Start-Up Evaluation of a Full-Scale Wastewater Treatment Plant Consisting of a UASB Reactor Followed by Activated Sludge

: UASB (upﬂow anaerobic sludge blanket) reactors have been recognized as a viable option for sewage treatment. However, in order to improve the UASB efﬂuent quality, some type of post-treatment must be implemented. The aims of this study were (i) to establish a start-up methodology of a full-scale anaerobic–aerobic system treating sewage, (ii) to evaluate the concentrations of different constituents in the inﬂuent and efﬂuent of the anaerobic and aerobic reactors as well as the removal efﬁciencies in every step of the system, and (iii) to deﬁne relevant operative aspects of the anaerobic and aerobic reactors. The Tunja (Colombia) wastewater treatment plant consists of three modules with preliminary treatment followed by UASB reactors with post-treatment of activated sludge. The results of this investigation showed that the efﬂuent system meets the Colombian environmental legislation with average removal efﬁciency values of BOD (88 +/ − 5%), COD (87 +/ − 4%), and TSS (94 +/ − 5%). The UASB reactor start-up was conducted without an inoculum, requiring a period of 120 days. The evaluation of the combined systems was conducted over 300 days. Moreover, a methodology to operate the system during and after the start-up of the anaerobic reactor was deﬁned. It was demonstrated that the anaerobic efﬂuent can deteriorate the sludge in the aerobic tank. In order to avoid this, important operational aspects must be considered during the operation of the system, such as the implementation of a raw wastewater bypass higher than 15% and monitoring of the anaerobic efﬂuent settleable solid concentration (<0.3 mL/L).


Introduction
It is recognized that anaerobic systems are suitable for the treatment of high-strength wastewaters (COD < 4000 mg/L), and that aerobic systems are applicable for the treatment of low-strength wastewaters (COD < 1000 mg/L), bringing out an effluent quality generally higher than the anaerobic system [1]. Therefore, a combination of both systems (anaerobic + aerobic) is a viable option for sewage treatment. When compared with conventional aerobic technologies, based on activated sludge processes, this combination offers lower energy consumption and lower excess sludge production [2].
Activated sludge and upflow anaerobic sludge blanket (UASB) reactors are among the most used technologies for sewage treatment. The latter has been used mainly in Latin America and middle-income countries [3,4] such as Brazil, Colombia, and India [5][6][7].
According to [6], UASB reactors are compact and economical units with low operation costs and low excess sludge production. At present, UASB reactors are recognized as the average removal efficiencies of the systems were as follows: COD, 86 and 83%; BOD, 94%; and TSS, 61 and 64%.
The recommendations and procedures for the start-up and operation of UASB reactors treating sewage have been well-documented previously [25]. However, there is scarce information about the steps and methods that can improve the start-up of full-scale systems that include UASB reactors followed by aerobic treatment for sewage treatment. The authors of [14] evaluated the operation of a pilot UASB-AS system, recommending a global hydraulic retention time of 7.9 h distributed as follows: 4.0 h UASB; 2.8 h aerobic reactor; and 1.1 h final clarifier. The authors of [26] presented an operational strategy for a combined pilot-scale UASB-AS system treating an influent with the typical sewage composition from the city of Belo Horizonte (Brazil). The selected control variables were the solid concentration in the effluent and the sludge wastage rate. The best control strategy was based on two cascaded PI controllers for the solid concentration in the effluent and a look-up table for the excess sludge rate.
In particular, regarding the combination of UASB and AS, even though it is considered a viable post-treatment option [6,11] offering high removal efficiencies for BOD, COD, and TS [27], there are few studies evaluating full-scale systems [28]. Moreover, these studies are oriented to the evaluation of the systems but not the start-up methodology. Therefore, this investigation aimed to define a start-up methodology for full-scale UASB-AS systems treating sewage, including guidelines to improve the performance and relevant operative aspects. The novelty of this research is focused on the identification of the relevant operative aspects that affect the start-up and operation of the combined system. This is particularly important due to the influence of the anaerobic effluent on the aerobic system. This research was conducted in a real-scale sewage treatment plant (STP) in Tunja (Colombia). The treatment system is a combined anaerobic (UASB)-activated sludge system. The results of this study will allow for establishing a start-up methodology that can be adopted worldwide for similar wastewater treatment systems, with the proposed combination being a valuable contribution because it has been recognized as a viable option for sewage treatment.

Description of the Study Area
Tunja is the capital of the county of Boyaca in Colombia. It is located in the eastern chain of the Andes. The average annual temperature in Tunja is 16 • C. The sewage treatment plant (STP) is located at an altitude of 2750 m above sea level. The raw wastewater is transported to the STP by a gravity-combined sewer. The treatment processes at the Tunja STP have three modules operating in parallel. Each module consists of a train operating in parallel that includes coarse and fine screening, grit removal, a UASB reactor, an activated sludge aerobic aeration tank, and secondary sedimentation. Actually, it is the largest Colombian wastewater plant using this type of treatment sequence. The designed STP flow is 120 L/s per module. The produced anaerobic and aerobic excess sludge is taken, after it is thickened and dewatered, to the final disposal. Figure 1 (produced with a Capdeworks 4.0 academic license) shows the flowsheet of a module of the Tunja STP.
This research was conducted in one of the modules of the STP. Each UASB reactor had a working volume of 2895 m 3 (area: 579 m 2 ). The tank water depth was 5.0 m. Each aeration tank had a volume of 2469 m 3 (area: 588 m 2 ). The total liquid depth was 4.2 m. The effluent of the aeration tank was received by two circular secondary settlers (diameter: 23.5 m) with a liquid depth of 4.5 m. Table 1 shows the main design parameters of the STP units. This research was conducted in one of the modules of the STP. Each UASB reactor had a working volume of 2895 m 3 (area: 579 m 2 ). The tank water depth was 5.0 m. Each aeration tank had a volume of 2469 m 3 (area: 588 m 2 ). The total liquid depth was 4.2 m. The effluent of the aeration tank was received by two circular secondary settlers (diameter: 23.5 m) with a liquid depth of 4.5 m. Table 1 shows the main design parameters of the STP units. The influent flow was measured continuously with a Pulsar flowmeter (FlowCert Model) that uses an Ultra Data Logger unit. The maximum average flow during this evaluation was 90 L/s. The average temperature of the raw wastewater was 19 +/− 0.7 °C. The dissolved oxygen in the aeration tanks was measured online with a HACH sc200 universal controller equipped with a HACH 9020000 probe. The aeration tank was operated with intermittent aeration (on: 45 min; off: 15 min), maintaining the dissolved oxygen concentration between 1.5 mg/L and 2.5 mg/L. However, the aeration time was changed when the dissolved oxygen concentrations were lower than the minimum value.

The Start-Up of the Anaerobic-Aerobic System
The water quality parameters measured in the liquid phase (raw wastewater; UASB effluent; final effluent) were temperature, pH, biochemical oxygen demand (BOD), and chemical oxygen demand (COD) as well as total suspended solids (TSS), settleable solids, organic nitrogen, and ammonia nitrogen. Flows of proportional composite samples were collected at the points presented above. The procedures for collecting and preserving as well as the analytical methods that were implemented were based on the recommendations of [29].  The influent flow was measured continuously with a Pulsar flowmeter (FlowCert Model) that uses an Ultra Data Logger unit. The maximum average flow during this evaluation was 90 L/s. The average temperature of the raw wastewater was 19 +/− 0.7 • C. The dissolved oxygen in the aeration tanks was measured online with a HACH sc200 universal controller equipped with a HACH 9020000 probe. The aeration tank was operated with intermittent aeration (on: 45 min; off: 15 min), maintaining the dissolved oxygen concentration between 1.5 mg/L and 2.5 mg/L. However, the aeration time was changed when the dissolved oxygen concentrations were lower than the minimum value.

The Start-Up of the Anaerobic-Aerobic System
The water quality parameters measured in the liquid phase (raw wastewater; UASB effluent; final effluent) were temperature, pH, biochemical oxygen demand (BOD), and chemical oxygen demand (COD) as well as total suspended solids (TSS), settleable solids, organic nitrogen, and ammonia nitrogen. Flows of proportional composite samples were collected at the points presented above. The procedures for collecting and preserving as well as the analytical methods that were implemented were based on the recommendations of [29].

UASB Start-Up
The start-up of the combined system was conducted starting with the aerobic reactor, a process that took twenty days. Subsequently, the UASB reactor required a period of 120 days to finish the start-up despite the psychrophilic operation condition. The combined system evaluation was conducted over 300 days. The procedure is described in the following sections.
In this project, the start-up of the UASB reactor was followed, after taking into consideration the impacts of the anaerobic effluent on the operation of the aeration tank. In UASB reactors treating sewage, certain studies found that the start-up process can be conducted without the addition of seed sludge [5,30]. Taking this into consideration, the UASB start-up was carried out without an inoculum.
The UASB start-up was evaluated by considering: (1) the growth of the sludge mass while monitoring the sludge bed and measuring the sludge total solids (TS) and (2) the change in the removal efficiency of organic matter (COD and BOD 5 removal), TSS, and settleable solids in the UASB effluent. Table 2 shows the parameters and operation variables that were measured with their frequencies in the raw wastewater and the UASB effluent. The procedures for collecting the samples and the analytical methods were carried out following the recommendations of [29]. Biogas production was not quantified as the evacuation systems were being installed during the evaluation period. The statistical analysis tool used in this work was Minitab ® 21 software from Minitab Inc. The normal probability distribution using the Kolmogorov-Smirnov test for the TSS, BOD 5 , and COD concentrations in the raw sewage, UASB effluent, and effluent did not reject the null hypothesis at p > 0.05.
The sludge bed profile was measured weekly at 6 sampling points located at different depths of the UASB digestion compartment (0.82; 1.33; 1.82; 2.33; 2.83; 3.34 m). The specific methanogenic activity (SMA) of the anaerobic sludge was measured weekly, by taking samples from the middle of the digestion compartment. The assay was implemented with agitation (120 rpm) using an IKA KS4000 incubator shaker at 20 • C. Both parameters were determined using the methods described by [31][32][33].

Activated Sludge Start-Up
The start-up of the aeration tank was conducted previously with a flow of 6653 m 3 /d. It took around 21 days, reaching a BOD removal of around 80% and the established criteria in the aeration tank for MLSS of 3500 mg/L [34] and SVI of 80 mL/g [35]. The evaluation of the combined system was then started.
During the first 22 days, the UASB influent flow was 6653 m 3 /d (85% of the total flow). In this flow, deterioration of the sludge in the aeration tank was observed. For this reason, the strategy for the UASB start-up was to reduce the UASB influent flow to 5443 m 3 /s, and to monitor the UASB sludge blanket growth, the UASB effluent settleable solids, and the MLSS in the aeration tank. This operation mode allowed the adaptation of the aerobic reactor to the anaerobic influent in terms of the SVI and a dissolved oxygen concentration between 1.5 and 2.0 mg/L [31].
It is expected that a low organic removal efficiency at the beginning of the operation of the UASB reactor is the best option in order to avoid a low F/M ratio in the aerobic reactor. In the steady-state condition, 85% of the raw wastewater flow goes to the UASB reactor (following the recommendation of [22]), and 15% is transported by the bypass to the aeration tank. Table 3 shows the flow treated during the evaluation. The maximum flow treated during the steady-state condition was 7776 m 3 /d, where 6653 m 3 /d was treated in the UASB and 1123 m 3 /d was transported via the bypass to the aeration tank. In this operation condition, the aeration was implemented in a sequence of 45 min on and 15 min off. The sludge recycling had a flow of 3024 m 3 /d for 1 h every 2 h. Table 4 shows the parameters, methods, and the frequency measured in the aeration tank. The SVI was evaluated following the procedure suggested by [36]. The F/M and VOL were calculated using the BOD mass balance between the UASB effluent and the bypass.

Final Water Quality Verification
The water quality of the influent and effluent was evaluated at the end of the evaluation period with a 24 h composite sampling. This was carried out to evaluate the STP efficiency, taking into consideration the Colombian regulatory limits. The following parameters, with their analytical methods, were measured: Acidity (SM 2310 B); total alkalinity (SM 2320 B); COD (SM 5220 D); BOD 5

Start-Up of the Anaerobic-Aerobic System
The start-up of the STP was initiated with the aerobic system without an inoculum. This procedure took 21 days and showed fast aerobic biomass growth. The authors of [37] mentioned that the typical duration for the start-up time of activated sludge systems without an inoculum is 30 days.
After the aeration tank start-up, the UASB start-up was initiated, taking 85% of the total raw wastewater flow as the UASB influent and the rest as a bypass to the aeration tank. This period of the combined anaerobic-aerobic system took 121 days. The final period was between 122 and 300 days and corresponded to the system operation in a steady state. Deterioration of the sludge quality was observed in the first 30 days of the start-up and was attributed to the impact of the anaerobic effluent on the aeration tank sludge. Figure 2 shows details of the F/M, VOL, SVI, and sludge color variation in the aeration tank during the start-up and steady-state phases. The sludge color is a good indicator of the process stability; this qualitative evaluation parameter is very useful as a fast method for making decisions during the aeration tank operation [38,39]. The STP operators were instructed to tune the operation parameters according to their visual observation as described below: • Brown: earthy odor indicates normal operation; no adjustment required. • Light tan: odor indicates extremely young sludge; decreased sludge wasting. • Dark brown: earthy odor indicates old sludge with high solid concentration; increased sludge wasting. • Black: rotten-egg odor indicates septic conditions with a low dissolved oxygen concentration; increased aeration.
During the start-up phase in the aeration tank, the predominant sludge colors were black and dark brown, indicating unstable operational conditions. This was related to the influence of the anaerobic effluent on the aeration tank sludge. This deficiency was managed by controlling the UASB effluent settleable solid concentration by increasing the bypass or by increasing the aeration time.
After the start-up ended (100-120 days), the sludge color improved (light tan), evidencing the stability of the process in concordance with the reference literature values for F/M (0.2-0.6 kg BOD 5 /MLVSS * d) and SVI (100-200 mL/g) for conventional activated sludge systems [34].
During the steady-state phase (120-300 days), the predominant sludge color was light tan, an indication of young sludge in the stable activated sludge process. Between days 263 and 289, a color change (light tan to dark brown), characterized by F/M ratios below 0.2 kg BOD 5 /kg MLVSS * d and a considerable increase in the SVI (>200 mL/g), was observed. This was an indication of adverse conditions that would lead to deterioration of the aerobic sludge.
The low values of F/M and VOL between days 263 and 289 were associated with the UASB efficiency increase for organic matter removal (BOD 5 , 68%). This caused COD and BOD 5 effluent average concentrations of 230 mg/L and 102 mg/L, respectively. This is evidence of the high sensitivity of the aerobic system to the anaerobic effluent. Despite this, the robustness of the combined system was demonstrated by the stability of the global removal efficiencies. During the start-up phase in the aeration tank, the predominant sludge colors were black and dark brown, indicating unstable operational conditions. This was related to the influence of the anaerobic effluent on the aeration tank sludge. This deficiency was managed by controlling the UASB effluent settleable solid concentration by increasing the bypass or by increasing the aeration time.
After the start-up ended (100-120 days), the sludge color improved (light tan), evidencing the stability of the process in concordance with the reference literature values for F/M (0.2-0.6 kg BOD5/MLVSS * d) and SVI (100-200 mL/g) for conventional activated sludge systems [34].
During the steady-state phase (120-300 days), the predominant sludge color was light tan, an indication of young sludge in the stable activated sludge process. Between days 263 and 289, a color change (light tan to dark brown), characterized by F/M ratios below 0.2 kg BOD5/kg MLVSS * d and a considerable increase in the SVI (>200 mL/g), was observed. This was an indication of adverse conditions that would lead to deterioration of the aerobic sludge.
The low values of F/M and VOL between days 263 and 289 were associated with the UASB efficiency increase for organic matter removal (BOD5, 68%). This caused COD and BOD5 effluent average concentrations of 230 mg/L and 102 mg/L, respectively. This is evidence of the high sensitivity of the aerobic system to the anaerobic effluent. Despite this, the robustness of the combined system was demonstrated by the stability of the global removal efficiencies. Taking into account the efficiency of the UASB organic matter removal, treating sewage is important to avoid low values of VOL (<0.4 kg BOD/m 3 * d) in the influent of the aeration tank that cause low F/M ratio values. This can be avoided by increasing the raw wastewater bypass to the aeration tank. In the aeration tank, the control of the F/M ratio and the volumetric organic loading can be conducted by implementing a raw wastewater bypass higher than 15% of the influent flow to the aeration tank. Other corrective measures are the extraction of excess sludge from the UASB reactor and an increase in the aeration time and the raw wastewater bypass.
The aeration tank showed high-efficiency removal during the start-up and steady-state phases, as shown in Figure 3.
During the start-up phase, the aeration tank removal efficiencies were as follows: COD (74 +/− 5.9); BOD (68 +/− 14); TSS (69 +/− 15). The size of the box indicates a low data dispersion, and the similar mean and median values are an indication of the data's normal probability distribution. The atypical values are related to the deterioration of the effluent caused by the high settleable solid concentration in the anaerobic effluent.
In the steady-state phase, the average removal efficiencies were as follows: COD (77 +/− 12); BOD (76+/− 17); TSS (81 +/− 18). In both phases, the low efficiencies were related to the UASB effluent settleable solid concentration, which was higher than 0.3 mL/L. Figure 4 shows the variation in the UASB effluent settleable solid concentration, the total solid mass in the UASB sludge blanket, and the dissolved oxygen concentration in the aeration tank during the start-up and steady-state operation of the combined system. wastewater bypass to the aeration tank. In the aeration tank, the control of the F/M ratio and the volumetric organic loading can be conducted by implementing a raw wastewater bypass higher than 15% of the influent flow to the aeration tank. Other corrective measures are the extraction of excess sludge from the UASB reactor and an increase in the aeration time and the raw wastewater bypass.
The aeration tank showed high-efficiency removal during the start-up and steadystate phases, as shown in Figure 3. In the steady-state phase, the average removal efficiencies were as follows: COD (77 +/− 12); BOD (76+/− 17); TSS (81 +/− 18). In both phases, the low efficiencies were related to the UASB effluent settleable solid concentration, which was higher than 0.3 mL/L. Figure 4 shows the variation in the UASB effluent settleable solid concentration, the total solid mass in the UASB sludge blanket, and the dissolved oxygen concentration in the aeration tank during the start-up and steady-state operation of the combined system. During the UASB sludge blanket formation, corresponding to the first 90 days, the anaerobic effluent settleable solid concentration was below the analytical DL (<0.1 mL/L). As expected, with the sludge blanket growth, the UASB effluent settleable solid concentration increased.
Between days 95 and 200, there were events with a low dissolved oxygen concentration in the aeration tank that were associated with a settleable solid concentration higher than 0.3 mL/L in the anaerobic effluent. This is evidence of the high sensitivity of the aerobic tank to the anaerobic effluent in terms of the anaerobic effluent settleable solid concentration. To solve this problem, starting from day 200, measurements of the settleable solid concentration in the anaerobic effluent and the steady UASB sludge bed profile were used as control variables to maintain a dissolved oxygen concentration in the aeration tank of 2 mg/L.
According to [7,40], the maximum sludge storage capacity of UASB reactors can be identified using the sludge profile and the settleable solid measurements. For the latter criterion, these authors recommended a maximum value of 1 mL/L. In this case, it was found that the UASB effluent with settleable solids higher than 0.3 mL/L caused anaerobic conditions in the aeration tank.
The UASB start-up without an inoculum was completed in 4 months (120 days), despite the wastewater temperature corresponding to psychrophilic conditions (minimum 17.6 • C, maximum 19.6 • C). The average effluent pH was 7.0 +/− 0.3. At this time, the sludge profile showed that the UASB maximum sludge storage capacity in the digestion zone was 60-ton TS (average concentration: 38 kg TS/m 3 ). Figure 5 shows the UASB sludge bed profile after 300 days of the operation period. A clear distinction can be seen between the sludge bed and the sludge blanket, which is an indication of the good sludge quality [32]. In this case, 83% of the sludge mass was in the sludge bed. During the UASB sludge blanket formation, corresponding to the first 90 days, the anaerobic effluent settleable solid concentration was below the analytical DL (<0.1 mL/L). As expected, with the sludge blanket growth, the UASB effluent settleable solid concentration increased.
Between days 95 and 200, there were events with a low dissolved oxygen concentration in the aeration tank that were associated with a settleable solid concentration higher than 0.3 mL/L in the anaerobic effluent. This is evidence of the high sensitivity of the aerobic tank to the anaerobic effluent in terms of the anaerobic effluent settleable solid concentration. To solve this problem, starting from day 200, measurements of the settleable solid concentration in the anaerobic effluent and the steady UASB sludge bed profile were used as control variables to maintain a dissolved oxygen concentration in the aeration tank of 2 mg/L.
According to [7] and [40], the maximum sludge storage capacity of UASB reactors can be identified using the sludge profile and the settleable solid measurements. For the latter criterion, these authors recommended a maximum value of 1 mL/L. In this case, it was found that the UASB effluent with settleable solids higher than 0.3 mL/L caused anaerobic conditions in the aeration tank.
The UASB start-up without an inoculum was completed in 4 months (120 days), despite the wastewater temperature corresponding to psychrophilic conditions (minimum 17.6 °C, maximum 19.6 °C). The average effluent pH was 7.0 +/− 0.3. At this time, the sludge profile showed that the UASB maximum sludge storage capacity in the digestion zone was 60-ton TS (average concentration: 38 kg TS/m 3 ). Figure 5 shows the UASB sludge bed profile after 300 days of the operation period. A clear distinction can be seen between the sludge bed and the sludge blanket, which is an indication of the good sludge quality [32]. In this case, 83% of the sludge mass was in the sludge bed. The average operational criteria of the UASB during the evaluation were an upflow velocity of 0.5 +/− 0.04 m/h and an organic volumetric load of 1.3 +/− 0.2 kg COD/m 3 * d. The anaerobic sludge SMA changed during the evaluation from 0.02 g COD-CH4/g VSS * d (start-up) to 0.2 g COD-CH4/g VSS * d (steady state), a typical value for UASB flocculent sludge [41].
The amount of UASB sludge to waste was defined based on the measured sludge The average operational criteria of the UASB during the evaluation were an upflow velocity of 0.5 +/− 0.04 m/h and an organic volumetric load of 1.3 +/− 0.2 kg COD/m 3 * d. The anaerobic sludge SMA changed during the evaluation from 0.02 g COD-CH 4 /g VSS * d (start-up) to 0.2 g COD-CH 4 /g VSS * d (steady state), a typical value for UASB flocculent sludge [41].
The amount of UASB sludge to waste was defined based on the measured sludge blanket, taking the volume corresponding to the sludge present in the upper sampling point (3.34 m) of the UASB digestion compartment. The sludge extraction point was selected as it allows the withdrawal of the less concentrated sludge with the worst settling conditions and the preservation of the more concentrated, best-quality sludge.
The UASB reactor excess sludge production at the end of the evaluation period (290 d) was 0.8 ton TS/d. As is already known, the temperature has an impact on the required UASB sludge age. In this evaluation, the wastewater average temperature was 19 • C, the calculated UASB sludge considered the excess sludge, and the UASB effluent TSS load was 100 days, a value that is coherent with calculations performed in previous studies [41].
Although the UASB reactor is considered a secondary treatment, it can provide a removal efficiency of BOD 60% and COD 70% [42]. In this STP, the expected organic removal efficiency of the UASB reactor is less in order to maintain an appropriate F/M ratio in the aeration tank. Figure 6 shows the COD, BOD, and TSS removal efficiencies in the UASB reactor during the two evaluation phases. The average removal efficiencies of COD in the UASB reactor during the start-up and steady state were 59 +/− 7 and 66 +/− 8%, respectively. The confidence intervals based on the median allow us to conclude that, with a 95% confidence interval, the UASB COD removal efficiencies were 56-62% at the start-up and 63-71% in the steady state. A statistical two-sample t-test showed (p-value < 0.05) significant statistical differences between the two phases for COD removal. In the start-up and steady-state phases, the UASB COD effluent values had median values of 401 mg/L and 226 mg/L, with 95% confidence intervals based on the medians of 275-348 mg/L and 208-271 mg/L, respectively.
The UASB DBO5 removal efficiencies evaluated during the start-up and steady state were 55 +/− 18 and 60 +/− 12%, respectively. The confidence intervals based on the median allow us to conclude that, with a 95% confidence interval, the UASB BOD5 removal effi- The average removal efficiencies of COD in the UASB reactor during the start-up and steady state were 59 +/− 7 and 66 +/− 8%, respectively. The confidence intervals based on the median allow us to conclude that, with a 95% confidence interval, the UASB COD removal efficiencies were 56-62% at the start-up and 63-71% in the steady state. A statistical two-sample t-test showed (p-value < 0.05) significant statistical differences between the two phases for COD removal. In the start-up and steady-state phases, the UASB COD effluent values had median values of 401 mg/L and 226 mg/L, with 95% confidence intervals based on the medians of 275-348 mg/L and 208-271 mg/L, respectively.
The UASB DBO 5 removal efficiencies evaluated during the start-up and steady state were 55 +/− 18 and 60 +/− 12%, respectively. The confidence intervals based on the median allow us to conclude that, with a 95% confidence interval, the UASB BOD 5 removal efficiencies were 39-72% at the start-up and 49-70% in the steady state. There was no significant statistical difference (p-value > 0.05) in the BOD between the start-up and steadystate phases. In the start-up and steady-state phases, the UASB BOD effluent concentrations had median values of 160 mg/L and 165 mg/L, with 95% confidence intervals based on the medians of 135-192 mg/L and 108-182 mg/L, respectively.
The UASB TSS removal efficiencies during the start-up and steady state were 75 +/− 9 and 79 +/− 10%, respectively. The confidence intervals based on the median allow us to conclude that, with a 95% confidence interval, the UASB TSS removal efficiencies were 73-80% at the start-up and 79-85% in the steady state. A statistical two-sample t-test showed (p-value < 0.05) statistically significant differences between the two phases in UASB TSS removal. In the start-up and steady-state phases, the UASB TSS effluent concentrations had median values of 67 mg/L and 38 mg/L, with 95% confidence intervals based on the medians of 56-78 mg/L and 35-48 mg/L, respectively. The median effluent value in the UASB is an indication of the high TSS and the relative fast growth of the reactor sludge blanket. It is also connected with the 4-month UASB start-up time.
It was evident that the combination of technologies (UASB-AS) provides robustness and reliability to the global system performance in terms of the removal efficiency. Despite the failures during the start-up phase, the global removal efficiencies were not affected. Figure 7 shows the COD, BOD 5 , and TSS global removal efficiencies of the STP during the two phases. It was evident that the combination of technologies (UASB-AS) provides robustness and reliability to the global system performance in terms of the removal efficiency. Despite the failures during the start-up phase, the global removal efficiencies were not affected. Figure 7 shows the COD, BOD5, and TSS global removal efficiencies of the STP during the two phases. The removal efficiencies of COD, BOD, and TSS of the combined system during the start-up and steady-state conditions were evaluated. A statistical two-sample t-test showed (p-value < 0.05) significant differences between the two phases in COD and TSS removal. The 95% confidence intervals based on the median were as follows: COD (startup: 83-86%; steady state: 87-90%); TSS (start-up: 89-91%; steady state: 94-97%). In contrast, no significant statistical difference was found in the BOD5 removal efficiency (pvalue > 0.05), with a 95% confidence interval (start-up: 83-94%; steady state: 84-92%).
The average COD removal efficiency (UASB-AS) was 83 +/− 4% at the start-up and 88 +/− 4 in the steady state. These values are similar to those reported in a previous study by [15] in a UASB-AS system, which ranged from 85 to 93%. Moreover, the average DBO5 The removal efficiencies of COD, BOD, and TSS of the combined system during the start-up and steady-state conditions were evaluated. A statistical two-sample t-test showed (p-value < 0.05) significant differences between the two phases in COD and TSS removal. The 95% confidence intervals based on the median were as follows: COD (start-up: 83-86%; steady state: 87-90%); TSS (start-up: 89-91%; steady state: 94-97%). In contrast, no significant statistical difference was found in the BOD 5 removal efficiency (p-value > 0.05), with a 95% confidence interval (start-up: 83-94%; steady state: 84-92%).
The average COD removal efficiency (UASB-AS) was 83 +/− 4% at the start-up and 88 +/− 4 in the steady state. These values are similar to those reported in a previous study by [15] in a UASB-AS system, which ranged from 85 to 93%. Moreover, the average DBO 5 removal efficiencies of the STP were 87 +/− 7% at the start-up and 87 +/− 5% in the steady state. These values are similar to the values reported by [22]. In terms of the TSS removal efficiency of the combined system, it was 90 +/− 4% and 94 +/− 5%, respectively, at the start-up and in the steady state. Table 5 shows the COD, BOD 5 , and TSS concentrations in the raw sewage as well as the UASB effluent and the effluent of the STP. The COD, BOD 5 , and TSS effluent concentrations are below the Colombian legislation required standard (Ministerio de Vivienda y Territorio, Resolución 631 de 2015) are as follows: COD = 180 mg/L; BOD 5 = 90 mg/L; TSS = 90 mg/L. Table 5. STP effluent COD, BOD, and TSS concentrations at the start-up and in the steady state.  Table 6 shows a comparison of the results of this investigation with other full-scale UASB-AS systems treating sewage in terms of the HRT applied in the UASB and the aeration tank as well as the final effluent concentrations and removal efficiencies of BOD, COD, and TSS. As shown in this evaluation, the performance of the Tunja sewage treatment plant demonstrated higher efficiencies than the experiences in India, with similar results to those reported in Brazil.
Based on the reported experiences, for the UASB+AS combination, it is evident that the UASB operates in a low HRT. In this investigation, the UASB was operated with a lower flow than the design value; therefore, the HRT was higher than the design value (6 h).

Final Water Quality Verification
The steady-state COD balance at the end of the evaluation period was built with the free STAN software [43]. This is a tool for material flow analysis of goods, substances, and energy that allows the user to calculate the unknown flows considering data uncertainties through error propagation. In this case, the unknown values were the amounts of influent COD converted to methane in the UASB reactor and to CO 2 in the aerobic reactor. The other flow values with their uncertainties were measured. Figure 8 shows that the total average amount of COD converted was 85%. The most important COD average flow removal in the STP is the production of aerobic sludge (32%), followed by the anaerobic sludge production in the UASB reactor (18%). The COD converted to CH 4 in the anaerobic reactor totaled 27%. At the wastewater temperature used in this study, the amount of methane released in the effluent or entrapped within the sludge bed is non-negligible [44]. This is because the CH 4 solubility at a low temperature is high (104 mg/L at 15 • C). The influent average ( +/− SD) concentration of grease and oil was 102 +/− 53 mg/L. Meanwhile, its effluent concentration was always less than the method detection limit (DL = 10 mg/L). The average raw wastewater pH and total alkalinity were 8.0 +/− 0.3 and 7.0 +/− 1.2 meq/L, respectively. The effluent had a pH of 7.0 +/− 0.2 and a total alkalinity of 5.6 +/− 0.5 meq/L. Table 7 shows an additional influent and effluent characterization conducted at the end of the evaluation period with a 24 h composite sampling. It can be seen that, as expected, and because of the low-sludge-age operation, there was no nitrogen removal. The BOD effluent was less than 30 mg/L, a value that is widely recognized as a valid quality criterion for conventional activated sludge system effluents [10]. The influent settleable solid concentrations were between 2.8 and 7.0, but their effluent concentrations were less than the method DL (0.1 mL/L). The heavy metals that were present in the influent were aluminum, zinc, and copper, but their effluent concentrations were less than their DLs. The iron removal efficiency was 82%.  The influent average (+/− SD) concentration of grease and oil was 102 +/− 53 mg/L. Meanwhile, its effluent concentration was always less than the method detection limit (DL = 10 mg/L). The average raw wastewater pH and total alkalinity were 8.0 +/− 0.3 and 7.0 +/− 1.2 meq/L, respectively. The effluent had a pH of 7.0 +/− 0.2 and a total alkalinity of 5.6 +/− 0.5 meq/L. Table 7 shows an additional influent and effluent characterization conducted at the end of the evaluation period with a 24 h composite sampling. It can be seen that, as expected, and because of the low-sludge-age operation, there was no nitrogen removal. The BOD effluent was less than 30 mg/L, a value that is widely recognized as a valid quality criterion for conventional activated sludge system effluents [10]. The influent settleable solid concentrations were between 2.8 and 7.0, but their effluent concentrations were less than the method DL (0.1 mL/L). The heavy metals that were present in the influent were aluminum, zinc, and copper, but their effluent concentrations were less than their DLs. The iron removal efficiency was 82%.

Start-Up Strategies and Operation of UASB-AS Systems Treating Sewage
This investigation showed that the start-up of sewage treatment systems using UASB reactors (anaerobic technology), followed by conventional activated sludge (aerobic technology), can be implemented taking into account the operative aspects that look for the harmonization of both processes. Despite the fact that a low-temperature wastewater regime was predominant during this study (psychrophilic conditions, 18.6 +/− 1.0 • C), the UASB start-up time was 141 days (4.7 months). The aerobic reactor was also started without an inoculum in 21 days.
The results of this investigation allow us to establish recommendations for the start-up of these combined sewage treatments. Moreover, it was possible to define the operational limits and the key monitoring parameters for the successful start-up of the system, as shown in Figure 9. Because of its fast response, it is recommended that the start-up of the system must be conducted starting with the start-up of the aerobic tank. Once it is stable, the UASB start-up can be initiated because it will take more time to reach the steady-state condition.
The start-up end of the aeration tank can be identified by the following parameters' values: MLSS design concentration (3500 mg/L); dissolved oxygen (2.0 mg/L); SVI (80 mL/g); and sludge color (brown).
The start-up of the UASB reactor can be initiated by maintaining a raw wastewater bypass flow of 15% to the aeration tank in order to guarantee an appropriate F/M ratio value in the aeration tank (0.2 kg BOD5/kg MLVSS * d). This operation mode also reduces the risk of sludge and effluent deterioration caused by the presence of filamentous sludge in the aeration tank.
In the case of unfavorable sludge color changes in the aeration tank with a reduction in the F/M values and the dissolved oxygen concentration, it is recommended to increase the bypass percentage flow and the aeration time. This deterioration of the sludge can be caused by the high suspended solid concentration in the UASB effluent. Therefore, it is important to measure the UASB effluent settleable solid concentration daily. The recommended value is less than 0.3 mL/L.
It is important to monitor the UASB sludge blanket profile to evaluate the sludge blanket formation during the start-up phase. In this case, the UASB start-up end allowed for estimating the maximum mass that can be stored in the UASB digestion compartment, which was 60-ton TS. This condition can be associated with the increment in the UASB effluent settleable solid concentrations (>0.3 mL/L) and the need to take out excess sludge Because of its fast response, it is recommended that the start-up of the system must be conducted starting with the start-up of the aerobic tank. Once it is stable, the UASB start-up can be initiated because it will take more time to reach the steady-state condition.
The start-up end of the aeration tank can be identified by the following parameters' values: MLSS design concentration (3500 mg/L); dissolved oxygen (2.0 mg/L); SVI (80 mL/g); and sludge color (brown).
The start-up of the UASB reactor can be initiated by maintaining a raw wastewater bypass flow of 15% to the aeration tank in order to guarantee an appropriate F/M ratio value in the aeration tank (0.2 kg BOD 5 /kg MLVSS * d). This operation mode also reduces the risk of sludge and effluent deterioration caused by the presence of filamentous sludge in the aeration tank.
In the case of unfavorable sludge color changes in the aeration tank with a reduction in the F/M values and the dissolved oxygen concentration, it is recommended to increase the bypass percentage flow and the aeration time. This deterioration of the sludge can be caused by the high suspended solid concentration in the UASB effluent. Therefore, it is important to measure the UASB effluent settleable solid concentration daily. The recommended value is less than 0.3 mL/L.
It is important to monitor the UASB sludge blanket profile to evaluate the sludge blanket formation during the start-up phase. In this case, the UASB start-up end allowed for estimating the maximum mass that can be stored in the UASB digestion compartment, which was 60-ton TS. This condition can be associated with the increment in the UASB effluent settleable solid concentrations (>0.3 mL/L) and the need to take out excess sludge from the top part of the UASB digestion compartment (3.32 m). The UASB excess sludge in the steady-state condition was 12 TS ton/d every 15 days.
It was observed that a higher UASB effluent settleable solid concentration deteriorates the aeration tank sludge quality and reduces its dissolved oxygen concentration. In order to solve this, it is necessary to increase the aeration time. Therefore, the minimum control parameters were the daily UASB settleable solid concentration, the hourly dissolved oxygen concentration in the aeration tank, and the weekly determination of the removal efficiencies, as well as the F/M ratio and the observation of the sludge color.
Finally, this study demonstrated the importance of implementing efforts to strengthen local capacities to reach the appropriate start-up and operation conditions of sewage treatment systems in developing countries. This was evident during the training of the system operators, which was carried out considering the operational and technical aspects as well as the design criteria, control variables, and concepts that allowed for improving the decision-making process. The accompaniment to the operators during the evaluation promoted a technical knowledge appropriation that allowed the proper operation of the system.

Conclusions
Combined anaerobic and aerobic treatment is an efficient combination for sewage treatment that can reach global efficiencies of COD 87 +/− 4%, BOD 88 +/− 5%, and TSS 94 +/− 5%. The start-up of both systems was conducted without an inoculum, starting with the aerobic reactor. Despite the psychrophilic wastewater temperature, the UASB reactor start-up took 120 days.
It was demonstrated that the aerobic tank is greatly affected by the anaerobic effluent. These potential adverse effects can be overcome with the daily measurement of the anaerobic effluent settleable solid concentration (<0.3 mL/L), or, in the aeration tank, the sludge color (brown), sludge volumetric index (100-200 mL/g), and dissolved oxygen (2.0 mg/L). Moreover, it is recommended to implement a raw wastewater bypass flow of higher than 15% to the aeration tank to maintain proper values of the F/M ratio and volumetric organic loading.
The performance of the combined systems at the end of the evaluation period showed that, as expected, and because of the low-sludge-age operation, there was no nitrification. The BOD effluent was less than 30 mg/L, a value that is widely recognized as a valid quality criterion for conventional activated sludge system effluents. The influent settleable solid concentrations were between 2.8 and 7.0, but their effluent concentrations were less than the method DL (0.1 mL/L). The heavy metals that were present in the influent were aluminum, zinc, and copper, with effluent concentrations less than their DLs. The iron removal efficiency was 82%. Hydrocarbons were present in the influent, but their effluent concentrations were less than their DLs. Funding: This project was founded by Veolia-Tunja and Universidad de Boyaca. Agreement 20_2020.

Informed Consent Statement: Not applicable.
Data Availability Statement: Data sharing not applicable. No new data were created or analyzed in this study. Data sharing is not applicable to this article.