Operation strategy for repeated batch/fed-batch processes considering uncertainty = 불확실성을 고려한 반복되는 회분식/유가식 공정의 운전 전략

Cited 0 time in webofscience Cited 0 time in scopus
  • Hit : 12
  • Download : 0
A model-based optimization can be a systematic way to improve performances of batch/fed processes considering their inherent nonstationary characteristics and nonlinear dynamics while satisfying the desired constraints of operation and product quality. However, an open-loop optimization may give a highly suboptimal recipe leading to constraint violations. Since uncertain parameters can seldom be directly measured, they should be estimated using measurements but the estimation of all uncertain parameters often becomes ill-conditioned problem. The conventional optimization approaches for handling uncertainty have the following limitations: robust optimization (RO) based on prior knowledge can result in performance loss due to its conservative nature, and measurement-based optimization (MBO) approaches do not typically deal with the residual uncertainty after parameter identification. This dissertation has addressed the above-mentioned challenges for designing an effective and efficient operation strategy for repeated batch/fed-batch processes considering uncertainty. An improved parameter estimation method is developed to increase the accuracy of estimates in the ill-conditioned estimation problems, and its performance is demonstrated through statistical analysis and then through case studies of linear and nonlinear regressions. A new framework of robust batch-to-batch optimization (RB2BO) with scenario adaptation is developed to overcome the limitations of the conventional RO and MBO. The iterative scenario adaptation using new measurements and robust optimization mechanism can reduce the conservativeness of the recipe obtained from the conventional RO and consider the residual uncertainty after the estimation of the uncertain parameters. In addition, the developed estimation method and an optimal sampling point design are adopted to alleviate the ill-conditioning problem in the estimation, resulting in the improvement in the quality of the adaptation. Therefore, the proposed framework provides less conservative and more robust recipes than the conventional RO and MBO, respectively. Time and cost for measurement and optimization can be reduced by adopting the suggested termination criteria of the batch-to-batch correction.
Advisors
Lee, Jay Hyungresearcher이재형researcher
Description
한국과학기술원 :생명화학공학과,
Publisher
한국과학기술원
Issue Date
2019
Identifier
325007
Language
eng
Description

학위논문(박사) - 한국과학기술원 : 생명화학공학과, 2019.2,[vi, 119 p. :]

Keywords

Batch/fed-batch process▼aDynamic optimization under uncertainty▼aParameter estimation▼aRobust batch-to-batch optimization▼aScenario adaption; 회분/유가식 공정▼a불확실성하에서의 동적최적화▼a매개 변수 예측▼a강건한 batch-to-batch 최적화▼a시나리오 개선

URI
http://hdl.handle.net/10203/264885
Link
http://library.kaist.ac.kr/search/detail/view.do?bibCtrlNo=842122&flag=dissertation
Appears in Collection
CBE-Theses_Ph.D.(박사논문)
Files in This Item
There are no files associated with this item.

qr_code

  • mendeley

    citeulike


rss_1.0 rss_2.0 atom_1.0