Tuesday, May 10, 2016

[DMANET] PhD position in Nancy (France)

A PhD financed by the University of Lorraine, Nancy (France), is available.
The PhD project is the result of the collaboration between two laboratories in Nancy: LRGP and LORIA, and Università di Roma Tor Vergata (Italy).
The financing will be for 3 years, starting from October 2016.
To obtain the double title (Italian and French), it is necessary to participate to the Italian competition for the PhdD school in Computer science, control and geoinformation at Università di Roma Tor Vergata.
The application for the Italian competition must be sent before the 30th of May (we can provide assistance in order to produce the documents)

Below you can find a description of the thesis.

For any additional questions, you can contact:
Veronica Piccialli ( veronica.piccialli@uniroma2.it )
Bernardetta Addis ( bernardetta.addis@loria.fr ), co-director of the thesis with Christophe Castel (LRGP - Nancy)

Veronica Piccialli​ and Bernardetta Addis

-----------------------------------------------------------
Mixed Integer Nonlinear programming for membrane system Design

Introduction
Membrane gas separation processes are currently applied in numerous industrial sectors (energy,
chemistry, pharmacy, electronics, aeronautics) [B2004] and are considered as one the most
attractive technology for the development of sustainable processes for industries [H2008, TR2005].
Membrane separations indeed combine a series of key advantages: continuous operation without a
regeneration step (simple process), no phase change (energy efficient process), compact units due
to the large specific surface area of modules (intensified process), physical separation process
without chemical reactions involved and without wasted chemicals losses (environmental friendly
process) [HK1997].
The design of membrane gas separations processes for a target application has received
considerable attention for decades [B2004, H2008, EEA1998]. A state of the art analysis shows that
the separation performances of a single membrane module can, in a large majority of cases, be
correctly predicted through a standard chemical engineering modelling and simulation approach
[EEA1998, KKS2000]. Nevertheless, the achievement of the target process performances (i.e.
product purity and recovery, energy efficiency, productivity) often requires multistage processes to
be designed, including recycling loops [B2004, KM2000, B2002]. The resulting process architecture
can be complex and the identification of the optimal design (number of stages, number and location
of recycling loops) addresses a formidable computational challenge [KKS2000, KM2000].

>From an optimization point of view, the design of membrane gas separations processes can be
modelled as a Mixed Integer Nonlinear Programming (MINLP) problem. Continuous variables and
constraints allow to model a single membrane behaviour, and discrete variables are necessary to
parametrize the number and connection of different membranes (and eventually the type of
membrane involved). In recent years, some works addressed the problem of membrane design
using MINLP models [QH2000­SWM2014], and proposing solution techniques, but all of them
impose strong restrictions on the possible design and/or focus on a specific application.
Defining a suitable MINLP model able to represent all the spectrum of interesting configurations is
already a challenge, and to the best of our knowledge there is no work addressing this issue.
Effectively solving the design problem asks for finding an appropriate algorithm to solve the
resulting optimization model. MINLP is NP­hard, and state of the art solvers either require some
hypothesis (as convexity or structure of the problem) or are able to solve only small size instances.

All the existing algorithms for solving MINLP have some limitations, for example some can deal only
with convex optimization, others ask for algebraic representation of the constraints and the
objective function and impose some limitation on their form. Problems that can be solved at
optimality are still of quite small size, compared to the ones solved by Mixed Integer Programming
solvers and Nonlinear Programming solvers.
For these reasons, researchers deal with MINLP tailoring and combining existing tools, exploiting
specific knowledge on the considered system. This allows to reduce the search space, improving the
algorithm performances in terms of computational time and quality of the solution [Betal2012, FCRL2008, LLPV2005].

Objective
The objective of the thesis is to develop a tailor made computer program in order to identify the
optimal multistage membrane process design (architecture) for a set of target performances (i.e.
purity, recovery, energy efficiency, cost).

To this aim adequate optimization models to describe a general membrane system and optimization
methods to find optimal or nearly optimal solutions of such models must be studied. To the best of
our knowledge, no systematic simulation and optimization tool is available today in order to achieve
this purpose.

To reach this objective a key role is played by the interdisciplinarity of the collaboration in act
between the LRGP and LORIA in Nancy and the Università di Roma "Tor Vergata", Italy.
In fact, conception and development of efficient methods for difficult MINLP problems need
combination of skills in advanced optimization techniques (LORIA, Università di Roma "Tor
Vergata") and a deep knowledge of membranes systems (LRGP).

The thesis will be co­directed by Christophe Castel (LRGP ­ Nancy), Bernardetta Addis
(LORIA­Nancy) and Veronica Piccialli (Università di Roma "Tor Vergata").

References
[B2004] R.W. Baker (2004) "Membrane Technology and Applications" ​, 2nd Ed. John Wiley & Sons, Chichester, England.
[H2008] H.W. Häring (2008) ​"Industrial Gas Processing", Wiley VCH.
[TR2005] "Materials research for separations technologies: energy and emission reduction
opportunities", ​US Department Of Energy, Industrial Technologies Program, Tech Report, May 2005.
[HK1997] J.L. Humphrey, G.E. Keller (1997) "Separation Process Technology" ​, MacGraw Hill Ed., New York.
[EEA1998] H.M. Ettouney, H.T. El­Dessouky, W. Abou Waar (1998) "Separation characteristics of air by polysulfone hollow fiber membranes in series", Journal of Membrane Science, 148, 105­117.
[KKS2000] S.P. Kaldis, G.C.Kapantaidakis, G.P. Sakellaropoulos (2000) "Simulation of
multicomponent gas separation in a hollow fiber membrane by orthogonal collocation – Hydrogen recovery from refinery gases", Journal of Membrane Science, 173, 61­71.
[KM2000] W.J. Koros, R. Mahajan (2000) "Pushing the limits on possibilities for large scale gas separation: which strategies?", ​Journal of Membrane Science, 175, 181­196.
[B2002] R. W. Baker (2002) "Future directions of membrane gas separation technology" ​, Industrial & Engineering Chemistry Research, 41, 1393.
[QH2000] R. Qi, M.A. Henson (2000) " ​Membrane system design for multicomponent gas mixtures via mixed­integer nonlinear programming", Computers & Chemical Engineering, 24, 2719­2737.
[SWM2014] M. Skuborowski, J. Wessel, W. Marquardt (2014) "Efficient Optimization­Based Design of Membrane­Assisted Distillation Processes", Industrial & Engineering Chemistry Research, 53, 15698­15717.
[Betal2012] C. Bragalli, C. D'Ambrosio, J. Lee, A. Lodi, P. Toth (2012) "On the optimal design of water distribution networks: a practical MINLP approach", Optimization and Engineering, 13, 219–246.
[FCRL2008] T. Farkas, B. Czuczai, E. Rev, Z. Lelkes (2008) "New MINLP model and modified outer approximation algorithm for distillation column synthesis", Industrial & Engineering Chemistry
Research, 47(9), 3088–3103.
[LLPV2005] G. Liuzzi, S. Lucidi, V. Piccialli, M. Villani (2005) ​"Design of induction motors using a mixed­ variable approach. Computational Management Science", ​2 (3), 213­228.

**********************************************************
*
* Contributions to be spread via DMANET are submitted to
*
* DMANET@zpr.uni-koeln.de
*
* Replies to a message carried on DMANET should NOT be
* addressed to DMANET but to the original sender. The
* original sender, however, is invited to prepare an
* update of the replies received and to communicate it
* via DMANET.
*
* DISCRETE MATHEMATICS AND ALGORITHMS NETWORK (DMANET)
* http://www.zaik.uni-koeln.de/AFS/publications/dmanet/
*
**********************************************************