Abstracts Track 2026

Area 1 - Intelligent Sensor Systems and Edge-Cloud Computing

Nr:	97
Title:	A Two-Layer ANN Framework for Sensor-Driven Intelligent Control of a Residential Microgrid: A Real-World Cyber-Physical Implementation
Authors:	Anthony Fabian Nyoyoko, Peter Mark Jansson and Samuel Oduduabasi Ekpo
Abstract:	This work presents the design and implementation of a two-layer Artificial Neural Network (ANN) framework for intelligent control of a real, operational residential microgrid at Bucknell University. Unlike many existing studies that rely on simulation-based environments, this work is grounded in a fully restored cyber-physical system, where sensing, data acquisition, and control are performed on low-cost embedded hardware using real-time operational data. The proposed architecture follows a structured Predict → Decide → Protect philosophy. The first layer performs next-hour prediction of electricity prices in the PJM Interconnection, a regional transmission organization that coordinates wholesale electricity markets and grid operations across multiple states in the United States, using historical Day-Ahead and Real-Time Locational Marginal Price (LMP) data. This prediction addresses a fundamental constraint of electricity markets, where Real-Time prices are only available after operation, requiring decisions to be made under uncertainty. The second layer translates these predicted price signals into actionable control decisions within the microgrid. Real-time electrical measurements, including voltage, frequency, current, and power, are continuously collected through an AcuRev smart meter and logged via a Raspberry Pi-based controller at five-minute intervals. These measurements, combined with predicted price signals and environmental data, serve as inputs to the intelligent controller, which determines operating modes such as normal operation, load management, pre-heating or pre-cooling, and islanded operation. A key contribution of this work is the integration of artificial intelligence with real-world system constraints. While the ANN-based controller demonstrates measurable improvements in economic performance compared to traditional rule-based strategies, experimental observations reveal that unconstrained optimization can introduce violations in voltage stability. This finding highlights a critical limitation of purely data-driven control in power systems. To address this, the framework incorporates a hybrid control philosophy in which AI-driven decision-making is augmented with rule-based safety constraints that enforce voltage and frequency limits. This ensures that economic optimization does not compromise system reliability. This work is supported by an operational microgrid platform and prior validated results in the price prediction stage. The results demonstrate that effective AI-based microgrid control is not solely a modeling problem but a cyber-physical systems challenge requiring coordinated sensing, communication, and control.

Area 2 - Modeling and Simulation Methodologies

Nr:	92
Title:	A Multi‑Bottleneck System Dynamics Model for Integrated Carbonation–CCU Systems
Authors:	Beatriz Royo Agustin and M. Teresa De la Cruz Eiriz
Abstract:	Purpose Current modelling approaches do not capture the dynamic interaction between demand, resource availability, and operational constraints in carbonation–CCU systems. This work proposes a system-level dynamic modelling framework to analyse integrated carbonation–CCU systems under realistic conditions. Design/Methodology/Approach The system is modelled as a multi-bottleneck structure, where production is determined by the most restrictive factor among demand, reactor capacity, slag availability, and CO₂ supply. A system dynamics approach integrates carbonation, CCU, and logistics subsystems. Production follows a demand-driven logic with inventory feedback, while constraints are implemented through a minimum-function formulation. Baseline and verification scenarios are used to identify regimes and transitions. Findings Results show that behaviour depends on the dominant constraint. Under baseline conditions, the system is demand-driven, leading to underutilisation (~25%) and high unit costs despite high material efficiency (~85%). Scenario analysis reveals transitions between demand-, slag-, reactor-, and CO₂-constrained regimes. The CCU subsystem introduces buffering effects that increase short-term resilience but may cause abrupt performance degradation when depleted. Research Limitations/Implications The model relies on simplified assumptions and excludes stochastic variability and market dynamics. However, it provides a basis for extending analysis toward realistic industrial scenarios. Social/Technical Impact The framework supports integrated analysis of material flows, constraints, and operational decisions, contributing to the evaluation and scaling of circular CO₂ utilisation systems. Final Comments/Originality/Value The contribution lies in framing carbonation–CCU systems as dynamic multi-bottleneck processes, enabling the identification of regime shifts and supporting design and scaling decisions beyond static representations.

Area 3 - Simulation Technologies, Tools and Platforms

Nr:	32
Title:	Predicting Reconstitution of Powders using a Hybrid Modelling Approach
Authors:	Yogesh Harshe, Marina Latino, Alberto Di Renzo and Francesco Paolo Di Maio
Abstract:	When dealing with the solubilization of food powders in a liquid medium, proper handling of the reconstitution process is essential. However, to date, the complexity of the overall process – comprising different steps and phenomena - is accompanied by fragmented and still incomplete knowledge. No unified model currently exists that can adequately predict its performance, as most investigations have been conducted primarily through experiments. Based on these considerations, the present work investigates the reconstitution of food powder in pure water from a modelling perspective. The proposed approach enables the prediction of the reconstitution time across different types of food powders. Specifically, the study considers powders belonging to two broad food categories (matrix A and matrix B) and exhibiting four distinct structural classes determined by their manufacturing processes (submatrices 1, 2, 3 and 4). The model also incorporates the effect of different temperatures on reconstitution behaviour. The modelling methodology explores different hybrid modelling techniques. A serial hybrid structure was first developed, followed by an alternative parallel framework to enable further comparisons and investigations. The Noyes–Whitney equation was adopted as the basis of the mechanistic component, which primarily focuses on the dissolution step to ensure consistency with the dissolution-oriented experimental conditions of the available datasets. Different regression algorithms were tested for the Machine Learning component. The models were evaluated using standard statistical performance metrics, primarily the coefficient of determination (R²) and the root mean square error (RMSE). Distinct behavioural clusters were identified. For some of them, the proposed model achieved promising predictive performance. However, this was not the case for all the clusters, as preceding reconstitution steps may still strongly influence the overall behaviour. Their effect was indirectly incorporated through a lumped parameter, the dissolution kinetics, which was synthetically generated through an optimization problem. Overall, compensation effects among different clusters helped the hybrid model converge toward accurate predictions, enabling promising predictive performances even in a simplified formulation.

Area 4 - Application Domains

Nr:	93
Title:	Using Motivation to Explain the Outcome of a Cyber-Attack
Authors:	Erjon Zoto
Abstract:	1. Purpose In cybersecurity, motivation helps understand what pushes the different actors to initiate and/or resist and counter ongoing attacks. This study covers several theories that can be used to explain the underlying factors behind motivation and how to model it within a cyber-attack simulation tool. 2. Methodology Using a combination of qualitative and quantitative methods, this study aims to help answer the main research question: -Can motivation explain the outcome of a cyberattack? More specifically, this study will answer the following subquestions: 1. Which theories are most relevant in explaining actors' motivation in a cyber-attack? 2. How can we map relevant motivation theories in a cyber-attack simulation tool? The selected theories should cover attack and defense scenarios, and will then be applied to model agents' motivation within a simulation tool. 3. Findings The study focused on the following theories: the protection motivation theory (defense side), the self-determination theory (attack), and Schwartz's theory of basic human values (attack). Mapping the theories into the simulation tool was achieved through allocating specific values to the underlying factors and subfactors. 4. Research Limitations/Implications Three theories were selected for this study, with one covering the defense side and two the attack side. The simulation tool used is CyberAIMs, with pre-defined attributes and interactions, where only the “Motivation” attribute for specific agents was affected by the mapping process. Several assumptions were made to apply each theory in the tool for a given scenario. 5. Social & technical Impact The updated "motivation" attribute will help the tool increase its coverage and usability, hence provide better education and training on security awareness for all relevant stakeholders. 6. Value The study has found relevant theories explaining actors' motivation in cyberspace, and succeeded in applying them in a cyber-attack simulation tool.

Nr:	90
Title:	Python‑Based Simulation for Stress Testing and Reconfigurability Assessment in Industrial Supply Chains
Authors:	Beatriz Royo Agustin and M. Teresa De la Cruz Eiriz
Abstract:	Purpose This work presents a Python-based stress testing framework to assess reconfigurability and resilience limits in industrial supply chains. Within a Supply Chain Digital Twin (SCDT) context, it evaluates system behavior under a do-nothing scenario, identifying hidden vulnerabilities and structural limits. Design/Methodology/Approach The approach uses open-source discrete-event simulation (DES) in Python to build a dynamic replica of a make-to-stock system integrating sourcing, production, inventory, and distribution. Stress scenarios capture demand variability and material shortages. Performance is assessed through financial, operational, service, and inventory KPIs following SCDT stress testing principles. Findings Results show high vulnerability without reconfiguration. Demand drops cause disproportionate profit losses (>45% for −30% demand), while demand surges degrade service levels. Material shortages generate cascading effects across inventory, capacity, and fulfillment, revealing breaking points. Resilience cannot be inferred from baseline performance alone. Research Limitations/Implications Limitations include deterministic assumptions, aggregated data, and simplified dynamics. However, the framework supports extension with stochastic approaches and reinforces SCDTs as diagnostic tools for resilience assessment. Social/Technical Impact The work demonstrates how open-source Python models enable transparent, modular, and reproducible simulation. It is especially relevant for SMEs, providing accessible tools to understand vulnerabilities and support data-driven decisions without relying on proprietary solutions. Final Comments/Originality/Value The contribution lies in an open-source, reproducible, and transferable framework for SCDT-based stress testing. By targeting SME applicability, it promotes wider adoption of digital twins and supports scalable supply chain resilience analysis.