Photosynthetic oxygenation for urine nitrification

In the context of sustainable resource management, the transition towards energy-efficient and resource-recovery focused wastewater treatment is pivotal. For nutrients such as nitrogen (N) and phosphorus (P), the diluted characteristics of domestic wastewater make direct nutrient recovery technologically challenging and resource intensive. Nutrient recovery is therefore recommended for more concentrated waste streams, such as manure, digestate or urine (Verstraete et al. 2016). Urine accounts for only 1% of the total domestic wastewater volume, while it contains approximately 40% of the phosphorus load, 69% of the nitrogen load, and 60% of the potassium load arriving at municipal wastewater treatment plants (Kujawa-Roeleveld & Zeeman 2006). This makes it a valuable target stream for nutrient recovery.

While direct application of urine as fertilizer has been common practice in many rural societies around the world, the high water content makes transportation and application costly, while the high urine salinity causes soil salinization (Basakcilardan-Kabakci et al. 2007). In addition, potential presence of pathogens and pharmaceuticals further increases health concerns towards farmers and consumers (Udert et al. 2006). More advanced chemical and biological nutrient recovery strategies have therefore been demonstrated, such as ammonia stripping, struvite precipitation and microalgae cultivation (Maurer et al. 2006; Tuantet et al. 2013). Urine is, however, highly unstable, as urea hydrolysis through bacterial urease results in the production of ammonia and bicarbonate. The subsequent pH rise induces nutrient losses through unwanted phosphate precipitation and ammonia volatilization and concomitant odour and toxicity issues (Udert et al. 2003).

Chemical stabilization, for example through acid dosage, or biological stabilization through nitrification, has therefore been suggested as a pre-treatment step prior to nutrient recovery (Maurer et al. 2006; Feng et al. 2008). Biological urine stabilization through nitrification converts volatile ammonia to nitrate, thereby allowing for long-term storage and further use as agricultural fertilizer or for microalgae cultivation. However, oxygen requirements associated with the nitrification process result in large energy demands due to energy-intensive conventional aeration. Based on an oxygenation efficiency of 1 kg O₂ kWh^–1 (Metcalf & Eddy 2002) and a urine chemical oxygen demand (COD)/N ratio of 0.8, the electricity need for nitrification and COD oxidation with conventional aeration is 31 kWh m⁻³ urine (solids retention time (SRT): 6.66 days; 25 °C).

An alternative for conventional aeration is photosynthetic oxygenation, or in situ oxygen production by photosynthetic organisms such as microalgae. By providing the oxygen demand for autotrophic nitrification and heterotrophic carbon oxidation through in situ photosynthesis while consuming heterotrophically produced carbon dioxide, electricity costs and greenhouse gas emissions affiliated to conventional aeration and wastewater treatment can be reduced (Praveen & Loh 2015). In addition, the produced microalgal-bacterial biomass can be used for energy production (anaerobic digestion), as a resource for the production of high-value biochemical and biofuels or it can be applied as slow-release fertilizer or microbial protein (Tuantet et al. 2013; Coppens et al. 2016).

The combination of photosynthetic oxygenation with nitrification or nitrification–denitrification has been described in both high-rate algal ponds (HRAPs) and in photo-bioreactors (PBRs), for various waste streams (Table 1). Karya et al. (2013) achieved a volumetric nitrification rate of 43 mg N L⁻¹ d⁻¹ and photo-oxygenation rate of 140 mg O₂ gVSS⁻¹ d⁻¹ (VSS: volatile suspended solids), while van der Steen et al. (2015) reported a volumetric nitrification rate of 46 mg N L⁻¹ d⁻¹ and photo-oxygenation rate of 234 mg O₂ gVSS⁻¹ d⁻¹. Furthermore, direct microalgae cultivation on urine has been established (Adamsson 2000; Yang et al. 2008; Tuantet et al. 2013) and superior microalgal growth on nitrified urine compared to untreated urine has been demonstrated (Coppens et al. 2016).

Photosynthetic oxygenation for urine nitrification has so far not been documented. The goal of this study was to develop a microalgal-bacterial consortium for urine nitrification through photosynthetic oxygenation. The influence of salinity and ammonium concentration on microalgal growth was determined and subsequently, a PBR was operated for urine nitrification in the absence of external aeration.

The inoculum comprised commercially available nitrifying activated sludge (Avecom, Belgium), the microalgal species Chlorella sp., Haematococcus sp., Desmodesmus sp., Ankistrodesmus sp., Pediastrum duplex, Chlorella vulgaris and Nannochloropsis sp. (Marine Biology research group, Ghent University) and microalgae acquired from a grassland pond (Ghent, Belgium). The initial biomass concentration was set at 0.7 g VSS L⁻¹ of which half was microalgal biomass, the other half activated sludge. Prior to inoculation, all microalgae were cultivated at 25 ± 1 °C in enriched seawater artificial water (ESAW) medium made with artificial seawater (Harrison et al. 1980), to provide a salinity level in the range of that of urine. The cultures were aerated with sterile air and continuously illuminated (Philips TL-D 90 De Lux 36 W).

A mixture of all microalgae was exposed to ESAW medium containing 1, 2, 3.5 and 5 g L⁻¹ NaCl and 50, 100, 200 and 1,000 mg NH₄⁺-N L⁻¹, brought to pH 6 with a phosphate buffer. The experiment was realized in quadruplicate in 96 well microtiter plates (300 μL per well). The optical density (OD) was measured daily at 620 nm with a plate reader (Tecan Infinite), after shaking. Each well was inoculated between 0.10 and 0.17 (OD620). The plates were incubated in an orbital shaking incubator at 25 °C for 160 hours.

A Plexiglas, gas-tight, bubble column PBR with a working volume of 4 L (ø 12 cm; height 50 cm), was operated in semi-continuous mode (Figure 1). The illuminated surface area was 0.16 m² resulting in a volume to surface ratio of 0.025 m³ m⁻². Three fluorescent growth lamps (Grolux T5, 24 W, Sylvania) provided a photon flux density of 300 μmol PAR m⁻² s⁻¹ (PAR: photosynthetically active radiation), measured in a straight angle from the lamp at the outer reactor wall (Fieldscout quantum light meter, USA). Liquid recirculation at 0.5 L min⁻¹, to facilitate pH and dissolved oxygen (DO) measurements and mixed liquor sampling, and headspace gas recirculation at 6 L min⁻¹ took place. CO₂ and N₂ gas were dosed in a 10/90 (v/v) ratio (Bronkhorst high-tech EL-FLOW select mass flow meter/controller) as 1% of the gas recirculation flow. Where mentioned, co-aeration with compressed air was applied at 5 L h⁻¹. The pH was controlled between 6.5 and 7 (Prominent dulcometer; pH electrode dulcotest PHEP-112). The temperature was kept at 25 ± 1 °C by operating in a temperature controlled room.

Initially, 10% diluted synthetic urine medium (Brooks & Keevil 1997) with additional trace element solutions (Kuai & Verstraete 1998) was fed, implying a nitrogen loading rate of 50 mg N L⁻¹ d⁻¹. The influent electrical conductivity was altered to that of a 33% dilution of urine (6.66 mS cm⁻¹) using NaCl. The nitrogen loading was distributed over time using three cycles of 8 hours per day. Influent was dosed in the first 8 minutes of each cycle during gas recirculation, while effluent was extracted in the last 8 minutes of each cycle, at the end of the settling phase.

Every 2 days, samples of influent and bulk reactor solution at the end of a cycle were filtered (0.45 μm Chromafil Xtra, Machery-Nagel, PA, USA) and stored at 4 °C. The concentration of ammonium was determined following the standard method of Nessler (APHA 1992) or according to the standard methods (APHA 1992). Kjeldahl nitrogen was also analysed according to the standard methods (APHA 1992). Organic nitrogen was determined as the difference between Kjeldahl nitrogen and ammonium nitrogen. Nitrite and nitrate were analysed with ion chromatography (IC 761, Compact, Methrom AG, Switzerland). COD was determined photometrically (Nanocolor COD 160; Machery-Nagel, PA, USA). Total suspended solids (TSS) and VSS were measured according the standard methods (APHA 1992). The dissolved oxygen (DO) concentration was measured daily (Hach HQ40d) and electrical conductivity weekly (Consort C833). Reactor biomass was analysed with light microscopy (Zeiss Axioskop 2 plus). During PBR operation, ammonia-oxidizing bacteria (AOB) and nitrite-oxidizing bacteria (NOB) nitrification activities were determined regularly following a standardized protocol (Coppens et al. 2016).

Since fresh unhydrolysed urine contains a high salinity (±20 mS cm⁻¹) and urea nitrogen concentration (5 g N L⁻¹ for synthetic urine), the influence of different salt and ammonium concentrations was investigated on the growth of a mixture of all microalgal species. Four different salinities (1, 2, 3.5 and 5 g L⁻¹ NaCl, corresponding to 4.54, 6.55, 9.95 and 13.60 mS cm⁻¹, respectively) and ammonium concentrations (50, 100, 250 and 1,000 mg NH₄⁺-N L⁻¹), were applied (Figure 2). A longer lag phase was observed at salt concentrations equal to or higher than 3.5 g NaCl L⁻¹. This might be due to the predominance of fresh water microalgal species. Results indicate that a concentration of 50 mg NH₄⁺-N L⁻¹ is preferred. Since no significant change in growth was observed for the mixed culture up to 2 g L⁻¹ NaCl, corresponding to a salinity of 6.55 mS cm⁻¹, these tests indicate that the salinity of roughly 2 mS cm⁻¹, for a 10% dilution of synthetic urine, will not inhibit microalgal growth. In contrast, ammonium concentrations (500 mg NH₄⁺-N L⁻¹ for a 10% diluted urine solution) could result in slower microalgal growth if ammonium accumulates in the PBR, depending on the nitrogen loading rate and nitrification activity.

During the first 15 days of reactor operation, nitrogen removal efficiencies between 60% and 95% (Figure 3(a)) were achieved, but no nitrate accumulation was observed. COD removal efficiencies fluctuated between 61% and 93% (Figure 3(d)) with a maximum COD removal rate of 45 mg COD g VSS⁻¹ d⁻¹ at day 15. The DO concentration fluctuated around 3.6 ± 2.6 mg L⁻¹, indicating sufficient photo-oxygenation. An estimated 22% of the incoming nitrogen was assimilated, 51% accumulated in the reactor or was unconverted and removed with the effluent (Figure 4). The remaining 27% of incoming nitrogen was assumed to be lost mainly due to ammonia stripping. Indeed, free ammonia concentrations were the highest (around 0.6 mg N L⁻¹ as calculated from 100 mg NH₄⁺-N L⁻¹ at pH 7 and 25 °C). It is unlikely that denitrification contributed to the nitrogen loss, as no consistent nitrification was observed, and as average DO concentrations were rather high (around 2 mg O₂ L⁻¹), limiting anoxic zones.

To stimulate nitrification, at day 15, the reactor was re-inoculated with 1 g VSS L⁻¹ of commercially available nitrifying inoculum (Avecom, Belgium). The high calcium carbonate content of the inoculum resulted in a decrease in VSS/TSS ratio from 0.83 to 0.42. Although DO concentrations between 0.6 and 2 mg L⁻¹ indicated successful photo-oxygenation, re-inoculation did not result in nitrification. However, activity tests demonstrated the low nitrification potential of the PBR biomass in both fresh medium and reactor supernatant. Additionally, no inhibitory effect was observed of reactor supernatant on new nitrifying inoculum (Figure 5(a); Figure S1, Supplementary material, available with the online version of this paper).

To increase oxygen concentrations and stimulate nitrification, co-aeration with compressed air was started at day 25 at a flow rate of 0.25 L O₂ L_reactor⁻¹ h⁻¹. Complete nitrogen removal was obtained (Figure 3(a)) and nitrogen assimilation (77%) in microbial biomass remained the dominant nitrogen removal pathway, with 76% assimilated as microalgal and 1% as chemoheterotrophic biomass (Figure 4).

From day 55 onwards, accumulation of nitrate indicated successful nitrification, while the COD removal efficiency was situated between 44% and 83%. Starting from day 110, nitrification efficiencies higher than 50% were obtained. Therefore, at day 118, external aeration was stopped and since oxygen levels remained above 6 mg O₂ L⁻¹, photosynthetic oxygenation was sufficient to sustain nitrification and organic carbon oxidation. At day 134, nitrification efficiency reached 64% at a nitrogen loading rate of 97 mg N L⁻¹ d⁻¹ (12% synthetic urine) (Figure 3(b)). Furthermore, well-settling microbial biomass was obtained, as indicated by the low sludge volume index (SVI₃₀) of 58.06 mL g⁻¹.

The increase of nitrogen loading rate to 100 mg N L⁻¹ d⁻¹ (15% dilution of synthetic urine) at day 134 resulted in a nitrification efficiency decline to values lower than 30% and reduction in biomass concentration to 2.14 gVSS L⁻¹ at day 145. Nitrification batch activity tests demonstrated maximum biomass specific ammonium and nitrite oxidation rates of 67 and 48 mg N gVSS⁻¹ d⁻¹ (Figure 5(b); Figure S2, Supplementary material, available online) (or volumetric oxidation rates of 122 and 87 mg N L⁻¹ d⁻¹, respectively), which are higher than the nitrogen loading rate. A DO concentration below 1 mg L⁻¹ indicated a drop in photo-aeration rate.

Between day 146 and 167, nitrification efficiency fluctuated between 0 and 30% and, at day 168, the nitrogen loading rate was decreased to 35 mg N L⁻¹ d⁻¹. From day 168 until 180, nitrification efficiencies remained between 52% and 53%. At day 180, a last activity test was performed (Figure 5(b); Figure S2, Supplementary material), demonstrating that both AOB and NOB activity decreased to 12 mg NH₄⁺-N gVSS⁻¹ d⁻¹ and 11 mg NH₄⁺-N gVSS⁻¹ d⁻¹, respectively.

The maximum achieved volumetric nitrification rate was 67 mg N L⁻¹ d⁻¹ at a loading rate of 97 mg N L⁻¹ d⁻¹, which is higher than in the outdoor HRAP described by de Godos et al. (2009), where a volumetric nitrification rate of 9 mg N L⁻¹ d⁻¹ was achieved at a loading rate of 30 mg N L⁻¹ d⁻¹. Karya et al. (2013) nitrified artificial wastewater supported by photo-oxygenation in an algal-bacterial consortium at a nitrogen loading rate of 50 mg N L⁻¹ d⁻¹ and maximum volumetric nitrification efficiency of 43 mg N L⁻¹ d⁻¹. This artificial wastewater did, however, not contain COD and all nitrogen was present as ammonium. Van der Steen et al. (2015) reached a comparable volumetric nitrification rate of modified BG-11 medium of 46 mg N L⁻¹ d⁻¹ (Table 1).

The observed sudden nitrification activity in this study could be the consequence of adaptation of the nitrifying community to light, since it is known that AOB and NOB are inhibited by light at an intensity of 500 and 1,250 μmol m⁻² s⁻¹ (Vergara et al. 2016). Guerrero & Jones (1996) studied light inhibition on marine AOB and NOB and concluded that photo-inhibition is species-specific and dependent on light intensity, lighting period and wavelength. AOB were found to be more sensitive to blue light than NOB, while cool-white fluorescent light inhibited AOB activity but did not influence NOB. Abeliovich & Vonshak (1993) observed complete nitrification inhibition during 4 days of exponentially growing Nitrosomonas europeana after 1 hour light exposure, while ammonia presence provided some protection. Alleman et al. (1987) observed that light with a wavelength in the range of 410–415 nm is responsible for Nitrosomonas inhibition, during a period without respiration and nutrient absence. The basis for light sensitivity is assumed to be damage to the many cytochromes of AOB and NOB, involved in the nitrification energy transduction pathways (Ward 2011). No records were found on a longer period of light irradiation and potential adaptation to light-inhibiting conditions. Since in this study, no light–dark cycle was used to enable recovery from light inhibition and, additionally, the selected light source to stimulate microalgal growth was rich in the inhibitory wavelengths, conditions were unfavourable for nitrifiers.

Theoretically, nitrifying bacteria require 4.57 g oxygen to convert 1 g NH₄⁺-N to 0.95 g NO₃⁻-N and 0.05 g biomass-N, while heterotrophic bacteria consume 0.69 g oxygen to oxidize 1 g of organic carbon to CO₂ (SRT: 6.67 days; 25 °C). At day 134, the moment of optimal nitrification, the oxygen consumption and thus production rate was 160 mg O₂ gVSS⁻¹ d⁻¹, calculated based on the stoichiometry of nitrogen and organic carbon oxidation. This value is comparable to the photo-oxygenation rates calculated based on the results in other studies (Table 1). Additionally, in this study, the light supply rate was 1.04 mol photons L⁻¹ d⁻¹. Together with the volumetric oxygen production rate of 429 mg O₂ L⁻¹ d⁻¹ (recalculated from 160 mg O₂ gVSS⁻¹ d⁻¹), this results in an oxygen quantum yield of 0.013 mol O₂ mol photons⁻¹. This is lower than the experimental maximal quantum yield of 0.1 mol O₂ mol photons⁻¹ under low light (Zijffers et al. 2010), indicating room for improvement in photosynthetic efficiency. However, the real oxygen quantum yield could be slightly higher due to a lower photon flux density than the measured 300 μmol photons m⁻² s⁻¹, since perfect homogenous light distribution was lacking.

To check if photosynthetic oxygenation was sufficient, DO was monitored during one cycle at day 134 (Figure S3, Supplementary material, available online). To assess the oxygen consumption rate without algal oxygen production, the lights were turned off during the first 2 hours. When the DO reached values lower than 3 mg O₂ L⁻¹, lights were turned on. Initially the DO remained constant while nitrification was still ongoing, indicating a sufficient photo-oxygenation rate due to the simultaneous oxygen production and consumption. When ammonium and nitrite were fully oxidized (measured with ammonium and nitrite test strips), DO increased at a rate equal to the oxygen production rate minus the heterotrophic oxidation rate. The measured oxygen production rate was 4.24 mg O₂ L⁻¹ h⁻¹ taking into account the ongoing organic carbon oxidation. This oxygen production rate is within the range of other microalgal-bacterial systems (Karya et al. 2013). However, 95 mg O₂ L⁻¹ was consumed for nitrification and 23 mg O₂ L⁻¹ was consumed for COD oxidation leading to a total theoretical oxygen consumption rate of 15 mg O₂ L⁻¹h⁻¹. The difference between this rate and the oxygen consumption rate based on the DO profile might be due to the direct consumption of oxygen within the floc entity.

Since each involved type of micro-organism has its own function, nutritional requirement and metabolic rate, the potential nitrification rate can be calculated. Taking into account the stoichiometry of photosynthesis, nitrification and heterotrophic COD oxidation, ideally without nitrogen losses, 23% of incoming nitrogen should be assimilated by the microalgae, resulting in a sufficient amount of oxygen to oxidize the leftover 77% of the ammonium nitrogen and present COD (Figure 6). Considering algal oxygen production rates in an optimal designed PBR between 128 and 192 mg O₂ L⁻¹h⁻¹ (Javanmardian & Palsson 1992), it is theoretically possible to nitrify urine at a nitrogen influent loading rate of 1.14 g N L⁻¹ d⁻¹ and 0.91 g COD L⁻¹ d⁻¹ (urine COD/N ratio of 0.8). In cheaper HRAPs, however, optimal oxygen production rates up to 9.55 mg O₂ L⁻¹h⁻¹ are reported (Arbib et al. 2017), indicating a maximum influent loading rate of 57 mg N L⁻¹ d⁻¹.

Several operational parameters influence the activity of the consortium organisms. The specific growth rate and activity ratio between microalgae and nitrifying bacteria is influenced by parameters such as available nutrients and light and is related to the substrate affinity constants (K_s) for ammonium nitrogen and inorganic carbon. Due to the higher nitrifier K_s values for inorganic carbon (21.4 mg C L⁻¹; Guisasola et al. (2007)) in comparison to microalgae (0.58 mg C L⁻¹; Shelp & Canvin (1980)), higher CO₂ concentrations favour nitrifiers. The same can be assumed in terms of ammonium concentration due to the lower K_s value for microalgae (0.01 mg N L⁻¹; Hein et al. (1995)) compared to AOB (0.42 to 4.47 mg N L⁻¹; Kayee et al. (2016)).

Next to high substrate concentrations, an SRT increase favours slow growing nitrifiers and increases nitrification activity. To obtain better settling characteristics, dense floc formation could be stimulated by a low hydraulic retention time or high volumetric exchange ratio (VER), causing poor-settling biomass to wash out. The microalgae could form the outer layer of the granule while the nitrifying bacteria are protected from light inhibition in the inner layers. Nevertheless, granule formation is a complex process which depends on many parameters. Additionally, the anoxic core could facilitate denitrification, resulting in an unwanted nitrogen loss. With an increase in VER from 15% to 50%, by reducing the number of cycles from three to two and increasing the influent flow rate to 2 L d⁻¹ of 10% dilution of urine, a nitrogen loading rate of 250 mg N L⁻¹ d⁻¹ could be reached. Attention has to be paid to potential free ammonia formation when the same nitrogen loading rate is applied with fewer cycles.

Since parameters such as pH, nutrient quantity and quality, stirring and temperature were optimal for most microalgae, higher oxygen production rates may be achieved by increasing the illuminated volume. Light attenuates exponentially as it penetrates into the culture medium, estimated by Lambert–Beer's law (Lee 1999). Ogbonna & Tanaka (2000) observed light penetration of only 2 cm in a photo-bioreactor containing a biomass concentration of 1 g L⁻¹ with a light absorption coefficient of 200 m² s⁻¹ and an illumination intensity of 500 μmol PAR m⁻² s⁻¹. In this study the photon flux density was 300 μmol PAR m⁻² s⁻¹ and biomass density fluctuated around 2 g VSS L⁻¹, indicating a light penetration depth less than 1 cm or an illuminated volume of less than 30%.

Microscopic observation after 180 days of reactor operation revealed that microalgal-bacterial flocs were formed with a size of 50–100 μm (Figure 7). Scenedesmus sp. was the dominant green microalga present, but also Chlorella sp. and the cyanobacteria Synechocystis sp. and Leptolyngbya sp. were detected. Between the large microalgal cells, the bacteria were present in low numbers. The electrical conductivity in the PBR was fluctuating around 8.5 mS cm⁻¹ corresponding to 100 mM NaCl. Among the inoculated microalgae, only Nannochloropsis sp. was a distinct salt water species, although all fresh water species were initially cultivated in a salt water medium (ESAW) and, depending on species, microalgae can be halotolerant and show adaptation (Hart et al. 1991). Chlorella sp. presents optimal growth at a salinity level of 100 mM (Abdel-Rahman et al. 2005), indicating feasible conditions in the PBR. In contrast, for Scenedesmus obliquus optimal growth was observed in BBM medium supplemented with 25 mM NaCl (Salama et al. 2013). This shift towards Scenedesmus sp. in a concentrated wastewater solution was observed before by Koreiviene et al. (2014), suggesting the larger surface area to volume ratio enabled quicker nutrient uptake. Karya et al. (2013) inoculated with Scenedesmus quadricauda and selection took place towards cyanobacteria; however, reactor salinity was not mentioned.

Since the energy gain of photosynthetic oxygenation would completely disappear in a PBR with synthetic light supply, HRAPs and PBRs in natural sunlight are more suitable. Although electricity costs and associated carbon footprint are reduced and electricity independence allows application in developing countries, the volumetric nitrification rate of 67 mg N L⁻¹ d⁻¹ achieved in this study is low compared to the urine nitrification rate of 450 mg N L⁻¹ d⁻¹ achieved with conventional aeration in a membrane bioreactor (Coppens et al. 2016). Due to this lower nitrification rate and outdoor day–night light regime, the required surface area and reactor volume are higher compared to systems relying on conventional aeration. Outdoor experiments on a larger scale should be performed to get a better insight on the true energy gain of photosynthetic oxygenation.

It was experimentally demonstrated that biological oxidation of all nitrogen present in urine to nitrate is a promising pre-treatment stabilization step to substitute for expensive conventional aeration, before further nutrient recovery. A consortium of microalgae and nitrifying bacteria was successfully developed in which the photosynthetic oxygenation rate was sufficiently high to nitrify urine at a volumetric nitrification rate of 67 mg N L⁻¹ d⁻¹. Additionally, a maximum biomass specific photo-oxygenation rate of 160 mg O₂ gVSS⁻¹ d⁻¹ was achieved and microscopic observations revealed that Scenedesmus sp. was the dominant microalga after 180 days of reactor operation. Finally, outdoor experiments on a larger scale should be performed to estimate the true energy gain of photosynthetic oxygenation.

The authors would like to thank Dr Claudio Sili from the ISE-CNR, Firenze for determination of the microalgae and Dr Marc Spiller and Dr ir. Erik Van Eynde for the critical discussions. This study was also supported by the European Space Agency (ESA) and the Belgian Science Policy (BELSPO) in the framework of the MELiSSA project.