Environmental Technology & Innovation 35 (2024) 103661

Contents lists available at ScienceDirect

Environmental Technology & Innovation

journal homepage: www.elsevier.com/locate/eti

Ultrafiltration of anaerobically digested sludge centrate as key

process for a further nitrogen recovery process
Magdalena Cifuentes-Cabezas a, *, Ester Pérez-Valiente a, María-José Luján-
Facundo a, b, María-Amparo Bes-Piá a, b, Silvia Álvarez-Blanco a, b, José
Antonio Mendoza-Roca a, b
a
Institute for Industrial, Radiophysical and Environmental Safety (ISIRYM), Universitat Politècnica de Valencia, Camino de Vera s/n, Valencia
46022, Spain
b
Department of Chemical and Nuclear Engineering, Universitat Politècnica de València, Camino de Vera s/n, Valencia 46022, Spain

A R T I C L E I N F O A B S T R A C T

Keywords: In a wastewater treatment plant, the centrate generated in the dehydration process of the
Anaerobic digestion sludge centrate anaerobically digested sludge has a high potential for nutrients recovery because of its high
Ultrafiltration nutrients concentration (specifically nitrogen and phosphorous). However, the high organic
Nitrogen recovery
matter and solids content may make difficult its management. Membrane fouling is the most
significant challenge limiting the application of membrane techniques for wastewater treatment.
Results in this article highlight the potential of conventional filtration followed by ultrafiltration
as pretreatment for further nutrients recovery using emerging membrane technologies (as
membrane contactor, membrane distillation or forward osmosis). Thus, a conventional filtration
(with 1, 5 and 60 μm filters) followed by an ultrafiltration process were tested. The UF mem­
branes studied were two PES membranes (5 kDa and 0.01 μm) and one PVDF membranes
(100 kDa). Results demonstrated that a conventional filtration with a cartridge filter of 60 μm and
a subsequent UF process with a 0.01 μm RAY100 membrane (Orelis, France) achieved the best
results in terms of the maximum organic matter (66% COD removal) and solids (over 97% sus­
pended solids removal) separation, hardly varying nutrients concentration (13% NH+ 4 removal).
Concerning membrane fouling, FESEM-EDX confirmed salt precipitation on the membrane sur­
face, which has to be controlled to avoid loss of performance.

1. Introduction

Nowadays wastewater treatment plants (WWTPs) can be transformed into a source of potential resources, which has been
expressed as a “paradigm shift” (Gherghel et al., 2019; Puchongkawarin et al., 2015). In this context of circular economy, energy and
substances recovery should be enhanced (Di Costanzo et al., 2021). The application from July 2022 of the EU Regulation 2019/1009
about fertilizing products implies a harmonization in the use of digestate materials as fertilizers, enhancing the recovery of nutrients
from waste streams as components of fertilizers and biostimulants. This ensures the presence in the market of the bioproducts that can
be obtained from the sludge line of the WWTPs, which is nowadays of great importance due to the restrictions of the market caused by
the war in Ukraine (Chojnacka et al., 2023).

* Corresponding author.
E-mail address: [email protected] (M. Cifuentes-Cabezas).

https://doi.org/10.1016/j.eti.2024.103661
Received 21 March 2024; Received in revised form 25 April 2024; Accepted 5 May 2024
Available online 7 May 2024
2352-1864/© 2024 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
M. Cifuentes-Cabezas et al. Environmental Technology & Innovation 35 (2024) 103661

The recovery of nutrients from wastewater (WW) is crucial for sustaining the agricultural production in an eco-friendly manner,
since 50% of global agriculture depends on ammonia-based fertilizers. It has been reported that nitrogen and phosphorus contained in
WW would satisfy, respectively, close to 14% and 7% of the world’s demand for fertilizers (Devos et al., 2023). As Qadir et al. (2020)
pointed out that the recovery of nutrients from WW embarks on the achievement of Sustainable Development Goals (SDG) established
by the United Nations, especially SDG 12 through number 12.5 target, whose aim is a “substantial reduction in waste generation
through prevention, reduction, recycling and reuse”. This also emerges as an alternative to ammonia-based fertilizers that are obtained
through the Haber-Bosch process, which consumes a large amount of energy. This process requires H2 and N2 at high temperature and
pressure, which is responsible of 1.4% of global greenhouse gas emissions, and alone contributes to 1% of global energy demand. On
the other hand, around 72% of the H2 used comes from steam methane reforming, which requires high pressures and temperatures
(25–35 bar and 850–900, respectively), which greatly increases net greenhouse gas emissions (Clark et al., 2024; Wang et al., 2021).
As noted by Beckinghausen et al. (2020) explained in their review article different techniques to recover nitrogen from wastewater.
They remarked that nitrogen recovery is the next step to improve the wastewater treatment process. Regarding urban WW process, it is
known that nitrogen and phosphorous are released from the bacterial cells during the anaerobic digestion of the sludge (Cruz et al.,
2019). Thus, the liquid stream separated from the sludge in the subsequent dewatering stage (sludge dewatering liquor) will be rich in
nutrients. This liquid, known as centrate, since centrifugation is the most common process for sludge dewatering, can be treated in
order to recover nutrients in a concentrated stream that could be used in agriculture (Lubensky et al., 2019). Although some authors
reported about phosphorous recovery as struvite at the end of the 20th century (Williams, 1999), many papers in the first decade of this
century were focused on a separated treatment of the anaerobic digestion sludge centrate (ADSC) for nitrogen and phosphorous
elimination (Fux et al., 2006; Guo et al., 2010). Thus, it was avoided their recirculation to the WWTP entrance, which implies, in the
case on nitrogen, an increase in the oxygen demand in the biological reactor. Additionally, the significant contribution of ADSC
(recirculated to the WWTP entrance) to the total ammonium nitrogen and phosphorus entering the biological reactor (around 25% and
8%, respectively) is also important to be considered (Soler-Cabezas et al., 2018).
It is well-known that phosphorous can be recovered from ADSC by precipitation as struvite. The scarcity of phosphorous natural
resources and the gradual change to a policy prioritizing recycling over elimination led to a deep research on this topic in the last two
decades (Pastor et al., 2008). Struvite is magnesium ammonium phosphate (MgNH4PO4⋅6 H2O), which means that a part of the
ammonium-nitrogen in ADSC is also separated. However, the viability of this process is limited to WWTPs with enhanced biologically
phosphorous removal. In other cases, phosphate concentration in ADSC is not high enough to make the recovery feasible from an
economic point of view, since addition of magnesium and pH adjustment is needed for struvite crystallization (Sichler et al., 2022).
In the last years, the use of membrane contactors for nitrogen separation from ADSC has been studied by some researchers. In this
process, a porous hydrophobic membrane which acts as a selective barrier between the ADSC (after rising the pH to convert
ammonium into ammonia) and an acidic solution (stripping solution). Ammonia is transferred through the membrane from the ADSC
to the acidic solution (Darestani et al., 2017; Noriega-Hevia et al., 2020; Vecino et al., 2019).
Other strategy to be considered, is the concentration of the ADSC. Unlike the process commented above, a particular nutrient is not
separated but it is obtained a concentrated stream rich in nutrients, which could be used for fertirrigation or even as a fertilizer if the
needed concentrations are reached. For this purpose, some authors have applied forward osmosis (FO). This process is based on water
permeation through a membrane due to the difference in osmotic pressure between both sides of the membrane. Soler-Cabezas et al.
(2018) proposed the use of FO for ADSC concentration using as draw solution wastewater with a high ammonium sulfate concen­
tration. Therefore, the phenomenon of salt reverse flux (in this case ammonium reverse flux) was used to enhance the concentration of
this nutrient. Vu et al. (2019) used seawater as draw solution for the same purpose. Other process for the concentration of the ADSC is
reverse osmosis (RO) (Munasinghe-Arachchige and Nirmalakhandan, 2020), which has the advantage over FO of being a well-known
membrane technology with availability of many commercial membranes. However, the energy costs are higher due to the high
transmembrane pressure required.
Summarizing, membrane technologies (membrane contactors, FO, RO) are promising techniques for nutrients recovery from the
ADSC. However, membrane fouling is very severe if ADSC is fed directly to the membranes due to the high suspended solids and
organic matter concentration.
The process of recovering nutrients from any wastewater consists of three stages. The first is the preprocessing of the wastewater,
followed by further concentration and ending with recycling of the concentrate. According to Qin et al., (2023) regarding nitrogen
recovery, most studies focus on the last two stages, leaving aside the treatment, a fundamental stage for the effectiveness, stability, and
economic viability of the entire process. On the other hand, although membrane processes, such as membrane contactors, can be
extremely efficient, consideration must be given to the pretreatment for reducing the concentration of contaminants that could foul the
contactor. As noted above, membrane technologies are sensitive to contaminants (especially solids or other materials that can cause
fouling) and, as noted by other authors, working conditions need to be carefully chosen and optimized, as well as the process combined
with pretreatments if it is necessary. There is a lack of literature on fouling of gas-permeable membranes used for ammonia recovery, as
reported by some authors (Munasinghe-Arachchige et al., 2021).
For instance, Suleman et al. (2022) observed a high flux decline caused by ADSC in a FO process. In the same way, some researchers
also observed the need of a pretreatment for membrane contactors for nitrogen recovery (Aguilar-Moreno et al., 2022;
Cifuentes-Cabezas et al., 2023b; Reig et al., 2022) and it is well known that RO membranes have to be fed with streams meeting
turbidity values lower than 1 NTU. Thus, in order to minimize membrane fouling in any of the above-mentioned processes, ultrafil­
tration (UF) could be the appropriate technology, being the key process for further nutrients recovery or concentration.
The aim of the UF process is to provide the needed quality for the subsequent membrane process with the minimum loss of nu­
trients. However, the control of the UF membrane fouling will be a key factor for the feasibility of the process. UF as a pretreatment of

2
M. Cifuentes-Cabezas et al. Environmental Technology & Innovation 35 (2024) 103661

another membrane process is commonly used for the treatment of secondary effluents from municipal WWTPs (Yasar et al., 2022),
seawater desalination (Brover et al., 2022) and for reclamation of industrial effluents (Carbonell-Alcaina et al., 2018). However, to the
best of our knowledge there are no papers focused on UF as a first step to further nutrients separation from the ADSC in WWTP.
However, in similar fields of study, UF has presented good results as a pretreatment, because it is capable of retaining suspended solids,
colloids, emulsions, bacteria and viruses (Deemter et al., 2022). In a work conducted by Zacharof et al. (2019), anaerobically digested
sludge was subjected to sedimentation, dilution with tap water and acidification, evaluating the removal of soluble components from
sludge using a UF diafiltration (DF) process. The study involved different UF sequences, which consisted of first concentrating and then
diluting the sludge with tap water using a 500 kDa PVDF membrane. With this process, the separation between phosphate and
ammonium was achieved, resulting in a permeate rich in ammonium, but limited in phosphate, which was the objective of the
research. A more recently work studied the effect of different pretreatment strategies (centrifugation alone or in combination with
coagulation with polyaluminum chloride, biocoagulation with chitosan and flocculation with polyacrylamide) of sugar beet pulp
digestate on fouling mitigation of a UF membrane (PVDF, 150 kDa). With a pretreatment consisting of centrifugation followed by
flocculation with polyacrylamide, a high yield of the UF process was achieved. In particular, the removal of chemical oxygen demand
(COD) and total solids were of 89 and 80%, respectively, maintaining and a high content of ammonium nitrogen in the permeate
(Chuda and Ziemiński, 2023).
Summarizing, UF has been hardly studied for the treatment of ADSCs from WWTP as a first step for nutrient recovery, and only a
few papers deal with other types of ADSCs. As mentioned above, fouling of the UF membrane is expected. Therefore, ADSC pre­
treatment and membrane fouling are key factors to be assessed in view of a future implementation at industrial scale. Thus, in this
study, the removal of organic matter using conventional cartridge filters was firstly considered. Three cartridge filters (with pore sizes
of 1, 5 and 60 μm) were evaluated and the selection was based on the efficiency of COD and solids removal. Then, three UF membranes
made of different materials (two from PES and one from PVDF) were evaluated in total recirculation mode under different trans­
membrane pressures (TMP). Two membranes (one of each material), which presented the best results in terms of permeate flux, COD
and nutrient removal, were studied in depth in batch operating mode.

2. Materials and methods

2.1. ADSC samples

The wastewater used in this work came from a municipal WWTPs located in Comunidad Valenciana (Spain) and corresponds to
anaerobic digestion sludge centrate taken from the centrifuge outlet pipe. Once the samples were received, they were stored in the
refrigerator at about 5ᵒC until the characterization analyses were carried out, so that their properties were preserved. ADSC samples
were characterized before and after the pretreatment and UF process. Each sample was characterized in terms of pH, conductivity,
2+
turbidity, total nitrogen (Ntotal), ammonium (NH+ 4 ), total phosphorous (Ptotal), soluble and COD, potassium (K ), calcium (Ca ),
+
2+
magnesium (Mg ), alkalinity, suspended solids (SS) and total solids (TS) content.

Fig. 1. Scheme of filtration tests carried out.

3
M. Cifuentes-Cabezas et al. Environmental Technology & Innovation 35 (2024) 103661

2.2. Cartridge filter selection for pretreatment

As can be seen in Fig. 1, in a first stage, a conventional filtration study was carried out to eliminate the suspended solids contained
in the sample in order to protect the ultrafiltration membranes subsequently tested. Three polypropylene filters of different microns
(1 µm, 5 µm and 60 µm) were tested. The same company (Aquatronica, Italy) supplied all the filters. Filtration was performed by height
difference (potential energy). The resulting filtrated stream was characterized in terms of COD, SS, TS, Ntotal, and Ptotal, using the same
methodology as that used to characterize the raw ADSC samples. Filter selection was determined based on organic matter and solids
removal efficiency.

2.3. Ultrafiltration tests

2.3.1. Ultrafiltration at constant concentration

Ultrafiltration experiments were carried out in a laboratory-scale membrane plant. The feed stream was pumped from feed tank
(with a capacity of 8 liters) to a Rayflow membrane module (Orelis, France), which contained a flat sheet membrane (0.0125 m2 total
active surface area). Permeate flux was measured using a precision balance (0.01 g readability) (Kern, Germany) and the collected data
were recorded using a data acquisition system. The feed tank was provided with a refrigeration system to maintain the temperature at
25 ◦C. Three flat sheet membranes and were tested; the characteristics are presented in Table 1. Both the filters and UF membranes
were selected based on bibliography (Judd and Judd, 2011; Munasinghe-Arachchige et al., 2021; Noriega-Hevia et al., 2020), as well
as studies previously carried out by us (Cifuentes-Cabezas et al., 2023b).
All the membranes were initially compacted before characterizing them by determining their hydraulic permeability (K); which
was obtained with the slope of the straight line obtained by plotting the hydraulic permeate flux (obtained with osmotized water with
conductivity < 40 μS/cm as feed) versus the applied TMP, following the theoretical basis of Darcy’s law (Eq. 1):
Δp
J = K • ΔP = (1)
μ • Rm

Where J represents the water permeate flux at a specific TMP (ΔP), µ the viscosity of the permeate and Rm the intrinsic membrane
resistance.
After the characterization of the membranes, a test was carried out at constant concentration mode (recirculation of the concentrate
and permeate streams back to the feed tank) to study the influence of TMP on each membrane performance. The working conditions
used for the UF tests at total recirculation mode were a fixed cross flow velocity (CFV) of 2 m/s, varying the TMP between 1 and 2 bar.
Due to limitations, both in the laboratory plant and the instability of the pressure, tests were only carried out at 1 and 1.5 bar with the
PVDF membrane. The performance of the membranes was evaluated from the permeate flux and the rejection of COD, Ntotal, Ptotal, K+
and NH+ 4 , which was calculated using the following equation (Eq. 2):

CF − CP
Rejection(%) = • 100 (2)
CF

Where CF and CP are the concentration of the different parameters in the feed and in the permeate, respectively. On the other hand, for
a better evaluation of the results, a ratio (r) between the concentration of the different parameters in the permeate and feed streams
was also considered (Eq. 3).
CP
r= (3)
CF

2.3.2. Ultrafiltration at variable concentration

With the selected PES and PVDF membrane (Fig. 1), tests with variable concentration (concentration mode, recirculation only of
the concentrate stream) were performed to observe the influence of the different membrane materials. The test were performed at a
fixed CFV of 2 m/s and TMP of 1.5 bar (best TMP condition of test at constant concentration). In this case, a final conversion factor (ɳ)
of 75% was considered for the tests. It was calculated by the ratio between the obtained volume of permeate (VP) and feed volume (VF),
as Eq. 4 shows.

Table 1
Membrane Characteristics provided by manufacturer.
RAY100 UP005 FORM06

Manufacturer Orelis Microdyn Nadir Orelis

Configuration Flat sheet Flat sheet Flat sheet
Material PES PES PVDF
MWCO 0.01 µm 5 kDa 100 kDa
Area (m2) 0.01 0.01 0.01

4
M. Cifuentes-Cabezas et al. Environmental Technology & Innovation 35 (2024) 103661

VP
ɳ(%) = • 100 (4)
VF
The idea was to study the evolution of the permeate flux under more aggressive conditions, as well as the characteristics of the
permeate and concentrate streams obtained, mimicking the operation at a higher scale.
Finally, the surface of the selected membrane was characterized after the experiments at concentration mode. Field emission
scanning electron microscope (FESEM) and energy dispersive X-ray spectroscopy (EDX) were used to analyze the morphology and
composition of the fouled membranes. The equipment used for the characterizations was a ZEISS Ultra 55 FESEM.

2.3.3. Membrane cleaning

After the tests, the membranes were cleaned to restore the initial characteristics of the membrane. For this, a first cleaning was
applied with osmotized water at 1 bar and a CFV of 2.2 m/s. After 30 minutes, permeate flux was measured and compared with the
permeate flux determined for the pristine membrane. If the initial value was still not reached, a second cleaning was carried out. The
second cleaning consisted of a chemical cleaning using P3-Ultrasil 110 (Ecolab, Spain) 1% (v/v) with the aim to recover (at least) 90%
of the initial flux. The operating conditions were the same, 1 bar and a CFV of 2.2 m/s during 30 minutes. Then, a rinsing with
osmotized water was performed for 10 minutes to remove possible chemical residues on the membrane and in the lab-plant.

2.4. Analytical characterizations

Feed, concentrate and permeate streams from each test were analyzed. pH and conductivity were measured using a pH-Meter GLP
21+ and EC-Meter GLP 31+ (Crison, Spain), respectively. Turbidity was analyzed by means of a turbidimeter model TL 2310 (Hach,
3- 2+ 2+
Spain). Ntotal, NH+
4 , Ptotal, PO4 , Ca , Mg , K and COD were measured using kits from Merck and following the experimental
+

protocol described by the manufacturer. The total alkalinity (TA) was measured following the method 403 cited in the Standard
Methods (APHA, 2005); and TS, volatile total solids (VTS) and SS were measured also according to Standard Methods (APHA, 2005),
taking a sample of 50 mL for each one. All the measurements were duplicated.

3. Results and discussion

3.1. ADSC characterization and cartridge filter selection

Table 2 presents the characterization of three ADSC samples used in the study (mean value and standard deviation are shown),
taken on different days. The high content of organic matter is manifested by total COD (close to 1 g/L) and total solids (greater than
3 g/L). Although these values agree with those expected in ADSC (Holloway et al., 2007), they show a large variation among samples.
This was also observed by Soler-Cabezas et al. (2018), who adjudicated the variability of SS to the efficiency of the centrifugation
process in the WWTP, which could be related to the high variability of the total COD values. On the other hand, these authors also
observed a large variability in phosphorus content, which they attributed to the spontaneous precipitation of Ca3(PO4)2 and struvite
that could occur at the outlet of the anaerobic digester. Finally, the high values of organic matter and TS support the need for a
pretreatment prior to the UF process for the care of the membranes. It has been reported that wastewater with so large organic matter
content, especially in the reuse of municipal wastewater, cause remarkable fouling and biofouling, significantly affecting the per­
formance of the membrane (Tow et al., 2022). On the other hand, salts not only affect the membrane due to a possible precipitation on
its surface, but also favor the aggregation of organic (Zhang et al., 2020).
Regarding the presence of possible nutrients to be recovered, it is observed that the concentration of NH+ 4 was the highest, in
comparison to those of PO3- 2+
4 , K and Mg . The concentration of NH4 is in the typical concentration range (from 250 – 1500 mg/L)
+ +

mentioned by others authors (Lubensky et al., 2019; Thornton et al., 2007; Uysal et al., 2010).
Fig. 2 shows the reductions achieved for each parameter according to the filter used. As mentioned, the objective of this stage was to
achieve a high reduction of both SS and organic matter without reducing the nitrogen content in order to recover it in subsequent
stages with membrane techniques. As expected, conductivity and pH did not present changes after filtration.
Comparing the values obtained from the filtered samples with those of the unfiltered sample, it can be observed that SS was the

Table 2
Complete characterization of ADSC samples (average values and standard deviation from three samples).
Parameter Value Parameter Value

pH 7.68 ± 0.25 Ntotal (mg/L) 517.0 ± 71.36

Conductivity (mS/cm) 7.15 ± 1.38 Nsoluble (mg/L) 429.50 ± 84.21
Turbidity (NTU) 166.15 ± 11.85 NH+ 4 (mg/L) 472.75 ± 94.50
CODtotal (mg/L) 988.5 ± 715.5 Ptotal (mg/L) 119.75 ± 46.94
CODsoluble (mg/L) 414.2 ± 201.0 PO3-
4 (mg/L) 99.88 ± 50.59
TA (mg/L) 3062.5 ± 784.6 K+ (mg/L) 178.50 ± 44.13
TS (mg/L) 3138.7 ± 576.2 Mg2+ (mg/L) 85.15 ± 42.93
VTS (mg/L) 681.7 ± 282.1 Ca2+ (mg/L) 168.75 ± 42.31
SS (mg/L) 368.5 ± 341.7

5
M. Cifuentes-Cabezas et al. Environmental Technology & Innovation 35 (2024) 103661

Fig. 2. Removal percentages for the samples filtered by means of the different filters considered.

parameter with the highest removal percentage. By contrast, there is hardly any difference in terms of N, as expected, due to its very
small size. Comparing the diverse filters used, it is observed that the separation of nutrients did not show great differences. In the case
of Ptotal, a reduction over 50% for all the filters was reached. In the case of TS, no differences were observed among the different filters
tested. The high removal of Ptotal was expected, since, according to the bibliography (Camilleri-Rumbau et al., 2015), it is estimated
that 30% of Ptotal is mainly attached to particles with a diameter greater than 10 μm. In this case, the similar removal percentages of
Ptotal observed for the different filters indicate that the majority of this compound would be bound to particles larger than 60 μm. In
fact, it was determined that 39.84% of Ptotal was not soluble. On the other hand, precipitation of calcium phosphate in the filter would
have contributed to the high separation of Ptotal by the filters.
Curiously, the 5 µm filtrate presents practically the same reduction of SS as the 60 µm filtrate, which may be due to the fact that the
solids contained in the sample are mostly larger than 60 µm. Regarding COD, it is observed that the filter that achieves the highest
removal is the tightest one. For turbidity, a clear relation between separation and filter aperture is observed, with the highest reduction

Fig. 3. Evolution of permeate flux with time for constant concentration tests using anaerobic digestion sludge centrate (ADSC) as feed at different
transmembrane pressures. A: RAY100; B: UP005; C: FORM06 (21 ± 1 ◦ C).

6
M. Cifuentes-Cabezas et al. Environmental Technology & Innovation 35 (2024) 103661

reached by the filter with the smallest pore size.

The results for the 1 µm filter indicate that it is a filter with a behavior similar to a microfiltration stage, with high reductions in
suspended solids, turbidity and COD. However, with this filter, greater fouling was observed, having to stop the system to perform
more frequent cleanings. Similar problems were exhibited by the 5 µm filter. Although the clogging was lower than that observed for
the closest filter, it was still necessary to carry out frequent cleanings. On the other hand, the results obtained with the 5 and 60 µm
filters were relatively similar in terms of suspended solids and nutrients removal. Thus, 60 µm filtration was selected as pretreatment
for the ADSC. It is important to note that the filters were recovered only with pressurized water cleaning and a low concentration of
bleach (< 0.1%).
Regarding the comparison of the results obtained with other pretreatment proposals, Zacharof et al. (2019) proposed a discon­
tinuous process to separate and fractionate phosphate and ammonia from digestate of agricultural origin. The process included a
pretreatment, followed by a DF system with UF and NF, consisting of concentration and dilution stages (with tap water). The pre­
treatment consisted of an acidification stage, followed by settling, dilution and mixing stages, and, prior to NF, coarse filters varying in
pore size between 1.045 mm to 0.5 mm. The pretreatment achieved a reduction of 46.7% of SS and 15% of color. However, the process
required a high footprint and needed to add a large amount of water to perform the DF process.

3.2. Constant concentration tests

First, the permeability of the membranes was measured. As expected, the membrane with the largest pore size showed the greatest
water permeability, however the UP005 PES membrane with a MWCO of 5 kDa presented higher hydraulic permeability than the
FORM06 membrane with 100 kDa of MWCO (53.05 ± 1.50 L⋅h− 1⋅m− 2⋅bar− 1, 22.74 ± 0.47 L⋅h− 1⋅m− 2⋅bar− 1 and
17.10 L⋅h− 1⋅m− 2⋅bar− 1 for RAY100, UP005 and FORM06 membranes, respectively). Then, the constant concentration tests were
performed in order to select the best two membranes. Fig. 3 shows the variation of the permeate flux with time for all the membranes at
the different TMP tested.
A very different behavior can be observed between the membranes. At the lowest TMP, all membranes presented a stable permeate
flux. Then, as expected, an increase in TMP generated an increase in permeate flux. However, for the RAY100, the increase from 1.5 to
2 bar did not generate any change in the permeate flux. Moreover, for this membrane, it can be observed that, at the highest TMP
tested, permeate flux presented higher decay, decreasing notably during the first 30–40 minutes of the test, from 80 L⋅h− 1⋅m− 2 to
60 L⋅h− 1⋅m− 2. Finally, a steady permeate flux close to 55 L⋅h− 1⋅m− 2 was reached. This behavior can be due to several reasons: at low
TMP, as the driving force is small, the transport of solute molecules towards the membrane surface is less intense. Therefore, severe
fouling is not expected, resulting in a permeate flux more stable during the test. However, when the TMP was raised, greater fouling
was observed due to the greater increase in the concentration at the membrane surface because of concentration polarization. As a
consequence, flux decline in the first minutes of operation was observed due to pore clogging until the gel layer was formed. This
pronounced fouling caused that an increase in the TMP did not produce any increase in the permeate flux, what implies that limiting
flux conditions were reached. The critical flux is defined as the flux that leads to a first deviation from the linearity of the flow with
transmembrane pressure, being a criterion for the transition between concentration polarization and fouling (Giacobbo et al., 2018). It
has been reported that an increase in TMP generates an increase in the total hydraulic resistance, due to the accumulation of larger
amounts of solutes on the surface of the membrane and, therefore, causing the formation of a gel layer (Zielińska et al., 2020). This
little difference in membrane fluxes with increasing TMP was also observed by Yue et al. (2021) at testing four PES membranes with
different MWCO (50, 20, 10 and 5 kDa). Although these authors used an anaerobic digestate from swine manure as feed, they reported
that it could be due to concentration polarization phenomenon, which is more intense at higher TMPs. In their study, the digestate
contained a large concentration of salts (conductivity = 7.7 ± 0.05 mS/cm), similar to the value measured in the ADSC in this study
(conductivity = 7.15 ± 1.38 mS/cm). This concentration polarization would lead to the formation of a more selective gel layer on the
membrane surface, thereby influencing the permeate flux. Once a gel layer is formed, an increase in TMP does not lead to an increase in
flux, but instead the thickness of the gel layer increases (Cifuentes-Cabezas et al., 2021). On the other hand, this membrane is, within
the PES membrane, the one with the highest MWCO. A study on membrane contamination by dissolved organic matter revealed that
membrane contamination with the largest pores was the most severe due to pore blocking (Lin et al., 2014). This could also be
responsible for a decrease in permeate flux over time.
The opposite behavior was exhibited by the UP005 membrane, presenting an almost constant permeate flux from the beginning of
the test, for each operating TMP. With this membrane, an increase in permeate flux was observed with increasing TMP. The con­
centration polarization is more intense the greater the permeate flux. Therefore, membranes with a larger pore size (RAY100) have
higher concentration values at the membrane surface and are therefore more likely to form a gel layer than in membranes with a pore
size smaller. Thus, for the membrane with a smaller pore size, such as UP005, the variation of permeate flux with TMP presents a linear
relationship. In the case of the UP005 membrane, significant fouling was not appreciated for any of the TMPs tested, with a stable
permeate flux during time, unlike the RAY100 membrane. This means that the membrane with the smallest pore size suffered less
fouling, and may be due to the size of the particles present in the ADSC. Organic matter in the ADSC includes cellular debris from the
digestion process. Zheng et al. (2009) studied the treatment of secondary effluents with UF, which also contained proteins and car­
bohydrates from cellular debris. They concluded that, although all fractions (particles, large colloids, and dissolved organics) lead to
membrane fouling, more than 50% of the total fouling resistance was caused by dissolved substances within the 0.45–0.026 µm
fraction. As the pore diameter of the RAY100 membrane (0.01 µm) is closer to this fraction than that of the UP005 membrane, when
pressure increases, the particles could collide and break, being more probable the clogging of this membrane. This could confirm the
drop in permeate flux observed for the RAY100 membrane in the first minutes of operation. It is important to note that, although the

7
M. Cifuentes-Cabezas et al. Environmental Technology & Innovation 35 (2024) 103661

RAY100 membrane presented a greater decrease in permeate flux, it managed to stabilize, so it did not suffer severe fouling, specially
at the lowest TMP tested.
Regarding the PVDF membrane testes, in general terms, the FORM6 membrane presented a behavior more similar to the UP005 PES
membranes, with a more stable flux and a tendency of increase the value of permeate flux as the TMP raised. At a TMP of 1 bar, it
begins with a permeate flux of 19.68 L⋅h− 1⋅m− 2, decreasing rapidly in the first 25 minutes, to then reach a steady permeate flux of
12.28 L⋅h− 1⋅m− 2, corresponding to a 37.6% decrease in flux. Then, by increasing the TMP to 1.5 bar, the membrane manages to reach
equilibrium after 15 minutes, decreasing the permeate flux by 35.5%, similar than at 1 bar. Although it was not possible to work at
higher TMP with this membrane due to experimental problems, they did not present significant differences in terms of fouling
compared to the TMP of 1 bar, such as the RAY100 membrane.
Regarding the characteristics of the permeate obtained after the different tests carried out, Fig. 4 shows the results in terms of
rejection percentages. It can be observed that Ntotal rejection exhibited by the membranes was similar and did not significantly change
with pressure, fluctuating between 6% and 14%. Another important parameter is the rejection of COD, which also presented quite
similar results for these membranes, varying for all the tests between 57% and 68%. As commented in the previous section, a large
removal of Ptotal (between 28% and 41%) was expected in comparison with other nutrients, such as K+, Ntotal and NH+ 4 (which are
mostly in dissolved form), since, according to Camilleri-Rumbau et al. (2015), 20% of P is mostly attached to particles with a large
diameter (> 0.45 µm), being mostly retained by microfiltration and ultrafiltration membranes.
To determine the two best membranes to continue the study, two factors were considered: i) high values of permeate flux and ii) the
appropriate characteristics in the permeate stream. Taking these factors into account and observing that the membranes presented
similar rejection percentages, the RAY100 membrane made of PES material and the FORM06 PVDF membrane were selected. The
RAY100 membrane was the one with the best performance and considering that UP005 and FORM06 presented similar performance,
the latter was selected to have the material as a more considered variable in the variable concentration tests. Both presented a low and
similar reduction in NH+ 4 content, of 10.7% and 10.2%, but a high COD rejection, of 62.5% and 60.2%, respectively for PES and PVDF
membrane. These results were presented under conditions of 1.5 bar of TMP, with permeate flow values of 53.25 ± 2.26 L⋅h− 1⋅m− 2
and 23.15 ± 0.6 L⋅h− 1⋅m− 2, respectively, for RAY100 and FORM06 membranes (Fig. 3a, Fig. 3c).
Although there are no ADSC studies with UF, other studies propose a pretreatment of ADSC using pH adjustment, sedimentation
and conventional filtration (0.45 µm), without a subsequent UF step. Although they achieved a similar loss of NH+ 4 (close to 15%), they
reached lower removal of COD than in this study (53% vs. 64.3%) (Noriega-Hevia et al., 2020). This is important since, as these authors
stated, the main bottleneck in membrane contactors or forward osmosis processes, for NH+ 4 recovery, is membrane fouling.
An example of using UF as a pretreatment is the integrated N-Free system presented by Ledda et al. (2013), to treat a concentrated
cattle digestion of manure. They managed to concentrate ammonium by 72% with the system, which includes (among other stages)
ultrafiltration (TMP between 3.5 and 4.5 bar) and reverse osmosis.
Thus, UF seems to be needed as pretreatment, as our study also suggests. Our work is aimed to find the appropriate UF conditions
for the elimination mainly of organic matter and SS, without affecting the concentration of potential nutrients to be recovered. As
Munasinghe-Arachchige et al. (2020) rightly comment, the gradual accumulation of fouling on the membrane can increase energy
consumption, maintenance cost and reduce the useful life of the membrane, making the process unviable for continuous operation.
Therefore, the selection of the appropriate membranes and operating conditions is crucial.

Fig. 4. Rejection of the different parameters for the ultrafiltration membranes in the constant concentration tests at the different transmembrane
pressure tested.

8
M. Cifuentes-Cabezas et al. Environmental Technology & Innovation 35 (2024) 103661

3.3. Variable concentration tests

3.3.1. Permeate flux variation

Once the optimal membranes were selected (RAY 100 and FOMR06 membrane), it was decided to study the impact of the
membrane material on permeate flux and solute rejection. Studies carried out by other authors indicate that the 0.01 μm RAY100 PES
membrane can be considered with MWCO of 100 kDa. Since they demonstrated that the RAY100 membrane retains almost all proteins
from 100 to 1000 kDa and allows the passage of those from 10 to 100 kDa (Li et al., 2015). Therefore, RAY100 and FORM06 mem­
branes with similar MWCO could be considered. The comparison between these two membranes is important because PVDF ultra­
filtration membranes are the dominating commercial membranes in wastewater treatment processes. PES membranes usually present
higher permeate flux than PVDF membranes due to their hydrophilic character. However, a higher flux through the membrane leads to
greater concentration of solutes at the membrane surface, which can cause higher fouling and the formation of a gel layer, resulting in
the need for more intense cleaning protocols and therefore a shorter useful life (Chen et al., 2021). But it is also important to take into
account that the more hydrophobic character of the PVDF membrane is generally more susceptible to fouling compared to more
hydrophilic membranes (Susanto and Ulbricht, 2005).
Both membranes were compared under tests in concentration mode with the ADSC, until a conversion of 75% was reached (starting
with an initial feed volume of 10 liters).
In Fig. 5, the normalized permeate flux is plotted over time for both membranes until the desired conversion (75%) was reached. A
first decrease in flux due to membrane fouling was observed for both cases. Then, as time increased a subsequent stabilization was
noted. This curve is typical of UF processes and is due to the concentration polarization phenomenon, followed by a smooth stabili­
zation or decay of the flux until the end of the process. Specifically, in the first hour there was a flux reduction of 20% and 40% for the
PES and PVDF membranes, respectively, being more pronounced for the PVDF membrane (FORM06). Afterwards, the PES membrane
maintained a stable flux until 6 hours of testing, to then decrease by approximately 8% in the subsequent two hours. Then, it remained
stable until 15 hours, when a constant decrease started. At the end of the test, permeate flux was 17% smaller than the stable flux
reached in the first hours of testing. In the case of the PVDF membrane, after the sharp drop at the begging of the run, flux slowly
decreased throughout the test, resulting in a 26% permeate flux loss at the end of the test (in comparison with the permeate flux
observed after the first severe decay). This indicates a greater impact of fouling for the PVDF membrane, as previously mentioned, due
to the more hydrophobic profile. Both, natural organic matter (NOM), composed of humic substances, as well as proteins and poly­
saccharides, are the main fouling agents that cause the fouling of the membranes in wastewater and natural waters treatment (Zhao
et al., 2022). These foulants are also found in ADSC, due to their presence in primary and secondary sludge (Wilson and Novak, 2009).
Therefore, the continuous drop of the permeate flux shown by both membranes may be due to the macromolecules present in the feed
solution. Since large macromolecules that are completely retained by ultrafiltration membranes (eg, bovine serum albumin (BSA))
have been reported to cause the formation of a continuously growing gel layer on the membrane surface, a moderate but constant
increase in the resistance to filtration occurs (Haberkamp et al., 2008). Flux decline can also be due to the deposition of precipitates on
the membrane surface, such as calcium phosphate, struvite or sulfates.
Due to the nature of each membrane, RAY100 would be more affected by the hydrophilic fraction of the organic matter, while
FORM06 by the hydrophobic one. However, it is important to point out that flux decline for both membranes was low enough to not
significantly affect the membrane performance.
A large difference in the time needed to reach the conversion of 75% was observed. On the one hand, the PES membrane reached
the desired conversion after 17 hours, while the PVDF membrane needed almost 32 hours. This was expected, since, regardless of the
fact that both membranes have a similar MWCO, the PVDF membrane has lower permeability due to the strong hydrophobic nature of
the material (Subasi and Cicek, 2017).

Fig. 5. Evolution of the normalized permeate flux of PES (RAY100) and PVDF (FORM06) membranes in variable concentration tests using anaerobic
digestion sludge centrate (ADSC) as feed until a conversion of 75% was reached (J0: pure water permeate flux after membrane compaction).

9
M. Cifuentes-Cabezas et al. Environmental Technology & Innovation 35 (2024) 103661

Analyzing the performance of the RAY100 membrane, when the permeate flux was compared with that obtained at constant
concentration, it dropped by about 10 L⋅m− 2⋅h− 1 (at the same conditions, TMP 1.5 bar and 2 m/s), but it continued to maintain a stable
trend. On the other hand, it should be noted that a decrease in the permeate flux is expected when the concentration of the feed
increases.

3.3.2. Permeate characterization

Fig. 6 shows the ratio between the value of certain parameters in the permeate and in the filtered ADSC (before UF). Thus, a ratio of
1 means the same concentration in the UF permeate and feed stream, while values greater than 1 may correspond to the value being
within the standard deviation or to an increase in the permeation of some ions to maintain electroneutrality in the permeate. Although
this is a phenomenon more associated with nanofiltration processes, there are studies that indicate that the percentage of ion retention
during UF can decrease with increasing salinity or ionic strength. Which could be attributed to electrostatic repulsion by the negatively
charged membrane. Since, when the concentration of the feed solution exceeds the fixed charge concentration on the membrane, the
sorption of ions into the membrane phase is less hindered and could lead to decreased rejection (Guo et al., 2001; Reddy et al., 2003).
Therefore, by concentrating the solution and a possible negative surface membrane (due to PES material), a greater passage of ions into
the permeate could be generated. However, this variation was low, so we leaned over a possible values in the standard deviation. It is
important to remember again that the objective of the UF as a pretreatment of the ADSC is to maintain the maximal NH+ 4 (value close to
1) concentration in the permeate stream, with the lowest value of COD.
For the RAY100 membrane, a similar trend is observed for most of the measured parameters, decreasing sharply upon reaching a
conversion factor of around 9% and then remaining relatively constant as the conversion factor increases. However, NH+ 4 (ratio of 1.08
in conversion of 27.77%, corresponding to a concentration greater than 0.65 g/L in the permeate) as well as Ntotal presented an in­
crease when conversion raised from 9% to 45%. The maximum ratio for Ntotal was observed at 45% conversion, to then present a sharp
decline.
On the other hand, it is observed that the calculated ratio for K+ is around 1. This could also be an opportunity, since potassium,
another nutrient present in the ADSC, could be also recovered from the UF permeate as a precipitate (Gao et al., 2018).
Regarding FORM06 membrane, it can be seen in Fig. 6.B how divalent ions and COD present a similar behavior to that observed for
the RAY100 membrane. However, curiously, it can be seen that the ratio for the COD increased slightly up to 0.65 at the end of the test
performed with this membrane, this being prejudicial for the permeate quality. Although at the beginning of the test the concentration
of Ntotal and NH+4 remained almost unchanged, afterwards the ratios for these compounds continuously decreased, presenting a value
of about 0.72 at the end of the test. Nonetheless, the same as for the RAY100 membrane, K+ presented a relatively constant con­
centration (ratio around 1) throughout the test.
Regarding RAY100 membrane, it was possible to obtain a permeate stream without hardly loosing nutrients, which could be further
treated to recover nitrogen. Finally, RAY100 membrane is selected due to its higher permeate flux exhibited and to the greater COD

Fig. 6. Relationship between the concentration of the main parameters in the permeate and in the anaerobic digestion sludge centrate (ADSC),
during the concentration tests for both membranes. A: RAY100 and B: FORM06.

10
M. Cifuentes-Cabezas et al. Environmental Technology & Innovation 35 (2024) 103661

removal efficiency without significant NH+ 4 loss.

The characterization of the concentrate for RAY100 membrane was carried out to calculate the mass balance at the end of the test,
which is shown in Table 3. Delta represents the difference between the solute mass in the feed and that in permeate+retentate (in
mmol).
Some differences in the mass balance are always expected mainly due to the experimental errors and the withdrawal of small
volumes of samples for analysis. Thus, the small differences in the mass balance for NH+4 and K means that there were not losses of
+
2-
these ions during the process. Concerning SO4 , the value of delta is very small (0.69 mmol), but it corresponds to around 10% of the
concentration in ADSC, which means that there could be some losses in the process in form of precipitate. The high reduction in the
concentration of PO3-4 , Ca
2+
and Mg2+ indicate that there has been precipitation of calcium phosphate and magnesium phosphate
either in the tanks and/or on the membrane. To corroborate this matter, FESEM and EDX characterization of the membrane surface
was performed.

3.4. FESEM and EDX analysis

By means of FESEM (Fig. 7A), it was verified that there was precipitation on the membrane (images of different membrane sectors).
Then, to determine the characteristics of the fouling layer, EDX analysis was performed (Fig. 7B). It was observed that the most
predominant elements were carbon (43.9 ± 0.7% w/w), oxygen (20.4 ± 0.4% w/w), Ca (9.8 ± 0.2% w/w), S (5.5 ± 0.1% w/w), and P
(5.17 ± 0.1% w/w). The presence of carbon, oxygen and nitrogen was expected due to both, the membrane material, and the coating
necessary to analyze the sample. On the other hand, part of the fouling of the membrane was due to the deposited organic matter. The
results of the EDX analysis suggest the precipitation of Ca3(PO4)2 and Mg3(PO4)2 on the membrane. This highlights the need of periodic
cleaning of the membrane, both with water and chemicals to remove the precipitate. Although EDX spectra shows the presence of S on
the surface of the membrane, this could be mainly attributed to its material (PES), since, according to the mass balance displayed in
Table 3, SO2-4 was hardly lost in the process.

3.5. Membrane cleaning

After each test, the membranes were subjected to the cleaning protocol indicated in section 2.4.2. The permeability of the
membranes was measured after each cleaning step and compared with the value obtained for the pristine membranes. The established
criterion to consider the membrane as clean before performing the next test was a recovery of at least 90% of the initial permeability.
It can be seen in Fig. 8 that, in almost all tests a recovery close to 90% was achieved by cleaning only with pure water, with the
UP005 membrane being the one that presented the highest recovery (average value of 92.9 ± 1.7%). Then, after cleaning with P3
Ultrasil 115, 90% recovery was achieved in all cases. The cleaning efficiency of this chemical used for membrane cleaning has been
widely commented in other studies with different foulants (Regula et al., 2014). It is important to note that, the percentages presented
are an average of the cleanings after each test carried out. Therefore, it must be considered that in some tests performed under more
severe conditions (the highest TMP), the recovery with water was lower, hence the value of the error bars. Curiously, the FORM06
membrane was the one that reported the greatest variation in permeability recovery, with an average of 90.78 ± 4.08% and 96.6 ±
3.1% for the recovery with pure water and Ultrasil, respectively. This implies that, when this membrane works at a higher TMP, fouling
is greater and therefore the recovery percentages obtained after cleaning are lower. The result was expected as this membrane pre­
sented the greatest decline in permeate flux with time (Fig. 5). All this implies a more severe pore blocking due to foulant adhesion
inside the membrane pores and, therefore, greater difficulty to recover the initial membrane permeability (Cifuentes-Cabezas et al.,
2023a). This is an important factor to consider since the best working condition for flat membranes was TMP of 1.5 bar. It is important
to highlight that the data provided corresponds to both the total recirculation test and the variable concentration test. Therefore, the
membranes managed to be cleaned in both tests. Although differences were observed, the first cleaning with pure water was the one
that represented the greatest variability (approximately a 9% variation).
Finally, the cleaning protocols used managed to reach the established permeability recovery criterion. For this purpose, water
rinsing and chemical cleaning had to be combined, and each protocol depended on the membrane.

Table 3
Material balance for RAY100 membrane at the end of the test (75% conversion).
Conversion (%) 0 75 Delta

Amount of substance(mmol) ADSC P+R

Ca2+ 38.25 ± 1.91 23.06 ± 2.59 15.19

Mg2+ 27.81 ± 1.39 16.01 ± 0.12 11.81
NH+ 4 335.37 ± 16.68 342.13 ± 8.15 -6.76
K+ 126.15 ± 7.23 129.06 ± 5.53 -2.9
SO2-
4 6.45 ± 0.86 5.77 ± 0.19 0.69
PO3-
4 12.53 ± 0.62 5.58 ± 0.87 6.94

ADSC: anaerobic digestion sludge centrate; P+R= permeate + retentate; delta= difference between ADSC and P+R at 75% conversion.

11
M. Cifuentes-Cabezas et al. Environmental Technology & Innovation 35 (2024) 103661

Fig. 7. A: FESEM images at 8000 magnification and B: EDX spectra of the RAY100 membrane surface after the variable concentration tests.

Fig. 8. Recovery of permeate flux after membrane cleaning.

4. Conclusions

In this work, the use of ultrafiltration as a potential pretreatment of ADSC for a further nitrogen recovery step (for example by
membrane contactors) was assessed. Nitrogen recovery is aimed in the municipal wastewater treatment in the frame of circular
economy and to produce liquid biofertilizers that could replace commercial products made with a high carbon footprint. Three
membranes of different materials and MWCO were tested in experiments at constant and variable concentration. The main conclusions

12
M. Cifuentes-Cabezas et al. Environmental Technology & Innovation 35 (2024) 103661

achieved are the following:

• The 100 kDa MWCO membranes, made of different materials (PES and PVDF), were selected to perform variable concentration test
due to the highest permeate flux with a similar solute rejection. The selected TMP was 1.5 bar.
• The FORM06 PVDF membrane needed approximately twice as long as the RAY100 membrane to achieve 75% conversion.
• The RAY100 PES membrane was selected due to the highest permeate flux exhibited and the highest COD removal efficiency
without a significant loss of NH+
4.
• The mass balances carried out and the FESEM and EDX analyses performed demonstrated the presence of calcium, magnesium and
phosphorus on the surface of the RAY100 membrane, which can be associated with the formation of precipitates in the form of
Ca3(PO4)2 and Mg3(PO4)2.

Summarizing, UF was demonstrated to be an appropriate pretreatment of the ADSC for a further nitrogen recovery process.
However, special attention has to be paid to salts precipitation, due to the presence of phosphates in ADSC.

CRediT authorship contribution statement

Magdalena Cifuentes-Cabezas: Writing – review & editing, Writing – original draft, Visualization, Validation, Methodology,
Investigation, Formal analysis, Conceptualization. María Amparo Bes-Piá: Writing – review & editing, Project administration,
Methodology, Conceptualization. Silvia Álvarez-Blanco: Writing – review & editing, Visualization, Supervision, Project adminis­
tration, Methodology, Conceptualization. Ester Pérez-Valiente: Validation, Methodology, Investigation, Formal analysis, Concep­
tualization. María-José Luján-Facundo: Writing – review & editing, Visualization, Supervision, Methodology, Conceptualization.
José Antonio Mendoza-Roca: Writing – review & editing, Supervision, Project administration, Methodology, Conceptualization.

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.

Data availability

Data will be made available on request.

Acknowledgements

The authors thank FACSA (SOCIEDAD DE FOMENTO AGRÍCOLA CASTELLONENSE, S.A.), BIOVIC CONSULTING S.L. and INDE­
REN (INGENIERÍA Y DESARROLLOS RENOVABLES, S.L.) companies for supporting this work in the frame of the project BIOFERES
funded by Agencia Valenciana de Innovación de la Generalitat Valenciana.

References

Aguilar-Moreno, M., Vinardell, S., Reig, M., Vecino, X., Valderrama, C., Cortina, J.L., 2022. Impact of sidestream pre-treatment on ammonia recovery by membrane
contactors: experimental and economic evaluation. Membranes 12. https://doi.org/10.3390/membranes12121251.
APHA, 2005. Standard Methods for the Examination of Water and Wastewater. In: Standard Methods for the Examination of Water and Wastewater. American Public
Health Association, Washington, DC, p. 21 (st. ed).
Beckinghausen, A., Odlare, M., Thorin, E., Schwede, S., 2020. From removal to recovery: an evaluation of nitrogen recovery techniques from wastewater. Appl.
Energy 263, 114616. https://doi.org/10.1016/j.apenergy.2020.114616.
Brover, S., Lester, Y., Brenner, A., Sahar-Hadar, E., 2022. Optimization of ultrafiltration as pre-treatment for seawater RO desalination. Desalination 524, 115478.
https://doi.org/10.1016/j.desal.2021.115478.
Camilleri-Rumbau, M.S., Norddahl, B., Wei, J., Christensen, K.V., Søtoft, L.F., 2015. Microfiltration and ultrafiltration as a post-treatment of biogas plant digestates for
producing concentrated fertilizers. Desalin. Water Treat. 43, 238–244. https://doi.org/10.1016/j.watres.2008.10.011.
Carbonell-Alcaina, C., Álvarez-Blanco, S., Bes-Piá, M.A., Mendoza-Roca, J.A., Pastor-Alcañiz, L., 2018. Ultrafiltration of residual fermentation brines from the
production of table olives at different operating conditions. J. Clean. Prod. 189, 662–672. https://doi.org/10.1016/j.jclepro.2018.04.127.
Chen, M., Ding, W., Zhou, M., Zhang, H., Ge, C., Cui, Z., Xing, W., 2021. Fouling mechanism of PVDF ultrafiltration membrane for secondary effluent treatment from
paper mills. Chem. Eng. Res. Des. 167, 37–45. https://doi.org/10.1016/j.cherd.2020.12.021.
Chojnacka, K., Skrzypczak, D., Szopa, D., Izydorczyk, G., Moustakas, K., Witek-Krowiak, A., 2023. Management of biological sewage sludge: fertilizer nitrogen
recovery as the solution to fertilizer crisis. J. Environ. Manag. 326, 116602 https://doi.org/10.1016/j.jenvman.2022.116602.
Chuda, A., Ziemiński, K., 2023. Ultrafiltration of digestate liquid fraction by hollow-fiber membranes: Influence of digestate pre-treatment on hydraulic capacity and
nutrient removal efficiency. Chem. Eng. J. 473 https://doi.org/10.1016/j.cej.2023.145426.
Cifuentes-Cabezas, M., Bohórquez-Zurita, J.L., Gil-Herrero, S., Vincent-Vela, M.C., Mendoza-Roca, J.A., Álvarez-Blanco, S., 2023a. Deep study on fouling modelling of
ultrafiltration membranes used for OMW treatment: comparison between semi-empirical models, response surface, and artificial neural networks. Food
Bioprocess Technol. https://doi.org/10.1007/s11947-023-03033-0.
Cifuentes-Cabezas, M., Carbonell-Alcaina, C., Vincent-Vela, M.C., Mendoza-Roca, J.A., Álvarez-Blanco, S., 2021. Comparison of different ultrafiltration membranes as
first step for the recovery of phenolic compounds from olive-oil washing wastewater. Process Saf. Environ. Prot. 149, 724–734. https://doi.org/10.1016/j.
psep.2021.03.035.
Cifuentes-Cabezas, M., Luján-Facundo, M.J., Cuartas-Uribe, B., Iborra-Clar, A., Mendoza-Roca, J.A., 2023b. Nitrogen recovery from sludge centrate by membrane
contactor: influence of operating parameters and cleaning conditions. J. Environ. Manag. 341 https://doi.org/10.1016/j.jenvman.2023.118051.

13
M. Cifuentes-Cabezas et al. Environmental Technology & Innovation 35 (2024) 103661

Clark, B., Sharma, N., Apraku, E., Dong, H., Tarpeh, W.A., 2024. Ligand exchange adsorbents for selective phosphate and total ammonia nitrogen recovery from
wastewaters. Acc. Mater. Res. https://doi.org/10.1021/accountsmr.3c00290.
Cruz, H., Law, Y.Y., Guest, J.S., Rabaey, K., Batstone, D., Laycock, B., Verstraete, W., Pikaar, I., 2019. Mainstream ammonium recovery to advance sustainable urban
wastewater management. Environ. Sci. Technol. 53 https://doi.org/10.1021/acs.est.9b00603.
Darestani, M., Haigh, V., Couperthwaite, S.J., Millar, G.J., Nghiem, L.D., 2017. Hollow fibre membrane contactors for ammonia recovery: current status and future
developments. J. Environ. Chem. Eng. 5, 1349–1359. https://doi.org/10.1016/j.jece.2017.02.016.
Deemter, D., Oller, I., Amat, A.M., Malato, S., 2022. Advances in membrane separation of urban wastewater effluents for (pre)concentration of microcontaminants and
nutrient recovery: a mini review. Chem. Eng. J. Adv. 11, 100298 https://doi.org/10.1016/j.ceja.2022.100298.
Devos, P., Filali, A., Grau, P., Gillot, S., 2023. Sidestream characteristics in water resource recovery facilities: a critical review. Water Res 232, 119620. https://doi.
org/10.1016/j.watres.2023.119620.
Di Costanzo, N., Cesaro, A., Di Capua, F., Esposito, G., 2021. Exploiting the nutrient potential of anaerobically digested sewage sludge: a review. Energies 14, 1–25.
https://doi.org/10.3390/en14238149.
Fux, C., Velten, S., Carozzi, V., Solley, D., Keller, J., 2006. Efficient and stable nitritation and denitritation of ammonium-rich sludge dewatering liquor using an SBR
with continuous loading. Water Res 40, 2765–2775. https://doi.org/10.1016/j.watres.2006.05.003.
Gao, Y., Liang, B., Chen, H., Yin, P., 2018. An experimental study on the recovery of potassium (K) and phosphorous (P) from synthetic urine by crystallization of
magnesium potassium phosphate. Chem. Eng. J. 337, 19–29. https://doi.org/10.1016/j.cej.2017.12.077.
Gherghel, A., Teodosiu, C., De Gisi, S., 2019. A review on wastewater sludge valorisation and its challenges in the context of circular economy. J. Clean. Prod. 228,
244–263. https://doi.org/10.1016/j.jclepro.2019.04.240.
Giacobbo, A., Bernardes, A.M., Rosa, M.J.F., De Pinho, M.N., 2018. Concentration polarization in ultrafiltration/nanofiltration for the recovery of polyphenols from
winery wastewaters. Membranes 8. https://doi.org/10.3390/membranes8030046.
Guo, L., Hunt, B.J., Santschi, P.H., 2001. Ultrafiltration behavior of major ions (Na, Ca, Mg, F, Cl, AND SO4) in natural waters. Water Res 35, 1500–1508.
Guo, C.H., Stabnikov, V., Ivanov, V., 2010. The removal of nitrogen and phosphorus from reject water of municipal wastewater treatment plant using ferric and nitrate
bioreductions. Bioresour. Technol. 101, 3992–3999. https://doi.org/10.1016/j.biortech.2010.01.039.
Haberkamp, J., Ernst, M., Makdissy, G., Huck, P.M., Jekel, M., 2008. Protein fouling of ultrafiltration membranes - Investigation of several factors relevant for tertiary
wastewater treatment. J. Environ. Eng. Sci. 7, 651–660. https://doi.org/10.1139/S08-038.
Holloway, R.W., Childress, A.E., Dennett, K.E., Cath, T.Y., 2007. Forward osmosis for concentration of anaerobic digester centrate. Water Res. 41, 4005–4014. https://
doi.org/10.1016/j.watres.2007.05.054.
Judd, S., Judd, C., 2011. Chapter 3. Design, operation and maintenance. : MBR Book 209–288. https://doi.org/10.1016/b978-0-08-096682-3.10003-4.
Ledda, C., Schievano, A., Salati, S., Adani, F., 2013. Nitrogen and water recovery from animal slurries by a new integrated ultrafiltration, reverse osmosis and cold
stripping process: a case study. Water Res 47, 6157–6166. https://doi.org/10.1016/j.watres.2013.07.037.
Li, C., Cabassud, C., Guigui, C., 2015. Effects of carbamazepine in peak injection on fouling propensity of activated sludge from a MBR treating municipal wastewater.
J. Memb. Sci. 475, 122–130. https://doi.org/10.1016/j.memsci.2014.10.017.
Lin, T., Shen, B., Chen, W., Zhang, X.B., 2014. Interaction mechanisms associated with organic colloid fouling of ultrafiltration membrane in a drinking water
treatment system. Desalination 332, 100–108. https://doi.org/10.1016/j.desal.2013.11.001.
Lubensky, J., Ellersdorfer, M., Stocker, K., 2019. Ammonium recovery from model solutions and sludge liquor with a combined ion exchange and air stripping process.
J. Water Process Eng. 32, 100909 https://doi.org/10.1016/j.jwpe.2019.100909.
Munasinghe-Arachchige, S.P., Abeysiriwardana-Arachchige, I.S.A., Delanka-Pedige, H.M.K., Cooke, P., Nirmalakhandan, N., 2021. Nitrogen-fertilizer recovery from
urban sewage via gas permeable membrane: Process analysis, modeling, and intensification. Chem. Eng. J. 411, 128443 https://doi.org/10.1016/j.
cej.2021.128443.
Munasinghe-Arachchige, S.P., Cooke, P., Nirmalakhandan, N., 2020. Recovery of nitrogen-fertilizer from centrate of anaerobically digested sewage sludge via gas-
permeable membranes. J. Water Process Eng. 38, 101630 https://doi.org/10.1016/j.jwpe.2020.101630.
Munasinghe-Arachchige, S.P., Nirmalakhandan, N., 2020. Nitrogen-fertilizer recovery from the centrate of anaerobically digested sludge. Environ. Sci. Technol. Lett.
7, 450–459. https://doi.org/10.1021/acs.estlett.0c00355.
Noriega-Hevia, G., Serralta, J., Borrás, L., Seco, A., Ferrer, J., 2020. Nitrogen recovery using a membrane contactor: modelling nitrogen and pH evolution. J. Environ.
Chem. Eng. 8, 103880 https://doi.org/10.1016/j.jece.2020.103880.
Pastor, L., Marti, N., Bouzas, A., Seco, A., 2008. Sewage sludge management for phosphorus recovery as struvite in EBPR wastewater treatment plants. Bioresour.
Technol. 99, 4817–4824. https://doi.org/10.1016/j.biortech.2007.09.054.
Puchongkawarin, C., Gomez-Mont, C., Stuckey, D.C., Chachuat, B., 2015. Optimization-based methodology for the development of wastewater facilities for energy
and nutrient recovery. Chemosphere 140, 150–158. https://doi.org/10.1016/j.chemosphere.2014.08.061.
Qadir, M., Drechsel, P., Jiménez Cisneros, B., Kim, Y., Pramanik, A., Mehta, P., Olaniyan, O., 2020. Global and regional potential of wastewater as a water, nutrient
and energy source. Nat. Resour. Forum 44, 40–51. https://doi.org/10.1111/1477-8947.12187.
Qin, Y., Wang, K., Xia, Q., Yu, S., Zhang, M., An, Y., Zhao, X., Zhou, Z., 2023. Up-concentration of nitrogen from domestic wastewater: a sustainable strategy from
removal to recovery. Chem. Eng. J. 451, 138789 https://doi.org/10.1016/j.cej.2022.138789.
Reddy, A.V.R., Mohan, D.J., Bhattacharya, A., Shah, V.J., Ghosh, P.K., 2003. Surface modification of ultrafiltration membranes by preadsorption of a negatively
charged polymer: I. Permeation of water soluble polymers and inorganic salt solutions and fouling resistance properties. J. Memb. Sci. 214, 211–221. https://doi.
org/10.1016/S0376-7388(02)00547-1.
Regula, C., Carretier, E., Wyart, Y., Gésan-Guiziou, G., Vincent, A., Boudot, D., Moulin, P., 2014. Chemical cleaning/disinfection and ageing of organic UF membranes:
a review. Water Res 56, 325–365. https://doi.org/10.1016/j.watres.2014.02.050.
Reig, M., Vecino, X., Aguilar-Moreno, M., Valderrama, C., Cortina, J.L., 2022. Ammonia valorization by liquid–liquid membrane contactors for liquid fertilizers
production: experimental conditions evaluation. Membranes. https://doi.org/10.3390/membranes12070663.
Sichler, T.C., Adam, C., Montag, D., Barjenbruch, M., 2022. Future nutrient recovery from sewage sludge regarding three different scenarios - German case study.
J. Clean. Prod. 333, 130130 https://doi.org/10.1016/j.jclepro.2021.130130.
Soler-Cabezas, J.L., Mendoza-Roca, J.A., Vincent-Vela, M.C., Luján-Facundo, M.J., Pastor-Alcañiz, L., 2018. Simultaneous concentration of nutrients from
anaerobically digested sludge centrate and pre-treatment of industrial effluents by forward osmosis. Sep. Purif. Technol. 193, 289–296. https://doi.org/10.1016/
j.seppur.2017.10.058.
Subasi, Y., Cicek, B., 2017. Recent advances in hydrophilic modification of PVDF ultrafiltration membranes – a review: part II. Membr. Technol. 2017, 5–11. https://
doi.org/10.1016/S0958-2118(17)30233-1.
Suleman, S.Bin, Hai, F.I., Mukhtar, H., Duong, H.C., Ansari, A.J., 2022. Influence of operating parameters and membrane fouling on nutrient transport by FO
membrane. J. Water Process Eng. 47, 102699 https://doi.org/10.1016/j.jwpe.2022.102699.
Susanto, H., Ulbricht, M., 2005. Influence of ultrafiltration membrane characteristics on adsorptive fouling with dextrans. J. Memb. Sci. 266, 132–142. https://doi.
org/10.1016/j.memsci.2005.05.018.
Thornton, A., Pearce, P., Parsons, S.A., 2007. Ammonium removal from digested sludge liquors using ion exchange. Water Res. 41, 433–439. https://doi.org/
10.1016/j.watres.2006.10.021.
Tow, E.W., Rad, B., Kostecki, R., 2022. Biofouling of filtration membranes in wastewater reuse: In situ visualization with confocal laser scanning microscopy. J. Memb.
Sci. 644, 120019 https://doi.org/10.1016/j.memsci.2021.120019.
Uysal, A., Yilmazel, Y.D., Demirer, G.N., 2010. The determination of fertilizer quality of the formed struvite from effluent of a sewage sludge anaerobic digester.
J. Hazard. Mater. 181, 248–254. https://doi.org/10.1016/j.jhazmat.2010.05.004.
Vecino, X., Reig, M., Bhushan, B., Gibert, O., Valderrama, C., Cortina, J.L., 2019. Liquid fertilizer production by ammonia recovery from treated ammonia-rich
regenerated streams using liquid-liquid membrane contactors. Chem. Eng. J. 360, 890–899. https://doi.org/10.1016/j.cej.2018.12.004.

14
M. Cifuentes-Cabezas et al. Environmental Technology & Innovation 35 (2024) 103661

Vu, M.T., Price, W.E., He, T., Zhang, X., Nghiem, L.D., 2019. Seawater-driven forward osmosis for pre-concentrating nutrients in digested sludge centrate. J. Environ.
Manag. 247, 135–139. https://doi.org/10.1016/j.jenvman.2019.06.082.
Wang, M., Khan, M.A., Mohsin, I., Wicks, J., Ip, A.H., Sumon, K.Z., Dinh, C.T., Sargent, E.H., Gates, I.D., Kibria, M.G., 2021. Can sustainable ammonia synthesis
pathways compete with fossil-fuel based Haber-Bosch processes? Energy Environ. Sci. 14, 2535–2548. https://doi.org/10.1039/d0ee03808c.
Williams, S., 1999. Struvite precipitation in the sludge stream at slough wastewater treatment plant and opportunities for phosphorus recovery. Environ. Technol. 20,
743–747. https://doi.org/10.1080/09593332008616869.
Wilson, C.A., Novak, J.T., 2009. Hydrolysis of macromolecular components of primary and secondary wastewater sludge by thermal hydrolytic pretreatment. Water
Res 43, 4489–4498. https://doi.org/10.1016/j.watres.2009.07.022.
Yasar, A., Can Dogan, E., Ayberk, H.S., Aydiner, C., 2022. Water recovery from urban wastewater for Irrigation using Ultrafiltration and Nanofiltration: optimization
and performance. Clean. - Soil Air Water 50. https://doi.org/10.1002/clen.202200280.
Yue, C., Dong, H., Chen, Y., Shang, B., Wang, Y., Wang, S., Zhu, Z., 2021. Direct purification of digestate using ultrafiltration membranes: Influence of pore size on
filtration behavior and fouling characteristics. Membranes 11, 1–16. https://doi.org/10.3390/membranes11030179.
Zacharof, M.P., Mandale, S.J., Oatley-Radcliffe, D., Lovitt, R.W., 2019. Nutrient recovery and fractionation of anaerobic digester effluents employing pilot scale
membrane technology. J. Water Process Eng. 31, 1–10. https://doi.org/10.1016/j.jwpe.2019.100846.
Zhang, Z., Xu, Z., Song, X., Zhang, B., Li, G., Huda, N., Luo, W., 2020. Membrane processes for resource recovery from anaerobically digested livestock manure
effluent: opportunities and challenges. Curr. Pollut. Rep. 6, 123–136. https://doi.org/10.1007/s40726-020-00143-7.
Zhao, L., Liu, Z., Soyekwo, F., Liu, C., Hu, Y., Niu, Q.J., 2022. Exploring the feasibility of novel double-skinned forward osmosis membranes with higher flux and
superior anti-fouling properties for sludge thickening. Desalination 523, 115410. https://doi.org/10.1016/j.desal.2021.115410.
Zheng, X., Ernst, M., Jekel, M., 2009. Identification and quantification of major organic foulants in treated domestic wastewater affecting filterability in dead-end
ultrafiltration. Water Res 43, 238–244. https://doi.org/10.1016/j.watres.2008.10.011.
Zielińska, M., Bernat, K., Mikucka, W., 2020. Membrane bioreactor technology: the effect of membrane filtration on biogas potential of the excess sludge. Membranes
10, 1–16. https://doi.org/10.3390/membranes10120397.

15