The SEMIZENTRAL approach is an infrastructure solution for the challenges of high growth dynamics and resource scarcity in fast growing urban areas. The integration of water, wastewater, waste, and energy in one system increases resource efficiency. District-wise realization enables the infrastructure system to grow at the same rate as the city. The concept has been realized for the first time on a scale of 12,000 population equivalent in Qingdao, China. Greywater and blackwater are collected separately; treated greywater is reused for toilet flushing. Reclaimed blackwater is used for irrigation. The analyses of the wastewater composition reveal significant differences in comparison to design values as well as to literature values for greywater and blackwater. Unexpected user behaviour, as well as cross-connections, are likely reasons. The greywater and blackwater treatment processes in the Resource Recovery Center were adapted to the influent's characteristics, so that legal effluent limits are fulfilled, despite changes in influent quality. Small systems often show higher influent variability. Design data for systems with source separation are still lacking. Measurement campaigns in areas similar to the planning area are recommendable, but might not always be possible. In addition, there is a risk of cross-connections between blackwater and greywater, though this can be reduced. For these reasons, there is (possibly high) uncertainty regarding design values for greywater and blackwater. Correspondingly, the treatment processes need to be designed flexibly. For future implementation, technical risks deriving from source separation have to be weighed against the expected higher acceptance of reuse of treated greywater in households. Intra-urban reuse of total wastewater, in combination with extensive public relations programs, might be an alternative.
Challenges regarding water infrastructure in fast-growing urban areas
Urbanization is one of the major trends in the 21st century. The rapidity at which urban growth occurs, usually hand in hand with regional water scarcity, represents a particular challenge for infrastructure planning, especially in terms of the realization of physical (pipe-based) infrastructure. China, for example, with its many fast-growing urban areas, is increasingly affected by these challenges. Rapid economic growth, as well as population growth and urbanization, are leading to regional water scarcity (Yi et al. 2011). Insufficient treatment of wastewater from industry and from the urban population causes decreasing water quality (Hu & Cheng 2013). To cope with the challenges of high dynamics and limited resources in fast-growing urban areas, infrastructure needs to be flexible and adaptable, as well as more resource-efficient. A possible solution is the SEMIZENTRAL approach. The integration of water, wastewater, and waste enables higher resource efficiency: non-potable service water is produced from wastewater to replace tap/drinking water where lower water quality is sufficient. Moreover, the co-digestion of biodegradable waste with the sewage sludge enables an energy self-sufficient operation of the supply and treatment plant, the so-called Resource Recovery Center (RRC). In addition to its resource efficiency, SEMIZENTRAL is characterized by its size, which lies between central (entire city) and decentral (single building) dimensions. If a new district is developed, a new RRC is built for exactly this district. With this district-wise realization, the system reaches full capacity more rapidly. Planning only needs to be undertaken for the currently developed district and not for an entire city, thus reducing uncertainties and subsequent planning and investment risks. In contrast to decentralized systems, a professional operation is economically feasible; hence, high treatment and product safety can be ensured. (Bieker et al. 2010).
Implementation of SEMIZENTRAL in Qingdao, P.R. China
The SEMIZENTRAL approach has been implemented for the first time, worldwide, with the RRC in the Chinese coastal city of Qingdao. Greywater (wastewater from e.g. showers, wash basins, laundry) and blackwater (from toilets and kitchen sinks) from 12,000 population equivalent (design value, based on 100 gCOD/(C·d)) are collected and treated separately. Greywater is processed to provide service water for toilet flushing; treated blackwater is reused for irrigation. Energy is recovered through the co-digestion of food waste and sewage sludge, enabling energy self-sufficient operation (Tolksdorf et al. 2016).
The catchment area consists of two residential areas and ShiYuan village, where office buildings, a canteen, and hotels/guest houses are located. In addition, two more hotels are connected to the RRC. ShiYuan village, as well as one of the hotels, are connected via a pressure pipe to the RRC (because of the topography). The wastewater from the residential areas and the other hotel is discharged into a gravity sewer system.
The greywater treatment process consists of a pre-storage tank, followed by a 1 mm sieve, a membrane bioreactor (MBR) with external filter chambers, and chlorine disinfection. Blackwater passes through a pre-storage tank and is then mechanically pre-treated with a screen, a grit chamber, a pre-clarifier, and a 1 mm sieve. The biological treatment takes place in an MBR. For denitrification, a combination of pre-denitrification and post-denitrification with acetic acid dosage has been chosen. Phosphorous is removed by two-point precipitation (dosage of FeCl3 at the intake of the pre-clarifier and the aeration basin). After biological treatment, blackwater is disinfected by chlorine. Treated greywater and blackwater can be stored in a post-storage tank for equalization between production and usage of service/irrigation water. Further details about the design of the RRC are presented in Tolksdorf et al. (2016).
Measurement campaigns with the aim of characterizing the influent flows to the RRC in Qingdao have been conducted. 24 h-composite samples (sampling every 10 min) are taken from the effluent of the blackwater pre-storage tank; greywater samples are taken at the influent to the aeration basin (after the pre-storage tank and sieve). In addition, samples are taken from the sewer system at various representative locations within the catchment area between 6 am and 10 pm. Every half hour, a grab sample is taken from the pumping station at a hotel and at the influent shaft of the RRC and the two grab samples are mixed to from a composite sample. At the residential area and ShiYuan village, samples are taken every 15 minutes from the sewer and 4 samples are combined to form a composite sample. All samples are analysed for chemical oxygen demand (COD), total nitrogen (TN), and total phosphorous (TP) with cell tests from Merck KGaA (Darmstadt, Germany) within 24 hours; temperature and conductivity are measured directly after sampling.
RESULTS AND DISCUSSION
Wastewater characteristic at the influent to the RRC
The characteristics of the influent grey- and blackwater are highly variable, diurnally as well as from day to day.
Average influent concentration to the RRC
Design values for greywater and blackwater are derived from inhabitant-specific pollutant loads, according to the Chinese standard for the design of wastewater treatment plants (GB 50101-2005). The distribution of the loads between grey- and blackwater, as well as the inhabitant-specific flow rates, are calculated according to Bi (2004; cf. Tolksdorf et al. 2016). The measured influent concentrations differ considerably from these design values (cf. Figure 1); greywater is more concentrated than expected, whereas blackwater has lower concentrations.
Meinzinger & Oldenburg (2009) analysed literature data on grey- and blackwater composition. The medians for COD and TP concentration of greywater in their study lie only slightly lower than the medians measured at the influent to the greywater treatment in the Qingdao RRC (cf. Figure 2:  and ). Nonetheless, lower concentrations were expected because it was intended to exclude kitchen wastewater from greywater. Based on the COD and TP concentrations, kitchen wastewater is probably included in the greywater. However, the measured TN influent concentration of 55 mg/L is significantly higher than literature values for light greywater (DWA 2008), as well as for total greywater (Meinzinger & Oldenburg 2009). Cross-connections between grey- and blackwater, as well as unexpected user behaviour are possible explanations for the difference between design or literature values and the measured concentrations. The relatively low C/N ratio of greywater (median 10) indicates that proportionally more nitrogen enters into the greywater; in comparison, Todt et al. (2015) calculated a C/N ratio of approximately 17. Discharge of excreta into greywater may be an explanation. Colony-forming units (CFU) of E. coli were measured six times in the influent of the blackwater and greywater module. With values between 3.3·105 and 4.9·106 CFU/mL for greywater and 4.3·105 to 4.5·106 CFU/mL for blackwater, the results do not differ between both types of wastewater. This supports the assumption of cross-connections.
The median COD concentration of blackwater to the RRC is, at 744 mg/L, low compared to values reported by Chen et al. (2010) and Knerr (2012) for raw blackwater. The same applies to the median of the TN and TP concentrations (cf. Figure 3). It was expected that blackwater to the RRC is less concentrated, compared to the literature data, due to the inclusion of kitchen wastewater. Nonetheless, the measured values are still lower than the design values. Literature values for blackwater including kitchen wastewater were not found. The relatively low concentration might result from higher water consumption, infiltration water and/or from discharge of greywater to the blackwater sewer. On the basis of the relatively high blackwater influent flow between 2 and 4 am (cf. Figure 4), it can be concluded that part of the higher dilution is explained by infiltration water, probably groundwater, because the groundwater level in the area is relatively high. Assuming a constant infiltration rate over the day and observed months, and excluding the infiltration, a concentration of 150 mg/L TN, 915 mg/L COD and 13 mg/L TP was calculated, equalling 93%, 71% and 65% of the design values, respectively. The greater difference for COD can be explained by the discharge of kitchen wastewater to greywater. TP, in contrast, seems to be too low.
With the aim of verifying the assumptions regarding total loads (sum of greywater and blackwater), the concentration of total wastewater was calculated, using data for representative effluent flows from pre-storage tanks. For example, data from days with lower effluent flow, due to maintenance work such as membrane cleaning, were excluded. TN and COD concentrations are near the expected value for total wastewater (102% and 90%). With respect to COD, it has to be taken into account that greywater samples are taken behind a 1 mm sieve; hence, COD is partly removed (according to the design calculation: approx. 10%). TP concentration is lower than expected (approx. 30%). Apart from excreta, phosphorous in wastewater is mainly derived from detergents. The usage of detergents with a low P content might be an explanation. Moreover, at one of the hotels, only clothes of guests are washed within the hotel; the main laundry load is washed outside the catchment area. At ShiYuan village, wastewater from laundry might also be of minor importance (low numbers of guests). With the exception of TP, the assumptions for total inhabitant-specific loads seem relatively good and differences between assumed and measured concentration of blackwater and greywater may mainly be caused by cross-connections.
Diurnal variation of influent flow and characteristics
The blackwater influent flow increases between 6 and 8 am and is then relatively constant during the day until 10 pm, when it decreases (cf. Figure 4). Greywater, in contrast, shows more pronounced fluctuations, with several peaks between 8 am and 10 pm. For blackwater, a peak would have been expected in the morning, at a time when people are getting up (cf. Penn et al. 2012). Nonetheless, a relatively constant flow pattern during the day for toilet wastewater is to be expected. For greywater, in contrast, greater fluctuations due to various activities such as cooking, cleaning, and showering/bathing can be anticipated. Thus, it can be assumed that (although, on the basis of concentration levels, misconnections are likely) blackwater still consists, to a large extent, of toilet wastewater, whereas greywater is mainly influenced by wastewater from washing, showers, and kitchens.
The temperatures of grey- and blackwater at the influent of the RRC show large diurnal variations of between 10 and 17°C; surprisingly, blackwater is sometimes warmer than greywater (cf. Figure 5). The concentration of COD, TN, and TP also varies but, for both temperature and concentration levels (cf. Figure 7), not in an expected pattern. The variation does not reflect the typical changes expected from changing activities during the day (e.g., showering, cooking). The characteristic is mainly influenced by where within the catchment area the wastewater comes from during the sampling: higher temperature levels have usually been measured during times when wastewater arrives from the pressure pipe connecting one of the hotels and ShiYuan village. To evaluate the differences in the wastewater characteristics in the catchment area, additional samples were analysed.
Wastewater characteristics in the catchment area
The main wastewater flows to the RRC during sampling times originate from one of the hotels, one of the residential areas, and ShiYuan village. Only a few people currently live in the second residential area.
The temperature of the greywater and blackwater is low, between 10 and 11 °C (cf. Figure 6). Higher temperatures during the evening, due to shower wastewater, were not observed, contrary to expectations. Hence, wastewater from showers represented a minor proportion of the inflow during sampling times. According to a survey in October, residents save water, because this is the first time that they have had to pay for it (ISOE 2016). Moreover, older residents in particular are used to taking showers in public bathhouses, which they might, during the winter, do only once a week (ibid.). Although there are no demographic data, observations suggest that the proportion of older people is high. In conclusion, water consumption in the residential area is probably low, resulting in higher concentrations.
For greywater, the COD is higher during the first sampling day than during the second (cf. Figure 7). On the first day, the greywater was partly damned up at the shaft, due to a blockage in the sewer. A grease layer developed at the water surface, which explains the higher COD. This indicates that, in contrast to expectations, considerable amounts of kitchen wastewater are probably included in the greywater. The high nitrogen concentration can also be explained by kitchen wastewater; e.g., Chen et al. (2010) determined an average concentration of 43 mg/L TN and a range of 11 to 197 mg/L. Nonetheless, misconnection cannot be excluded, as it seems unlikely that kitchen wastewater is discharged during the whole day at the observed rate.
In the morning, the high nitrogen concentration of up to 200–300 mg/L in blackwater lies within the range of literature values (cf. Figure 3). The COD, at 1,500 mg/L, and TP, at 15 to 25 mg/L, are also comparable to blackwater samples analysed by Chen et al. (2010). Whereas TN is relatively constant after the morning, at about 120 mg/L, COD varies (although the two days do not reveal a distinct pattern). Compared to the results of Chen et al. (2010), the average concentrations of all parameters are lower, indicating dilution due either to greywater or to infiltration water. Blackwater contained visible food remains (either from kitchens or flushed down toilets).
The occupancy of the hotels differed on the two sampling days. During the first sampling, the hotel had only few guests, whereas on the second sampling day it was very busy because of a conference. According to an interview with the hotel engineer, the kitchen wastewater is discharged to greywater (after passing a grease trap). Due to the sampling point (pumping station), there is probably an equalization effect. The temperature level for both wastewater flows is higher, compared to the housing area (cf. Figure 6). At noon on the first day, the concentration of COD, TN, and TP in blackwater drops while the temperature rises. From this, as well as from the slight increase of the C/N ratio, the discharge of greywater to blackwater is assumed. A grab sample from the blackwater sewer before the pumping station on the second sampling day supports this assumption, because the blackwater temperature was 30°C, although the colour and odour clearly indicated a high proportion of toilet wastewater. It has to be noted that the blackwater samples from the pumping station are not representative due to the presence of a thick layer of sludge. Especially on the second sampling day, some samples included a higher proportion of solids and had COD concentrations above the measurement range (1,500 mg/L). The development of the sludge indicates that only part of the total load is pumped to the RRC (mainly COD and TP). The composition of greywater is relatively constant: the TN, at approx. 30 mg/L, is lower compared to greywater from the housing area. On the second sampling day, the C/N ratio is higher, possibly due to a higher percentage of kitchen wastewater. COD and TN concentrations are somewhat higher, compared to the literature data for total greywater (cf. Figure 2). Although misconnections cannot be excluded (supposedly, blackwater in the service area is discharged to greywater), the influence of kitchen wastewater from the hotel restaurant might explain the differences between the literature and the measured concentrations. The ratio of shower water, laundry, and kitchen wastewater in a hotel might not be comparable to that of a housing area.
The COD concentration of greywater demonstrates large fluctuations and, compared to other locations in the catchment area, very high concentrations (up to more than 2,000 mg/L). The TN concentration is slightly lower than that of greywater from the hotel. Greywater from ShiYuan village includes a high proportion of kitchen wastewater (from canteens and restaurants), because the guest houses and hotels have only few guests and, in office buildings, only small amounts of greywater are to be expected. At least one of the kitchens does not have a grease trap, resulting in the high COD concentration. Nonetheless, only part of the grease is pumped to the RRC, because a thick grease layer developed in the pumping station. Blackwater from the ShiYuan village (as well as from the hotel) is relatively warm (cf. Figure 6). In the biggest office building, the water is stored in the relatively warm basement, which explains the high temperature.
Conclusion regarding wastewater characteristics
The wastewater characteristics in the different parts of the catchment area varies, because of user habits, different predominant wastewater sources, and most likely, different degrees of cross-connection. The high variability of the influent characteristics (also in 24-h composite samples) is probably caused by changing percentages of wastewater flows from the different parts of the catchment area. In particular, the hotel has high fluctuations in occupancy, resulting in varying amounts of wastewater. At least during the winter, the percentage of shower wastewater at the housing area seems to be low, resulting in smaller amounts of greywater. The misconnections and unexpected user behaviour led to considerable deviations from design values. This strongly affects the operation of the RRC in Qingdao. Heat recovery is, conceptually, a possible component of the SEMIZENTRAL approach. Thus, a high potential was assumed for greywater. According to the measurements in the catchment area, greywater has relatively low temperature levels, and blackwater is sometimes even warmer. This demonstrates the importance of more detailed data collection before planning such a system.
Operation of greywater and blackwater modules
Currently, the grey- and blackwater flows lie considerably below the design capacity. The main reasons are only partial occupation of the residential area, as well as vacant office buildings and guest houses in ShiYuan village. Additionally, the hotels have varying numbers of guests. For this reason, the operation data presented here are a snapshot, representative of the current conditions. Water volumes and pollutant loads might change over time.
Under the assumption of a TN influent concentration of less than 5 mg/L for the design of the greywater treatment within the RRC (cf. Figure 1), nitrogen removal was not planned. For the currently measured influent characteristics, in contrast, nitrification and denitrification is necessary to meet the legal limit of 15 mg/L for TN (according to the Chinese standards GB 18918-2002 (Class 1A) and GB 18920-2002). The greywater treatment occurs in an MBR. The design sludge age was chosen to be 25 days (Tolksdorf et al. 2016), but is higher under the actual operation conditions; hence nitrification occurs as long as the aeration capacity is sufficient. The aeration basin consists of two consecutive chambers, each with an independent aeration grid. Additionally, mixers have been installed as the mixing energy from aeration alone would be insufficient. This design enables the establishment of pre-denitrification by turning off the aeration in the first chamber.
The strong variation of greywater flow and characteristics during the day (cf. above) is partly equalized, due to the pre-storage tank. The hydraulic retention time is, on average, 13 ± 9 hours. Following the initial start-up, the effluent standard can now be met (cf. Figure 8). Low effluent COD concentrations, often in the range of 20 mg/L, indicate good biodegradability. The average influent flow during the observed period is about 30% of the design capacity. The COD influent load (123 kg/d) amounts to 83% of the design value. The TN influent load, in contrast, lies well above the design value of 12.8 kg/d, which equals 376% of the design value. Thus, the oxygen demand is higher compared to the design calculation, but can be supplied by the cross-flow aeration. Nonetheless, without the aeration in the aeration basin, the ammonia effluent concentration starts to increase after one week (cf. Figure 8 after 70 days). Thus, cross-flow aeration on its own is not sufficient under these loading conditions. Because the oxygen concentration does not exceed 0.15 mg/L and is often considerably lower, simultaneous nitrification/denitrification can be assumed in the aeration basin. The sludge loading rate is, at approx. 0.07 kg COD/(kgMLSS·d), low; therefore, aerobic sludge stabilization can be assumed.
The greywater flow and loads can increase, if more residents move to the housing areas. Higher occupancy of the hotels/guest houses in the summer can also increase wastewater amounts. The treatment capacity of the greywater module may thus be insufficient at some point in the future. In this case, part of the greywater can be bypassed to the blackwater module.
The average influent flow during the observed period is 344 m3/d and equals 43% of the design capacity. The COD load is 289 kg/d, and the TN load 36 kg/d (26% and 25% of the design values). The aeration basin consists of two parallel lines, of which only one was commissioned, because of the lower influent flow and load. The low COD effluent concentration of about 25 mg/L (cf. Figure 8) indicates good biodegradability. The sludge age is well above 25 days, indicating aerobic sludge stabilization. The lower TN influent concentration results in lower required removal rates: An influent TN concentration of 162 mg/L was expected and, therefore, 91% of TN should have had to be eliminated. In reality, with the average TN influent concentration at 110 ± 29 mg/L during the observed 90-day period, only 86% have to be removed. Hence, less nitrate has to be denitrified. No external carbon dosage is necessary for denitrification, in contrast to the planned design. It has to be remembered, however, that the pre-clarifier, which would reduce the C/N ratio, has been bypassed during this time. The high denitrification capacity is explained by the high proportion (67%) of denitrification volume in the aeration basin (60%, excluding the post-denitrification). The influent load will increase when the co-digestion of food waste and sludge is in full operation and sludge liquor from sludge dewatering is discharged to the influent of the blackwater module. Currently, the digestion is commissioned. Therefore, the operational data given here are representative only for the actual loading conditions and might change over time.
The implementation of the RRC has revealed considerable differences between the assumed and actually measured greywater and blackwater characteristics. Even within the catchment area, notable differences for blackwater and greywater composition have emerged, depending on user behavior, predominant area usage, and possible cross-connections. Literature data for greywater and blackwater characteristics also show high concentration ranges. Although many studies have been conducted to characterise separated wastewater streams, the deduction of design values for source-oriented infrastructure is still a challenge. There are only a few well documented data with similar boundary conditions; the variability of influent data is high (DWA 2014). Moreover, smaller systems usually show higher variability regarding wastewater characteristics and flow. Apart from variations in inhabitant-specific loads due to user habits (such as diet), uncertainties regarding the distribution to the separated wastewater flows have to be taken into account. Cross-connections are an additional risk and measures for their reduction are recommendable. For example, more detailed information for planners of building technologies, architects, and workers should be supplied. Because a certain degree of misconnection might not be avoidable, but difficult to determine beforehand, the treatment processes should be flexible, enabling the adaptation to differences between design values and actual influent concentrations and loads. MBR plants, for example, make it possible to adapt the amount of sludge in the system, at a constant basin volume, by altering the sludge concentration in a wide range. Bypasses that are located between grey- and blackwater flows offer additional flexibility. Pre-storage tanks are recommendable in case high variations of inflow loads and flow occur. Intra-urban water reuse is essential to ensure sufficient water supply in fast-growing urban areas. For the implementation in Qingdao, greywater and blackwater separation was chosen, because higher public acceptance for the reuse of treated greywater was expected. Nonetheless, this advantage has to be weighed against the technical challenges deriving from uncertainties/risks caused by the separation. The reuse of treated total wastewater is technically feasible and (with appropriate public relations work) a viable alternative. The SEMIZENTRAL approach is not limited to specific material flows and process technologies; both have to be adapted to local conditions.
The SEMIZENTRAL approach can contribute to sustainable water supply and treatment in fast-growing urban areas. Water reuse and energy self-sufficient operation enable high resource-efficiency. Because implementation is related to city district development, the infrastructure can develop in parallel with the growth rate of the city. The implementation of the first semicentralized RRC has yielded valuable experience in terms of planning of integrated systems for the supply and treatment of water and wastewater.