Publication Date



Open access

Embargo Period


Degree Type


Degree Name

Doctor of Philosophy (PHD)


Civil, Architectural and Environmental Engineering (Engineering)

Date of Defense


First Committee Member

James D. Englehardt

Second Committee Member

Helena Solo-Gabriele

Third Committee Member

Gang Wang

Fourth Committee Member

Amir R. Rahmani


The US National Research Council recently recommended direct potable water reuse (DPR), or potable water reuse without environmental buffer, for consideration to address rising US water demand. In addition, conveyance of wastewater and water to and from centralized treatment plants currently consumes on average four times the energy of treatment in the US. Moreover, scaling of DPR systems involves tradeoffs beyond those of treatment facility economy-of-scale versus cost and energy of conveyance. In particular, additional factors favoring distributed DPR include minimization of energy for upgradient distribution of treated water, and retention of wastewater thermal energy. Therefore, a network modeling study addressing the optimal scale of DPR plants, considering variability in population density and topography, is presented in this dissertation. First, information on the cost of unit treatment processes potentially useful for DPR versus system capacity is reviewed, converted to constant 2012 US dollars, and synthesized. A logarithmic variant of the Williams Law cost function is proposed as applicable over orders of magnitude of system capacity, for the subject processes: activated sludge, membrane bioreactor, coagulation/flocculation, reverse osmosis, ultrafiltration, peroxone and granular activated carbon. Results are then demonstrated versus 10 DPR case studies. A generalized model of the cost of DPR water as a function of treatment plant scale, assuming futuristic, optimized conveyance networks, is then proposed for purposes of developing design principles. Fractal landscapes representing flat, hilly, and mountainous topographies were simulated, with urban, suburban, and rural housing distributions placed by modified preferential growth algorithm. Treatment plants were allocated by agglomerative hierarchical clustering, networked to buildings by minimum spanning tree. Simulation results indicate total DPR capital and operation & maintenance (O&M) costs, assuming new urban facilities with 20-year design life capable of mineralizing chemical oxygen demand to below detection limits, is competitive with current water/wastewater service costs at scales of ca. one plant per 10,000 residences. Costs for rural systems are high and dominated at most scales by the cost of capital for pipeline installation, while urban/suburban system cost is driven by a balance between pipeline installation and treatment equipment capital. The optimal scale of mineralizing DPR systems is projected to range widely in rural areas, and to range to service populations at least as small as 100 homes in suburban areas and 1000 residences in urban areas. Therefore, distributed DPR systems are recommended for consideration for municipal water and wastewater system capacity expansion projects, particularly in new construction zones. Finally, the proposed model is applied and demonstrated to evaluate the feasibility and optimal scale of DPR plants versus current plans for treatment capacity expansion in Miami-Dade County, Florida. Local data on the distribution of population and housing structures, and topography, were input, to evaluate four scenarios for the expansion service area: (a) proposed new wastewater treatment plant (WWTP) and assumed new water treatment plant treating County-projected flow; (b) central DPR treating flow expected under generalized conditions; (c) central DPR treating County-projected flow; (d) optimal distributed DPR treating expected generalized flow; and (e) new central water and wastewater systems treating expected generalized flow. Results suggest that DPR systems which mineralize organics so as to essentially eliminate discharge of endocrine-disrupting compounds to the environment may represent a practical alternative in many applications. Total cost was minimized at a scale of 46 plants for the service population of 671,823(4,810 per plant). Though DPR capital cost is projected at approximately twice that of the current plan, the total unit cost of $13.00/1000 gallons when added to O&M costs is approximately 51% higher than might be estimated for the current plan, and is less than reported for several major US cities and Florida municipalities. Overall, the model presented in this work confirms DPR as a potential water management alternative to address increasing water demand in the future, and presents an optimization approach that may be useful in planning studies. General design principals regarding the scale of DPR systems include the use of 100-10,000-home DPR systems in urban/suburban areas, and consideration of systems that return nutrients to agricultural sectors in rural areas.


Direct potable water reuse; optimization; water