Dewatering of Short-Fibre Digestates from Paper Recycling Mills: Liquid Fraction and Mass Distribution Profiles

Abstract

The paper sector is characterised by high freshwater consumption and a strong need for improved resource efficiency. In this context, industrial digestates derived from short-fibre residues in paper recycling mills represent a promising substrate for water recovery within a circular economy framework. This study investigated the dewatering of short-fibre digestates as a pre-treatment for downstream membrane processes, aiming to maximise the liquid fraction (LF) recovery while minimising dry matter (DM) content. Seven scenarios were studied: sedimentation (S0); pre-sedimentation with chemical addition using iron(III) chloride (FeCl₃) + polydiallyldimethylammonium chloride (polyDADMAC) (S1), FeCl₃ + starch (S2), Nanofloc^® (S3), and polyDADMAC (S4); and direct dewatering without pre-sedimentation using polyDADMAC with cloth filtration (S5) and centrifugation (S6). With reference to the sedimentation supernatant, S4 achieved the highest DM separation efficiency of 76%, followed by S1 (64%), whereas S2 and S3 were below 40%. However, LF recovery relative to the initial digestate was limited in scenarios S1–S4 to 17% (170 g/kg_digestate), with DM concentrations of 2.0–4.8 g/kg_LF. In contrast, direct dewatering increased LF recovery substantially, with centrifugation (S6) achieving up to 690 g_LF/kg_digestate and cloth filtration (S5) 420 g/kg_digestate, while maintaining a low DM (1.7 g/kg_LF). Chemical oxygen demand (COD) and phosphorus (Ptot) were largely separated from the liquid fractions in all the scenarios. Nitrogen (Ntot) and ammonium (NH₄-N) in the LF remained more variable, ranging from 22 to 153 and 5 to 22 mg/kg_digestate, respectively. These results indicate that centrifugation with polyDADMAC is the most effective approach, suggesting that mechanical force with a chemical additive can be used for the efficient dewatering of short-fibre digestates.

1. Introduction

The paper and pulp industry is one of the largest industrial sectors in terms of raw material throughput. In Europe, approximately 850 paper mills produce around 80 million tonnes of paper and board annually as of 2024, with Germany accounting for roughly 25% of European output [1]. Paper recycling rates in the EU have reached nearly 79%, reflecting efforts to close material loops and reduce environmental impact [2]. Despite these advances, paper recycling remains resource-intensive, requiring on average 5 L of water and 2.5 kWh of energy per kilogram of paper produced.

A key challenge in modern paper recycling is the management of short-fibre residues, such as deinking sludge, pellet sludge, fibre rejects, and short-fibre sludge. These organic-rich waste streams cannot be reused in papermaking due to reduced fibre length and altered material properties. They contain cellulose fibres, mineral fillers (e.g., calcium carbonate), inks, adhesives, and other process chemicals. Currently, these residues are predominantly incinerated in Germany, while in other countries they are landfilled [3,4,5].

Short-fibre residues are promising feedstocks for biogas production, which can substantially reduce greenhouse gas emissions compared to fossil fuels [6]. Continuous biogas production from these residues has been successfully demonstrated at lab and pilot scales, with yields up to 400 NL/kg organic dry matter (oDM) from various streams, including deinking sludge [7,8,9,10]. While biogas production supports climate and sustainability goals, the management of the resulting short-fibre digestates remains a critical bottleneck for achieving full process optimisation.

Short-fibre digestates principally differ from conventional agricultural digestates due to their mineral fillers, process additives, and industrial contaminants. Their direct application to land is prohibited under the German Fertiliser Ordinance (Düngemittelverordnung, 2020; §7(3)19; §8(4)9), requiring engineered treatment strategies [11]. Illegal land application has caused environmental incidents, including contamination by per- and polyfluoroalkyl substances (PFASs), affecting soil and groundwater [2]. These risks point to the need for controlled and compliant management pathways for short-fibre digestates.

Given the high freshwater demand in paper recycling, short-fibre digestates represent a potential internal water resource, but direct reuse is not feasible. Dewatering is applied first as a pre-treatment step to separate solids and produce a clarified liquid fraction. This liquid fraction can then be treated in downstream processes for water recovery. The existing studies on short-fibre digestates, particularly on deinking sludge, mainly focused on their solid content and potential applications in cement and concrete production [12,13,14]. Short-fibre digestates differ in composition and particle size from agricultural digestates, raising questions about the suitability of conventional digestate dewatering methods [15].

Digestate management generally follows two approaches: (a) dewatering, which reduces digestate volume by separating solids and liquids using filtration, centrifugation, screw presses, or chemical conditioning, producing a solid fraction suitable for further processing, [16] and (b) purification, which recovers valuable components such as ammonium (NH₄-N) and Phosphor (Ptot) and treats the liquid to meet discharge or reuse standards through techniques such as membrane filtration, struvite precipitation, ammonia stripping, or vacuum evaporation [17,18,19,20].

To systematically address digestate management, the present study applies classic dewatering strategies for short-fibre digestates. Preliminary experiments from the current study (Figures S1–S7, Supplementary Materials) tested different chemical additives, dosage rates, and combinations with respect to floc formation and clear solid–liquid separation. Based on these results, four chemical additives and their most effective dosage rates were selected for the main study. Seven dewatering scenarios were investigated in the current study, grouped into with and without pre-sedimentation approaches. The baseline scenario consisted of simple sedimentation (S0), while the four dewatering scenarios (S1–S4) combined pre-sedimentation with each of the selected chemical additives. Two scenarios without pre-sedimentation applied mechanical dewatering, i.e., cloth filtration (S5) and centrifugation (S6), using the chemical additive identified as most effective among S1–S4. For all scenarios, mass distribution profiles of fresh matter (FM), dry matter (DM), chemical oxygen demand (COD), total nitrogen (Nₜₒₜ), NH₄-N, and Pₜₒₜ were determined to evaluate the dewatering performance.

This study provides a detailed characterisation of a mixture of short-fibre digestates derived from different biogas fermenters fed with diverse residues from German and European paper recycling mills. The analysis includes PFAS contaminants, metals, and trace elements to inform safe handling and downstream utilisation. The main objective of the present study was to identify the optimal dewatering approach among the seven scenarios that yields the highest liquid fraction recovery with the lowest DM concentration, suitable as a pre-treatment for downstream membrane treatment.

2. Results and Discussion

2.1. Performance of Chemical Additives with Pre-Sedimentation

Preliminary experiments (Figures S1–S7, Supplementary Materials) tested five individual chemical additives (FeCl₃, lime, starch, Nanofloc, and polyDADMAC) and two sets of coagulant–polymer combinations: FeCl₃ or lime (coagulants) each combined with four polymers (starch, Nanofloc, polyDADMAC, and polyacrylamide) with each chemical or combination at three dosage rates. Given the large number of tests, each experiment was visually evaluated based on floc formation, settling rate (30 min), and clarity of the supernatant. Additives producing rapid sedimentation, compact flocs, and a clear supernatant were considered for the main study scenarios S1–S4.

S1 and S2 used FeCl₃ with polymeric or starch, while S3 applied Nanofloc alone and S4 polyDADMAC alone. These scenarios represent different dewatering mechanisms such as metal-induced charge neutralisation and precipitation (FeCl₃), polymer bridging (polyDADMAC, Nanofloc), and polysaccharide-assisted flocculation (starch). Mass distributions are calculated relative to the supernatant of pre-sedimented digestate (SPSD), serving as 100% reference prior to chemical conditioning.

2.1.1. FM and DM Distribution

The mass distribution profiles of FM, DM, COD and nutrients varied strongly between scenarios S1 to S4, as shown in Figure 1. In this section, all percentage values are calculated relative to the respective component present in the SPSD. Solid fractions were not measured directly due to their high dry matter content, which made cuvette-based methods (Hach kits) unsuitable for measurements. Therefore, compositions were estimated through mass balance from the measured digestate and liquid fraction and should be considered indicative rather than absolute values.

The main goal was to maximise liquid recovery while maintaining a low solid carry-over in the liquid fraction for downstream membrane filtration. S3 achieved the highest liquid FM recovery at 778 g/kg_SPSD, followed by S2 (726 g/kg_SPSD), S4 (648 g/kg_SPSD), and S1 (437 g/kg_SPSD). The supernatants used for S1–S2 (SPSD from S0a) and S3–S4 (SPSD from S0b) differed in initial composition (

Table 1. Composition of liquid fractions in scenarios S0–S6.

Parameter	Unit	S01	S02	S1	S2	S3	S4	S5	S6
DM	Wt. %	0.54 ± 0.002	0.54 ± 0.002	0.4 ± 0.066	0.5 ± 0.02	0.43 ± 0.01	0.2 ± 0.003	0.17 ± 0.04	0.17 ± 0.006
oDM	Wt. %	0.26 ± 0.0	0.26 ± 0.0	0.09 ± 0.05	0.13 ± 0.02	0.2 ± 0.01	0.07 ± 0.00	0.09 ± 0.03	0.04 ± 0.01
COD	mg/L	4640 ± 0	6990	1744 ± 79	1808 ± 464	1263 ± 65	997 ± 147	1488 ± 460	1148 ± 226
Ntot	mg/L	297 ± 0	1250 ± 0	288 ± 50	243 ± 1	441 ± 42	1424 ±136	135 ± 30	133 ± 17
NH4-N	mg/L	94 ± 0	55 ± 0	133 ± 2.75	130 ± 1.4	55 ± 2	48 ± 8	49 ± 4	75 ± 3
Ptot	mg/L	30 ± 0	79± 0	12 ± 2	16 ± 0.002	1 ± 0.03	11 ± 5	27 ± 1	32 ± 2
PO4-P	mg/L	-	-	3.4 ± 0.6	8.4 ± 0.6	6.1 ± 0.3	21 ± 0.4	<1	25.7 ± 1.3
pH	-	7.7 ± 0.2	7.3 ± 0.2	7.4 ± 0.01	7.4 ± 0.002	7.4 ± 0.02	7.4 ± 0.02	7.9 ± 0.1	7.6 ± 0.1
EC	µS/cm	3350 ± 0	3580 ± 0	7310 ± 140	7233 ± 223	5307 ± 401	3200 ± 223	3967 ± 413	3277 ± 122

^{1, 2} correspond to S0¹ and S0², respectively, indicating the two supernatant batches. Both S0¹ and S0² originated from the same digestate at different time periods. Liquid fraction from S0¹ (supernatant of sedimentation) was used for S1–S2, while liquid fraction from S0² was used for S3–S4. EC—electrical conductivity. Values represent mean ± standard deviation. For S0¹ and S0², the standard deviations appear as “0.0” for some of the components because the supernatant, obtained from the mother digestate, was analysed in two determinations immediately before the jar tests were performed. The values therefore differed by less than one decimal place, resulting in very small standard deviations when rounded.

). Therefore, the observed variations in liquid recovery, DM content, and nutrient distribution are influenced not only by the applied treatment but also by this compositional difference.

Although S3 and S2 achieved high liquid recovery, DM in the liquid remained high (3.3–3.5 g DM/kg_SPSD), giving low dewatering efficiencies of 35–39% relative to DM in SPSD. S4 reduced DM to 1.3 g DM/kg_SPSD (ηDM 76%) (Figure 2b,i). In S1, the liquid fraction contained 1.9 g DM/kg_SPSD, but the solid fraction reached the highest DM content (9 g/kg_SF) compared with S4 (5 g/kg_SF) (Figure 2h). These trends correspond with the additive mechanisms. PolyDADMAC (S4) destabilises negatively charged colloids via charge neutralisation, forming dense, rapidly settling flocs and the lowest DM in the liquid (2 gDM/kg_LF) (Figure 2g). In S1, the combination of FeCl₃ and polyDADMAC likely caused charge reversal, leading to floating flocs and a brown supernatant (Figure S8, Supplementary Materials).

Zeta potential directly measures the particle surface charge, reflecting the effectiveness of charge neutralisation. In this study, only EC was measured. EC indicates the ionic strength, which affects electrostatic interactions and flocculation. Compared with the initial SPSDs (S01: 3350 µS/cm; S02: 3580 µS/cm), FeCl₃ addition in S1 and S2 increased EC (7200–7310 µS/cm). This indicates a high release of dissolved ions from ferric hydrolysis, suggesting overdosing. Elevated ionic strength can compress the electrical double layer but may also lead to charge reversal or colloidal re-stabilisation under excessive dosing conditions [21]. It is consistent with the observed floating flocs and incomplete solid transfer in S1. Phase-specific pH differences observed further in S1 and S2 (neutral liquid pH 7.4, acidic solid pH 3) further support localised floc acidification. Such conditions can alter particle density and gas formation, potentially promoting flotation rather than effective settling [22]. S4 (polyDADMAC alone) slightly reduced EC (3200 µS/cm), indicating that dissolved ions largely remained in solution while charge neutralisation and bridging promoted the formation of dense and stable floc [23]. S3 (Nanofloc) showed an intermediate EC increase (5307 µS/cm), which may be associated with excess polymer addition or impurities contributing to the ionic strength.

FeCl₃ + starch in S2 aimed to combine charge neutralisation and floc bridging. However, particle destabilisation remained limited, leaving high residual DM (4.8 g/kg_LF) and COD (1292 mg/kg_SPSD) in the liquid fraction. The high FeCl₃ dosage (≈2900 mg/L) likely limited starch’s synergistic effect, unlike in simpler water systems where low dosages achieve >90% turbidity and P reduction [24,25].

2.1.2. COD and P Distribution

COD removal from the liquid fraction ranged from 72 to 91% and was highest in S4 (646 mg/kg_SPSD remaining in liquid), followed by S3 (86%, 982 mg/kg_SPSD), S1 (76% and 813 mg/kg_SPSD) and S2 (72%, 1292 mg/kg_SPSD).

Ptot followed a similar trend. S3 separated 99% of Ptot from the liquid fraction (0.8 mg/kg_SPSD remaining in the liquid fraction), followed by S4 (91%, 6.7 mg/kg_SPSD). S1 and S2 showed lower removal rates, with 82% and 62%. Lower efficiency in S1 and S2 can be linked to the limited Fe³⁺-mediated precipitation, floc formation and pH [23,26]. Comparison of nutrient distributions should be interpreted cautiously, as minor variations in the initial SPSD may also affect results.

2.1.3. N Distribution

Most NH₄-N remained in the liquid fraction, with 100% in S2, followed by S3 (72%), S1 (62%), and S4 (57%), relative to the concentrations in initial SPSDs. Ntot removal from the liquid fraction was higher in polymer-based S3–S4 (73–74%) than in FeCl₃-containing S1–S2 (42–59%). This pattern is partly explained by differences in the initial SPSD compositions (S0¹ and S0² in Table 1).

The polymers used in S3 and S4 are expected to aggregate particulate N_organic, while soluble NH₄-N remains largely unaffected. Despite their higher Ntot reduction in the liquid fraction (343 and 905 mg/kg_SPSD), neither S3 nor S4 substantially reduced dissolved NH₄-N to its fullest from the liquid fraction. In scenarios containing FeCl₃ (S1, S2), Fe³⁺ can hydrolyse to Fe³⁺ to form ferric hydroxide precipitates that enmesh particulate N, while soluble NH₄-N does not readily adsorb onto ferric hydroxide under the typical pH (near-neutral) and dosing conditions. The presence of floating flocs in S1 and S2 (Figure S8 in the Supplementary Materials) indicates this phenomenon yet confirms that dissolved NH₄-N predominantly persisted in the liquid fraction.

These observations are consistent with the literature findings. Flocculation alone does not significantly reduce NH₄-N, and high removal (>80%) is typically achieved only when combined with processes such as ammonia stripping [27]. Studies using FeCl₃ at comparable dosages (≈24,330 mg/L) report that 90% of NH₄-N remains in the liquid, while up to 93% of Ptot is transferred to solids, confirming that Fe³⁺ primarily removes particulate components [28]. Similar behaviour is observed for FeCl₃-polymer systems, which improve COD and Ptot removal but have a limited effect on NH₄-N [29]. These results confirm that both cationic polymers and ferric salts mainly remove particulate N, whereas dissolved NH₄⁺ remains in the liquid.

It should be noted that solid fractions were not analysed directly but were estimated through mass balance. The direct analysis of solids could provide information on their reuse potential. However, the calculated distribution profiles are sufficient to assess the dewatering performance with respect to the quality of the liquid fraction, since the main focus was on recovering the liquid fraction for downstream membrane treatment. Future studies could include direct solid fraction analyses to better guide solid utilisation from short-fibre digestates.

2.2. Performance of Mechanical Methods Without Pre-Sedimentation

The performance of cloth filtration (S5) and centrifugation (S6) was evaluated using polyDADMAC derived from S4, which had previously achieved the highest liquid fraction recovery and the lowest DM concentration. Figure 3 shows the mass distribution of FM, DM, COD, P_tot, N_tot, and NH₄-N between liquid (measured) and solid (estimated) fractions. Percentages refer to the initial raw digestate (100% reference). Images of liquid fractions are shown in Figure S9 in the Supplementary Materials.

2.2.1. FM and DM Distribution

Compared to S1–S4, which yielded <200 g LF/kg_digestate, S5 and S6 showed a higher FM recovery of the liquid fraction. Centrifugation (S6) recovered 686 g/kg_digestate (68%), while cloth filtration (S5) recovered 424 g/kg_digestate (42%). Consequently, more water remained in solids for S5 (429 g/kg_digestate) than S6 (180 g/kg_digestate). These differences are partially affected by slight variations in the initial digestate composition used for each scenario (

Table 2. Composition of mixture of short-fibre digestates used in the experiments of dewatering scenarios S0–S6.

Parameter	Unit	S0	S1	S2	S3	S4	S5	S6
DM	Wt. %	16 ± 0.09					14.7 ± 0.25	13.6 ± 0
COD	mg/L	45,213 ± 192					9950	37,150
Ntot	mg/L	1344 ± 776					401	387
NH4-N	mg/L	280 ± 83					73	97.2
Ptot	mg/L	362 ± 93					110	306
pH	-	7.5 ± 0.4					8 ± 0	8.5 ± 0
EC 1	µS/cm	-					4320 ± 0	19,050

¹ EC—electrical conductivity. Values represent mean ± standard deviation.

).

Both methods achieved low DM in the liquid fraction (1.7 g/kg_LF), indicating efficient separation. However, centrifugation (S6) concentrated more solids (444 g/kg_SF) than cloth filtration (S5, 260 g/kg_SF) due to the higher liquid recovery. The DM-free liquid fraction was highest in S6 (686 g/kg_digestate). Figure 4 illustrates this trend by plotting DM content in the liquid fraction (gDM/kgLF) against FM recovery (gLF/kgdigestate), showcasing the poor (S0–S4), intermediate (S5), and optimal (S6) performance.

Compared to S1–S4, both S5 and S6 showed an improved DM separation. Centrifugation without polymer (S00A, see Table S1 in Supplementary Materials) achieved a higher liquid recovery (810 g/kg_digestate) but much higher DM in the liquid (5 g/kg_LF). It limits its suitability for downstream reverse osmosis. This highlights the role of polyDADMAC in improving solid removal, particularly under centrifugation, which achieved the lowest residual water content of 180 g/kg_digestate in the solid fraction.

In agricultural digestates, mechanical systems often recover more than 90% of FM from the liquid fraction but achieve moderate DM separation (12–75%) (Table S2, Supplementary Materials) [30,31,32]. In contrast, this study showed lower FM recovery (42–68%) but very high DM separation (>99%). This emphasises that digestate characteristics influence separation performance, and that high FM recovery alone does not ensure effective dewatering performance, but must be considered together with reduced solids in the liquid fraction.

2.2.2. COD and P Distribution

COD was mainly transferred into the solid fraction. In S5, 94% was removed, leaving 599 mg COD/kg_digestate in the liquid fraction (Figure 5). In S6, 97% was removed, with 786 mg COD/kg_digestate remaining. The higher COD in S6 reflects a higher liquid recovery, despite the better separation efficiency.

The reported COD removal in agricultural digestates ranges from 85 to 96% with polymer-assisted separation [33]. Centrifuges generally outperform simpler systems when chemical additives are used [34]. In this study, baseline sedimentation (S0) alone, already separated 98% of COD from the supernatant, indicating a strong influence of digestate properties. The lower DM in the liquid fraction corresponded to lower COD in the present study, as seen in S4 (0.2 gDM/kg_digestate and 112 mg COD/kg_digestate), while a higher DM increased COD. Direct comparisons across scenarios should therefore be made cautiously, since the initial digestate composition varied between S1–S6 (Table 2).

Ptot followed a similar pattern to COD. S5 removed 89% and S6 removed 93%, leaving 11 and 22 mg Ptot/kg_digestate in the liquid, respectively. Pre-sedimentation (S1–S4) achieved nearly complete P removal, although baseline sedimentation (S0) had already separated 99% from the supernatant.

Importantly, Ptot in the liquid fraction does not completely denote Ptot speciation. Most residual P in the liquid was orthophosphate (PO₄-P), while particulate or organic P (Porg = Ptot − PO₄-P) was mainly removed from the liquid. In S6, the liquid fraction contained 32 mg/L Ptot, of which 25.7 mg/L was PO₄-P and the remaining 6.3 mg/L was Porg. In S5, Porg dominated the residual Ptot in the liquid fraction, while PO₄-P was nearly negligible. Chemical treatments in S1–S4 enhanced particulate P removal but left small amounts of PO₄-P. These trends, along with the full liquid fraction compositions reported in Table 1, showcase the differences in dewatering efficiency across the scenarios.

This behaviour is similar to conventional manure or agricultural digestates. P distribution depends on both technology and conditioning. A case study on a full-scale biogas plant using a screw press with additives reported that 52–67% of Ptot, including Porg, was transferred to solids, whereas unconditioned separation achieved only a 35% retention in solids [35]. Decanter centrifuges generally perform better, with reported Ptot separation efficiencies of 51–71%. Notably, the short-fibre digestate in the present study had much lower Ptot concentrations (<300 mg/L) available during the time of investigation than typical agricultural digestates.

Short-fibre digestates contain both fine and coarse particles, unlike agricultural digestates dominated by fine fractions [15]. Fine particles hinder P and organic matter separation without conditioning, whereas larger particles facilitate the more effective capture of solids and Ptot, even with simpler separation methods and lower polymer doses.

Potential indicators of membrane fouling in short-fibre digestates include organic matter (COD, oDM), solids (DM), divalent cations (Ca²⁺, reflected in water hardness), and P. These can promote both organic fouling and calcium-phosphate scaling on membranes. Scenarios using polyDADMAC (S4) and assisted mechanical separation (S5–S6) achieved lower residual DM, oDM, COD, P, and EC, with S6 performing better. This aligns with the reported SEM analysis of RO membranes, where S6 pre-treatment resulted in minimal scaling and fouling [36].

2.2.3. N Distribution

In S5, the digestate contained 401 mgNtot/kg_digestate and 73 mg NH₄-N/kg_digestate, indicating that most N was particulate or organic. Cloth filtration transferred 56 mg Ntot/kg_digestate and 21 mg NH₄-N/kg_digestate to the liquid, corresponding to 14% and 29% of the initial amounts. NH₄-N accounted for 38% of Ntot in the liquid fraction. In S6, centrifugation increased transfer to 91 mg Ntot/kg_digestate and 51 mg NH₄-N/kg_digestate, corresponding to 24% and 53%, respectively, of the initial amounts. NH₄-N comprised 56% of Ntot. It should be noted that percent distributions were calculated relative to the initial concentration. Therefore, Ntot and NH₄-N trends are not directly proportional. Overall, particulate and organic N dominated Ntot in the raw digestate.

Compared with S1–S2, where FeCl₃ was combined with polyDADMAC or starch, the NH₄-N fraction in the liquid fraction was similar, corresponding to 48–54% of Ntot, despite differences in chemical treatment. Chemical coagulation in S1–S2 enhanced particulate N separation from the liquid, increasing total N recovery, while NH₄-N remained largely in the liquid fraction. However, Yadav et al. [28] reported a higher NH₄-N recovery (90%) into the liquid fraction when treating swine manure using FeCl₃. This shows that, in principle, higher NH₄-N recovery in the liquid fraction can be achieved through chemical coagulation.

S3–S4 (polymer-only treatments) transferred limited NH₄-N to the liquid fraction. In S3, 7 mg NH₄-N/kg_digestate was recovered from 59 mg Ntot/kg_digestate, and, in S4, 5 mg NH₄-N was recovered from 154 mg Ntot/kg_digestate. NH₄-N accounted for only 12% (S3) and 3% (S4) of Ntot in the liquid after dewatering. This confirms that polymer-only treatment mainly removes particulate N and does not enhance NH₄-N recovery.

As a baseline, S0 transferred 50 mg Ntot/kg and 16 mg NH₄-N/kg_digestate to the liquid fraction (5% of Ntot and 6% of NH₄-N relative to the initial digestate). NH₄-N represented 32% of Ntot in the liquid fraction. This low recovery shows that additives improve particulate N separation. NH₄-N distribution is governed by solubility, the initial concentration available and dewatering efficiency.

Residual water in the solid fraction strongly influences NH₄-N recovery. S6 had the lowest residual water (180 g/kgSF), followed by S5 (429 g/kgSF), in the separated solid fractions. It enables a greater release of dissolved NH₄-N into the liquid. In contrast, S1–S4 retained more residual water (707–814 g/kgSF), limiting NH₄-N transfer. This accounts for the higher NH₄-N recovery observed in S6, followed by S5. The combination of chemical additives and efficient phase separation in S6 using centrifugation effectively maximised the transfer of dissolved NH₄-N into the liquid fraction.

2.3. Perspectives on Investigated Dewatering Scenarios for Model Short-Fibre Digestate

The experiments described in Section 3.1 and Section 3.2 demonstrated that the dewatering performance of short-fibre digestates is influenced by both the composition of the digestate and the separation method. Due to heterogeneous starting materials, direct scenario comparisons were limited. To allow consistent comparisons, a standardised model short-fibre digestate was defined (Table S3, Supplementary Materials) based on real digestates from biogas fermenters of different scales (10 L, 70 L, 1500 L), operating periods, and feedstocks, including deinking sludge, pellet sludge, and short-fibre rejects from German and European paper recycling mills [15].

Table S3 in the Supplementary Materials shows the compositional range of model digestate based on the percentiles from digestates analysed in 2022–2025 (Min, Q1, Median, Q3, Max). This model enables a comparison of liquid fraction quality across all scenarios.

The main goal of dewatering is to maximise liquid fraction recovery while maintaining a low DM content. The dewatered liquid fraction is intended for downstream membrane filtration. Using the model digestate and the mass distribution profiles from Section 3.1 and Section 3.2, liquid fractions in all scenarios (S0–S6) were assessed (Figure 6, values in Table S4 in the Supplementary Materials). This allows scenario comparison beyond the limits of Section 3.1 and Section 3.2.

The model digestate differs from conventional agricultural digestates in parameters affecting dewatering behaviour. DM ranges from 4.8 to 16% (median 10%), slightly higher than agricultural digestates, due to industrial fibre and recycled process waters. Organic matter indicators such as COD and TOC are higher and more variable (30,000–80,000 mg/L) compared with agricultural digestates (10,000 mg/L), reflecting heterogeneous industrial feedstocks.

Nutrients (Ntot: 356–1800 mg/L; NH₄-N: 40–1320 mg/L; Ptot: 187–1012 mg/L) show a broad variability, and C:N ratios range from 5 to 99. Electrical conductivity (EC) is elevated up to 20,500 µS/cm, likely from residual salts, while conventional digestates are generally 2000–10,000 µS/cm.

Metals and chemicals in short-fibre digestate originate from paper recycling, including calcium (Ca) from fillers, sodium (Na) from pulping chemicals, sulphur (S) from sulphonation/sulphite pulping, Zinc (Zn), cadmium (Cd), and copper (Cu) from inks and coatings, and PFAS from plastics/lamination. Additional metals accumulate through screening and wastewater residues, increasing the mineral content compared with agricultural digestates.

DM in the liquid fraction can be expressed per kg LF, reflecting liquid quality, or per kg digestate, reflecting total DM distribution. Both perspectives help evaluate dewatering efficiency and suitability for reuse. Scenario comparison (Figure 6) shows differences in DM per kgLF. From here on, the absolute values denote the median. S0 exhibited the highest DM (3.1 g/kg_LF), reflecting poor sedimentation. S1–S3 followed (2.7, 2.0, and 2.4 g/kg_LF, respectively), while S4–S6 achieved lower DM (0.9–1.1 g/kg_LF), indicating more efficient separation and higher-quality liquid fractions.

Expressing DM per kg digestate changes the perspective, as DM depends on the recovered liquid volume. S0, recovering the smallest fraction of the liquid fraction (180 g/kg_digestate), shows low DM (0.6 g/kg_digestate). In contrast, S5 and S6 recover larger volumes (FM-LF 484 and 686 g/kg_digestate, respectively), with similar DM per kg digestate (0.5 and 0.7 g/kg_digestate, respectively). S4, with moderate liquid recovery (FM-LF 120 g/kg_digestate), achieves low DM (0.1 g/kg_digestate). Consequently, S4–S6 produce low DM, despite differences in recovery volumes, with S6 providing the highest volume suitable for reuse.

Residual water in the solid fraction reflects liquid recovery. S0 retained the most water (726 g/kg_digestate). S1–S4 improved little (766–878 g/kg_digestate), indicating that flocculation alone is insufficient. Mechanical separation in S5 and S6 reduced residual water in solids to 422 and 224 g/kg_digestate, respectively, with DM concentrations in solids increasing to 182 and 295 g/kg_SF, respectively, comparable to industrial centrifuges.

COD in the liquid fraction was highest in S0 (445 mg/kg_digestate), intermediate in S1–S4 (94–175 mg/kg_digestate), and elevated in S5–S6 (1507 and 530 mg/kg_digestate, respectively). Mechanical separation mobilises more liquid-bound COD. Ptot shows similar trends, being low in S1–S4 (0–5 mg/kg digestate) and higher in S5 and S6 (71 and 50 mg/kg digestate).

N dynamics in the liquid fraction were influenced by both separation efficiency and liquid recovery. In S0, Ntot and NH₄-N were moderate (47 and 23 mg/kg_digestate, respectively) due to low liquid recovery (FM 180 g/kg_digestate). S1–S4 partially improved N separation, with Ntot ranging from 38.5 to 55.3 mg/kg_digestate and NH₄-N ranging from 11 to 31 mg/kg_digestate.

Scenarios S5 and S6 achieved the highest Ntot (121 and 202 mg/kg_digestate) and NH₄-N (116 and 212 mg/kg_digestate), while DM per kg_LF remained low (0.9–1.0 g/kg_LF). In S6, NH₄-N exceeded Ntot in the liquid fraction. This is a methodological artefact caused by the model digestate approach; Ntot and NH₄-N were treated independently using percentile distributions, allowing NH₄-N to be allocated more to the liquid fraction than Ntot.

Nevertheless, these results show that mechanical separation (filtration or centrifugation) increased liquid fraction recovery (FM 484–686 g/kg_digestate) and enhanced the transfer of soluble nitrogen (NH₄⁺) into the liquid fraction. Chemical additives alone did not produce high-quality liquid fractions.

The model digestate approach has limitations. The FM distributions from Section 3.1 and Section 3.2 were fixed, ignoring variations in the total digestate mass. Treating NH₄-N independently from Ntot can create minor inconsistencies. Thus, NH₄-N should be considered the primary indicator of dissolved N in the liquid fraction, while Ntot reflects N retained in the solids.

A preliminary energy and economic assessment of centrifugation combined with polyDADMAC demonstrates its practical feasibility at an industrial scale. A paper recycling mill fermenting 40,000,000 kg/a of dry short-fibre residue (53 Wt. % DM) generates 151,811,237 kg/a (8 Wt. % DM). Decanter centrifugation from Wuxi Dajiang Environmental Technology Co., Ltd. (Wuxi, China) could process 23 m³/h, of digestate, producing 37,563,290 kg of solids and 114,642,656 kg of liquid fraction annually. Continuous operation requires 30 kW, equivalent to 0.002 kWh/kg digestate. The PolyDADMAC (20 Wt. %) dosage is 0.0026 kg/kg digestate, totalling 395,735 kg at a cost of 309,000 €/a. These estimates indicate that centrifugation with polyDADMAC is energetically and economically feasible, producing clarified liquid suitable for downstream membrane treatment.

3. Materials and Methods

3.1. Digestate

For the experimental study in the present work, digestate from short-fibre residues was investigated. A total of eight lab-scale anaerobic fermenters, four main and secondary fermenters each, of type CSTR-10S, BPC Instruments AB, Lund, Sweden (10 L, 0–300 rpm, residence time 50 days), were operated at the Institute NOWUM-Energy, Aachen University of applied sciences, Germany. The lab-scale fermenters were fed with short-fibre residues such as deinking sludge, short-fibre sludge and pellet sludge for biogas production. These residues were generated during various stages of paper production from waste paper at four different paper recycling mills in Germany. A uniform mixture of these four digestates (Figure 7) from the lab-scale secondary fermenters was used for the experiments in this study. The digestate mixture was stored at room temperature.

3.2. Dewatering Scenarios

Seven dewatering scenarios were investigated, as shown in Figure 8. The baseline scenario (S0) consisted of simple sedimentation, while the four scenarios (S1–S4) combined pre-sedimentation with each of the selected chemical additives. Scenarios without pre-sedimentation applied mechanical dewatering with cloth filtration (S5) and centrifugation (S6) using the chemical additive identified as most effective among S1–S4. The composition of the digestate mixture during the investigation of scenarios is presented in Table 2.

3.2.1. Chemical Additives (With Pre-Sedimentation)

In scenarios S1–S4, raw digestate was subjected to pre-sedimentation (S0) for 24 h at room temperature (25 °C) to obtain a supernatant fraction (Figure 8a). The supernatant was then flocculated using four chemical additives or their combination: S1: 0.5 vol.% FeCl₃ + 0.25 vol.% polyDADMAC; S2: 0.5 vol.% FeCl₃ + 2.5 vol.% starch; S3: 0.75 vol.% Nanofloc^®; S4: 0.25 vol.% polyDADMAC. More details on the used additives are shown in Table S1 in the Supplementary Materials.

Jar tests were performed in triplicate, each with 450 mL of input experimental volume of supernatant from pre-sedimentation at 25 °C using a six-beaker apparatus. Following chemical addition, samples were stirred at 300 rpm for 2 min, then at 50 rpm for 10 min, and subsequently allowed to sediment for 120 min, with no changes found in terms of sludge volume index after 30 min. The resulting supernatant was decanted for analysis.

3.2.2. Mechanical Dewatering with Chemical Additive (Without Pre-Sedimentation)

In scenarios S5 and S6, raw digestate was directly flocculated using the most efficient chemical additive identified from S1–S4. In S5, the flocculated digestate was subjected to cloth filtration through a 1 µm filter. The flocculated digestate in S6 was centrifuged at 4000 rpm for 10 min. All experiments were conducted in triplicate with 450 mL input experimental volume of raw digestate at 25 °C.

3.3. Analytical Methods

Raw digestate and the liquid fractions from each stage of the separation experiments were analysed for the following parameters: COD (LCK 3114, according to DIN 38409-H41–H44) [37], Ptot (LCK 350, according to DIN EN 6878) [38], Ntot (LCK 238, according to EN ISO 11905-1), NH₄-N (LCK 303, according to DIN 38406 E5-1) [39], and water hardness (LCK 327, according to DOC312.53.94045). Measurements were performed using cuvette tests and the barcode reader DR2800 (Hach Lange GmbH, Düsseldorf, Germany). The pH value was determined using a pH electrode (Greisinger^® 600,770) in accordance with DIN EN ISO 10523-C5 [40], and electrical conductivity and total dissolved solids (TDS) were measured using an electrode (Model H 198311, Hanna Instruments^®, Woonsocket, RI, USA) following DIN EN 27888 [41]. The determination of DM and oDM was carried out according to DIN 15935 [42]. Only the raw digestate and the liquid fractions were analysed experimentally. The composition of the solid fractions was calculated through mass balance from the measured digestate and liquid fraction data.

3.4. Performance Calculations, Definitions and Statistical Analysis

3.4.1. Mass Distribution and Performance Calculations

The mass of each component such as fresh matter (FM), dry matter (DM), COD and nutrients (N_tot, NH₄-N, P_tot) in the initial digestate and the separated liquid fraction was calculated by multiplying the measured concentration, $C,$ (mg/L), determined using Hach kits, by the corresponding sample volume, $V,$ (L).

$$Mass=C·V$$

(1)

The percentage fraction of a component in the liquid fraction ($X_{LF}$) and solid fraction ($X_{SF}$) was calculated using

$$X_{LF}={\displaystyle \frac{C_{LF}·V_{LF}}{C_D·V_D}}$$

(2)

$$X_{SF}=1-X_{LF}$$

(3)

where $X$ is the fraction of the component of interest (e.g., COD, N_tot, NH₄-N, P_tot); $C_D$ and $V_D$ represent the concentration [mg/L] and volume [L] of a component in the initial digestate, respectively; and $C_{LF}$ and $V_{LF}$ represent the concentration [mg/L] and volume [L] in the separated liquid fraction, respectively. The composition of the solid fraction, $X_{SF}$, was calculated by difference using a mass balance approach.

For DM, the distribution was calculated based on the measured FM and DM concentrations using Equations (4) and (5):

$${DM}_{LF}={\displaystyle \frac{{FM}_{LF}·{DM}_{LF}}{{FM}_D·{DM}_D}}$$

(4)

$${DM}_{SF}=1-{DM}_{LF}$$

(5)

where ${FM}_D$ and ${FM}_{LF}$ are the fresh matter of the original digestate before separation and liquid fraction, respectively [g], and ${DM}_D$ and ${DM}_{LF}$ are the DM concentrations in the digestate and liquid fraction, respectively. ${DM}_{SF}$ was calculated by difference using a mass balance approach. The solid fraction (${DM}_{SF}$) represents the fraction of total DM transferred into the solid fraction and can also be referred to as the DM separation efficiency (ηDM).

FM was directly determined gravimetrically as the total (wet) mass of the sample without further treatment. DM represents the residual mass after drying the FM sample according to DIN 15935.

3.4.2. Definitions

Definitions of mass terms that are used in the results Section 2 are listed in

Table 3. Definition of mass terms.

Parameter	Definition	Units	Notes
FM	Total mass of a fraction including both wet and dry contents	g/kg initial material (SPSD 1 or digestate)	FM-LF 2 = water in LF + residual DMLF FM-SF 3 = DMSF + residual water in SF
DM	Only dry solids in the fraction	g/kg initial material or g/kgLF or g/kgSF	DMLF = dry solids in LF DMSF = dry solids in SF
DM-free LF	Only water content in liquid fraction	g/kgLF or g/kgSF	DM-free LF = FMLF − DMLF
Residual water in SF	Only water content in solid fraction	g/kgLF or g/kgSF	Residual water in SF = FM-SF − DM-SF

¹ SPSD—supernatant of pre-sedimented digestate, relevant for Section 3.1; ² LF—liquid fraction; ³ SF—solid fraction.

.

In this study, FM refers to the measured wet mass of each fraction (sample), while DM represents the corresponding dry solid content.

4. Conclusions

This study evaluated dewatering strategies for short-fibre digestates from paper mills to maximise liquid fraction recovery and minimise DM content for downstream reverse osmosis. Sedimentation alone, or as a pre-step in scenarios S1–S4, proved ineffective. Although chemical conditioning after sedimentation improved DM separation, it did not substantially enhance liquid fraction recovery, implying that insufficient primary sedimentation limits the overall dewatering performance. Among additives tested in scenarios with pre-sedimentation, polymer-based flocculation with polyDADMAC alone outperformed FeCl₃, starch, and Nanofloc in reducing solid transfer to the liquid fraction. Mechanical force combined with chemical conditioning was essential for effective phase separation. Centrifugation combined with polyDADMAC achieved the highest liquid fraction recovery and lowest DM content, followed by cloth filtration. The composition of the solid fraction was not measured directly but was estimated through mass balance from the initial digestate and liquid fraction, representing a methodological limitation with some uncertainty in solid-phase interpretation.

The model short-fibre digestate, developed from digestate compositions analysed over four years of biogas fermenter operation, enabled a systematic comparison of liquid fraction quality across scenarios. The results indicate that short-fibre digestates are comparable to conventional agricultural digestates in terms of DM and nutrient concentrations but differ substantially in mineral content and trace contaminants originating from paper recycling processes. While Ptot and COD were largely transferred into the solid fraction across all scenarios, NH₄-N distribution depended strongly on separation efficiency and residual water in the solids. Scenarios achieving lower residual water in the solid fraction resulted in a higher NH₄-N transfer to the liquid fraction.

Overall, mechanically assisted dewatering with chemical conditioning is critical for high liquid fraction recovery with low DM content suitable for membrane-based water reuse. Future studies should investigate other additives, including FeCl₃, with centrifugation to compare performance with polyDADMAC.