FNRS

Beyond Incentive Regulation

Coordinators: Per Agrell (CORE and ILSM, UCL), Axel Gautier (Université de Liège and CORE) and Sergio Perelman (Université de Liège)
Date: 2012 - 2015
Project: FRFC

Traditionally, public authorities implement separate policies to regulate financial and quality performance of network utilities. In addition, governmental policies put more emphasis on the former than on the latter. The trade-off between service level and cost efficiency is implicit and its properties are poorly examined. Moreover, new developments in these industries -- e.g. smart grids in electricity distribution and environmental concerns in water distribution -- stress even more the need for a new holistic approach to regulation.

It is customary to evaluate network utilities' performance based on the amount of output provided or total number of customers served. In essence, the current regulation concentrates on the quantity dimension of service provision. One of our goals is to develop models where network operators' performances are assessed using a quality normalized multi-output perspective. In this alternative context, quality features correspond to these additional dimensions included in the analysis. In particular, we expect to elaborate empirical studies that analyze the regulation of quality for energy (electricity, gas) water and railways infrastructure provision. Our aspiration is to contribute to policy-making with the development of a family of optimal incentive mechanisms defined simultaneously by service quality and financial performance.

Control and Optimization of Networked Systems with Poorly Understood Couplings

Coordinators: François Glineur (CORE and ICTM, UCL) and Julien Hendricks (INMA, UCL)
Date: October 2012 - September 2016
Project: FRIA

The deployment of wind turbines has experienced a dramatic increase over the recent years, and wind power is expected to represent a significant part of the electricity production in the future. For practical and economical reasons, it is often advantageous to deploy several wind turbines in a same location, in a "wind farm". Certain of these farms contain up to 100 turbines.

The complex couplings and interactions between turbines in these farms create new challenges in control and optimization (and indeed in fluid dynamics): in order to produce electricity, a turbine takes energy from the wind, which may decrease the strength of the wind arriving at some of the other turbine, and create complex perturbations and turbulences. In this context, nothing guarantees that a selfish maximization of each turbine’s production is globally optimal, and indeed current simulation results show that it is not. Similar problems arise when one is trying to stabilize the production of the wind farm under varying conditions, in order to avoid sharp variations on the electricity network. Understanding exactly how to control or optimize the production of these farms is challenging, as the coupling between the turbines involve very complex fluid dynamics phenomena, the accurate simulation of which is already an important research subject on its own. In particular, it appears that an accurate simulation of the interferences taking place in a large farm will be beyond reach for long, and would be extraordinarily costly from a computation point of view. On the other hand, a range of simplified models are available, but it is unclear how their (in)accuracy may affect the performances of the control laws relying on them. On the one hand it would not make sense for the control law should not be able to take all dynamical effects into account (wind speed is a very fast-varying input, and pitch actuators have a very limited speed), and on the other hand, the model should be accurate enough to allow the controller to be robust to its prediction faults.

Our research goal is to develop control and optimization methods for such networked systems with unknown or poorly understood coupling, which is thus directly related to wind farms.

Game Theoretic Approaches in Supply Chain Management

Coordinators: Per Agrell (CORE and ILSM) and Constantin Blome (CORE and LSM)
Date: 01/01/2012 - 31/12/2016

The project aims at developing modelling and experimental approaches for strategic behaviour in supply chains using cooperative game theory. Relatively little positive work has been published in the area of game theory in supply chain coordination, in spite of clearly identified opportunities for applications (limited number of well-defined strategic decision makers, observable decisions of durable character, repeated interaction, etc).

Development of a Tool for Planning and Optimizing Active Autonomous Electrical Distribution Grids

Coordinators:Emmanuel De Jaeger (MCTR, UCL) and François Glineur (CORE and INMA, UCL)
Date: October 2013 - September 2017
Project: FRIA

The project of research is intended to develop a tool for the planning of active low-voltage distribution networks. The active nature covers four aspects: the presence of distributed generation, the flexibility of the load, the storage capacities and the active management of the grid thanks to an increased level of automation. Furthermore, this study will focus on the case of autonomous networks in a context of rural electrification for developing countries. Those autonomous grids are self-sufficient in terms of production capability and may have the possibility to be connected to a higher voltage level.

The development of the planning tool will occur in two steps. First, it is necessary to model the load, the distributed generation, the storage and the grid with appropriate models. In particular, a fraction of the load and the dispersed generation based on renewables have a stochastic nature. Probabilistic previsional models need to be used at this level. Then, a multi-objective optimization must be carried out based on a set of predefined technico-economic criteria and constraints such as the reliability of the supply, the minimization of the losses in the lines or the optimal location of storage elements. The integration of a set of models for the aspects above and an optimization tool, along with a process of decision support, should lead to a flexible and built-in tool. This tool will work by making a comparison with the traditional planning methods and will be tested on real study-cases in order to make the necessary adjustments.

High Dimensional Econometrics

Coordinator: SébastienVan Bellegem (CORE, UCL)
Date: 01/07/2013 - 30/06/2018
Project: FRESH

High dimensional models arise today in a lot of economic studies. In a linear regression model, for instance, it corresponds to the situation where the number of covariates is large, i.e. close or larger than the sample size. In a multivariate time series setting, high dimensionality refers to the high number of time series that are studied jointly. Due to the ease of data collection today, the empirical researcher faces more frequently such large data sets. High dimensionality also appears in linear models when the covariate is not a random variable but a random function. Examples are given by the spot electricity prices that are observed continuously over time, or fertility curves used in development economics to measure the density of birth rate over mother's age.

The wide availability of large data sets has increased the hope to address empirically major substantive questions. In the two cited examples, they are, for instance, the impact of electricity spot price on future contracts or the impact of the shift in fertility curve on economic growth. However, these new promising directions of research are also hampered by several major methodological obstacles. Classical methods of modeling and inference (such as, e.g. estimation by GLS) are not robust to a large increase in the dimension of the econometric model.

The goal of the present research program is to address a number of those methodological obstacles, in view of providing workable and theoretically justified econometric methods and efficient inferential tools. In particular, the program is organized around three interlocking aspects : the failure of stationarity in time series collection, the modelling of large dimensional
covariance matrices and the endogeneity of covariates in high dimensional regression.

Optimal Fertility, Health and Education in Market Economies

Coordinator: Julio Dávila (CORE, UCL)
Date: 01/07/2013 - 30/06/2018
Project: PDR

The proposed research aims at identifying policies that allow improving the steady state efficiency and welfare upon the laissez-faire market outcome. This is achieved exploiting the externalities existing between the households' decisions on savings, fertility, health, and education on their own incomes at the aggregate level. Such externalities are disregarded by households at the market equilibrium under laissez-faire, which creates room for improvement through policy intervention.