Open concepts to think on

NASA traced the char loss on the Artemis I heat shield to gases forming inside the Avcoat ablator that could not vent fast enough because the material was not permeable enough. For Artemis II, NASA changed the reentry path rather than the shield, and says Artemis III upgrades aim to make the Avcoat blocks uniformly permeable.

The idea goes one step further: treat permeability as a design variable that can vary across the shield, not a single material property. A coupled multiphysics model, linking heat conduction, pyrolysis gas generation, porous flow and internal stress, would map where pressure is most likely to build up for a given entry trajectory. Regions that run hotter or longer, or that sit near seams and fittings, could then be given more open venting paths, while other regions keep a denser structure for strength and insulation.

Manufacturing methods that control porosity locally, such as additive manufacturing or graded infiltration, could turn that map into hardware. Arc-jet tests and post-flight scans would close the loop by checking predicted pressure and recession against what the material actually did, and by recalibrating the model for the next design.

Quality inspection and repair are usually separate steps: a defect is found, logged and fixed later, often by a person. Self-healing polymer composites raise a different possibility: parts that can repair some damage themselves when triggered.

The idea: connect an autonomous quality intelligence system directly to self-healing materials. When sensors or vision detect micro-damage, the system decides whether the damage is within the healable range, triggers the repair (for example with local heat), and then re-inspects to confirm it worked.

Every cycle becomes training data, teaching the factory which defects can be healed in place and which need a human expert.

Today, deciding whether a returned booster can fly again relies heavily on inspection. Additively manufactured parts, such as nozzle cooling channels, make this harder because their internal geometry is difficult to inspect.

The idea: design each critical part together with its own sparse sensor layout and a physics-informed prognostic model. During every flight, a handful of strain and temperature readings update a Bayesian estimate of accumulated fatigue damage, with honest uncertainty bands.

The result would be a 'remaining flights' gauge for each component, so refurbishment decisions are driven by evidence rather than conservative fixed limits.

Shunts for normal pressure hydrocephalus are usually set using population-level guidance, yet the risk of over-drainage, and with it subdural hematoma, depends on each patient's anatomy, brain stiffness and posture.

The idea: build a lightweight fluid-structure interaction 'digital twin' from a patient's routine MRI. It would estimate how cerebrospinal fluid drainage reshapes the brain and stresses bridging veins at different valve settings, then suggest the setting with the lowest predicted risk.

Over time, outcomes from real patients would recalibrate the model, turning a one-off simulation into a learning system that gets safer with every case.