Florian Hesselink

Project title: Uncertainty in heating transition modelling
Mathematical modelling of energy systems is used more and more often to inform policy making for the energy transition. One such model, developed by CE Delft, is CEGOIA. Dutch governments use it to create strategies for the energy transition in the built environment with the aim of completely phasing out the use of natural gas for heating purposes by 2050. Integrating local data, this model assesses the costs for refurbishing building insulation and switching to sustainable heating methods of residential and utility buildings. Models like CEGOIA necessarily have to make many (>1000) assumptions regarding economic, technological and other factors. As models become bigger and more complex, these assumptions can become problematic. Modellers should always be able to interpret and contextualize results in order to translate them to robust advice. We can’t do this if the relations between inputs and outputs are uncertain. In my research I will use CEGOIA as a case study to assess the way in which this uncertainty manifests and translates into model-based policy advice.

About me: I am curious and passionate about many things and always want to know exactly how something works, which is probably why I ended up focussing on sustainability in the field of systems engineering. In my free time I read a lot and write occasionally. During this pandemic I’ve been missing being able to travel, which I not only enjoy for the people and sights but particularly for the cuisine, seeing as that I would like to think I am a great chef. To keep active I enjoy cycling and swimming.

Committee: Emile Chappin, Els van Daalen, Nina Voulis

Florian Hesselink

MSc programme: Complex Systems Engineering and Management (Energy Track)

Research resume

The Dutch heating transition involves changing the heating systems of eight million buildings to a sustainable alternative by 2050. Many heating system technologies are available, but deciding which systems are cheapest for all these buildings is a difficult question to answer. Local policymakers are increasingly making use of heating transition models that estimate the feasibility and costs of systems in municipal neighbourhoods. The applicability of these models is limited by the degree of uncertainty about the future as well as the complexity in communicating the model results to policymakers. Sensitivity Analysis (SA) is a tool with which the most influential model uncertainties can be identified, quantified and communicated. So far, limited energy transition model studies have extensively used this method. A case study of SA on the CEGOIA heating transition model was performed to fill this gap and evaluate SA’s value.

CEGOIA calculates the costs of a variety of heating systems and optimizes the allocation of scarce energy carriers such as green gas and hydrogen to find the lowest societal costs. Sensitivities of eight heating system options were analysed in different archetypical neighbourhood contexts using Fractional Factorial analysis, the Method of Morris and the Sobol’ Method. Out of an initial set of 953 parameters, a subset of less than a dozen highly influential variables – consistent between neighbourhoods of different physical characteristics – was identified for each heating system option. High sensitivities indicate that changing the value of a parameter leads to a large change in total costs. These sets, therefore, describe exactly what uncertainties are crucial to evaluating what heating system is the cheapest possible solution. Variables in these sets include, but are not limited to, the price and infrastructure costs of electricity and gas, heating installation costs and insulation costs. Interviews with other heating transition model owners further illustrated that the use of systematic SA as done in this analysis is not the norm.

Besides results and insights from the CEGOIA SA, further applications for SA in heating transition modelling is postulated to be able to improve the modelling process, as well as better, understand complex model dynamics. One recommendation is, therefore, to include SA as part of the toolkit for the large heating transition models currently being used in the Netherlands. The main barrier for doing so with CEGOIA is the computational time of the model, which limited the number of parameters that could be evaluated as well as the SA techniques that could be used. Still, a more systematic analysis of sensitivities in heating transition models will provide insights that ultimately aid Dutch policymakers in making robust decisions.