Economic Optimization of Spray Dryer Operation using Nonlinear Model Predictive Control

Research output: Chapter in Book/Report/Conference proceedingArticle in proceedingsResearchpeer-review

Abstract

In this paper we investigate an economically optimizing Nonlinear Model Predictive Control (E-NMPC) for a spray drying process. By simulation we evaluate the economic potential of this E-NMPC compared to a conventional PID based control strategy. Spray drying is the preferred process to reduce the water content for many liquid foodstuffs and produces a free flowing powder. The main challenge in controlling the spray drying process is to meet the residual moisture specifications and avoid that the powder sticks to the chamber walls of the spray dryer. We present a model for a spray dryer that has been validated on experimental data from a pilot plant. We use this model for simulation as well as for prediction in the E-NMPC. The E-NMPC is designed with hard input constraints and soft output constraints. The open-loop optimal control problem in the E-NMPC is solved using the single-shooting method combined with a quasi-Newton Sequential Quadratic Programming (SQP) algorithm and the adjoint method for computation of gradients. The E-NMPC improves the cost of spray drying by 26.7% compared to conventional PI control in our simulations.
Original languageEnglish
Title of host publicationProceedings of the 53rd IEEE Conference on Decision and Control
Number of pages7
PublisherIEEE
Publication date2014
Pages6794-6800
ISBN (Print)978-1-4799-7746-8
DOIs
Publication statusPublished - 2014
Event53rd IEEE Conference on Decision and Control (CDC 2014) - Los Angeles, United States
Duration: 15 Dec 201417 Dec 2014
http://control.disp.uniroma2.it/CDC2014/index.php

Conference

Conference53rd IEEE Conference on Decision and Control (CDC 2014)
Country/TerritoryUnited States
CityLos Angeles
Period15/12/201417/12/2014
Internet address

Fingerprint

Dive into the research topics of 'Economic Optimization of Spray Dryer Operation using Nonlinear Model Predictive Control'. Together they form a unique fingerprint.

Cite this