Power system planning is a strenuous exercise for both people and computers. As carbon goals and geopolit- ical complexities intensify, the number of scenarios for electricity system development is limitless. In search of robust policy and profitable investments, policymakers and investment managers rely on long-term elec- tricity system simulation models. Those seeking to simulate myopic investment behaviour in systems reliant on intermittent generation, batteries and seasonal storage then encounter a computational barrier. The sole model detailed enough to conduct such simulations – MIDO (Myopic Investment Detailed Operational) – takes over 12 hours to run a 20-year scenario. Exploring avenues to cut back computational time is key to un- locking the potential to meaningful scenario analyses. Also, due to its novelty, many of its assumptions and modeling choices were not validated with stakeholders. Research into methodologies to reduce the computational burden at an acceptable compromise to model behaviour thus holds merit.

To this end, this methodological thesis explores myopic optimisation, developing the MODO methodology: Myopic Optimisation, with Detailed Operational analysis of system performance. We have realised an 85% improvement in computational efficiency with a sufficient degree of accuracy for it to be considered as an alternative to MIDO. The essence of our approach is the replacement of an iterative investment loop with a single optimisation. In doing so, we transfer insights from the field of agent-based modeling into the field of myopic optimisation, potentially enabling the revitalisation of a research area which has seen little activity in recent years.

MODO has been implemented as a modular toolkit in Linny-R, in such a way that users need not have any technical modeling knowledge to be able to construct their own system and/ or analyse their own scenarios. The implementation in Linny-R greatly enhances model transparency and usability.

We create a projection for the state of the Dutch electricity system by 2030 and use MODO to simulate the transition path from 2030-2050. Preliminary results suggest the significant potential of both seasonal (hy- drogen) and battery (lithium-ion) storage to lower electricity prices below 2030 levels, bringing down annual consumer spending to EUR 15 bn annually in a scenario with mid-range commodity price forecasts. Addi- tionally, results affirm the functionality of MODO in more complex electricity systems.

We identify three exciting avenues for future research. The risk of contemporary investment decisions is largely mitigated through long-term contracting, such as bilateral power purchase agreements or forward ENDEX markets. Ideally, academic work would converge to the point where a first model simulates a range of potential portfolio choices based on long-term contracting, after which - based on risk appetite - the upside potential of offering excess capacity on spot markets is assessed with a methodology such as MODO. However, as the data on such long-term contracting is often kept secret, research into ways to approximate the data or work with it using ’black-box’ models could prove fruitful. Secondly, researchers could choose to build upon this work by leveraging it to conduct large-scale exploratory analyses. The case studies in this work illustrate the potential of MODO, but only provide a preliminary insight into system development. Whilst model devel- opment has been informed by a stakeholder consultation, the scenarios analysed have not. This is simplified by the highly accessible nature of MODO’s implementation in Linny-R. Lastly, we suspect that qualitative re- search into the empirical applicability of power system simulation models might prove invaluable. Research that treats decision-makers and policy analysts as clients to discover their modeling needs could result in a roadmap for the participatory development of power simulation models in order to better foster their adop- tion, and that of their findings.