Field Investigations and Service Life Assessment of Geosynthetic Filters in Tidally Influenced Revetments in Cases of Ochre Clogging

Abstract

In recent decades, there have been individual cases of damage to the revetments of the German North Sea estuaries due to clogging by precipitated ochre products. This process is defined as ochre clogging and has been extensively explained in the literature. The primary aim of the investigations was to better understand the clogging process under in situ conditions and the causative environmental conditions surrounding the filters. Extensive in situ investigations were therefore carried out. It was found that a permeability reduction in geotextile filters can be strongly accelerated by ochre clogging, which appears to be a biogeochemical process. This describes a combined action of the chemical precipitation of iron and manganese, precipitation by microorganisms, and physical clogging. A further aim of this study was to establish limit values for the decisive ochre clogging parameters, which could be used to quantify the susceptibility to ochre clogging. It was shown that the determination of the iron and manganese content of the groundwater, as well as the redox capacity of the groundwater, is sufficient to assess the tendency for ochre clogging. To minimise the negative impact on filter performance, recommendations for an adapted filter design have been developed as a guide for planners.

1. Introduction

In Germany, approximately 340 km of all rivers and waterways are under the tidal influence of the North Sea. Where extensive riverbank protection is required, the riverbanks are generally secured with heavy technical revetments consisting of a permeable armour layer and a filter layer [1] (see Figure 1). These are exposed to various mechanical and transient hydraulic loads, such as wind- and ship-induced waves, water level fluctuations, and the resulting erosion processes, throughout their service lives. Therefore, revetments on inland waterways must be robustly designed to attain a service life of at least 50 years. Coastal revetments, on the other hand, are designed for a service life of up to 100 years [2]. A key component of revetments is the filter layer, which contributes to the stability of the revetment. It retains the fine material of the subsoil while maintaining the hydraulic conductivity to prevent the build-up of pore-water pressure [1]. The permeability of a filter can decrease with ageing due to physical and biochemical mechanisms [3,4,5]. The severe reduction in permeability exerted high hydrostatic pressure on the armour layers of various revetments in the German North Sea estuaries of the Ems and Weser, causing the armour layers to uplift. This has been described, for example, by Abromeit [6] in the case of revetment damage, in which the permeability was reduced after 8 to 9 years to such an extent that an essential filter function, i.e., permeability, was no longer fulfilled. This and other cases of damage have been attributed to ochre formation [6,7].

Ochre clogging, in general, has been extensively described in the literature and is summarised below based on the literature review [3,4,5,8,9,10,11,12,13,14,15]. Ochre formation is a chemical and/or biological process involving the precipitation of iron, manganese, and other substances in various chemical compounds [16]. The most important prerequisite for ochre formation, which is initiated by oxidation, is the presence of oxygen and/or iron or manganese bacteria in the soil, groundwater, or soil water [11], as well as dissolved iron or manganese. Mendonca et al. [13] described the problem and process as an increase in pore-water pressure in the soil or filter element associated with instability of the riverbank soil structure or a change in flow direction. These conditions can occur where, as a result of a change in flow direction, groundwater containing iron or manganese comes into contact with oxygen in the air or with oxygen-rich surface water. This can include the geotextile or granular filters in the revetments of tidal waterways [9]. The accumulation of the ochre-clogging products on the filter structure can be classified as internal clogging [9]. This can result in damage to the revetment if it fails due to the swelling of the filter (see Figure 1). Weidner [5] and Weidner et al. [8] investigated vertical granular filter wells in open pit mines, which are also prone to clogging due to hydrochemical conditions. As part of their work, the properties of various granular well-pack materials were evaluated concerning long-term Fe-clogging affinity. A comprehensive literature review on physical and geochemical processes in coastal areas related to iron precipitation in porous media was carried out by Cao et al. [15]. They described several approaches to model Fe precipitation. Mendonca et al. [13] carried out a study of ochre clogging of geotextiles, taking into account different levels of iron concentration, available dissolved oxygen, and pH value. The fundamental importance of iron bacterial activity, geotextile type, and environmental conditions for ochre formation was demonstrated in the laboratory tests. This study was later extended by Correia et al. [4]. The results showed that ochre formation decreased significantly in the submerged filters due to reduced oxygen availability for aerobic microbial activity. Liu et al. [14] conducted a series of column tests to study the mechanism of polyurethane foam to reduce the chemical clogging of nonwoven geotextile filters. They installed a polyurethane foam under the geotextile to reduce the oxygen availability in the filter layer. As a result, the extent of the chemical clogging of the geotextile was reduced.

As the cases of revetment damage occur locally and fully intact revetments can be found in close proximity to the damaged areas, the environmental conditions surrounding the filters need to be investigated in addition to the structural and filter properties. The environmental conditions were analysed in preliminary studies using external data sources [9]. Some filter technological adjustments implemented to slow down the accumulation of ochre-clogging products were described in preliminary experimental studies [3].

The primary aim of this study was to investigate the chemical (abiotic) and (micro)-biological parameters, as well as the hydraulic and structural boundary conditions, in estuaries with ochre formation phenomena, in order to record the environmental conditions surrounding ochre-clogged filters. Therefore, the results of this study are divided into a comprehensive description of the study areas and field observations. The environmental conditions surrounding the ochre-clogged filters are then described with reference to extensive in situ sampling and long-term studies. Criteria describing the ochre clogging tendencies of granular and geotextile filters are derived from these investigations. Finally, a workflow is proposed for the design of geotextile filters where they are susceptible to ochre clogging. This workflow is intended to make a significant contribution to the functional use of geotextile filters, even when they are susceptible to ochre clogging. The aim is to achieve the desired technical service life and economic and ecologic optimisation.

2. Study Areas and Methods

Based on an extensive literature review, enquiries about known damaged areas were made to all relevant national bodies (namely, the Federal Waterways and Shipping Administration (WSV)) whose areas of responsibility included the German North Sea estuaries. From these and a review of existing expert reports (unpublished; written by the Federal Waterways Engineering and Research Institute (BAW)), it was possible to compile a list of all known areas where ochre clogging damage had occurred. In this study, the current instances of damage could only be localised within the area of the Ems estuary north of Herbrum and in the area of the Weser estuary near Neuenkirchen. As a result, the following two study areas were defined for the investigations described: one at the Ems estuary and one at the Weser estuary.

2.1. Study Area: Ems Estuary

The study area was located within the larger hydrogeological area of the north and central German unconsolidated aquifer and the sub-area of the Hunte-Leda Moor Lowlands [17]. The sandy upper unconsolidated aquifer of the Moor Lowlands has high permeability and is characterised by siliceous or siliceous/organogenic rock. The study area was locally limited to the area of the Ems estuary between Herbrum and Papenburg (see Figure 2). Revetment damage was mainly found in the area of Rhede (Ems). The Ems estuary was subjected to a strong tidal range of 3.35 m (Mn) at the Rhede (Ems) gauge in the period of 2013–2022 [18]. The mean water levels in the Ems estuary at the Rhede (Ems) gauge for the same period are shown in

Table 1. Mean water levels or tidal data, 2013–2022 [18,20].

Water Level	Tide Gauge Rhede (Ems)	Tide Gauge Elsfleth
MHW [MAMSL]	1.96	2.19
MLW [MAMSL]	−1.39	−1.72
Mn [m]	3.35	3.91
RD [h]	3:21	5:50
FD [h]	9:03	6:35
MLWS [MAMSL]	−1.52	−1.84

Note: MHW: mean high water, MLW: mean low water, Mn: mean range of tide, RD: rise duration, FD: fall duration, MLWS: mean low water spring tide.

.

2.2. Study Area: Weser Estuary

The study area was located within the larger hydrogeological area of the north and central German unconsolidated aquifer and the sub-area of the Lower Weser Marshes [17]. The sandy upper unconsolidated aquifer of the Lower Weser Marshes has high permeability and is considered saline. The rock is siliceous, siliceous/carbonate, or siliceous/organic. The study area was locally limited to the area between Bremen and Brake (Unterweser) (see Figure 2). Revetment damages were mainly found near Neuenkirchen (Schwanewede). The Weser estuary was characterised by a strong tidal range of 3.91 m (Mn) at the Elsfleth gauge in the period of 2013–2022 [20]. The mean water levels of the Weser estuary at the Elsfleth gauge for the same period are also shown in Table 1.

2.3. Research Programme

In order to investigate the chemical (abiotic) and (micro)biological parameters, as well as the hydraulic and structural boundary conditions, in ochre-clogged revetments, extensive data sampling was necessary. The characteristics of the groundwater (GW) and river water (RW) were determined through quarterly sampling over a period of one year. In addition, the relevant revetment components, i.e., the ochre-clogged filters and the riverbank material, where the armour layer was already damaged, were removed and analysed. Other external data were used to complete the data.

Table 2. Research programme and investigations conducted.

Station	Investigations
GW1-E, GW2-E, GW1.1-W, GW1.2-W, GW2.1-W, GW2.2-W (groundwater)	- Analysis of hydraulic and groundwater quality data (2000–2022); - Quarterly collection of hydraulic and groundwater quality data (Q2/2022–Q1/2023; i.e., pH value, dissolved oxygen (DO), redox potential (EH value), temperature, electric conductivity (EC), rH value, dissolved iron (Fe2+), ferric iron (Fe3+), total iron (Fetot), sodium (Na+), nitrate (NO3-N), chloride (Cl−), dissolved manganese (Mn2+), total manganese (Mntot), dissolved organic carbon (DOC), total organic carbon (TOC)); - DNA extraction and amplicon sequencing to determine types of bacteria in groundwater.
RW-E RW-W (river water)	- Analysis of river water quality data (2000–2022); - Quarterly collection of hydraulic and river water quality data (Q2/2022–Q1/2023; i.e., see line above, redox potential (EH value) and rH value not relevant for river water).
E1, E2, E3, W1, W2 (aquifer material)	- Determination of geotechnical and filter parameters (i.e., particle size distribution, hydraulic conductivity (kT,10 °C value), loss on ignition); - Determination of iron and manganese content; - Electron microscopy observations (i.e., scanning electron microscopy (SEM), energy-dispersive X-ray spectroscopy (EDXS)).
E1, E2, E3, W2 (geotextile filter)	- Determination of geotechnical and filter parameters (i.e., hydraulic conductivity (k value) or velocity index (VIH50), loss on ignition, mass per unit area); - Determination of iron and manganese content; - Electron microscopy observations (i.e., scanning electron microscopy (SEM), energy-dispersive X-ray spectroscopy (EDXS)); - DNA extraction and amplicon sequencing to determine types of bacteria in geotextile.
W1 (granular filter)	- Determination of geotechnical and filter parameters (i.e., particle size distribution, hydraulic conductivity (kT,10 °C value), loss on ignition).

gives a detailed overview of the research programme.

Incubation experiments were carried out with some geotextile samples to determine the proportion of biological iron oxidation. For this purpose, sodium azide (NaN₃) was added to parts of the geotextile samples to inhibit the activity of the iron-oxidising bacteria. Other parts were spiked with hydrochloric acid (HCl) to stop both biological and chemical iron oxidation. The proportion of biological and chemical oxidation could have been determined by comparing the Fe²⁺ concentration at different times. The results were insufficient, and no further description of the incubation experiments is therefore provided.

2.4. Groundwater Investigations

Groundwater measurement stations from previous evidence conservation measures were existent in the study areas. Prior to the sampling campaign, the measurement stations were thoroughly flushed with river water. Pump samples were taken during the sampling. Comparable flow conditions were always considered, i.e., exfiltration conditions were always present, and the groundwater flowed into the river water. Accordingly, the samples were always taken when the water level was low. At two measurement stations (GW2-E and GW1.2-W), continuous pumping was not possible, so qualified samples were taken after multiple flushes (≥5 times) of the water column. Groundwater sampling was carried out by certified samplers. The sampling dates are shown in

Table 3. Dates of groundwater and river water sampling.

Stations	No. of Sampling
	1	2	3	4
GW-E (all stations)	6 May 2022	16 August 2022	29 October 2022	25 March 2023
RW-E
GW-W (all stations)	4 May 2022	3 August 2022	14 November 2022	14 February 2023
RW-W

. All groundwater analyses were performed according to the national (German) and European standards. The pH value, dissolved oxygen (DO), redox potential (E_H), temperature, electrical conductivity (EC) and dissolved iron (Fe²⁺ and Fe³⁺ ions) were determined directly onsite (see Figure 3a). For the determination of the other parameters, the samples were preserved according to the aforementioned standards and analysed in the laboratory.

2.5. River Water Investigations

The river water samples were taken as random samples in the main stream at low tide. The pH, DO, temperature, EC, Fe²⁺ ions, and Fe³⁺ ions were determined directly onsite. As in the case of the groundwater investigations, the samples were preserved according to the standards for the determination of the other parameters and analysed in the laboratory. The sampling data corresponded to those of the groundwater analyses (Table 3).

2.6. Aquifer Material Sampling

The aquifer material (or riverbank subsoil) could only be sampled at low tide. First, the filter layer was removed (see below). This was performed with extreme care to ensure that the contact layer between the filter and the aquifer material remained intact. Various undisturbed samples were then taken to carry out the analyses listed in Table 2 (see Figure 3b). Some samples were immersed in epoxy resin to produce a 30-micron polished section.

The particle size distribution was determined according to DIN EN ISO 17892-4 [21]. To determine the k_{T,10 °C} value, undisturbed samples were used in accordance with DIN EN ISO 17892-11 [22]. For the determination of the loss on ignition according to DIN EN 17685-1 [23], the samples were annealed at 550 °C to a constant mass. The iron and manganese contents were determined in an accredited laboratory via inductively coupled plasma mass spectrometry (ICP–MS) according to DIN EN ISO 17294-2 [24]. The samples were prepared via aqua regia digestion. Electron microscopic observations (see Figure 3c) and high-spatial-resolution chemical analyses were performed on a JEOL 6510-LA scanning electron microscope (SEM; Akishima, Japan) equipped with a tungsten filament (20 kV acceleration voltage). Energy-dispersive X-ray spectroscopy (EDXS) was performed with the same settings, using the internal JEOL software “Analysis Station” (version 4, 13, 0, 79).

2.7. Granular Filter Sampling

The methods used to determine the geotechnical and filter parameters of the granular filter were similar to those used for the aquifer material. Only the sampling itself was different, as undisturbed sampling was not applicable due to the grain size of the granular filter.

2.8. Geotextile Filter Sampling

The geotextile filter samples were taken at the same time as the granular samples. Organic and inorganic adhesions were initially not removed, in order to obtain almost undisturbed samples. Some of the samples were then prepared in the laboratory, so that half of the samples were labelled as undisturbed and half as disturbed. The disturbed samples were prepared in several steps. The aim was to remove mobile particles from the textiles. First, they were dried at 45 °C and then carefully vacuumed and saturated in a heat bath (45 °C). The samples were then dried again, vacuumed, and treated briefly (2 min) in an ultrasonic bath. They were then dried and vacuumed again. The samples were then prepared for the required analyses via punching. Some samples were immersed in epoxy resin to produce a 30-micron polished section.

The permeability or velocity index was determined according to ASTM D 5887 [25] and DIN EN ISO 11058 [26]. For the determination of the loss on ignition, the sample was annealed until the mass was constant (see above). The mass per unit area was determined in parallel with all geotextile analyses according to DIN EN ISO 9864 [27]. The iron and manganese contents were also determined via ICP–MS (see above). The samples were prepared via aqua regia digestion. Microwave-assisted digestion was used for selected highly loaded, undisturbed samples. The methods adopted for the electron microscopic observations and EDXS were the same as described above.

For DNA extraction, the textile samples were cut into small fragments using sterile scissors, and the total genomic DNA was extracted using the Qiagen DNA Power Soil Pro Kit (Venlo, The Netherlands), following the manufacturer’s instructions, with some adjustments. Specifically, 25 μL of proteinase K (22 mg/mL) and 10 μL of RNase were added, and the mixture was incubated at 37 °C for 1 h. The gene for the 16S rRNA, a common phylogenetic marker for archaea/bacteria, was amplified using primer pairs 515F-Y (5′-GTGYCAGCMGCCGCGGGTAA) and 926R (5′-CCGYCAATTYMTTTRAGTTT), targeting the hypervariable regions V4 and V5, as described by Parada et al. [28]. Subsequently, library preparation, sequencing, and data analysis were performed by Microsynth AG (Balgach, Switzerland). The PCR products were sequenced using a v2 500 cycle kit on the Illumina MiSeq platform (San Diego, CA, USA). Further bioinformatic analysis was performed as described by Bakenhus et al. [29].

3. Results

3.1. Examination of River Water and Groundwater

Table 4. Difference in water level (Δh_W = h_GW−h_RW) during groundwater sampling.

Stations	No. of Samplings
	1	2	3	4
GW1-E	0.87	1.03	0.41	1.05
GW2-E	0.28	0.59	1.19	−0.15
GW1.1-W	2.37	1.34	1.51	1.67
GW1.2-W	−0.60	1.99	2.05	2.22
GW2.1-W	n/a	1.56	1.8	1.04
GW2.2-W	n/a	0.66	1.73	2.19
all values in m

shows the differences in the water level during the groundwater sampling. Positive values indicate the exfiltration of the groundwater into the river water, and negative values indicate infiltration. The problems associated with ochre clogging, including damage to the armour layer, occur mainly when exfiltration occurs and the groundwater flows into the estuary at low water levels. A maximum difference in the water level of 1.19 m was recorded for the gauges along the Ems estuary during the study period. In the Weser estuary, a maximum difference in the water level of 2.37 m was recorded during the sampling period. Taking into account the mean water levels in the study areas (Table 1) and the mean position of the groundwater surface in the vicinity of the study areas at the Ems and Weser estuaries, a mean water level difference according to the EAU (2020) [30] can be roughly determined. The following approximation can be used:

$$\mathrm{Difference} \mathrm{in} \mathrm{water} \mathrm{level} (\mathrm{BS}–\mathrm{P}): {\mathsf{\Delta }\mathrm{h}}_{\mathrm{w}}=\mathrm{a}+0.30 \mathrm{m}+\mathrm{d}$$

(1)

• a: 0.5 × (h_MHW−h_MLW);
• d: h_MLW−h_MLWS;
• h_MHW: mean high water level;
• h_MLW: mean low water level;
• h_MLWS: mean low water spring tide level.

The position of the groundwater surface can be specified as h_GW,mean(E) = 0.89 m a.s.l. [31] near the Ems estuary (E) and h_GW,mean(W) = 1.67 m a.s.l. [32] near the Weser estuary (W) for the time series of 2013–2022. Accordingly, the mean difference in the water level according to Equation (1) in the study area at the Ems estuary (E) for Δh_W(E) = 2.11 m and that for the area at the Weser estuary (W) for Δh_W(W) = 2.38 m.

Table 5. Relevant parameters derived via sampling (mean value + standard deviation).

Parameter	Unit	GW1-E	GW2-E	GW1.1-W	GW1.2-W	GW2.1-W	GW2.2-W	RW-E	RW-W
pH value	-	6.8 ± 0	6.7 ± 0.1	7.0 ± 0	7.3 ± 0.3	7.3 ± 0.1	7.2 ± 0.1	7.7 ± 0.1	7.7 ± 0.4
EH value	mV	172.5 ± 51.2	−8.5 ± 44.6	15.6 ± 85.5	43 ± 98.3	43 ± 92.4	27.5 ± 103.9	217.5 ± 22.8	n/a
Temp.	°C	11.3 ± 1.1	12.1 ± 1.9	12.1 ± 0.9	15.1 ± 2.8	14.8 ± 1.3	14.5 ± 4.1	14.5 ± 5.9	14.2 ± 5.9
DO	mg/L	3.3 ± 1.1	3.8 ± 1.6	0.9 ± 1.2	5.2 ± 2.8	3.8 ± 2.8	2.5 ± 1.6	9.2 ± 2.4	10.0 ± 2.1
EC	µS/cm	990 ± 108	862 ± 107	1730 ± 12	1743 ± 78	1347 ± 5	1697 ± 61	690 ± 58	1153 ± 263
rH value	-	19.7	13.1	14.6	16.1	16.1	15.4	23.0	n/a
Fe2+	mg/L	≤0.1 ± 0	10.1 ± 5.6	25.4 ± 2.9	23.7 ± 5.7	11.9 ± 5.4	23.6 ± 1.4	≤0.1 ± 0	≤0.1 ± 0
Fe3+	mg/L	≤0.1 ± 0	≤0.1 ± 0	1.1 ± 1.4	0.4 ± 0.3	0.6 ± 0.2	1.1 ± 0.9	≤0.1 ± 0	≤0.1 ± 0
Fetot	mg/L	0.1 ± 0	9.9 ± 5.2	26.3 ± 0.8	27.8 ± 6.0	10.4 ± 6.6	14.6 ± 7.8	38.3 ± 22.3	0.5 ± 0.7
Mn2+	mg/L	≤0.1 ± 0	0.5 ± 0	4.4 ± 0.1	3.8 ± 0.5	2.9 ± 0	3.2 ± 0.3	≤0.1 ± 0	≤0.1 ± 0
Mntot	mg/L	≤0.1 ± 0	0.5 ± 0	4.5 ± 0	3.9 ± 0.4	2.9 ± 0.1	3.3 ± 0.2	1.7 ± 0.8	≤0.1 ± 0
Na+	mg/L	56.0 ± 12.2	85.8 ± 23.1	89.1 ± 3.6	72.9 ± 6.6	117 ± 3.7	147 ± 2.2	44.5 ± 10.4	100.9 ± 34.6
Cl−	mg/L	82.5 ± 49.1	99.9 ± 38.7	151.8 ± 4.4	158.8 ± 9.2	219.0 ± 8.5	317.3 ± 20.9	82.5 ± 21.3	193.8 ± 60.1
Nitrate	mg/L	2.9 ± 1.8	≤0.1 ± 0	≤0.1 ± 0	0.2 ± 0.1	≤0.1 ± 0	≤0.1 ± 0	0.8 ± 0	2.6 ± 0.7
DOC	mg/L	5.6 ± 0	5.8 ± 0	11.9 ± 0.2	10.6 ± 0.7	5.3 ± 0.5	5.1 ± 0.8	8.6 ± 0	4.5 ± 0.7
TOC	mg/L	5.7 ± 0.1	5.9 ± 0.5	12.6 ± 0.6	17.2 ± 6.2	4.7 ± 1.3	5.7 ± 0.9	28.2 ± 18.1	10.3 ± 5.4

shows the summarised results of the river water and groundwater investigations. In their theoretical preliminary study, Tophoff et al. [9] described relevant limit values for the decisive ochre clogging parameters (DOCP) of the groundwater, namely, the concentration of Fe²⁺ ions (>0.2 mg/L), DO (<0.7 mg/L), redox potential E_H (<0 mV), and rH value (<17). When comparing the limit values with the actual values of the groundwater behind the revetment, it is noticeable that the parameters do not fully match at the measurement stations where the ochre formation actually occurred. In particular, the DO is not below the limit at any of the measurement stations and the redox potential is not consistently less than or equal to 0.

The amplicon sequencing results for the groundwater sample are shown in Figure 4a. The figure lists all genera with a relative abundance > 1% and shows a highly diverse range of bacteria.

3.2. Examination of Revetment Components

Mainly, the aquifer material and filter materials were considered in the analysis of the revetment components. The armour layer itself was only considered in terms of the design and mass per unit area.

Table 6. Overview of the sampled revetment components.

Station	Date of Sampling	Year of Construction	Armour Layer	Filter	Aquifer Material
E1	17 August 2022	1987	45 cm sandstone riprap (weight per unit area: 550–600 kg)	Geotextile A	Sand, 0.1–0.5 mm
E2	17 August 2022	1983		Geotextile B	Sand, 0.1–0.6 mm
E3	23 November 2022	1987		Geotextile B	Sand, 0.1–0.5 mm
W1	04 August 2022	Unknown	22 cm reinforced concrete slabs + 10 cm drainage (weight per unit area: 670 kg)	Granular filter	Sand, 0.15–1.5 mm
W2	04 August 2022	1983/84		Geotextile C	Sand, 0.15–1.5 mm

provides an overview of the revetment components investigated.

3.3. Aquifer Material

The filter parameters of the aquifer material are given in

Table 7. Filter parameters and iron and manganese contents of aquifer material samples.

	E1	E2	E3	W1	W2
Grain size [mm]	0.1–0.5			0.15–1.5
Coefficient of uniformity [Cu]	2.1–2.4			2.2
Hydraulic conductivity (kT,10 °C value) [m/s]	1.2 × 10−5			4.5 × 10−5
Loss on ignition (LOI) [%]	0.4			0.3
Dry density [g/cm3]	1.69			1.75
Iron content [g Fe/kg DS]	10.1	3.2	33.2	7.0	1.6
Manganese content [g Mn/kg DS]	0.9	0.9	1.2	0.2	0.2

. They were comparable within each study area, so they are summarised in part for each study area. Uniform sands (Sa) were found in both areas. The iron and manganese contents varied greatly depending on the sampling point. These values are given in Table 7. The amounts of both iron and manganese were higher in the aquifer material of the Ems estuary than in that of the Weser estuary.

Figure 5a shows an undisturbed sample of the Ems aquifer material at sampling point E1. Black-brown structures can be seen, particularly in the contact zone with the geotextile (upper area). Further down, several brownish-coated grains are visible. Electron microscopy (EDXS) found that these structures were high in Si, Fe, and Mn relative to the grain particles.

3.4. Granular Filter

A granular filter was found exclusively in the Weser estuary (W1; see Figure 5b). The filter sample showed a very heterogeneous, slightly silty, sandy gravel (Gr, sa, si), with a grain size distribution of <0.063–63 mm. The coefficient of uniformity C_u was 61.1. High hydraulic conductivity (k_{T,10 °C} value) of 7.0 × 10⁻² m/s was empirically determined according to the method of Seiler [33]. The loss on ignition (LOI) of the granular filter material was 1.9%. The iron and manganese content could not be determined representatively due to the sizes of the individual grains and the maximum permissible digestion quantity.

3.5. Geotextile Filter

Table 8. Material and filter parameters of geotextile filters.

		Unit	E1	E2	E3	W2
	Geotextile	-	A	B	B	C
Virgin	Mass per unit area	g/m2	>1700	1800	1800	1400
	VIH50	mm/s	17	10	10	21
	k value	m/s	4.1 × 10−3	2.6 × 10−3	2.6 × 10−3	3.9 × 10−3
	Loss on ignition	%	≥99.9	≥99.9	≥99.9	≥99.9
Loaded (disturbed)	Mass per unit area	g/m2	3408	3470	2316	1527
	k value	m/s	1.3 × 10−6	1.1 × 10−6	1.4 × 10−6	1.0 × 10−6
	Residual permeability	%	0.032	0.042	0.054	0.026
	Loss on ignition	%	FL: 45.9 AL: 79.4	FL: 45.3 AL: 79.4	-	FL: 25.8 AL: 56.8
	Iron content	g/m2	51.1	55.4	60.9	12.6
	Manganese content	g/m2	87.9	109.5	-	3.5
Loaded (undisturbed)	Mass per unit area	g/m2	7382	6884	6984	3672
	k value	m/s	1.3 × 10−7	3.4 × 10−8	-	2.7 × 10−7
	Residual permeability	%	0.003	0.001	-	0.007
	Loss on ignition	%	FL: 32.8 AL: 54.3	FL: 48.4 AL: 39.6	-	FL: 51.1 AL: 92.0
	Iron content	g/m2	273.0	162.7	313.5	75.9
	Manganese content	g/m2	466.1	253.1	-	11.5

Note: FL: filter layer. AL: additional layer.

summarises the material and filter parameters of the geotextile filters. The table shows various parameters for the unloaded (virgin) and loaded (disturbed and undisturbed) samples. The same textiles were used in E2 and E3. All textiles, A, B, and C, were composites consisting of a filter layer (FL) to fulfil the filter function and an additional layer (AL) to increase the frictional resistance on the aquifer material (see Figure 5c). It was clear that the loaded (undisturbed) samples weighed up to 4.3 times more than the virgin geotextiles due to the accumulations and adhesions. The loaded (disturbed) samples also had up to two times larger mass per unit area. The hydraulic conductivity (VI_H50 and k value) decreased significantly. The reduction in hydraulic conductivity is shown in Table 8 as the residual permeability relative to the virgin (or unloaded) k value. The LOIs decreased to varying degrees, indicating that significant inorganic inclusions were present in the textiles. The iron and manganese contents of the geotextile filters also varied. It was clear that the iron and manganese contents in the textiles from the Ems estuary were significantly higher than in those from the Weser estuary.

Figure 5. Revetment components. (a) An undisturbed sample (polished section) of the aquifer material (E1) at 250× magnification. The upper area shows the contact zone with the geotextile. (b) The granular filter before sampling (W1). (c) An undisturbed geotextile sample before preparation (W2).

Figure 5. Revetment components. (a) An undisturbed sample (polished section) of the aquifer material (E1) at 250× magnification. The upper area shows the contact zone with the geotextile. (b) The granular filter before sampling (W1). (c) An undisturbed geotextile sample before preparation (W2).

Figure 6 shows details of the geotextiles after sampling and preparation, magnified up to 100 times. It is clear that the disturbed samples (see Figure 6a,d), from which the mobile particles have been largely removed, show significantly fewer adhesions than the undisturbed samples (see Figure 6b,c,e,f). The filter layers have considerably thinner filaments (30–40 µm) than the additional layer (120–150 µm). Figure 6c,e show that many granular particles are trapped in the geotextile structure. It can be assumed that the filter layer of the undisturbed geotextile W2 (see Figure 6f) is less clogged than the filter layer of the undisturbed geotextile E1 (see Figure 6c). EDXS investigations were carried out to determine the (chemical) compositions of the deposits and adhesions. Representative extracts of the results are shown in Figure 7. EDXS showed that the structures partially enveloping the filaments contained mainly C, O, Si, Fe, and Mn. Traces of Al, P, K, and Ca were also found. The elemental distributions of the samples (E1: see Figure 7a;W2: see Figure 7b) were comparable, but there were some differences. Mn was not found in W2, while the amount of Al detected was higher in W2 than in E1. The amount of Fe detected was higher in E1 than in W2. Figure 8 shows an elemental distribution image, or SEM–EDXS mapping, of an undisturbed sample from E1. Here, the distribution of the elements Al, C, Fe, Mn, O, and Si is shown with the additional SEM image. Again, the transition from the filter layer to the additional layer is shown. It can be seen that the adhesion is composed of different elements (including Al, Fe, Mn, O, and Si). A trapped quartz particle (<0.1 mm) can be seen at the centre of the image.

The microbiological examination revealed iron oxide particles (approximately 5 μm) with attached microbial cells. In addition, individual bacterial cells (approximately 1 μm) and isolated diatoms (approximately 20 μm) were always present in these samples. The amplicon sequencing results (see Figure 4b) showed a large proportion of prokaryotes (bacteria and archaea) of low abundance below 1% for both the Weser (W2) and Ems (E3) geotextiles. Additionally, the Weser geotextile contained a large proportion of genera with no cultured representatives. The reliability of the amplicon sequencing was confirmed by the consistent results obtained in the sequence analysis of 12 replicates from the Weser geotextile.

4. Discussion

As a preliminary note, it should be mentioned that the complicated sampling and highly resource-demanding microbiological analysis did not permit an increased number of tests. The results are therefore subject to statistical uncertainty.

4.1. Environmental Conditions of Filters

The environmental conditions surrounding the filters, particularly the geotextile filters, were derived through the mentioned investigations of the groundwater, river water, and aquifer material.

4.2. River Water

As expected, the surface water of both estuaries was highly oxic (DO > 9 mg/L). Accordingly, no dissolved iron or manganese ions were present. In the surface water of the Ems (RW-E), high levels of ferric precipitates were found (Fe_tot = 38.3 ± 22.3 mg/L). The values were significantly lower in the surface water of the Weser (RW-W; Fe_tot = 0.5 ± 0.7 mg/L). Manganese precipitates were found in the RW-E (Mn_tot = 1.7 ± 0.8 mg/L), whereas no precipitates were found in the RW-W. This could have been due to the high suspended sediment concentration of up to 30 kg/m³ [34] in the Lower Ems near Rhede (Ems). Precipitated oxides and/or hydroxides can adhere to suspended sediments. At the same time, suspended sediments can also accumulate in the geotextile filters and increase the reduction in the permeability [35]. In summary, it can be concluded that sufficient oxygen can be provided in the surface water for oxidation. However, the reaction is much faster under aerobic conditions than under submerged conditions [36]. Other quality parameters are not relevant to the ochre reaction and can be neglected when assessing the tendency for ochre formation. It is also important to monitor the water levels and tidal range to determine differences in the water levels (see below).

4.3. Groundwater

The groundwater investigations showed that reducing groundwater was found in the areas where damage had occurred as a result of ochre-clogged geotextiles. For example, weakly reducing groundwater (as defined by Hoelting and Coldewey [37]) was found at measurement station GW2-E and at all measurement stations along the Weser estuary. The high oxygen concentrations of DO = 0.9–5.2 mg/L are remarkable. This can be explained by the fact that the groundwater and surface water in the revetment are in constant exchange due to the tides. The fact that there is still sufficient dissolved iron and manganese available for the ochre formation reaction indicates very high initial concentrations of Fe²⁺ and Mn²⁺ ions (Table 5). The half-life times (see [3]) for chemical iron oxidation over the arithmetic mean of all sampling points are approximately 14 min (min: 0.5 min; max: 180 min). These times are sufficient for chemical oxidation during the low water phase. Considering the formation characteristics of iron and manganese oxides and hydroxides (iron: E_H value of approximately 0 to 500 mV, manganese: E_H value of approximately 600 to 1200 mV [10]), iron oxidation can be chemically induced, as the “real” redox potential in the filter will be between that of the river water (E_H ≈ 220 mV) and groundwater (E_H ≈ −10 to 170 mV). Chemical manganese oxidation is not likely, as oxidation generally requires redox potential E_H ≥ 600 mV. However, as the manganese levels were measured in the textile, these may have been precipitated by microbial processes (see below). Measurement station GW1-E showed indifferent conditions; no dissolved iron and no dissolved manganese could be found, so it can be assumed that oxidising conditions prevailed.

The groundwater microbiome revealed a certain abundance of the genera Gallionella and Curvibacter (see Figure 4). The bacteria of the genus Gallionella are well-known iron oxidisers [38]. The members of the genus Curvibacter are also iron-oxidising bacteria and typically occur where iron-rich, oxygen-poor groundwater enters an oxygen-rich environment [39]. As such conditions were found in the vicinity of the Weser estuary, it is feasible that the ochre formation was favoured by microbial influences.

The monitoring of the water levels and tidal range is also relevant for the determination of water level differences (see below).

4.4. Aquifer Material

Veldhuijzen Van Zanten and Thabet [7], in their investigations of ochre-clogged revetment samples, found that changes in the permeability of the aquifer material were largely limited to the upper 5 mm. This observation was also made by Tophoff et al. [3] in an experimental study. In the Ems aquifer material (E1, see Figure 5a), it was also found that only the contact zone with the geotextile had high iron and manganese content (Table 7). Based on the original mass, these values were approximately 0.2–3% for iron and approximately 0.02–0.1% for manganese, which appears to be comparable with the natural content. A direct comparison with natural sand could not be made. The hydraulic conductivity (k_{T,10 °C} value) of the aquifer material beneath the ochre-clogged geotextiles was approximately 10 to 1000 times higher than that of the textiles. In this respect, no relevant changes in the hydraulic performance of the aquifer material were expected due to the high permeability and the observation that the ochre formation occurred exclusively in the contact zone. In further investigations, the hydraulic conductivity of the aquifer material should always be considered, as all damage caused by ochre-clogged geotextiles occurs where sandy soils are found. However, medium- to low-permeability marsh soils are usually found on the riverbanks of the German North Sea estuaries [17]. The preliminary tests for the experimental studies by Tophoff et al. [3] also showed that ochre formation is faster when the subsoil material is more permeable. This is probably due to the greater availability of oxygen and the higher loads of Fe²⁺ or Mn²⁺ ions and biomass.

4.5. Armour Layer

In order to determine the armour layer weight required to prevent the ochre-clogged or impermeable geotextile filter from lifting the armour layer, the buoyancy design according to MAR (2008) [1] is useful. As the maximum difference in the water level can only occur at low tide and under atmospheric conditions, no buoyancy for the armour layer itself needs to be considered.

$$\mathrm{Weight} \mathrm{of} \mathrm{armour} \mathrm{layer}: {\mathrm{G}}_{\mathrm{R}}=\frac{{\mathsf{\gamma }}_{\mathrm{A}}\times {\mathsf{\Delta }\mathrm{h}}_{\mathrm{w}}\times {\mathsf{\gamma }}_{\mathrm{w}}}{\cos \mathsf{\beta }}$$

(2)

• γ_A: safety factor for buoyancy;
• Δh_W: difference in water level (Equation (1));
• γ_W: weight density of water;
• β: slope angle.

With γ_A = 1.0, Δh_W(E) = 2.11 m or Δh_W(W) = 2.38 m, and β = 18.4° (inclination ratio 1:3), the required armour layer weight of 21.8 kN/m² (E) or 24.6 kN/m² (W) can be roughly determined. These values exceed the existing armour layer weights (Table 6) by a factor of up to 3.7 so that stability is no longer given, and the revetment can be uplifted as a consequence. According to the latest standards, permeable armour layers (riprap) should weigh at least 900 kg [1]. Another source gives a maximum weight of 907 kg (2000 pounds [40]). However, it would be uneconomical to design the armour layer based on the weight as it would need to be significantly higher. It should, nevertheless, be high enough to ensure that the geotextile filter rests firmly on the subsoil throughout its service life [41] so that the naturally formed filter cake on the underside of the geotextile is not destroyed. This can also form under a long-term cyclic flow [42].

4.6. Ochre Clogging in Granular Filters

The studies by Tophoff et al. [3] showed that no relevant permanent change in the permeability of granular filters due to iron ochre clogging under experimental conditions was measured.

The in situ samples in W1 revealed that the granular filter was highly permeable (k_{T,10 °C} value) and that the total amount of organic substances (as determined by LOIs) was low. The hydraulic filter stability against the soil is ensured when k_filter > 25 × k_soil [33]. This was the case here (k_filter = 7.0 × 10⁻² m/s; 25 × k_soil = 1.1 × 10⁻³ m/s), although many ochre-clogging products were visible (see Figure 5b). The service life of the filter and the iron and manganese content were unknown.

It is probable that a granular filter located in tidally influenced revetments generally does not experience a significant permanent reduction in permeability due to ochre clogging. A temporary reduction in permeability may be followed by local rearrangement, which could lead to an increase in the permeability to the original level (see also Weidner [5] and Tophoff et al. [3]).

4.7. Formation and Characteristics of Ochre-Clogging Products in Geotextile Filters

In their studies, Miszkowska et al. [43] showed that nonwoven geotextiles with an alternating flow are more likely to cause clogging than geotextiles with a unidirectional flow. The fines, in particular, play an important role [42]. The greater the total amount of fines in the filter environment, the faster the reduction in permeability due to physical clogging.

Palmeira et al. [44] observed that the ochre clogging was concentrated in the first few layers of the filaments. Abromeit [6] showed that the ochre-clogging products were attached to the underside of the geotextile. Microscopic examinations of the cross-sections of the undisturbed samples were used to determine where and how the ochre-clogging products were able to accumulate in the filter structure. The majority of the clogging products were concentrated in the first few layers of the geotextile filaments. In geotextile W2, most of the ochre-clogging products were attached to the underside of the filter layer (see Figure 5c). Only a few ochre-clogging products were attached to the filaments of the additional layer. The opposite was observed for geotextile E1. In these samples, most of the ochre-clogging products were found on the top side of the geotextile in the filter layer. Furthermore, the filaments of the additional layer showed more adhesions in E than in W. An explanation for this can be found by considering the mass per unit area and the resulting permeability. Due to the almost doubled amounts of accumulation and adhesion in the Ems textile compared to the Weser textile (and the very low permeability), it can be assumed that the underside of the geotextile was not exposed to oxygen for a particularly long time and showed completely anaerobic conditions. From these observations, it seems likely that the ochre-clogging products do not generally accumulate on the underside but in the first few filament layers, where sufficient oxygen is available. The enveloping ochre-clogging products (see the disturbed specimens in Figure 6a,d) were largely immobile because they could not be completely removed during the preparation process.

The extensive physicochemical investigations allowed a satisfactory analysis of the nature of the ochre-clogging products. As in the experimental investigations by Tophoff et al. [3], the ochre-clogging products consisted of oxides and/or hydroxides (in this case, iron and manganese oxides and/or hydroxides) and particles from the filter environment. In contrast to the experimental studies, the Ems and Weser samples showed, as expected, a significantly wider range of different elements (see Figure 7 for E and W; see Figure 8 for E). These included silicon, calcium, and aluminium, which occur naturally in phyllosilicates. The silt in the brackish water of the estuaries contains these phyllosilicates, which can then accumulate in the filter. Potassium also occurs naturally in the silt or in suspended matter in brackish water. Traces of phosphorus can originate from other marine silicate minerals, such as apatite.

The amplicon sequencing results showed that the microbial communities of the geotextile samples were different from those of the groundwater sample. The presence of certain types of nitrifying prokaryotes, such as Nitrospira and MDN1 of the Nitrosomonadaceae family, in the geotextile samples from the Weser and the Ems, respectively, suggests that ammonium is being oxidised to nitrate in these locations (see Figure 4). It is worth noting that there was no clear evidence of sulphate-reducing or sulphur-oxidising prokaryotes in either the Weser or Ems samples, despite the presence of sulphate due to marine influences. The comparison of the Weser samples (groundwater and geotextile) in Figure 4 shows no significant similarities. This is remarkable considering that the groundwater and the textile come into contact with each other during each tidal cycle. The filter seems to represent its own environment. The comparison of the microbiomes of the Ems and Weser geotextile samples reveals only the following four common genera (see Figure 4b): Nitrosarchaeum, UTCFX1 of the Anaerolinaceae family, Hyphomicrobium (<1%, therefore not shown in Figure 4b; see European Nucleotide Archive (ENA), project ID PRJEB76146), and Curvibacter (<1%). Of these, Hyphomicrobium is capable of oxidising manganese [45], and Curvibacter is able to oxidise iron [46]. Thus, it seems that macroscopically very similar phenomena (the ochre clogging of geotextiles) can be caused by different microbiomes, showing only minor overlaps. It can be assumed that the salt content (Cl and Na), which differs by a factor of 2.3 (RW-W and RW-E), is mainly responsible for the composition of the different microbiomes. However, even if the microbiomes are very different, they can be physiologically equivalent and may favour similar biogeochemical processes.

The chemical analysis of the textiles revealed that the textile samples from the Ems contained large quantities of manganese and iron oxides and/or hydroxides (Table 8). Parts of it may have been oxidised as a result of biological processes. This finding indicates that the revetment damages observed can be increased by the activity of different microorganisms. It was not possible to quantify the microbiological contribution to ochre formation in this study.

Thus, the process of ochre clogging in the Ems and Weser estuaries seems to be a biogeochemical clogging process. This refers to the combined action of the chemical precipitation of iron and manganese and precipitation by microorganisms. Physical clogging due to the accumulation of particles from the filter environment in the geotextile filter seems to be favoured by ochre formation, as the pore volume is reduced by adhesions. According to the authors, the reaction rates for chemical clogging (see [3]) and other models describing the permeability of geotextiles (e.g., [15,47]) cannot be used in isolation because they do not consider the combined biogeochemical and physical clogging process.

4.8. Application of Geotextile Filters in the Case of Ochre Clogging Tendency

A model to estimate the accumulation of ochre products in a geotextile filter over time is not possible with the available results. However, the effect of ochre clogging on the service lives of geotextile filters can be statistically estimated. Figure 9 summarises all the available data (logarithmic representation) on the reduction in the permeability of ochre-clogged filters. In order to harmonise the values, the residual permeability of each geotextile was calculated. The experimental data describe the chemical iron precipitation and show that the residual permeability is still 59% after approximately 0.5 years. Investigations by the BAW in 1993 [35] and 1998 [48] showed residual permeability of 0.31% after 5 years, 0.54% after 9 years, and 0.8% after 11 years at different sampling locations. The current sampling campaign showed residual permeability of 0.03% to 0.05% after 35 to 39 years at different sampling points. A “mean” was formed using a non-linear regression analysis to describe the reduction in filter permeability over time. The assumption that the residual permeability is non-linear is described in detail in Tophoff et al. [3]. This is supported by the hypothesis that the iron and manganese loads, as well as the biomass, are significantly involved in the ochre formation (see above). As the permeability of the textile decreases and the influent loads are reduced, the rate of permeability reduction is also reduced. To determine the upper and lower bounds, the chemical reaction rate was factorised using the extreme values of the groundwater quality and oxygen availability.

Further research is required to provide a quantifiable estimation of the combined biogeochemical and physical clogging process, but at present, crucial indications for the optimum design of the geotextile filter can be described. Limit values for the susceptibility to ochre clogging of geosynthetic filters in tidally influenced revetments have so far been established by Tophoff et al. [9]. Taking the in situ investigations into account, the limit values for the groundwater quality need to be updated. The redox potential (E_H) and dissolved oxygen (DO) are no longer considered in isolation, as these limits were clearly exceeded at the damaged sites. Both parameters influence the rH value, whose limit value (rH > 17) was confirmed by the sampling. The limit value for the concentration of iron (Fe²⁺ ions < 0.2 mg/L) was also confirmed. Additionally, the manganese content (Mn²⁺ ions) has to be considered with the same limit value of <0.2 mg/L. If these limits are complied with, a state-of-the-art filter design can be obtained. If these limits are not complied with, an adapted state-of-the-art filter design that can reduce the damage caused by ochre clogging can be proposed, as below.

Palmeira [49] summarises various studies on the behaviour of geotextile filters under severe and critical conditions and measures to minimise failures. He describes that nonwoven needle-punched geotextiles appear to be less susceptible to clogging than heat-bonded or woven geotextiles. This means that nonwoven, non-heat-bonded geotextiles should be used where there is a risk of clogging. Tophoff et al. [3] recommend selecting the maximum allowable characteristic opening size when designing filters with a risk of iron precipitation. Other studies support this recommendation, although the opening size is not directly related to the clogging potential ([47,50] according to [51]). Heibaum [51] recommends the use of the permeability criterion k_geotextile ≥ 100 × k_soil according to Rüegger and Hufenus [52] when intense clogging is expected. This recommendation is supported by Tophoff et al. [3] and Palmeira [49]. As ochre clogging can favour physical clogging due to the associated reduction in pore space, other non-clogging criteria should also be considered when designing filters. Giroud [53] and the USBR [41] suggest that the number of constrictions should be between 25 and 40. As Tophoff et al. [3] recommend a smaller inner surface area in the case of possible iron precipitation, the number of constrictions should be as low as possible within the scope of the state-of-the-art filter design, as this is related to the mass per unit area and the layer thickness of the geotextile and thus the inner surface area [54]. Tophoff et al. [3] also recommend selecting a geotextile constructed from a mixture of materials with hydrophilic behaviour. Thus, an additional layer with a large inner surface area on the top side of the geotextile could slow down the negative effects of ochre clogging due to the decreased oxygen availability in the filter layer [14]. The weight of the revetment should be as high as possible within an economical framework. Figure 10 summarises the adapted filter design by proposing a procedure for filter design if ochre clogging is likely. It should be emphasised that these suggestions only apply to geotextiles that meet the other requirements for hydraulic filter materials. According to Heibaum [51], criteria relating to geometric and hydraulic properties, resistance to clogging, and survivability must be fulfilled. This applies to selected nonwovens with a mass per unit area of 500 g/m² or more. National regulations and guidelines are to be considered.

5. Conclusions and Prospects

Extensive theoretical and in situ investigations were carried out to better understand the environmental conditions surrounding granular and geotextile filters in estuaries in the case of ochre clogging. Physical, chemical, and biological tests were therefore carried out to analyse the filter and the surrounding environmental conditions in detail. The main aim was to improve the knowledge of the ochre clogging process under in situ conditions in geotextile filters.

Our research shows that ochre formation in the Ems and Weser estuaries is favoured by microbiological influences. The geotextiles receive organic material through the groundwater in limited quantities but at high concentrations. This can lead to the growth of a biofilm in the geotextile, which further compounds the reduction in permeability caused by chemical and mechanical processes.. Therefore, the process appears to be a biogeochemical clogging process, i.e., the combined action of the chemical precipitation of iron and manganese and precipitation by microorganisms. The filter itself seems to be a self-contained microbiological environment. The precipitates appear to be largely immobile and accumulate in the first few filament layers, where sufficient oxygen is available, and reduce the pore volume of the filter. These ochre-clogging products envelop the filaments until the permeability of the geotextile filter is very strongly reduced. In the Ems estuary, for example, the residual permeability of the geotextile filter after a service life of 35 to 39 years is between 0.05 and 0.001%, and it is therefore almost impermeable.

According to the current state of research, physical clogging in the geotextile filter is favoured by ochre formation. Fine particles, which would be passable in their original state, can thus clog the filter structure as deposits in the ochre-clogging products.

Based on the preliminary studies and the current study, measures were developed to minimise the problems of reduced filter performance due to ochre clogging. An assessment of the tendency for ochre formation is essential for this. For the first time, limit values for the groundwater quality were established to quantify the ochre clogging tendency. In order to maximise the service life of the geotextile filter in the event of ochre clogging, recommendations for an adapted state-of-the-art filter design have been developed as a guide for planners.

A quantifiable and generally valid assessment of the reduction in filter permeability as a result of the combined biochemical clogging process could not yet be derived from the research results obtained. Nevertheless, we have been successful in describing the bio-geochemical conditions of the filter environment almost completely, although the results are subject to statistical uncertainty because only a limited number of tests were possible. The authors see a need for further research in this area. Adapted geotextiles and filter constructions should be investigated under experimental and in situ conditions. Therefore, further accelerated experimental studies are recommended. The in situ investigations should cover periods of 1, 5, and 10 years. The aim of both investigations is to gain further knowledge and to validate or recalibrate the described model for estimating the service life of geotextile filters in case of ochre clogging tendency.