ICEC 2021 poster presentation abstracts 

Ardnt, Michael

Technische Hochschule Mittelhessen, Germany

Static and dynamic setpoint optimization in residential heat pumps based on learned user behaviour

P.-R. Adam, S. Erkel, M. Arndt

Commercial residential heat pumps mainly work on static, manually adaptable setpoints for heating and hot water generation. Often, these setpoints remain at the standard factory settings. Thus, there is potential to reduce the energy consumption by individualizing and dynamizing these setpoints.
In this paper we describe the development of two methods to adapt the hot water temperature setpoint statically and dynamically by learning the user behavior and the installation conditions. The methods were then evaluated by system simulations. The optimal hot water temperature setpoint depends on the building type, the hot water demand and the inlet temperature to the heat pump. Therefore, in our approach, the heat pump will first run with various settings for a limited learning period. From the collected data, the resulting energy consumption for each hot water heat up cycle was calculated. Thus, at the end of the learning period, a dataset existed, which was used to find the optimal hot water temperature setpoint under given circumstances.
Method 1 calculates, based on this dataset, one individual optimal hot water temperature setpoint for one specific heat pump installation and user profile. This new static setpoint was then used in the simulation.
Method 2 generates categories (e.g. periods like day, night, working day or weekend) and calculates an optimal setpoint for each of these categories. Using a schedule, these setpoints were then used dynamically in the simulation. A variation of method 2 uses an additional demand prediction algorithm, to further improve the energy efficiency.
The energy reduction potentials for method 1 and 2 were finally quantified using a building simulation system which includes a TRNSYS Building model (Type 56), randomized and normed (DIN EN 16147) user hot water consumption and a reference model of the hot water tank and the heat pump. The simulation results show an energy reduction potential of 3.5% for method 1 and at least 6.5% for method 2 compared to the factory settings. The presented concept could reduce the carbon footprint of heat pumps, as it can be implemented by adaptation of the control software of existing residential heat pumps.

Garner, Robert

Brunel University, UK

Solar PV and hybrid energy storage Virtual Power Plant for smart energy communities: an optimal parallel services concept

R. Garner, G. Jansen

Renewable Distributed Energy Resources (DER) are becoming more popular as governing bodies introduce ambitious climate policy objectives to curb emissions production. The Virtual Power Plant (VPP) concept can be used to increase the visibility of DER and provide grid support services to the grid, as well as promote the continual increase in renewable energy usage and additional revenue to participants. In this model, a prosumer Energy Community is constructed consisting of surplus PV solar generation, flexible loads, and a hybrid battery and Hydrogen Fuel Cell (HFC) Energy Storage System (ESS). A non-linear optimisation strategy based on a Genetic Algorithm is used to find the optimal day ahead energy dispatch to ensure optimal revenue for each participant within the VPP. Excess energy storage provides internal peer-to-peer remuneration as well as peak demand management and frequency control services to the grid. Hybridising battery and HFC provides optimal synergy between the technologies, making use of the strength of each technology by allowing the battery to provide short duration storage and fast response for frequency control, while the HFC can provide long duration storage, and improve the overall resilience and economic performance of the ESS. The hybrid storage is sized to reduce the Levelised Cost of Electricity (LCOE) by finding the optimal battery and HFC capacity ratio. Specific social, regulatory, and technical barriers to implementation are presented and discussed in relation to the proposed VPP concept. Further research of the system’s economic performance and potential business models in combination with this work could prove the commercial feasibility of the Energy Communities concept.

Grisiute, Ayda

Singapore-ETH Centre, Singapore


Creating multi-domain urban planning indicators using a knowledge graph: a district energy use case in Singapore

A. Grisiute (SEC), H. Silvennoinen (SEC), Z. Shi (SEC), A. Chadzynski (CARES), S. Li (SEC), M. Q. Lim (CARES), F. Sielker (CARES), A. von Richthofen (SEC), P. Herthogs (SEC), S. Cairns (SEC), M. Kraft (CARES)

In Singapore, decision-makers from multiple ministries and government agencies participate in urban planning and management. This often results in siloed datasets, different data formats and domain specific software: a lack of interoperability that hinders the integration of (big) data in planning. Semantic Web Technology (SWT) can help to solve data interoperability issues. With SWT, computers can infer semantic relationships between heterogeneous data that are linked using ontologies (i.e. ‘common languages’). Knowledge Graph (KG) data structures allow such linking, and thus support SWT applications. The Cities Knowledge Graph (CKG) research project uses a KG to facilitate the use of multi-domain data in city planning. We present a use case that demonstrates how KGs enable the creation of planning indicators, building on various openly available datasets in Singapore. The first step was to transform datasets containing geospatial and regulatory data on zoning, parcels and buildings to CityGML. Then it was loaded into a KG structured using a CityGML-based ontology. Retrieving raw data values or composite metrics (essentially manifold combinations of queries on different datasets) was done using SPARQL. However, our goal was to develop indicators that could benefit urban planners – examples include ‘GPR potential’ (unused Gross Plot Ratio per zone) or ‘allowable programmes per plot’ (which uses could exist on a plot, given its zone). These and other indicators have many potential applications, including in urban energy modelling. For instance, we developed an indicator for district cooling potential based on geometric parcel data as well as zoning and density data. We demonstrated how these indicators can be used to analyse a part of downtown Singapore, and visualised the results. We show that KG technology allows planners to analyse cities through multi-domain indicators, which would be difficult to develop based on individual datasets. Another benefit of KGs is highly malleable data architecture, allowing data to be updated, expanded or created. Future work includes integrating new datasets into our CKG and developing new analyses. Ultimately, these could be carried out autonomously by a multi-agent system.

Hjalmarsson, Johannes

Uppsala University, Sweeden


Large Scale Energy Storage in Uppsala, Sweden: an analysis of voltage fluctuations and a service stacked portfolio

J. Hjalmarsson, K. Thomas, C. Boström, F. Carlsson, A. Berlin

1. Subject, motivation and objectives
Extensive electrification of several sectors of the society has created an increased demand of electricity, considering both energy and power. As a result, a handful of regions in Sweden indicate a critical lack of distribution capacity to larger urban areas. One of these regions is Uppsala, where the DSO Vattenfall has decided to connect a 5MW/20MWh Li-ion battery energy storage system (BESS) to temporarily ease the congestion issue. Additional applications of high relevance are flexibility services and frequency support to ensure safe and efficient distribution of high-quality electric power. As of the connection of the BESS, it is also of great interest and importance to investigate how the power quality in the area will be affected by the operation of the BESS. For this study, the voltage stability of the grid and variations over time would be of most particular interest. Furthermore, an optimal operation strategy of the battery would also be of interest by combining the services of interest.
2. Method and approach
By measuring the local voltage level over time, it is possible to indicate how the power quality is affected by the operation of the BESS. The local grid voltage level will be measured before, during and after service provision of the BESS. Further, by considering a smart planning of the BESS operation it enables the feature of combining two or more services in parallel to increase the benefit of the energy storage unit during and between seasons.
3. Results and conclusions
Results from the measurements indicate very small or no voltage fluctuations during neither flexibility nor frequency response service provision. The next step in the verification process is to test more stressful services like fast frequency response (FFR) which requires approx. 30 times shorter activation time of the storage. Also, it is also of interest to further investigate and implement service stacking in an intra-day and intra-season timescale. Energy storage in distribution grids will certainly become an important tool for distribution system operators as the market evolves even further.

Hofmeister, Markus

University of Cambridge, UK


Sustainable operation of district heating systems using dynamic hierarchical optimisation

M. Hofmeister

District heating is expected to play an essential role in the cost-effective decarbonisation strategy of many countries. However, resource-optimised management of district heating networks needs to consider a wide range of factors, including demand forecasting, flexibility of the heat provision mix, and volatile market conditions. While traditional approaches often rely on static models and rather simple heuristics, dynamic cross-domain interoperability that allows the consideration of all these factors is essential to holistically optimise thermal grid operations. A hierarchical optimisation approach based on the merit-order principle is developed and embedded in a model predictive control framework to allow the system to incorporate most recent information and react to disturbances promptly. Solving the unit commitment and load dispatch problem using a heat merit-order stimulates the inclusion of (industrial) waste heat and intermittent renewable energy sources by increasing transparency on marginal generation cost from different sources. Simulation-based optimisation is used to determine the short-term heat generation mix based on data-driven gas consumption models and day-ahead forecasts for the energy demand and grid temperatures. Seasonal autoregressive integrated moving average models with exogenous predictor variables (SARIMAX) are found to be sufficiently accurate and precise. The forecasting errors of the best-fitting SARIMAX models are shown to have no decisive influence on the generation optimisation results when compared to actual historical data. A detailed sensitivity study is conducted to identify key design criteria and input parameters and assess the impact of anticipated changes in regulation and market conditions. The effectiveness of the approach is demonstrated for an existing heating network of a midsize city in Germany, where a reduction of approximately 20% and 40% compared to baseline operational data is obtained for operating cost and CO2 emissions, respectively. Scrutinising real industry data has revealed several severe data quality issues and emphasises the importance of a semantic representation to foster cross-domain interoperability between heterogeneous data and allow for automatable artificial intelligence applications. This work demonstrates early progress towards a knowledge graph-based approach to heat generation management within a broader semantic digital ecosystem for municipal utilities.

Sielker, Franziska

Cambridge University, UK


Can different urban information systems speak to one another? – An innovative use case of the Knowledge Graph technology for Singapore’s energy planning

H. Quek, F. Sielker

As governments and cities aspire to be ‘smarter‘ and ‘more digital‘, the knowledge graph technology aims to advance these aspirations for optimal planning outcomes by solving the lack of interoperability between different urban information systems such as Building Information Management and Geographic Information Systems. This poster presents the results of the ‘Consumer Energy Usage Data in Smart City Development‘ funded by the Singapore National Research Foundation, and provide evidence of the opportunities to use knowledge graph technology to link different data systems.
In an era driven by digital innovation, technological tools, like cloud computing and big data, offer cities new possibilities in tackling the complex environmental and socio-economic challenges of today. Despite the potential of digitalisation, governments at all levels across the globe are not prepared to fully use the opportunities of digital developments into their planning systems. Considering the diversity of urban information systems available, interoperability and level of detail remains the key obstacle, resulting in barriers such as lack of universal standards, coordination, and expertise. Moreover, different levels in the planning system requires different detail and scope of data. For example, energy planning requires the geo-spatial coordinates for the power grids at a national level and consumer switch rooms at a neighbourhood level. Thus, the existing siloed urban management approaches impede cities and planning systems from capitalising on the latest technological developments to encourage good data practices and sharing, while providing more efficient higher quality services. This poster showcases the use of knowledge graph technology as a solution to the interoperability issue through ‘The World Avatar’ project in Singapore. The knowledge graph represents a network of interlinked descriptions for real-world entities – objects, events, or concepts. We present how the knowledge graph can be used as a one-stop solution platform to correspond between various urban information systems through the case of energy use in Singapore. Specifically, this poster demonstrates the proof-of-concept when ‘The World Avatar’ project is integrated with three planning technologies – Building Information Management, Common Information Model and Geographic Information Systems – to address planning issues.

Shariq, Muhammad Hasan

University of Hull, UK


Automatic building inspection techniques through drones and transfer learning on fused thermal and colour imagery of building facades

H. Shariq

The building sector is responsible for 40% of overall global energy consumption. Building defects such as heat losses, moisture, and air leakages account for inefficient space heating and cooling that results in excessive energy consumption and the associated greenhouse gas emissions. Building inspections and energy surveys are performed to detect and diagnose these building defects, however, recent inspection methods have been highly labour intensive, time consuming, and not suitable for large-scale audits. The research and development in this paper aims to automate building inspection techniques through artificial intelligence (AI). A drone-based low-powered embedded machine learning (ML) system is proposed and implemented that utilises pre-trained model via transferring learning (TL) on fused thermal and colour imagery of façades to detect and identify faults. The experiment sets were performed and compared on heat losses around windows through (a) only thermal imagery and (b) on fused thermal and colour data set. Results showed a 20% higher accuracy of successful fault detection on the fused data set compared to applying ML on thermal data alone. The use of transfer learning in building façade survey allowed the ML model to be run on a low-power embedded microcontroller, which provide massive future scope for an autonomous drone-based building survey at large-scale with automated and real-time fault detection on building façades.

Walker, Johanna

University of Southampton, UK


Co-creating Smart City Solutions in Low Data Maturity Cities

J. Walker

The Smart City Innovation Framework Implementation is a framework to enable medium sized cities at the beginning of their open data/smart city journey to engage with the private sector to pilot new products and services in the energy, mobility and environment spaces. This framework supports cities to activate the commercial market and test outcomes, in a low risk, low investment environment. It was implemented in 2 rounds over 3 years in 4 cities in Northern Europe.
The implementations followed the below process:
Business cases (25) > Challenges published to commercial market (16) SMEs selected to co-create solutions (14) > Pilots developed > Successful procurements (4)
The process yielded a 20% rate of successful procurements, despite low levels of data and technology within the city governments. The sustainable smart watering solution implemented in Saint Quentin has won 2 national awards.
Some key findings: By co-creating rather than outsourcing or procuring, cities were able to develop their technical knowledge in a sustainable manner. Although it was anticipated that all data used would be open data, in reality risks of triangulation and the creation of new personal data were present, and required management. Integrating the solutions into existing deployments often had unforeseen challenges. A waste collection algorithm, while optimal in terms of reducing overfull bins and unnecessary trips, was difficult for the collection company, a social enterprise, to use with their employees. Cities found some emergent areas challenging in terms of setting up appropriate contracts, intellectual property rights and confidentiality protection, especially around sensor data. Cities can be aware of best practices of open data, but still have issues complying, and little idea of the pitfalls when diverging. Some legislative aspects of open data, such as mandated data set opening, can be onerous. Non-sector specific) non-legislative measures, such as awareness-raising and the sharing of data-driven innovation best practice, are an appropriate policy solution. There is a need for a kind of ‘virtual intermediary,’ which cities could consult for guidance, rather than withdrawing from challenging situations.

Zou, Jasmine

University of Victoria, Canada


A Systems Approach to More Resilient Data Management for Municipalities

J. Zou

Throughout the last year, city leaders have increasingly been focused on ensuring the safety of their citizens and adapting city planning functions to support the delivery of critical services during unprecedented circumstances. To support operations and data-driven decision making during this time, city staff have increasingly relied on data collection, management, and producing insights from data. The process of which is easier said than done, requiring significant infrastructure for data collection, data governance and processing expertise, and the tools necessary to produce insights from this information. To this extent, this poster submission will highlight best practices and outcomes from UrbanLogiq’s work supporting City Halls across North America to facilitate more streamlined data management and analytics functions to inform decision making and policy.
Poster contents will detail applications of UrbanLogiq’s technology: our Data Platform, a feature-rich data lake that leverages multiple regional, local, and private data streams for more informed city services. This includes illustrating our work with municipal agencies such as the City of San José, California, the City & County of Honolulu, Hawaii, and the City of Ottawa, Ontario, to institute data management best practices and derive outcome-oriented insights. Highlights will focus on tangible quick wins that municipal agencies can implement to collect data that’s easier to consume, processes for transforming data, and what city officials should be considering when scaling data collection and analysis functions. In addition, the presentation will showcase outcomes from recent projects, such as a data analytics project completed with the City of San José Department of Transportation, which used machine learning to project road safety patterns over a 5-year period so as to direct city planning resources more efficiently and effectively to save lives.