conductivity tracer STUDIES FOR a FLUIDIZED-BED BIOREactor

S.-Y. Leung and R.L. Segar Jr.

Department of Civil Engineering, University of Missouri-Columbia, Columbia, MO 65211, Phone: (573) 882-0075, FAX: (573) 882-4784

ABSTRACT

An automated conductivity tracer test was developed to measure the residence time distribution (RTD) of a cometabolic fluidized-bed bioreactor (FBBR). The FBBR contained sand-core bioparticles grown with phenol and it provided high (70% to 80%) removal of trichloroethene (TCE) at short (3 minute) detention times. The tracer test apparatus was constructed with "off-the-shelf" components controlled with a PC-based data acquisition system. Non-disruptive hydrodynamic testing was obtained during normal operation of the FBBR. The conductivity of injected brine (sodium chloride) pulses was monitored at the reactor inlet and outlet. Dispersion numbers and detention times were computed by fitting the advection-dispersion model to the tracer curves. Typical dispersion numbers attributed to the fluidized-bed of bioparticles ranged from 0.07 to 0.11. In simplified modeling of the FBBR, dispersion was found to have little effect on TCE removal. Based on the dispersion of brine pulses, it was determined that phenol feed pulses injected at inhibitory concentrations over 2 g/L would be rapidly dispersed in the biological bed to non-inhibitory concentrations.

Keywords: conductivity tracer, fluidized-bed bioreactor, dispersion

INTRODUCTION

The performance of a chemical reactor depends on the intrinsic kinetics of the reactions and the transport processes occurring in the reactor. The study of bioremediation processes with tracer tests is essential for linking fluid transport conditions with biokinetic phenomenon The rate of mass transfer is strongly influenced by the degree of mixing and dispersion (Weber and DiGiano, 1995). The residence time distribution (RTD) curves are useful for quantitatively evaluating the degree of mixing and can provide information on the flow pattern. The RTD can be determined experimentally by using tracer techniques (Levenspiel, 1972; Riemer, et al., 1980; Stevens, et al., 1986; and Clark, 1996). Tracer tests are conducted by injecting a non-reactive tracer upstream and measuring its concentration downstream as a function of time. Then curve-fitting techniques are applied to the tracer data to obtain the parameters of hydrodynamic models.

In this research, an automated conductivity tracer test was developed to measure the RTD of a cometabolic FBBR. The FBBR contained sand-core bioparticles grown with phenol under aerobic conditions and provided 70% to 80% removal of 0.1 mg/L trichloroethene (TCE) at a two-minute detention time (Segar, et al., 1997). Measurement of the hydraulic residence time, dispersion number, and bioparticle volume was needed for bioreactor characterization and optimization. Also, an understanding of the dispersive behavior of concentrated phenol pulses used in periodic feeding strategies was needed to optimize cometabolic removal of TCE. Thus, the research objectives were to develop an automated conductivity tracer test for biofilm reactors, apply the dispersion model to analyze the tracer curves, and measure the dispersion number in the FBBR with clean media and biological media for various operating conditions.

MATERIALS AND METHODS

Tracer Technique

Figure 1 shows the configuration of the tracer test apparatus and the FBBR. The apparatus was constructed with "off-the-shelf" components, including a PC-based data acquisition system, a flow-through conductivity probe, a conductivity meter/transmitter, a peristaltic pump for sample withdrawal, and a diaphragm pump for tracer-pulse injection. To initiate the conductivity tracer test, a pulse of sodium chloride brine was injected into the FBBR supply line. Conductivity was continuously monitored at either the top of the inlet packing or at the top of the FBBR bed to obtain the tracer curves. A series of tests were conducted at the FBBR inlet; then, testing was repeated at the FBBR outlet to obtain the effect of the reactor bed. Table 1 lists the operating conditions for tracer tests.

Tracer Data Analysis

The data analysis involved smoothing raw conductivity data by post-processing filters to eliminate sporadic signals caused by biofilm particulates passing through the conductivity sensor, reduction of the raw data frequency, and converting tracer data from conductivity to sodium chloride concentration. The E-curve was calculated directly from the C-curves for pulsed inputs (Levenspiel, 1972). Detention time (), and dispersion number D_h/uL (D_h is the hydrodynamic dispersion coefficient; u is superficial flow velocity; and L is reactor length of tracer curves) were determined with the dispersion model by a nonlinear curve fitting software (Jandel, 1994). The one-parameter model did not fit the tracer data well, so a two-parameter approach was used where both and D_h/uL were varied. The detailed procedure of tracer data analysis is described elsewhere (Leung, 1996).

RESULTS AND DISCUSSION

Inlet Tracer Curves

Tracer curves obtained at the top surface of the inlet packing without fluidized media present were used to define the actual tracer perturbation applied to the fluidized bed. In many previous studies reported in the literature, tracer input was assumed to be an ideal, infinite-concentration impulse of defined mass. However, Figure 2 shows that the measured FBBR inlet pulse was much different from the ideal input pulse. Therefore, the tracer pulse should be measured at the FBBR inlet and outlet dispersion numbers corrected for the non-ideal input.

Input tracer curves were fitted extraordinarily well (R² ranged from 0.96 to 0.99) with the two-parameter dispersion model. Tracer mass recovery ranged from 76% to 101%. Detention times (travel time from concentrated tracer injection point to top of inlet packing) ranged from 2.2 to 3.9 seconds, depending on the flow rate. Dispersion numbers ranged from 0.021 to 0.044 and corresponded to a reduction in pulse peak concentration from 200 g/L for the injected brine to 3 g/L entering the FBBR bed (Leung, 1996). Analogously, the phenol pulses fed in pulsed-feed biological experiments were injected at a concentration of 2 g/L; thus, phenol entered the bed at a concentration of about 30 mg/L. Based on these results, dispersion of injected pulses within the delivery system should be considered in substrate delivery and tracer testing.

Clean Media Studies

The E-curves from FBBR outlet tests with equal volumes of clean 30/35 mesh quartz sand and 40/50 mesh garnet are shown in Figure 3 for 50% bed expansion. Both tracer curves were fitted well with the two-parameter dispersion model (R² ranged from 0.98 to 0.99). The ratio /_H , where _H is the theoretical hydraulic retention time, was examined as an indication of the degree in which the tracer behavior deviated from the dispersion model. The ratio ranged from 1.2 to 1.3, indicating that tracer movement was slightly retarded by media adsorption. Ratios of the first moment of tracer mass to _H ranged from 1.4 to 1.5, confirming the retardation conclusion. Tracer recovery ranged from 112% to 116%.

Bioparticle Bed Studies

The E-curves from a tracer test with biofilm-covered quartz sand are shown in Figure 4. The tracer curves had very high signal noise due to particulate material in the samples and signal filtering was necessary. The two-parameter dispersion modeling fit for these tracer curves was very good (R² = 0.99). The ratios of the first moment tracer mass to _H were higher with bioparticles than with clean sand, ranging from 1.6 to 2.1. Thus, tracer retardation was greater with biofilm present than for clean media. No clear effect of the bioparticles on tracer recovery was observed, as the recovery was highly variable with a range from 40% to 170%.

Table 2 lists the analyzed results of the series of tracer tests. Reported dispersion numbers have been corrected for the dispersion of inlet pulses. Details of the computation procedure may be found in Leung (1996). Higher dispersion numbers were obtained with the bioparticle bed than with the clean sand beds. The presence of biofilm has several effects on tracer curves. Its presence changes the reactor bed hydraulics. The biofilm may interact with the tracer by reversible and irreversible adsorption. Because the biofilm is porous and has a high water content, tracer may diffuse into and out of the biofilm. Such behavior will distort the tracer signal and may indicate greater dispersion than is actually present.

Bioparticle bed dispersion numbers, ranging from 0.07 to 0.11, were used to predict the effect of dispersion on TCE removal in a fixed-bed reactor (as an approximation to the FBBR behavior), assuming homogenous, first-order kinetics. For conditions representative of the FBBR, a 1% to 2% reduction in TCE removal is attributed to dispersion effects when the overall TCE removal is in the range of 90% to 95%. By comparing the reduction in peak conductivity between inlet and outlet tracer curves, phenol pulses would be expected to be diluted by 90% across the bed without any biological uptake. Thus, high stock feed concentrations of phenol, an inhibitory substrate, could be introduced to the reactor as part of a pulsed feeding strategy because of dilution within the feed line and within the fluidized bed.

CONCLUSIONS

The automated conductivity tracer method was useful for obtaining the RTD in the FBBR. Tests were ran rapidly without disrupting operation of the FBBR and the tracer curve data was collected automatically. The two-parameter dispersion model was appropriate for determining the detention time and dispersion number from the tracer curves. Corrections for the effect of the inlet pulse were necessary to produce dispersion numbers attributed to the FBBR bed. The presence of biofilm retarded the movement of the brine tracer through the reactor, presumably due to absorption and diffusion of tracer ions within the biofilm. Thus, tracer curves had longer detention times and higher dispersion numbers with biofilm-covered sand than with clean sand. However, the issue of consistency in and degree of tracer mass recovery in the presence of biofilm remains to be resolved.

ACKNOWLEDGMENTS

The research described in this paper was funded in part by the MU Research Board and in part by the U.S. Environmental Protection Agency under assistance agreement R-819653, through the Great Plains/Rocky Mountain Hazardous Substance Research Center headquartered at Kansas State University. It has not been subjected to the agency's peer and administrative review and therefore may not necessarily reflect the views of the agency, and no official endorsement should be inferred.

REFERENCES

Clark, Mark M. 1996. Transport Modeling for Environmental Engineers and Scientists. John Wiley & Sons, Inc., New York, NY.

Jandel Scientific. 1994. Transforms and Curve Fitting. SigmaPlot for Windows v2.0. Reference Manual.

Leung, S.Y. 1996. "An Automated Conductivity Tracer Test for a Fluidized-Bed Bioreactor." M.S. Thesis, University of Missouri-Columbia, Columbia, MO.

Levenspiel, O. 1972. Chemical Reaction Engineering. 2nd ed., John Wiley & Sons, Inc., New York, NY.

Riemer, M., G. Holm Kristensen, and P. Harremoes. 1980. "Residence Time Distribution in Submerged Biofilters", Water Research, 14: 949-958.

Segar, R.L. Jr., S.Y. Leung, and S.A. Vivek. 1997. "Treatment of Trichloroethene (TCE)-Contaminated Water with a Fluidized-Bed Bioreactor." Annals of the New York Academy of Sciences, 829: 83-96.

Stevens, David K., P.M. Berthouex, and T.W. Chapman. 1986. "The Effect of Tracer Diffusion in Biofilm on Residence Time Distributions", Water Research, 20(3): 369-375.

Weber Jr., W.J., and F.A. DiGiano. 1995. Process Dynamics in Environmental Systems. 1st ed., John Wiley & Sons, Inc., New York, NY.

Table 2. Summarized results of tracer tests.

Media Type	Grade (mesh)	Bed Volume (L)	Flow Rate (L/min)	(sec)
Clean Quartz	30/35	0.43	1.3	23.1	0.052
Clean Garnet	40/50	0.43	1.4	24.1	0.076
Quartz Bioparticles	25/30	1.30	1.2	123.4	0.108

Figure 1. Configuration of FBBR with tracer-testing apparatus.

Figure 2. Conductivity Tracer Injection Input Signals:

(top) Theoretical and Estimated Input Signals at Brine Injection

Point; (middle) Expected Input Pulse at Top of Inlet Packing;

(below) Measured Input Pulse at Top of Inlet Packing.

Figure 3. FBBR outlet E-curves with clean media: (a) 30/35 quartz media; (b) 40/50 garnet media.

Figure 4. FBBR outlet curves with a bioparticle bed: (a) conductivity tracer curve; (b) E-curve after filtering.