REFINER BIODEGRADES SEPARATOR-TYPE SLUDGE TO BDAT STANDARDS

R. Lee Vail
Murphy Oil USA Inc.
El Dorado, Ark.

Bioremediation has been successfully used to treat sludge from a refinery stormwater surge basin at the Murphy Oil USA Inc. refinery in Meraux, La.

The technique can be used for numerous basins recently reclassified as hazardous waste units because they contain listed waste, specifically "primary and secondary oil/water/solids separation sludge," or F037 and F038.1 Approximately 9,000 cu yd of high oil and grease sludge accumulated in the stormwater basin at the Meraux refinery (Fig. 1). ENSR Consulting & Engineering performed a feasibility study on the sludge to determine if it degraded at a rate sufficient to justify a full-scale operation. Data from the feasibility study was used to scale-up to a 730,000 gal (in situ) reactor.

The residual from the sludge degradation process was:

• Noncharacteristically toxic-based according to the toxicity characteristic (TC) rule

• Below land-ban standards for API separator bottoms (a listed refinery waste of similar composition).

BACKGROUND

Most refineries, especially older ones, have commingled storm water and process water systems. Contaminated process waters containing organics, metals, and solid particles are transported through an open or closed sewer system to an oil/water/solids separator.

The Resource Conservation and Recovery Act (RCRA) gives the U.S. Environmental Protection Agency (EPA) the authority to list waste as hazardous if EPA has reason to believe that a waste stream is typically hazardous. K051 is the EPA source designation for API separator sludge from the petroleum industry.

EPA expects API separator bottoms to contain unacceptable levels of lead and chromium. These anticipated levels were the basis for listing the waste as hazardous.

Because the storm water and process water sewers are commingled, during periods of heavy rainfall the water volume often exceeds treatment capacity. Surge capacity is necessary and is typically supplied by an impoundment or basin.

The commingled water often contains oily process water, and rain water contains suspended solids. When the two are mixed, oil-contaminated solids are produced. These solids are transported to the basin, where they settle and accumulate on the bottom. Free oil accumulates on the top surface of the basin water, and the water is pumped out for treatment and eventual release.

EPA recently promulgated a rule listing "primary and secondary oil/water/solids separation sludge from the petroleum refining industry" as a hazardous waste.1 It was argued that any petroleum refinery sludge resulting from primary or secondary oil/water/solids separation should be considered hazardous-regardless of the separation method.

This definition can include solids that settle in a storm runoff pond, depending on the configuration of the pond in the waste water treatment process.

The 9,000 cu yd of oily sludge in the bottom of the stormwater surge basin at the Meraux refinery contained over 2,000 mg/kg of polynuclear aromatic hydrocarbons (PAHS) and approximately 200 mg/kg of total benzene, toluene, ethylbenzene, and xylene (BTEX). This is similar in composition to API separator bottoms, a listed hazardous waste.

At the time of this research, these oily solids were not listed as hazardous waste because the basin is not an API separator. However, it was anticipated that such waste would eventually be regulated, and the sludge accumulation was recognized by management as a significant problem that needed to be addressed.

Because the stormwater/process water basin sludge is similar to API separator bottoms, any solution to the disposal of the Meraux waste is likely to be applicable to K051 waste as well.

EPA issued "landban" pretreatment standards for listed hazardous wastes.3 These restrictive standards were promulgated on Aug. 8, 1988, and became effective Aug. 8, 1990.

The regulation requires that listed waste be pretreated prior to land disposal using a technology that equals or exceeds treatment standards established for a best demonstrated available technology, or BDAT.

All listed wastes have, or will have, specific limits for individual organic components or metals in their extracts. To meet these requirements for API separator bottoms, in excess of 90% of several base-neutral PAHs such as anthracene and pyrene must be removed, and leachable metals must be stabilized.

At this time, EPA has not established landban criteria for primary separation sludge; however, it is logical that the eventual requirements will be similar to those for API separator bottoms.

EPA has requested information on the treatment of organics in F037 and F038 by biological treatment methods, with the intention of developing BDAT standards for these newly listed wastes.4

FEASIBILITY STUDY

The ENSR feasibility study was divided into two activities: primary degradation screening and batch-reactor testing.

For the primary degradation screen, five reactor mixtures (with varying nutrient concentrations) and a control (with no added nutrients) were incubated on a Labline orbit shaker for 25 days. Degradation was monitored by measuring several parameters including hydrocarbon oil and grease (HO&G) and chemical oxygen demand, and by performing a Microtox bioassay.

All mixtures showed significant signs of degradation. The mixture with the 5:1 nitrogen/phosphorous nutrient addition was selected as the best candidate for confirmation testing, based on organics removal and toxicity reduction.

A batch reactor was used to confirm the optimum nutrient mixture and to determine degradation parameters, which were used in the design of a full-scale reactor.

Three 15-1. reaction mixtures containing sludge, water, and nutrients were prepared and added to three reactors. A small bubble diffuser at the deepest point of each reactor provided aeration and mixing.

Reactor A was used to determine whether the addition of phosphorus after the "lag period" (meaning lag in bacterial growth) further stimulated degradation. Reactor B was used to confirm the optimum nutrient treatment program determined by the primary degradation screen, i.e., 45 ppm N:9 ppm P, or 5:1. Reactor C was used to evaluate the effects of an emulsifier. (Emulsifiers commonly increase degradation by increasing oil-to-water contacting.)

All three reactors showed similar reduction in HO&G-about 98% (Fig. 2).

The Microtox bioassay measures the effective concentration of a sample that causes a 50% decrease in bacterial activity (called EC50), as measured by light output.5 As samples become less toxic, %EC50 (as measured by Microtox) increases, up to a maximum of 100%.

Both Reactor A and Reactor B reached a %EC50 of 100% by the end of the 21-day batch reactor biodegradation test. The addition of phosphorous to Reactor A did not produce a significant difference in efficacy. However, the addition of the emulsifier to Reactor C appeared to interfere with the toxicity reduction.

Fig. 2 also shows %EC50 for the three reactors, illustrating that as HO&G is removed, the mixtures become less toxic. The Microtox bioassay also shows that the degradation products are not more toxic than the raw sludge.

Another important test result used in the design of the bioreactor was the maximum dissolved oxygen uptake rate (DOUR) of 18 mg/l./hr, which occurred in Reactor B after a lag period of 4-7 days (Fig. 3).

Because bacteria consume oxygen in the process of converting carbon to carbon dioxide, it is desirable to design the reactor to satisfy the maximum DOUR at all times.

REACTOR DESIGN

A 120-ft x 100 ft x 9 ft reactor was built in the earthen impoundment called the stormwater surge basin. The basin originally held 4.9 million gal of surge capacity.

The reactor was built by constructing two sheet pile walls across the surge basin using 20-ft long sheet piles (Fig. 4). The basin's clay bottom was 9 ft below the water surface. The sheet piles were therefore driven into approximately 11 ft of solid clay. Three feet of sludge overlayed the solid clay bottom.

To assure the integrity of the walls, no more than a 4-ft difference was allowed between the water level inside the reactor and the water level outside.

A stormwater/process water mixture flows into the stormwater surge basin through a weir box located on the east leg of the "horseshoe" (Fig. 5). This particular basin receives only wet-weather flow. Water is pumped out of the surge basin via the west leg of the horseshoe.

The reactor walls cut off the normal flow route from the inlet to the outlet. A 36-in. pipe, or equalization line, was laid across the bottom of the basin in a trench to hydraulically connect the east and west legs of the horseshoe, as shown in Fig. 5.

Six 20 hp aspirator-type floating aerators were installed in the reactor for mixing and aeration at a rate of 0.16 hp/1,000 gal mixed (Fig. 1). This was designed to achieve the maximum DOUR of 18 mg/l./hr.

Mixing was adequate to assure good mass transfer of oxygen throughout the reactor cell. Ten days after start-up the dissolved oxygen was measured at multiple locations and depths. The results were consistent, at 3.2 t0.2 mg/l. dissolved oxygen.

The fixed solids concentration measured 2 ft below the surface was about 30-40% of the total theoretical value. Although the basin was not completely mixed, the level of HO&G associated with the solids was found to be consistent.

Fig. 6 shows the solids HO&G content in the mixed liquor as compared to the HO&G in the sludge. The HO&G in the sludge is calculated on a dry weight basis. The HO&G in the mixed liquor is calculated as the total HO&G less the soluble HO&G, the difference being divided by the total suspended solids (TSS) concentration. This calculation simulates the HO&G level in the mixed liquor.

The fact that the two calculated HO&G levels show the same basic trends indicates that mixing was adequate.

FULL-SCALE TEST

Full-scale testing of the biodegradation process was performed in two phases.

In Phase 1, about 1,200 cu yd of existing sludge was biologically degraded in the reactor area (average depth, 3 ft). An aerobic process was initiated using indigenous (native) bacteria on Feb. 17, 1989. No exogenous (nonnative) bacteria were added to augment the biomass.

The reactor temperature started at 60 F. and peaked at 98 F. before decreasing to 90 F. (Fig. 7).

Operating variables were monitored for two reasons:

• To ensure that nutrient levels were optimal, thus maximizing activity

• To ensure that the process was working.

The concentrations of PAHs and BTEX were determined every 14 days by obtaining and testing a composited mixed-liquor sample from the basin. The composite was produced by sampling the reactor with a 600 ml thief, 2 ft below the surface, in multiple locations until a 1 gal sample was accumulated.

In addition, sludge samples were collected and analyzed for benzene, toluene, ethylbenzene, xylene, and several polynuclear aromatics, for comparison to EPA landban standards.

The second phase of testing began at the completion of the first phase, on Apr. 25, 1989. This phase lasted for 7 weeks, until the middle of June.

Approximately 500 cu yd of fresh sludge was added to the partially degraded sludge from Phase 1. No treated sludge was removed. The ratio of organics to total fixed suspended solids (TFSS) therefore started out lower in Phase 2 than in Phase 1.

The mixed-liquor temperature in Phase 2 ranged from 92 to 113 F. The higher reactor temperatures in Phase 2 were primarily caused by higher ambient temperatures and higher biological activity. No equipment or procedures were used to modify the temperature.

The DOUR followed the plate count trends during both phases of experimentation (Fig. 8). This shows that both measurements are good indicators of biological activity.

In Phase 1, the reactor temperature increased above 75 F. at Week 5, causing an increase in biological activity. In Phase 2, reactor temperatures of about 110 F. occurred from Weeks 2 through 7. DOUR and the plate count decreased after Week 2 because of high temperatures, lack of food, or both.

RESULTS

Most of the HO&G in the mixed liquor was associated with the solids. The ratio of soluble HO&G (run on filtered sample) to total HO&G immediately increased from 0.05 to 0.205 because of the initial mixing of the fresh sludge at the start of Phase 1. As more soluble HO&G biodegraded, the ratio decreased, until a final level of 0.01 was reached.

Eliminating the extreme high and low values, the HO&G concentration on the solids was one to two orders of magnitude greater than the concentration in the aqueous phase at equilibrium.

To standardize the HO&G results in the mixed-liquor samples, it was necessary to report the HO&G concentration in relation to a unit mass of TFSS. The concentration of a particular sample was given as: Ca = HO&G (mg/l.) TFSS (mg/l.). TFSS is equal to total suspended solids minus total volatile suspended solids (or bacteria).

In Phase 1, the HO&G/TFSS ratio decreased by 0.018 per week in terms of mg/l. For an average concentration of 20,000 mg/l. fixed solids, the HO&G in the mixed liquor decreased 50 mg/l./day.

As stated previously, all the solids were not suspended in the mixed liquor. The author suggests that the solid particles were constantly settling and being remixed, thus keeping the mixture rich in organics.

The initial theoretical TFSS (i.e., complete mixing) for Phase 1 was 93,000 mg/l. Based on a removal rate of 0.018 mg/l. HO&G per mg/l. TFSS per week, an HO&G removal rate of 240 mg/l./day is calculated. This compares well with the ENSR batch test results of 230 mg/l./day.

However, 3-4 g oxygen are theoretically required to convert 1 g carbon from a paraffin Oil to CO2.6 This means that no more than 135-180 mg/l./day of HO&G grease could have been consumed.

This fact suggests that the determined HO&G removal rates are not precise, but are of the proper magnitude. Possible reasons for the discrepancy are too numerous and complex to address in this article.

In Phase 1, total BTEX in the sludge decreased from 193.6 mg/kg to 0.465 mg/kg. The total PAHs in the sludge were reduced from 2,710 mg/kg to 1,194 mg/kg. The residue at the end of Phase 2 contained only 0.62 mg/kg total PAHS.

Test results for specific landban organics are presented in Table 1, along with BDAT standards for API separator bottoms (K051).

The toxicity of the sludge in the stormwater basin was reduced through aerobic biodegradation as measured by the reduction in HO&G, BTEX, and PAHS.

HO&G in the sludge decreased from 16.9 dry wt % to 6.5 dry wt % in Phase 1. After the addition of fresh sludge (11.1 dry wt % HO&G) to the Phase 1 residual, HO&G in the sludge decreased to 5.1 dry wt % by the end of Phase 2.

The residual sludge at the end of Phase 1 still contained a significant level of PAHS. At the lower temperatures, the bacteria showed little ability to biodegrade five and six-ring PAHS. Higher temperatures in Phase 2 resulted in higher degradation rates and in operational problems caused by severe foaming.

Under the conditions of Phase 2, four and five-ring aromatics, including carcinogenic PAHS, were readily biodegraded with half-lives of 36 days. The residual sludge contained BTEX and PAH levels below the proposed landban restrictions.

VOLUME REDUCTION

Two separate methods were used to calculate sludge volume reduction. The first method was simply to shut down the aerators, wait for the solids to settle (about 2 hr), and physically measure the depth of the sludge.

The sludge depth was measured from a flat boat at 15 predetermined locations defined by a 5 x 3 grid. A piece of PVC pipe was lowered vertically into the water and pushed down until stiff resistance was felt from the clay layer.

The PVC pipe was clearly stained from the oily sludge. Measuring the length of the stain determined the sludge depth.

This type of survey was performed before Phase 1, at 5 weeks, and at 9 weeks. The volume of sludge was 57% less after 5 weeks and 73% less after 9 weeks. The foam was too deep after the first phase to properly survey the reactor cell.

Volume reduction can also be calculated from the concentrating effect on either the fixed solids or specific metals in the sludge.

Fixed solids were measured during the first 18 weeks of operation (including Phase 1, Phase 2, and two weeks thereafter). The original volume of sludge was approximately 1,200 cu yd (TFSS = 23.4%).

After 9 weeks, 500 cu yd of fresh sludge (TFSS 18.2%) was added. At Week 16, 180 cu yd of dense reacted material (TFSS = 42.5%) was removed. One day later, 200 cu yd of fresh sludge (TFSS = 16.9%) was added to the cell. The final concentration of TFSS in the sludge at 18.4 weeks was 35.3%.

Based on the concentrating effect of the fixed solids, the volume reduction was 48%. This analysis assumes that the fixed solids represent all of the material in the sludge that is completely inert.

Applying the same type of analysis to metal concentrations (specifically nickel and lead) by substituting the metal analysis for fixed solids concentration produces another conclusion.

Whereas the fixed solids concentration increased from 23.4% to 35.3%, the nickel concentration increased from 5.86 ppm to 19.6 ppm and the lead, from 53.6 ppm to 282 ppm. Because Mississippi River water was added to the reactor to make the slurry, the only possible source of these metals was the sludge.

Typically, the Mississippi River contains about 10 ppb each of nickel and lead.7 The final concentrations of soluble nickel and lead in filtered water from the reactor was 2 ppb and 41 ppb, respectively. The calculated volume reduction based on nickel was 68 vol %, and based on lead was 70 vol %.

FOAMING

Foaming was a severe problem throughout the experiment. Foaming in Phase 1 was usually associated with high dissolved oxygen in the water (above 5 mg/l.) and quickly disappeared upon turning off some of the aerators.

The foam in Phase 2 was extremely stable and did not disappear after turning off all the aerators. Even chemical antifoaming agents did not control the foam.

The foam layer was typically 2-3 ft deep and provided efficient insulation for the water surface. During this period, the mixed-liquor temperature reached 113 F.

The foam was more than a nuisance to the project. During Phase 2, four out of six aerator shafts broke because the foam crept up the air intake and plugged the annular space between the rotating shafts and the outer cover.

Dissolved oxygen was very low throughout Phase 2. Because foaming is often the result of an inadequate oxygen supply to the bacteria, blowers were installed to add air to the reactor late in the project.

BASIN CLEANUP

At the conclusion of the experimental phases, a continuous operation was instituted, utilizing a dredge to transfer sludge from the bottom of the rest of the basin into the reactor. With this approach, most of the sludge was removed and degraded without taking the entire stormwater basin out of service.

Approximately 6 in. of sludge remained in the basin at the end of the project. The residual sludge contained less than 1 mg/kg total PAHs and easily passed toxic characteristic leaching procedure (TCLP) limits.

COST

This project demonstrated that through biodegradation, BDAT levels for API separator bottoms could be achieved for refinery stormwater/process water primary sludges. Residence time required for adequately mixed systems with mixed liquor TSS of 20,000-50,000 mg/l. to achieve BDAT standards is approximately 3 weeks (for sludge of similar composition).

Capital cost for this project was approximately $200,000, with operating costs (dredge to move material, nutrients, and labor) of $50,000. Electrical cost to operate the aerators during the 1-year biodegradation of approximately 7,000 tons of sludge was another $50,000. Total cost was less than $50/ton (assuming no residual value for capital, although the aerators had significant value at the end of the project).

If the system operated for 10 years degrading 7,000 tons/year, and the capital was depreciated over that 10-year period, the operating cost would be $20/ton. Given that incineration costs are typically about $1,400/ton, biodegradation of oily sludges can be a cost-effective alternative for petroleum refiners.

REFERENCES

1. Federal Register, Vol. 55, Nov. 2, 1990, pp. 46354-97.

2. Federal Register, Vol. 45, May 19, 1980, p. 33073.

3. Federal Register, June 1, 1990.

4. Federal Register, Vol. 56, May 30, 1991, pp. 24444-52.

5. Bulich, A.A., "A Practical Reliable Method for Monitoring the Toxicity of Aquatic Samples," Proc. Biochem, March/April 1982.

6. Hornick, S.B., Fisher, R.H., and Paolini, P.A., "Petroleum Waste," Land Treatment of Hazardous Waste, ad. by Parr, J., Noyes Data Corp., 1983, p. 323.

7. U.S. Geological Survey, "Water Resources Data Mississippi Water 1982," Water-Data Report MS-82-1, 1982, p. 168.

Copyright 1991 Oil & Gas Journal. All Rights Reserved.