; For presentation at the Air Waste Management Association th
Learning Center
Plans & pricing Sign in
Sign Out
Your Federal Quarterly Tax Payments are due April 15th Get Help Now >>

For presentation at the Air Waste Management Association th


  • pg 1
									Paper 98-TA20B.04. In: Proc. Annual Meeting and Exhibition of the Air and Waste Management
Association, San Diego, CA, June 14-18, 1998. AWMA, Pittsburgh, PA. 12 pp.

The Use of CAT Scanning to Characterize Bioreactors for
Waste Air Treatment
Marc A. Deshusses and Huub H.J. Cox
Department of Chemical and Environmental Engineering
University of California
Riverside, CA 92521
Daniel W. Miller
Loma Linda University Medical Center
Dept. of Radiation Medicine
11234 Anderson St. B121
Loma Linda, CA 92354
Computed Axial Tomography (CAT) scanning was used for the first time to characterize the
structure of packed beds biofilters and biotrickling filters used for waste air treatment. A detailed
analysis of CAT scan images demonstrated the heterogeneities of the channels through which air
-and water, in the case of biotrickling filters- flow. Images of cross sections of a biotrickling filter
containing about 70% vol. of biomass revealed that large areas were completely filled with
biomass, whiles others had only little. The size of air/water channels ranged from less than 5
mm2 to380 mm2. Numerical analysis of a CAT scan image of a biotrickling filter showed that the
interfacial area was 101 m2 m-3. Although this value is higher than expected for a reactor that was
close to clogging by biomass, it is consistent with the hypothesis that biofilm roughness
contributes to a significant extent to the increasing the interfacial area for pollutant mass transfer.
CAT scanning appears to be a promising technique to quantitatively characterize air flow
dynamics in biofilters and biotrickling filters.
Biological waste air treatment is a new and promising technique which utilizes mixed microbial
populations to degrade gaseous chemicals. Pollutant vapors are sorbed into an aqueous phase
prior to biodegradation.1 This technique is most suited for the treatment of large air streams
containing low concentrations of odoriferous compounds and/or volatile organic chemicals,
primarily solvents. Biological waste air treatment is gaining interest in the US with the future
implementation of the Clean Air Act Amendments. The most promising reactors for biological
waste air treatment are biofilters and biotrickling filters.
Interestingly, very little is known about biofilm-air interfaces and three-dimensional structures of
packed beds in biofilters and biotrickling filters. This causes a problem, for example in the
development of mathematical models for biofilters or biotrickling filters where knowledge of the
interfacial area and of the biofilm thickness is required. Usually, these parameters are difficult to
obtain experimentally and are either guessed or fitted. In rare cases, either visual observation or
random sampling for micrometric analysis lead to major approximations. Further, it has been

recently demonstrated that biofilms are far from the planar geometry generally assumed for
modeling purposes,2,3 but to date, no study has attempted to quantify biofilm roughness and the
implication for pollutant mass transfer. Finally, bioreactors for waste air treatment are also
subject to changes over time. This is mostly due to biomass growth in the case of biotrickling
filters, and due to bed compaction in biofilters. These changes have never been characterized
from a structural point of view, even though they have drastic effects on the efficiency of
pollutant removal4 and on the costs of waste air treatment. In this paper, preliminary results of a
new approach to characterize packed bed structures of biofilters and biotrickling filters and to
determine the interfacial area are presented and discussed. The approach uses Computed Axial
Tomography (CAT) scanning, a technique usually reserved for medical diagnostics. During
scanning of an object, a circular array of x-ray diffraction scans is acquired. Using deconvolution
techniques, a cross-section of the scanned object is constructed. CAT scanning is non-destructive
and it has a high resolution. Since x-rays absorption is roughly proportional to the density of the
material being investigated, CAT scans images allow to differentiate air and biofilm from
support material. Recent developments in medical imaging have further allowed to combine the
information of adjacent “slices” to reconstruct three-dimensional structures. Clearly, there is an
unexplored potential to use CAT scanning techniques for the characterization of gas phase
bioreactors such as biofilters and biotrickling filters.
Biofilter and Biotrickling Filter
Details of the biofilter and biotrickling filter used in this study have been described
elsewhere.4,5,6 In summary, the reactors were made of 15.2 cm internal diameter clear PVC
tubing and had a packing height of 1.3 m for the biotrickling filter, and three separated sections
of 50 cm for the biofilter. The packing for the biotrickling filter was 2.5 cm polypropylene Pall
rings (Flexirings®, Koch Engineering, Wichita, KS), while the biofilter was filled with an 80/20
by volume mixture of wood chips (1-3 cm) and compost. Both the biofilter and the biotrickling
filter were operated in a downflow mode and effectively degraded toluene as a model pollutant.
Prior to CAT scan, the biotrickling filter was allowed drain for at least 3 hours. CAT scans
discussed in the present paper were acquired for the following conditions:
1. Biofilter operating since about one year, medium moisture content about 65% wet weight
   (scans at about mid reactor height)
2. Biotrickling filter nearly clogged by growing biomass, biomass content: about 70% (scans at
   mid reactor height and reactor bottom)
CAT Scanning and Image Treatment
CAT scans were performed on a General Electric clinical scanner at Loma Linda University
Medical Center. In general, a spacing between adjacent scans of 3 mm was chosen. The 512 x
512 pixels images had a resolution of approximately 0.3 mm. ACR-NEMA format images were
analyzed using the Osiris 3.1 software (University Hospital of Geneva, Switzerland). For specific
calculations such as the determination of single air channel area and perimeter, a custom C++
code was developed at UC Riverside.

A typical image of a biotrickling filter cross-section is shown in Figure 1. On this image, the
biofilm (in off-white) covering the Pall rings (gray rings with cross internal structure) can easily
be distinguished from air/water channels (black spots). At the time of the CAT scans, this
bioreactor was partially clogged with biomass (70% of reactor volume) which explains the high
content of biomass. Obviously, Figure 1 shows that biomass distribution within the biotrickling
filter is very heterogeneous. While some Pall rings appear to be covered by only a relatively thin
biofilm (A, on Figure 1), some regions of the reactor are completely filled with biomass (B). In
this latter regions, oxygen limitation and/or pollutant transfer limitations are very likely to occur.
This may explain why compounds such as trichloroethylene (TCE) or perchloroethylene (PCE)
recalcitrant under aerobic conditions but degradable under anaerobic conditions have sometimes
been shown to be degraded in gas phase bioreactors.6 A more thorough evaluation of Figure 1
shows that in addition to a number small channels, relatively large channels exist. The exact
dynamic of trickling water and air in these channels during treatment would require simultaneous
scanning and reactor operation: an impossible task at this time. However, a rough comparison of
the actual space used by the trickling water (C on Figure 1, calculated as the average dynamic
hold-up divided by the bed height), with the total area of air channels (sum of black spots) shows
that most of the free space in the bioreactor was occupied by trickling water. Hence there was
probably not much direct air-biofilm contact. Consequently, as biomass grows in the biotrickling
filter, the pollutant transfer path may switch from essentially gas-biofilm at low biomass content,
to gas-liquid-biofilm at high biomass content. This will certainly have significant implication as
far as pollutant elimination performance is concerned.
In a similar manner as for the biotrickling filter, a scan of typical cross section of a biofilter is
shown in Figure 2 and a close view is shown in Figure 3. While distinguishing biofilm from
compost is very difficult in these figures because of their similar absorption of x-rays, a number
of wood chips (B, on Figure 2) and some fine structures can be distinguished. Interestingly
Figure 2 suggests that air undergoing treatment probably travels inside a few major channels,
rather than in small diameters pore as it would in homogenous porous medium. This may be due
to the fact that the biofilter had been operated for more than a year and that bed compaction may
have occurred over time. Also, at the time of the measurement, the medium moisture content was
high (65%) so that significant amounts of water was held by capillary forces in small interstitial
pores of the medium.
Gas-biofilm or gas-medium interfaces in the Figure 2 biofilter can qualitatively be viewed on
Figures 4 and 5, where the pixel densities in the CAT scan corresponding to interfaces appear in
white and gray while air and support medium remains black. These figures highlight the
heterogeneous nature of the biofilm surface even at the sub millimeter scale (Figure 5). They also
indicate that most air channels are on the side (Figure 4), hence that significant wall effect must
have occurred in this biofilter. Further work, including scanning at a higher resolution is
A quantitative determination of the interfacial area is possible by analyzing each image and
defining the perimeter and the surface of each air channel. The interfacial area can then be
evaluated using Equation 1.

Interfacial area = Sum of the perimeter of all air channels / Reactor cross-section area (m-1) (1)

This was performed on the biotrickling filter image showed on Figure 6 (left) using a custom
C++ code. While further calibration of the exact threshold pixel value for the air-biofilm
boundary is still needed, preliminary results of channel size distribution are shown in Figure 7.
81 air channels were observed that correspond to the white spots in Figure 6 (right side). Figure 7
reveals that most of the channels fall within the range of 0 to 20 mm2 and that a few large
channels (>100 mm2) existed. This is most probably specific to the biotrickling filter that was
analyzed. In order to generalize such data, analysis of several images and statistical treatment of
the results is presently being performed.
A bed porosity of 10% was be determined by dividing the sum of all channels area by the cross
section area. This value corresponds to the bed porosity determined by comparing the weight of
the reactor with and without biomass (12-17%). Further, Equation 1 allowed to determine the
interfacial area which was found to amount 101 m2 m-3 in the present case. This is a fairly high
value considering that the original support area was 220 m2 m-3 and that significant reduction of
interfacial area was expected from the biomass overgrowth at the time of the CAT scan
measurements. A possible explanation to the high interfacial area value observed, is that the
interfacial surface area could be greatly enhanced by microscopical heterogeneities of the
biofilm. This is supported by independent confocal laser microscopy studies which found that
large channels existed from the air interface to the substratum.3 Cumulative interfacial area data
computed using Equation 1 are shown in Figure 8. The data show that most of the interfacial area
is obtained in air channels of surface smaller than 60 mm2. Again, further analyzes are needed to
determine whether this is specific to the reactor that was analyzed or if this can be extrapolated to
biotrickling filters in general. At this time, a statistical treatment on various CAT scan images
obtained for biofilters and biotrickling filters operated under different conditions is being
performed in order to improve the accuracy of interfacial area determination and define trends so
that a general theory can be developed.
CAT scanning of biofilters and biotrickling filters provides a unique opportunity to progress in
our fundamental understanding of bioreactors for air pollution control. While for the first time,
the interfacial area was experimentally determined, ongoing work at UC Riverside involves the
reconstitution of the three-dimensional structure of the various air channels, so that the air flow
dynamics can be simulated. This will ultimately allow to quantify the interfacial surface area
which is in continuous contact with the contaminated air, and improve the predictability of
biofilter models.
1. Cox, H.H.J.; Deshusses, M.A. Biological waste air treatment in biotrickling filters. Current
   Opinion in Biotechnology 1998, vol 9/3: in press.
2. Hugler, W.C.; Cantu-De la Garza, J.G.; Villa-Garcia, M. Biofilm analysis from an odor-
   removing trickling filter. In Proc. of the 89th Annual Meeting and Exhibition of the Air &
   Waste Management Association, Air & Waste Management Association, Pittsburgh, PA,
   1996; paper 96-RA87A.04: 20 pp.

3. Moller, S; Pedersen, A.R.; Poulsen, L.K.; Arvin, E.; Molin, S. Activity and three-dimensional
   distribution of toluene-degrading Pseudomonas putida in a multispecies biofilm assessed by
   quantitative in situ hybridization and scanning confocal laser microscopy. Appl. Environ.
   Microbiol. 1996, 12, 4632-4640
4. Cox, H.H.J.; Deshusses, M.A. Elimination of toluene vapors in biotrickling filters:
   Performance and carbon balances. In Proc. of the 91th Annual Meeting and Exhibition of the
   Air & Waste Management Association, Air & Waste Management Association, Pittsburgh,
   PA, 1998; paper 98-20B.03: 12 pp.
5. Cox, H.H.J.; Deshusses, M.A. The use of protozoa to control biomass growth in biological
   trickling filters for waste air treatment. In Proc. of the 90th Annual Meeting and Exhibition of
   the Air & Waste Management Association, Air & Waste Management Association,
   Pittsburgh, PA, 1997; paper 97-RA71C.05: 10 pp.
6. Deshusses, M.A.; Johnson, C.T.; Hohenstein, G.A.; Leson, G. Treating high loads of ethyl
   acetate and toluene in a biofilter. In Proc. of the 90th Annual Meeting and Exhibition of the
   Air & Waste Management Association, Air & Waste Management Association, Pittsburgh,
   PA, 1997; paper 97- WA71A.07: 13 pp.
7. Devinny, J.S.; Webster, T.S.; Torres, E.; Basrai, S. Biofiltration for removal of PCE and TCE
   vapors from contaminated air. Hazardous Waste & Hazardous Materials, 1995, 12, 283-293

Our thanks are due to the management of the Radiation Medicine at the Loma Linda University
Medical Center for the generous allocation of scanner time, and to Edmundo Anton Chow, UC
Riverside Computer Science for the development of the C++ code.


                   Reactor wall

     Pall Ring

                       Air channel


Figure 1. CAT scan of the cross section of a biotrickling filter (I.D.=15.2 cm). The dashed
square represents the average cross-section area of the dynamic hold-up in the biotrickling filter.
Conditions: biotrickling filter near clogging, lower section of the biotrickling filter. (A) Pall ring
covered with only little biomass; (B) clogged regions; (C) approximate average space occupied
by the trickling water.

                                               LOMA LINDA UNIV MED CTR
0-00000                                                    CT HLA2OC0
                                                           CT HLA2OC0




Figure 2. CAT scan of the cross section of a biofilter (I.D.=15.2 cm). Conditions: wood chips-
compost biofilter after one year operation (moisture content approximately 65%), mid-height
section. (D) a wood chips. The inset is magnified in Figure 3.

                            5 mm

Figure 3. Detail of Figure 2.

REACTOR AFTER INF SEG, O                                               LOMA LINDA UNIV MED CTR
                                                                       LOMA LINDA UNIV MED CTR
0-00000                                                                            CT HLA2OC0


Figure 4. CAT scan of the cross section of a biofilter. The white area corresponds to the gas-
biofilm or gas-medium interface. Conditions: as in Figure 2.

                                5 mm

Figure 5. Detail of Figure 4.

REACTOR, O               LOMA LINDA UNIV MED CTR                              REACTOR, O                                LOMA LINDA UNIV MED CTR
23 23 23                             CT HLA2OC0
                                     CT HLA2OC0                               23 23 23                                              CT HLA2OC0
                                                                                                                                    CT HLA2OC0

1996.12.19                                                                    1996.12.19

Figure 6. CAT scan cross-section (left) and processed image (right) for the analysis of air
channel distribution and interfacial area of Figure 7. Conditions: biotrickling filter near clogging,
scan was performed at mid reactor height.

          25                                                                                                                          400
          20                                                      Occurrence

                                                                                                                                            Sum of channel area
          15                                                                                                                          250

          10                                                                                                                          150
             0                                                                                                                        0

                                      Channel area range (mm2)

Figure 7. Distribution of air channels for the biotrickling filter section shown in Figure 6. The
number of air channels and the sum of the channel area is reported. Conditions: as in Figure 6.









                                               Channel area (mm2)

Figure 8. Cumulative interfacial area as a function of the size of the air channels. Conditions: as
in Figure 6.


To top