Feasibility and Performance of Blockchain Dual-purpose Mining-heating Systems: Comparative Numerical Analysis of Bang-bang, Hybrid, AI-based, and Traditional Electric Heating Models

Combining blockchain mining with building heating offers a viable way to increase energy efficiency and
accelerate decarbonization. A high-fidelity numerical assessment of dual-purpose mining-heating systems was carried out
using the Bitmain Antminer S21 Pro, comparing bang-bang (on/off), hybrid (modulating), AI-based (reinforcement learning),
and conventional electric resistance heating. Thermal and economic performance across representative room volumes
(60-340 m³ ) in the European comfort zone (20-23 °C) is captured by the simulation framework. All mining-heating techniques,
in most cases, produce net profitability while maintaining comfort. The AI-based controller outperformed traditional heuristics
by further minimizing cycling and preserving near-continuous comfort within 20-23 °C. Bang-bang generated the highest raw
profit (€108.41 at 160 m³) using the updated constants: network hashrate 915 EH/s, pool fee 2 %, electricity €0.0946/kWh,
BTC €102701.28, and block reward 3.125 BTC. With fewer cycles and lower duty (energy
2418.24 kWh, uptime 100.00 %, duty 95.69 %, time-in-band 99.94 %, cycles/day 0.0), Reinforcement Learning (RL) reached
€103.73 at 160 m³ (economic efficiency 1.4534). With only a slight profit trade-off, hybrids virtually eliminated cycling and
cut energy use by about 7.5 % (2336.71 vs. 2527.20 kWh). Despite being the most energy-efficient, conventional electric
heating always resulted in net operating losses (e.g., −€116.77 at 60 m³). This study establishes the groundwork for intelligent,
sustainable crypto-integrated heating and compares conventional and AI-integrated miner heating