A toxicity reduction evaluation for an oily waste treatment plant exhibiting episodic effluent toxicity

The Science of the Total Environment 218 Ž1998. 141]152

A toxicity reduction evaluation for an oily waste treatment plant exhibiting episodic effluent toxicity Mujde Erten-Unal a,U , Allen B. Gelderloos b , Jane S. Hughes b,1 a

Old Dominion Uni¨ ersity, Department of Ci¨ il and En¨ ironmental Engineering, KDH 135, Norfolk, VA 23529-0241, USA b Malcolm Pirnie Inc., 400 Monroe, Suite 340, Detroit, MI 48226, USA Received 1 December 1997; accepted 30 April 1998

Abstract A Toxicity Reduction Evaluation ŽTRE. was conducted on the oily wastewater treatment plant ŽPlant. at a Naval Fuel Depot. The Plant treats ship and ballast wastes, berm water from fuel storage areas and wastes generated in the fuel reclamation plant utilizing physicalrchemical treatment processes. In the first period of the project ŽPeriod I., the TRE included chemical characterization of the plant wastewaters, monitoring the final effluent for acute toxicity and a thorough evaluation of each treatment process and Plant operating procedures. Toxicity Identification Evaluation ŽTIE. procedures were performed as part of the overall TRE to characterize and identify possible sources of toxicity. Several difficulties were encountered because the effluent was saline, test organisms were marine species and toxicity was sporadic and unpredictable. The treatability approach utilizing process enhancements, improved housekeeping, and operational changes produced substantial reductions in the acute toxicity of the final effluent. In the second period ŽPeriod II., additional acute toxicity testing and chemical characterization were performed through the Plant to assess the long-term effects of major unit process improvements for the removal of toxicity. The TIE procedures were also modified for saline wastewaters to focus on suspected class of toxicants such as surfactants. The TRE was successful in reducing acute toxicity of the final effluent through process improvements and operational modifications. The results indicated that the cause of toxicity was most likely due to combination of pollutants Žmatrix effect. rather than a single pollutant. Q 1998 Elsevier Science B.V. All rights reserved. Keywords: Toxicity Reduction Evaluation ŽTRE.; Marine species; Acute toxicity; Toxicity Identification Evaluations ŽTIEs.; Oily wastewater

U 1

Corresponding author. Tel.: q1 757 6834412; fax: q1 757 6835354. Previously with Malcolm Pirnie.

0048-9697r98r$19.00 Q 1998 Elsevier Science B.V. All rights reserved. PII S0048-9697Ž98.00208-3

142

M. Erten-Unal et al. r The Science of the Total En¨ ironment 218 (1998) 141]152

1. Introduction and background The Amendments to the 1972 Clean Water Act incorporated a combination of chemical specific and whole effluent toxicity limits into National Pollutant Discharge Elimination System ŽNPDES. permits. The US Environmental Protection Agency ŽEPA. recommended that whole effluent toxicity tests be used in State Toxic Management Programs. The presence of effluent toxicity on a less than regular basis will trigger the requirement for a permitted discharger to conduct a Toxicity Reduction Evaluation ŽTRE. in most of the US. The Virginia Department of Environmental Quality ŽDEQ. adopted the Toxics Management Regulation ŽTMR. which governs the control of toxics in treatment plant effluents. The overall goal of the TMR was to prevent the discharge of toxic levels of pollutants into the State’s waters. It also provided water quality standards for certain pollutants. Based upon toxicity test results indicating noncompliance with the criteria in the TMR, a TRE was required at a US Naval Fuel Depot’s oily wastewater treatment plant ŽPlant.. For compliance with the TMR no more than 25% of the acute toxicity tests can exhibit LC 50 values of less than 100%. The Plant effluent demonstrated a failure rate of 38% Ž6 out of 16 tests. and consequently the DEQ required the performance of a TRE at the Plant. A TRE plan was developed and consisted of extensive chemical characterization of the wastewater throughout the treatment process, monitoring for acute toxicity, evaluation of the entire treatment processes and the plant operations. The EPA’s Phase 1 Toxicity Identification Evaluation ŽTIE. procedures were also performed to characterize effluent toxicity. From this evaluation, improvements were made to the treatment system which produced reductions in acute toxicity below the permitted failure rate. This phase of the project is referred to as Period I. In order to verify that the plant improvements reduced the effluent toxicity, additional toxicity testing was performed. Along with toxicity testing, some of the compounds which were identified in

Period I as suspected toxicants were evaluated through additional chemical analyses and laboratory studies. In addition, more detailed monitoring of plant operations was conducted during this period Žreferred to as Period II. to evaluate the effects of different source wastewaters on effluent toxicity. The results of this period confirmed Period I results; however, the specific causeŽs. of toxicity could not be identified. This paper describes the approach taken in conducting the TRE, and presents the results of the toxicity tests along with the effects of plant improvements on the toxicity reduction during this project. The results of special tests performed in an attempt to focus on suspected toxicants are also discussed. 1.1. Toxicity reduction e¨ aluation approaches As described by the EPA ŽUS Environmental Protection Agency, 1989., a TRE is an investigation designed to isolate the sourceŽs. of effluent toxicity, including specific causative pollutants if possible, and to determine the effectiveness of various control options in reducing toxicity. The approach used for the TRE in this study was similar to the general approach developed by the EPA for performing a TRE at an industrial facility. A step-wise or tiered process is used which combines toxicity testing, chemical testing, and treatment evaluations to identify either the actual cause of toxicity or treatment methods that will reduce toxicity. As shown in Fig. 1, the first tier includes the collection of background information about the Plant and its past operating history. The second tier involves the evaluation of remedial actions to optimize the operation of the facility so as to reduce effluent toxicity. If these improvements are successful in reducing effluent toxicity to acceptable levels, then the TRE can be considered completed, provided that follow-up monitoring confirms these results. If the remedial actions fail to solve the toxicity problem, the TRE proceeds to the third tier, which includes the TIE. The overall objective of a TIE is to characterize andror identify the causeŽs. of effluent toxicity. The approach presented in Fig. 1 is a general one which is to be used as guidance for per-

M. Erten-Unal et al. r The Science of the Total En¨ ironment 218 (1998) 141]152

143

Fig. 1. General approach of a TRE.

forming a TRE. The approach used in this study combined the effort of tiers two and three by performing TIEs during the same period as the treatment plant evaluations and improvements. Through simultaneous performance of tiers two and three, our approach was designed to determine both the effect of treatment improvements on effluent toxicity and the causeŽs. of toxicity. One reason for conducting the TRE in this manner was that more stringent toxicity limits were anticipated and measures to control toxicity in the future could be more easily implemented, if the causeŽs. of toxicity could be identified. 1.2. Plant processes and operations The Plant was specifically designed to remove oil and grease to less than 10 mgrl at an average design flow rate of 1 million gallonsrday. Prior to treatment through the Plant, various oily waste waters are collected in one of two 1.5-million-gallon equalization tanks which contain oil skimming mechanisms to remove the floating oil from the

water surface and transfer the oil to an on-site fuel reclamation plant. The physicalrchemical treatment system consists of oil separators to protect the Plant from accidental discharges of oil from the equalization tanks, aeration tanks, rapid mixrflocculation tanks, dissolved air flotation units and pressure sand filtration for suspended solids removal, and granular activated carbon column ŽGAC. for dissolved organics removal. A process flow schematic of the existing Plant is shown in Fig. 2. Treatment of wastewater from the equalization tanks is a continuous flow but intermittent operation. Currently, the plant operates for an average of 2]3 days per week at a flow rate of between 200 and 300 gallonsrmin Žgpm.. 1.3. Influent wastewater sources and characteristics The Plant treats a variety of different strength wastewaters including ship ballast and bilge water, storm water Žalso called berm water., and washwater from fuel tank cleaning. Bilge water

144

M. Erten-Unal et al. r The Science of the Total En¨ ironment 218 (1998) 141]152

Fig. 2. Process flow schematic of the oily waste treatment plant.

primarily contains oily wastes from the oils and greases used in a ship’s engines; however, it can also contain a wide variety of wastes, including degreasers, cleaning solvents, or any other wastes generated on a ship. Ballast water is sea water with some small quantities of fuel and is used by ships to improve stability and control draft. Berm water is the rain water collected from the diked fuel storage areas and contains low concentrations of oil and grease or BOD5 . Washwater from fuel tank cleaning originates from the cleaning of ships’ fuel tanks with high pressure water and

occasionally degradable detergents. The ballast water and washwater along with some of the bilge water cause the wastewater to be slightly saline Ž2]9 ppt.. The Plant influent wastewater strength is highly variable and depends on the amount of each source of wastewater within the equalization tanks. During this study, the influent TOC concentrations ranged from 29 mgrl to 900 mgrl, and the BOD concentrations ranged from 29 to 770 mgrl. On average, the influent wastewater had a TOC of 180 mgrl and a BOD5 of 309

M. Erten-Unal et al. r The Science of the Total En¨ ironment 218 (1998) 141]152

mgrl. Suspended solids were present in low concentrations Ž- 50 mgrl., implying that the organic material was primarily in the dissolved or soluble form. Results of other chemical analyses did not reveal any compounds in unusually high concentrations or concentrations that would likely cause toxicity. A summary of influent wastewater characteristics is shown in Table 1. 2. Materials and methods 2.1. Aquatic toxicity monitoring All toxicity tests during this study were conducted according to the EPA’s toxicity testing procedures ŽUS Environmental Protection Agency, 1985.. The organisms used during this study were Mysidopsis bahia Žmysid shrimp. and Cyprinodon ¨ ariegatus Žsheepshead minnows.. Results were reported in terms of LC 50 Žmedian lethal concentration } the percentage of effluent that is lethal to 50% of the exposed organisms at a specific time of observation.. Due to the intermittent nature of the treatment operation, some modifications were made to the original TRE plan which called for renewal bioassays to be conducted. All testing was conducted under static non-renewal conditions. In order to conduct toxicity tests with marine species, the salinity of the samples Žwhich ranged from 2 to 9 ppt. was adjusted to 20 ppt with synthetic sea salts. Also, the toxicity test results from Period I indicated that the sheepshead minnow was relatively insensitive to toxicity; therefore toxicity tests using only mysid shrimp were performed in Period II. Extensive chemical characterization of the wastewater throughout the treatment process was

145

also performed during the entire project. All chemical analyses were performed in accordance with the EPA methods. Each sample was analyzed for priority pollutant inorganics and organics, and for selected non-priority pollutants. 2.2. Toxicity identification e¨ aluation procedures The US EPA has published manuals describing procedures for conducting Phase I, II and III TIEs using freshwater organisms ŽMount and Anderson-Carnahan, 1988, 1989; Mount, 1989; Norberg-King et al., 1991.. However, during this study EPA has not yet published methods for conducting TIEs using marine or estuarine organisms. The advice given by EPA for those effluents discharged to marine waters is to use freshwater species in Phases I and II of the TIE and then in Phase III attempt to demonstrate that the freshwater species are responding in the same manner to the same toxicants as the marine species. These procedures can be used if the effluent is freshwater, as most effluents are. However, the Plant effluent is sufficiently saline that it cannot be tolerated by freshwater organisms. Thus, the Phase I procedures had to be performed using marine species. During this study, the TIE procedures developed by EPA were followed as closely as possible; however, because of wastewater salinity, some modifications to the fractionation steps were necessary. The modifications to the effluent sample during the TIE fractionation steps were performed after the adjustment of the salinity and addition of test organisms. One of the modifications included the use of a cation exchange resin instead of EDTA chelation

Table 1 Source wastewater and Plant influent characteristics Parameter

TOC Žmgrl. BOD5 Žmgrl. Oil and grease Žmgrl. TSS Žmgrl. Note. NA, not analyzed.

Bilge water

1000 2050 390 NA

Berm water

12 - 13 1.2 NA

Plant influent Average

Minimum

Maximum

180 309 18 45

28 29 1.0 15

900 770 71 77

146

M. Erten-Unal et al. r The Science of the Total En¨ ironment 218 (1998) 141]152

test since chelation is not suitable for saline samples. The graduated pH test was also modified by performing the test at pH 7, 8 and 9. Performance of toxicity tests at pH 6 was omitted because this pH is too stressful for the marine organisms. Since ammonia toxicity significantly varies over this range of pH values, the relative differences in toxicity could still be examined. In Period II, the TIE procedures were further modified to focus on surfactants as a possible toxicant. A procedure was added to the standard pH adjustmentraeration test which involved rinsing the aeration glassware to assure that all the surface foams were collected. The toxicity of the rinse water was determined and was compared to the toxicity of the pipetted solution. TIEs are designed to be conducted on effluents that demonstrate consistent toxicity. Typically, the Phase I procedures would be scheduled for a certain date, the sample would be obtained and the initial toxicity tests would be set up immediately. However, because of the sporadic toxicity of the Plant effluent, there was no foreknowledge of which samples would exhibit toxicity and should be evaluated through the TIE. Thus, it was not possible to begin the TIE manipulations on the day of sample arrival. The results of the routine monitoring tests served as the initial toxicity tests. Furthermore, when toxicity was observed in the routine tests, it was not seen until the end of the tests Ž48 h.. Thus, the baseline tests for the TIE could not be initiated until much later after sample collection than recommended in the EPA protocol. During the Phase I characterization procedures, the LC 50 values for sodium thiosulfate for the oxidant reduction test had to be determined since these values were not available in the literature for the test species. The LC 50 values determined in this study were 0.5 mgrl for M. bahia and 42 mgrl for C. ¨ ariegatus. In the Phase I test, varying concentrations of sodium thiosulfate were added to the effluent at a concentration four times the 24-h LC 50 Žfour times the 24-h LC 50 was greater than 100% for all Plant effluent samples.. This fractionation step and graduated pH test were omitted in the second Period of the

project as there was no indications that toxicity was caused due to oxidants or ammonia. Finally, in the second Period of the project, TIEs were performed only on Plant effluent samples demonstrating an initial toxicity with an LC 50 F 75%. This decision was made because the results of Period I indicated that when TIEs were performed on marginally toxic samples, the toxicity ‘disappeared’ Ži.e. LC 50 values ) 100%. in the TIE baseline test. The possible cause of the disappearance is discussed further in the following sections. 3. Results 3.1. Aquatic toxicity and chemical monitoring of the Plant effluent During Period I of the TRE, significant process and operational improvements were implemented at the Plant. The major improvements are discussed in detail elsewhere ŽGelderloos et al., 1991. and included the installation and operation of a new pH adjustment equipment, relocation of the alum feed point, DAF improvements and replacement of the media in one of the granular activated carbon ŽGAC. columns. These improvements produced reductions in toxicity. The acute toxicity failure rate, obtained from toxicity test results when there were no modifications to the Plant operations, declined from 35% Ž7 out of 20 tests. to 9% Ž3 out of 32 tests. after the Plant operational improvements were implemented in Period I. Continued toxicity monitoring in Period II confirmed the reduction of toxicity. The failure rate during this Period was 17% Ž3 out of 18 tests.. The overall failure rate after modifications to the treatment process during both periods was 12% Ž6 out of 50 tests.. The results, illustrated in Table 2, demonstrate that improvements to the treatment system have reduced toxicity such that the final effluent was in compliance with the target failure rate in the Toxics Management Regulation. Another indication that toxicity had been reduced was that, for those samples which demonstrated toxicity, the level of toxicity Žrepre-

M. Erten-Unal et al. r The Science of the Total En¨ ironment 218 (1998) 141]152

147

Table 2 Summary of Plant effluent toxicity test results

Toxicity Failure rate Range of LC50 s

DEQ study 1987]1988 38% Ž6 out of 16. 3.4%]) 100%

Period I Pre-modifications

Post-modifications

35% Ž7 out of 20. 19%]) 100%

9% Ž3 out of 32. 55%]) 100%

sented by LC 50 values. was also reduced. For example, prior to the modifications, the most toxic samples had LC 50 values of 3.4 and 19%. After the modifications, however, the most toxic sample had an LC 50 value of 55%. Chemical analyses conducted on the Plant effluent for selected non-priority pollutants indicated that the wastewater contained a high soluble organic fraction represented by the nonspecific parameters TOC, COD and BOD5 . The chemical analyses results showed that acetic acid was the primary organic fraction in the effluent which was most likely produced as a by-product of the natural degradation of the oily wastewaters in the equalization tanks. The results of priority pollutant inorganics analysis showed that most metals were detected in the Plant effluent but at relatively low concentrations in relation to water quality standards for most metals. Analysis of priority pollutant organics detected the typical compounds for this type of wastewater, specifically benzene, toluene, phenols and naphthalene. A regression analysis was performed to examine possible correlations between the levels of toxicity and the detected concentrations of the conventional pollutants, metals and priority pollutant organics. However, none of the compounds showed significant correlations with acute toxicity. 3.2. Toxicity reduction results through the Plant During Period II, toxicity tests were performed at the major unit treatment processes to examine the effects of each major process on toxicity reduction. The sampling locations were designed to evaluate the performance of the DAF units, sand filters and carbon columns. Examination of the toxicity reduction through

Period II after modifications 17% Ž3 out of 18. 55%]) 100%

Overall

12% Ž6 out of 50. 55%]) 100%

Table 3 Toxicity reduction through unit processes Sample number

LC50 Ž% effluent. Storage tank effluent

DAF effluent

Sand fitter effluent

GAC effluent

II-2 II-4 II-6 II-8 II-9 II-10 II-11

39 9 15 15 44 61 68

59 18 54 25 33 71 ) 100

) 100 17 29 29 ) 100 ) 100 ) 100

) 100 ) 100 ) 100 55 ) 100 ) 100 ) 100

individual unit treatment processes are presented in Table 3. Due to the low number of corresponding toxicity test results and the variability of the wastewater, only the general trends observed in each unit process are discussed below. In general, the DAF units, sand filters and carbon columns reduced toxicity. The removal through the DAF units was slight, but consistent. Of the seven results with corresponding DAF influent and effluent samples, six showed a reduction in toxicity through the DAF units. The reduction in toxicity through this unit process may be a result of the removal of the chemically conditioned material from the DAF units which would tend to implicate that metals were contributing to toxicity. The aeration associated with the DAF process may also be responsible for the removal of toxicity which would implicate volatile compounds as a cause of toxicity. Regardless of the cause of toxicity, the DAF units, in all but one case, were unable to remove toxicity completely. The removal of toxicity through the sand filters exhibited more inconsistent results than the DAF units. Because one of the DAF effluent samples

148

M. Erten-Unal et al. r The Science of the Total En¨ ironment 218 (1998) 141]152

had an LC 50 value of greater than 100%, there were only six corresponding results with sand filter influent and effluent samples. Of these six samples, three showed no significant change, while the other three samples showed a decrease in toxicity. The wide range of toxicity removal through the sand filters ŽLC 50 s from 17% to greater than 100%. may be a result of the variability of the wastewater toxicity. When the sand filter produced non-toxic effluent, the toxicity could be in a precipitate form which would be removed by filtration. When the sand filter did not remove toxicity, the toxicity could be in a soluble form which filtration will not remove. For examination of toxicity removal through the carbon columns, only a limited number of data points were available. Because four of the sand filter effluent samples had LC 50 values of greater than 100%, there were only three corresponding results with carbon column influent and effluent samples. Of these three samples, each showed a significant removal of toxicity with two of the samples producing non-toxic results Žfrom LC 50 values of 17% and 29% to LC 50 values greater than 100%.. Although toxicity was removed in these three sampling events, the overall results for Period II showed that the carbon columns could not be relied upon to consistently produce non-toxic effluent. Because three of the 16 final effluent samples contained toxicity ŽLC 50 s - 100., the carbon column was unable to completely remove all of the toxicity. In conclusion, each major unit treatment process reduced the toxicity levels. Because of the limited number of samples, however, it was not possible to accurately characterize the class or classes of compounds responsible for causing toxicity. 3.3. TIE results During Period I, three effluent TIEs were conducted on samples which demonstrated initial toxicity to mysids with LC 50 values of 55%, 88% and 83%. However, at the end of the baseline tests, the effluent exhibited no toxicity ŽLC 50 values greater than 100%.. The disappearance of toxicity

during the time between the initial toxicity tests and the baseline toxicity tests during this period made it nearly impossible to obtain any definitive results from the TIE procedures. Since the toxicity reduction from the fractionation steps was supposed to be compared to the toxicity observed in the baseline test, a baseline test with little or no observed toxicity rendered the TIE results largely useless. In the second period, TIEs were conducted on two Plant effluent samples. The first TIE was conducted on a sample with an initial toxicity of 55%. The baseline test, which started 5 days after the sampling for the initial test, showed an LC 50 Table 4 Effluent TIE results in Period II Sampling event II-8 LC50 Ž% effluent. Initial toxicity Baseline toxicity pH adjustment test pH 3 pH 11 pHraeration test pH 3 pH ŽI. pH 11 pHrfiltration test pH 3 pH ŽI. pH 11 pHrC18 test pH 3 A B pH ŽI. A B pH 11 A B pHrion exchange test pH 3 A B pH ŽI. A B pH 11 A B pHrsurfactant test pH 3 pH ŽI. pH 11 a

Sampling event II-15 LC50 Ž% effluent.

55 80

71 76

71 71

62 71

46 91 80

) 100 ) 100 55

80 71 31a

) 100 71 91

) 100 ) 100 ) 100 91 ) 100 ) 100

) 100 71 85 46 71 80

) 100 ) 100 ) 100 ) 100 ) 100 ) 100

80 ) 100 62 71 ) 100 ) 100

80 71 ) 100

) 100 100 35

Exhibited an atypical dose]response relationship.

M. Erten-Unal et al. r The Science of the Total En¨ ironment 218 (1998) 141]152

of 80%. The second effluent TIE was conducted on a sample with an initial toxicity of 71%. The baseline test began 4 days after the sampling for the initial test and produced an LC 50 of 76%. This delay, while not desirable, was unavoidable due to the Plant operation schedule. A summary of the effluent TIE results obtained during Period II are shown in Table 4. In the first TIE, most of the fractionation test results had similar LC 50 values as in the baseline test except for the solid phase extraction ŽC-18 column. and pHrion exchange tests where both tests removed toxicity. Toxicity reduction through these tests would implicate non-polar organics and cationic metals as a cause of toxicity. In the second TIE, the solid phase extraction and ion exchange tests did not completely remove toxicity at all pH levels which is not supportive of the first TIE results. In the pH adjustment test and pH adjustmentraerationrsurfactant test, there was no relative difference in toxicity from the baseline test in the first TIE. The removal of toxicity by pH adjustmentraeration test in the second TIE may have indicated the presence of volatile organics or hydrogen sulfide. Both pH adjustment followed by aeration and pH adjustmentrsurfactant tests removed toxicity at pH 3 and at pH ŽI. in the second TIE. If toxicity had not been removed by the pHrsurfactant test but removed in pHraeration test, then surfactants would have been the probable cause of toxicity. However, the results of TIEs conducted on both Plant effluents did not provide any evidence of surfactants as potential toxicants. The pH adjustmentrfiltration tests exhibited similar toxicity as in the baseline test implying that toxicity was not associated with filterable materials. Since mortality was so slight in both baseline tests and the number of organisms used in the TIE was so small Žfive organisms per test., no conclusions could be drawn as to whether aeration, solid phase extraction or cation exchange resin had a definite impact on toxicity removal. Therefore, the TIE procedures performed during this project were not successful in identifying the specific causative pollutantŽs. or the sourceŽs. of toxicity.

149

4. Discussion 4.1. Factors affecting toxicity During the entire project, suspected toxicants were investigated as possible causeŽs. of effluent toxicity. For some compounds, special studies were performed and for others the chemical results obtained throughout the study were correlated to the toxicity test results. The results derived from these studies are discussed in the following sections. 4.2. Influent wastewater composition During Period I, it was believed that the bilge water was much more toxic than berm water, since bilge water has a higher organic strength as indicated by TOC ŽTOC of approx. 1000 mgrl.. Also, berm water was thought to be diluting the bilge water, hence reducing its toxicity. Therefore detailed monitoring of the various source waste water quantities in the plant influent Žequalization tanks. was included in Period II in order to determine the most toxic wastewater source and to investigate whether the Plant was capable of removing toxicity associated with this source. During Period II, the two major sources of wastewater, berm and bilge water, were mixed in the equalization tanks prior to treatment. Due to operational constraints, some of the sand filter and carbon column backwash water was pumped into the equalization tanks. For a short period of time, however, the two wastewaters were segregated Žsix sampling events. into two tanks, with one tank holding bilge water, and the other tank holding berm water. The segregation of the wastewaters allowed for the evaluation of toxicity removal through the Plant for the highest strength, and presumed most toxic wastewater. Examination of the influent toxicity values showed that berm water was as toxic as bilge water and that there was no correlation between the levels of toxicity and the relative portions of bilge water or berm water within the equalization tanks. The toxicity of the bilge water was comparable to that of the berm water. The fact that the bilge water was not as toxic as expected can most

150

M. Erten-Unal et al. r The Science of the Total En¨ ironment 218 (1998) 141]152

likely be explained by the variability of the bilge water. For example, the 100% bilge water samples were not as toxic as the ‘mixed’ Ži.e. berm and bilge water. sample, indicating that the level of toxicity of the bilge water could be highly variable. On the other hand, since berm water is just rainwater collected from the diked fuel storage areas, it is not believed to be as toxic as these results indicate. Even though the equalization tanks were flushed twice prior to sampling during the segregated period, the berm water toxicity could have been influenced in some manner. One possibility was that the backwash water from the plant Žwhen treating bilge water. which was pumped back into the equalization tank holding berm water contained some toxicity. Another possibility was that the contaminants within the sludge layer on the bottom of the tanks may have leached into the water and thus contaminated the berm water and caused increased levels of toxicity.

freshwater organisms. The second test was conducted on an actual Plant effluent sample with spiked amounts of sulfide. The results showed that for H 2 S concentrations up to 1.38 mgrl of H 2 S, there was not enough toxicity to determine an LC 50 . It is believed that the organisms were probably exposed to much lower H 2 S concentrations than the predicted concentrations, most likely, due to precipitation of the spiked sulfide in the laboratory studies. The maximum H 2 S concentration was 1.8 mgrl in the Plant effluent during Period II, as shown in Table 3, and that sample exhibited no toxicity ŽLC 50 greater than 100%.. These results suggest that H 2 S is not as toxic in actual wastewater as compared to laboratory prepared water. Also, based on the actual Plant effluent results, no correlation was found between effluent H 2 S concentrations and the level of effluent toxicity. 4.4. Surfactants

4.3. Hydrogen sulfide Because hydrogen sulfide was identified as a suspected toxicant in the Plant wastewater, the effects of hydrogen sulfide ŽH 2 S. on the effluent toxicity were evaluated through a literature review. The literature review revealed several studies on H 2 S toxicity to freshwater organisms and showed that H 2 S was generally toxic at levels below 0.5 mgrl. Studies showed the 96-h LC 50 concentrations of H 2 S ranging from 57.2 m grl at pH 7.1 to 14.9 m grl at pH 8.7 for fathead minnows ŽBroderius et al., 1977. and found the 96-h LC 50 concentrations of 47.8 m grl for juvenile bluegill ŽBroderius and Smith, 1976.. No information, however, was available on the toxicity of H 2 S to mysid shrimp. Because of the lack of toxicity data in the literature, toxicity tests were performed in the laboratory to determine the LC 50 values of H 2 S on mysid shrimp. Hydrogen sulfide concentrations were calculated from dissolved sulfide and pH measurements. The first test was conducted using laboratory dilution water and exhibited a hydrogen sulfide LC 50 concentration of 0.14 mgrl. This value is within the range found in literature for

Surfactants were thought to be one possible class of toxicants in the Plant effluent because of a number of reasons, including the occasional foaming observed in the aeration tanks and DAF units, the use of certain cleaning agents on shipboard which contain relatively high concentration of surfactants, and the disappearance of toxicity between the initial toxicity tests and TIE baseline toxicity tests. The disappearance of toxicity is commonly observed in samples which contain degradable surfactants. Analysis of both non-ionic Žas CTAS. and anionic Žas MBAS. surfactants were incorporated into the sampling plan. However, no surfactants as CTAS were detected and surfactants as MBAS were present in very low concentrations ŽTable 3.. Linear regression analysis of the results found that there was no correlation between effluent surfactant concentrations Žeither anionic or non-ionic. and the level of toxicity. The literature showed the toxicity of individual surfactants ranging from 0.71 mgrl to 116 mgrl for mysid shrimp ŽHall et al., 1989.. In general, the acute toxicities of many surfactants to different species were within the range of 1]100 mgrl,

M. Erten-Unal et al. r The Science of the Total En¨ ironment 218 (1998) 141]152

with the majority being less than 20 mgrl ŽPatoczka et al., 1990.. However, it was not possible to relate any of the reported values to the toxicity observed in the Plant effluent because the surfactants in the Plant effluent were measured as general classes ŽCTAS and MBAS. rather than individual compounds. 4.5. Priority pollutants At the beginning of the TRE study, it was thought that acute toxicity was associated with the presence of heavy metals and volatile organic compounds. The analysis of priority pollutant inorganics was included in all sampling events during this study. The data collected during Period I and Period II did not indicate that these compounds or any other measured inorganics was the cause of toxicity because statistical analyses of metal concentrations vs. toxicity showed insignificant correlations. Therefore the priority pollutant inorganics measured in the Plant effluent do not appear to be responsible for the effluent toxicity. Priority pollutant organic compounds were also analyzed in Period I for both of the Plant influent and effluent samples and only for the Plant effluent samples in Period II. Linear regression analysis with individual compounds did not exhibit any significant correlations between these compounds and toxicity. An additional analysis was performed to determine if a certain combination of compounds could be used to predict the level of effluent toxicity. The analysis included summing the concentrations of detected priority pollutants by class Ži.e. base neutrals, acid extractables and volatiles. and performing a regression analysis using these summed concentrations and toxic units Ž100rLC 50 . of the failing effluents. This analysis also did not produce any correlations but given the large number of combinations of priority pollutant organics and inorganics, some other combination of pollutants may correlate with toxicity. 5. Conclusions The TRE conducted on the oily wastewater

151

treatment plant was successful in reducing the acute toxicity of the final effluent through process improvements and operational modifications. Continued monitoring following these improvements and modifications confirmed the reduction of toxicity. Attempts to identify the causeŽs. of toxicity by regression analysis, TIE fractionation tests and special studies conducted on suspected toxicants were generally unsuccessful. Based on these results, the cause of toxicity is most likely due to combination of pollutants Žmatrix effects . rather than a single pollutant. Given the unlimited number of possible combinations and the relatively limited chemical data, the possibility of determining the cause of toxicity appears to be highly remote. Also, considering the TIE results obtained during this study, the possibility of characterizing the toxicity through additional TIE tests appears to be unlikely. Although the causeŽs. of toxicity were not identified, the toxicity failure rate stayed below the existing State’s criteria due to treatment improvements and modifications. However, the Navy anticipated more stringent toxicity requirements. To meet more stringent requirements there are generally two options which could be pursued. The first option is to continue chemical and toxicity testing in an effort to identify the causeŽs. of toxicity and then implement the procedures to control the compounds in the wastewater. As previously noted, based on the results of this study, the identification of the causeŽs. of toxicity appears unlikely. The second option is to add one or more treatment processes to control toxicity. The DEQ informed the Navy that the existing Plant must be upgraded to provide treatment for biochemical oxygen demand ŽBOD.. The target BOD limit Žbased on a standard 5-day BOD. test proposed by the State on the Plant discharge was 26 mgrl of BOD5 . Negotiations with the DEQ resulted in permit compliance schedule which stipulated that a biological treatment system, for BOD removal would be constructed at the Plant at the end of 1996 ŽLeach, 1994, unpublished paper on Biological and Ad¨ anced Oxidation Treatment of High Strength Bilge Water and Oily Waste.. In addition, the schedule included a spe-

152

M. Erten-Unal et al. r The Science of the Total En¨ ironment 218 (1998) 141]152

cial condition which required that a treatability study be conducted to investigate various biological treatment technologies and determine their technical feasibility for supplemental organics removal at the Plant. This option would not require identification and characterization of toxicity. Pilot testing of various additional unit processes would help to identify the appropriate treatment technology that renders an acceptable effluent quality for BOD. The State also stipulated that the discharge from the plant must be non-toxic to Mysidopsis bahia. At least 50% or more of the mysid shrimp should survive 48 h in a sample of 100% Plant effluent. References Broderius SJ, Smith LL Jr. Effect of hydrogen sulfide on fish and invertebrates: part II } hydrogen sulfide determination and relationship between pH and sulfide toxicity. EPA600r3-76-0626. Duluth, MN: US Environmental Protection Agency, 1976. Broderius SJ, Smith LL Jr, Lind DT. Relative toxicity of free cyanide and dissolved sulfide forms to the fathead minnow Ž Pimephales promelas.. J Fish Res Board Can 1977;34: 2323]2332. Gelderloos AB, Hughes JS, Leach BG. A toxicity reduction evaluation and wastewater treatability study on an oily wastewater treatment plant. Proceedings of the 64th An-

nual Water Pollution Control Federation Conference, Toronto, Canada, 1991. Hall WS, et al. Acute toxicity of industrial surfactants to Mysidopsis bahia. Arch Environ Contam Toxicol 1989;18: 765]772. Mount DI, Anderson-Carnahan L. Methods for aquatic toxicity identification evaluations: Phase I toxicity characterization procedures. EPA-600r3-88r034. Duluth, MN: US Environmental Protection Agency, 1988. Mount DI. Methods for aquatic toxicity identification evaluations: Phase III toxicity confirmation procedures. EPA600r3-88r036. Duluth, MN: US Environmental Protection Agency, 1989. Mount DI, Anderson-Carnahan L. Methods for aquatic toxicity identification evaluations: Phase II toxicity identification procedures. EPA-600r3-88r035. Duluth, MN: US Environmental Protection Agency, 1989. Norberg-King TJ, Mount DI, et al. Methods for aquatic toxicity identification evaluations: Phase I toxicity characterization procedures. EPAr600r6-91r003. 2nd ed. Washington DC: US Environmental Protection Agency, 1991. Patoczka J, et al. Biodegradation and secondary effluent toxicity of ethoxylated surfactants. Water Res 1990;24:965]972. US Environmental Protection Agency. Generalized methodology for conducting industrial Toxicity Reduction Evaluations ŽTREs.. EPA-600r2-88-070. Cincinnati, OH: Risk Reduction Engineering Laboratory, 1989. US Environmental Protection Agency. Methods for measuring the acute toxicity of effluents to freshwater and marine organisms. EPAr600r4-85r013. 3rd ed. Cincinnati, OH: Environmental Monitoring and Support Laboratory, 1985.