with our final newsletter for 2022 we are again sharing a short overview of recently published papers about district heating which have caught our attention in the last few weeks. Like for the entire year, we saw a large number of interesting papers being published on a wide range of topics. From these, we picked papers with a focus on topology optimization and network dimensioning, as well as an interesting method to find patterns in anomaly datasets.
Published just a few days ago, Wack et al. describe an Economic topology optimization of District Heating Networks using a pipe penalization approach. This work builds on a previous paper by the same authors on topology optimization. But with the current paper, the authors have significantly improved the economic evaluation as a crucial part of the optimization. They argue that because the high initial investment is the major challenge for district heating, a more detailed economic evaluation in the design phase can help to lead more projects to success. In addition, the authors combine this economic perspective with a non-linear heat transport model in an effort to find a balance between physical accuracy and scalability of their optimization model.
To reach a fast solution (10 minutes optimization for a network with 160 connected buildings), the method starts with a continuous function and then pushes towards a discrete solution for each pipe's diameter with a penalization approach. This approach is demonstrated in the paper for a network supplying the aforementioned 160 buildings with heat from 2 potential heat sources. In this setup, the results demonstrate how different heat generation costs and temperature levels of the 2 heat sources influence the optimal network topology solution. Thus, the authors present an interesting method to optimize the net present value of a district heating project with an optimal network topology.
Our second paper selected for this newsletter also addresses district heating network design challenges, but with a focus on the hydraulic network operation and using a detailed thermo-hydraulic simulation model. In the paper Dimensioning of low-temperature district heating grids with geothermal heat sources using detailed simulation software SIM-VICUS, Hirsch et al. present their simulation tool SIM-VICUS and demonstrate how it can be used to evaluate different pump control strategies.
As a demonstration use case, the authors present the ultra-low temperature network in Bad Nauheim, which is supplied by a large horizontal ground heat exchanger and thus operates at network temperatures between -5 and 20 °C. To work with these network temperatures, the buildings are equipped with heat pumps which use the network as a low temperature heat source. But the main focus of the paper is an evaluation of the electricity needed to operate the central network pump to provide a sufficient pressure head for all buildings. For that, the authors show that compared to a constant supply pressure, a worst-point pressure control can save around 50 % of pump electricity.
Moving from optimization and simulation models towards data-driven methods we wanted to include the paper Pattern detection in abnormal district heating data by Mbiydzenyuy and Sundell in this issue of our newsletter, even though it is still a pre-print version. The first remarkable thing about the paper in our opinion is that the paper starts with understanding district heating as a good example of an industrial scale application with large amounts of sensor data. While this is certainly not (yet) the case for every district heating network, that's a clear sign that we're moving towards greater amounts of data being available for analysis. And to make best use of this data, fault detection and diagnostics is a promising field which other publications have explored already.
Yet, we think that this current paper is addressing a specific challenge when working with such data as we will probably see more and more in district heating applications: There may be measurement datasets including faults and abnormal operation, but there is a general lack of labeled fault data. This means that unsupervised anomaly detection is needed, which can identify faults in the measurement data. But as these faults are not labeled, it would still require manual effort to know *what* is wrong with a given dataset. To mitigate this problem, the authors present an approach to find patterns in such anomaly data and thus suggest likely causes for anomalies automatically even if only one or very few instances of labeled fault data points are available.
In addition to these papers, Garay-Martinez et al. recently published a new Handbook of Low Temperature District Heating with several interesting contributions. And even though it is not strictly focused on district heating, we found that the paper Accurate metering and billing of ambient loop systems adds an interesting perspective on measuring heat balancing effects like we can also find in 5GDHC networks.
And on a final note, we saw that two recent papers (here and here) took their motivation to develop new district heating models from concerns about the performance of existing Modelica models in the Modelica IBPSA Library. To contribute to that discussion, our experience at heatbeat is different: While it can be challenging to use those models efficiently, we found these models to be a great foundation to develop efficient workflows and system models also for large scale applications with hundreds of connected buildings.
Our next newsletter will be published January 4, 2023. Until then we wish you happy holidays and a great start to the new year!
© heatbeat engineering GmbH & heatbeat nrw GmbH 2017-