Modelling the spatial distribution of decentralised renewable energy investment decisions using system dynamics and agent-based modelling approach

Zara Abba and Nazmiye Ozkan

Cranfield University, UK

Background

The path to net zero and universal energy access for 568 million people in Sub-Saharan African countries (SSA) living without clean and reliable electricity requires a scale-up in decentralised renewable energy (DRE) deployment (e.g., solar mini-grids). Despite global trends in technology advancement and decreased cost, there remains a deficit in investments in SSA. According to the IEA, SSA received only 1.5% of global investments in renewables between 2000-2020. Energy planning studies have enabled the identification of suitable DRE technologies for specific locations and analysing of policy actions to encourage private investments. However, the studies do not incorporate location attractiveness factors with heterogeneous investor preferences and attributes that drive investment decisions. This can better inform the identification of actions that catalyse investments. This paper introduces a hybrid system dynamics and agent-based model (SD-ABM) that assesses the emerging spatial distribution of DRE investments by heterogeneous investors (using solar mini-grids as a case study). The model allows for examining the impact of policy actions on investment patterns, consequent generation, and CO₂ emission avoidance.

Methods

An integrated SD-ABM model was developed to investigate emerging investment patterns of solar mini-grids. The model framework considers heterogeneity in location and technology attractiveness within a country. It additionally considers that investors perceive risks differently and make investment decisions in locations with the highest expected utility and attractiveness. The model thus simulates the spatial investment patterns based on the investor decisions resulting from dynamic interactions. These interactions are between electricity demand, supply, heterogeneous investor attributes (including investment goals, technology attractiveness, risk and financial attributes), location characteristics that drive attractiveness and representation of spatial consumer ability to pay. SD was used to model supply and demand dynamics, while yearly investment decisions were modelled using ABM according to heterogenous rules and attributes. The investor agents considered in the model include development finance institutions (debt), commercial banks (debt), developers (equity), impact investors (concessional debt) and combinations of the investor types.

The model was developed in the following steps:

1. A conceptual model was developed based on concepts and structures from literature and interaction with academia. The conceptual model was validated via expert consultation, testing the model structure and assumptions. The use of structures in available literature provides confidence in the theoretical validity of the model structure.
2. A mathematical model was developed involving formulating equations and agent rules and creating stock and flow diagrams based on literature, GIS data, survey data, and interaction with experts. The intermediate results were verified throughout the model development process. The model was built using Anylogic software.
3. Further model structural and behavioural validity was checked using a case study of solar mini-grids in Nigeria.
4. Scenario analysis was conducted to test the implications of actions on investment outcomes.

Results

Results showed that concessional debt investors dominated investment decisions without intervention, while commercial banks made the least investments. Moreover, scenario analysis was conducted to investigate policy actions that can improve investments. Results showed that while some locations are attractive without interventions, others do not attract significant investments for the tested interventions due to their inherent characteristics. Targeted local-based policies and actions may enable investments for such locations.

Conclusions

This study contributes to ongoing research by developing a model to support investment planning and decision-making by incorporating investor preferences and location attractiveness. This model is helpful for policymakers to test location-specific policies that can improve investments and for investors to identify viable locations based on their preferences and attributes. We conclude that aside from policy interventions, improvement in electricity access requires a combination of actions that improve project economics and attractiveness and reduce the impact of risks.