TL;DR
This paper introduces a new mathematical model for designing "heat-integrated water networks" (HIWNs) that efficiently reuse water and heat within industrial processes. By combining a novel superstructure for water and heat exchanger networks, they developed a more streamlined optimization approach that can tackle complex, large-scale problems, leading to significant savings in water and energy.
What Problem Does This Paper Address?
Industrial processes consume vast amounts of water and energy, often generating wastewater and emissions that harm the environment. To combat resource scarcity and climate change, industries need to optimize their water and energy usage. Heat-integrated water networks (HIWNs) offer a solution by systematically reusing water and recovering heat within a plant. However, designing these complex networks, especially for large-scale operations with multiple contaminants and water sources, is challenging. Existing methods often struggle with complexity or require multiple optimization steps. This paper aims to simplify and improve the synthesis of HIWNs by proposing a new, more efficient modeling framework.
Methods
The core of this work is a novel superstructure that integrates a water network (WN) with a modified heat exchanger network (HEN). Unlike previous approaches that might use a stage-wise HEN superstructure, this new design simplifies the HEN by using a limited number of heat exchangers and their associated heaters/coolers. This simplification allows for more flexible water stream splitting and mixing, as well as various heat exchanger configurations (serial, parallel, bypass).
Based on this superstructure, the authors developed a mixed-integer nonlinear programming (MINLP) model. This model captures the mass and heat balances of all components within the HIWN. A key innovation is the one-step solution strategy. Instead of breaking down the problem into multiple stages (like targeting and design), this approach solves the MINLP model directly. This strategy also incorporates random initializations and iterative solving to generate multiple potential solutions, from which the best one (minimizing Total Annualized Cost - TAC) is selected.
Datasets
The paper evaluates its model using five case studies, drawing data from existing literature. These case studies vary in complexity, involving:
- Example 1: Four process water-using units, one contaminant.
- Example 2: Four process water-using units, two contaminants (modified from an isothermal problem).
- Example 3: Four process water-using units, three contaminants.
- Example 4: Eight process water-using units, one contaminant.
- Example 5: Fifteen process water-using units, one contaminant (a "pinched" problem requiring cooling utility).
The data for these examples (flow rates, temperatures, contaminant loads, costs) are taken from previously published works, allowing for direct comparison of results.
Key Insights and Ideas
- Simplified HEN Superstructure: By reducing the complexity of the heat exchanger network representation within the HIWN superstructure, the authors achieve a more manageable optimization problem without sacrificing significant heat integration opportunities.
- One-Step Solution Strategy: Moving from multi-step to a single-step optimization approach simplifies the solution process and potentially leads to more globally optimal solutions by considering all aspects simultaneously.
- Improved Scalability: The proposed model and solution strategy demonstrate effectiveness for large-scale HIWN problems, which have historically been difficult to solve.
- Reduced Number of Heat Exchangers: The optimal designs often feature fewer heat exchangers compared to traditional HEN designs, a benefit stemming from the direct heat transfer possibilities (mixing) within HIWNs.
Key Results
The proposed model and solution strategy were applied to five case studies, demonstrating its capability to solve HIWN problems of varying complexity. The results obtained are in good agreement with those reported in the literature, often achieving comparable or slightly improved outcomes in terms of freshwater consumption, utility usage, and Total Annualized Cost (TAC).
For instance:
- Example 1: Achieved minimum freshwater consumption (25 kg/s) and hot utility usage (1050 kW) with a compact HEN design.
- Example 2: Resulted in a reduced HEN investment cost (≈41% lower) and TAC (6.25% lower) compared to a previous study, despite slightly higher freshwater and hot utility consumption.
- Example 5: Showed reduced hot and cold utility consumption compared to one reference, at the expense of slightly higher heat exchanger investment costs, but still achieved a competitive TAC.
The model statistics indicate efficient computation, with average CPU time per iteration being less than 2 seconds for the largest example.
Limitations and Future Work
The authors acknowledge several areas for future work:
- Piping Design: The current model does not explicitly include piping design, which is crucial for practical implementation and can significantly impact investment costs and pumping requirements.
- Additional Model Extensions: Incorporating more advanced heat exchanger types, managing practical water flowrates, and including pumping costs are suggested.
- Wastewater Treatment: Extending the model to include wastewater treatment units would enable more comprehensive water reuse and recycling strategies, particularly for large-scale HIWN problems where this aspect is often under-addressed.
- Control and Management: Future work could also focus on the operational aspects of managing and controlling the designed networks.
Summary updated