Skip to Main content Skip to Navigation
Theses

LNG portfolio optimization, approach by stochastic programming techniques

Abstract : The work presented in this Ph.D dissertation is motivated by the problem of management of a fleet of cargos transporting lique ed natural gas (LNG) initially proposed by Total. The holder of the portfolio has to meet its commitments towards its counterparts while trying to generate pro ts through arbitrating different commodities market. Thus, the management of portfolio can be modeled as a stochastic, dynamic and integer optimization problem. This Ph.D dissertation is organised as follows: 1 We introduce a numerical method for solving continuous relaxation problem. We propose an algorithm based on the combination of the vectorial quantization method as discretization method and the dual dynamic programming approach. We show the convergence of numerical schema and give the error analysis on the discretization by quantization. Some numerical tests on real energy market problem are performed. 2 We also study the risk averse optimization by using conditional value at risk (CVaR) as criterion. We show that the algorithm proposed is also adapted to such formulation. Furthermore, we propose the technique of changes in probability measure in stochastic programming in order to improve rare scenario simulation. Same numerical test as in risk neutral problem is performed in order to make comparison. 3 We study the sensitivity of the portfolio with respect to several parameters in the market price model. We proposed a numerical method to compute sensitivity value based on Danskin's theorem. The convergence of sensitivity value of discretized problem to the one of original problem is proved. Comparison between result obtained by our algorithm and other classical methods are provided. 4 We study the stochastic integer programming problem. The integrality cutting plane method is applied to approximate the integer problem. We show that it is impossible to converge to the integer solution because of the non convexity and discontinuity of the Bellman value function. We apply an heuristic method and propose a small improvement. Some numerical tests are also provided.
Document type :
Theses
Complete list of metadatas

https://pastel.archives-ouvertes.fr/pastel-00645441
Contributor : Zhihao Cen <>
Submitted on : Monday, December 5, 2011 - 5:01:36 PM
Last modification on : Wednesday, March 27, 2019 - 4:08:30 PM
Long-term archiving on: : Friday, November 16, 2012 - 2:20:24 PM

Identifiers

  • HAL Id : pastel-00645441, version 1

Collections

Citation

Zhihao Cen. LNG portfolio optimization, approach by stochastic programming techniques. Optimization and Control [math.OC]. Ecole Polytechnique X, 2011. English. ⟨pastel-00645441⟩

Share

Metrics

Record views

715

Files downloads

393