Energy system design under uncertainty

The current context leads energy system design to very demanding objectives, due to their variety. Indeed, despite an increasing energy demand, environment indicators are becoming always more important. So that for a given service, emission (and then associated consumption as well) is desired to decrease. Improving systems efficiencies is then a important step. Such a problem is formulated as an optimization. It is based on numerical models. Every models differs by definition from reality. This difference can be translated into uncertainties. Usually, they are considered at their most probable value. However, their variation can lead to consequences between a performance decrease and plant inoperability. It is then critical to take into account the deviation due to uncertainties when optimizing an energy system. The optimization problem will be described. It will introduce the description of functions and variables involved in energy system design. The formulation of the optimization under uncertainty will be developed, as well as mathematical methods for uncertainty propagation. Finally, an innovative method taking advantage of the high number of iterations due to the chosen solver will be described. In this study, pinch analysis has been applied. Its limits related to uncertainties treatment will be presented. Methods described here will be applied to an hybrid system of a fuel cell coupled with gas turbines. Results will be compared to a conventional optimization solutions. It will demonstrate that, despite sub-optimal objectives, the sensitivity of the system to uncertainties has been improved.

Urban Energy System Design from the Heat Perspective using mathematical Programming including thermal Storage

Jakob Moritz Fabian Rager

Energy planning recently received more attention in Switzerland through the new strategy phasing out nuclear energy by 2034. Often however the energy planning is only done from the electrical side. This work takes a different angel and helps communities and energy utilities planning tomorrows energy system from a heat based perspective. After the data collection and structuring, the methodology presented here designs an energy system. Based on the quality of the collected data, the approach to define the energy demand should be chosen. In order to reduce calculation time, a data reduction approach is developed to reduce the input data without loosing significant information and precision. In particular, the methodology focuses on the integration of a stochastic resource, in this case solar thermal heat production, in combination with thermal energy storage. The thermal energy storage can be used as a short or long term thermal energy storage. The framework compares design solutions for the two storage types considering either a total cost approach or a life cycle assessment approach using the cumulative exergy demand (CExD). The proposed mathematical programming framework is based on a mixed integer linear programming (MILP) approach, that can work on different levels of detail between building to community or city level. The optimization problem can also be further simplified to a linear problem, increasing the size of problem that can be solved while reducing or keeping a constant computation time. The discussed cases show an interest in further investigating storage solution using both, the short and long term storage at once, because they allow to reduce the system's overall costs or CExD significantly. The framework is then extended to consider buildings as an energy storage. The building's internal temperature can be raised from 20 \degree C up to 23 \degree C, giving a comfort temperature range that can be used for storing heat. The integrating of both storage types, the thermal energy storage and the building as an energy storage, show no significant impact on the energy system design. However, costs or CExD can be reduced. In addition, the heat demand can be modified through the decision of optimal energy retrofitting strategy for a group of buildings. The framework decides which of the building to refurbish based the overall CExD including CExD used for retrofitting the building. Finally, a method is proposed to integrate uncertainty of the model's input parameters into the system design. A global sensitivity analysis evaluates the impact of each uncertain parameter onto the system, allowing to focusing on the outputs of interest. Robust optimization is applied with a simulation-based approach, the additional costs for a robust design are calculated, as well as the different unit sizes. The low complexity of the developed models allows for an easy integration of new data collected during the development of a project, which is often the case in urban energy planning applications.

EPFL2015

Multi-objective design and optimization of district energy systems including polygeneration energy conversion technologies

Céline Weber

In the present context of finding ways to decrease CO2 emissions linked with human activity, district energy systems including polygeneration energy conversion technologies are likely to play a major role. District energy systems meet the heating, hot water, cooling and electricity requirements of a district. Because they meet several types of energy requirements, and for more than one single building, district energy systems represent good opportunities to implement polygeneration energy conversion technologies. Polygeneration energy conversion technologies indeed provide different energy services simultaneously, helping to decrease the CO2 intensity compared to energy conversion technologies that meet only one energy service. Moreover, when providing energy to a whole district, polygeneration energy conversion technologies can take advantage of the various load profiles of the buildings by compensating the fluctuations and having therefore a smoother operation. A district energy system comprises essentially two parts: the plant with the polygeneration energy conversion technologies, and the distribution networks (heating and cooling). When designing the energy system for a district, one has therefore to define which type of polygeneration energy conversion technologies are best suited for the district, as well as which building are worse connecting to the system and which buildings shouldn't be connected (for instance if they are located too far away from the other buildings or if they have too small requirements to justify a connection from the plant). Moreover the operation strategy needs to be defined. In the present thesis, a method is developed that helps designing and optimizing district energy systems, from the structuring of the information available for the district (energy consumption profiles, location of the buildings, available energy sources, possible layouts for the pipes,...), over the thermo-economic modelling of the energy conversion technologies, the design of the network and the simulation of its operation strategy, and finally the evaluation of the results in terms of CO2 emissions and costs. The design and optimization of the district energy system is a multi-objective Mixed Integer Non Linear Programming problem. To solve this problem, a decomposition strategy including a master and a slave problem was developed. The master optimization problem takes care of the energy conversion technologies, whereas the slave optimization problem optimizes the network part. The two sub-problems are solved iteratively and result in the definition of a Pareto optimal curve that gives the trade-offs between the emissions and the costs for various configurations satisfying the requirements of the district. A configuration is characterized by given types and sizes of energy conversion technologies, their location in the district, the network layout, as well as the operation strategy of the technologies. Due to the time dependent energy consumption profiles and the geographical location of the buildings and plant, the method developed combines two well known types of problems, namely the multi-period optimization problems and the network problems. The method developed allows to take into account various constraints such as limited availability of energy sources, forbidden connections between buildings (for instance if a large river separates these two buildings), or else space limitations in underground technical channels. The capabilities of the method are demonstrated by means of a test case, as well as a real case in the Canton of Geneva. The results show the importance of considering all the energy services together (and not separately). Energy systems including a gas engine or a gas turbine combined cycle, together with heat pumps, indeed help decreasing both the emissions and the costs compared to the actual configurations. In the Geneva case study for instance, emissions can be decreased by up to 45%, with a simultaneous costs reduction of 24%. However, the method only deals with water networks, while in some cases space limitations and safety issues make the use of water impossible. A new type of district energy system based on CO2 as energy transfer medium (instead of water), is therefore developed in order to take such issues into account. This new system, that led to the submission of a patent, meets all the different types of energy requirements with only two pipes (instead of three or four like in conventional water based system), and uses the latent heat of CO2 as driving force, instead of the specific heat.

EPFL2008

Méthodologie de la synthèse des systèmes énergétiques industriels

Raffaele Bolliger

This work presents a synthesis method that leads to the preliminary design of industrial energy systems. Such systems are composed of several technologies that transform, through a set of physical unit operations, raw materials and energy into products and energy services. The purpose of the preliminary synthesis is to define which technologies form the system, to calculate their technical characteristics, their operating conditions and their interactions, based on performance indicators such as for example the cost of the system or its environmental impact. This area is studied since a long time and many synthesis approaches are available. However, these methods have significant limitations, particularly in their ability to face the study of more and more complex systems. The main goal of the method presented in this work is to enable the study of large systems and promote the use of the experience gained from previous studies in the form of models and methodological approaches. The synthesis method of industrial energy systems is based on the overall system optimization using a model that is obtained by assembling several modules, representing each a set of physical unit operations and whose equations are formulated with the black box technique, and an integration model that can represent the possible interactions within the system. Optimization is performed using a multi-objective evolutionary algorithm, whose objective functions are defined on the basis of calculation of several performance indicators based on the model of the system. To simplify the resolution of this complex system, the optimization problem is decomposed into a master problem, responsible for calculating the characteristics of the units and their operating conditions, and an optimization slave subproblem, which selects the units being part of the system and their interconnections. To ensure its robustness, the slave subproblem is formulated as a mixed-integer linear optimization problem. The slave subproblem is formulated by using process integration techniques, which are extended in this work to allow the synthesis of multiple heat and mass transfer networks. The synthesis problem can thus be defined using an explicit superstructure or by one generated automatically, either implicit or explicit. This work also introduces a technique that greatly reduces the number of degrees of freedom of the integration model. Instead of separately optimizing each ΔTmin/2 associated with heat streams, a formula is applied to calculate their value from a reference case. The ΔTmin/2 optimization is thus reduced to the optimization of a single decision variable related to the reference case, regardless of the size of the problem. The proposed method uses a set of heterogeneous elements, including flowsheeting software for modeling physical unit operations, mathematical programming tools for the formulation of the integration problem, methods for calculating performance indicators and calculation tools such as the optimization algorithm. This work introduces new tools developed to systematically apply the proposed methodology and to automate recurring operations of data transfer and the call of the various used software. In particular, a syntax description is defined as an abstraction layer to describe and to structure the exchanged information. A computing platform has been created to support the application of the method and to ensure the data transfer between its components. Two case studies are presented to illustrate the various aspects of the synthesis method. A first case, involving the synthesis of two combined cycles, has been chosen to illustrate the different application stages of the method and to show the potential reuse of certain modules. Through the integration techniques, it has been possible to identify potential heat recovery that can increase the performance of one of the cycles beyond what had been expected by experts using conventional simulation techniques. The second case study is about the treatment of waste generated by an industrial site active in the field of fine chemicals. Waste treatment can recover different materials and energy services useful for process units, thereby reducing the quantities purchased in the market. The model of multi-network integration can easily solve the complexity of the problem of waste management in developing strategies for allocating waste to the various treatments available for different objective functions related to operating costs and environmental impact.

EPFL2010