Acid mine drainage (AMD) results from biogeochemical oxidation of iron and sulfur minerals in flooded mine shafts and waste rock piles. In the presence of water and oxygen, sulfur rich minerals are oxidized to produce sulfate ions and cations such as ferrous, copper and aluminum ions, which lower the pH. The low pH further dissolves and releases other metals in the ore resulting in a high concentration of metal ions in the water. The solute concentration in a stream is a result of dynamic coupling between a number of chemical processes and transport processes. Creating a model encompassing all the processes occurring at an AMD affected area increases the complexity of the model multifold times. Different approaches toward modeling highly contaminated systems such as AMD sites have been employed over the past few decades. Two distinct styles of coupled reactive transport modeling have developed over time to model the contaminated systems such as AMD affected streams. This research evaluates the use of parsimonious coupled models of both types - kinetic and equilibrium to predict the fate and transport of four major ions - iron, aluminum, zinc and sulfate in the Effluent Creek at the Davis Mine Site. Model performances are accessed in terms of the ability of the model to accurately match the observed concentration in the Creek, along with the stability of the model determined by Akaike and Bayesian information criteria. In addition to the fundamental question of comparing kinetic and equilibrium model, we model processes affecting the fate and transport and model complexity. The simple kinetic models-precipitation, sorption and precipitation-sorption; outperform the equilibrium models for the three cations - iron, aluminum and zinc. For sulfate, both the kinetic and equilibrium models give comparable predictions. The simple kinetic model is however unable to define exact processes that causes the chemical transformations. The equilibrium sorption model predicts the expected chemical reactions at the given pH conditions. The chemical rates in the hyporheic zone are a couple orders of magnitude higher than the stream, except for the anion, sulfate, which shows comparable reactivity in the hyporheic zone and the stream.

As prolifically documented in the literature, wet weather events have a significant effect on the physical, chemical and biological properties of streams, thereby affecting the overall water quality of the stream as well as the receiving body of water. These effects are of particular importance when the receiving water body is a reservoir used as a source of drinking water in which a decrease in water quality could pose a significant risk to human health. Pathogens such as Cryptosporidium and Giardia are of particular importance to water supply authorities for their specific and direct threat to human health. Cryptosporidium and Giardia are known waterborne pathogens, whose effects on immunocompromised individuals can be life threatening. Because of the risk posed to human health by these pathogens, it is imperative to gain a solid understanding of the environmental factors affecting die-off and transport, and thus the ultimate concentrations in receiving waters. Testing for these pathogens is both expensive and time consuming. It is therefore also important to determine if other, more easily measured parameters can act as surrogates for these pathogens. Documentation of outbreaks resulting from exposure to these resistant pathogens is plentiful and has in part been paralleled by the study described herein. However, this research was necessary owing to the reliance of Southern New England on surface water to supply drinking water to its population and has several focus points that have yet to be evaluated in the existing literature. Specific objectives of this research were to (1) identify and better understand factors affecting Cryptosporidium and Giardia concentrations entering the studied surface water supplies and; (2) develop a sampling protocol for future use by researchers and water utilities to most accurately determine nutrient and microbial concentrations entering a water supply. Factors evaluated included event rainfall, antecedent rainfall, streamflow, land use characteristics, and the validity of indicator organisms to predict the presence of Cryptosporidium and Giardia. The study sites for this research were four small streams in Massachusetts in watersheds with varying land use characteristics. This research focused on gaining a better understanding of the effects of rainfall on in-stream parameter concentrations for streams receiving water running off of areas with distinct land-uses. The effects of rainfall were investigated through field measurements and statistical analyses. The transport of Cryptosporidium and Giardia to receiving waters, as well as understanding the factors impacting their detection, were main focus areas of this research. Study results suggest that the occurrence of Cryptosporidium and Giardia in the tributaries of small watersheds is not easily explained by basic hydrologic conditions, precipitation, land use, or the presence/absence of standard in-stream indicator organisms. Total event rainfall did not significantly affect the presence, absence or concentration of Cryptosporidium or Giardia. However, 72-hr antecedent rainfall totals greater than 20 mm were observed for all but one of the events where Cryptosporidium and/or Giardia were detected. Clostridium perfringens concentrations greater than 100 cfu/100mL was the only water quality parameter (both nutrient as well as microorganism) to provide an accurate indication for the potential presence of Cryptosporidium or Giardia. If present in the near stream area, even small amounts of precipitation when paired with wet antecedent conditions, were likely to result in transport and thus elevated in-stream concentrations of Cryptosporidium and Giardia. Concentrations were typically higher on the rising limb or near the peak of the storm hydrograph, suggesting that the nutrients and microorganisms evaluated were carried into the stream with the first flush. This flushing phenomenon was further evidenced by the clockwise hysteresis loops produced for all events that presented a single peak hydrograph. Cryptosporidium and Giardia were generally (6 out of 7 events) observed at higher concentrations at the downstream sampling location. Occurrence of Cryptosporidium and Giardia in surface waters appeared to be largely a factor of infectivity rates of wild and domesticated animals living within the watershed and their annual cycles. The animal population may have increased the probability of detecting Cryptosporidium or Giardia during the late winter and early spring. For watersheds where reservoirs are the major source of water withdrawals, sampling should be based on hydrodynamics of the reservoir and travel times from the tributaries to the intake. Attention should also be paid to antecedent rainfall conditions and time of year. This dissertation outlines the research objectives, reviews pertinent literature, describes the research design and methods as well as presents results and conclusions.

Urban sustainable development requires an in-depth and holistic understanding of the complex Food-Energy-Water-Waste (FEWW) nexus facing population growth, climate change, economic development, and pollution control. Following the United Nations (UN) 17 Sustainable Development Goals (SDGs), goal 11 aims for sustainable cities and communities to be resilient, sustainable, and safe cities to achieve social, economic, and environmental sustainability. Explicitly, SDG target 11.3 concentrates on inclusive and sustainable urbanization for sustainable planning and management. As the UN Food System program aims for inclusive, sustainable, and resilient network, promoting sustainable food systems becomes a necessity to echo social, economic, and environmental sustainability. The goal of this study is placed on analyzing sustainability patterns via different types of FEWW nexuses from building to regional scale in urban regions of Orlando and Miami, FL leading to deepen the holistic understanding of urban sustainable development. A series of system dynamic modeling analyses were conducted to create actionable options for mitigating climate change impact, improving carbon emission reduction, enhancing urban farming and green energy harvesting, and promoting carbon neutrality and decarbonization. By examining the governance structure and function of each city, this study generates actionable information by analyzing the distributed production and storage of materials and energy flows into, out of, and within a community/city given their consumption patterns and supply chains associated with various FEWW nexuses. Findings indicate the importance of sustainable technology hub integration via decentralized decision support in different types of urban FEWW nexuses. The two case studies present feasible alternatives via green building retrofit options and urban green infrastructure integration with respect to low impact developments, renewable energy harvesting, urban farming, and stormwater reuse. This information will be used to understand the role of key technologies in different types of nexuses associated with different urban planning scenarios affecting the final urban sustainable solutions.

This dissertation reports on research related to trace organic compounds (TrOCs) in surficial groundwater supplies and their subsequent removal from nanofiltration (NF) membranes. The research was conducted along coastal South Florida in cooperation with the Town of Jupiter Water Utilities, Jupiter, FL (Town). The focus of the research was to determine the extent of reclaimed water impacts on surficial groundwater supplies and subsequent effects on the Town's NF water treatment plant. Routine monitoring of fourteen TrOCs in reclaimed water and at the water treatment facility revealed varying degrees of TrOC detection in the environment. Certain TrOCs, including caffeine and DEET, were detected in a majority of the water sampling locations evaluated in this work. However, subsequent dilution with highly-treated reverse osmosis (RO) permeate from alternative supplies resulted in TrOCs below detection limits in potable water at the point-of-entry (POE). Pilot testing was employed to determine the extent of TrOC removal by NF. Prior to evaluating TrOC removal, hydraulic transients within the pilot process were first examined to determine the required length of time the pilot needed to reach steady-state. The transient response of a center-port NF membrane process was evaluated using a step-input dose of a sodium chloride solution. The pilot was configured as a two-stage, split-feed, center-exit, 7:2 pressure vessel array process, where the feed water is fed to both ends of six element pressure vessels, and permeate and concentrate streams are collected after only three membrane elements. The transient response was described as a log-logistic system with a maximum delay time of 285 seconds for an 85% water recovery and 267 gallon per minute feed flowrate. Eleven TrOC pilot unit experiments were conducted with feed concentrations ranging from 0.52 to 4,500 ?g/L. TrOC rejection was well-correlated with compound molecular volume and polarizability, with coefficient of determination (R2) values of 0.94. To enhance this correlation, an extensive literature review was conducted and independent literature sources were correlated with rejection. Literature citations reporting the removal effectiveness of an additional sixty-one TrOCs by loose NF membranes (a total of 95 data points) were found to be well-correlated with molecular volume and polarizability, with R2 values of 0.72 and 0.71, respectively. Of the TrOC's detected during this research, the anthropogenic solute caffeine was selected to be modeled using the homogeneous solution diffusion model (HSDM) and the HSDM with film theory (HSDM-FT). Mass transfer coefficients, K_w (water) K_s (caffeine), and k_b (caffeine back-transport) were determined experimentally, and K_s was also determined using the Sherwood correlation method. Findings indicate that caffeine transport through the NF pilot could be explained using experimentally determined K_s values without incorporating film theory, since the HSDM resulted in a better correlation between predicted and actual caffeine permeate concentrations compared to the HSDM-FT and the HSDM using K_s obtained using Sherwood applications. Predicted versus actual caffeine content was linearly compared, revealing R2 values on the order of 0.99, 0.96, and 0.99 for the HSDM without FT, HSDM-FT, and HSDM using a K_s value obtained using the Sherwood correlation method. However, the use of the HSDM-FT and the Sherwood number resulted in the over-prediction of caffeine concentrations in permeate streams by 27 percent and 104 percent, respectively.

The enhanced biological phosphorus removal (EBPR) process efficiency relies on different operational and process conditions especially the type of carbon source available in the wastewater. Acetic acid and propionic acid are the two major volatile fatty acids (VFAs) found in domestic wastewater which can drive biological phosphorus (P) removal to the desired level. However, often domestic wastewater does not have a sufficient amount of VFAs. Due to high acetate and propionate production-cost, it is not economic to add acetate and propionate directly in full-scale wastewater treatment plants. This brought up the idea of using external carbon sources (e. g. molasses has been used successfully) in EBPR systems that can be converted to VFAs through a fermentation process. On the other hand, biodiesel fuels have been produced increasingly over the last decade. Crude glycerol is a biodiesel production major by-product that can be used as an external carbon source in wastewater treatment plant. Therefore, the main objective of this research is to optimize the glycerol/biodiesel waste fermentation process' operational conditions in pursuit of producing more favorable fermentation end-products (i. e. a mixture of acetic acid and propionic acid) by adding glycerol to a prefermenter versus direct addition to the anaerobic zone or fermentation with waste activated sludge. For this reason, different prefermenter parameters namely: mixing intensity, pH, temperature and solids retention time (SRT), were studied in a small scale fermentation media (serum bottles) and bench scale semi-continuous batch prefermenters. Experimental results revealed that glycerol/biodiesel waste fermentation resulted in a significant amount of VFAs production with propionic acid as the superior end-product followed by acetic acid and butyric acid. The VFA production was at its highest level when the initial pH was adjusted to 7 and 8.5. However, the optimum pH with respect to propionic acid production was 7. Increasing the temperature in serum bottles favored the total VFA production, specifically in the form of propionic acid. Regarding the mixing energy inconsistent results were obtained in the serum bottles compared to the bench scale prefermenters. The VFA production in mixed serum bottles at 200 rpm was higher than that of un-mixed ones. On the other hand, the unmixed or slowly mixed bench scale prefermenters showed higher VFA production than the mixed reactors. However, the serum bottles did not operate long enough to account for biomass acclimation and other long-term effects that the prefermenter experiments could account for. As a consequence one of the most important and consistently results was that VFA production was significantly enhanced by reducing mixing intensity from 100 rpm to 7 rpm and even ceasing mixing all together. This was true both for primary solids and glycerol. In addition propionate content was high under both high and low intensity, and adding glycerol also increased the fraction of primary solids that formed propionic acid instead of acetic acid. Increasing the SRT from 2 to 4 days increased the VFA production about 12% on average. In order to investigate the effect of glycerol on EBPR process efficiency two identical A2/O systems were monitored for 3 months. Experimental results suggested that glycerol addition could increase the P removal efficiency significantly. Adding glycerol to the prefermenter rather than the anaerobic zone resulted in a lower effluent soluble ortho phosphorus (SOP) (0.4 mg-P/L vs. 0.6 mg-P/L) but the difference was apparently statistically significant. Future experimentation should be done to determine if this effect is consistent, especially in carbon poor wastewaters. Also it would be desirable to conduct a longer pilot study or a full scale study to determine if this improvement in effluent SOP remains true over a range of temperature and changing influent conditions.

Trichloroethene (TCE) is a chlorinated volatile organic compound (CVOC) that can be found in industrial and household products. It is typically used as a solvent or degreaser. TCE can have detrimental health impacts and is known to be carcinogenic to humans. Federal and state regulatory drivers determine the need to assess and remediate soil and groundwater contaminated with CVOCs. There are many different methods for remediation; however, bioremediation has the ability to breakdown TCE all the way to harmless gasses (ethene and ethane). Bioremediation requires dechlorinating microbes (indigenous or augmented), electron donor (food source), and an electron acceptor (CVOCs). Electron donors are typically injected into the target area and are distributed naturally throughout the subsurface. A partitioning electron donor (PED) has the ability to partition from the dissolved phase into low permeability zones and/or dense non-aqueous phase liquids (DNAPLs) (i.e. source zones), and then be slowly released and readily metabolized at the DNAPL:water interface. This thesis summarizes the first field scale PED implementation with the main research objective of evaluating whether utilizing a PED for bioremediation of a TCE source zone is achievable. Based on laboratory studies, n-butyl acetate (nBA) was selected as the PED for application in a TCE source area, selected at Cape Canaveral Air Force Station's Launch Complex 34, identified as Hot Spot 1. Hot Spot 1 has a zone of high concentration TCE in a low permeability clay layer at a depth of approximately 40 feet below land surface (ft BLS). Implementation included the recirculation of groundwater above and below the clay layer without PED injection for comparative analysis (baseline flux), then with PED injection in, above, and below the clay layer (system operation phase). The groundwater was recirculated using a solar powered recirculation system, which consisted of a pair of extraction wells in the center of the treatment area, screened above and below the low permeability layer, and a set of five peripheral injection well pairs, similarly screened, used to create an inward hydraulic gradient and promote horizontal flow across the top and base of the clay layer. Groundwater concentrations in the treatment area were monitored using three monitoring well clusters (each with six depth intervals ranging from 23 to 61 ft BLS) and existing monitoring wells in the treatment area. The groundwater recirculation system was operated, without addition of PED, for approximately four weeks to establish the baseline flux condition. PED was then introduced to the subsurface by injecting 34,000 gallons of a solution containing nBA (3,000 mg/L) and conservative tracers (bromide and/or iodide) using direct push technology (DPT) at 20 locations from approximately 23 to 62 ft BLS. Confirmation sampling (DPT groundwater and monitoring well sampling) was conducted to assess the PED distribution after injection activities. The recirculation system remained off after PED injection for approximately four weeks to allow the PED to partition into the DNAPL and to facilitate the acclimation and establishment of biomass within the treatment area. The recirculation system was then restarted and operated for approximately one year. Groundwater sampling was performed regularly to assess mass flux and microbial reductive dechlorination. PED amendment was successfully injected above, in, and below the low permeability layer, as evidenced by positive detections of nBA from soil and groundwater sampling within the treatment area immediately following the injection event. The implementation was also successful in reducing contaminant mass from both soil and groundwater. CVOC mass removed during the baseline flux phase (pre-PED injection; 14 March 2011 to 18 April 2011) was calculated based on groundwater sampling data and totaled 14 pounds (lbs). All of the mass removed during the baseline flux phase was from the high permeability layer, indicating that mass removed was dissolved phase mass above and below the clay layer. Mass removal was likely a result of extraction and dilution from operation of the recirculation system. The mass removal rate during the baseline flux phase was approximately 0.40 pounds per day (lbs/day). CVOC mass removed during the system operation phase (post-PED injection; 9 August 2011 to 11 September 2012) was calculated based on groundwater and soil sampling data and totaled 110 lbs. Of the 110 lbs removed, 78 lbs of CVOC mass was removed from the high permeability layer and 32 lbs was removed from the low permeability layer, indicating that not only dissolved phase mass in the high permeability layer was removed, but source zone material sorbed into the low permeability layer was removed as well. Mass removed from the low permeability layer was likely a result degredation (ie. reductive dechlorination) at and around the DNAPL:water interface. The mass removal rate during the system operation phase was approximately 0.28 lbs/day. The higher rate of removal during the baseline flux phase is likely due to the initial removal of a significant amount of dissolved phase CVOCs and not the mass sorbed into the low permeability layer. In general, TCE and cis-1,2-dichloroethene (cDCE) concentrations decreased during the baseline flux phase with no increase in vinyl chloride (VC) concentration, indicating removal via extraction and dilution and not reductive dechlorination. Following the PED injection, TCE and cDCE concentrations generally decreased with increases observed in VC concentrations, indicative of reductive dechlorination. Ethene concentration was monitored to assess complete dechlorination from TCE to ethene. Average ethene concentration detected in samples collected from treatment zone monitoring wells increased from 52.8 micrograms per liter (?g/L) (pre-injection; April 2011) to 408 ?g/L (September 2012), indicating complete dechlorination of CVOCs was occurring. In addition, dechlorinating microbial biomass increased significantly, as evidenced by increases in average Dhc (dechlorinating microbial culture) and vcrA (specific gene of culture responsible for breaking down VC through to ethene) concentrations detected in samples collected from treatment zone monitoring wells; Dhc increased from 8.5x106 gene copies/L (pre-injection; April 2011) to 5.0x107 gene copies/L (September 2012) and vcrA increased from 5.0x103 gene copies/L (April 2011) to 6.8x107 gene copies/L (September 2012). TOC concentration was shown to generally increase following the injection activities, then decrease through the system operation period, indicating the electron donor was successfully injected into the subsurface, and was being utilized by the indigenous dechlorinating microbial population. Remaining TOC at the site was minimal, with an average TOC concentration of 21 mg/L (September 2012) detected in samples collected from treatment zone monitoring wells, decreasing from 250 mg/L (August 2011) just following injection. If reductive dechlorination were to continue to occur, more electron donor would be needed. The reduction of CVOC concentrations at the site are likely due to reductive dechlorination as a result of the PED amendment injection, as evidenced by: (i) the production of daughter products relative to the degradation of TCE; (ii) the production of ethene; (iii) the production of dechlorinating microbial mass; and (iv) the reduction of electron donor. Although effective, nBA was utilized and depleted quicker than an industry electron donor would be expected to last, depleting within 12 months, as opposed to two to three years. Based on this alone, it appears that nBA would not be a good candidate for full scale implementation at this or other sites; however, to provide a true comparative analysis, side-by-side test plots would be recommended at the site, one utilizing nBA and one utilizing a standard substrate. This would ensure both electron donor options are being subjected to the same geophysical and geochemical settings and the same or similar contaminant concentrations.

This study evaluated the short-term effects of glycerol addition on readily biodegradable (RB) chemical oxygen demand (COD) in a carbon limited wastewater influent. The presence of an RB fraction provides with a suitable substrate for microorganisms to produce volatile fatty acids (VFA). The oxygen utilization rate (OUR) has been used to evaluate the oxygen consumption for RB substrate in wastewater. Wastewater with low organic content contains limited RB substrate, and thus, additional carbon source is required to improve biological treatment capability. Acetate, propionate, methanol, and glycerol are the commonly available carbon sources for biological treatment process. However, the cost of acetate and propionate are relatively high, and it is not economical to use these carbon sources in the wastewater plant. The use of methanol as a carbon source inherently poses safety issues in field applications due to its toxic and flammable properties. On the other hand, crude glycerol is the byproduct of biodiesel, which is an excellent carbon source alternative. However, crude glycerol contains impurities and requires a certain degree of purification to enhance the performance. The samples for the study were collected from the Iron Bridge Wastewater Reclamation Facility (Oviedo, FL) designed for treating municipal wastewater. The total COD (TCOD) of the sample influent was in the range of 237 to 408 mg COD/L, and RBCOD value was between 38 and 80.5 mg COD/L, containing up to 10 mg COD/L of VFA. This study also demonstrates the relationship between the glycerol concentration and OURs during the diauxic growth phase from the addition of glycerol. The growth was due to the existence of RB substrate and availability of glycerol for the microorganisms. TCOD increased from 284 to 378 mg COD/L and from 284 mg COD/L to 323 mg COD/L by spiking approximately 30 and 15 mL of glycerol stock solution (6.67 g/L), respectively. RBCOD increased from 45 to 89 mg COD/L and 55 mg COD/ L by spiking 30 mL and 15 ml glycerol stock solution, respectively. The initial influent heterotrophic active biomass (ZBH) increased from 5.4 to 15.8 mg VSS/L (8 to 23.4 mg COD/L) due to the addition of glycerol, indicating that the glycerol may be an adequate carbon source. The COD of wastewater with limited VFA (e.g., 10 mg COD/L) increased up to 2,502 mg COD/L where propionic acid (2,468 mg COD/L) exists as the primary end product with a small quantity of acetic acid (34 mg COD/L). Propionic acid was the main VFA component fermented from the glycerol addition. Glycerol addition led to increased RBCOD accompanied by high VFA production. This research investigated the short-term effect of glycerol addition on existing RBCOD in wastewater. It is recommended to explore the effect of increased RBCOD by the addition of glycerol to the effluent N and P for future study.

Bench-scale water treatment testing of three next generation hollow-fiber (HF) nanofiltration (NF) membranes was conducted to characterize divalent ion rejection capabilities and investigate removal mechanisms. Existing mathematical models were investigated to describe solute transport using synthetic magnesium sulfate solutions including the size exclusion model, homogenous solution diffusion (HSD) model, dimensional analysis, and the HSD model incorporating film theory. Solute transport for two of the membranes were described by HSD theory and were predictive of their 90% divalent ion removal. A third membrane was more accurately modeled using size exclusion and was found to be predictive of its 40% divalent ion rejection. Feed ionic strength variation was shown to significantly impact rejection. In this work, semi-empirical models were developed to describe solute transport under varying feed ionic strength conditions. Bench-scale testing of aerated groundwater confirmed the HFNF membrane divalent ion rejection capabilities. Pilot testing of a commercially available HFNF membrane was shown to remove divalent ions and dissolved organic carbon (DOC) by 10% and 25%, respectively. Financial evaluations indicated that HFNF offered cost savings over traditional spiral-wound (SW) NF, $0.60/kgal versus $0.85/kgal operating costs, respectively. Traditional SWNF membranes produced superior water quality achieving 90% divalent ion removal and 96% DOC removal but required media and membrane filtration pretreatment. When considering the costs of constructing a new 2 million gallon per day (permeate) HFNF process, conceptual cost comparisons revealed that HFNF technologies could reduce capital costs by approximately $1 million, and operating costs by $0.27/kgal for an 85% recovery plant.

This dissertation reports on research related to groundwater and contaminant transport to the Volusia Blue Spring (VBS), an Outstanding Florida Water Body located in Volusia County (Florida). The integration of springshed water quality and contaminant fate and transport (CFT) modeling played key roles in the evaluation of anthropogenic recharge impacts on VBS. To study anthropogenic recharge into the karst limestone aquifer, wastewater effluent, golf course ponds, septic tanks, groundwater monitoring wells, and VBS discharge were sampled for boron, nitrate-nitrogen, nitrate-oxygen and their isotopes spatially throughout the VBS springshed. Data related to natural water features, rainfall, land use, water use, treated wastewater discharge, and septic tank effluent flows was used as inputs to the three-dimensional CFT model developed from an integration of MODFLOW-2000 and MT3DMS. The model was calibrated and validated from field observed water levels and water quality taken throughout the springshed. The purpose of this model is to understand groundwater and spring water quality throughout the VBS springshed. Water quality and model results indicate that water from the surficial aquifer in surrounding urban areas contributed to the flow and water quality at the spring's boil. Protection scenarios that included wetland treatment systems and the conversion of targeted septic systems to sewer were simulated to estimate future reductions of anthropogenic nutrients transported to the Spring. Of the scenarios evaluated in this study, targeted septic system removal results in the greatest benefit with a 36% nitrate decrease in a forty-year projection of spring discharge water quality. Results from this combined water quality and model development approach is expected to contribute an understanding of anthropogenic impacts from the urbanized developments overlying and surrounding the karst VBS aquifer.

Enhanced biological phosphorus removal (EBPR) as well as biological nitrogen removal require a carbon source to be carried out. Volatile fatty acid (VFAs) (mainly acetic and propionic acids) are the major driving force for EBPR. Many domestic wastewaters have an insufficient amount of VFAs. However, carbon sources such as acetic and propionic acids can be produced using primary solids fermentation process. Due to the cost of VFA production, an external carbon source can be added to the biological nutrient removal (BNR) system that can be fermented to provide the desired VFAs. Glycerol (biodiesel by-product) offers a solution to reduce carbon addition cost if can be fermented to acetic and propionic acid or can be used directly as an external carbon substrate for EBPR and denitrification. Using glycerol in wastewater treatment can also offset the biodiesel plant disposal cost and reduce the BNR chemical cost. The main objective of this study was to optimize the prefermentation process and optimize the BNR system using glycerol as an external carbon source. In this work, Optimization of the prefermentation process using glycerol, mixing, and hydrogen gas addition was evaluated. EBPR performance within an A2O-BNR system was evaluated using either a direct glycerol method to the anaerobic zone or by co-fermentation with primary solids. Also, optimization of the nitrogen removal (specifically denitrification) efficiency of a 5-stage BardenphoTM BNR system using either a direct glycerol method to the second anoxic zone or by co-fermentation with primary solids was evaluated. It was found in this study that glycerol was an efficient external carbon substrate for EBPR as well as biological nitrogen removal. The prefermentation experiment showed that glycerol co-fermentation with primary solids produced significantly higher (p< 0.05) VFAs than primary solids fermentation alone, even more than the possible value from the added glycerol (427 mg-COD/L). The increased VFAs imply that the glycerol addition stimulated additional fermentation of primary solids. Lowering the prefermenter mixing energy (50 to 7 rpm) resulted in a significant increase in VFAs production (80%). Also, purging the headspace of the prefermenter with hydrogen gas did not lead to more VFAs, but significantly (p< 0.05) increased the propionic acid to acetic acid ratio by 41%. In the A2O-BNR pilot plant experiment, it was found that glycerol is a suitable renewable external substrate to drive enhanced EBPR as well as denitrification. The results from both locations of glycerol addition (direct vs. fermented) were beneficial to the BNR system. Both systems had similar effluent quality and achieved total nitrogen (TN) and total phosphorus (TP) removals up to 86% and 92% respectively. The 5-stage BardenphoTM BNR experiment investigated the location of glycerol addition (direct vs. fermented) on the performance of denitrification in the second anoxic zone and the overall performance. The results from both systems were that glycerol was beneficial to the BNR system and had virtually similar effluent quality. Both systems achieve complete denitrification and excellent removal of TN and TP up to 95% and 89% respectively. Also, the pilot that received fermented glycerol had significantly higher VFAs loading and lower observed yield. The side-stream prefermenter effluent flowing to the second anoxic reactor did not cause high effluent ammonia (NH3) concentration. In summary, the location at which glycerol was added did not affect effluent quality for nitrogen and phosphorus. However, glycerol addition and mixing energy did impact prefermenter performance and effluent quality.