Grid congestion has become a structural constraint on the Dutch energy transition, yet the flexible grid connection agreements introduced to facilitate Battery Energy Storage Systems (BESS) under congested conditions remain poorly understood as integrated institutional arrangements. While these agreements are intended to unlock network capacity through non-firm access, they also create an institutional paradox: the technologies best suited to alleviate congestion are increasingly constrained by access arrangements originally designed for passive network users. This thesis investigates how the institutional design of flexible connection agreements influences the financial viability, operational behaviour, and system alignment of utility-scale BESS in the Netherlands.

An exploratory sequential research design is adopted, combining qualitative institutional analysis with quantitative policy evaluation. First, 33 Dutch regulatory and stakeholder documents are analysed to identify recurring institutional frictions, which are synthesised into three structural tensions: temporal misalignment, asymmetric risk and value allocation, and a disconnect between operational control and financial responsibility. These tensions are subsequently operationalised as four contractual design properties: compensation structure, firmness, volume orientation, and directive control. Their effects are evaluated using a Mixed-Integer Linear Programming (MILP) Policy Testbed that simulates the operation of a 10 MW / 20 MWh utility-scale BESS across 200 independent Monte Carlo market realisations. The evaluation covers four existing Dutch flexible connection agreements, the Alternative Transport Right (ATR85), Capacity Limiting Contract (CBC), Capacity Steering Contract (CSC), and Time-Based Transport Right (TBTR), together with two prospective contract designs, Volume Bound and Firmness Gradient, developed to isolate the independent effects of individual contractual properties.

The results demonstrate that the principal barrier to battery bankability lies upstream of contract design itself. The high-voltage transport tariff alone absorbs approximately 70% of gross market revenues, rendering the unconstrained benchmark structurally non-bankable. Among the evaluated agreements, only the Capacity Limiting Contract consistently restores financial viability. However, it achieves this through a compensation coverage ratio of up to 1,596% of demonstrated operational risk, indicating that bankability is restored through financial transfers sufficiently large to offset the underlying tariff burden rather than through proportionate risk compensation. By contrast, fixed tariff-discount arrangements, particularly the Time-Based Transport Right, remain financially unviable because rigid transport windows conflict with high-value market opportunities.

Beyond financial performance, no single contractual design property simultaneously optimises all evaluation dimensions. Directive control, implemented through the Capacity Steering Contract, delivers the strongest congestion alignment but depends critically on accurate DSO forecasting and activation timing. Volume-oriented arrangements reduce battery degradation and extend operational lifetime, while their contribution to congestion management is regime dependent, improving outcomes in urban, demand-driven networks but potentially worsening them in rural, generation-driven systems. Overall, the findings demonstrate that the effectiveness of flexible connection agreements is determined by their underlying design properties rather than by contractual labels alone. More fundamentally, they indicate that existing agreements are compensating for structural shortcomings in the Dutch grid tariff architecture, suggesting that the greatest opportunity for future reform lies in grid-access pricing rather than incremental contract redesign. ... Read more

Introduction

The Dutch electricity system increasingly needs flexibility, because electricity generation from variable renewable sources such as solar and wind does not always match electricity demand. Battery Energy Storage Systems (BESS) can provide this flexibility by storing electricity and releasing it later. They respond quickly, scale across system sizes, and are already commercially proven. These properties have led market actors to increasingly deploy utility-scale BESS. At household level, residential BESS, or home batteries, are also growing rapidly and make up ~45% of installed capacity in the Netherlands.

Despite technical similarity, installed capacity of the two types of BESS grows in fundamentally different ways. Utility-scale capacity growth results from a limited number of professional investment decisions and is therefore relatively predictable. Residential BESS capacity growth is more difficult to predict, because it emerges from many separate household decisions shaped by financial and non-financial considerations. Various actors can only influence the conditions under which households evaluate one. In this thesis, these conditions are conceptualized as home battery attributes, such as investment costs, payback time, certainty of payback time, self-consumption and emergency power.

Literature Review

Existing literature shows that residential BESS adoption follows from many household decisions to invest in a home battery. This decision should not be understood as purely techno-economic. Financial drivers such as costs, payback time and lower electricity bills matter, but non-financial drivers such as energy autonomy, emergency power and environmental motives also shape the household decision. A structured literature search identified a broad set of adoption drivers, which were organized using a framework for categorizing drivers of household investment decisions. Although literature is rich in possible drivers, it does not clearly show which home battery attributes matter most to households, how important these attributes are relative to one another, nor how changes in these attributes affect choices. Existing prioritizations rely mainly on stated importance rather than observed choice trade-offs, and evidence is concentrated in Australian studies, limiting transferability to the Dutch context.

The main research question of this thesis is therefore: How do changes in home battery attributes affect the household investment decision to buy a home battery? It is answered through three sub-questions. First, the thesis identifies the most important attributes affecting the household investment decision. Second, it estimates their relative importance. Third, it explores how changes in these attributes affect the probability that a household chooses a home battery.

Methodology

The research combines a literature analysis with Discrete Choice Modeling (DCM). The literature analysis screened the broad set of drivers down to a set of 12 attributes, from which we derived a subset of five key home battery attributes: investment costs, payback time, certainty of payback time, energy autonomy, and backup power.

A discrete choice experiment (DCE) was designed where respondents repeatedly chose between Battery A, Battery B and no battery. Each battery was described by a set of these attributes and their levels. The final experiment included eight choice tasks and was completed by 529 usable respondents. From the total pool of respondents, based on household characteristics, a representative sample for the average Dutch household was made. A panel mixed logit model estimated relative importance of the key attributes and the choice probabilities for choosing a baseline home battery with consistent attributes.

Results

A panel mixed logit model estimated on the choice data shows that all five attributes have a statistically significant effect on household choices. The model provides significant explanatory power (ρ² = 0.2477). The relative importance of the attributes for the average Dutch household is shown in Figure . A choice probability analysis shows the probability for choosing a baseline home battery is 0.60 in a data sample with solar PV owners, and 0.53 in the representative sample. These values should not be interpreted as real market shares, but as controlled model outputs.

Conclusion

Changes in home battery attributes affect the household investment decision by changing the probability that a household selects the battery alternative. The five selected attributes are therefore relevant "knobs" through which residential BESS adoption can be understood or influenced. However, the model only tests the included attributes: it shows that these five matter, not that they are the five most important in every real-world decision.

From the key attributes and their relative importances we found that the household investment decision is strongly financial, but not purely economically rational. Costs, payback time and certainty of payback time together account for most of the relative importance, yet households do not appear to evaluate a battery as a simple profitability calculation. Payback time, a key metric for professional investors, is the least important attribute in the representative sample, and households did not accept more risk in exchange for shorter payback.

Investment costs and certainty of payback time are the strongest attributes. This has direct implications for policy and market actors. Measures that lower upfront costs, such as subsidies or tax rebates, target the most important part of the household decision. Measures that stabilize or guarantee returns target the second most important attribute. This is relevant because current Dutch policies mainly affect payback time and self-consumption, certainty of payback time is affected only indirectly. Barely any policies affect investment costs and certainty of payback time directly. Grid operators could use this insight by offering contracts that give households more certain returns in exchange for control over battery operation to stabilize grids.

The attributes do not tell the full picture. Comparing the representative and solar samples shows that targeting the right households can shift choice probabilities by a similar magnitude as large attribute changes. Actors seeking to promote adoption should therefore not only improve the battery proposition, but also focus on groups already more willing to adopt.

The results should be read within the limits of the method. The experiment simplifies a complex real-world decision, measures stated rather than actual choices, and only tests the included attributes. Future work should expand and formalize the attribute selection, test transferability across countries, quantify how policies change home battery attributes, and connect these preference estimates to diffusion models that translate attribute changes into capacity growth. ... Read more

Infrastructure planning problems are typically guided by long-term target states, but it is not always clear which earlier system conditions are necessary to reach those targets. Backcasting methods enable policymakers to reason backward from a desired future target state back to the present. Although backcasting literature is well established, the actual backward inference step is often the least formalised, or carried out in a non-formalised or intuitive way. As a result, policymakers lack methodological support to infer earlier system-state pathways in a systematic and reproducible way. This thesis addressed this problem by developing, implementing, and evaluating a DBN-inspired backward inference methodology for reduced infrastructure system. It starts from a specified target condition, and works backward to infer admissible earlier system-state pathways. The proposed methodology is operationalised through a reduced case study of residential electric vehicle charging in Norway. The results show that the methodology can produce multiple admissible pathways in a structured and explicit manner. However, the results are limited to the context of the reduced case study, and follow-up research is needed to test the methodological appropriateness in cases where no feasible pathways are identified, possibly on larger and more complex infrastructure systems with a longer time horizon and other inference techniques. ... Read more

Greenhouse horticulture is a highly energy-intensive sector that supports year-round food production but remains strongly dependent on fossil fuels. In the Netherlands, greenhouse heating and electricity supply are commonly supported by natural gas boilers and combined heat and power (CHP) units. Increasing energy-market volatility, carbon-related costs, grid constraints, and climate-policy pressure are changing the conditions under which these systems operate. At the same time, electrified heating, renewable electricity generation, thermal storage, and interconnection through shared infrastructure provide potential transition pathways. The value of these options, however, depends on how they interact within the complete greenhouse energy system rather than on the performance of individual technologies in isolation.

This research is conducted in collaboration with Certhon, a Dutch greenhouse specialist, and develops and applies a mixed-integer linear programming (MILP) optimisation framework to evaluate alternative energy system configurations and the resulting hourly operational dispatch strategies for a greenhouse cluster consisting of five greenhouse facilities located within the same geographical area. Operational data were obtained through Certhon, meeting with the greenhouse operators, expert consultation and a site visit to the participating greenhouse facilities. The framework compares individual operation with interconnected configurations using a shared heating network and shared renewable electricity production assets. It represents hourly heat and electricity demand, greenhouse-specific technologies, storage operation, grid imports and exports, renewable generation profiles, direct CO$_2$ emissions, carbon-related costs, and relevant Dutch market and regulatory conditions. The evaluated configurations are assessed under both 2025 known market conditions and projected 2030 energy market transition conditions. A reduced set of promising configurations based on performance in terms of cost, is subsequently tested under alternative market developments through in-optimisation sensitivity analysis.

The results show that the planned future installation already creates a significant transition pathway for the greenhouse cluster. Under individual operation, the current system's 2030 configuration, including the confirmed installations of the subsidised air-source heat pumps, reduces total gas use and direct CO$_2$ emissions by approximately 49.4\% relative to the 2025 baseline. However, this electrification pathway also increases electricity grid dependency, showing that greenhouse decarbonisation cannot be assessed only through gas reduction. Renewable electricity integration and grid interaction become central to the economic performance of future greenhouse energy systems.

Within the defined model boundaries, the lowest-cost configuration is Scenario~4, the individual operation pathway with local wind generation. Under projected 2030 market conditions, this configuration reduces the cluster's annual total cost by approximately 59.7\% relative to the 2025 baseline. This result is important because it shows that shared coordination does not automatically lead to the lowest annual cost. In this case, direct local renewable electricity availability reduces grid dependency and supports electricity export more effectively than the evaluated shared heating network alternatives. The configuration combining a shared heating network with a shared PVT installation, ATES and wind generation is the strongest interconnection alternative. Under standard 2030 conditions, it reduces annual costs by 48.6\% relative to the 2025 baseline and achieves the lowest direct emissions among the selected configurations.

The shared heating network results show that interconnection creates different operational roles across the greenhouse facilities. Some greenhouses operate as net heat suppliers, while others benefit from net extraction from the network. This demonstrates the potential value of coordinated heat exchange, but also shows that shared infrastructure requires transparent compensation mechanisms and clear agreements on infrastructure costs, ownership, and operational responsibility.

Overall, the research shows that greenhouse energy transition planning should be approached as a configuration-oriented decision problem. Heat pumps, renewable generation, storage, grid interaction, and shared infrastructure need to be evaluated together under changing market and regulatory conditions. The developed framework provides a structured decision-support basis for comparing realistic greenhouse energy transition pathways and identifying priorities for further engineering and investment analysis. ... Read more