High-Strength, Chemical Industry Wastewater Treatment Feasibility Study for Energy Recovery

Abstract

This paper presents an experimental study on the treatment of industrial chemical wastewater with a high organic load; it is aimed at process cost optimization and possible energy and resources recovery. The facility generates five separate streams of liquid waste, which range in organic strength from practically nil to 400,000 mg/L, with individual flows ranging from 2 to 1400 m³/d. The combined strength and the flow of all the streams are approximately 1500 mg/L COD and 1500 m³/d, respectively; however, excluding the cleaner one (the cooling and condensation water), the maximum average COD concentration rises to 115,000 mg/L, at a flow of 16 m³/d. These wastes are currently trucked away for external processing, at a high cost. The aim of the study was to evaluate the feasibility of the onsite treatment of the facility’s waste streams with energy recovery and water reuse. Various approaches were examined, including anaerobic treatment for biogas recovery. The preliminary characterization, however, showed strong inhibition toward anaerobic biomass, and in particular to methanogenesis, induced by some waste components. Further testing led to the conclusion that stream segregation and separate processing could represent the most efficient approach for the management of the facility’s liquid discharges and the optimization of resources recovery. A final solution that maximizes treatment efficiency and resources reuse by combining separate aerobic and anaerobic treatment is proposed.

1. Introduction

The chemical industry is one of the main players of the manufacturing sector worldwide, with over 100,000 industrial compounds currently produced, for a direct commercial value of over EUR 4 trillion, which is about 7% of the world’s GDP. Global production of chemicals has increased 50-fold since 1950 and is set to further triple by 2050, compared to the year 2010. The European Union’s (EU) chemical industry is the second largest in the world, with 15% of the global market share and almost 4 million employees [1]. Italy is Europe’s third largest chemical producer, and the 12th worldwide (Figure 1).

The chemical production sector has a relevant environmental footprint, especially in terms of water resources, accounting for 11% of all freshwater use in the EU alone [2], and in terms of energy as it is the largest industrial energy consumer; it is also the third largest industry sector in terms of direct CO₂ emissions [3]. In addition, the chemical industry requires large amounts of raw ingredients and reagents; consequently, large volumes of wastewaters are generated that are generally characterized by a particularly high pollutant content.

Chemical industry wastewater is different from that of other common high-load industrial streams, like those of food, pulp and paper, etc.; these are well suited to biological treatment due to the presence of organic contaminants (mostly of natural origin and with a high BOD/COD ratio) which consist mainly of sugars and carbohydrates [4]. Due to the presence of complex, biodegradation-resistant natural or man-made contaminants, chemical industry wastewaters generally present a low BOD/COD ratio (an index of poor biodegradability) and may include toxic organic compounds and inorganic salts. As such, they tend to be degradation-resistant, toxic, mutagenic, teratogenic, and/or carcinogenic; hence, they are particularly challenging to treat [5,6,7] and potentially hazardous to humans and the environment. Still, they contain resources that could be recycled or reused as carbon sources or to extract bioenergy, using the proper approaches.

Various methods are available for the onsite treatment of high-strength chemical wastewaters; these can be divided into three main classes:

-

Physical technologies, such as adsorption, extraction, filtration, and evaporation: these involve the mechanical removal of pollutants (“transfer” rather than “degradation”). Physical technology purification yields vary depending on the type of effluent and will usually require additional treatment, in most cases. They usually imply high operational costs and may produce significant quantities of residuals. With the exception of evaporation technology, they are generally not applied to highly concentrated solutions [8]. Evaporation is uniquely different from other physical processes since it removes the water from the contaminants, rather than the contrary. Evaporation reduces bulk liquid waste to solid and condensate liquid fractions. As COD is not destroyed by evaporation, the residuals require further processing. As a drawback of evaporative technology is represented by its relatively high costs, it is generally used only if all other possible treatment methods have failed or are not applicable; in chemical wastewaters contaminated by solvents, it may be possible to recover and regenerate them after evaporation [9].

-

Chemical methods, such as advanced oxidation processes (AOPs) generally demonstrate better efficiency than others since they can achieve thorough contaminant degradation or even the complete mineralization of refractory compounds [10]. Fenton oxidation is among the earliest AOPs identified for the removal of refractory compounds and has been shown to effectively degrade a variety of contaminants derived from dye, textile, pesticide, and pharmaceutical processes [11]. The high effectiveness of AOPs derives from the generation of hydroxyl radicals (OH) and other strong oxidants in solution. As not all contaminants are susceptible to degradation by oxidation, advanced oxidation/reduction processes (AORPs) that generate both oxidative and reductive radicals simultaneously are being introduced [12]. Chemical wet oxidation can be effective in treating all organic high-load aqueous wastes, even toxic or refractory ones [13,14]. AOPs and wet oxidation could be highly expensive in terms of energy, reactants, and operational requirements.

-

Biological methods are adopted for the degradation of a variety of organic pollutants through large-scale wastewater treatment operations by means of naturally available microorganisms. The conventional activated sludge (CAS) process is the most common biological process for a variety of applications due to its efficiency and relative simplicity [5], despite its high energy demand and residual by-product generation [15]. Anaerobic digestion for the treatment of chemical wastewater is an appealing option due to its capacity for bioenergy recovery, its low by-product (excess sludge) yield, and its strong tolerance to high organic loading. Anaerobic processing overcomes most of the drawbacks of CAS and has become a mainstream process for the treatment of many high-strength organic wastewaters [16]. Both these methods are commonly adopted in wastewater treatments of different origins and do not generally present particular operational challenges, although some compounds, including long-chain fatty acids, phenols and alkyl phenols, alcohols, polynuclear aromatic hydrocarbons (PAH), halogenated hydrocarbons (AOX), BTEX, pesticides, and dyes, have been reported to be toxic to microbial populations when present over certain threshold concentrations [5,17]. It is usually considered that at COD concentrations above approximately 50,000 mg/L, biological treatment may have limited success, due to the possible presence of refractory components [18,19].

Figure 2 summarizes the common applicability range of the methods cited above in relation to liquid waste organic strength and specific COD removal costs, based on current available technology.

This paper reports the results of an experimental case study concerning the treatment of the high-load wastewater of one chemical industry; it is aimed at process cost optimization and possible energy recovery. The initial process strategy contemplated the adoption of anaerobic treatment of the liquid waste; however, the pilot treatability tests revealed that, under the circumstances, the segregation of streams and their separate processing could represent the best approach for the management of liquid discharges and resources recovery optimization. The final proposed solution for the facilities’ liquid waste management is illustrated and discussed.

2. Materials and Methods

2.1. Context Description

The facility originating the waste streams under study belongs to a chemical supply company that produces prime chemical ingredients for many final consumer product manufacturers in the food, cosmetic, pharmaceutical, tanning, paper, plastic, and cleaning products industries. The facility is located in an industrial district in northern Italy. Process wastewater requiring treatment can be broadly divided into five separate internal streams:

• Streams 1 and 2: effluents from ester refining processes. There are two such separate streams with different characteristics, both containing high COD loads and variable amounts of oils that may separate over time and solidify at ambient temperature. The composition of these streams may vary in time, depending on the active production cycles at each given moment.
• Stream 3: process rinsewater. This is the lighter of the four streams in terms of pollutant load. This is the waste generated from the cleaning of reactors after each production cycle.
• Stream 4: process reaction water. It is characterized by medium COD load consisting mainly of soluble alcohols’ residuals.
• Stream 5: cooling and condensation water. This is water extracted from an onsite industrial well. Due to its very mild characteristics (negligible organic content), this stream could be reused within the facility with minimal or no processing, according to required process water characteristics.

The main characteristics of streams 1–5, as derived from long-term (at least 2 years) records of onsite monitoring, are summarized in

Table 1. Main characteristics of the single wastewater streams generated in the facility.

Stream No.	Source Type	Flow	COD a Average (Range) [mg/L]	Total Susp. Solids [mg/L]	pH	Notes
1	Benzoic acid ester refining	15 ± 10 m3/week	300,000 (200,000–400,000)	7300 (±15%)	basic	Molecular composition, see Table 2
2	Non-benzoic ester refining	25 ± 7 m3/week	190,000 (130,000–250,000)	9500 (±15%)	basic	Molecular composition, see Table 3
3	Rinsewater	25 ± 5 m3/week	10,000 (8000–13,000)	400 (±25%)	basic	Process rinsewater, no added chemicals
4	Process water	50 ± 7 m3/week	80,000 (50,000–100,000)	1800 (±20%)	basic	Contains mainly soluble alcohols (≤1%)
5	Cooling and condensation	≈10,000 m3/week	negligible	negligible	neutral	Suitable for internal reuse, as is or after mild processing

^a depending on actual ongoing production processes.

.

As seen in the table, streams 1 and 2 show the highest strength (and composition) variability in time, depending on the production cycles. For this reason, no detailed analytical fractionation of the individual compounds in streams 1 and 2 was performed. However, their content can be inferred from the production processes involved, as recommended by the IPPC guidelines [19]. The CAS classification codes of the potential contaminants that could be present in these streams, according to the possible production cycles, are reported in

Table 2. Stream 1 likely components ^a.

Molecule	CAS [21]	IUPAC Name [22]
INCI: C12-15 ALKYL BENZOATE	68411-27-8	Benzoic acid, C12-15-alchil esters
BENZOATE DIPROPYLENGLYCOL	27138-31-4	Oxydipropyl dibenzoate
DIETHYLENE GLYCOL DIBENZOATE	120-55-8	Oxydiethylene dibenzoate
BENZYLBENZOATE	120-51-4	Benzyl benzoate
2-ETHYLHEXYLBENZOATE	5444-75-7	2-ethylhexyl benzoate
BENZOATE PROPYLENGLYCOL	19224-26-1	1,2-propan-diyl dibenzoate
ETHYLENGLYCOL DIBENZOATE	94-49-5	2-(benzoyloxy)ethyl benzoate

^a As inferred from the involved production processes.

and

Table 3. Stream 2 likely components ^a.

Molecule	CAS [21]	IUPAC Name [22]
INCI: CAPRYLIC/CAPRIC TRIGLYCERIDES	73398-61-5	mixed-decanoyl-and-octanoyl glycerides
2-ETHYLHEXYL STEARATE	22047-49-0	2-ethylhexyl stearate
INCI: ETHYLHEXYLPALMITATE	29806-73-3	2-ethylhexyl-palmitate
ISOBUTYLSTEARATE	646-13-9	di-isobutyl stearate
N-BUTYLSTEARATE	85408-76-0	fatty-acids, -C16-18,-butyl-esters
INCI: ISOPROPYL MYRISTATE	110-27-0	isopropyl-myristate
DIBUTYLSEBACATE	109-43-3	dibutyl-sebacate
DIOCTYLSEBACATE	122-62-3	bis(2-ethylhexyl) sebacate
INCI: CETEARYLISONONANOATE	111937-03-2	Isononanoic acid, C16-18-alkyl esters
INCI:CETEARYL ETHYLHEXANOATE	90411-68-0	Hexanoic acid, 2-ethyl-, C16-18-alkyl esters
INCI: ISOCETYL STEARATE	25339-09-7	Isohexadecyl stearate
INCI NAME:ETHYLHEXYL HYDROXYSTEARATE	29710-25-6	2-ethylhexyl 12-hydroxyoctadecanoate
INCI:ISOPROPYL PALMITATE	142-91-6	Isopropyl palmitate
INCI: ISOSTEARYL NEOPENTANOATE	58958-60-4	Propanoic acid, 2,2-dimethyl-, 2-octyldodecyl ester
INCI: DIISOPROPYL ADIPATE	6938-94-9	Diisopropyl adipate
DIOCTYLADIPATE	103-23-1	Bis(2-ethylhexyl) adipate
FORMAL BUTYLCARBITOL	143-29-3	bis(2-(2-butoxyethoxy) ethoxy)methane
BIS(2-(2-BUTOXYETHOXY)ETHYL) ADIPATE	141-17-3	bis(2-(2-butoxyethoxy)ethyl) adipate
BIS[2-[2-(2-BUTOSSIETOSSI)ETOSSI]ETIL] ADIPATE	65520-46-9	bis[2-[2-(2-butoxyethoxy)ethoxy]ethyl] adipate
ACETYL TRIBUTYL CITRATE	77-90-7	Tributyl O-acetylcitrate
2-ETHYL HEXYL ACETATE	103-09-3	2-ethylhexyl-acetate
COMPLEX BLEND OF FATS	91031-31-1	fatty acids,-C16-18,-esters-with-ethyleneglycol
FATTY ACIDS, C16-18, ESTERS WITH PENTAERYTHRITOL	85116-93-4	fatty-acids,-C16-18,-esters-with-pentaerythritol
NEOPENTYL GLYCOL HEPTANOATE	27841-04-9	biseptanoate-to-di-2,2-dimethylpropane-1,3-diyl
DIISONONILADIPATE	33703-08-1	Diisononyl adipate
DIISODECYL ADIPATE	27178-16-1	diisodecyl-adipate
DIISOTRIDECYLADIPATE	26401-35-4	diisotridecyl adipate
INCI: ISONONYL ISONONANOATE	59219-71-5	3,5,5-trimethyl ethanoate-di-3,5,5-trimethyl ethyl
2-ETHYL HEXYL ISONONANOATE	71566-49-9	2-ethylhexyl-isononanoate
INCI: C12-15 ALKYL BENZOATE	68411-27-8	-do-benzoic acid, -C19-13-alkyl-esters
INCI: DIISOSTEARYL MALATE	81230-05-9	Butanedioic acid, 2-hydroxy-, 1,4-diisooctadecyl ester
INCI-DICAPRYL ADIPATE	108-63-4	bis(1-methylheptyl) adipate
TRIETHYLCITRATE	77-94-1	tributyl-citrate

^a As inferred from the involved production processes.

. While less relevant than the quality parameter COD for the purpose of wastewater treatment, these lists suggest that both liquid streams could potentially be harmful/toxic to aquatic life, with long-lasting effects, and harmful if swallowed and could damage fertility or fetuses and induce skin and eye irritation [20].

The strength variability of streams 3 and 4 is more limited. Stream 3 may contain any of the compounds listed in Table 2 and Table 3, but at much lower concentrations. Stream 4 contains mainly soluble alcohols.

Currently, all the generated wastewater (streams 1–4 for a total of 115 ± 29 t/week, at an average combined COD load of ≈115,000 mg/L) is stored onsite and trucked away as special waste for ex situ handling. The increasing disposal fees and the generally reduced industrial profit margins due to foreign competition in the sector, however, are making this option poorly sustainable in the long run. A more efficient onsite waste management scheme, incorporating energy recovery and water reuse or direct discharge possibilities, which is aimed at reducing disposal costs, was investigated.

2.2. Preliminary Analysis and Selection of Possible Treatment Approaches

Based on the previous technological considerations, possible treatment approaches for the facility’s wastewater are summarized in

Table 4. Examined technological alternatives for waste management in the facility.

Alternative	Cost	Pros	Cons	Comments
Wet oxidation	High	Usually effective for high strength wastes. Heat energy recovery possible	Complexity of equipment and operation	Alternative ruled out due to perceived feasibility problems. Energy self-sufficiency obtained at COD load >12–15 g/L.
Evaporation	Medium to high	Volumes are reduced. Concentrate and condensate can be energetically valuable.	Residuals require onsite or offsite post-treatment (thermal valorization).	Evaporation tests were carried out on stream 2
Biological treatment	Low	Well-known processes. Anaerobic digestion can recover energy as biogas.	Application hindered by high initial waste concentration. Aerobic treatment expensive in energy terms.	Best option if combined with dilution of original waste streams to reduce possible microbial toxicity and enhance biotreatability.

. Mainstream physico-chemical treatment was immediately ruled out due to the waste streams’ strength, but evaporative treatment was considered as a possible option given the high strength of the waste streams and factoring in the possibility of achieving the solvents’ recovery. Wet oxidation was also ruled out as an option due to the perceived feasibility problems, which are mainly related to the initial costs and the complexity of the equipment and operation.

The most immediately applicable approach seemed to be the one based on biological treatment. Among state-of-the-art wastewater treatment technologies, anaerobic processes are considered the most cost-effective techniques for both domestic and industrial liquid wastes due to low energy requirements and enhanced resources recovery potential [15]. In these reactors, a series of synergistic biological processes performed by a complex microbial consortium in the absence of air, converts organics to methane and carbon dioxide, without the need for an energy-expensive oxygen supply, and reduces the scope 2 greenhouse gas (GHG) emissions from the process.

In particular, UASB (Upflow Anaerobic Sludge Blanket) technology is particularly suited to the processing of liquid wastes at varying degrees of organic load (both municipal and industrial) and environmental conditions [23]. UASB technology is based on an anaerobic biomass developed in a granular form, which is kept in dynamic suspension within a thick sludge blanket [16]. UASB-IC (internal circulation) is one of the several refinements of the original UASB technology; it is aimed at improving digestion rates and gas yields by means of controlled internal flow recirculation. This achieves enhanced biomass/substrate mixing with a smaller footprint compared to traditional UASBs [24]. At a theoretical CH₄ production of 0.35 m³/kg COD removed [25], the UASB processing of streams 1–4 could in principle yield over 600 m³_CH4/d that could be burned on site in combined heat and power (CHP) generation (both of which could be locally used) or converted into biomethane for other uses [26,27].

As anaerobic metabolism is slower than aerobic metabolism, these processes generate low excess sludge production, simplifying its handling and disposal. For these reasons, anaerobic processes have been adopted in many circumstances for the efficient treatment of organic wastewater, even in non-optimal conditions [28,29,30,31]. A considerable number of technology reviews has been published on anaerobic industrial wastewater treatment, highlighting its benefits [32,33].

In the present situation, however, given the waste streams’ strength, no type of biological treatment could be easily implemented without a substantial degree of dilution. When combined, streams 1–4 exhibited an average strength of 115,000 mg_COD/L, which is normally considered beyond the possibility of any biological process. However, when also considering the existence of stream 5, the partial dilution of the combination of the highly polluted streams 1–4 with an aliquot of stream 5 would make anaerobic biological treatment not only feasible but also attractive in view of possible energy recovery. The dilution of streams 1–4 with stream 5 at a 1:10 ratio would bring the COD load of the combined stream within approximately 10,000 to 20,000 mg/L, which is well within the biological treatment application range. The treated flow would be between 180 and 250 m³/d and thus easily handled by a compact onsite plant.

At a combined initial average strength of 115,000 mg_COD/L, the total energy content of waste streams 1–4 could be estimated to be approximately 0.45 kWh/L, or 51 MWh/week, assuming complete COD mineralization [34]. Even considering inevitable process inefficiencies and incomplete COD mineralization, a substantial amount (60–70%) of that energy could still be available for recovery.

2.3. Pilot Treatability Tests

As the objective of this study aimed at waste stream processing cost reduction and possible energy recovery, the implementation of anaerobic treatment of the facilities’ wastewater was examined as the most desirable approach. The wastewater streams’ biochemical methane potential (BMP) was ascertained to determine the feasibility of embedded chemical energy recovery through biogas production.

BMP tests are widely used to characterize a substrate’s influence on anaerobic digestion processes performance. Notwithstanding the lack of method standardization, BMP can be considered essential to evaluate a priori the suitability of a substrate to anaerobic digestion and its prospective methane yield [35,36]. Therefore, BMP assays were carried out on streams 1–4, both individually and as proportionally composite samples, at appropriate dilutions (10-fold), with pH adjustment and the addition of nutrients (3 mg urea per 100 mg COD and 6 mg P per L inoculum) for 35 days in the mesophilic temperature range (35 °C), according to the VELP respirometric procedure (UNI UNI/TS 11703) [37]. The results of the BMP assessment are summarized in Figure 3.

The most significant outcome of the BMP test is the exposition of a strong inhibitory effect on the anaerobic process effectiveness of stream 2. This is shown by a negative CH₄ yield, meaning that a “blank reactor”, i.e., a non-inoculated sample, produces more methane than one with inoculum, due to the adverse effect of its composition on methanogenic bacteria. The graph of Figure 3 also provides evidence that the inhibitory influence of stream 2 also extends to the weighted mixture of all four streams; overall, the methane generation pattern in the composite sample follows closely that observed for stream 2, although, overall, it is mildly in the positive range due to the contribution of all the other streams.

In fact, the individually tested streams 1–3 show different, but consistently positive methanation yields: stream 1 shows a significant initial methanation lag, with strong production only after 15 days, approaching the theoretical maximum (350 mL/g COD) on the 35th day; stream 4 exhibits an almost constantly increasing production, approaching 250 mL/g (less that the theoretical maximum) on the 34th day; and stream 3 presents an almost ideal trend, with sustained methane production since the first day, reaching a maximum yield of 250 mL/g.

Based on this evidence, the maximum expected methane production of the combined streams 1–4 could be estimated as close to 190 m³/d, that of stream 1 alone as approximately 230 m³/d, and that of streams 1, 3, and 4 combined as 380 m³/d.

In conclusion, the BMP assays showed that the overall methane generation capacity would be under the best conditions (streams 1, 3, and 4 combined) at about 60–70% of the theoretical expected maximum (based on the COD load). It also highlighted that stream 2, due to its composition, has a strong methanogenesis inhibition capacity, even when diluted and mixed with the others.

2.3.1. Pilot Anaerobic Treatability Tests

Following the BMP tests, actual pilot tests of anaerobic treatability were conducted in a heated UASB-IC reactor with 60 L total volume, fitted with a pH pre-buffering unit, a programmable logic controller (PLC) for online control of the process parameters, and a gas meter (Figure 4). The PLC-controlled reactor’s temperature was set at 37 ± 0.5 °C, in order to conduct the process in optimal mesophilic range conditions. As the establishment of a viable granular sludge population may require some time [38], the reactor was initially seeded with granular sludge from a nearby UASB facility treating food industry wastewater. It was then fed with a mix of stream 3 and internal sanitary wastewater in varying proportions for one month, in order to promote biomass acclimation to the substrate composition.

The pilot reactor was operated directly within the chemical facility’s grounds: streams 1–4 were fed to the reactor in different combinations and flow rates, varying from 20 to 30 L/d; the process cooling water (stream 5) was used for the dilution of these streams at a 1:10 ratio. Each test with a specific stream combination lasted 2 weeks. These UASB-IC reactor trials further demonstrated that streams 1, 3, and 4 could be successfully anaerobically treated with good COD removal and energy recovery and that a full-scale in-house UASB facility, appropriately improved, could easily meet the required effluent limits. The results are reported and discussed in detail in Section 3.1.

The tests also demonstrated that the composition of stream 2 affected anaerobic digestion as a whole, achieving limited COD degradation and having a particularly strong inhibitory role on methanogenesis, with apparent toxicity to methane-producing organisms. It was therefore confirmed beyond doubt that this stream alone was not suitable for anaerobic processing. Hence, additional options for its separate treatment were investigated.

2.3.2. Pilot Tests for Segregated Stream 2 Aerobic Treatment

Following the preliminary BMP findings and the results of the anaerobic treatability tests, an investigation on the aerobic treatability of stream 2 was implemented. For this purpose, an aerobic reactor with a net total volume of 54 L was set up. The unit, with a 44 L aerated compartment and adjacent 10 L sedimentation compartment, separated by an adjustable submerged weir, was continuously aerated by porous bottom air diffusers. The design of the unit was such that it could be operated in two modes: as a conventional activated sludge (CAS) reactor (Figure 5 left) and as an MBR reactor (Figure 5 right).

In CAS mode, the Eckenfelder-type reactor design achieves biomass recirculation from the settling to the aerobic compartment through aeration-induced mixing. This can be controlled by manually adjusting the submerged weir level. The treated effluent flowrate was regulated by an outflow weir.

In MBR mode, a microfiltration membrane unit (SICFM-00145, Cembrane A/S, Lynge, Denmark), operated by means of a permeate extraction pump, was inserted into the tank by means of a suspension frame, and the sedimentation compartment was isolated by lowering the mobile weir. A ceramic membrane (Silicon Carbide material) was chosen due to the reported characteristics of this type of medium as thermally stable, resistant to chemicals and long lifespan, which makes it ideal for the treatment of industrial wastewaters [39]. The characteristics of the membrane medium are reported in

Table 5. Characteristics of the microfiltration membrane medium (source: Cembrane technical specifications brochure).

Membrane Supplier and Model	Cembrane SICFM-00145
Material	Silicon Carbide (SiC)
Size: LxH [cm]	25 × 14
Thickness [cm]	0.6
Porosity [µm]	0.1
Permeability [L/m2·h·bar]	10,000
Max. operating temperature [°C]	80
Max. backflush pressure [bar]	3
Operating range [bar]	0–0.7

. The effluent was extracted by a pre-programmed peristaltic pump connected to the membrane unit. To optimize membrane performance and avoid continuous suction through the membrane, the pump was operated on a 1:3 min on/off continuous sequence.

The air flow from a central compressed air line was manually regulated in excess of the stoichiometric requirements, since it also provided biomass mixing, separation of the nonbiodegradable fats contained in the influent stream, and membrane surface scouring (in the MBR mode). The operating temperature was controlled by means of a programmable resistance immersed in the reactor.

The aerobic reactor was initially seeded with an inoculum of activated sludge from a municipal wastewater treatment plant, then fed with domestic wastewater at increasing aliquots of stream 2 waste for over a month. After biomass acclimation, the reactor was continuously fed with pre-diluted stream 2 wastewater by means of an automatic control peristaltic pump at a constant rate (7 L/d).

As these tests were conducted at an offsite laboratory, the pre-diluted influent substrate was prepared by mixing 1 part of stream 2 wastewater with 10 parts of tap water (instead of stream 5 water) before each test. The sample was stored in a covered, continuously stirred vessel at ambient temperature and fed to the reactor via a peristaltic pump. Each sample batch of 100 L lasted 14 days; a new batch was prepared prior to each following test. The actual strength of treated batch samples varied slightly from test to test due to variations in stream 2 composition at the time of collection and to the preparation and dilution procedures.

2.3.3. Stream 2 Evaporation Tests

Following the BMP tests outcome, evaporation tests were carried out on stream 2 as a possible processing alternative to biological aerobic treatment. Aliquots of 250 mL of undiluted stream 2 samples were processed in a laboratory vacuum evaporator with a rotary Soxlet extractor for approximately 3 h at a time, or until the collected condensate volume reached a constant level. From each original 250 mL sample volume, approximately 200 mL of condensate was collected, on average, and a complementary concentrate sample. COD and mass balance analyses of the condensate and concentrate were carried out to assess the characteristics of the two final products. The results are reported and discussed in Section 3.3.

2.4. Process Monitoring and Parameters Characterization

The process monitoring during biological treatment consisted of the periodic measurement of the process temperature, pH, and D.O. concentration (for aerobic tests only) in the reactor with a multi-parametric probe (HACH HQ40d). The measurements were carried out and registered at least every 4th day, with 3 consecutive readings within a 30 min interval, and occasionally in spot mode. The temperature was also continuously monitored by means of a digital probe inserted in the reactors.

COD was determined by spectrophotometric methods (Hanna Instruments IRIS HI801, with the appropriate kits) on the influent and effluent samples. The influent was characterized after each pre-diluted batch sample preparation; the effluent COD was tested every other day on filtered spot samples from the effluent line and at the end of each test phase on a composite sample taken from the effluent collection tank.

The MLSS in the aerobic (CAS/MBR) reactor were determined by gravimetric analysis every 4th day, according to standard methods procedures; during the CAS mode operation, the SVI was also determined to assess sludge settleability.

3. Results

3.1. Pilot UASB Process

After biomass acclimation to the new substrate, waste streams 1–4 were subject to UASB digestion tests individually and in combination, as summarized in

Table 6. Chemical wastewater pilot UASB-IC treatment tests.

Test No.	Substrate Composition a	Qin [L/d]	CODin [mg/L]	Qrecycle [L/d]	CODout [mg/L]	VLR [g/L-d]	Qbiogas [L/d]	qbiogas [L/gCOD]	ηCOD
1	1, 510	26	12,240	9.9	832	5.37	114.4	0.382	0.933
2	1, 2, 510	33	13,890	20.3	2676	7.63	37.4	0.132	0.810
3	1, 510	19	20,760	8.6	6150	6.43	8.2	0.031	0.705
4 b	1, 510, C2H3NaO2	19	19,720	12.2	4883	6.06	59.5	0.202	0.753
5	1, 510	20	18,000	20.9	652	6.08	130	0.404	0.964
6	1, 3, 4, 510	20	16,820	21.2	544	5.79	143.7	0.484	0.968
7	1, 3, 4, 510	52	22,040	45.7	623	19.2	612.3	0.554	0.972

^a The number refers to the stream numbering in Table 1. Streams 1–4 always mixed proportionally to their initial production ratios. Stream 5 pedix indicates its diluting fraction (e.g., 10 indicates 1 part of ‘other’ stream to 10 parts Stream 5). ^b Sodium acetate (C₂H₃NaO₂) added to the influent to promote biomass recovery from the inhibitory effects of stream 2.

. The digestion assays of stream 1 alone (tests 1 and 5), diluted with stream 5 aliquots, achieved COD abatements from the initial influent range of 12,000–18,000 mg/L to effluent levels of ≈650–830 mg/L. The COD removal efficiency ranged from 93.3 to 96.4%. The specific biogas yield ranged from 0.382 to 0.404 L/g_COD. The digestion of combined streams 1, 3, and 4 (also diluted with stream 5) in tests 6 and 7 achieved COD removal from the initial concentration of ≈16,800–22,000 mg/L to the effluent concentrations of ≈540–620 mg/L, with COD removal efficiency in the range of 96.8–97.4%. The biogas yield ranged from 0.484 to 0.554 L/g_COD; based on the stoichiometric mass balances, the CH₄ content of the produced biogas was estimated at between 60 and 65%.

On the other hand, a digestion test run including stream 2 (test 2) showed a sharp reduction in COD removal efficiency (down to 81%), with effluent concentrations reaching 2676 mg COD/L. The following tests 3 and 4, conducted with stream 1 only, showed that the biomass toxic inhibition induced by stream 2 persistently affected the system, with process efficiency further decreasing. In test 3, effluent COD increased to 6150 mg/L (with removal efficiency dropping to 70.5%), and biogas production reached a minimum of 0.03 L/g_COD, an order of magnitude lower than previously observed, despite the high COD concentration of the fed substrate. During test 4, sodium acetate (C₂H₃NaO₂) was added to the influent to promote biomass recovery from the toxic shock: COD removal efficiency increased to 75.3% and biogas yield recovered to 0.202 L/g_COD. Good process conditions were reinstated during test 5, as indicated by the process performance data.

It can be observed that, except for the tests involving stream 2, UASB processing achieved effluent COD concentrations that were very close to the public sewer discharge limit (500 mg/L). It is thus assumed that with some process parameter adjustment, this process could be easily implemented at full scale and achieve full compliance with the discharge requirements.

3.2. Pilot Aerobic Processes

After the preliminary biomass acclimation phase, the aerobic reactor was operated in CAS mode; after six weeks of testing, its configuration was switched to MBR mode, significantly improving COD removal efficiency. In order to further improve effluent quality, a third experimental phase was added, in which granular activated carbon (GAC) adsorption was introduced as a tertiary treatment. The sequence of experimental phases and the summary of the process efficiency are summarized in

Table 7. Summary of the experimental phases of pilot aerobic treatment of stream 2.

Test a	Flow [L/d]	CODin [mg/L]	CODout b [mg/L]	Process Temperature [°C]	Notes
Phase I: CAS operation mode
1	7 c	10,190	1210	32 ± 3	MLSS d = 4000 mg/L OLR = 0.405 mg/mg-d
2	7 c	13,270	1190	35 ± 1	MLSS d = 4100 mg/L OLR = 0.515 mg/mg-d Biomass bulking observed
3	7 c	12,340	1190	35 ± 0.5	MLSS d = 4200 mg/L OLR = 0.467 mg/mg-d Biomass bulking reduced by reduction in air insufflation
Phase II: MBR operation mode
4	7 e	11,700	850	35 ± 1	MLSS d = 5150 mgSS/L OLR = 0.361 mg/mg-d
5	7 e	9420	720	35.5 ± 0.2	MLSS d = 5300 mg/L OLR = 0.283 mg/mg-d
6	7 e	9750	760	35.5 ± 0.2	MLSS d = 4600 mg/L OLR = 0.337 mg/mg-d
7	7 e	11,150	660	35.5 ± 0.2	MLSS d = 4300 mg/L OLR = 0.412 mg/mg-d
8	7 e	10,550	570	35.5 ± 0.2	MLSS d = 4200 mg/L OLR = 0.319 mg/mg-d
Phase III: MBR with final GAC filtration operation mode
9	7 e	11,400	190	32 ± 3.0	MLSS d = 5500 mg/L OLR = 0.330 mg/mg-d
10	7 e	12,360	160	30 ± 1.0	MLSS d = 6100 mg/L OLR = 0.322 mg/mgd

^a Each test lasting 2 weeks. ^b Value in the composite effluent sample from each test. ^c Continuous flow. ^d Average during each test. ^e Influent and effluent flow regulated by a dual-channel peristaltic pump.

.

3.2.1. CAS Operation Mode

The CAS operation phase of the reactor highlighted some process performance issues: while COD removal was generally around 90%, biomass bulking was apparent, with generally poor settleability, resulting in SVI values in excess of 850 mL/L. Considerable solids loss with the effluent was observed, accumulating in the final storage tank. Although the process operating temperature was gradually increased during this phase to boost microbial activity, this had minimal effect on process efficiency. Soluble effluent COD averaged approximately 1200 mg/L (Figure 6).

Table 8. Summary of average operating conditions and performance in CAS process mode.

QIN = Qout [L/d]	7.00
pHin	8.64
CODIN [mg/L]	11,930 [10,190–13,270]
Tox [°C]	34.4 [26.9–35.8]
SVI30min [mL/L]	856 [850–900]
MLSS [mg/L]	4100.00 [4000–4200]
D.O. [mg/L]	5.4 [2.5–7]
Organic loading rate [mgCOD/mgSS·d]	0.463
pHout [-]	8.83 [8.21–9.34]
CODOUT [mg/L]	1197 [1150–1260]
ηCOD	0.898

summarizes the main process and the operational data of the CAS operational phase. Figure 7 summarizes the main process parameters observed during Phase I.

Contrary to expectations, the CAS operation did not achieve an effluent quality suitable for discharge into the public sewer. Despite the fact that no biomass was extracted from the sedimentation compartment (although accumulation was evident within), the MLSS concentration remained practically constant during this test: bulking and solid loss with the effluent were observed. The dispersed solids accumulation that was visible at the bottom of the effluent collection tank was indicative of such loss. Floating, coalescent oil and grease globes formed continuously during this phase on the surface of the aerated compartment and were skimmed out manually.

Specific bulking and foaming phenomena, accompanied by the development of specific filamentous bacteria strains and bioactivity reduction were previously correlated with the presence of industrial compounds, especially oil, fat, and grease, and the high, slowly biodegradable COD derived therefrom [40]. In such cases, settling problems, white foaming, and a thin waxy layer on the biomass surface were reported and were similar to what was observed in the collected biomass samples.

3.2.2. MBR Operation Mode

Given the modest performance of the CAS treatment, which was mainly attributed to poor growth and excessive loss of biomass, an immersed MBR process configuration was tested. The 0.1 µm Cembrane UF medium was inserted directly into the aeration tank. A membrane substitutes the CAS biomass’s gravity separation and recirculation with a physical barrier; this reduces the overall footprint of the process and provides more consistent effluent water quality, independently of the sludge characteristics [41]. MBRs can be particularly useful in the treatment of poorly biodegradable substrates since they promote biomass retention within the reactor, increasing the MLSS concentration, continuing the biomass selective acclimation to the substrate, and resulting in the general increase in organic degradation efficiency [42].

Table 9. Summary of average operating conditions and performance in MBR process mode.

QIN = Qout [L/d]	7.00
pHin	8.34
CODIN [mg/L]	10,290.00 [9420–13,500]
Tox [°C]	35.55 [30.5–36]
MLSS [mg/L]	5000 [4200–6100]
D.O. [mg/L]	6.87 [4.5–12.2]
Organic loading rate [mgCOD/mgSS·d]	0.327
pHout	8.89 [8.4–9.5]
CODOUT [mg/L]	777 [160–850]
ηCOD	0.924

summarizes the Phase II experimental results.

During MBR operating mode, both the COD removal and the biomass concentration increased compared to the CAS operation, although to a lesser degree than could initially be expected. The average COD removal increased to 92.4%, with average effluent concentrations approaching the 500 mg/L discharge limit towards the end of Phase II. The MLSS concentration increased to a maximum of just over 6000 mg/L, which was still a low value for MBR processes but was 22% higher than in the CAS mode. This indicates that, notwithstanding the long biomass acclimation (the same as from the previous CAS phase) and further selection during the MBR operation, the microorganisms still had some difficulty digesting the substrate. It seems that, regardless of the biomass selection normally occurring in the MBR systems, the characteristics of stream 2 as a substrate are still preventing the development of a highly efficient biomass for its uptake. Biomass bulking, though less relevant for a successful process operation in the MBR units, was reduced but not fully eliminated. Figure 8 and Figure 9 summarize the behavior of the MBR process during Phase II (day 1 to 70) and Phase III (starting day 71) experiments.

Microscopic biomass observations substantially confirmed the visual similarity of the active populations in CAS and MBR modes. Stream 2, despite proving to be more susceptible to aerobic than anaerobic processing, clearly manifests its bioresistant nature, inducing a clear deflocculation behavior in the active biomass. As a consequence of the high concentration of slowly biodegradable COD, a situation of nutritional imbalance occurs. Unfortunately, the typical solutions adopted in such instances when treating conventional substrates, such as a reduction of the slowly biodegradable COD fraction in the wastewater influent, are not applicable in this case. In MBR operational mode, however, the treatment efficiency was able to achieve an effluent profile close to the desired permit limitations.

3.2.3. MBR+GAC Operation Mode

Despite the higher COD removal efficiency, approaching the required effluent parameters for sewer discharge, the MBR effluent maintained the same slight yellow-brownish hue previously observed in the CAS effluent (Figure 10) due to the presence of residual, poorly degradable soluble organic matter.

In order to further reduce the effluent COD concentration and eliminate the residual color, granular activated carbon (GAC) filtration was added as a tertiary polishing unit to the MBR effluent. The filter, containing commercial grade, 1–3 mm GAC (CARBOSORB 1230, Comelt S.p.A. Milan, Italy) was a simple cylindrical vessel with a net volume of 700 mL (HRT of 2.4 h). The filter was operated during test phases 9 and 10.

The additional step completely removed the observed discoloration and reduced the effluent final COD load by up to an additional 80% compared to the MBR operation alone, to a concentration range between 80 and 200 mg/L (increasing the removal efficiency of the combined MBR+GAC system to up to over 99%). This occurred despite the decrease in the process temperature from ≈35.5 °C to approximately 30 °C, due to the failure of the immersed thermostat.

The effectiveness of GAC for natural organic matter color removal is well known [43,44]; in this case, the residual color was due to dissolved residual COD components, as highlighted by the sharp decrease in final effluent concentration after GAC adsorption.

Finally, the performance of the membrane UF module (Figure 11) should be discussed. Membranes are normally subject to progressive fouling phenomena that limit their ability to permeate clean water from a mixed liquor tank, especially when processing high fats, grease, and COD substrates [45]. Fouling reduces a membrane’s performance capacity by increasing the necessary trans-membrane pressure (TMP) to extract the same amount of liquid. Excessive fouling accumulation can lead to membrane failure due to unrecoverable permeability loss, eventually resulting in the need to substitute the filtration medium, at considerable expense [46]. Various cleaning methods, based on the addition of chemical reagents and appropriate operational procedures are applied to membrane media for fouling control [47]. Compared to the more common polymeric membranes, ceramic ones are less prone to organic fouling, particularly of the irreversible type, due to their hydrophilic lower surface roughness and lower surface negative charge properties. Studies have found that the fouling rate of silicon carbide ceramic membranes was much lower than that of polymeric ones under the same conditions [48,49].

In this study, the membrane was operated discontinuously according to ON/OFF recurring cycles to avoid the constant pressure build-up on its external wall that could favor pore clogging by sludge or suspended and colloidal particles. As the membrane medium was also exposed to a continuous flux of air bubbles with a scouring effect, the potentially obstructing particles were constantly removed from its surface. Full-scale MBR units are built so that backflushing either with clean permeate or compressed air occurs at regular intervals during operation to maintain filtration capacity. Programmed maintenance and recovery with chemical cleaning are also needed at regular intervals.

As this experimental setup was not built for automated membrane cleaning, the medium was periodically exchanged with a twin module, with the possibility to perform manual cleaning of it, if necessary. Initially, the exchanges occurred prior to each feed cycle (14 days) then, after observing that no significant performance degradation occurred, this interval was doubled (28 days). The modules were tested for permeation capacity in clean water after each substitution by monitoring the permeate flux achievable under controlled conditions. All the permeation tests showed that, in spite of the particularly aggressive composition of the substrate, the membrane modules maintained their original filtration performance throughout the study.

3.3. Stream 2 Evaporation Tests

As initially planned, stream 2 undiluted waste was subject to a vacuum evaporation treatment as a possible alternative to aerobic processing, according to the procedure described in Section 2.3.3. After evaporation, a colorless, slightly cloudy condensate residual with COD of approximately 17,000 mg/L, down from the initial 180,000, was obtained. Although there was a reduction in COD in the liquid phase, the process did not reduce its volume significantly (only ≈ 20%). This high-load residual would still require considerable post-processing, either by incineration or wet oxidation, since the treatment did not improve its biotreatability, either aerobically or anaerobically, as demonstrated by batch tests. These solutions would be expensive and would require external handling of the waste, a situation similar to the present one. The solid concentrate consisted of a whitish, sticky oily mass (Figure 12).

As evaporation does not degrade COD, the concentrate’s load can be calculated by a simple mass balance, as shown in

Table 10. Mass balance of the evaporation process.

	Daily Volume [m3]	COD Load [mg/L]	Notes
Stream 2	3.6	180,000
Condensate	2.9	16,900	BMP test on the concentrate showed results similar to those obtained from the original stream 2 tests. Vacuum evaporation did not improve condensate anaerobic treatability.
Concentrate	0.7	850,000	Since evaporation does not degrade COD, its concentration in the solid residue can be calculated as indicated

. The concentrate consists mainly of the waste’s residual, non-volatile compounds. The LCW of the concentrate was not tested but could be inferred to be approximately 14 MJ/kg [50]. The valorization of this fraction by energy recovery could be achieved by thermal incineration, which would be handled by specialized firms.

Given the observed outcome, the evaporation option was also excluded from further consideration.

4. Discussion

Cost-effective, sustainable treatment of high-load chemical industry wastewater can be a challenging endeavor, both from the technical and economic points of view, due to the variety of processes and raw materials involved in its generation and the resulting characteristics, and because of the complex processes that may be required. Proper industrial wastewater processing should aim at the cost-effective production of a harmless effluent, both for human health and the environment, depending on its final use and compatibly with local regulations [51]. Water reuse and resources (energy and materials) recovery are also important discriminants for the choice of the proper choice of treatment process sequence [52]. Selecting the most sustainable wastewater treatment technology is a very complex task because the choice must integrate and balance economic, environmental, and social criteria. Two major factors driving treatment configuration choices are legal requirements and volumes. The former must consider the limits applied by local municipal plants to allow industrial effluent discharge or internal and external reuse. The latter affect process sustainability, unit treatment costs, and discharge fees.

In this study, an investigation of the possible alternatives for the onsite treatment of an industrial facility’s wastewaters was conducted under the constraints of management ease, treatment cost containment, and energy recovery and water reuse opportunities.

Table 11. Waste streams treatment alternatives.

Waste Source	Process Option	Pros	Cons	Comments
Stream 5 (cooling and condensation water)	No treatment	Zero cost	Few. May not be suitable for direct discharge to surface waters.	This stream is suitable for direct internal or district reuse. If partly mixed with other streams could facilitate treatment.
All streams combined	Aerobic treatment (MBR)	Easy to implement	High volumes and costs (plant construction and energy). No energy recovery.	Stream 5 could be mixed in part (about 10 + 20%) with other streams (1–4), reducing total treated volume. Still, aerobic treatment of high flows implies high energy demand.
Streams 1–4 (≈1:10 dilution with stream 5)	Anaerobic treatment (UASB-IC)	UASB-IC allows high organic loads. Easy to implement, low energy demand, energy recovery	Stream 2 proved refractory to anaerobic treatment.	Combined treatment of streams 1–4 resulted in much lower COD removal efficiency of biogas production than from stream 1 alone.
Streams 1–4 (no dilution)	Onsite storage and externalized treatment	No hassles	Costly. Risks due to temporary storage safety.	This is the currently adopted solution for which an alternative is sought.
Streams 1, 3, and 4 (≈1:10 dilution with stream 5)	Anaerobic treatment (UASB-IC)	UASB-IC allows high organic loads. Easy to implement, low energy demand, energy recovery.	Achieved good COD removal, may require refinement or further polishing prior to discharge or reuse.	Achieved high COD removal and biogas production. About 380 m3/d methane recoverable from the process. Dilution ratio could be further adjusted to optimize the process.
Stream 2 (≈1:10 dilution with stream 5)	Aerobic MBR treatment	Aerobic treatment proved effective for stream 2 degradation. Relatively easy to implement.	Highly energy intensive, but relatively low volumes will not make this a major problem. MBR alone does not achieve COD removal to discharge limits.	Treatment efficiency was improved by adding a downstream effluent GAC filtration unit. Dilution ratio could be further adjusted to optimize the process. Effluents from MBR+GAC process combination are directly suitable for discharge or reuse.
Stream 2 (≈1:10 dilution with stream 5	Aerobic CAS treatment	Easy to implement, but not very effective for stream 2 degradation.	Highly energy intensive, but relatively low volumes will not make this a problem. Does not achieve COD removal to discharge limits	Treatment efficiency insufficient for discharge or reuse.
Stream 2 (no dilution)	Onsite storage and externalized treatment	No hassles. Much lower volumes involved than in present solution.	Costly. Risks due to temporary storage safety. Implies dual approach to liquid waste management.	This is the currently adopted solution for all streams. Reducing trucked volumes will reduce total costs.

summarizes the possible treatment alternatives that were examined and tested.

Among the options examined, anaerobic UASB digestion proved to have the potential to treat a considerable fraction (≈80%) of the generated wastewater volumes and to make public sewer discharge and energy recovery in the form of biogas possible. However, the presence of a single waste stream that proved refractory to anaerobic treatment hindered the initial attempt to pursue a completely anaerobic approach and required additional considerations. Upon further investigation, this stream proved susceptible to satisfactory aerobic treatment with a membrane bioreactor. After extensive testing, internal stream segregation and separate anaerobic and aerobic processing, according to the scheme depicted in Figure 13, seemed to present the optimal approach.

5. Conclusions

Possible alternatives for the onsite treatment of an industrial facility’s high-load chemical wastewater were investigated in this study under the constraints of treatment costs containment, energy recovery, and water reuse opportunity development. Part of the embedded chemical energy (as COD) in the wastewater could be recovered as methane (biogas) by separated anaerobic treatment of segregated selected waste streams, while the refractory waste fraction was tested for separated processing under aerobic conditions in the MBR process mode.

The adopted strategy allows the recovery of an estimated 380 m³/d of methane gas for internal energy use and the treatment of all streams to sewer discharge standards or for internal reuse. The proposed approach would improve industrial sustainability while reducing current waste disposal costs.