## 90C15 Stochastic programming

### Refine

#### Document Type

- ZIB-Report (12)
- Master's Thesis (1)

#### Language

- English (13)

#### Has Fulltext

- yes (13)

#### Is part of the Bibliography

- no (13)

#### Keywords

- KKT recursion (2)
- Multistage Stochastic Programs (2)
- discrete dynamics (2)
- tree-sparse QP (2)
- Chemical Processes (1)
- Column Generation (1)
- Convex program (1)
- Flugzeugumlaufplanung (1)
- Gas Network Planning (1)
- Hierarchical KKT Sparsity (1)

#### Institute

Today's gas markets demand more flexibility from the network operators which in turn have to invest into their network infrastructure. As these investments are very cost-intensive and long-living, network extensions should not only focus on one bottleneck scenario, but should increase the flexibility to fulfill different demand scenarios. We formulate a model for the network extension problem for multiple demand scenarios and propose a scenario decomposition. We solve MINLP single-scenario sub-problems and obtain valid bounds even without solving them to optimality. Heuristics prove capable of improving the initial solutions substantially. Results of computational experiments are presented.

Mobile communication is nowadays taken for granted. Having started
primarily as a service for speech communication, data service and
mobile Internet access are now driving the evolution of network
infrastructure. Operators are facing the challenge to match the
demand by continuously expanding and upgrading the network
infrastructure. However, the evolution of the customer's demand is uncertain.
We introduce a novel (long-term) network planning approach based on
multistage stochastic programming, where demand evolution is considered as
a stochastic process and the network is extended as to maximize the
expected profit. The approach proves capable of designing large-scale
realistic UMTS networks with a time-horizon of several years. Our
mathematical optimization model, the solution approach, and computational
results are presented in this paper.

Application of Multistage Stochastic Programming in Strategic Telecommunication Network Planning
(2010)

Telecommunication is fundamental for the information society. In both, the
private and the professional sector, mobile communication is nowadays taken
for granted. Starting primarily as a service for speech communication, data
service and mobile Internet access are now driving the evolution of network
infrastructure. In the year 2009, 19 million users generated over 33
million GB of traffic using mobile data services. The 3rd generation
networks (3G or UMTS) in Germany comprises over 39,000 base stations with
some 120,000 cells. From 1998 to 2008, the four network operators in
Germany invested over 33 billion Euros in their infrastructure. A careful
allocation of the resources is thus crucial for the profitability for a
network operator: a network should be dimensioned to match customers
demand. As this demand evolves over time, the infrastructure has to evolve
accordingly. The demand evolution is hard to predict and thus constitutes a
strong source of uncertainty. Strategic network planning has to take this
uncertainty into account, and the planned network evolution should adapt to
changing market conditions. The application of superior planning methods
under the consideration of uncertainty can improve the profitability of the
network and creates a competitive advantage. Multistage stochastic
programming is a suitable framework to model strategic telecommunication
network planning.
We present mathematical models and effective optimization procedures for
strategic cellular network design. The demand evolution is modeled as a
continuous stochastic process which is approximated by a discrete scenario
tree. A tree-stage approach is used for the construction of non-uniform
scenario trees that serve as input of the stochastic program. The model is
calibrated by historical traffic observations. A realistic system model of
UMTS radio cells is used that determines coverage areas and cell capacities
and takes signal propagation and interferences into account. The network
design problem is formulated as a multistage stochastic mixed integer
linear program, which is solved using state-of-the-art commercial MIP
solvers. Problem specific presolving is proposed to reduce the problem
size. Computational results on realistic data is presented. Optimization
for the expected profit and the conditional value at risk are performed and
compared.

Robust Tail Assignment
(2010)

We propose an efficient column generation method to minimize the probability of delay propagations along aircraft rotations. In this way, delay resistant schedules can be constructed. Computational results for large-scale real-world problems demonstrate substantial punctuality improvements. The method can be generalized to crew and integrated scheduling problems.

Unnecessarily conservative behavior of standard process control techniques can be avoided by stochastic programming models when the distribution of random disturbances is known. In an earlier study we have investigated such an approach for tank level constraints of a distillation process. Here we address techniques that have accelerated the numerical solution of the large and expensive stochastic programs by a factor of six, and then present a refined optimization model for the same application.

Scenario tree models of stochastic programs arise naturally under standard nonanticipativity assumptions. We demonstrate how tree-sparse programs cover the general case, with \emph{arbitrary} information constraints. Detailed examples and intuitive interpretations illuminate the basic thoughts behind the abstract but elementary construction.

Tree-Sparse Convex Programs
(2001)

Dynamic stochastic programs are prototypical for optimization problems with an inherent tree structure inducing characteristic sparsity patterns in the KKT systems of interior methods. We propose an integrated modeling and solution approach for such tree-sparse programs. Three closely related natural formulations are theoretically analyzed from a control-theoretic viewpoint and compared to each other. Associated KKT solution algorithms with linear complexity are developed and comparisons to other interior approaches and related problem formulations are discussed.

Standard model predictive control for real-time operation of industrial production processes may be inefficient in the presence of substantial uncertainties. To avoid overly conservative disturbance corrections while ensuring safe operation, random influences should be taken into account explicitly. We propose a multistage stochastic programming approach within the model predictive control framework and apply it to a distillation process with a feed tank buffering external sources. A preliminary comparison to a probabilistic constraints approach is given and first computational results for the distillation process are presented.

Mathematical optimization techniques are on their way to becoming a standard tool in chemical process engineering. While such approaches are usually based on deterministic models, uncertainties such as external disturbances play a significant role in many real-life applications. The present article gives an introduction to practical issues of process operation and to basic mathematical concepts required for the explicit treatment of uncertainties by stochastic optimization.

Interior point methods for multistage stochastic programs involve KKT systems with a characteristic global block structure induced by dynamic equations on the scenario tree. We generalize the recursive solution algorithm proposed in an earlier paper so that its linear complexity extends to a refined tree-sparse KKT structure. Then we analyze how the block operations can be specialized to take advantage of problem-specific sparse substructures. Savings of memory and operations for a financial engineering application are discussed in detail.