BMPA FUSION LAB
PGOIS · PHASE ARRAY CONTROL
NEXT GENERATION FUSION CONTROL

Fusion plasma,
controlled by phase arrays.

An interactive simulator extending BMPA (Phase Array + AI Closed-Loop Optimization) to tokamak MHD instability-mode control. Compare mode amplitude, phase correction, and heat-load proxies under the same initial conditions and disturbances for baseline and BMPA control.

DUAL-REACTOR MODEIdentical disturbances · Different control strategiesT + 000.0 s
CONTROL TARGET MHD 2/1 TEARING
01 / DIGITAL TOKAMAK TWIN

Real-Time Plasma Control Comparison

SYNCHRONIZED INPUTS
Conventional Control (Baseline)
Fixed gain · Delayed mode correction
FIXED RESPONSE
TOROIDAL PLASMAMODE ACTIVITY
MHD MODE AMPLITUDE0.000NORMALIZED · |a|
MONITORINGBASELINE
Mode growth rate+0.00 /s*
Cumulative heat-load proxy0.0 a.u.
BMPA Closed-Loop Control
12-channel phase array · Online adaptation
PHASE ADAPTIVE
TOROIDAL PLASMAPHASE LOCK
MHD MODE AMPLITUDE0.000NORMALIZED · |a|
CLOSED LOOP ACTIVEBMPA / PGOIS
Phase correction error0.0 °
Cumulative heat-load proxy0.0 a.u.
Amplified mode / concentrated heat loadPhase correction / mode suppressionSimulated disturbances and perturbations
REAL-TIME PHASE SIGNALSBaseline mode component ━ · BMPA mode component ━ · Array correction ━
Mode Suppression · Relative Comparison
0.0 %
BMPA mode-amplitude reduction versus baseline at the same time
BMPA Mode Stability Index
0 /100
A smaller mode amplitude yields a higher normalized score
Cumulative Heat-Load Reduction
0.0 %
Integral comparison of MHD-related heat-load proxies
Array Control Output
0.0 a.u.
Normalized magnitude of the 12-channel phase synthesis
03 / TIME-DOMAIN ANALYSIS

MHD Mode Amplitude · Time Response

Baseline Control BMPA Closed Loop
The same input and disturbance time series are applied to both models. Vertical axis: normalized mode amplitude |a| · Horizontal axis: model time.
04 / PHASE ARRAY DIAGNOSTICS

12-Channel Phase Array

CLOSED LOOP

Combines circumferential coil phases to synthesize a correction vector for the selected MHD mode.

Estimated Mode Phase0.0°
Control Field Phase0.0°
Target Cancellation Offset180°
Online Control StatusADAPTING
05 / BMPA CONTROL ARCHITECTURE

Phase Array × AI Closed Loop

FUNCTIONAL FLOW
STAGE 01 · SENSE
Multi-Sensor Diagnostics

Estimate complex MHD mode amplitude and phase from simulated magnetic diagnostics, such as Mirnov coils.

STAGE 02 · ANALYZE
Phase & Growth-Rate Analysis

Monitor estimated phase, mode amplitude, disturbance response, and prediction residuals.

STAGE 03 · OPTIMIZE
Online Phase Search

Evaluate candidate phases and gains to minimize predicted loss and update the correction angle.

STAGE 04 · ACTUATE
Distributed Coil Array

Synthesize 12 phase-control channels and feed a bounded correction field back to the plasma model.

06 / EVENT TELEMETRY

Simulation Event Log

LAST 3 EVENTS
T+000.0Initial states synchronized: baseline and BMPA use identical starting modes and disturbance seeds.
07 / RESEARCH REVIEW PORTALINDEPENDENT RESEARCH PROPOSALKSTAR-RELEVANT TOPIC · NOT AFFILIATED

From interactive demo to testable research proposal.

This research-oriented extension documents the reduced model, implemented algorithms, limits of the comparison, and requirements for further validation. It is an independent proposal relevant to KSTAR's MHD research interests; it is not an official study, endorsement, or validation by KSTAR or the Korea Institute of Fusion Energy (KFE).

08 / TECHNICAL OVERVIEW

An explicit model and control law.

What the browser actually computes, with the model boundary stated clearly.

REDUCED COMPLEX-MODE MODEL
01 / MODEL

Complex mode state

a=x+iy; |a| is normalized mode amplitude. γ=0.08+mode term+0.40D−0.28B. ω is selected by mode. The 2/1, 3/2 and 1/1 choices are reduced-model labels.

02 / CONTROL

Prediction and online search

Predict a mode state and apply an approximately opposing phase. Five candidate offsets are evaluated online. This is a lightweight adaptive optimizer, not a trained neural network.

03 / ACTUATION

Idealized 12-channel array

Channel positions θⱼ=2πj/12 and phases φⱼ=φₜ−nθⱼ. Real coil transfer functions and electromagnetic coupling to a tokamak plasma are not modeled.

Current baseline

A simplified fixed-gain, fixed-offset feedback example. It does not represent a real KSTAR controller or a validated industrial PID/MPC implementation.

Current BMPA

Predictive phase correction with an idealized coil array. The baseline and BMPA have different observer filters and output limits, so this is not a fair performance benchmark yet.

Scientific scope: Mode amplitude, growth rate, phase error, heat-load proxy and stability score are internal normalized metrics. This is not a solver for MHD equilibrium/transport, real coil currents, divertor heat flux, D–T fusion yield, or neutron damage.
09 / VALIDATION PROTOCOL

A reproducible validation pathway.

Proposed gates before any claim of real tokamak-control performance.

PROPOSED · NOT COMPLETED
01
Equalized comparison conditions

Use matched modes, initial states, disturbances, observation delays, sensor noise, actuator limits and control-energy budgets across PID, MPC and BMPA.

02
Pre-registered test plan

Freeze seeds, pulse amplitudes/timing, test windows and allowed tuning ranges before examining performance outcomes.

03
Mode-control metrics

Compare peak amplitude, modal growth, recovery time after the pulse, phase estimation error, integrated control effort and failure rate.

04
Independent validation

Retest on held-out seeds and parameter sets. Move to offline data, then validated physics solvers, before considering any experimental integration.

GateComparator / inputRequired artifactsStatus
V0 · Web demonstratorFixed feedback vs BMPA in a reduced oscillatorSettings JSON, amplitude CSV, event logImplemented
V1 · Fair benchmarkPID / MPC / BMPA under equal sensor and actuator limitsMulti-seed statistics, control effort, failure ratesPlanned
V2 · Physics integrationValidated MHD solver and coil-response modelMode stability, delay/saturation sensitivityUnverified
V3 · Research diagnosticsOffline analysis of authorized KSTAR/public diagnosticsPhase-estimation errors, reproducibility reportNo data link
10 / REPRODUCIBILITY KIT

Export settings alongside the signals.

The experiment runs locally in your browser. Download configuration metadata and a proposed validation protocol.

LOCAL-FIRST · CLIENT-SIDE
Random seed24389
IntegratorEuler
Model step0.025
Nominal history interval0.2
Reproducibility boundary: Resetting the same browser code with the same inputs and seed reinitializes the deterministic RNG. Mid-run slider edits and pulse clicks are logged separately. The JSON is a review snapshot of settings and user actions, not a complete automated replay engine or third-party reproducibility certification.
11 / RESEARCH STATUS & EVIDENCE

Implemented, hypothesized, or unverified?

A transparent evidence ledger for technical reviewers.

EVIDENCE LEDGER · 2026
ItemEvidence availableStatus
Browser simulatorImplemented visualization, disturbance injection, CSV export and numerical oscillatorImplemented
Phase-array mode suppressionHypothesis testable in this reduced model; no demonstrated tokamak performanceHypothesis
Superiority over PID/MPCNot compared under equal sensor/actuator constraintsUnverified
KSTAR diagnostic / device integrationNo authorized data integration, institutional partnership or device experimentNot connected
Confinement, heat flux, neutron damageNo physics-based calculations or experimental validationUnverified
Public institutional context: Korea Institute of Fusion Energy (KFE) ↗This independent page does not imply KFE/KSTAR approval, sponsorship or affiliation.
12 / RESEARCH COLLABORATION ROADMAP

A staged proposal for KSTAR-relevant research.

Begin with a small, reviewable offline study rather than a request for device access.

Propose a review ↗
PHASE 01Technical review

Discuss physical meaning, observable mode phases and available coil authority

PHASE 02Offline reproduction

Compare PID, MPC and BMPA under jointly agreed constraints

PHASE 03Physics-model validation

Evaluate predictions using authorized diagnostics, MHD and coil responses

PHASE 04Next-step decision

Define any further study through the institution's own safety and review processes

13 / RESEARCH & BUSINESS PARTNERSHIP

Partner with us to
advance BMPA fusion control.

We welcome collaboration with research institutions, companies, engineering teams, and strategic partners — from validation of fusion-plasma phase-array control to co-development and commercialization.

01
Joint Research & ValidationMHD mode suppression, tokamak data integration, model and experiment validation
02
Technology & LicensingPhase-array actuation, control algorithms, simulation integration
03
Business & InvestmentProofs of concept, research equipment, commercialization and strategic collaboration

Please do not include trade secrets or confidential technical materials in your initial message. Your submission is processed by Formspree to enable us to respond to your inquiry.

Partnership & Research Inquiry

Share your proposed collaboration below. Fields marked * are required.