Models for BTEX evaluation in cases of oil spill on the sea, using Experimental Desing

The objective of this study was to evaluate and obtain mathematical models, using the experimental design methodology, capable of predicting the contamination of benzene, toluene, ethylbenzene and xylenes (BTEX) in cases of oil spillage at sea, and to evaluate the acute toxicity of the soluble petroleum fraction (FSA). The factorial experimental planning was developed to describe the concentration of each compound according to the variables: oil °API, contact time oil/seawater after spillage and ambient temperature. The models presented can be used to quantitatively predict BTEX contamination in seawater with accuracy greater than 99%, within the studied ranges. The contact time oil seawater was the most determining factor in the concentration/contamination of BTEX in the Petroleum FSA. According to CL50% the soluble fraction of oil of °API 39.8 showed higher toxicity, with a LC50% of 46.07 with 1 h, and 11.38 with 96 hours of contact time oil/seawater, while the oil °API 32.8 showed lower toxicity, however presented higher concentrations of BTEX in FSA, this is possibly due to the synergistic effect of other hydrocarbons that became bioavailable to Mysidopsis juniae during the tests. The samples demonstrated potential environmental risk in cases of oil spillage, and these results can help in the development of a risk assessment of oil spills and serve as a useful analytical tool for emergencies.


Introduction
The growing global demand for oil and gas has driven the advance of exploration and production operations (Harfoot et al., 2018). In the last 20 years Brazil has experienced a great increase in maritime traffic, which has contributed to the increase in the number of accidents with oil tankers (Tournadre, 2014;Marques et al., 2017). In 2019-2020 the north and southeast coast of Brazil suffered great environmental and socioeconomic impact with a spill of more than 200 tons of crude oil, reaching about 3,000 km long. Being considered the largest extension of contaminated coast in the world recorded in 30 years (IBAMA, 2019;Cardoso et al., 2020). The impacts on the marine environment nature are of incalculable magnitude, this without assessing the long-term damage due to the recalcitrance of some of the heavier components of the oil (Kanjilal, 2015;Rekadwad & Khobragade, 2015;Turner et al., 2014;Fingas, 2011;Stout & Wang, 2007;Peters et al., 2005).
In cases of spillage or leakage of crude oil in a marine environment, the oil is immediately subject to a variety of physical-chemical and biological processes, which includes: scattering, dispersion, evaporation, solubilization, emulsification, photo-oxidation, biodegradation, chemical transformation and sedimentation (Reddy et al., 2012;Liu et al., 2012;McNutt et al., 2012;Gutierrez et al., 2013;Hall et al. , 2013;Liu & Liu, 2013;Kiruri et al., 2013;Aeppli et al., 2014;Ruddy et al., 2014;Chanton et al., 2014;Lewan et al., 2014;Cravo-Laureau & Duran, 2014;Gros et al., 2014;Bacosa et al., 2015;King et al., 2015;Stout et al., 2016). Oil is a mixture of complex hydrocarbon compounds that vary in volatility and solubility, which are potentially toxic to marine organisms, particularly monoaromatic hydrocarbons such as benzene, toluene, ethylbenzene and xylenes (BTEX) that pose threats to humans and marine species (Meador and Nahrgang, 2019). In regions of intense oilrelated activities, these compounds are present in appreciable concentrations (sometimes even in the milligram per liter range) (Saeed et al., 2013). Brazil, being an oil producing country, is heavily involved in production, refining and transportation activities, this can inevitably result in oil spills. Development, v. 10, n. 3, e44610313555, 2021 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v10i3.13555 3 During a crude oil spill accident in the ocean, an essential part of the planning and decision-making processes is driven by predictions of integrated oil and gas spill models (Nissanka & Yapa, 2018). Therefore, improvements in models for ocean oil spill cases become invaluable in the contingency planning and mitigation of environmental damage.
Numerical/computational models provide a critical view of the destination and transportation of oil and subsequent weathering processes, which are important in emergency responses and oil mitigation attempts in order to minimize environmental impacts.
However, there not much information regarding models of prediction of contaminants in the soluble phase of petroleum (FSA) at sea in the literature.
Therefore, the objective of this work is to develop predictive models that allow quantifying BTEX concentrations in cases of oil spillage or leakage in seawater, in order to assist in the evaluation of environmental impacts and contingency planning of accidental oil spills. In addition to the procedures for assessing the potential risk in the event of oil spill accidents, the acute toxicity assessment of the FSA was also carried out using the Mysidopsis juniae.

Experimental Planning
In order to obtain mathematical models capable of predicting the concentration of BTEX compounds in seawater after contact with crude oil, a statistical modeling was performed using a factorial experimental planning matrix 2³ establishing three factors: the API degree of oil, the ambient temperature and the contact time of oil and seawater, as independent variables and as dependent variables the concentrations of BTEX. For each factor, two levels were selected: low (-1) and high (+1), according to Table 1. Independent factors were selected based on Saeed et al. (2013) and Saeed et al. (1998). The tests were carried out in duplicates in a total of 16 trials. Moreover, to evaluate the statistical significance of the proposed models, a study of variance analysis (ANOVA) was carried out at 95% confidence. All analyses, statistical calculations and graphs were obtained with the software STATISTICA 7.0. Research, Society and Development, v. 10, n. 3, e44610313555, 2021 (CC BY 4.

Soluble fraction of oil
The soluble fraction of oil in seawater was based on Saeed et al. (2013) and Saeed et al. (1998), First, the oil was added into a glass jar with about 2 Liters of seawater (adjusted to 34 ± 1º‰), being transfered slowly with the aid of a glass rod, until it formed a 5 mm film (oil:water ratio 1:25 v/v). The system was maintained in low agitation without vortex formation, using a magnetic stirrer.
In order to evaluate the concentration of monoaromatic hydrocarbons (BTEX) in the soluble fraction of petroleum (FSP) in seawater, the contact time of oil and seawater of 1h and 96h and at room temperature of 20 °C and 30 °C (± 0.5 °C) was varied, according to the predetermined experimental matrix previously determined. After the defined times, water samples were collected with the aid of a BT 100S peristaltic pump (Lead Fluid, Hebei, China) with low flow (50 mL/min) and transferred to 44 mL glass vials, with PTFE/silicone thread and septum. A sample was also collected to analyze the toxicity of the FSA of the petroleum in seawater in amber glass vials of 1 L.

Analysis of BTEX by chromatography
The chemical analysis of BTEX in the soluble fraction of the oils was based on the USEPA 8260B (Volatile Organic Compounds by Gas Chromatography-Mass Spectrometry (GC/MS) method, using the gas chromatograph coupled to a Clarus 600 MS mass spectrometer (PerkinElmer, Waltham, USA), and in the USEPA 5021A. Chromatographic separation performed with a stationary phase TG-624 SILMS (Thermo Fisher Scientific, Massachusetts, USA) column of 6% cyanopropylphenyl and 94% dimethylpolysiloxane measuring 30 m in length, 0.25 mm in internal diameter and 1.40 μm film thickness. The drag gas used was helium (99.999% purity, Linde Gases Ltda, Brazil) constant flow of 1 mL/min.
The chromatographic conditions used were: initial temperature of 40 °C maintained for 2 min, following a heating ramp of 100 °C, rate of 10 °C/min. The temperature was then raised to 200 °C, rate of 30 °C/min, remaining at this temperature for 4 min. The running time was 15.33 minutes, the temperature of the injector and the transfer line were 220 °C and 200 °C, respectively. The detection of the BTEX was performed in a quadrupole mass spectrometer, with electrical ionization at a temperature of 200ºC, in the total scanning mode of the spectra (full scan), with mass-over-load ratio (m/z) ranging from 35~270. Quantification was performed with external standardization and the identification of analyses was performed comparing the retention and confirmation times of the mass spectra with the compounds of the standard mixture.

Acute toxicity test
Mysidopsis juniae were cultivated in the ECOTOXLAB (Aquatic Ecotoxicology Laboratory) of the Primary Processing and Reuse Center of Produced Water and Waste at UFRN. The crop was maintained under controlled conditions in aquariums (8 L) with filtered natural seawater (filter of 0.5 μm) and gender ratio of 20 males to 60 females. The abiotic conditions were as follows: salinity of sea water at 35 ± 1º‰; photoperiod of 12h light / 12h dark; temperature of 25 ± 2 ° C; constant aerate. Mysidopsis were fed with the nauplios of the microcrustacean Artemia sp. enriched with cod liver oil and fish oil (age: 48h) daily.
The acute toxicity tests were based on NBR 15308 of the Brazilian Association of Technical Standards (ABNT, 2017).
Ten juveniles of M. juniae (1-8 days of age) were exposed in triplicate for 96 hours to five concentrations of the soluble fraction of oil in seawater (6.25, 12.5, 25, 50 and 100%), in addition to a negative control of filtered seawater. The test organisms were fed daily with Artemia sp. (age: 48h). Organisms were considered dead when immobilized after mild mechanical stimulation. The pH, salinity and dissolved oxygen concentration (DO) were measured at the beginning and end of each experiment using a Hanna Multiparameter -HI9829 (Hanna Instruments, Limena, Italy). The validation criteria of the test Research, Society and Development, v. 10, n. 3, e44610313555, 2021 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v10i3.13555 5 were based on the survival of the control, which should be higher than 90%, and on the sensitivity to zinc as a reference substance, at concentrations of 0.10; 0.18; 0.32 and 0.56 mg/L (ABNT, 2017).
After 96h of exposure the tests with the M. Juniae were terminated and the final count of the individuals was performed, followed by the calculation of the median lethal concentration at 50% of the tested organisms (LC50), using the Trimmed Spearman-Karber program (Hamilton et al., 1977). The data of the LC50 were calculated for 96h, with a 95% confidence interval. The percentage of survival of the organisms at the end of each test was also evaluated and for the validation of the results, only trials in which the mortality of the organisms in the control was less than or equal to 10% was adopted as a criterion of acceptability. Table 2 presents the configuration of the experiments, as well as the experimental concentrations obtained for benzene (B), toluene (T), ethylbenzene (E), xylenos (X) and the sum of BTEX in the soluble fraction oil in seawater, related to factorial planning. The levels of significant factors and effects of interactions between the factors that influence the responses were analyzed by factorial planning 2 3 . Three important factors were selected, as described in the literature: the oil °API, contact time of oil/seawater and temperature of the environment, both at two levels, planned according to the matrix described in Table   2. Figure 1 shows

Regression Analysis
Regression analysis was applied in order to obtain mathematical models that resulted in second-order polynomies equations and that express the relationship between each response and significant factors and interactions. The following models were obtained by experimental design. Eq. 5

Statistical validation of the models obtained
A very important step is the validation of the constructed models, that is, their adequacy to the experimentally obtained responses, which will dictate their predictability (Breitkreitz et al., 2014). Moreover, the validation of the models was determined by the Variance Analysis (ANOVA), presented in Table 3, and the correlation measure used to estimate the model was the coefficient of determination (R 2 ). Research, Society and Development, v. 10, n. 3, e44610313555, 2021 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v10i3.13555 7 Table 3. Experimental planning matrix 2 3 . indicates a good regression and, consequently, a mathematical model that satisfactorily represents its experiment (Barros Neto et al., 2010). The models showed significant regression at a 95% confidence level (Fcalc > Ftab) with R 2 > 0.99 for all models,

Source of Variation
showing that they explained above 99% of the variation of the experimental data (Table 3). The Fcalc value of the regressions ranged from 42.33 to 614.68 times higher than Ftab, to a confidence level of 95% for all cases analyzed, implying useful models to perform predictions within the studied range (predictive models).

Quantitative evaluation of proposed models
In order to quantitatively evaluate the prediction of the proposed models, it was chose to use the mathematical equations to calculate the concentrations of BTEX compounds in the experimental conditions previously performed in this study. Thus, the values of the variables °API, time and temperature in the mathematical equations and the concentration of BTEX compounds then calculated were replaced. Table 4 shows a relationship between the actual values, obtained by the mean between the experimental tests in duplicate, and the values calculated by the models.

Effect of variables on BTEX solubilization
The effects of the oil °API, contact time oil/seawater after spillage and ambient temperature, were evaluated using Pareto diagrams. Variables with statistical significance in the process were identified for a confidence level of 95%. Figure 2 shows the Pareto diagrams for BTEX.
By analyzing the Pareto diagrams ( Figure 2) one can observe that the oil °API, oil contact time in seawater after the spill and temperature have significant effects, in isolation and/or combined in the solubilization of BTEX in seawater, within the studied intervals. Taking into account the effect of the isolated variables and combined with the other factors, it is observed that the time variable showed greater influence on the solubilization process of BTEX, followed by temperature and oil °API, respectively. For all contaminants, the variable time presented an effect directly proportional to the response, that is, the longer the time, the higher the concentration of BTEX in seawater after the oil spill. According to Meador and Nahrgang (2019), Marques et al. (2017) and Sammarco et al. (2013), about 15% of the oil mass is composed by volatile hydrocarbons (VOCs) that are quickly lost by evaporation in cases of oil spillage at sea. In this study, we noticed the increase in BTEX Development, v. 10, n. 3, e44610313555, 2021 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v10i3.13555 9 concentrations in the soluble fraction of oil in seawater with the increase of contact time oil from seawater from 1 h to 96 h, from ≅ 2 to 15 times. This may have occurred possibly, because all experiments were carried out in a laboratory with controlled ambient conditions, without weathering. For the temperature variable, the inversely proportional effect was perceived for all models (Figure 2), which implies that the lower the temperature, the higher the solubilization of BTEX in seawater. As BTEX are volatile compounds, temperature directly influences the volatilization of these compounds, due to the high vapor pressures (Gebara et al., 2013), thus for marine environments with temperatures higher than 20 ºC, to a tendency of greater volatilization of BTEX, consequently decreasing the amount solubilized in water. In relation to the oil °API, different behaviors were verified: for benzene, the higher the oil °API, the lower the presence of these contaminants, for Xylees, the higher the oil °API, the greater Development, v. 10, n. 3, e44610313555, 2021 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v10i3.13555 10 the presence of this contaminant in seawater after oil spillage. For toluene, ethylbenzene and BTEX, °API does not influence the process in isolation. In these cases, only when combined with the other variables, the °API will influence the solubilization of these compounds in seawater.

Acute toxicity -Mysidopsis juniae
Using the test method with Mysidopsis juniae, according to NBR 15308 (ABNT, 2017), it allowed to determine the LC50%, which is the median lethal concentration that causes mortality or immobility at 50% of M. juniae after the 96-hour exposure period, being it calculated by the Trimmed Spearman-Karber method (Hamilton et al., 1977). Making inference about the acceptable internal laboratory quality control, in all experiments, mortality did not exceed 10% in the negative control and sensitivity of M. juniae to zinc, a LC50-96 h of 0.26 ± 0.05 mg/L was reached, within the values reported by Figuerêdo et al. (2016a), Figuerêdo et al. (2016b, Nilin et al. (2019) and Nascimento et al. (2020), ranging from 0.25 to 0.46 mg/L. Thus, the sensitivity of organisms is a determining factor of environmental impact, as well as the intensity and frequency of exposure (Connon et al., 2012). Table 6 and Figure 3 show the number of dead organisms and the mortality rate of Mysidopsis juniae after exposure to the soluble fraction of oil in seawater at 6.25, 12.5, 25, 50 e 100% (v/v).  Source: Authors (2021). Development, v. 10, n. 3, e44610313555, 2021 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v10i3.13555 11 The parameters (salinity, pH and DO) were measured during the ecotoxicological tests to verify the minimum survival conditions of M. juniae. The physical-chemical parameters were measured at the beginning and end of each experiment (including controls), and were within the acceptable range (ABNT 2017). The mean pH was 8.0 ± 0.3 (n = 24), the mean DO concentration was 4.8 ± 0.6 mg/L (n = 24) and the mean salinity was 34.2 ± 0.6 (n = 24). Therefore, the authors chose not to renew the test solutions, aiming to achieve exposure conditions that most closely resemble dweller in a case of accidental oil spillage in the environment.
Petroleum hydrocarbons are chemical compounds widely known for their toxicity and mobility in water. In order to evaluate the toxicity of the soluble fraction of the oils studied in seawater, the acute toxicity was evaluated by the mortality rate of Mysidopsis juniae organisms in 96 h of exposure to soluble fraction of petroleum. Figure 3 showed an increase in the mortality rate of organisms with an increase in the contact time of oil and seawater from 1 h to 96 h, corroborating the results found in the Concentrations of BTEX of FSA. However, it was observed (Figure 3) that the FSA of the oil sample of ºAPI 39.8 with a temperature of 30 ºC and contact time oil seawater of 96h, which presented average concentration of benzene 33.3 μg/L, toluene 288.0 μg/L, ethylbenzene 25.8 μg/L, xylenes 326.6 μg/L and total BTEX 673.7 μg/L much lower than the FSA sample of the same oil with room temperature of 20ºC and the contact time oil sea water of 96h, but presented a similar mortality rate of 72%. This high mortality rate is possibly due to other larger chain hydrocarbons, such as polyaromatics (HPAs) that have been solubilized in seawater with increased temperature, or due to petroleum compounds called complex mixtures of unresolved aromatics (MCU), as reported by Melbye et al. (2009), which are persistent pollutants, biocomulant and toxic, generally neglected in petroleum contamination (Booth et al. 2007). Table 7 shows the median lethal concentration (LC%) calculated with a 95% confidence limit for all oil soluble fraction samples studied. There is an increase in toxicity with the contact time of oil in seawater from 1 hour to 96 hours, this condition occours since with the increase of time, hydrocarbons become more bioavailable to Mysidopsis juniae, possibly causing greater toxicity of the soluble fraction of the oil.

Conclusion
The proposed models can be used to quantitatively predict BTEX contamination in seawater with accuracy greater than 99%, within the range of contact time seawater oil, ambient temperature and °API addressed in this work. The approach of the statistical modeling technique using experimental planning was constituted, in this experiment, as a potential tool in the prediction and dimension of contamination of BTEX compounds in seawater after oil spill age and its possible environmental impacts. From the developed models, it is possible to estimate the concentration of BTEX during an oil spill or spill, and this information can assist in emergency plans, as well as in the taking of mitigation and remediation actions of contaminated areas.
Moreover, regarding toxicity, it was verified that the contact time of oil with seawater directly influences the increase in Development, v. 10, n. 3, e44610313555, 2021 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v10i3.13555 12 toxicity due to hydrocarbons becoming more bioavailable to Mysidopsis juniae, possibly causing greater toxicity of the soluble fraction of the oil. The experimental analyses indicate that the concentrations of BTEX in seawater and the toxicity of FSA are favored at the lowest temperatures, because at 30°C it is observed the volatilization of these compounds with 96h, decreasing their concentration in seawater, this information being a warning point for regions with temperatures between 20 °C and 30 °C, where accidents occur with these types of oil.