GOALI: Multi-scale Optimization for the Design, Capacity Planning and Operation of Power Intensive Process Networks under Uncertain Electricity Prices and Market Demands

GOALI：电价和市场需求不确定下电力密集型过程网络的设计、容量规划和运营的多尺度优化

• 批准号: 1159443
• 负责人: Ignacio Grossmann
• 金额: $ 30.2万
• 依托单位: Carnegie-Mellon University
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 2012
• 资助国家: 美国
• 起止时间: 2012-05-01 至 2016-04-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1159443&HistoricalAwards=false
• 关键词: GOALI; Multi; scale; Optimization; Design

1159443GrossmannSummary: Power intensive industries such as air separation, cement and chlor-alkali manufacturing will face increased electricity prices in the future due to environmental pressures to reduce CO2 emissions. Furthermore, another challenge is that since electricity markets became deregulated in the 1990s, electricity prices have been subject to hourly as well as seasonal variations. These are also likely to become more acute as renewable sources of energy like wind and solar are introduced for power generation. These trends have led to a considerable amount of uncertainty and variability in the daily operating expenses of power intensive industries, which in turn affect their competitiveness and long term planning. The aim of this GOALI proposal, which will be performed in collaboration with researchers from Praxair, is to develop a multi-scale modeling framework that can be used as a decision-making support tool to optimize designs so as to introduce flexibility in plant operations to effectively address uncertain hourly variations in electricity prices and uncertain product demands. In order to tackle the problem of uncertain electricity prices and product demands, we consider as a first step the development of an optimization methodology for the deterministic case when the electricity prices and demands are assumed to be known (e.g. in terms of forecasts). We propose a short-term mixed-integer operational optimization model that is based on offline computations or measured plant data, and that is integrated with the design and long-term capacity planning problem, which involves decisions on installing or upgrading equipment, or increasing storage capacity. To solve the resulting multi-scale mixed-integer linear program (MILP) model for a single plant, we plan to develop a tailored bi-level decomposition algorithm. Also, to consider the case of process networks consisting of several plants, we intend to investigate the solution of the large-scale model with a Lagrangean decomposition scheme based on a novel hybrid cutting plane and subgradient method to accelerate convergence. As a second major step, we will address the treatment of uncertainties of product demands and electricity prices for which we intend to investigate a novel hybrid stochastic programming/robust optimization approach. The basic idea is to model the long-term design decisions and uncertain demands with multistage stochastic programming, and the short term operating decisions and uncertain prices through robust optimization. The proposed models will be tested with process models and real world data of air separation plants supplied by Praxair.Intellectual Merit: The major intellectual challenges in this project lie in the multi-scale integration of the short-term operational model with the design and long-term capacity planning problem, the treatment of uncertainties in product demands and electricity prices, and the development of effective computational algorithms for solving large-scale optimization models. In order to address these challenges, we propose a strategy for multi-scale integration that effectively incorporates the operational model with the design and capacity planning model. Furthermore, we propose decomposition schemes that have the potential of effectively tackling large-scale deterministic models for realistic process networks of power intensive plants, particularly for air separation plants. Finally, we propose a potentially promising hybrid stochastic programming/ robust optimization model and solution method in order to anticipate the effect of uncertainties in the product demands and electricity prices.Broader Impact: The proposed GOALI project has the potential of yielding significant economic savings in the enterprise wide optimization of power intensive industries to make them more competitive. The proposed GOALI project will involve summer internships for the Ph.D. student and visits by the PI to Praxair. The project also has the potential of collaboration and dissemination of the basic methodologies to several petroleum, chemical and engineering/software companies of the Center for Advanced Process Decision-making (CAPD) at Carnegie Mellon. We also intend to collaborate with the Electric Energy Systems Group at Carnegie Mellon. We plan to involve undergraduates for documenting case studies that will be made available through the internet in our cybersite on MINLP. Finally, we also plan to participate in the University outreach program at Carnegie Mellon where we intend to expose high school students to technology on air separation and major issues in operation under fluctuating electricity prices by using simple day-to-day examples like deciding what appliances to turn on and off at their houses if the utilities charged electricity prices that changed on an hourly basis.

1159443Grossmann总结：由于减少二氧化碳排放的环境压力，空气分离、水泥和氯碱制造等电力密集型行业未来将面临电价上涨。 此外，另一个挑战是，自 20 世纪 90 年代电力市场放松管制以来，电价一直受到每小时和季节性变化的影响。随着风能和太阳能等可再生能源被引入发电，这些问题也可能变得更加严重。这些趋势导致电力密集型行业的日常运营费用存在相当大的不确定性和可变性，进而影响其竞争力和长期规划。该 GOALI 提案将与普莱克斯的研究人员合作执行，其目的是开发一个多尺度建模框架，可用作优化设计的决策支持工具，从而为工厂运营带来灵活性有效解决不确定的电价每小时变化和不确定的产品需求。为了解决电价和产品需求不确定的问题，我们认为第一步是在电价和需求已知的情况下（例如在预测方面）开发确定性情况的优化方法。我们提出了一种短期混合整数运营优化模型，该模型基于离线计算或测量的工厂数据，并与设计和长期容量规划问题相结合，其中涉及安装或升级设备或增加存储的决策容量。为了解决单个对象的多尺度混合整数线性规划（MILP）模型，我们计划开发一种定制的双层分解算法。此外，考虑到由多个工厂组成的过程网络的情况，我们打算研究采用拉格朗日分解方案的大型模型的解决方案，该方案基于一种新颖的混合切割平面和次梯度方法来加速收敛。作为第二个主要步骤，我们将解决产品需求和电价不确定性的处理问题，为此我们打算研究一种新颖的混合随机规划/鲁棒优化方法。其基本思想是通过多级随机规划对长期设计决策和不确定需求进行建模，并通过鲁棒优化对短期运营决策和不确定价格进行建模。所提出的模型将使用普莱克斯提供的空气分离设备的过程模型和真实世界数据进行测试。 智力优点：该项目的主要智力挑战在于短期运行模型与设计和长期运行模型的多尺度集成。 - 长期容量规划问题，产品需求和电价不确定性的处理，以及开发用于解决大规模优化模型的有效计算算法。为了应对这些挑战，我们提出了一种多尺度集成策略，将运营模型与设计和容量规划模型有效地结合起来。此外，我们提出了分解方案，该方案具有有效处理电力密集型工厂（尤其是空气分离工厂）的实际过程网络的大规模确定性模型的潜力。最后，我们提出了一种潜在有前途的混合随机规划/鲁棒优化模型和解决方法，以预测产品需求和电价的不确定性的影响。更广泛的影响：拟议的 GOALI 项目有可能在以下方面产生显着的经济节约：企业范围内优化电力密集型产业，提高竞争力。拟议的 GOALI 项目将包括博士生的暑期实习。学生和 PI 访问普莱克斯。该项目还具有向卡内基梅隆大学高级过程决策中心 (CAPD) 的多家石油、化学和工程/软件公司合作和传播基本方法的潜力。我们还打算与卡内基梅隆大学的电力能源系统小组合作。我们计划让本科生参与记录案例研究，这些案例研究将通过互联网在我们的 MINLP 网站上提供。最后，我们还计划参加卡内基梅隆大学的大学外展计划，通过使用简单的日常示例（例如决定使用哪些设备），让高中生了解空气分离技术以及电价波动下运行中的主要问题如果公用事业公司收取每小时变化的电价，则可以在他们的房屋中打开和关闭。