (normal, residual, half normal plots
checked)
• Effects of nuisance factors were checked & found negligible
• Limited time & resources: physical experiments not done in 2nd
stage
• Anticipated results comply best with real results
• Future Works : perform real experiments & verify the results
21

22. Conclusions
• Heel minimized & Pores maximized for moderately high regeneration
temperature, lower molecular weight & smaller kinetic diameter
VOCs
• Optimum results (heel = 4.1%, pore volume = 0.82 cm3/g) identified
for:
• VOCs molecular weight : 79.2 g/mol , kinetic diameter: 0.28 nm
• Heating temperature: 259°C
• Recommendation
• To reduce no. of runs: Fractional factorial in 1st stage
• Taguchi method : better results in least no. of runs
22

23. 23
• Professor Kajsa Duke
• Professor Zaher Hashisho & all my colleagues from Air
Quality Control group
• Ford Motor Company for financial support
Acknowledgement

24. 24
1. Kim, B. R., 2011, “VOC emissions from automotive painting and their
control: A review”, Environ. Eng. Res., 16 (1), pp.1 – 9).
2. D.C. Montgomery, Design and analysis of experiments. 8th edition, John
Wiley & Sons, New York, 2014.
3. Hou, P., Byrne, T., Cannon, F. S., Chaplin, B. P., Hong, S., and Nieto-
Delgado, C., 2014, “Electrochemical regeneration of polypyrrole-tailored
activatedcarbons that have removed sulfate”, Carbon 7 9, pp 4 6 –5 7
4. Dong, L., Liu, W., Jiang, R., Wanga, Z., 2014, “Physicochemical and
porosity characteristics of thermally regenerated activated carbon
polluted with biological activated carbon process”, Bioresource Technol,
171, pp. 260 - 264
References

25. 25