Offshore produced water treatment by a biofilm reactor on the seabed The effect of temperature and matrix characteristics

In many industrial processes a large amount of water with high salinity is co-produced whose treatment poses considerable challenges to the available technologies. The produced water (PW) from offshore operations is currently being discharged to sea without treatment for dissolved pollutants due to space limitations. A biofilter on the seabed adjacent to a production platform would negate all size restrictions, thus reducing the environmental impact of oil and gas production offshore. The moving bed biofilm reactor (MBBR) was investigated for PW treatment from different oilfields in the North Sea at 10 ◦ C and 40 ◦ C , corresponding to the sea and PW temperature, respectively. The six PW samples in study were characterized by high salinity and chemical oxygen demand with ecotoxic effects on marine algae S. pseudocostatum (0.4% < EC 50 < 2.7%). In continuous operation over a year, MBBR achieved a stable COD removal of 64 ± 5% at 10 ◦ C and 68 ± 8% at 40 ◦ C. Batch experiments revealed that most dissolved compounds were removed (up to 63%) within 3 h of treatment. High temperature (40 ◦ C) was a key parameter to achieve a faster kinetics with degradation constant rate (k) up to eight-fold faster compared to 10 ◦ C. Alongside contaminants removal, PW toxicity was also reduced (64 – 89%) during MBBR at both temperatures, hot and cold. The toxicity reduction was most likely related to the elimination of dissolved organic compounds, such as phenols, naphthalenes and BTEX. The biofilm was able to handle PW with high oil in water content from unstable production, as well as high salinity. Thus, MBBR seems to be a realistic solution to treat PW with complex and variable composition by removing harmful components towards the zero harmful discharge goal.


Introduction
The extraction of oil and gas produces a natural by-product known as produced water (PW).PW is a complex mixture of formation water (water trapped for millions of years in a geologic reservoir), condensation and injection water.In addition to the inherent compounds present in PW, production chemicals such as biocides, corrosion inhibitors, sulfide scavengers and methane-hydrate inhibitors can be added at different stages during production (Neff et al., 2011).Over the years, the amount of PW generated has been steadily increasing with the ageing of oilfield production wells.Even though PW reinjection is considered the Best Environmental Practice for PW management, discharge to the sea is the most common management solution.In the oceanic area covered by the Oslo-Paris Convention for the Protection of the Marine Environment of the North-East Atlantic (OSPAR), almost 300 million m 3 of PW was discharged to sea in 2020, which represents double the amount of PW reinjected (OSPAR Comission, 2020).PW contains numerous dissolved and particulate organic and inorganic substances, with many of these substances reported to possess the ability to cause adverse effects to marine organisms (Farmen et al., 2010;Hansen et al., 2019).For example, it has been reported that wild-caught fish species from the areas close to oil platforms have few indications of exposure of contaminants from PW discharge (Grøsvik et al., 2009).
The conventional PW treatment processes installed at offshore platforms are composed of physical separation technologies such as gravity settlement and hydrocyclones.The phase separation technologies are able to clean the water to meet current legislation during stable production conditions (OSPAR, 2013).The main treatment objective is to reduce oil to 30 mg L − 1 , but dissolved compounds will remain in the water phase and might be harmful to the marine environment (Bento and Campos, 2020;Beyer et al., 2019).To improve treatment efficiency and meet stringent discharge limits there is the need to retrofit existing technologies.Moreover, there is also a requirement for oil and gas operators to seek continuous improvement, i.e., reduce the environmental impact to As Low As Reasonably Practicable (ALARP) applying the Best Available Technology (BAT).
Biological processes are the most common techniques used to treat water, especially for the removal of organic compounds in wastewatertreatment plants (di Biase et al., 2019).It has the main advantages of lower operation costs, no usage of chemicals and higher removal efficiencies for a wide spectrum of dissolved compounds compared with physicochemical treatments (Lusinier et al., 2019).Nevertheless, wastewater with a high salt content is more difficult to treat than another industrial wastewater.A great challenge, mainly for offshore application, is associated with the large footprints of most biological processes, which do not allow for implementation of large bioreactors.Therefore, research has been focusing on biological processes that could yield high treatment performance with simultaneous high flow rates (e.g., 50,000 m 3 /d), and small footprint.The complex composition of real PW samples might also jeopardize the implementation of biological processes in oil-production facilities.PW salinity is thought to affect the performance of biological treatment processes because introduces microorganisms to a challenge of dealing with osmotic stress and it might strongly inhibit the biological activity inside the bioreactor (Camarillo and Stringfellow, 2018;Mannina et al., 2016;Nakhli et al., 2014).Among biological technologies to treat oil and gas PW, fixed-film reactors and membrane bioreactors are among the most investigated (32% and 20%, respectively).However, these methods seem to be inadequate when highly toxic recalcitrant compounds are present in water (Camarillo and Stringfellow, 2018).
The moving bed biofilm reactor (MBBR) has been widely applied to treat both urban and industrial wastewaters due to the advantages of the biofilm process i.e., compact, stable removal efficiency and simplicity of operation.The capacity of biomass retention inside the reactor is a key factor to avoid washout or loss of slow-growing bacteria.The high active biomass concentration on the carriers decreases sensitivity to aggressive conditions and also increases resistance to high organic loading comparing with traditional biological technologies (Camarillo and Stringfellow, 2018;Hasanzadeh et al., 2020).Even though MBBR has been effectively used for wastewater treatment, the number and scope of studies of offshore PW treatment are limited with only few studies published (Dong et al., 2011;Hasanzadeh et al., 2020;Lusinier et al., 2021).Onshore it has been proposed to incorporate biological removal of organic contaminants from PW because of cost-effectiveness of biological compared to physical/chemical treatment (Fakhru'l-Razi et al., 2009a;Lefebvre and Moletta, 2006;Lu et al., 2009).To our knowledge, there are no literature that addresses the potential of MBBR to treat real offshore PW.The current literature in PW treatment mostly uses synthetic water and treatment performances are not representative of real field conditions, since better removals are achieved when synthetic PW was used instead of real samples (Camarillo and Stringfellow, 2018).Furthermore, besides complex composition of PW, this water is characterized by high temperatures at offshore platforms (Yang, 2011).It is well known that temperature is an essential operating factor with significant effects on biological process performance, since affects reaction rates of most chemicals and biological degradation rates (Alisawi, 2020;Yang, 2011).Nevertheless, there is a lack knowledge on the impact of the temperature upon the behavior of MBBR (Alisawi, 2020;Majid, 2019).When addressing temperature influence on the MBBR performance, the studies are mostly conducted at room or very low temperatures (20-25 • C; 1 • C) and focusing on ammonia removal from municipal/domestic wastewater (Abujayyab et al., 2022;Ashkanani et al., 2019;Hoang et al., 2014;Shore et al., 2012).In case of PW, some studies reported that temperature stabilization was necessary prior biological treatment experiments (Ji et al., 2007).Understanding the impact of temperature on MBBR performance will increase the knowledge of the best ways related to high removal efficiencies and ensure operational stability.This study investigated the potential of MBBR at two different very distinct temperatures (10 • C and 40 • C) to reduce the dissolved organic compounds and inherent PW toxicity.The biological treatment was tested with different offshore PW samples from the North Sea aiming to cover distinct characteristics linked to composition of the wells in production, possible malfunction of equipment's at offshore production (unstable production), or influence of chemicals injection during production.

Offshore produced water samples collection
Produced water samples were collected from three different offshore platforms in the North Sea.In each different platform, the samples were taken from two different sampling points: degasser outlet (discharge point) and upstream hydrocyclone to simulate a stable and unstable production, respectively.PW samples were collected on the production platform following protocols established by the operators.The samples were collected in 10 L plastic bottles filled up to the top without headspace, which avoids evaporation of organic analytes or dissolution of oxygen.The samples were transported to onshore as quickly as possible.The characterization of each sample was carried out upon arrival at Technical University of Denmark and stored in the dark at 4 • C to prevent biological growth until further use for MBBR batch experiments.

MBBR experimental setup and continuous operation
Two MBBR (cylindrical glass reactors Ø = 11 cm and a height of cm) were operated in parallel under two different representative temperatures: 10 • ± 0.5 • C (average temperature in the North Sea) and ± 0.5 • C (expected temperature of PW in the bioreactor at offshore platform since PW temperature can be as high as 90 • C (Yang, 2011).The temperature of each reactor was controlled by circulating water from cooling or heating thermostat baths through the water jackets of the reactors.The reactors were covered by an insulation material to avoid temperature fluctuation.Each reactor had an operating volume of 3 L with a filling ratio of 50% (500 AnoxKaldnes® K5 biofilm carriers; 800 m 2 m 3 protected surface area of each carrier).The biofilm carrier movement was caused by air bubbles produced by the air pump that adds oxygen from the bottom of the reactor.Dissolved oxygen (DO) was always higher than 4 mg L − 1 .The biofilm carriers used in this study were already established from a previous project also related with PW.
The reactors have been operating at 10 • C and 40 • C in a continuous flow mode with onshore PW from oil and gas production.This allows a specialized microbial consortium to growth and adapt to oil-field PW, not only in terms of salinity but also to the organic compounds usually present in this type of water.The feed was pumped through a peristaltic pump to the top of the reactor.The effect of different operational variables i.e., salinity and HRT on systems performance was evaluated as total COD removal.The reactors started to operate with 1 day HRT (COD loading of 0.5 kg COD m − 3 d − 1 ) and salinity of ~69 g Cl/L.Nutrient dosage was adjusted in the feed by adding ammonium bicarbonate (NH₄HCO₃) and potassium phosphate monobasic (KH 2 PO 4 ), in order to have C:N:P (100:10:1) throughout the continuous feed operation mode.The salinity gradually increased over the time up to ~90 g Cl/L to mitigate the detrimental effect of PW characteristics with higher salinity on microbial activity.Commercially available sea salt was added to adjust conductivity levels.Changes were kept for at least one-week operation before the start of any batch experiment to ensure that biofilm was perfectly acclimated to the new conditions.It was assumed that the steady state condition occurred when variation in the removal efficiency was ±20% compared to previous measurements.Operating parameters such as pH, conductivity, DO, temperature and total COD in the reactors were periodically monitored.The pH of the feed ranged between 7.0 and 8.0 and no adjustments were required.

MBBR batch experiments
Batch experiments were carried out aiming to study the organic compounds removal kinetics based on soluble COD removal.The batch experiments lasted for 24 h and samples (i.e., COD, BTEX, VFA, nontarget hydrocarbons and toxicity) were taken from each reactor in predefined times (0, 0.5, 1, 2, 3, 4, 6, 8, 24 h).The MBBR batch experiment started by withdrawing half of the water from each reactor with raw PW.This prevents any shock for the biofilm to the raw offshore PW and enables the study of degradation kinetics for COD and target specific compounds.The pH and conductivity in each reactor were also monitored.There were no pH adjustments nor addition of nutrients during the batch experiments.Deionized water was added if needed in the reactors to counteract water evaporation and keep the same conditions inside the reactors such as salinity levels.

Abiotic experiments
To assess the extent of removal by abiotic factors such as temperature, experiments without carriers at 10 • C and 40 • C were performed.The experiments were carried out with PW in 1 L reactor (same setup as the MBBR batch experiments) and lasted for 24 h.Samples were taken from each reactor over the time (0, 0.5, 1, 2, 3, 4, 6, 8, 24 h) for COD, BTEX and VFA.

Eco-toxicity measurements
Ecotoxicity tests were performed to assess the potential toxicity of PW before and after MBBR.Standardized tests used were the Microtox® (Vibrio fischeri) acute toxicity test (ISO Standard 11348-3, 2007) and growth inhibition test with the unicellular temperate marine algae S. pseudocostatum (ISO 10253, 2006).

Microtox® bioassay
Test kits (ABOATOX, Finland) were used for quantification of decrease in bacterial luminescence following the ISO 11348-3 standard with modifications.Freeze-dried bacteria, Aliivibrio fischeri (formerly Vibrio fischeri) were reconstituted in 12 mL saltwater solution (2 wt%) and left for 20 min before the experiment was initiated.Five test concentrations were prepared from the PW or MBBR treated PW (10 and 40 • C) by dilution with the 2% saltwater solution, resulting in a salinity between 2.0 % and 3.5 %.The salinity was measured using a conductivity meter (Cond 315i, WTW, Xylem Analytics, Germany).The exposure concentrations used for all samples were 2.5, 5, 10, 20, 40% (v/v).The stock solutions were prepared in 10 mL measuring vials at a concentration corresponding to twice the test concentration to allow for dilution with the reconstituted bacteria solution.The background luminescence of the 2 % saltwater solution was corrected for by measuring 200 μL of the solution in 2 mL glass vials (Thermo Fisher Scientific™).Initial bacterial luminescence was measured by pipetting 100 μL of bacteria solution into 2 mL glass vials (Thermo Fisher Scien-tific™) and immediately measuring the luminescence using a luminometer (Luminoskan TL Plus, Thermo Labsystems) denoting the luminescence at time 0. Following the luminescence measurement, 100 μL of the double concentrated test solution was added to the bacteria solution yielding the final test concentrations, i.e., 2.5%-40% (v/v).The experiment was carried out with duplicates for each tested concentration and the controls.Luminescence measurements were repeated after 15 and 30 min.The tests were considered valid if the parallel determination of the controls did not deviate more than 3% and 30 min exposure to 3.4 mg L − 1 of 3,5-dichlorophenol caused between 20% and 80% decrease in luminescence.A correction factor (f kt ) was calculated for control solutions according to ISO Standard 11348-3:2007 to determine the water-dependent decrease in luminescence (Equation ( 1)) and for the test to be valid the correction factor should be between 0.6 and 1.8: where f kt is the correction factor at time t (15 or 30 min), I kt is the luminescence at time t (15 or 30 min) and I 0 is the luminescence at time 0. The relative decrease in luminescence for each sample was calculated according to Equation (2).
where Rel t is the relative decrease in luminescence at time t (15 or min), I kt is the luminescence at time t (15 or 30 min) and I 0 is the luminescence at time 0. The relative decrease in luminescence at 15 and 30 min was plotted in the statistical software R loaded with the drcpackage and used to estimate concentration-response curves, ECvalues and their corresponding 95% confidence intervals using a lognormal function.

Algal growth inhibition test
Algal growth inhibition test was conducted with marine algae S. pseudocostatum (formerly Skeletonema costatum) obtained from the Norwegian Institute for Water Research, Norway (NIVA-BAC 1).It was grown continuously in 100 mL bluecap bottles containing filtered (pore size 0.45 μm, Whatman®, Merck Life Science) natural seawater (obtained from DTU Aqua, Technical University of Denmark) enriched with nutrient following ISO 10253:2006(ISO, 2006).The media was continuously aerated with atmospheric air to allow for sufficient CO flux and keep the algae in suspension.The setup was illuminated from the side with fluorescent tubes (30W/33; Philips Amsterdam, The Netherlands) with an intensity of 61 ± 4 μmol m − 2 s − 1 measured by LI-189 Quantum/Radiometer/Photometer (LI-COR, Nebraska, USA) at a temperature of 20 • C ± 2 • C. The 72 h growth inhibition test was conducted with modifications of the ISO 10253:2006standard (ISO, 2006).Test concentrations were prepared from filtered samples collected from two MBBR temperatures (10 and 40 • C) and different reaction times (0h and 24h) diluted in ISO 10253:2006 algal media based on synthetic seawater.The dilutions were inoculated with exponentially growing algal cultures to obtain an initial density of 2⋅10 4 cells/mL.The cell density was quantified by coulter counter (Beckman Multisizer™ 3, Indianapolis, USA).In practice, 250 μL of an exponentially growing algal culture of 2⋅10 6 cells/mL was added to 25 mL of each exposure concentration and 4 mL was transferred to 20 mL scintillation vials (n = 3), and incubated on an orbital shaker (IKA® Schüttler MTS 4) mounted with a rack and continuously illuminated from below with fluorescent tubes (30 W/33; Philips, Amsterdam, The Netherlands) with an intensity of 79 ± 5 μmol m − 2 s − 1 measured by LI-189 Quantum/Radiometer/ Photometer (LI-COR, Nebraska, USA).
The concentration tested for the MBBR treated samples were 0.45, 0.9, 1.8, 3.75, 7.5 and 15% (v/v).For each concentration three replicates were made, and six replicates were used for the control group.All samples were incubated at similar conditions as described above, and validity criteria stated in the ISO 10253:2006 standard were met for all tests.i.e., control growth of minimum 0.9 day − 1 and a maximum change in pH of 1 unit during the 72 h of incubation.After 0h and 72h aliquots of 0.4 mL from each sample were extracted with 1.6 mL acetone.The algal growth rates were calculated based on the in vitro fluorescence of algal pigments as surrogate for biomass as described by (Mayer et al., 1997).The fluorescence was 430 nm and 670 nm for excitation and emission wavelengths.Background fluorescence was corrected by measuring a blank sample containing seawater and acetone.To avoid interference with precipitates the supernatant of each sample was gently transferred to a new vial before fluorescence measurement.The algal growth rates were calculated assuming exponential growth following Equation (3): where μ is the growth rate (d − 1 ), N 0 is the initial biomass, N n is the final biomass and t d is the length of the test period (d).Additionally, the inhibition was calculated as the growth rate of the control related to the growth rate in each individual exposure following Equation (4).
where I i is the percentage inhibition of growth for concentration i, and μ i is the mean growth rate for concentration i and μ c is the mean growth rate for the control.Growth inhibition based on growth rates was plotted in the statistical software R loaded with the drc-package and used to estimate concentration-response curves, EC-values and their corresponding 95% confidence intervals using a log-normal function (Ritz and Streibig, 2005).

Analytical methods
All chemicals were of analytical grade except for NaCl, which was purchased commercially.The pH and conductivity were measured by Thermo Scientific Orion Star™ A215 pH/conductivity Benchtop Meter.Dissolved oxygen in biofilm reactors was measured using Thermo Sci-entific™ Orion™ DO Probe for Lab or Field.Chemical oxygen demand (COD) was measured using Hach Lange™ kit LCK1814.The samples for soluble COD analysis were previous filtered through a nylon 0.20 μm pore size filter (Agilent Technologies).Nutrients i.e., orthophosphate as phosphorus (PO 4 -P), nitrate nitrogen (NO 3 -N), nitrite nitrogen (NO 2 -N), ammonium nitrogen (NH 4 -N) were analyzed by SKALAR San++ continuous flow analyzer (by Skalar Analytical B.V. analysis, software San series FlowAccess™ V3).Prior analysis all samples were filtered by nylon 0.20 μm pore size filter (Agilent Technologies).Benzene, toluene, ethylbenzene and xylene, as well as volatile fatty acids were analyzed by gas chromatography (more information in Supplementary material).

Non-target hydrocarbons analysis
To have a broader view of the PW composition regarding dissolved hydrocarbons, a screening of the raw/untreated and biologically treated PW samples was carried out.Samples from the experiments were initially in aqueous solution and liquid-liquid extraction with dichloromethane (DCM) was performed to extract organic compounds.Samples were shaken with DCM over several hours and the organic phase collected using separating funnels.The DCM solvent was evaporated off under nitrogen and the samples subsequently re-dissolved in nhexane for analysis.In addition to the samples, hexane and DCM solvent blanks and a straight chain alkane retention time standard were prepared.The samples were analyzed on an Agilent GC-QTOF instrument (7200B QTOF and 7890 GC, Agilent Technologies).The 2D chromatography was achieved with a 3s modulation using a thermal modulator purchased from Zoex and using an Agilent DB-5MS column in the first dimension and a Restek Rxi-17Sil MS column in the second dimension.Retention times in the first and second dimension were 10-60 min and 2-3 min respectively.Data analysis was performed by GC Image software (GC Image, LLC) library search with the NIST Mass Spectral Database (2017).The library search revealed between 1400 and 2000 compounds in each sample.The compounds were grouped into different groups for comparison of the intensity of these groups between samples.The compound groups were normalized to the initial sample before any treatments to highlight changes between samples.

Biological oxygen demand
Biological oxygen demand (BOD) was measured by the manometric respirometric test.The pressure changes were measured by a manometer and converted to oxygen consumption by the respirometric OxiTop® measuring system.High salinity adapted carriers (K1, AnoxKaldnes™ carriers) used as inoculum.The detailed procedure is described in (Ferreira et al., 2022).

Data analysis
The removal percentage (%) of COD was calculated using Equation (5): For comparison purposes and in order to have a simplified evaluation of the data, the results were fitted to first-order degradation kinetics (Eq. (2) using the software GraphPad Prism 9.3.0(Graphpad Software, Inc., USA)).It is important to highlight that even though the first-order model was deemed adequate, and correlations (R 2 ) were higher than 0.94, for some cases may not be truly first-order, and other models could give a better fit to the data (Fig. S1).However, to have a simplified evaluation of the data, the first-order degradation kinetics was used for all the data (Equation ( 6)).
where C 0 is the initial COD concentration; Plateau is the C value at infinite times.

Offshore PW characteristics
A description of the main characteristics analyzed for the offshore PW samples in study are presented in Table 1.In total, three different oilfield platforms from the North Sea were considered.To simulate a stable and unstable process during production at offshore, the samples were collected from two different sampling points: degasser outletdischarge point (stable process) and before hydrocyclones (unstable process).The samples were called effluent (discharge point) and unstable (before hydrocyclones) according to the sampling point throughout the manuscript.The results showed that PW composition is field-dependent since the samples had different characteristics among them.Platform 1 exhibited a higher conductivity (~144 mS/cm) compared to the other two fields (65.5-79 mS/cm).The pH was similar among samples except for effluent from platform 2 , which showed a high pH (pH = 8.7), but still within the range reported in literature (4.3<pH < 10) (Liu et al., 2021).Moreover, this sample also exhibited a threefold increase in terms of COD and low biodegradability (BOD 5 /COD = 0.1) when comparing with the other samples, which can be attributed to the chemicals added during production process (corrosion inhibitors, biocides, emulsion breakers, asphaltene dispersant, etc.).Overall, PW samples showed a total COD average of 500 mg L − 1 with a high BOD 5 /COD ratio (>0.3) indicating a high biodegradable nature.An exception was observed for effluent from platform 2 with a total COD of 1794 mg L − 1 and low biodegradability compared to the other samples (BOD 5 /COD = 0.1).This might be mainly related to the addition of chemicals during production after hydrocyclones, which is also related to the toxicity results discussed below.
The high content of ammonium concentration from PW in the North Sea has been previously reported in literature, and it is greatly dependent on the geographic location of the field (Amakiri et al., 2022;Ferreira et al., 2022).The average concentration of ammonia in PW can range from 10 to 300 mg L − 1 throughout the world, with some authors even investigating potential approaches for ammonia recovery from PW (Chang et al., 2022;Hu et al., 2020;Quartaroli et al., 2017).Regarding VFA, all samples are in the low range of the very wide interval (2-4900 mg L − 1 ) reported in oil-field PW samples from over the world (Fakhru'l-Razi et al., 2009b).
Studies have shown that hydrocarbons in PW are dominated by the volatile aromatic fraction of the oil, namely benzene, toluene, ethylbenzene and xylene (Røe Utvik, 1999).As expected, higher concentration of dissolved and dispersed oil compounds was observed for PW samples collected before hydrocyclones (simulating an unstable process at offshore platforms).These measurements showed that concentrations of the quantified compounds were in the same order of magnitude as previously observed in PW from other oil fields including in the North Sea (Horner et al., 2011;Liu et al., 2021).In addition, non-target gas chromatography analysis revealed between 1400 and 2000 compounds in each sample, which could be grouped into 7 groups: alkanes, alkenes, alcohols, phenols, benzenes, naphthalenes and phenanthrenes.
Besides the naturally occurring compounds in offshore PW, chemicals added during production process, as well as the composition of oil and gas in the reservoir will have a very high impact on toxicity (Al-Ghouti et al., 2019).It has been noticed that production chemicals could increase the soluble fraction of oil components in PW, and thus also increase the PW toxicity (Henderson et al., 1999).This may justify the difference not only in terms of organic content, but also toxicity among the different PW samples in study.All the PW samples revealed to be to some extent toxic for the tested organisms.The Microtox acute toxicity using A. fischeri showed that EC 50 ranged from 29 ± 10% to >40% for untreated PW samples from platform 1 and 3, which means that higher toxicity using this test organism was observed for PW from platform 2 (EC 50 < 10%).For the tests using S. pseudocostatum, the EC 50 was below 3% for PW from the three platforms.The tests using algae revealed that the sensitivity of each microorganism can vary, and it is important to conduct eco-toxicity tests using distinct test organisms.

System performance-response to salt and loading fluctuation
MBBR started to operate with an inlet COD concentration of 0.5 kg COD m − 3 d − 1 and conductivity range of 87-88 mS/cm corresponding to ~69 g Cl/L.The results presented in Fig. 1 show that bioreactors operating at 10 • C and 40 • C had a stable performance for over a year measured as total COD removal efficiency.During this period, the average tCOD removal was 64% ± 5 and 68% ± 8 at 10 • C and 40 • C, respectively.No differences were observed between cold and hot temperatures with HRT greater than 24 h.During this period, the reactors were also exposed to different operational changes.After one-month operation, the biomass in the reactors was exposed to an increase of salt concentration that rose from ~69 g Cl/L to ~90 g Cl/L.Despite the salt increase, the results showed a steady state operation with similar COD removal rates for both reactors during this period (from month-two to month-five).It was also observed that increasing the HRT to 4 days at
A.R. Ferreira et al. month-eight rose the salinity with the conductivity increasing from 76.8 to 85.3 88 to 101 mS/cm in reactor operated at 40 • C.This was related to the water evaporation at low inflow rates.Nevertheless, MBBR kept the same performance in terms of COD removal.This means that biomass was always resistant to process variation, highlighting biomass acclimation to salinity.A review and meta-analysis study carried out by (Camarillo and Stringfellow, 2018) revealed that biological treatment of real PW samples reported an average COD removal of 73% for low salinity levels (TDS<50,000 mg L − 1 ) comparing with 54% when salinity increased (TDS>50,000 mg L − 1 ).Key issues were related to microbial acclimation, toxicity, biological fouling, and mineral scaling during biological treatment (Camarillo and Stringfellow, 2018).In the same study, the microbial inhibition by salinity was reported to be variable according to the treatment system (Camarillo and Stringfellow, 2018).
In case of MBBR, the present results showed that biomass immobilized on a free-support media is a key feature comparing to other treatments since microorganisms can be slowly acclimated to high salinity tolerating the changes on operational parameters without compromising treatment efficiency.Biomass gradually acclimated to salinity was previously reported as a key event to treat wastewater at elevated salinities by MBBR e.g.: (Nakhli et al., 2014).

MBBR batch experiments
The removal of soluble COD (sCOD), specific organic compounds i.e., BTEX and VFA, nutrients and toxicity were accessed during MBBR treatment of PW at two different temperatures.

Effect of temperature on COD removal by MBBR
The effect of temperature on MBBR performance was evaluated with offshore PW samples from degasser outlet, which simulates a stable production.As shown in Fig. 2, the overall kinetic data indicate that degradation rate of sCOD can be characterized by a first-order reaction kinetic model, as evidenced by a high correlation coefficient (R 2 > 0.94; Table 2).
For PW from platform 1 and 3, the biodegradable sCOD removal ranged from 34% to 63% within 3 h.A faster removal was observed for MBBR operated at hot temperature (40 • C) with degradation constant rate (k) up to eight-fold faster than MBBR at cold temperature (10 • C).The differences on sCOD removal between the two tested temperatures decreased over the time and upon 8 h of treatment, 52-74% and 64-78% of sCOD removal was achieved at 10 and 40 • C, respectively.After 8 h, no significant further removal was observed except for PW from platform 2. In this specific case, the reaction rate at 10 • C seems to be very close to zero-order decay with a constant sCOD removal over the time independent of the initial concentration.The results seem to indicate that biofilm was inhibited due to the presence of toxic compounds in PW such as production chemicals (e.g., biocides), which contributed to a higher COD in PW.Nevertheless, the removal rate at 40 • C was threefold faster than 10 • C.This result indicates that biofilm from MBBR operated at hot temperature seemed to be more robust and less sensitive to the presence of toxic compounds.Characterization of microbial communities in the biofilm was not on the scope of this study, but the present results are supported by literature that reports different microbial community structure in biological treatment depending on the  prevailing temperature (Alisawi, 2020;Martín-Pascual et al., 2016).Low temperatures decrease biological treatment efficiency due to physiological characteristics, low microbial growth rate and microbial activity (Zhou et al., 2018).The present results show that temperature is an essential MBBR operation parameter to attain sCOD removal and achieve a higher treatment efficiency with a lower footprint.However, it is also important to note that the effect of high temperature on removal rates was not observed in case of PW from platform 1A, which is characterized by the highest salinity (~90 g Cl/L).Even though conductivity was monitored during the batch experiment, an abrupt salinity increase within the first hour due to water evaporation at hot temperature might have inhibited to some extent the microbial activity.Therefore, during the first hours of MBBR batch experiment the reaction rate of sCOD removal reduced significantly when comparing to 10 • C due to osmotic shock from the salinity spike, since biofilm needed to quickly adapt.The negative impact of high salinity on microorganisms has been previously discussed by other authors e.g.(Lefebvre and Moletta, 2006).

Effect of high dispersed oil in water concentration on MBBR performance.
Nowadays, offshore PW treatment generally refers to oil treatment, namely the non-dissolved hydrocarbons that are removed before discharge to sea.However, it is important to study the MBBR performance if production operates under unstable conditions.PW samples from platform 1B, platform 2B and platform 3B were collected from upstream hydrocyclone and are characterized by higher concentration of particulates and dispersed oil compounds, like polycyclic aromatic hydrocarbons and alkylated (Liu et al., 2021).The sCOD removal kinetics for each PW sample during MBBR is presented in Fig. S3, and the degradation rate constants are in Table 2.The results show that higher concentration of oil in water had no remarkable influence on biofilm performance since no differences were observed comparing with stable conditions.In 3 h of MBBR treatment, more than 40% of sCOD decreased for the three PW samples.Like the results presented in Section 3.3.1.,hot temperature had a remarkable influence to achieve faster degradation rates in a shorter period (less than 3 h).In contrast with the present results, literature mostly reports that biodegradation rates tend to decrease with increasing molecular weight of the compounds, due to limitation of the oxygen transfer to the microorganisms (Lofthus et al., 2018).In the case of biofilm, the superficial layers of lower cohesion will be detached in presence of "disturbing" compounds providing a good protection for the lower layers, which are reported to be thicker at high than low temperatures (Majid, 2019).This reveals to be a key advantage of MBBR to have a stable efficiency even in the presence of disturbing compounds such as high concentration of dispersed oil.In addition, since the biodegradation of hydrocarbon substrates is more difficult due to their low solubility, the effect of high temperature in MBBR process might have a great importance since temperature is responsible for controlling the nature and extent of microbial metabolism in hydrocarbons as well as diffusion rates, bioavailability and solubility (Tellez et al., 2005).The present results highlight the potential of MBBR to handle high concentration of non-dissolved hydrocarbons, without interfering with the process efficiency unlike other biological treatment technologies, which are more sensitive to process variations.This is shown to be relevant since dispersed oil is considered a key risk component in the PW mixture to the environment and even after treatment, the offshore PW discharged typically contains dispersed oil (normally 10-100 mg L − 1 range) (Beyer et al., 2020).For example, 95.5 tons of dispersed oil was discharged in 2020 only in Denmark (OSPAR, 2020).

Dissolved components: BTEX and VFA removal
This section outlines the potential of MBBR to remove BTEX and VFA that are ubiquitous present in PW.Figs. 3 and 4 show the removal of benzene and acetic acid, the most representative in the PW samples, during MBBR treatment for 24 h.The results for all BTEX and VFA compounds analyzed are presented in Table S1 and Table S2, respectively.The results showed the same tendency previously observed for sCOD removal, with the major concentration of benzene and acetic acid (>50%) removed within 3 h of biological treatment.After 8 h, most of these compounds were below the limit of quantification.The higher concentration of BTEX present in PW samples from upstream hydrocyclone did not seem to affect the biofilm performance.An exception occurs for PW from platform 2 collected from degasser outlet, which the presence of toxic compounds seems to have inhibited to some extent the microbial activity and causing a slower removal.Monoaromatic compounds, such as BTEX are characterized by their high volatility related to their high vapor pressure (El-Naas et al., 2014).Therefore, their removal mechanism under aerobic processes should consider both biodegradation by microorganisms and volatilization through aeration.Nonetheless, abiotic experiments (without carriers) at both 10 • C and 40 • C were carried out and only 12 and 20% of soluble COD removal was observed after 24 h (Fig. S4).This means that biological degradation was the main removal mechanism.It is important to point out that current offshore discharge regulations do not address these constituents, but BTEX are included among the most common oil-based pollutants present in water bodies and have been classified as high priority chemicals (Amakiri et al., 2022;Rahman et al., 2018;World Health Organization, 2019.For example, in 2020 a total amount of 4307 tons of BTEX was discharged in OSPAR zones (UK 56%, 41% Norway, 2% Denmark, 1% Netherlands) (OSPAR, 2020).In any case, MBBR showed to be a potential solution capable of removing these compounds within a short period, with BTEX concentration below the Summary of predicted no-effect concentrations Table 2 Rate constants (k) and correlation coefficient (R 2 ) using one phase decay equation for the offshore PW samples after 24 h of MBBR treatment.

Availability of nutrients
In general, the results presented in Table 3 show that there was not a substantial removal of ammonia (NH 3 -N) during PW treatment by MBBR.The presence of nitrite (NO 2 -N) was only observed for the MBBR batch experiment at 40 • C with effluent from platform 2. This means that nitrifying microorganisms were able to oxidize NH 3 -N to NO − 2 -N even under stressing and inhibitory conditions due to presence of toxic chemicals in that PW sample (results presented in Table 1).The effect of temperature on nutrients removal by MBBR has been previously reported with higher activity of nitrifying bacteria at 20 • C when comparing to cold temperatures (1 • C) (Hoang et al., 2014).
Even without NO 2 -N generation, a reduction of NH 3 -N after 24 h was also observed for all the MBBR experiments operated at 10 • C and 40 • C (7-20% and 27-60%, respectively).It can be assumed that the main responsible mechanisms for the removal of NH 3 -N were assimilation into the bacteria biomass that is produced.Possible minor effects also reported in literature are volatilization and/or adsorption on the biofilm due to different electric charges (Gálvez et al., 2003).
It is also relevant to point out that NH 3 -N can also be eliminated through stripping at high pH (pH > 9.5) with a strong aeration (Bassin et al., 2011;Mujtaba et al., 2017).In the present study, the reactors were not monitored for ammonia stripping, but pH did not rise enough (pH ~ 8) to support this phenomenon and the aeration was not strong enough in our treatment system.Thus, ammonia stripping can be considered negligible.
Phosphate (PO 4 -P) was not significantly removed and the concentration even increased after 24 h, mainly in the MBBR operated at 10 • C. The concentration rise of PO 4 -P in both MBBR can be explained by the release of some stored phosphates from the bacterial biomass into the reactor.It is known that some bacterial species can uptake phosphate beyond the need for balanced growth in aerobic environment, converting them into polyphosphates (Mujtaba et al., 2017).In any case, the results show that biomass was not nutrient limited, and the addition of nutrients is not required.

Toxicity assessment
Reference tests with 3,5-dichlorophenol were carried out for S. pseudocostatum and A. fischeri resulting in an EC 50 of 1.5 [0.71-2.1]and 3.4 mg L − 1 , respectively.Both values were within the expected range for the substance and test organisms according to the international testing guidelines (2006).Table S3 shows the ecotoxicity measured as amount of PW in % required to effect 50% of the endpoint measured (EC 50 ) of the PW from two different sampling points and three different platforms prior to MBBR treatment with values ranging from 9.7 to 40% (v/v) and 0.83-2.7%(v/v) for A. fischeri and S. pseudocostatum, respectively.The samples collected from upstream hydrocyclone had a consistently higher toxicity towards the two organisms than effluent (discharge point) except for platform 3 where both samples were unable to result in an EC 50 at the highest concentration tested (40%, v/v).At higher concentrations the resulting salt content would be a contributing factor to ecotoxicity and was thus omitted.There is a clear trend that marine algae is more sensitive than the bacteria independent of PW tested which is to be expected as the test using marine algae also includes chronic effects rather than the acutely toxic response of the bacteria test.
After treatment with MBBR (10 and 40 • C) for 24h it was not possible to reach 50% inhibition of the bacteria at the highest tested concentration (40%, v/v) for any of the sampling points or platforms (Table S3).Even for most toxic PW before treatment a complete  For the algae growth inhibition test full concentration response relationships with corresponding 50% effect concentrations (EC 50 ) was established for all sampling points and platforms both before and after MBBR treatment (10 and 40 • C).Both treatments, hot and cold, had a relative high ecotoxicity removal ranging between 64 and 89% (P1 un- stable > P3 unstable > P2 unstable > P3 effluent > P1 effluent > P2 effluent ) and 38-88% (P3 unstable > P2 unstable > P1 unstable > P1 effluent > P3 effluent > P2 effluent ), respectively.In terms of absolute EC 50 -values the values for cold and hot treatment ranged from 4.8 to 8.5% and 4.4-11% (v/v) respectively (Fig. 5).Consequently, the required dilution of the waste stream to reach 50% inhibition would range from 12 to 23 and 9-21 times for cold and hot treatments, respectively.Using an assessment factor of 100, the required dilution factor in order to not expect any detrimental effects would require a 2300 time dilution which is to be expected within 300-500m of the discharge point for the platforms in this study.Currently, the exclusion zone for discharges associated with offshore extraction platforms is 500 m.Hence, when applying the MBBR treatment in both hot and cold setups the associated discharge would fulfill this criteria.
The toxicity reduction is most likely related to the elimination of dissolved organic toxic compounds, such as phenols, naphthalenes (Fig. S1) and BTEX (results presented in Section 3.3.3and Table S1).The non-target analysis showed is that phenol and naphtalenes species are reduced in the samples after treatmentparticularly for the high temperature.An example from one sample, effluent from platform 2, shows that compounds like 1,2-dihydroacenaphtylene, 1,5 dimethylnaphtalene, 2,3-dimethyl phenol, 1-methylphenol, 3-ethyl-1-methylnaphtalene and 1-ethyl-2-methylbenzene to name some examples were below the detection limit in treated PW sample.Even though the present study did not reveal which compounds were the main drivers for toxicity, BTEX is reported as one of the main compounds that attribute the main toxicity of PW and classified as high priority chemicals (Beyer et al., 2020;World Health Organization (WHO), 2019).It has been previously reported that water soluble organic compounds of PW were major contributors to oxidative stress and cytotoxicity, and effects was not correlated to the content of total oil in PW (Farmen et al., 2010).While a marked toxicity reduction is observed across the samples tested, it should be noted that especially for PW simulating unstable conditions from Platform 2 a relatively high residual toxicity was found even after 24h using the 40 • C treatment (Fig. 5).From Table 1 it should be noted that this platform had the highest concentration of BTEX and could require a longer treatment for further removal of the toxicity.The residual toxicity across platforms

Conclusion
The biological treatment has proved feasible to use salt-adapted biofilm carriers to remove dissolved organic compounds and associated toxicity from offshore produced water.Produced water composition was field-dependent since the samples collected from three different platforms in the North Sea had different characteristics among them.The moving bed biofilm reactors (MBBR) had a stable total COD removal of 64 ± 5% at 10 • C and 68 ± 8% at 40 • C in a continuous flow operation for over one year.Biofilm showed to be resistant to operational parameters fluctuations such as salinity and hydraulic retention time, as well as sample variability including higher concentration of oil in water from produced water samples collected before the hydrocyclones (simulating unstable production).Batch experiments showed that most of the COD was removed within 3 h with biodegradation kinetics highly dependent on the prevailing temperature.PW high temperature (40 • C) was revealed to be a key parameter to the treatment efficiency since up to eight-fold faster kinetics was achieved, and a minimum reactor footprint is required to be compatible with offshore restrictions.In addition to the COD reduction, MBBR was able to reduce produced water toxicity since it was not possible to reach 50% inhibition of the bacteria at the highest tested concentration (40%, v/v) for any of the sampling points or platforms.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Fig. 1 .
Fig. 1.Total COD removal (left vertical axis) and PW conductivity changes (right vertical axis) in the two MBBR operated in a continuous-flow mode for 12 months at 10 • C and 40 • C.

Fig. 2 .
Fig. 2. Soluble COD during MBBR batch experiment under 10 • C (cold) and 40 • C (hot) using offshore PW from three different platforms and collected from the degasser outlet (discharge point; effluent).The first-order degradation model was fitted to experimental data.Horizontal dashed lines represent the residual COD (background level) present at the start of the batch experiment at 10 • C (blue) and 40 • C (orange).(For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Fig. 3 .
Fig. 3. Benzene during MBBR batch experiments at hot and cold temperatures with effluent and unstable PW sample from three different platforms.

Fig. 4 .
Fig. 4. Acetic acid during MBBR batch experiments at hot and cold temperatures with effluent and unstable PW sample from three different platforms.

Fig. 5 .
Fig. 5. Ecotoxicity measurements with the marine algae S. pseudocostatum expressed as EC 50 (%) for PW from three different platforms and collected from two different sampling points: degasser outlet (effluent) and upstream hydrocyclone (unstable).Error bars represent the lower and upper values.

Table 3
Nutrients analysis of offshore PW samples collected during MBBR batch experiment operating at cold (10 • C) and hot (40 • C) temperature.couldbeduetoincomplete removal of production chemicals or matrix effects caused by e.g., heavy metals (values presented in Table1, Section 3.1.OffshorePW characteristics).Operators in the North Sea region most frequently reported BTEX (34%), heavy metals (25%) and production chemicals (26%) to be the largest contributor to environmental impact, including ecotoxicological effects(OSPAR, 2013).