Nebojša Dubočanin | Swiss Federal Institute of Technology (ETH) (original) (raw)

Reports by Nebojša Dubočanin

Research paper thumbnail of Laboratory Report - Operation and Characterization of a Laboratory Waste Water Treatment Plant

This study describes the operation of a small scaled (23 l) sequence batch reactor (SBR) fed by p... more This study describes the operation of a small scaled (23 l) sequence batch reactor
(SBR) fed by primary treated wastewater from the WWTP Kloten-Opfikon. Different sensors and measuring methods were used to monitor the reactor’s behaviour continuously and reliably. A special attention was paid on measuring nitrate and nitrite with an UV spectrometer. The experiment ran over seven weeks and was split into two phases. First, a steady state phase where the performance of the reactor during a constant operation mode was assessed. Second, a dynamic phase where experiments in terms of aeration variation, postdenitrification with ethanol addition and urea addition were conducted. The second phase was carried out in order to optimize the reactor performance. During steady state phase, nitrification is always complete whereas denitrification efficiency is quite low. Due do degradation of readily degradable substrate in the wastewater storage container, a real steady state is never reached. Results from the dynamic experiment show a negative correlation between oxygen transfer rate and TSS concentration, implying that there is a disturbing effect of TSS. Compared to low oxygen concentrations, aeration efficiency seems to be more sensitive to TSS at high oxygen concentrations. An addition of ethanol enhances the denitrification considerably. Additionally, simultaneous nitrification and denitrification is observed. In order to optimize denitrification using ethanol as carbon source, 5.36 gTOD/gNO3-N of ethanol should be added. An addition of urea had an remarkable impact on the nitrogen compounds and pH over several cycles. Results indicate an enhanced denitrification. However, no enhancing or inhibitory effect on nitrification can be deducted. An assessment of the feasibility of measuring nitrate and nitrite with an UV spectrometer shows an overestimation of the nitrate concentration. Nitrite measurements are plausible for the aerated phase.

Research paper thumbnail of Laboratory Report - Measuring Residence Time Distribution with a Simple Tracer Experiment

This experiment describes the measurement of the hydraulic residence time distribution in a lab-s... more This experiment describes the measurement of the hydraulic residence time distribution in a lab-scale system of cascaded CSTR reactors, using ammonia as tracer. Two different types of sensors, namely an ion selective electrode (ISE) sensor and a conductivity sensor, were therefore step-wisely calibrated. The inflow was set to a theoretical hydraulic residence time of 13.1 min. The ammonia solution was added into the first compartment. The obtained data show the ammonia flow-through distribution in each sector for both sensors. In general, the conductivity sensor reacts faster to an increase of concentration and therefore, has a shorter response time. The hydraulic residence time distribution measured by the two sensors are slightly differing. Both sensor measurements display a pattern similar to the modeled data obtained by a simulation with Berkeley Madonna. Whereas, the conductivity sensor’s data deviates less from the modeled data. A hydraulic residence time of 11.4 min (conductivity sensors) respectively 14.3 min (ISE sensors) was determined. Overall, the conductivity meter performed better than the ISE sensor for this experiment. However, real waste water consists of different ions, all contributing the conductivity, whereas, the ISE sensor selectively measures the ammonia concentration. As a consequence, depending on the application, ISE sensors might be the more reliable choice.

Research paper thumbnail of Urban Drainage Planning and Modelling - Hydraulic Performance

In this report the performance of the urban drainage system (UDS) of Fehraltorf is assessed by hy... more In this report the performance of the urban drainage system (UDS) of Fehraltorf is assessed by hydrodynamically simulating the waste water flows in the sewer pipes. The computation is done using the storm water management model for the catchment area over a time period of 30 years with rain data from Kloten (temporal resolution of 10 min). The rain data is verified whereas the actual model cannot be calibrated due to time restrictions. The UDS is insufficiently performing for one design storm event at eight manholes. Both the flooding and the backwater frequencies are examined at one critical area consisting of three manholes (565, 566 and 567). To reduce the current flooding frequencies (0.07-0.23 a-1) as well as the backwater frequencies (0.57-0.67 a-1) three mitigation measures are suggested: increasing the pipe diameters from 300 mm up to 500 mm or 800 mm respectively, or reducing the impervious areas from 26% down to 5%. The second strategy (800 mm pipes) shows the most promising results: no more flooding (0.00 a-1) and a backwater frequency of 0.03 a-1 for one manhole (0.00 a-1 for the others). The associated implementation costs are estimated to be approximately CHF 420’000. While the effects of climate change on the UDS are probably negligible, the population increase by 20-25% until 2040 is expected to have a significant impact. Therefore, running additional simulations considering different growth scenarios is suggested for all proposed strategies

Research paper thumbnail of Urban Drainage Planning and Modelling - The impact of Russikon on the Rohrbach during wet weather

This report quantifies the impact of the urban drainage system (UDS) of Russikon on the Rohrbach.... more This report quantifies the impact of the urban drainage system (UDS) of Russikon on the Rohrbach. By hydraulically modelling the immissions into the receiving Rohrbach, its water quality is assessed - this indirectly indicates the performance of the UDS itself. The used model is an upgraded version of the one created by Aerts et al. (2019), namely by introducing an additional natural catchment block for the Rohrbach. After the catchment is calibrated for a summer period including June and July 2018 (NSE = 0.634), the model is fed with precipitation data (1981-1996) to simulate the ecotoxic ammonia concentrations in the Rohrbach. Then, by applying a peak over threshold analysis, critical ammonia levels for two return periods are computed (5 and 7 years). Additionally, the hydraulic stress is characterised by counting the annual number of critical discharges in 2018 (risk of riverbed erosion). The compliance of both performance indicators to common water quality standards is examined. Furthermore, two potential mitigation strategies are compared to the current situation and discussed regarding their feasibility by conducting a multi-criteria decision analysis. One strategy is to increase the maximum effluent at RUB Morgental to reduce the spillovers. In the other scenario trees are planted to lower the water temperature, which then decreases the transformation from ammonium into ammonia.
While the strict 5-year threshold for ammonia (0.029 mg/l, VSA-STORM) is exceeded in all three scenarios by more than 250%, the laxer 7-year threshold (0.1 mg/l) from Lammersen (1997) is always met. There are no significant differences between the two mitigation scenarios and there is only a small improvement compared to the status quo (-0.005 mg/l). The number of critical discharges is always above the threshold (5 events, VSA-STORM) for both strategies. However, the higher maximum effluent strategy is quite close to meet the requirements (8 events). Both mitigation strategies are not fully satisfying since the higher maximum effluent is far too expensive (approximately CHF 100’000 per year) and planting trees shows only a tiny ecological benefit. Considering the ecological state of the Rohrbach - which should be improved but is not disastrous - it is recommended to evolve and examine other more effective mitigation strategies.

Research paper thumbnail of Urban Drainage Planning and Modelling - Emission of the Drainage System

The goals of this project are to assess the performance of the urban drainage system (UDS) of the... more The goals of this project are to assess the performance of the urban drainage system (UDS) of the Russikon catchment area and to suggest improved management strategies. In order to conceptually visualise and optimise this UDS, the simulation tool City Drain II is used. First, the UDS model is fed with crossverified precipitation, flow, pollutant and level sensor data. Then, it is calibrated and validated by means of the split sample technique with data from flow and level sensors over a time period of 18 months. Finally, a 30-year-simulation with four different management strategies is performed (1981 - 2010): The first strategy represents reality, that means, nothing is changed. In the second strategy the volume of the Morgental detention tank is doubled up to 560 m2. In the third strategy, maximum effluent flow at the Morgental detention tank is increased by one third to 100 ls-1. The fourth strategy is about reducing the amount of impervious area of each sub-catchment by 20%. These strategies are then evaluated and compared with four performance indicators: carry-on flow efficiency, number of overflows per year, yearly COD and NH4 emissions. Increasing the maximum effluent flow is the most promising management strategy regarding all four indicators. However, cost, hydraulic and political feasibility are outside the scope of this project and have to be considered as well.

Research paper thumbnail of Urban Drainage Planning and Modelling - Freshwater Ecology

The ecological state of the river Luppmen upstream, at and downstream of the outlet SK102 in Fehr... more The ecological state of the river Luppmen upstream, at and downstream of the outlet SK102 in Fehraltdorf is assessed on the 23 September 2019. The parameters physical appearance and aquatic plants are analysed at the location by a visual inspection of the river, the river banks and the vegetation. For the parameter benthic macroinvertebrates three representative samples are collected and analysed under the microscope in the laboratory. The physical appearance and aquatic plants (almost non-existing) are fulfilling the requirements, while for benthic macroinvertebrates the requirements are slightly not fulfilled. There are species such as leeches found which are fairly tolerant to pollution. Hence, it is recommended to conduct a more detailed analysis of the location. It is important to mention that the sampling data is strongly affected by a preceding rain event and thus might not be entirely representative. Nevertheless, according to the collected and examined data it can be derived that the river Luppmen is in a good overall ecological state with no clear trend in downstream direction.

Research paper thumbnail of Urban Drainage Planning and Modelling - Performance Assessment of the Urban Drainage Systems Fehraltorf and Russikon

The current states of the urban drainage systems (UDSs) Fehraltorf and Russikon are investigated ... more The current states of the urban drainage systems (UDSs) Fehraltorf and Russikon are investigated regarding water pollution and flood control. This examination is performed using hydrological as well as hydrodynamic models, risk assessments and freshwater ecology analysis. For the results of the current state of each catchment, viable mitigation measures are introduced, assessed and ranked using a multiple-criteria decision analysis (MCDA). The weighting of the utility scores includes ecology (30%), intergenerational equity (30%), system reliability (30%) and costs (10%). For the UDS Russikon (baseline utility: 0.44), the following mitigation measures are examined: doubling the volume of the detention tank RUB Morgental, increasing the sewer carry-on flow from RUB Morgental, reducing the impervious area and planting trees along the banks of the river Rohrbach. The MCDA favours doubling the volume of RUB Morgental (utility: 0.71) and decreasing the impervious area (utility: 0.69). For the UDS Fehraltorf (baseline utility: 0.48), the comparison is done between reducing the impervious area and increasing the pipe diameters to 500 mm or 800 mm respectively in a critical area close to a school. All three options show similar and promising utility scores between 0.74 (800 mm pipes) and 0.68 (impervious area). The differences in the utility scores between all scenarios are rather small and highly uncertain due to their sensitivity regarding the underlying value functions and the weighting. Since no urgent measures are necessary, it is recommended to perform more detailed models, assessments and cost estimations before deciding on any large investment.

Research paper thumbnail of Predictive Modelling of the Housing Prices in Melbourne

The goal of this project is to predict housing prices in Melbourne (Australia), using several sta... more The goal of this project is to predict housing prices in Melbourne (Australia), using several statistical/machine learning prediction models. The supervised type of machine learning is implemented for all models. In total, 5 statistical learning models are used, three of them being the variations of linear regression models, while the other two are decision tree-based models. The models based on the linear regression include the ordinary least squares (OLS), OLS with Ridge Regression and the LASSO model. The decision tree-based models are Decision Tree and Random Forest. The dataset used for the project is downloaded from the website www.kaggle.com. The dataset of 34'847 observations and 21 variables. First, data is cleaned and explained using principles of the Exploratory Data Analysis. The target variable from the given dataset is PRICE. The results of models based on linear regression gave solid results, with RIDGE Regression being the best one. The best performing model, among all models used for the project, is the Random Forest. The model improves results until 300 trees are reached, afterwards no significant changes have been observed. The MAPE of the Random Forest is 9.5 % and the MAPE of the RIDGE Regression was 25.5 %.

Research paper thumbnail of Differential SAR Interferometry - Detecting Object Movements and Digital Elevation Model Generation

In this report, the GPRI [GAMMA Portable Radar Interferometer] with two distinctive approaches is... more In this report, the GPRI [GAMMA Portable Radar Interferometer] with two distinctive approaches is used. The first one, the repeat-pass differential radar interferometry is used for measurements of the corner reflector displacements. The second one, the across-track differential interferometry is used to generate the DEM model. The displacements measured by the radar are close to the idealized corner reflector movements. However, since the precise field measurements of a corner reflector movements are not present, it was hard to judge the precision of the radar measurements. On the other hand, the generated DEM using across-track differential radar interferometry is quite successful. Overall the generated DEM matches the reference Swiss Alti3D DEM with a MAE [Mean Absolute Error] of 3.7m for the whole area of interest.

Research paper thumbnail of The Water Network Management Challenge

In this report, an optimal water management strategy is proposed for Vendejakelu, a fictitious ci... more In this report, an optimal water management strategy is proposed for Vendejakelu, a fictitious city with 110’000 inhabitants. The strategy is based on specific weights: costs and customer minutes lost account each for 50 percent of the customer satisfaction grade, while intergenerational equity is not considered. Indeed, the water distribution network will be decentralized in 50 years.
The optimal strategy emerging from a modeling in R is changing every decade. The simulation period is 50 years. The first decades, around 40 pipes are being replaced per year. The pipes replaced are either the ones most likely to fail or the oldest. In the last decades, less pipes are being replaced, since the time horizon considered is only 50 years. The lower investment towards the end of the simulation period reduces the costs considerably, which impacts the Customer satisfaction score positively. The combination of the two strategies results in an average satisfaction score as high as 0.89 with 1 being the highest achievable score.
Nevertheless, the uncertainties of the model are relatively high with 5% and 95% confidence intervals of 0.80 and 0.94 respectively. The real network could perform better as well as worse than the modeled network. This does not change the fact that the strategy found is probably the best strategy possible according to the weights given.

Research paper thumbnail of Inženjerska Etika

Etika je nauka o moralu; (filozofija morala), ona istražuje smisao i ciljeve moralnih vrijednosti... more Etika je nauka o moralu; (filozofija morala), ona istražuje smisao i ciljeve moralnih vrijednosti, osnovne kriterijume za moralno vrijednovanje, kao i zasnovanost i izvor morala. Etika predstavlja dio filozofije koja proučava ljudsko ponašanje koje je prihvaćeno pod određenim moralnim aspektima. Ona je normativna nauka, a norme odlučuju o specifičnom karakteru etike, na taj način se razlikuje od drugih nauka. U ovom radu čitalac je potaknut na razmišljanje o tome na koji način inženjerske odluke mogu uticati na život ljudi, korisnike inženjerskih projekata i naravno samu okolinu, prirodu, ili u najopštijem smislu stanište svih živih organizama planete Zemlje. Takođe obrađene su teme, o etici tih odluka, kako odlučiti da li je ispravno štetiti prirodu za rad dobra ljudske populacije. Čak šta više, razmotrena je etičnost odluka u situacijama gdje te odluke utiču na nas same, ili na interackije između ljudskih jedinki.

Research paper thumbnail of Capacity Upgrade of the Waste Water Treatment Plant Sihltal

Increasing growth of population and therefore increased loading of WWTP in the upcoming decades w... more Increasing growth of population and therefore increased loading of WWTP in the upcoming decades will be a major challenge, especially if space availability for an expansion of the WWTP is scarce. The WWTP Sihltal located in Zurich Leimbach, Switzerland is at its capacity limit at the moment and will be faced with an increased load of 30% by 2040. The goal of this report was to examine the feasibility of a microsieve as a pre-treatment step in terms of compliance with the effluent standards in 2040. Furthermore, the influence on the energy balance of the WWTP was assessed. A SUMO model and hand calculations were used for these purposes. After calibration and validation of the model with observed data, four scenarios were defined to perfom the simulations: Current state (2017), future state (2040) without microsieve, without microsieve but lower SRT and with microsieve. Furthermore, a sensitivity analysis was conducted by varying the COD removal efficiency of the microsieve. Both hand calculations and the SUMO model computed MLSS concentrations above the suggested upper value of 4’000g/m3 [14]. Therefore, the WWTP will be facing capacity problems in 2040 with the current plant configuration. This finding was confirmed by comparing the effluent concentrations with the legal requirements. Whereas at the moment the effluent standards for COD can be met, limits will be exceeded on 32 days in 2040. The implementation of a microsieve could reduce the exceedances to 7 days. Lowering the SRT improved the situation as well, but the legal requirements were not complied. Exceedances of TAN and NO2 were not reduced substantially by a microsieve, however effluent standards are met for TAN for all the scenarios. With increasing COD removal efficiencies of the microsieve, exceedances were reduced in terms of COD, but increased for TAN and NO2. Moreover, for high removal efficiencies COD can become limiting in terms of nitrogen incorporation. Despite improving COD effluent concentrations it is possible to improve the overall energy balance of the plant by the implementation of the microsieve. However a net energy positive WWTP won’t be achievable. At a maximum COD removal efficiency of 80 %, it was possible to lower the net energy demand by almost 75%. The results led to the conclusion that a microsieve as a pre-treatment step is a feasible solution to face increased loading in 2040 for the space-limited WWTP Sihltal. Also, its energy balance could be improved.

Research paper thumbnail of Laboratory Report - Operation and Characterization of a Laboratory Waste Water Treatment Plant

This study describes the operation of a small scaled (23 l) sequence batch reactor (SBR) fed by p... more This study describes the operation of a small scaled (23 l) sequence batch reactor
(SBR) fed by primary treated wastewater from the WWTP Kloten-Opfikon. Different sensors and measuring methods were used to monitor the reactor’s behaviour continuously and reliably. A special attention was paid on measuring nitrate and nitrite with an UV spectrometer. The experiment ran over seven weeks and was split into two phases. First, a steady state phase where the performance of the reactor during a constant operation mode was assessed. Second, a dynamic phase where experiments in terms of aeration variation, postdenitrification with ethanol addition and urea addition were conducted. The second phase was carried out in order to optimize the reactor performance. During steady state phase, nitrification is always complete whereas denitrification efficiency is quite low. Due do degradation of readily degradable substrate in the wastewater storage container, a real steady state is never reached. Results from the dynamic experiment show a negative correlation between oxygen transfer rate and TSS concentration, implying that there is a disturbing effect of TSS. Compared to low oxygen concentrations, aeration efficiency seems to be more sensitive to TSS at high oxygen concentrations. An addition of ethanol enhances the denitrification considerably. Additionally, simultaneous nitrification and denitrification is observed. In order to optimize denitrification using ethanol as carbon source, 5.36 gTOD/gNO3-N of ethanol should be added. An addition of urea had an remarkable impact on the nitrogen compounds and pH over several cycles. Results indicate an enhanced denitrification. However, no enhancing or inhibitory effect on nitrification can be deducted. An assessment of the feasibility of measuring nitrate and nitrite with an UV spectrometer shows an overestimation of the nitrate concentration. Nitrite measurements are plausible for the aerated phase.

Research paper thumbnail of Laboratory Report - Measuring Residence Time Distribution with a Simple Tracer Experiment

This experiment describes the measurement of the hydraulic residence time distribution in a lab-s... more This experiment describes the measurement of the hydraulic residence time distribution in a lab-scale system of cascaded CSTR reactors, using ammonia as tracer. Two different types of sensors, namely an ion selective electrode (ISE) sensor and a conductivity sensor, were therefore step-wisely calibrated. The inflow was set to a theoretical hydraulic residence time of 13.1 min. The ammonia solution was added into the first compartment. The obtained data show the ammonia flow-through distribution in each sector for both sensors. In general, the conductivity sensor reacts faster to an increase of concentration and therefore, has a shorter response time. The hydraulic residence time distribution measured by the two sensors are slightly differing. Both sensor measurements display a pattern similar to the modeled data obtained by a simulation with Berkeley Madonna. Whereas, the conductivity sensor’s data deviates less from the modeled data. A hydraulic residence time of 11.4 min (conductivity sensors) respectively 14.3 min (ISE sensors) was determined. Overall, the conductivity meter performed better than the ISE sensor for this experiment. However, real waste water consists of different ions, all contributing the conductivity, whereas, the ISE sensor selectively measures the ammonia concentration. As a consequence, depending on the application, ISE sensors might be the more reliable choice.

Research paper thumbnail of Urban Drainage Planning and Modelling - Hydraulic Performance

In this report the performance of the urban drainage system (UDS) of Fehraltorf is assessed by hy... more In this report the performance of the urban drainage system (UDS) of Fehraltorf is assessed by hydrodynamically simulating the waste water flows in the sewer pipes. The computation is done using the storm water management model for the catchment area over a time period of 30 years with rain data from Kloten (temporal resolution of 10 min). The rain data is verified whereas the actual model cannot be calibrated due to time restrictions. The UDS is insufficiently performing for one design storm event at eight manholes. Both the flooding and the backwater frequencies are examined at one critical area consisting of three manholes (565, 566 and 567). To reduce the current flooding frequencies (0.07-0.23 a-1) as well as the backwater frequencies (0.57-0.67 a-1) three mitigation measures are suggested: increasing the pipe diameters from 300 mm up to 500 mm or 800 mm respectively, or reducing the impervious areas from 26% down to 5%. The second strategy (800 mm pipes) shows the most promising results: no more flooding (0.00 a-1) and a backwater frequency of 0.03 a-1 for one manhole (0.00 a-1 for the others). The associated implementation costs are estimated to be approximately CHF 420’000. While the effects of climate change on the UDS are probably negligible, the population increase by 20-25% until 2040 is expected to have a significant impact. Therefore, running additional simulations considering different growth scenarios is suggested for all proposed strategies

Research paper thumbnail of Urban Drainage Planning and Modelling - The impact of Russikon on the Rohrbach during wet weather

This report quantifies the impact of the urban drainage system (UDS) of Russikon on the Rohrbach.... more This report quantifies the impact of the urban drainage system (UDS) of Russikon on the Rohrbach. By hydraulically modelling the immissions into the receiving Rohrbach, its water quality is assessed - this indirectly indicates the performance of the UDS itself. The used model is an upgraded version of the one created by Aerts et al. (2019), namely by introducing an additional natural catchment block for the Rohrbach. After the catchment is calibrated for a summer period including June and July 2018 (NSE = 0.634), the model is fed with precipitation data (1981-1996) to simulate the ecotoxic ammonia concentrations in the Rohrbach. Then, by applying a peak over threshold analysis, critical ammonia levels for two return periods are computed (5 and 7 years). Additionally, the hydraulic stress is characterised by counting the annual number of critical discharges in 2018 (risk of riverbed erosion). The compliance of both performance indicators to common water quality standards is examined. Furthermore, two potential mitigation strategies are compared to the current situation and discussed regarding their feasibility by conducting a multi-criteria decision analysis. One strategy is to increase the maximum effluent at RUB Morgental to reduce the spillovers. In the other scenario trees are planted to lower the water temperature, which then decreases the transformation from ammonium into ammonia.
While the strict 5-year threshold for ammonia (0.029 mg/l, VSA-STORM) is exceeded in all three scenarios by more than 250%, the laxer 7-year threshold (0.1 mg/l) from Lammersen (1997) is always met. There are no significant differences between the two mitigation scenarios and there is only a small improvement compared to the status quo (-0.005 mg/l). The number of critical discharges is always above the threshold (5 events, VSA-STORM) for both strategies. However, the higher maximum effluent strategy is quite close to meet the requirements (8 events). Both mitigation strategies are not fully satisfying since the higher maximum effluent is far too expensive (approximately CHF 100’000 per year) and planting trees shows only a tiny ecological benefit. Considering the ecological state of the Rohrbach - which should be improved but is not disastrous - it is recommended to evolve and examine other more effective mitigation strategies.

Research paper thumbnail of Urban Drainage Planning and Modelling - Emission of the Drainage System

The goals of this project are to assess the performance of the urban drainage system (UDS) of the... more The goals of this project are to assess the performance of the urban drainage system (UDS) of the Russikon catchment area and to suggest improved management strategies. In order to conceptually visualise and optimise this UDS, the simulation tool City Drain II is used. First, the UDS model is fed with crossverified precipitation, flow, pollutant and level sensor data. Then, it is calibrated and validated by means of the split sample technique with data from flow and level sensors over a time period of 18 months. Finally, a 30-year-simulation with four different management strategies is performed (1981 - 2010): The first strategy represents reality, that means, nothing is changed. In the second strategy the volume of the Morgental detention tank is doubled up to 560 m2. In the third strategy, maximum effluent flow at the Morgental detention tank is increased by one third to 100 ls-1. The fourth strategy is about reducing the amount of impervious area of each sub-catchment by 20%. These strategies are then evaluated and compared with four performance indicators: carry-on flow efficiency, number of overflows per year, yearly COD and NH4 emissions. Increasing the maximum effluent flow is the most promising management strategy regarding all four indicators. However, cost, hydraulic and political feasibility are outside the scope of this project and have to be considered as well.

Research paper thumbnail of Urban Drainage Planning and Modelling - Freshwater Ecology

The ecological state of the river Luppmen upstream, at and downstream of the outlet SK102 in Fehr... more The ecological state of the river Luppmen upstream, at and downstream of the outlet SK102 in Fehraltdorf is assessed on the 23 September 2019. The parameters physical appearance and aquatic plants are analysed at the location by a visual inspection of the river, the river banks and the vegetation. For the parameter benthic macroinvertebrates three representative samples are collected and analysed under the microscope in the laboratory. The physical appearance and aquatic plants (almost non-existing) are fulfilling the requirements, while for benthic macroinvertebrates the requirements are slightly not fulfilled. There are species such as leeches found which are fairly tolerant to pollution. Hence, it is recommended to conduct a more detailed analysis of the location. It is important to mention that the sampling data is strongly affected by a preceding rain event and thus might not be entirely representative. Nevertheless, according to the collected and examined data it can be derived that the river Luppmen is in a good overall ecological state with no clear trend in downstream direction.

Research paper thumbnail of Urban Drainage Planning and Modelling - Performance Assessment of the Urban Drainage Systems Fehraltorf and Russikon

The current states of the urban drainage systems (UDSs) Fehraltorf and Russikon are investigated ... more The current states of the urban drainage systems (UDSs) Fehraltorf and Russikon are investigated regarding water pollution and flood control. This examination is performed using hydrological as well as hydrodynamic models, risk assessments and freshwater ecology analysis. For the results of the current state of each catchment, viable mitigation measures are introduced, assessed and ranked using a multiple-criteria decision analysis (MCDA). The weighting of the utility scores includes ecology (30%), intergenerational equity (30%), system reliability (30%) and costs (10%). For the UDS Russikon (baseline utility: 0.44), the following mitigation measures are examined: doubling the volume of the detention tank RUB Morgental, increasing the sewer carry-on flow from RUB Morgental, reducing the impervious area and planting trees along the banks of the river Rohrbach. The MCDA favours doubling the volume of RUB Morgental (utility: 0.71) and decreasing the impervious area (utility: 0.69). For the UDS Fehraltorf (baseline utility: 0.48), the comparison is done between reducing the impervious area and increasing the pipe diameters to 500 mm or 800 mm respectively in a critical area close to a school. All three options show similar and promising utility scores between 0.74 (800 mm pipes) and 0.68 (impervious area). The differences in the utility scores between all scenarios are rather small and highly uncertain due to their sensitivity regarding the underlying value functions and the weighting. Since no urgent measures are necessary, it is recommended to perform more detailed models, assessments and cost estimations before deciding on any large investment.

Research paper thumbnail of Predictive Modelling of the Housing Prices in Melbourne

The goal of this project is to predict housing prices in Melbourne (Australia), using several sta... more The goal of this project is to predict housing prices in Melbourne (Australia), using several statistical/machine learning prediction models. The supervised type of machine learning is implemented for all models. In total, 5 statistical learning models are used, three of them being the variations of linear regression models, while the other two are decision tree-based models. The models based on the linear regression include the ordinary least squares (OLS), OLS with Ridge Regression and the LASSO model. The decision tree-based models are Decision Tree and Random Forest. The dataset used for the project is downloaded from the website www.kaggle.com. The dataset of 34'847 observations and 21 variables. First, data is cleaned and explained using principles of the Exploratory Data Analysis. The target variable from the given dataset is PRICE. The results of models based on linear regression gave solid results, with RIDGE Regression being the best one. The best performing model, among all models used for the project, is the Random Forest. The model improves results until 300 trees are reached, afterwards no significant changes have been observed. The MAPE of the Random Forest is 9.5 % and the MAPE of the RIDGE Regression was 25.5 %.

Research paper thumbnail of Differential SAR Interferometry - Detecting Object Movements and Digital Elevation Model Generation

In this report, the GPRI [GAMMA Portable Radar Interferometer] with two distinctive approaches is... more In this report, the GPRI [GAMMA Portable Radar Interferometer] with two distinctive approaches is used. The first one, the repeat-pass differential radar interferometry is used for measurements of the corner reflector displacements. The second one, the across-track differential interferometry is used to generate the DEM model. The displacements measured by the radar are close to the idealized corner reflector movements. However, since the precise field measurements of a corner reflector movements are not present, it was hard to judge the precision of the radar measurements. On the other hand, the generated DEM using across-track differential radar interferometry is quite successful. Overall the generated DEM matches the reference Swiss Alti3D DEM with a MAE [Mean Absolute Error] of 3.7m for the whole area of interest.

Research paper thumbnail of The Water Network Management Challenge

In this report, an optimal water management strategy is proposed for Vendejakelu, a fictitious ci... more In this report, an optimal water management strategy is proposed for Vendejakelu, a fictitious city with 110’000 inhabitants. The strategy is based on specific weights: costs and customer minutes lost account each for 50 percent of the customer satisfaction grade, while intergenerational equity is not considered. Indeed, the water distribution network will be decentralized in 50 years.
The optimal strategy emerging from a modeling in R is changing every decade. The simulation period is 50 years. The first decades, around 40 pipes are being replaced per year. The pipes replaced are either the ones most likely to fail or the oldest. In the last decades, less pipes are being replaced, since the time horizon considered is only 50 years. The lower investment towards the end of the simulation period reduces the costs considerably, which impacts the Customer satisfaction score positively. The combination of the two strategies results in an average satisfaction score as high as 0.89 with 1 being the highest achievable score.
Nevertheless, the uncertainties of the model are relatively high with 5% and 95% confidence intervals of 0.80 and 0.94 respectively. The real network could perform better as well as worse than the modeled network. This does not change the fact that the strategy found is probably the best strategy possible according to the weights given.

Research paper thumbnail of Inženjerska Etika

Etika je nauka o moralu; (filozofija morala), ona istražuje smisao i ciljeve moralnih vrijednosti... more Etika je nauka o moralu; (filozofija morala), ona istražuje smisao i ciljeve moralnih vrijednosti, osnovne kriterijume za moralno vrijednovanje, kao i zasnovanost i izvor morala. Etika predstavlja dio filozofije koja proučava ljudsko ponašanje koje je prihvaćeno pod određenim moralnim aspektima. Ona je normativna nauka, a norme odlučuju o specifičnom karakteru etike, na taj način se razlikuje od drugih nauka. U ovom radu čitalac je potaknut na razmišljanje o tome na koji način inženjerske odluke mogu uticati na život ljudi, korisnike inženjerskih projekata i naravno samu okolinu, prirodu, ili u najopštijem smislu stanište svih živih organizama planete Zemlje. Takođe obrađene su teme, o etici tih odluka, kako odlučiti da li je ispravno štetiti prirodu za rad dobra ljudske populacije. Čak šta više, razmotrena je etičnost odluka u situacijama gdje te odluke utiču na nas same, ili na interackije između ljudskih jedinki.

Research paper thumbnail of Capacity Upgrade of the Waste Water Treatment Plant Sihltal

Increasing growth of population and therefore increased loading of WWTP in the upcoming decades w... more Increasing growth of population and therefore increased loading of WWTP in the upcoming decades will be a major challenge, especially if space availability for an expansion of the WWTP is scarce. The WWTP Sihltal located in Zurich Leimbach, Switzerland is at its capacity limit at the moment and will be faced with an increased load of 30% by 2040. The goal of this report was to examine the feasibility of a microsieve as a pre-treatment step in terms of compliance with the effluent standards in 2040. Furthermore, the influence on the energy balance of the WWTP was assessed. A SUMO model and hand calculations were used for these purposes. After calibration and validation of the model with observed data, four scenarios were defined to perfom the simulations: Current state (2017), future state (2040) without microsieve, without microsieve but lower SRT and with microsieve. Furthermore, a sensitivity analysis was conducted by varying the COD removal efficiency of the microsieve. Both hand calculations and the SUMO model computed MLSS concentrations above the suggested upper value of 4’000g/m3 [14]. Therefore, the WWTP will be facing capacity problems in 2040 with the current plant configuration. This finding was confirmed by comparing the effluent concentrations with the legal requirements. Whereas at the moment the effluent standards for COD can be met, limits will be exceeded on 32 days in 2040. The implementation of a microsieve could reduce the exceedances to 7 days. Lowering the SRT improved the situation as well, but the legal requirements were not complied. Exceedances of TAN and NO2 were not reduced substantially by a microsieve, however effluent standards are met for TAN for all the scenarios. With increasing COD removal efficiencies of the microsieve, exceedances were reduced in terms of COD, but increased for TAN and NO2. Moreover, for high removal efficiencies COD can become limiting in terms of nitrogen incorporation. Despite improving COD effluent concentrations it is possible to improve the overall energy balance of the plant by the implementation of the microsieve. However a net energy positive WWTP won’t be achievable. At a maximum COD removal efficiency of 80 %, it was possible to lower the net energy demand by almost 75%. The results led to the conclusion that a microsieve as a pre-treatment step is a feasible solution to face increased loading in 2040 for the space-limited WWTP Sihltal. Also, its energy balance could be improved.