Pergamon

0956-053X(95)00036-4

Waste Management, Vol. 15, Nos 5/6, pp. 351-357, 1995 Copyright © 1996 Elsevier Science Ltd Printed in the USA. All rights reserved

0956-053XJ95 $9.50 + 0.00

ORIGINAL CONTRIBUTION

BENCH-SCALE STUDIES OF REACTOR-BASED TREATMENT OF FUEL-CONTAMINATED SOILS

Dennis D. Truax, Ronald Britto and Joseph H. Sherrard Department of Civil Engineering, Mississippi State University, Mississippi State, MS 39762, U.S.A.

ABSTRACT. Biological treatment of hazardous wastes in groundwater and soils has recently assumed great importance. In particular, biotreatment of wastes from accidental spills or underground storage tank leaks has generated interest in bioremediation as a natural, economical mechanism for site decontamination. Because of drawbacks related to batch sys- tems, and the successful use of continuous flow treatment of wastewater for several decades, it was felt that continuous treatment of such soils would be a feasible alternative treatment technique. To this end, bench-scale bioreactor treatability studies were conducted. This study used contaminated soil made in the laboratory using No. 2 diesel fuel and sand. Con- tamination levels studied were from 1335 to 6675 mg (TPH) as derived from No. 2 fuel oil per kg sand. Variation in mean cell age was obtained between reactors, with sufficient nutrients and oxygen made available to ensure the fuel oil organics were the only limit to microbial growth. A theoretical biokinetic model was formulated based on Monod's theory of limiting substrate and continuous cultures. Biokinetic constants and removal efficiencies were evaluated. The off-gases CO2 and volatile hydrocarbons were monitored to allow mass balance analysis of the process. The solids retention times for evaluating final TPH concentration of 100 mg/kg were also calculated. The results of this investigation showed that continuous bioreactor treatment is a viable option in the treatment of diesel-contaminated sandy soils. Removal efficien- cies of up to 91% were attained at a loading of 1335 mg TPH/kg wet sand, operated at a biological solid retention time (BSRT) of 60 days. Experiments also showed that TPH desorption and volatilization were not rate-limiting in the overall removal process. Sand-to-moisture ratios in excess of 3 : 1 were also shown to retard TPH removal rates very little. How- ever, biokinetic constants were found to vary over a range of values. This was particularly true at varying diesel loading levels. Nevertheless, significant removal efficiency (up to 86%) was noted at the highest loading level tested, 6675 mg TPH/kg wet sand.

INTRODUCTION

A significant quantity of petroleum compounds have been released into the soil environment. These releases have been either from leakage of underground storage tanks (USTs) or from accidental spills. The U.S. EPA has reported recently that 100,000 UST leaks have been identified, and that the number might triple over the next few years, t A wide range of methods have been used to treat these contaminated sites.

Treatment processes have incorporated physical, chemical or biological phenomena, or a combination of these processes. In recent years bioremediation has been claimed to be an inexpensive, natural method of clean-up of petroleum soils. Both in situ and ex situ treatment schemes of bioremediation have been shown to be feasible. Ex situ processes include landfarming, composting and bioreactors. To date, bioreactors developed for treatment of

RECEIVED 19 JANUARY 1995; ACCEPTED 20 JUNE 1995.

351

petroleum-contaminated soils have employed a batch-type configuration and have suffered from drawbacks such as: (1) protracted periods of time for microbial acclimation to take place; (2) extensive, above-ground surface area requirements; (3) lack of control of volatiles and hence imperfect material balances; and (4) poor understanding of microbial mechanisms and biokinetics, and the factors influ- encing them.

Thus far, continuous biotreatment of contaminated soils has not been implemented in practice. Laboratory continuous systems have been tested for a wide range of organic compounds, mostly at high solu- tion : soil r a t iosY It is hypothesized that continuous systems of soil treatment can be a viable alternative to batch methods. Further, it should be possible to use smaller solution:soil ratios (close to saturation moisture levels) and operate a completely mixed continuous type system of contaminated soil treat- ment. Finally, it should be possible (and it is in fact essential) to evaluate the kinetics of continuous

352 D . D . TRUAX ET AL.

treatment systems by conducting bench-scale studies. The purpose of this paper is to present preliminary findings of a pseudo-continuous reactor system for remediation of diesel-contaminated sandy soils to con- firm this hypothesis. In addition, biokinetics which conform with microbial physiology and growth kinetics were evaluated at steady-state through the use of a model presented here.

The selection of diesel oil and sandy soil for the study was made keeping in mind the main objective of this study: to focus on biological removal and to evaluate true biodegradation rates. Sand with a fixed gradation range between 75 and 600 ixm was used as the soil type. The percentage of organic matter in the soil was negligible (organic fraction < 0.001). The use of sand rather than fine soils considerably diminished the effects of media characteristics, and hence physical phenomena such as adsorption, which might otherwise have limited microbial removal rates. The low organic matter content also aided in defining the removal process as biokinetic-controlled, rather than sorption-controlled. The rate of desorption for this sandy soil is expected to be greater than the rate of substrate utilization, and will therefore not govern overall removal rates. The low soil distribution coefficients reported for this type of soil further diminished the importance of adsorption phenomena in this study, and helped focus on the primary goal of studying a microbial-based removal system.

The choice of diesel as fuel contaminant type was made because of its lower rate of volatilization than fuels such as kerosene and gasoline. Volatilization as a means of physical removal was thus considerably lessened and microbial phenomena could hence be evaluated correctly. Any diesel which volatilized, however, was trapped in an organic solvent and measured, as will be explained below.

To investigate the system, a series of bench-scale reactors was used to study biodegradation of diesel in sandy soils. Pseudo-continuous loading of reactors was achieved by addition of small portions of sand and diesel fuel. Removal of treated soil in amounts equal to this addition was performed concurrently and thus a constant reactor volume was maintained. By varying the amount of soil removed between reactors, a variation in mean cell age of the microbial population was obtained. Since the reactor system was a soil-water system, some fundamental assumptions were made. Growth kinetics were outlined, and the biokinetic constants at steady state are presented.

Growth Rate Mode l In both batch and continuous-flow systems, the rate of growth of the biomass can be defined by the following relationship

where rg = rate of bacterial growth (mass/unit mass soil/time), IX = specific growth rate (t ime l) and X = concentration of micro-organism (mass of total volatile solids (TVS) per unit weight dry soil).

The effect of a limiting substrate was outlined by Monod 4 in the following equation

ix = Ig~Se/(Ks +Se) (2)

where Ixm = maximum specific growth rate (time-l), Se = concentration of growth-limiting substrate in the system (mass/unit weight soil), and K s = half- velocity constant, substrate concentration at one- half the maximum growth rate (mass/unit weight soil). Hence, the rate of microbial growth can now be expressed as

r s = IXmXSe/(Ks + Se) (3)

In any biological system, the distribution of cells covers a range of physiological conditions. During any stage of growth, part of the organic material is oxidized for generation of energy and cell maintenance. Predatory organisms also need to be accounted for in the growth equation. All these factors are combined to develop what is generally termed as endogenous growth (rd), and can be formulated as

rd = kdX (4)

where k d = endogenous decay coefficient (time-l). Defining the net rate of bacterial growth, r'g, as the corrected rate of growth yields

r'g = rg - r~ (5)

o r

r'g = [I~XSe/(Ks +S,)]- k d X (6)

Mass Balance As can be recognized from the schematic diagram in Fig. 1, the cell mass balance of the system follows the general word equation

Accumulation = Inflow - Outflow + Net Growth (7)

The mathematical expression for the mass balance of the reactor system can therefore be expressed as

M(dX/d t ) = Q X o - Q X + Mr'g (8)

where dX/d t = the rate of change of micro-organism concentration in the reactor measured in terms of total volatile solids (mass TVS/unit weight soil/time);

So,Q _1 ]

OFf-gas /

M,X Se'Q

Re(lctor / r s = / ~ X (1) FIGURE 1. Mass balance of materials around reactor.

BENCH-SCALE STUDIES OF REACTOR-BASED TREATMENT 353

Vol, v e

~ C I ] 2 ADSI]RBER

( f o u n d on at[ u n i t s )

FIGURE 2. Reactor system set-up.

M = reactor mass contents; Q = mass loading rate (mass/time); Xo = concentration of micro-organisms in influent, (mass TVS per unit weight wet soil), and X = concentration of micro-organisms in reactor, (mass TVS per unit weight soil).

Using the equation representing the value of r'~ obtained previously, the resulting equation is

M(dX/dt) = QXo - QX + M{[l~XSJ(Ks + S~)] - kdX} (9)

It can be assumed that the concentration of the micro-organisms in the influent can be neglected (Xo = 0) and that steady-state conditions prevail (dX/dt = 0), and if the above equation is divided by X, the following equation will result

1/O = {/~mSe/(gs + Se)} - k d (10)

where O = average detention time, M/Q. Further, for a single-stage reactor, the average detention time within the reactor must equal the average biological solids retention time (BSRT), or cell age Oc.

Assumptions in the Growth Rate Model and Mass Balance The assumptions made in using Monod's model and the formulation of the mass balance are stated below. These assumptions were necessary to make preliminary deductions for biotreatment of diesel taking place in a dual media system.

1. Complete mixing of the reactor system contents took place at all times. Uniform concentrations prevailed throughout the reactor volume. The total petrol hydrocarbons (TPH) and biomass concentrations in the daily effluent withdrawn from the reactor were the same as the concen- tration at any point within the reactor volume with a margin of error of + 5%.

. Microbial degradation took placedue to both suspended and attached growth. In the absence of evidence indicating different rates of removal of suspended and attached micro-organisms, the assumption was made that growth and degradation patterns did not vary significantly between the two types.

MATERIALS AND METHODS

A system of five bioreactors was set up in the labo- ratory. These reactors were placed on a common shaft drive, as shown in Fig. 2. The reactors were equipped with rollers which facilitated continual rotation. Each reactor was 0.42 m (1.39 feet) in length and 0.25 m (0.825 feet) in diameter, giving a reactor volume of 20.9 liters (0.738 cubic feet). Fig- ure 3 is a schematic drawing of a single bioreactor unit.

Each of the reactors was operated at different bio- logical solids retention times (cell age), Oc. At steady state, biokinetic constants can be determined from known values of Q, Se and X and the following derived relationships in the slope-intercept form

[XO/(So-Se)] = [(K,/k)(I/Se)1 + [(I/k)] (11) and

[1/O1 : [Y(So-So)(1/xo)] - [kd (12)

The amount of moist soil in each reactor varied from 5.0 to 6.0 kg. Each reactor contained the same weight ratio of dry soil to water, 3 : 1. Reactor con- tents were removed from each reactor daily, varying in amounts from 150 to 300 g on average. The same amount of soil was then added to the reactors as was removed to establish a pseudo-continuous load- ing rate. Through the removal and addition of fixed

354 D.D. TRUAX ETAL.

Ba?~ies--~ A.~ I ~---L ~ 2ZZ2 O~'~-g as ~--~ t Air Inle± Du~Le~

J I' r----~ I ~..

AJ

' ~_Rollers_/"

Cross-sec±ion A-A

3aFFle~. ~ le~r Plexlgtass

i ha l± p~.i .. ...

- - Air inte±/ Ou±le± Nozzle

FIGURE 3. Details of unit reactor.

amounts of moist soil from each reactor, a variation in solids retention time was obtained. Entering air was previously stripped in a carbon dioxide absorp- tion flask, passed through a flowmeter to determine the amount of air that entered the manifold, and then distributed evenly to each reactor.

Analysis of Total Petroleum Hydrocarbons ( TPH) Off-gas from each reactor was sparged through a flask containing methylene chloride acting as an organic solvent to extract volatilized materials from the gas phase. Treated soil samples were washed repeatedly with this same solvent to concentrate fuel organics for analyses. The TPH procedure used to quantify organics in the off-gas and effluent followed the Soxhlet extraction process described by Martin et al. 5

Microbial Counting Microbial counts (colony-forming units, c.f.u.) were accomplished by the spread plate technique with the growth media comprising an oil agar, yeast extract and peptone.

Biomass Measurement The amount of biomass in the reactors was mea- sured by the loss-on-ignition method, analogous to

the measurement of volatile suspended solids (VSS) in wastewater treatment. 7 The amount of fuel that remained in the effluent was fractionally subtracted from the total volatile solids to obtain an approxi- mate cell constituent mass.

Carbon Dioxide Trapping Carbon dioxide in the off-gas was measured using a potassium hydroxide (KOH) trap and then back titrating with an acid.

Quality Assurance~Quality Control All analyses were performed in triplicate with a blank carried through the experiment. For the mea- surement of TPH, triplicate samples, blanks and an original fuel source sample were measured to verify the methodology.

RESULTS

The soil bioreactors were run for a 90-day period. Bioreactor steady state was reached at or before 70 days. Steady state was confirmed from constant (within _+ 5%) biomass and TPH concentrations in the soils removed from the bioreactors. During the period following steady state, data were collected repeatedly to facilitate mass balance analysis for each of the reactors. These data are presented in Table 1.

Since the TPH loading levels in reactors were as high as 6675 mg/kg wet sand, volatilization rates were measured both at the start of the study and at steady state. At the initiation of the study, volatiliza- tion of TPH from the reactors was continuously monitored for a period of 36 h. Results of these studies indicated a low fraction of volatilization at all influent diesel loading levels, with less than 0.5% of the TPH materials being carried out of the reac- tor in the off-gas. Later during the experiment, after steady-state conditions were attained, six of the reac- tors covering the range of BSRTs and the higher

TABLE 1 Summary of Laboratory Data

Test no. Solids Effluent Biomass

Retention Time Quality, S e Content, X (days) TPH (mg/kg)* (mg/kg)*

1 20.1 555 + 25 750 + 78 2 24.3 480 + 18 867 + 70 3 28.6 430 + 18 886 + 51 4 31.1 382 + 14 1,015 + 68 5 40.1 272 + 14 940 + 82

Initial contaminant level, S o , for reported values was 1751 nag TPH/kg wet soil. *Summarized effluent and biomass readings are the average values of seven trials.

BENCH-SCALE STUDIES OF REACTOR-BASED TREATMENT 355

Removal Efficiency (%) I00

80

6O !

40

20

D

D

O O

i t i i

0 10 20 30 40 50 BSRT (days)

FIGURE 4. Variation in removal efficiency with BSRT.

{X(BSRT)/(So-Se)] 4C

2£

1£

0 I i i i

0 1 2 3 ( I /Se ) x lO00

FIGURE 6. Determination of biokinetic constants.

TPH loading levels were studied for TPH content in the off-gas. These examinations delineated that between 1.0 and 2.3% of influent fuel was being removed by volatilization. These results were deemed to demonstrate clearly the insignificance of volatili- zation as a contributing factor in TPH removal during this study.

As part of the Quality Assurance/Quality Control (QA/QC) program, reactor contents were measured at the end of the study to ascertain that significant deviations did not occur. Marked variation in this parameter would necessitate corrections in mass balance analyses and biokinetic relationships. It was found that the dry mass in the reactor had increased from 5.0 kg initial mass to approximately 5.2 kg over the study period. This value equates to less than 3% deviation for this period.

Figure 4 depicts the substrate removal efficiency versus the biological solids retention time (BSRT). A general trend develops over the plot, showing that as BSRT increased the removal efficiency increased. Figure 5 is a plot of biomass versus BSRT. Again, a trend is detected showing the biomass concentration to increase as BSRT increases, reaching a maximum value and decreasing at the highest BSRT. It may be expected that at BSRTs greater than 40 days the biomass values could decrease further following typical growth and energy use patterns. Regarding both these figures, it should be noted that X and Se varied

randomly over the period following the 70-day mark, thus indicating that the estimate of 70 days as the time required to reach steady state is valid.

The amount of carbon dioxide trapped as off-gas from the reactors was found to have an average value of 265 mg/kg dry mass soil (as CO2-C) at steady state. This works out to be about 15% of organic carbon fed, the rest being converted to cell mass and other metabolic end-products.

The biomass counts taken on spread plates indi- cated high numbers, averaging 7.1 × l 0 9 c.f.u. per gram soil in reactors at steady state. The number of colonies increased from 6 × l08 on average in the reactor with lowest BSRT to 1.9 × l0 ~° in the reac- tor with the highest BSRT. These values correlate well with microbial densities observed in fuel spill sites under favorable environmental conditions and maximum microbial proliferation levels. 7

The limited studies available on adsorption, such as determination of soil-distribution coefficients, indicate that desorption of diesel oil from the sand media was not a governing factor in microbial removal. Approximate TPH desorption rates were compared with microbial TPH utilization rates. With the exception of the 10-day BSRT data, TPH desorption rates were noted to be higher than the microbial utilization rates. Therefore, though a frac- tion of the TPH is adsorbed to the soil grains, the free phase or aqueous phase fuel fraction (which is

Btomass, mg VS$/kg sand 1200

1000

800

600

400

200

0

0

O

0 0

0

i i i i

I0 20 30 40 BSRT (days)

FIGURE 5. Variation in biomass levels with BSRT.

50

/BSRT) x IO0 D

I 2 3 4 5 [X(BSRT)/(So-Se)) x 100

FIGURE 7. Determination of biokinetic constants.

356 D.D. TRUAX ET AL.

TABLE 2 Summary of Biokinetic Constants

Parameter Variable Value*

Maximum substrate utilization rate k 0.113 day l

Half-saturation constant K s 877 mg TPH/kg sand Maximum yield coefficient Y 0.98 mg TVS/mg TPH Endogenous decay coefficient k d 1 × 10-3-3 day -1 Maximum specific growth rate I/,m 0.11 day -1 BSRT for final TPH

concentration of 100 mg/kg Oc 1°° 97 days

* For an initial contaminant level, S o, of 1313 mg TPH/kg sand.

presumed to be more amenable to biodegradation) is available in sufficient concentrations to eliminate adsorption as a controlling factor.

Model Calibration To determine the kinetic constants for the model presented earlier, Figs 6 and 7 were prepared. These constants are presented in Table 2. Among the eval- uated biokinetic constants of interest are the high value of the half-saturation constant, Ks (877 mg/kg), and the low maximum growth rate, /zm (0.10/day). However, after reviewing available litera- ture values of biokinetic constants for industrial wastes the calculated values obtained are not unusual. 8 Low/Zm and high Ks values are typical of wastes that are biodegraded slowly in comparison to more easily degraded substrates such as those encountered in generic wastewater treatment. Growth-substrate plots based on classical Monod or Michaelis-Menten formulations show much flatter rectangular hyperbolas for wastes that are harder to degrade. Oily substrates such as No. 2 diesel fuel fall in this category, and hence the reported observations.

The calculated maximum yield is 0.98, which is higher than microbial yields reported for municipal wastes. A possible reason for this high value is that wastes with long-chain hydrocarbons such as diesel fuel are in a very reduced state and need less energy for synthesis of new cells)

Reactor Performance Data A second phase of this study was designed to test for effect of moisture content on effluent TPH values and biokinetic parameters at varying BSRTs. Oper- ating conditions were similar to those used above, with reactors maintained at solid retention times (BSRTs) ranging from 20 to 60 days, Soil:moisture ratio was varied between reactors; the highest ratio of sand : water was 3 : 1. This level produced a satu- rated soil. Other sand : water ratios used were 2.7 : 1, 2.4:1 and 2.1 : 1. The diesel loading level was 1500 mg diesel/kg wet sand.

A comparison was conducted of average biomass values and effluent TPH levels at different soil : mois- ture ratios and a single BSRT. The data indicate a very small variation of these parameters with chang- ing soil:water ratios. Hence, these results tend to confirm that transport and surface factors like advection, dispersion and sorption (all of which decrease as moisture content increases) did not impose limitations on substrate availability and biomass growth.

Continuous bioreactor treatment of diesel-con- taminated sandy soils is a feasible option. Operating at an influent TPH loading level, So, of 1335 mg/kg wet sand, laboratory studies produced an effluent TPH level of 118 mg/kg wet sand at a BSRT (Oc) of 63 days. This is equivalent to a TPH removal effi- ciency of 91.2%. At the same TPH loading condi- tions, the theoretical BSRT required to achieve an effluent TPH level of 100 mg/kg wet sand was deter- mined to be 93 days.

It was interesting to note that TPH removal was significant at a BSRT as low as 10 days. TPH re- moval itself, however, does not indicate the degree of metabolism or mineralization of the diesel com- ponents. At the 10-day BSRT, the removal efficiency showed an inconsistent pattern of change with increasing TPH loading levels. The first step in the initial microbial conversion of the readily degradable fraction of the TPH at the 10-day BSRT was rapid. Results indicated that about 53% of the TPH was quickly transformed, while the remaining fraction took longer to degrade. Two reasons are therefore offered for this observation. First, higher boiling fractions of diesel would naturally take a longer time to degrade. These higher fractions probably com- prise the remaining 43% fraction of the fuel after the initial rapid TPH removal. Second, the higher boil- ing fractions probably need a different consortium of micro-organisms for degradation. The develop- ment of the necessary microbial population takes a longer time. Longer times for microbial development result in the observed increased times for degrada- tion to take place.

Other results showed that effluent TPH levels are not independent of influent TPH loadings, So. The removal efflciencies, however, did not show a signifi- cant variation when the influent loading level was increased from 1335 to 6675 mg TPH/kg wet sand. From a practical standpoint, removal efficiency is often the primary objective. Consistent removal effi- ciencies can be thus be attained over a wide range of TPH influent loading levels. If TPH effluent level itself is the main target, however, it can be achieved by the adoption of a higher BSRT. From the trends observed it was hypothesized that removal efficien- cies as well as TPH effluent levels will be fairly con- sistent at operating BSRTs greater than 60-70 days.

B E N C H - S C A L E S T U D I E S O F R E A C T O R - B A S E D T R E A T M E N T 357

CONCLUSIONS AND RECOMMENDATIONS

Based on the results obtained from the bench-scale laboratory study, a pseudo-continuous system is effective in reducing the substrate concentration in soil over a given period of time. Steady-state condi- tions were attained in the different reactors with different biological solids retention times, thus show- ing that the kinetics of the degradation process fol- low those used in other continuous flow processes.

It is suggested that further studies need to be done with different high and low BSRTs. These values would cover the required ranges of BSRT, contami- nant removal efficiency, and effluent quality, thereby providing a broader data base for the determination of biokinetic coefficients. These values would also provide necessary data on biomass values at a wider range of BSRTs, thus covering stages of the micro- organism's growth process.

Evaluation of other environmental and nutritional factors is also recommended in future studies. Opti- mization of biodegradation based on nutrient requirement, moisture content and the possible Use of media conditioners or soil pretreatment tech- niques are areas where further research is necessary. Finally, one important parameter which needs to be considered is the media type itself. Sand, silt and

clay have vastly different surface properties and these play a role in the rate of biodegradation. An effort to study these varying soil characteristics, especially the rates of sorption and desorption, will demonstrate the effectiveness of continuous-flow treatment in reactor-based systems.

REFERENCES

1. O'Neill, E. J. Working to increase the use of innovative cleanup technologies. Water Environ. Technol. 3:333 (1991).

2. Phelps, T. J., Niedzielski, J. J., Schram, R. M., Herbes, S. E. and White, D. C. Cultures to remove organics from soils. J. Appl. Environ. Microbiol. 56:1702 (1990).

3. Li, K. Y., Zhang, Y. and Xu, Tian. Water-soil model pre- dicts best ratio for cleanup. Environ. Protect. 5: 12, 28 (1994).

4. Monod, J. The growth of bacterial cultures. Annu. Rev. Microbiol. 3 (1949).

5. Martin, J. H. Jr., Siebert, A. J. and Loehr, R. C. Estimating oil and grease content of petroleum-contaminated soil. J. Environ. Eng. 117:291 (1991).

6. Davies, B. E. Loss on ignition as an estimate of soil organic matter. Soil Soc. Am. 38:150 (1974).

7. Song, H. and Bartha, R. Effects of jet fuel spills on microbial community of soil. J. Appl. Environ. Microbiol. 56:646 (1990).

8. Grady, C. P. L. Jr. and Lim, H. C. Biological Wastewater Treatment. Marcel Decker, New York (1980).

Open for discussion until 28 June 1996