Understanding Gas Porosity: From Invisible Flaws to Predictive Power

A Perfect Surface Can Hide
A Critical Internal Failure

Gas porosity forms beneath the visible surface of a casting, creating microscopic or macroscopic void networks that reduce mechanical integrity. Components may pass visual inspection while already containing the seeds of future fatigue cracks, leakage failures, and catastrophic breakdowns.

Two Different Mechanisms.
Two Different Solutions.

Porosity Follows Predictable Physics

Gas porosity is not random. Its formation is governed by thermodynamics, fluid flow, pressure, surface tension, and the movement of the solidification front.

Core-Gas Bubbles

Heat-driven binder pyrolysis generates gases such as carbon monoxide, carbon dioxide, and hydrocarbons. Poor venting can inject these gases into the melt, trapping spherical pores near core surfaces.

Shrinkage Porosity

Metals contract during solidification. If liquid feed metal cannot reach isolated regions, the remaining liquid contracts inward and produces irregular interdendritic voids.

Nucleation Physics

Bubbles form when local gas pressure overcomes metallostatic pressure and surface tension. Inclusions, oxide films, grain boundaries, and dendritic gaps often provide preferred nucleation sites.

Binder chemistry, core venting, riser and feeder design, melt cleanliness, and controlled solidification are the main levers for reducing porosity.

After nucleation, bubbles may grow, merge, or migrate according to local melt velocity and solidification-front speed, determining their final size and distribution.

Moving Beyond Trial-and-Error

CFD simulation models the complete filling and solidification sequence. Engineers digitally optimize gating, risers, and mold geometry, replacing weeks of physical trials with hours of computation.

Multiphysics Coupling

Accurate porosity prediction requires coupling fluid dynamics, heat transfer, gas transport, and solidification metallurgy. This integration enables precise defect prediction beyond simplified models.

Industry-Leading Tools

ProCAST: Commercial platform with modules for investment, die, and sand casting. Predicts shrinkage porosity, gas entrapment, and microstructure evolution with high accuracy.

OpenFOAM: Open-source CFD platform using VOF methods to track metal-air interfaces. Customizable for advanced gas evolution and bubble transport studies.

A Critical Hydraulic Casting
Was Rejecting One Out Of Every Four Parts

After repeated physical modifications failed to resolve a persistent defect problem, Metalcast S.A. turned to simulation-driven engineering to identify root causes, eliminate uncertainty, and achieve rapid quality improvement without new hard tooling.

Physical Trial & Error

Multiple modifications to the gating system and riser design failed to significantly improve quality because the actual defect mechanisms remained hidden.

Unknown Root Cause

Conventional inspection identified faulty parts but could not explain precisely where the defects originated or why they developed.

From Reactive Inspection to Proactive Design Excellence

Supercomputing, artificial intelligence, and real-time monitoring are converging to predict and prevent casting defects before production, redefining quality as something engineered into the process.

Supercomputing Models

High-performance computing enables finer spatial and temporal resolution, revealing dendritic structures, thin walls, and micro-scale gas behavior that simplified models may miss.

AI Defect Prediction

Machine-learning models trained on simulation data can match live sensor patterns to defect signatures and flag porosity risks before parts leave the mold.

Autonomous Optimization

Reinforcement learning can explore pouring temperature, fill speed, and cooling settings, using sensor feedback to continuously reduce scrap risk.

Post-production sorting accepts defects as inevitable. Simulation-driven design instead identifies and removes defect mechanisms before physical trials, reducing waste and shortening development cycles.

Foundries that combine simulation, AI, and in-process monitoring will set tomorrow’s benchmarks for consistency, development speed, material efficiency, and quality performance.