Causal Yield Mapping and Evaluation - A Graph-Based Methodology for Process-Driven Project Steering in BIM

Current BIM quality assurance operates in isolated silos: IFC models, schedules, and issues exist in separate systems without causal linkage. A missing property is flagged as a defect, but its impact on downstream processes — which milestone it blocks, which contractual deadline it jeopardises — remains invisible. This paper introduces Causal Yield Mapping and Evaluation, a methodology that unifies IFC building data, process landscapes (IoC Process Ontology), schedules, and BCF issues in a single traversable knowledge graph. The graph models the project as it is, not as it was planned: it continuously assimilates new data and derives forecasts from the current state rather than from an outdated baseline. A causal score quantifies the schedule relevance of each issue by traversing the graph from issue to task to edge to downstream milestones. Bottleneck identification follows Goldratt's Theory of Constraints: the bottleneck is the edge with the highest score on the critical path, and resolving it shifts the constraint to the next critical location. A weighted bottleneck ranking (Rank 1: current, Rank 2: predicted next, Rank 3: diminishing buffer) replaces flat issue lists. A two-layer separation keeps the scale tractable: atomic rule findings — tens of thousands per milestone check — carry the measured defect load, while curated issues are the tracked, communicated, and scored unit. Probabilistic forecasting is designed in three stages — Earned Schedule, Monte Carlo simulation from an actor–constraint matrix, and Bayesian networks — of increasing demand on historical project data. The methodology bridges the gap between BIM quality assurance and schedule-driven project steering, shifting BIM management from reactive issue processing to predictive, data-driven control. It is grounded in CPM, PERT, TOC, and Lean Construction (LPS), but extends them through weighted edges, heterogeneous graph traversal, continuous updating, and active cybernetic control loops. A reference implementation exists in the cyme™ platform of the author's firm. Concept paper (preprint, not peer-reviewed). First public version, September 2026. The methodology described may be cited and referred to freely under the CC BY 4.0 licence of this document; the platform designation is a trade mark of compendium BIM und Kybernetik GmbH + Co. KG.