# A Conservative Multi-Physics Simulator for Modular Plasma-Field Shielding with Fractal Correction Engine Integration: Numerical Methods, Validation, and Honest Performance Bounds **Author: ** Adam L McEvoy **Version: ** 3. 0 --- ## Abstract I present a modular, multi-physics simulator for evaluating plasma-based magnetic shielding of a spacecraft against relativistic threat fluxes (charged particles, neutral micrometeoroids, and photons), integrated with the Fractal Correction Engine (FCE) — a domain-agnostic predict–compare–correct framework that I apply as a closed-loop stability and navigation layer. Version 3 replaces the previous 2D magnetohydrodynamic (MHD) solver with a conservative finite-volume scheme using background-field splitting, an HLL Riemann solver with relativistically capped signal speeds, hyperbolic divergence cleaning, a Boris semi-relativistic momentum correction, and a shock-gated dual-energy formalism. Every physical and numerical energy channel is tracked in an exact ledger. The result is a reduction of the flagship scenario's maximum energy-conservation error from 17. 3\% (V2) to 10^-8, with machine-precision closure (10^-11) in controlled tests, and second-order convergence of the magnetic-divergence error across a 10032 400128 resolution sweep. The particle-in-cell (PIC) boundary layer now carries its own validation chapter with quantitative conservation metrics. Honesty requires reporting two corrections to earlier claims: (i) the V2 "100% shield effectiveness" result is an artifact of a simulation that terminated before any threat could arrive (simulated time 1. 0\, s against a scanner-measured minimum time-to-impact of 1. 34\, s) ; the physically complete run yields 45. 8\% overall effectiveness, decomposing into 100\% deflection of the magnetically addressable (charged) threat class and near-zero mitigation of neutrals and photons, exactly as the underlying physics requires; and (ii) the legacy 1D solver is shown to be non-conservative at long run times (energy self-amplification by a factor of 8. 6 10⁷ and superluminal flow speeds), and its results should be considered qualitative only. The simulator demonstrates modeled mitigation under specified assumptions; it does not demonstrate a realizable shield. --- ## 1. Introduction Shielding a spacecraft at relativistic cruise speed is a multi-physics problem: an onboard magnetic field can deflect charged particles, a confined plasma can contribute collisional stopping, a material (Whipple) shield absorbs what fields cannot touch, and trajectory planning can steer the vehicle around the densest threat sectors. Each subsystem lives in a different physical regime — ideal MHD for the confined plasma, kinetic/electrostatic physics in the thin boundary layer, relativistic single-particle dynamics for the threats, hypervelocity impact mechanics for the material shield. **Speculative context** The application context — interstellar cruise at = v/c up to 0. 99 behind a 100\, T-class magnetic shield — is speculative. No known materials or power systems support such a vehicle. I use the scenario as a *stress test for the numerical methods and for the FCE control concept*, not as an engineering proposal. Throughout this paper, sections describing physically established models and their numerical verification are the load-bearing content; sections tied to the interstellar application or to the FCE's conceptual framing are labeled speculative or theoretical. This version (V3) is a scientific-rigor release. Its purpose is to fix, measure, and honestly report the numerical foundations of every claim: 1. **Conservation. ** The 2D MHD solver is rewritten in conservative finite-volume form with an exact, RK-weighted energy ledger. The V2 flagship conservation error of 17. 3\% becomes 10^-8, and the sources of the residual are individually itemized. 2. **Divergence control. ** is now monitored per cell, plotted over time, and — critically — correlated against FCE actuation, closing the possibility that field corrections silently inject magnetic-monopole artifacts. 3. **Kinetic-layer validation. ** The PIC boundary layer, previously untested in practice, now has dedicated torture scenarios and quantitative conservation metrics. 4. **Honest reinterpretation. ** Where the corrected system contradicts earlier headline numbers, I report the contradiction and its cause. --- ## 2. What the System Is The simulator (`plasmaₛhieldₛim`, Python/NumPy) advances a coupled set of subsystems on a shared clock: | Subsystem | Model | Role ||---|---|---|| Plasma shield | 2D axisymmetric semi-relativistic ideal MHD, conservative FV (this work) | Magnetic deflection medium confined between r₈₍=50\, m and r₎ₔₓ=70\, m || Threat ensemble | Relativistic test particles: protons/ions, neutral grains, photons | The incoming flux (up to 10³–10⁴ particles) || Particle tracking | Vectorized relativistic Boris pusher; Bethe–Bloch stopping; Compton attenuation | Trajectories, deflection, hull-hit accounting || PIC boundary layer | Cloud-in-cell deposition, spherical Poisson (red–black SOR), Boris push, pair production, charge exchange | Kinetic charge-separation physics at the shield's outer edge || Material shield | Multi-layer Whipple bumper: fragmentation, energy partition, per-sector damage map | Last line of defense against neutrals/photons || Look-ahead scanner | 2000\, km detection horizon, threat classification | Sensor feed for the FCE || Path optimizer | FCE-driven great-circle heading corrections over a 1224 angular threat landscape | Navigation-level mitigation || Directional coils | Per-sector current modulation with power conservation | Field concentration toward predicted impacts || Engineering module | Coil stress, stored energy, cryogenics, quench (static) | Feasibility bookkeeping (not yet coupled into the physics) || FCE layer | Nine analyzer/corrector instances over a shared adapter | Closed-loop stabilization, prediction, navigation (Section 3) | A single simulation step advances the MHD state, pushes all particles, runs the PIC cycle where particles occupy the boundary zone, processes material-shield impacts, and updates the energy budget. Every `fcecorrectionᵢnterval` steps (default 20), the FCE layer scans, predicts, and writes its corrections. --- ## 3. The Fractal Correction Engine **Theoretical/heuristic framework** The FCE is my domain-agnostic correction concept. It is *not* a validated physical theory; it is a control-and-diagnostic architecture whose value is measured here purely by its quantitative effect on the simulation (Sections 6–7). I label its conceptual claims accordingly. ### 3. 1 Concept The FCE treats any evolving system as a path in a state space. Between perturbation events, the path is assumed to be reconstructible from a baseline (free-motion) law plus local curvature information; the engine extracts this "fractal path" (using -scaled multi-resolution curvature analysis), predicts forward, and re-anchors whenever an observation disagrees with the prediction. The runtime is a closed loop: predict compare (residual) correct, with the residual treated as a *signal*: a near-zero residual means the model is complete (coast) ; a persistent, structured residual means an unmodeled process is acting and the affected subsystem should be decomposed further. One FCE instance is attached per independently evolving part, and instances are composed at interaction points. **Heuristic** The specific curvature extraction — local Menger/osculating curvature combined with multi-scale (-harmonic) decompositions, Lyapunov-exponent estimation, Poincaré-section analysis, and fractal dimension estimates — is a pragmatic toolbox rather than a derivation from first principles. Where an external FCE backend is unavailable, the adapter falls back to built-in NumPy implementations of each primitive. ### 3. 2 How the FCE is applied in this system Nine consumer instances share one adapter. The complete map, including whether each loop actually closes into the physics state (a distinction V3 makes explicit): | # | Instance | Surface observed | Actuation (writes back) | Loop ||---|---|---|---|---|| 1 | `StabilityCorrector` (1D) | B (r) profile history | B (r) rate into induction | Closed || 2 | `StabilityCorrector2D` | angular mode amplitudes of Bᵣ, B_, B_ | per-component B (r, ) rate | Closed || 3 | `AlfvenSpeedCorrector` | vA (r) profile | B, smoothing | Closed || 4 | `TrajectoryPredictor` | threat particle histories | impact predictions (feed-forward) | Open || 5 | `FieldModulator` | predictions from #4 | radial B + coil multipliers | Open (feed-forward) || 6 | `FCEPathOptimizer` | threat landscape | ship heading correction | Closed (per scan) || 7 | `PICFCEAnalyzer` | PIC c (r), E (r) | **V3: Eᵣ field smoothing** | Closed (new) || 8 | `MaterialShieldFCEAnalyzer` | damage/impact maps | maintenance recommendations | Diagnostic || 9 | Coil field-quality analysis | designed B (z) | ripple metrics | Diagnostic | Two rigor rules were added in V3 for every actuating instance: 1. **Energy transparency. ** Any FCE correction that modifies the magnetic field also enters the energy equation with its consistent power, P₅₂₄ = B Ḃ₅₂₄/₀, and is accumulated in the energy ledger (`fceᵢnjected`). FCE actuation can therefore never silently create or destroy energy. 2. **Charge transparency. ** In the PIC layer the FCE corrects *only the electric field* (analogous to the digital field filtering standard in PIC codes), never the deposited charge density, so it cannot violate charge conservation. This is verified
Adam L McEvoy (Sat,) studied this question.