A primal–dual decomposition for the Stochastic Refinery Operational Planning Problem

Speaker: Mr Tiago Andrade
PUC-Rio and RMIT

Title: A primal–dual decomposition for the Stochastic Refinery Operational Planning Problem

Date and time: Friday 28 July 2017, 3:00–4:00pm
Location: Building 8 Level 9 Room 66 (AGR) RMIT City campus

Abstract: This talk is divided in six parts:
i) Relaxations for the nonconvex (Mixed Integer) Quadratically Constrained Quadratic Programming (MIQCQP) will be reviewed.
ii) A recent relaxation technique, Normalized Multiparametric Disaggregation Technique (NMDT) will be improved and will be shown how to get a better relaxation without increasing the model size too much.
iii) How Lagrangian relaxation can be used to decompose MIQCQP problem with special structure and issues that appear when applied to the nonconvex case.
iv) How to use use Lagrangian relaxation together with the reformulated NMDT to create a decomposition scheme to the nonconvex MIQCQP with special structure.
v) Present the Stochastic Refinery Operational Planning Problem (SROPP).
vi) Apply the derived method to do a primal-dual decomposition to the SROPP.

Bio: Tiago Andrade is a PhD student at Puc-Rio in Industrial Engineering and a research visitor at RMIT. His research focus is on modeling and solving Mathematical Programming applied to the oil and gas industry. The models used in his research belong to the classes Mixed Integer Programming or Mixed Integer Nonlinear Programming.

Download Poster

Skip to toolbar