Finkenrath, Matthias
Refine
Year of publication
Document Type
- Article (12)
- conference proceeding (article) (11)
- Book (2)
- conference proceeding (presentation) (2)
- Handout (2)
- Part of a Book (1)
- Doctoral Thesis (1)
- Moving Images (1)
- Report (1)
Publication reviewed
- begutachtet (26)
- nicht begutachtet (7)
Keywords
- Fernwärme (11)
- Lastprognose (11)
- Maschinelles Lernen (11)
- KWK-Flex (10)
- deepDHC (10)
- Biomass (5)
- CCS (4)
- Deep Learning (4)
- Brennstoffzelle (2)
- Machine Learning (2)
- Simulation (2)
- Anlageneinsatzplanungsoptimierung (1)
- CO2 capture (1)
- Compressed Air Energy Storage (1)
- Deep learning (1)
- Digitalisierung (1)
- Dispatch optimisation (1)
- District Heating (1)
- Dynamische Belastung (1)
- Energy Storage (1)
- Fernwärmeversorgung (1)
- Festoxidbrennstoffzelle (1)
- KI (1)
- Kraft-Wärme-Kopplung (1)
- Kraftwerke (1)
- Kraftwerkstechnik (1)
- LNG (1)
- Load forecasting (1)
- NGCC (1)
- ORC (1)
- Organic Rankine Cycles (1)
- Power-to-Heat (1)
- Process Simulation (1)
- Process simulation (1)
- Prozessimulation (1)
- Prozesssimulation (1)
- SOFC (1)
- Temporal Fusion Transformer (1)
- low temperature heat recovery (1)
Institute
Das Wissen um den zukünftigen Wärmebedarf gewinnt, bei komplexer werdenden Fernwärmesystemen mit volatilen und heterogenen Erzeugungsanlagen, immer weiter an Bedeutung. Denn zur Vermeidung von ineffizienten und unwirtschaftlichen Betriebszuständen muss für eine passende Einsatzoptimierung eine möglichst präzise und zuverlässige Wärmelastprognose vorliegen. Maschinelle Lernverfahren, steigende Datengrundlagen und Verfügbarkeit von ausreichender Rechenleistung bergen hierbei erhebliches Verbesserungspotential. Das von der Hochschule für angewandte Wissenschaften Kempten durchgeführte Forschungsvorhaben „DeepDHC - Untersuchung und Weiterentwicklung modernster maschineller Lernverfahren für die hochgenaue Lastprognose in Fernwärmenetzen“ (FKZ: 03EN3017) befasste sich mit der Performance unterschiedlicher maschineller Lernverfahren zur Wärmelastprognose in Fernwärmenetzen. Unter Berücksichtigung unterschiedlicher Fernwärmenetztopologien, der Einbindung von Smart Meter Daten und der automatisierten Berücksichtigung von Veränderungen im Fernwärmenetz wurden hierbei besonders relevante Fragestellungen aufgegriffen und bearbeitet.
Decarbonisation of heat generation has become a priority for district heating network operators. In order to avoid the use of fossil-fired boilers, operators need to know peaks in heat demand in advance. Accurate thermal load forecasting is playing an increasingly important role in this respect. This paper presents the final results of the research project “deepDHC” (deep learning for district heating and cooling) funded by the German Federal Ministry for Economic Affairs and Climate Action (BMWK). The three-year project focused on systematically benchmarking thermal load forecasts for district heating networks, based on state-of-the-art machine learning methods. The analysis covers a variety of machine learning techniques, such as neural networks – including latest deep learning methods – (e.g. LSTM, TFT, ESN, RC), decision trees (random forests, adaptive boosting, XGB) and statistical methods (SARIMAX). In addition, the impact of combining methods by so-called “stacking” was investigated. Training and validation of the machine learning algorithms was based on historical operating data from the district heating network for the city of Ulm in Germany, in combination with historical weather data, and weather forecasts. Thermal load forecasts – typically for three days ahead – are presented and compared against one another. An automatic tuning routine was developed as part of the project, which enables regular re-training of the machine learning algorithms based on the latest operating data from the heating network. Furthermore, a web interface for real-time forecasting was developed and implemented at the power station.
KI-Anwendung in der Energietechnik: Einsatz maschineller Lernverfahren für die Wärmelastprognose
(2021)
Maschinelles Lernen gilt als eines der vielversprechendsten Teilgebiete der künstlichen Intelligenz (KI). Sein Einsatz hat in den vergangenen Jahren zu enormen Fortschritten sowohl in der Bild- und Texterkennung als auch bei Zeitreihenprognosen geführt.
Der Vortrag demonstriert dies am Beispiel von Wärmelastprognosen für die Fernwärmebranche. Dabei werden maschinelle Lernverfahren genutzt, um den Wärmebedarf in Fernwärme-netzen über mehrere Tage im Voraus hochgenau vorherzusagen. Auf diese Weise können Energieversorger den Einsatz ihrer Wärmeerzeugungsanlagen optimal planen. Beispielsweise können bei vorhersehbaren Lastspitzen Wärmespeicher frühzeitig mit erneuerbar erzeugter Wärme gefüllt und so der Betrieb fossiler Spitzenlastkraftwerke vermieden werden. Als Folge kann der Betreiber sowohl CO2-Emissionen als auch Emissions-, Brennstoff- sowie An- und Abfahrkosten einsparen.
Der Vortrag basiert auf Ergebnissen aus mehreren Forschungsprojekten, die an der Hochschule Kempten seit 2016 gemeinsam mit Fernwärmeversorgern durchgeführt wurden. Dabei wurden maschinelle Lernverfahren sehr unterschiedlicher Komplexität systematisch untersucht und bewertet – von „einfachen“ bis hin zu anspruchsvollen Verfahren aus dem Bereich des sogenannten „Deep Learning“. Die Wärmelastprognosen werden unter Verwendung historischer Betriebs- und Wetterdaten sowie von Wetterprognosen vollautomatisiert erstellt und dem Betreiber über eine Web-Schnittstelle zur Verfügung gestellt, die auch im Vortrag gezeigt wird.
Die vorgestellte Methode bietet erhebliche Einsparpotenziale für den Anlagenbetreiber. Sie ist zudem auch auf andere Branchen mit ähnlichen Zielgrößen bzw. Fragestellungen übertragbar.
Keynote auf dem AGFW-Expertenforum "Fernwärme Digital" (18. - 19.04.2023, Frankfurt am Main):
Der Vortrag gibt einen Überblick über die aktuell zum Thema Digitalisierung in der Fernwärmeforschung laufenden Aktivitäten. Zudem werden die Forschungsprojekte deepDHC, KWKflex und HeatSHIFT der Hochschule Kempten zu diesem Thema vorgestellt.
The use of natural gas has continuously increased and reached 24.7% of the worldwide primary energy supply in 2020. The same trend applies to Liquefied Natural Gas (LNG), which contributed to 52% of overall natural gas trades in the same year. In this context, the recovery of the cold energy available at LNG receiving terminals during the process of regasification is of a critical importance.
This paper addresses the integration of the regasification process with an Organic Rankine Cycle (ORC) in order to exploit the available LNG cold energy, by condensing the organic fluid. In addition, a gas turbine exploits differences between regasification and distribution pressures. The analysis covers different organic fluids and two ORC heating source configurations: a) a low-temperature one, using seawater, and b) a high-temperature one, using exhaust gas. In addition, the integration of a natural gas-fired topping gas turbine, which uses the LNG cold energy by compressor inlet air cooling, was simulated. The performance of a medium size regasification terminal (50 kg/s) was evaluated as a function of both the regasification and the natural gas distribution pressures.
Dedicated models have been developed using Aspen Plus software to simulate the regasification process and the integrated topping cycles (Organic Rankine and Brayton), and their mutual energy integrations.
The analysis shows that ORC power outputs from 2 MW up to 4.5 MW in case a) and from 6 MW up to 9 MW in case b) can be reached. The topping gas turbine benefits from the inlet air cooling and can add a power output of 35 MW to 40 MW. R125 was the best working fluid for a low-temperature ORC, while R600a showed the best performance for a high-temperature application.
Temporal Fusion Transformer for thermal load prediction in district heating and cooling networks
(2022)
Accurate forecasting of thermal loads is a critical factor for operating district heating and cooling networks economically,efficiently and with minimized emissions. If thermal loads are known with high accuracy in advance, use of renewable energiescan be maximized, and fossil generation, in particular in peaking units, can be avoided. Machine learning has already provento be an efficient tool for time series forecasting in this context. One recent advancement in machine learning is the "TemporalFusion Transformer" (TFT), which shows especially good results in the area of time series forecasting. This paper examinesthe performance of TFT in the concrete context of thermal load forecasting for district heating and cooling networks. First,a brief summary of differences between TFT and other machine learning methods is given. Secondly, it is described how themethod can be adopted to train a machine learning model for thermal load forecasting. The data to train and evaluate the neuralnetwork is based on 8 years of hourly operating data made available from the district heating network of the city of Ulm inGermany. The presented technique is used to produce 72 hours of heating load forecasts for three different district heating gridsin the city of Ulm. The results are compared to forecasts of other machine learning methods that have been previously madeas part of the publicly funded research project "deepDHC", in order to evaluate if TFT is an improvement to further reduceforecasting uncertainties.
Efficient operation of district heating networks requires a precise forecasting of the thermal loads and an optimised dispatch strategy for the available generation and storage portfolio. This paper presents a holistic modelling and optimisation approach: first, detailed process modelling and optimisation of power plants and thermal storages; second, a numerical model for dispatch optimisation; and third, machine-learning-based load forecasting. The work is based on operating data from the district heating network of the city of Ulm in Germany. The paper presents the modelling, validation and simulation results of stationary and instationary process simulation for a biomass-fired combined heat and power plant. The analysis identifies a potential to integrate additional renewable power by “power-to-heat” technologies into different parts of the process. The economic benefit is quantified by mixed-integer linear programming optimisation applied to the district heating network. In order to allow for real-time dispatch optimisation, a machine-learning-based thermal load forecasting method was developed and evaluated, based on a 72-h forecast horizon. In addition, the economic impact of prediction uncertainties is analysed with the numerical dispatch optimisation tool.
Precise forecasting of thermal loads is a critical factor for economic and efficient operation of district heating and cooling networks. If thermal loads are known with high accuracy in advance, use of renewable energies can be maximized, and – in combination with thermal storage units – fossil generation, in particular in peaking units, can be avoided. Machine learning has proven to be a powerful tool for time series forecasting, and has demonstrated significant advancements in recent years. This paper presents the scientific methodology and first results of the publicly funded research project “deepDHC”, which aims at a broad benchmarking of traditional and advanced machine learning methods for thermal load forecasting in district heating and cooling applications. The analysis covers autoregressive forecasting approaches, decision trees such as “adaptive boosting”, but also latest “deep learning” techniques such as the “long short-term memory” (LSTM) neural network. This work is based on data from the district heating network of the city of Ulm in Germany. First, different performance metrics for evaluating forecasting qualities are introduced. Second, approaches for data screening and results of a linear and non-linear correlation analysis are presented. Third, the machine learning tuning process is described. For thermal load forecasting, weather data are key input parameters. This work uses hourly weather forecasts from weather models provided by the German meteorological service. These weather data are updated automatically, and have been statistically corrected in order to represent very accurate point forecasts for up to ten days ahead. In addition, a user-friendly web interface has been developed for use by the district heating network operator. The performance of different machine-learning algorithms is compared based on 72 h heating load forecasts.
Decarbonisation of district heating networks requires a heat generation and storage portfolio that allows for maximised integration of renewable energies. However, it is also essential to have a precise forecast of the thermal loads to be expected in the network and – based on this forecast – an optimised dispatch strategy in order to best match the available generation and storage portfolio with the actual heat demands in the grid. This paper presents a holistic approach that combines modelling and optimisation activities related to these three aspects: first, detailed process modelling and optimisation of power plants and thermal storages; second, a numerical model for dispatch optimisation; and third, machine-learning-based load forecasting. This work, which was performed as part of the publicly funded research projects “KWKflex” and “deepDHC”, was based on operating data from the district heating network of the city of Ulm in Germany. The paper presents the modelling, validation and simulation results of a stationary and instationary process simulation for a 58 MW thermal biomass-fired combined heat and power plant. The analysis identifies a potential to integrate additional renewable power of up to 17 MW thermal power by “power-to-heat” technologies into different parts of the process. The economic benefit is quantified with a mixed-integer linear programming dispatch optimisation model of the district heating network. In order to allow for real-time optimisation of the power plant and thermal energy storage dispatch, a machine-learning-based thermal load forecasting method was developed. The performance of different machine learning algorithms, including decision trees and deep learning techniques, is compared based on a 72-hour forecast horizon. In addition, the economic impact of uncertainties in thermal load prediction is analysed with the numerical dispatch optimisation tool.
This working paper evaluates cost and performance trends related to carbon dioxide (CO2) capture from power generation, based on extensive analysis of data from major engineering studies published between 2006 and 2010. Since individual studies use different methodologies and boundary conditions, study estimates for over 50 CO2 capture installations are re-evaluated on a consistent basis and updated to current cost levels.
The paper discusses the need for further standardisation of evaluation methodologies and additional data for specific CO2 capture routes. Further analysis for non-OECD countries is considered crucial for global energy scenario models, and for improving the skills and knowledge developing countries need to evaluate the role of CCS in their national energy contexts.
Electricity generation from coal is still growing rapidly and energy scenarios from the IEA expect a possible increase from today's 1 600 GW of coal-fired power plants to over 2 600 GW until 2035. This trend will increase the lock-in of carbon intensive electricity sources, while IEA assessments show that two-thirds of total abatement from all sectors should come from the power sector alone to support a least-cost abatement strategy. Since coal-fired power plants have a fairly long lifetime, and in order to meet climate constraints, there is a need either to apply CCS retrofit to some of today's installed coal-fired power plants once the technology becomes available. Another option would be to retire some plants before the end of their lifetime. This working paper discusses criteria relevant to differentiating between the technical, cost-effective and realistic potential for CCS retrofit. The paper then discusses today's coal-fired power plant fleet from a statistical perspective, by looking at age, size and the expected performance of today's plant across several countries. The working paper also highlights the growing demand for applying CCS retrofitting to the coal-fired power plant fleet of the future.
In doing so this paper aims at emphasising the need for policy makers, innovators and power plant operators to quickly complete the development of the CCS technology and to identify key countries where retrofit applications will have the biggest extent and impact.
Energy scenarios suggest that CO2 capture and storage (CCS) from power plants might contribute significantly to global greenhouse gas emission reduction. Since CCS from power generation is an emerging technology that has not been demonstrated on a commercial scale, related cost and performance information is still uncertain. This paper presents a detailed analysis of the impact of adding CO2 capture and compression process equipment to fossil-fuelled power plants. For coal-fired power generation, no single capture technology outperforms available alternative capture processes in terms of cost and performance.
Past studies have used power plant dispatch models with fixed plant portfolios to demonstrate that flexible capture systems may allow operators to increase profits in current electricity markets by increasing plant output at times of high demand. Some of these studies have also speculated that flexible capture systems may be valuable in future electricity systems with high shares of renewable energy as they could allow fossil fuel plants with capture to rapidly respond to changes in residual load. However, few studies have actually examined the role that plants with flexible capture could play in future power systems. Thus, this study examines the role that power generation with flexible capture systems could play in a future European power system where 80% of generation (by energy) is supplied by renewables. The results show that conventional base- and mid-load capacity decreases while the peak-load capacity (i.e., open cycle gas turbines) increases. In European regions with high shares of renewables, the residual load duration curve become steeper, and hourly changes in residual load increase significantly and happen more frequently at low load levels. The shift towards peak capacity and general decrease in load factors places technologies with high capital costs at a relative disadvantage. In the scenario in which CO2 prices reach 650 per tonne CO2 in 2050, approximately one-fifth of the CCS capacity deployed is equipped with flexible capture. However, in a scenario in which CO2 prices reach € 100 per tonne CO2 in 2050, very little capacity is equipped with flexible capture systems as the cost of emitting CO2 offsets the value of flexibility.
This paper presents a methodology to estimate the biomass energy potential and its associated uncertainty at a country level when quality and availability of data are limited. The current biomass energy potential in Colombia is assessed following the proposed methodology and results are compared to existing assessment studies.
The proposed methodology is a bottom-up resource-focused approach with statistical analysis that uses a Monte Carlo algorithm to stochastically estimate the theoretical and the technical biomass energy potential. The paper also includes a proposed approach to quantify uncertainty combining a probabilistic propagation of uncertainty, a sensitivity analysis and a set of disaggregated sub-models to estimate reliability of predictions and reduce the associated uncertainty. Results predict a theoretical energy potential of 0.744 EJ and a technical potential of 0.059 EJ in 2010, which might account for 1.2% of the annual primary energy production (4.93 EJ).
Methodology for biomass energy potential estimation: Projections of future potential in Colombia
(2014)
This paper presents a novel method to estimate the future biomass energy potential in countries with domestic markets unable to influence international markets. As a study case, the biomass energy potential in Colombia is estimated for the period 2010–2030.
The prediction model is a scenario-based optimization algorithm that maximizes the yearly profit of locally producing and importing commodities in a country subject to certain constraints (domestic demand, limited area, etc.) as well as to demographic, macroeconomic and market data (e.g. domestic and international prices of commodities). The bioenergy potential associated to the production of commodities is calculated according to a methodology presented by the same authors. In order to provide a modeling framework consistent with other state-of-the-art projections, global scenarios for analysis are selected from the literature rather than formulated. Selected global scenarios highlight the influence of global biofuel use on agricultural prices, production and demand.
Results predict a theoretical bioenergy potential in Colombia 56%–69% larger in 2030 than in 2010 (1.31–1.41 EJ). A sensitivity analysis shows that while a higher global biofuel use leads to a higher local bioenergy potential, its influence is less pronounced than that of agricultural yields, demand and specific energy of biomass resources.
This paper presents a novel approach to address uncertainty and improve reliability of the estimation of the biomass energy potential at a country level, particularly suitable for situations when quality and availability of data are limited. The proposed methodology improves the prediction reliability by following four steps: 1) using a simple accounting framework, 2) using a robust selection of probability density functions, 3) using a probabilistic propagation of uncertainty and 4) using sensitivity analysis to identify key variables contributing to uncertainty as well as a root cause analysis and a set of sub-models to improve estimation of key variables.
The application of the methodology to the energy scenario in Colombia shows that the improved estimation of the theoretical energy potential has an almost identical mean value compared to the preliminary estimate, but the uncertainty is significantly lower (less than 50%). Moreover, the mean value of the technical energy potential obtained through the methodology is 25% lower than the preliminary potential and the uncertainty reduces by one third.
This paper presents a modeling framework to address the energy, economy, emissions and land use nexus when exploiting bioenergy in developing countries. The modeling framework combines a qualitative and a quantitative element. The qualitative element integrates two components: (1) technology roadmapping to identify long-term technology targets through expert judgment and (2) scenario analysis to investigate different future storylines. The quantitative element comprises four integrated tools, namely the energy system model (ESM), the land use and trade model (LUTM), an economic model, and an external climate model. An overview of the modeling framework, scenario analysis, structure of the models, modeling techniques, mathematical formulations and assumptions is presented and discussed. The modeling framework is applied to the particular context of Colombia, as a case study of a developing country with large bioenergy potential. In this study case, the impacts that an accelerated deployment of bioenergy technologies might cause on the energy demand and supply, emissions and land use until 2030 are evaluated. Results suggest that a plan to exploit bioenergy in Colombia should prioritize the deployment of technologies for biomethane production, power generation & CHP, which can reduce more GHG emissions and more emissions per incremental hectare of land than first-generation biofuels. Moreover, while the share of bioenergy in the primary energy demand decreases in all the analyzed scenarios, it is possible to envision significant increases in the share of bioenergy in road transport energy demand, power generation and natural gas supply for scenarios implementing roadmap goals. In addition, impacts of El Niño oscillation on the dependence of hydro for power generation can be partly mitigated by exploiting the complementarity of hydro and bioenergy, which might result in a reduction of up to 5–6% in the demand for fossil fuels used in power generation in dry years. However, despite the ambitious goals proposed here, bioenergy alone cannot significantly reduce emissions by 2030 (maximum 10% reduction relative to baseline) and effective climate change mitigation requires a portfolio of additional measures.
The growing concern about the role of man-made CO2 emissions with respect to global arming, in combination with the large increase in energy demand spurred by developing nations and a growing global population that is foreseen over the next 15 years have recently turned attention to potential CO2-neutral energy supply solutions.
Grid-compatible integration of typically fluctuating electrical energy sources, like wind and solar power, will be important in order to support the goal to reduce CO2 emissions. However, this will require substantial adjustments to the grids and power plant systems in order to cope with the upcoming new boundary conditions imposed by substantially increased utilization of renewable energies.
To respond to this imperative, GE and RWE Power have started to investigate new technologies for large-scale storage of electrical energy in Adiabatic Compressed Air Energy Storage power plants.
This concept offers efficient, local zero-emission storage based on compressed air held in underground caverns. The compression and expansion of air with turbomachinery help to balance power generation peaks that are not demand-driven on the one hand and consumption-induced load peaks on the other, allowing the optimal use of both traditional fossil fuels and renewables.
Before this concept can be implemented, however, numerous technical issues must be addressed, mainly in the field of turbomachinery and the heat storage device. This paper describes today’s technical capabilities of turbomachinery equipment, and evaluates the need for further development based on the requirements of advanced CAES technology. Ongoing development activities are described and initial results presented.