CAREER: Integrated Production Management and Process Control of Energy-Intensive Processes

1454433 - baldea提案包括一个综合研究和教育计划，以解决能源密集型化学过程系统与电网的相互作用。化学过程的电力使用可以通过改变生产时间表来调节，以适应其他电网用户需求的变化：在非高峰时间增加产量，而在产量降低的高峰时间储存和销售多余的产品。这种需求响应（DR）操作要求在短时间间隔（例如，每小时）内做出生产管理决策，其中过程动态和控制是高度相关的。基于此，本项目的研究部分旨在为连续DR过程的生产调度和过程控制的优化集成提供一个新的框架。该方法基于在调度模型中嵌入闭环过程动力学的降阶表示。该项目的教育部分将向工程专业的学生介绍化学过程系统和电网之间的联系。研究生和少数民族本科生都将参与研究活动。将开发一套新颖的实践学习活动（其中依赖于增材制造），并用于培养创造性思维，i)新的一年级工程课程和ii)高级过程控制课程。拟议的外展活动将使低收入家庭的中学生参与STEM学习，并支持他们成为第一代大学毕业生。提出的集成生产调度和过程控制框架将通过工业案例研究进行验证。预计将获得从业者的认可，并扩大参与DR的工业基础（包括空分、水泥、氯碱、铝等，占美国工业用电量的10%以上），从而大幅降低电网的净峰值电力需求。其他几种化学工艺（如聚合物、废水处理）也存在类似的调度和控制挑战，该项目的解决方案可以用于改善其操作。化学过程的调度和控制的集成是具有挑战性的，因为这两种活动之间的时间范围存在差异，并且在所有相关的时间尺度上描述系统行为所需的模型的大小。该项目探索了克服这些困难的新方向：PI提出了时间尺度桥接的概念，并开发了与调度相关的低阶动态模型，该模型可以捕获过程的闭环行为。然后将这些模型合并为调度公式中的约束。他还介绍了一个新的控制理论方向，调度-MPC，将这些思想扩展到广泛使用的模型预测控制（MPC）范式。预计这些发展将减少在调度DR化学过程中考虑动力学和控制所需的计算工作量，并赋予集成框架鲁棒性。此外，所提出的故障检测技术将为流程重调度决策提供新的机制。在更广泛的背景下，化学品供应链的能源效率和经济绩效的未来改善需要“更智能”的制造，基于共享信息和同步所有级别的运营决策，从监管和监督控制到生产调度和计划。调度与控制的集成是协调过程决策层次中制造管理层和控制层的关键，但迄今为止受到的关注相对较少。因此，拟议的研究解决了一个重要而开放的问题，并将开发智能制造框架中目前缺失的环节。

1454433 - BaldeaThe proposal comprises an integrated research and education plan to address the interaction of energy-intensive chemical process systems with the power grid. The electricity use of chemical processes can be modulated to accommodate the variation of the demand of other grid users by changing production schedules: production is increased during off-peak hours and products generated in excess are stored and sold at peak times, when production is lowered. This demand response (DR) operation calls for making production management decisions over short (e.g., hourly) time intervals, where process dynamics and control are highly relevant. Motivated by this, the research component of the project aims to provide a new framework for the optimal integration of production scheduling and process control of continuous DR processes. The approach is predicated on embedding reduced-order representations of the closed-loop process dynamics in the scheduling model. The educational component of the project will introduce engineering students to the nexus between chemical process systems and the electric grid. Both graduate and minority undergraduate students will be engaged in the research activities. A suite of novel hands-on learning activities will be developed (relying, amongst others, on additive manufacturing), and used to foster creative thinking in, i) a new first-year engineering course and, ii) in the senior process control class. The proposed outreach activities will engage middle school students from low-income families in STEM learning and support their efforts to become first-generation college graduates. The proposed integrated production scheduling and process control framework will be validated with industrial case studies. It is expected to gain practitioner acceptance and expand the industrial base participating in DR (including, e.g., air separation, cement, chlor-alkali, aluminum, which account for over 10% of industrial electricity use in the U.S.), leading to a sizable reduction in net peak power demand in the grid. Several other chemical processes (e.g., polymers, wastewater treatment) pose similar scheduling and control challenges, and solutions from this project can be deployed to improve their operations. The integration of scheduling and control for chemical processes is challenging due to the discrepancy in time horizons between the two activities and to the size of the models required to describe system behavior over all relevant time scales. The project explores a new direction to overcome these difficulties: the PI proposes the concept of time scale-bridging, and the development of scheduling-relevant low-order dynamic models that capture the closed-loop behavior of a process. These models are then incorporated as constraints in the scheduling formulation. He also introduces a new control-theoretical direction, scheduling-MPC, to extend these ideas to the widely used model predictive control (MPC) paradigm. These developments are expected to reduce the computational effort required to account for dynamics and control in scheduling DR chemical processes, and to impart robustness to the integrated framework. Additionally, the proposed fault detection techniques will provide novel mechanisms for making process rescheduling decisions. In a broader context, future improvements in the energy efficiency and economic performance of the chemical supply chain call for "smarter" manufacturing, based on sharing information and synchronizing all levels of operational decisions, from regulatory and supervisory control, to production scheduling and planning. The integration of scheduling and control, which is the pivotal point for coordinating the manufacturing management and control layers of the process decision-making hierarchy, has received relatively little attention to date. The proposed research thus addresses an important and open problem, and will develop a currently missing link in the smart manufacturing framework.