Abstract for: Monte Carlo Simulation Support in LunaSim: A Browser-Based System Dynamics Modeling Platform for Sensitivity Analysis

Existing system dynamics tools that support Monte Carlo sensitivity analysis, such as Vensim PLE+, require desktop installation and commercial licensing, creating barriers to access in educational and collaborative settings. There is a need for a free, browser-based platform capable of probabilistic sensitivity analysis. Monte Carlo simulation support was implemented in LunaSim using a parallel ensemble execution architecture via the Web Worker API. Three new structured stock types (Microwave, Conveyor, and Queue) were introduced to expose named parameters for direct uncertainty sampling, and live API integrations with Open-Meteo, the World Bank, and Finnhub were added to seed model parameters from real-world data. The parallel Web Worker implementation reduced execution time by up to 68.5% compared to sequential execution across ensemble sizes of 100 to 1,000 runs. Validation against an analytical investment portfolio benchmark confirmed numerical correctness across p5, p50, and p95 percentile levels at five timesteps, with a maximum deviation of 6.84%. The three stock type test cases produced confidence bands consistent with their respective input distributions. The additions shift LunaSim from a deterministic tool into a platform capable of probabilistic, real-world-grounded simulation entirely in the browser. Structured stock parameters enabled more direct uncertainty targeting without requiring modelers to rewrite equations, and live API integration allowed confidence bands to reflect current real-world conditions.