(283b) Optimal Design of Multi-Enterprise Industrial Waste-to-Energy Networks Under Uncertain Conditions

Energy-intensive industrial activities, as many chemical processes, are facing increasing pressures to decrease their dependence on non-renewable energy sources. This represents a challenge but also an opportunity to explore alternative and more reliable energy sources that were previously considered of secondary importance, such as industrial waste incineration. In fact, chemical and pharmaceutical industrial sectors generate significant amounts of high-calorific liquid waste residues that are often incinerated for simultaneous treatment and energy recovery. However, in case of energy deficits due to low energy content of residuals, additional fuels, such as natural gas, are required to sustain the combustion process. Recent studies have shown that a systematic management of waste solvent incineration in integrated chemical sites can lead to an important decrease of the site’s primary energy consumption, reducing both utility costs and environmental impact (Abaecherli et al., 2016; Wassick, 2009). Nonetheless, potential opportunities arising from a coordinated multi-site management have only been partially investigated, especially when considering their contribution to a more sustainable energy system. This is of particular interest for countries where the chemical and pharmaceutical industry consists in several smaller enterprises with a centralized treatment capacity localized in few independent incineration sites, and a resulting waste-to-energy network that can be considered as a multi-enterprise supply chain. In fact, such small businesses often have neither the resources nor the necessary waste amounts to operate own incineration facilities and have then to ship the generated waste to third companies for treatment. These are usually the aforementioned chemical sites with incineration facilities, with the addition of some cement plants that can also use some residues as substitution fuel. Hence, waste shipments play an important role concerning costs and energy consumption, in particular when the supply chain network consists of sparse enterprises.

As typical industrial waste-to-energy networks are made of a large number of enterprises, the generation of hazardous waste can vary significantly in time, both in terms of amounts and composition. This is especially true in presence of multi-purpose plants, where type and quantity of residuals completely depend on ongoing production campaigns. Such variation also affects management and operation of incineration facilities, as it is common practice to use combustible waste solvents to sustain the incineration of non-combustible residues, in order to minimize the consumption of auxiliary fuels, such as natural gas. Although former studies have already considered the simultaneous optimization of process operations and financial decisions in the case of chemical supply chains (Guillén et al., 2006), possible decisions are usually constrained to allocating and sequencing tasks to different process units. In this sense, a formulation combining management of waste incineration sites and supply chain logistics with a more detailed description of process conditions, such as upper and lower temperature bounds, has only been recently proposed (Bolis et al., 2016). Nevertheless, such methodology still lacks uncertainty tools to tackle variations in waste amounts and composition, which can critically affect both shipments and incineration processes.

This study proposes an optimization framework for the design of multi-enterprise industrial waste-to-energy networks, focusing on long-term investment strategies and synergies between multiple incineration sites and cement plants. The considered system is formulated as a mixed-integer linear problem (MILP) with discrete time representation and consists of: i) a set of incineration and cement plants with site-specific features; ii) a set of various types of hazardous liquid waste with different physicochemical properties; as well as, iii) a transportation network connecting waste producers and treatment facilities. The overall objective is to optimize the costs of the whole network by managing: i) the planning of waste shipments between enterprises; ii) the expansion of both incineration and temporary storage capacities; as well as iii) the management of incineration sites, comprising temporary storage, waste mixing and combustion processes. In order to obtain robust solutions to the problem, a combination between a scenario planning approach (Tsiakis et al., 2001) and random perturbations is implemented in the model to address parameter uncertainty. In practice, a set of scenarios is first defined for covering a broad range of plausible future megatrends concerning energy systems and the chemical industry. Whereas energy-related parameters define gas, electricity and shipment costs, together with the share of rail and road transportation, future levels for quantity and composition of the hazardous waste can be roughly estimated as a function of possible changes in the chemical and process industry, exploring highly different conditions from business-as-usual to strong focus in pharmaceutical and life-science branches. Concerning very long optimization horizons, investment are probably the most important decisions, since they affect the system for a long period, while limited adjustments are still possible for shipments in case of unforeseen events. For this reason, it is imposed that investment variables have to be equal for all the scenarios, whereas all other variables can assume different values in each scenario. In practice, this can be achieved by inserting an additional set for the scenarios and solving the problem simultaneously for all of them, with common investment decisions. Although it is in principle possible to insert weights related to the probability of each scenario to occur, in this case no clear priority could be defined because of the length of the considered horizon and the consequent high uncertain conditions. Hence, the new objective function has simply been defined as the sum of the single scenarios’ objectives. Next, random perturbations concerning waste amounts and composition are used to represent unforeseen events and the varying waste generation typical of multi-purpose plants. By comparing the results obtained under several random perturbations, it is possible to further investigated the robustness of the solution, which is not anymore limited to the given scenarios.

The methodology has been applied to the whole Swiss industrial waste-to-energy network, which comprises 5 independent incineration sites, 6 cement plants and several sparse waste-generating companies, grouped in 7 regions. A set of five different scenarios describing a broad range of plausible future conditions, divided in two consequent time horizons of 15 years, starting in years 2020 and 2035, has been investigated with the discussed uncertainty tools. Moreover, the influence of future uncertainty on the optimization results has been studied by applying three simulation approaches. First, a long-term strategy has been obtained with a single optimization over the whole horizon, including the two time points. Secondly, weights have been introduced for each time horizon, to consider the higher importance of near future in present decision-making. Finally, a stepwise procedure, solving the problem for the three horizons in series, has been applied to define a time priority for investments. The comparison of the different approaches indicates a significant difference in the timing of investments, and provides a complete overview for the design of industrial waste-to-energy networks.

References

M.L. Abaecherli, D. Santos González, E. Capón-García, K. Hungerbühler, 2016, Mathematical Optimization of Real-time Waste Incineration Scheduling in the Industry, 26th European Symposium on Computer Aided Process Engineering

V. Bolis, E. Capón-García, K. Hungerbühler, 2016, Optimal Design of Industrial Waste-to-Energy Networks, 26th European Symposium on Computer Aided Process Engineering

G. Guillén, M. Badell, A. Espuña and L. Puigjaner, 2006, Simultaneous optimization of process operations and financial decisions to enhance the integrated planning/scheduling of chemical supply chain, Computers & Chemical Engineering, 30, 421-436

P. Tsiakis, N. Shah and C. C. Pantelides, 2001, Design of Multi-echelon Supply Chain Networks under Demand Uncertainty, Industrial & Engineering Chemistry Research, 40, 3585–3604

J.M. Wassick, 2009, Enterprise-wide optimization in an integrated chemical complex, Computers & Chemical Engineering, 33, 1950-1963