Björn Geißler

Using Piecewise Linear Functions for Solving MINLP s

In this chapter we want to demonstrate that in certain cases general mixed integer nonlinear programs (MINLPs) can be solved by just applying purely techniques from the mixed integer linear world. The way to achieve this is to approximate the nonlinearities by piecewise linear functions. The advantage of applying mixed integer lin- ear techniques are that these methods are nowadays very mature, that is, they are fast, robust, and are able to solve problems with up to millions of variables. In addition, these methods have the potential of finding globally optimal solutions or at least to provide solution guarantees. On the other hand, one tends to say at this point “If you have a hammer, everything is a nail.”[15], because one tries to reformulate or to approximate an ac- tual nonlinear problem until one obtains a model that is tractable by the methods one is common with. Besides the fact that this is a very typical approach in mathematics the question stays whether this is a reasonable approach for the solution of MINLPs or whether the nature of the nonlin- earities inherent to the problem gets lost and the solutions obtained from the mixed integer linear problem have no meaning for the MINLP. The purpose of this chapter is to discuss this question. We will see that the truth lies somewhere in between and that there are problems where this is indeed a reasonable way to go and others where it is not.

Validation of nominations in gas network optimization: models, methods, and solutions

In this article, we investigate methods to solve a fundamental task in gas transportation, namely the validation of nomination problem: given a gas transmission network consisting of passive pipelines and active, controllable elements and given an amount of gas at every entry and exit point of the network, find operational settings for all active elements such that there exists a network state meeting all physical, technical, and legal constraints. We describe a two-stage approach to solve the resulting complex and numerically difficult nonconvex mixedinteger nonlinear feasibility problem. The first phase consists of four distinct algorithms applying mixedinteger linear, mixedinteger nonlinear, nonlinear, and methods for complementarity constraints to compute possible settings for the discrete decisions. The second phase employs a precise continuous nonlinear programming model of the gas network. Using this setup, we are able to compute high-quality solutions to real-world industrial instances that are significantly larger than networks that have appeared in the mathematical programming literature before.

Combination of Nonlinear and Linear Optimization of Transient Gas Networks

In this paper, we study the problem of technical transient gas network optimization, which can be considered a minimum cost flow problem with a nonlinear objective function and additional nonlinear constraints on the network arcs. Applying an implicit box scheme to the isothermal Euler equation, we derive a mixed-integer nonlinear program. This is solved by means of a combination of (i) a novel mixed-integer linear programming approach based on piecewise linearization and (ii) a classical sequential quadratic program applied for given combinatorial constraints. Numerical experiments show that better approximations to the optimal control problem can be obtained by using solutions of the sequential quadratic programming algorithm to improve the mixed-integer linear program. Moreover, iteratively applying these two techniques improves the results even further.

Mixed integer linear models for the optimization of dynamical transport networks

We introduce a mixed integer linear modeling approach for the optimization of dynamic transport networks based on the piecewise linearization of nonlinear constraints and we show how to apply this method by two examples, transient gas and water supply network optimization. We state the mixed integer linear programs for both cases and provide numerical evidence for their suitability.

A New Algorithm for MINLP Applied to Gas Transport Energy Cost Minimization

In this article, we present a new algorithm for the solution of nonconvex mixed-integer nonlinear optimization problems together with an application from gas network optimization, the gas transport energy cost minimization problem. Here, the aim is to transport gas through the network at minimum operating cost. The proposed algorithm is based on the adaptive refinement of a new class of MIP-relaxations and has been developed within an industry project on gas network optimization. Since therefore the implementation is not as general as it could be, our computational results are restricted to instances from gas network optimization at this point of time. However, as these problems are real-world applications and turn out to be rather hard to solve with the aid of state-of-the-art MINLP-solvers we believe that our computational results reveal the potential of this new approach and motivate further research on the presented techniques.

Solving power-constrained gas transportation problems using an MIP-based alternating direction method

Abstract We present a solution algorithm for problems from steady-state gas transport optimization. Due to nonlinear and nonconvex physics and engineering models as well as discrete controllability of active network devices, these problems lead to difficult nonconvex mixed-integer nonlinear optimization models. The proposed method is based on mixed-integer linear techniques using piecewise linear relaxations of the nonlinearities and a tailored alternating direction method. Most other publications in the field of gas transport optimization only consider pressure and flow as main physical quantities. In this work, we additionally incorporate heat power supplies and demands as well as a mixing model for different gas qualities. We demonstrate the capabilities of our method on Germanys largest transport networks and hereby present numerical results on the largest instances that were ever reported in the literature for this problem class.

Mixed Integer Optimization of Water Supply Networks

We introduce a mixed integer linear modeling approach for the optimization of dynamic water supply networks based on the piecewise linearization of nonlinear constraints. One advantage of applying mixed integer linear techniques is that these methods are nowadays very mature, that is, they are fast, robust, and are able to solve problems with up to a huge number of variables. The other major point is that these methods have the potential of finding globally optimal solutions or at least to provide guarantees of the solution quality. We demonstrate the applicability of our approach on examples networks.

Solving Highly Detailed Gas Transport MINLPs: Block Separability and Penalty Alternating Direction Methods

Detailed modeling of gas transport problems leads to nonlinear and nonconvex mixed-integer optimization or feasibility models (MINLPs) because both the incorporation of discrete controls of the network and accurate physical and technical modeling are required to achieve practical solutions. Hence, ignoring certain parts of the physics model is not valid for practice. In the present contribution we extend an approach based on linear relaxations of the underlying nonlinearities by tailored model reformulation techniques yielding block-separable MINLPs. This combination of techniques allows us to apply a penalty alternating direction method and thus to solve highly detailed MINLPs for large-scale, real-world instances. The practical strength of the proposed method is demonstrated by a computational study in which we apply the method to instances from steady-state gas transport including both pooling effects with respect to the mixing of gases of different composition and a highly detailed compressor station ...

Penalty Alternating Direction Methods for Mixed-Integer Optimization: A New View on Feasibility Pumps

Feasibility pumps are highly effective primal heuristics for mixed-integer linear and nonlinear optimization. However, despite their success in practice there are only a few works considering their theoretical properties. We show that feasibility pumps can be seen as alternating direction methods applied to special reformulations of the original problem, inheriting the convergence theory of these methods. Moreover, we propose a novel penalty framework that encompasses this alternating direction method, which allows us to refrain from random perturbations that are applied in standard versions of feasibility pumps in case of failure. We present a convergence theory for the new penalty based alternating direction method and compare the new variant of the feasibility pump with existing versions in an extensive numerical study for mixed-integer linear and nonlinear problems.

Welfare Maximization of Autarkic Hybrid Energy Systems

Hybrid energy systems become a promising way for electrification of off-grid rural areas. We consider an autarkic mini-grid of households equipped with local solar panels, diesel generators and energy storage devices. Our aim is to find an energy distribution that maximizes the global welfare of the whole system. We present an MIQP model for the hybrid energy system optimization problem together with some remarks on computational results .