(137d) Optimization-Based Methodology for the Development of Wastewater Facilities for Energy and Nutrient Recovery

A paradigm shift is currently underway from an attitude that considers wastewater streams as a waste to be treated, to a proactive interest in recovering materials and energy from these streams [Guest et al., 2009]. Until recently a majority of the activities related to resource recovery from wastewater have focused on waste sludge streams, which are a by-product of biological treatment. Because these streams have relatively low flows in comparison to the main wastewater streams, and are more concentrated, resources can be recovered from them with minimal changes to the wastewater treatment infrastructure. In contrast, the main focus of this presentation is on direct resource recovery from municipal and industrial wastewater (although the proposed methodology can also be extended to encompass wastewater sludge as well).

A diverse toolkit is available and is increasingly being applied by practitioners to promote sustainability of wastewater facilities. The increasing market value of wastewater components, such as ammonia and phosphorus, are acting as key drivers for resource recovery from wastewater. Organic carbon too can be recovered from wastewater using anaerobic digestion to produce a methane-rich biogas that can be combusted on-site for heat or electricity generation, or cleaned-up and sold. Other examples of newer technologies for carbon recovery include microbial fuel cells as well as the production of bioplastics.

Although there seems to be a general consensus that wastewater (and wastewater sludge) is a potential source of valuable resources, and that the technology needed for such resource recovery is maturing, it is the lack of decision-making tools and design methodologies that is the primary problem to identifying the most sustainable solutions in a given geographic and cultural context [Hamouda et al., 2009]. For the synthesis of sustainable resource recovery facilities, one must account for the trade-offs between capital and operating costs and sales on the one hand, and water quality and other environmental and social considerations on the other hand. But as the array of technical options grows, a simple enumeration of all possible alternatives quickly becomes unmanageable, let alone the fact that each technology has its own parameters to specify or optimize. Technical insights are useful to reduce the combinatorial problem and often allow us to arrive at promising solutions as in [Sutton et al., 2001], but they do not always provide all the information that is required for an optimal (or near optimal) system.

In this presentation, we advocate the use of systems engineering methods and tools to address this problem in a systematic way. A superstructure modelling and optimization approach [Biegler et al., 1997; Kokossis and Yang, 2010] is considered, whereby the objective is to determine an optimal resource recovery facility in terms of (i) its units, (ii) the piping interconnections between the units, and (iii) the flow rates and compositions in the interconnections. The optimal system configuration is one that maximizes a certain sustainability or economical index of the facility, for instance a weighted sum of LCA impacts or the net present value (NPV). Regarding constraints, material balances on flows and concentrations around the sources, the units, and the sinks are to be obeyed in addition to the discharge limits and certain design and structural specifications. This formulation leads to nonconvex MINLP problems for which state-of-the-art global optimization software such as BARON [Tawarmalani and Sahinidis, 2004] can be used.

Key to the success of this methodology is the development/selection of mathematical models for the units that are simple enough, yet provide reliable estimates of their performance and associated costs. The simplest approach considers fixed conversion, removal or split ratios in the treatment and separation units, as is typically assumed in water network synthesis problems [Ahmetovic and Grossmann, 2011; Khor et al., in press]. In this work, we advocate an alternative, potentially more accurate, approach that relies on detailed first principle models, as implemented in wastewater treatment simulators such as GPS-X (www.hydromantis.com/GPS-X.html). Following a surrogate-modeling approach, simple linear or piecewise-linear models can be fitted to the predicted performance from a process simulator at a large number of influent flow rates and compositions. Likewise, reliable capital and operating costs can be obtained from preliminary costing software, such as CapdetWorks (www.hydromantis.com/CapdetWorks.html).

To illustrate this methodology, we showcase the synthesis of a resource recovery facility for the treatment of 100 m3/h of an industrial wine distillery effluent. A small superstructure that consists of 2 biological treatment units (UASB, SMBR), 2 filtration units (sand filter, membrane unit), and 2 nutrient recovery units (struvite crystallizer, zeolite adsorber) is investigated. Besides the treatment/separation units, an auxiliary piece of equipment is the generator producing electricity from biogas. The objective is to maximize the NPV accounting for electricity, struvite and ammonia sales, while satisfying maximum discharge requirements as defined by the EU Directive 91/271/EEC on Urban Wastewater Treatment. Several scenarios are discussed, which demonstrate that significant improvements in the NPV can be achieved through resource recovery.

By and large, the proposed methodology should be seen as a decision support system for isolating, among hundreds or even thousands of alternatives, those promising resource recovery systems whose development is worth pursuing. Based on this preselection, further simulation and optimization studies can then be undertaken to refine the performance and cost prediction by taking into account detailed design and operation considerations, as well as process integration. Such decomposition is indeed warranted as current computational capabilities and available algorithms do not allow for the optimal design and operation to be solved in a single step due to complex unit configuration, multiple scales, time dependence, and uncertainty. Besides providing a rational tool for comparing alternatives, this approach can also be useful for resource allocation, for instance in determining which technologies are critical and deciding where further research and development effort should focus.

References:

- E. Ahmetovic and I.E. Grossmann (2011). Global superstructure optimization for the design of integrated process water networks. AIChE journal, 57(2):434-457.

- L.T. Bielger, I.E. Grossmann, and A.W. Westerberg (1997). Systematic Methods of Chemical Process Design. Prentice Hall.

- J.S. Guest, S.J. Skerlos, J.L. Barnard, M.B. Beck, G.T. Daigger, H. Hilger, S.J. Jackson, K. Karvazy, L. Kelly, L. MacPherson, J.R. Mihelcic, A. Pramanik, L. Raskin, M.C. Van Loosdrecht, D. Yeh, and N.G. Love (2009). A new planning and design paradigm to achieve sustainable resource recovery from wastewater. Environmental Science & Technology, 43(16):6126-6130.

- M.A. Hamouda, W.B. Anderson, and P.M. Huck (2009). Decision support systems in water and wastewater treatment process selection and design: A review. Water Science & Technology, 60(7):1757-1770.

- C.S. Khor, B. Chachuat, and N. Shah (in press). A superstructure optimization approach for water network synthesis with membrane separation-based regenerators. Computers & Chemical Engineering.

- A.C. Kokossis and A. Yang (2010). On the use of systems technologies and a systematic approach for the synthesis and the design of future biorefineries. Computers & Chemical Engineering, 34(9):1397-1405.

- P.M. Sutton, H. Melcer, O.J. Schraa, and A.P. Togna (2011). Treating municipal wastewater with the goal of resource recovery. Water Science & Technology, 63(1):25--61.

- M. Tawarmalani and N.V. Sahinidis (2004). Global optimization of mixed-integer nonlinear programs: A theoretical and practical study. Mathematical Programming, 99(3):563-591.