Concrete cracks. It always has and always will — the material is strong in compression but brittle in tension, and thermal cycling, settlement, and loading guarantee that tensile stresses will eventually exceed the fracture threshold somewhere. The construction industry's response has been external repair: find the crack, inject epoxy, patch the surface. This works but requires detection, access, and labor — all expensive for infrastructure like bridge decks and tunnel linings where cracks form in inaccessible locations.
Self-healing concrete embeds the repair mechanism inside the material itself. Bacteria (Bacillus sphaericus), polymers, or chemical compounds are encapsulated in glass, polymer, or ceramic shells distributed throughout the concrete mix. When a crack forms, it ruptures nearby capsules, releasing the healing agent into the crack, where it polymerizes or precipitates calcium carbonate to seal the damage.
The modeling challenge is that healing isn't instantaneous — it's a time-dependent process coupling moisture diffusion with damage evolution (arXiv:2510.19839). A crack that forms in dry concrete heals slower than one in wet concrete. A crack in a load-bearing member may reopen during healing. The healing rate saturates nonlinearly: early repair is fast, late repair asymptotically approaches a limit that depends on crack width, moisture, and bacterial viability.
A Polynomial Chaos Expansion surrogate model captures this temporal evolution from early variability to nonlinear saturation, enabling full-cycle prediction of crack repair rates without running expensive finite-element simulations for each scenario.
The deeper engineering insight: the most useful infrastructure isn't the strongest — it's the most forgiving. Self-healing concrete doesn't prevent cracks; it makes cracks less consequential. The material's intelligence isn't in resisting damage but in responding to it autonomously, transforming a catastrophic event (structural cracking) into a routine maintenance process (biological repair).