Open-source hardware that lives on open-source software. Updates are advancements. The repo is the growing web: each verified file, mix, ledger line, and skill is a node. No patents. Ever.
η = useful_joules / human_joules. Heat-engine η, act η, EROI, and sim scores are four different quantities (N14). We never claim greater than 100% thermodynamic efficiency.
Founded by: Jesse McMillen — Sikeston, Missouri Node Zero: southeast Missouri Archive: DOI 10.5281/zenodo.20549528 Contact: jrm8908@proton.me GitHub: https://github.com/jesseray718/openroot
- Open hardware library — CC-BY-SA 4.0 docs / GPL-3.0 code.
- Knowledge commons — handbook, skills, seeds, workflows. Anyone can run and improve them offline.
- PoPW / ACRE — claims minted only for verified physical work. No pre-mine. No speculation.
Core organs: AeroCement volumetric exchangers, AeroDisk solar stack-effect panels, ferrocement domes and double-catenary stressed-skin shells, Black Locust coppice RMH, UNE computational_flow, fractal lattice, PoPW ledger.
This README is the public map. Load-bearing numbers live in CLAIMS.md. What talks to what lives in INTEGRATION.md. What this tree is lives in MANIFEST.md.
Grid mechanical work from fuel is typically ~10–26% after plant, line, and motor conversions. That is an architecture tax, not a law of nature.
OpenRoot architecture:
- Capture solar (or coppice fire) as heat, not as grid electricity first.
- Move air by stack effect. Gravity is free. The sun (or the RMH) makes the density difference.
- Store in dirt and water. Two tanks. Never one.
- Use latent heat of water as a transport mechanism, not as a second sun.
- Extract shaft work from ΔT (Stirling + flywheel). Electricity only where electricity is required.
- Use heat as heat. Cold as cold. Work as work.
That is the whole argument. Everything else is organs, mix, and measurement.
Peak noon, 1 m² collector, AM1.5. MODEL, not a pad hang. One hour = 3.6 MJ of incident sun on that face.
Open-loop stack. Solar or RMH pays the chimney. Outdoor air and ground donate coolth. Heat used as heat. Cold used as cold. Shaft used as shaft. Electric only where electric is required.
| Book | Power | Joules in 1 peak hour | What it is |
|---|---|---|---|
| Sun on this face | 1000 W | 3.60 MJ | incident AM1.5 |
| Hot Tank A (collector heat parked) | 931 W | 3.35 MJ | sun that hit the face, stored as heat |
| Coolth into ground / Tank B (35°C outdoor → 12.8°C ground class) | 413 W | 1.49 MJ | weather + dirt, not extra sun |
| Parked heat + harvested coolth | 1344 W | 4.84 MJ | service ratio 1.34 vs sun-on-face |
| Latent at 0.5 kg/h | 314 W | 1.13 MJ | exported in vapor unless condensed — do not add as a second tank pile |
| Stirling + belt/alternator if you spend ΔT | ~60 W_e | 0.22 MJ electric | takes a slice of the hot book; do not add it on top of 3.35 MJ heat |
1.34 = (3.35 MJ heat + 1.49 MJ coolth) / 3.60 MJ sun-on-face. Multi-reservoir harvest. Not 134% efficient sunlight. Not 220%. Grade: MEASURED only after H-003.
| Path | What you hold after 1 peak hour on 1 m² | vs 3.60 MJ on the face |
|---|---|---|
| This node, grade-matched | 3.35 MJ heat + 1.49 MJ coolth | 1.34 service |
| Silicon PV module + inverter | 0.65–0.79 MJ electric (180–220 W) | 0.18–0.22 |
| That PV burned as resistance heat | 0.65–0.79 MJ heat | 0.18–0.22 |
| That PV into a heat pump COP 3–3.5 | 1.9–2.8 MJ heat or cold, not both from the same watt | 0.54–0.77 |
| That PV into a motor at 85% | 0.55–0.67 MJ shaft | 0.15–0.19 |
Plant 33–40% of fuel → electric. Wires keep ~94%. Motor keeps ~80% of that.
- Fuel → shaft at the house: ~0.10–0.26 of the primary joules (old coal + cheap motor at the low end; better plant + good motor at the high end).
- Fuel → resistance heat in the house: ~0.30–0.38 of the primary joules.
- Most household kWh are already heat. The tax is buying that heat as electrons.
PV is good at volts. Grid is good at shipping volts. Both force you to buy grade back. This node parks the sun as heat and mines the intake as cold on the same stack, with almost no conversion loss on those two streams. That is the consumer-side difference.
Do not print 1.34 as thermodynamic efficiency (N14). Do not add latent 314 W into the 4.84 MJ pile. 2°C / 35°F air is a target, not this table. Steam / RMH boiler is a separate organ and a separate book.
Opencell (AeroCement) is related to aircrete but does not collapse at critical foam mass.
A thixotropic surfactant gel locks the matrix when ordinary aircrete bubbles would pop and the pour would fall in. Mix direction (already documented): xanthan in alcohol + Dawn Ultra + water → gel. Nighthawkinlight method is the public reference. Ratio 1 part gel : 2 parts cement. Agitation entrains air. At the old collapse point the gel holds shape. Voids stay interconnected. The pour becomes a stable, breathable open-cell heat exchanger that can be cast, pumped, or printed.
Stator-motor mixing + closest packing. A rotor-stator drives bubble diameter down and bubble count up toward densest equal-sphere packing (HCP / FCC, coordination 12, packing fraction π/(3√2) ≈ 74%). Real foam is polydisperse, so true close-pack is a target, not a claim. Halving radius doubles surface area per volume. Finer cells raise capillary area and spread stress.
Alkali-resistant glass fiber, ≥20% zirconium. AR-GFRC in the paste. Combined with the open-cell matrix the mix can be lighter and stronger than ordinary concrete, and potentially pumpable over long distances with less labor.
Activated-carbon / charcoal load. Carbon in the matrix raises solar absorptivity and IR emissivity. Prototype charcoal-infused open-cell pours exist. α = 0.98 is a design target, not a published ASTM E903 hang.
That single material change turns a failed insulation foam into high-S/V thermal mass and structure.
| Property | Design range | Status |
|---|---|---|
| Compressive strength | H1 target ≥15 MPa after 21-day wet cure | OPEN until ASTM C39 |
| Dry density | 50–90 lb/ft³ foamed; AE-GFRC structural target ≤ 1,200 kg/m³ | OPEN |
| Open-cell porosity | 65–80% interconnected | OPEN |
| Pore diameter | 0.5–5 mm typical | OPEN |
| Thermal conductivity | 0.15–0.40 W/(m·K) vs ~1.7 for ordinary concrete | OPEN |
| Solar absorption α | target 0.98 carbon-doped | OPEN (ASTM E903 / C1371) |
| Permeability | 10⁻⁸–10⁻⁶ m² class | OPEN |
| Water absorption | 15–25% by weight (wet labyrinth) | OPEN |
| Internal S/V | target 500–2,000 m²/m³ in the filled labyrinth | OPEN |
Mix direction — NightHawkInLight gel (the public base). Credit: NightHawkInLight — Perfect Aircrete, Kitchen Ingredients. Channel: NightHawkInLight. OpenRoot did not invent this gel. We add carbon, AR-glass, and the 21-day wet clock on top of it.
Gel, per ~1 L batch (volume, not a certified mix): 15 g xanthan gum · splash of rubbing alcohol (or alcohol mouthwash) to pre-wet and kill clumps · 60 ml / ¼ cup Dawn Ultra (or a detergent with the same top ingredients) · 1 L water. Stir. Sit 5–10 min until it is a thick gel.
Pour: 1 part gel : 2 parts Portland cement by volume. It looks too dry, then goes liquid as bubbles form. More stirring → more air → lower density → more open cells. Less stirring → denser and stronger. Keep mixing past the closed-cell stage if you want interconnected voids for the labyrinth. Over-whip collapses the set.
OpenRoot deltas (optional, not in the kitchen video): activated carbon / charcoal in the paste for α · AR glass fiber ≥20% Zr, 2–5% by volume · rotor-stator if you are chasing finer cells · optional sand later (NightHawkInLight has tested sand:cement up to 2:1 by volume; that is his update, not a hang here).
One continuous lift. Cover and keep wet 21 days. N13 is not optional.
Do not substitute a pre-formed protein-foam / foam-generator yard mix and call it this gel. Those are two different air-entrainment paths.
Full notes: MATERIAL_SCIENCE_NOTES.md · wiki
Handbook seed: OPENROOT_HANDBOOK.md §4 (AERO-GFRC-001) · wiki
Same open-cell matrix, three jobs. Passive after construction. No grid fans. No pumps on the thermal loop.
- Heat Paint the open-cell surface black or load activated charcoal. Volumetric absorber. Air flows through the matrix, not over a plate. Phi-spiral / stack path. Dump heat into a copper coil inside an insulated ferrocement tank (Hot Tank A).
- Cold Same matrix, kept wet, air dried first by desiccant. Evaporative area is every pore, not only tunnel walls. Store cold in a second insulated ferrocement tank (Cold Tank B). Two tanks. Never combined.
- Work The ΔT drives a Stirling + flywheel for shaft work and a TEG only where electricity is actually required.
Hard geometry of the loop
Fresh air → desiccant → underground labyrinth FILLED SOLID with wet AeroCement (target 500–2,000 m²/m³) → Cold Tank B (radiative night-sky lid) → hot side / AeroDisk or RMH absorber → Hot Tank A → Stirling → back to desiccant.
At ~10 ft in temperate ground, soil is near 55°F year-round. That is the cold sink class. Output air aims at ground temperature on a hot day. Sub-wet-bulb numbers (35°F / 2°C) are an evaporative target, not a hang. Same physics class as a ground-source heat pump plus evaporative assist. We do not claim magic COP.
If humid air enters the wet labyrinth:
- Evaporative driving force collapses.
- Confined humidity rusts steel.
- Mold is a habitable-space failure.
Desiccant sits between hot-side outlet and labyrinth inlet. Regen from surplus heat, not from the useful cold stream.
Replace or back the solar absorber with a rocket mass heater fired on coppiced Black Locust. Locust coppice is easier to harvest than conventional firewood and regenerates. Same open-cell cascade: RMH supplies the hot end; labyrinth + wet AeroCement still do cold and storage; Stirling still takes the ΔT. Fuel is a renewable coppiced input — not a grid.
RMH + labyrinth comparison and H-003 live in projects/aerocement/ and the aerocement calc package (calc_solar uses 931 W/m² as the locked net-to-air design constant).
Peak AM1.5 design point, per m² of collector face. These are model numbers. They are not a pad measurement.
| Step | Quantity | Value | Grade |
|---|---|---|---|
| 1 | Incident solar | 1000 W | design AM1.5 |
| 2 | Absorbed at α=0.98 | 980 W | OPEN (needs spectral hang) |
| 3 | Net to airflow after ~5% face loss | 931 W | design constant used by aerocement_calc |
| 4 | Latent transport at 0.5 kg/h evaporation | ~314 W | transport, not new energy |
| 5 | First-Law budget from the sun | 931 W in = work + heat-to-ground + loss | conservation |
Do not add 931 + 314 and call it “1245 W created.” The 314 W figure is latent transport at a model evaporation rate. In a closed loop that would only be form change. This loop is open. Solar or RMH pays the stack pump. Stack-driven outdoor air then exchanges sensible and latent enthalpy with the wet labyrinth, the ground, and (if the lid is a sky-window emitter) the night sky. Those joules are real. They are not joules the collector face created.
Service from all reservoirs can exceed collector input. That is multi-reservoir harvest, not solar η > 100%.
Do not add heat + cooling + shaft work and call it 2197 W or “220% efficiency.” Moving 854 W of heat into ground mass is one physical stream. Calling that same stream “heating service” in winter and “cooling service” in summer is a service count. Service count is allowed in a grant packet if labeled as service. It is forbidden as thermodynamic efficiency (N14).
Honest split of the 931 W:
- Shaft work from a low-ΔT Stirling is a slice of the heat that actually crosses the engine, at a fraction of Carnot. Carnot for 350 K / 275 K is 21.4%. 60% of Carnot is 12.9% of the heat that goes through the engine — not 12.9% of 931 W unless you prove that much heat crosses the working fluid.
- Remainder is heat dumped to ground / Hot Tank A.
- Friction at 0.1 W is a placeholder, not a measured duct loss.
Passive transport ratio (heat moved / electrical watts on the loop) can be large because electrical watts on the loop are designed to be near zero. That ratio is not a heat-engine efficiency and must not be written as COP = 21,972 in a sentence that a reviewer will read as perpetual motion.
Replication math for builders: use 931 W/m² net-to-air and the two-tank geometry. Ignore headline multipliers.
Corrected ledger script: projects/aerocement/aerocement_ledger.py.
AeroDisk is not the underground labyrinth and not the RMH.
It is a panel / disk absorber for stack-effect solar air. Dark, high-S/V open-cell or plated faces sit in sun. Heated air rises through a designed throat. Chimney / stack effect is the pump. No fan.
Use AeroDisks:
- as the hot-side collector feeding Hot Tank A
- as roof or wall panels that preheat the cascade
- as ACRE-0001 “Seed Core + Aero-Disc absorption” artifacts
Do not pour the labyrinth and call it an AeroDisk. Disks are above-grade solar stack panels. The labyrinth is below-grade wet volumetric exchanger.
Order-of-magnitude stack drive (design, not hang): ρ(27°C) ≈ 1.177 kg/m³, ρ(77°C) ≈ 1.028 kg/m³. ΔP = g · H · (ρ_c − ρ_h) ≈ 14.6 Pa at H = 10 m. Mass flow from Q = ṁ c_p ΔT at 931 W and ΔT ≈ 50 K is ~0.018 kg/s per m² if that ΔT is actually achieved. Duct velocity target 1–3 m/s so loss stays under stack pressure.
Ledger pointer: seed-core/ledger/eta_ledger.jsonl
- Cardboard geodesic panels (acetone + silicone treated, flanged triangles): 1v emergency shelter to large geodesic.
- Double-catenary stressed-skin ferrocement shells — pure compression geometry. AeroCement as core inside the skins.
- Thin ferrocement tanks and domes as the insulated vessels on both sides of the cascade.
- GFRC skin: 3/8–1/2 in, ~5% AR glass, 6–8 ksi class compressive as a design range.
- Ferrocement tanks: 1–1.5 in mortar, 2–3 layers 1 in mesh, pneumatic hog-ring ties.
Monolithic stress-skin catenary arch: self-supporting shell, no frame. The open-cell core is fill, not the compression skin.
Not a thermal claim. Separate hangs.
- Black Locust keystone guild — coppice fuel + nitrogen + fence + bee forage. Same species that fires the RMH.
- Vertical quail towers — high protein per footprint, manure to aquaponics.
- Ferrocement aquaponics — closed loop on the same tank skill as Cold Tank B / Hot Tank A.
- Heirloom seed bank — offline, local, forkable.
Food joules do not get added to the 931 W solar ledger.
Primary material (charcoal open-cell) is prototyped. These supporting claims stay OPEN until the named test exists.
| ID | Claim | Falsifier |
|---|---|---|
| H1 | AE-GFRC with ≥20% Zr-class binder sub, air voids as aggregate, ≥15 MPa and ASTM C1550 toughness at dry density ≤ 1,200 kg/m³, pumpable 1,609 m without segregation | fails C39 / C1550 / C1716 at that density or distance |
| H2 | Spherical voids beat mined lightweight aggregate on strength-to-weight at equal or lower cement | measured specific strength below LWAC control |
| H3 | Purpose-built pneumatic placement ≥300 m³ per operator-hour | sustained rate stays in the 30–50 m³/h foam-pump class |
| H4 | N pumps place volume V in T = V/(300N) with zero hand placement | labor or blockage dominates |
| H5 | Delta-T vehicle: drag-cooled open-cell radiator sustains a useful Stirling ΔT | wind-tunnel + bench Stirling cannot hold work-positive ΔT |
| H6 | Purpose-built AE-GFRC pump ≥300 m³/h at ≤30 bar | foam collapses under sustained pressure |
| H-003 | Instrumented solar + labyrinth node in Sikeston climate matches the 931 W/m² class closely enough to beat a measured electrical baseline on η_act | pad sensors show otherwise |
Nulls H01 / H05 / H06 stay published next to the claims. A rejected hang is data.
Required tests: ASTM C39, C1550, C138, C231/C457, C1716, E903, C1371, wind tunnel 5/15/25 m/s, Stirling bench, 100→500→1000→1609 m pump trial.
H5 numbers in old drafts (180–300 kW, 1500 kW radiator exchange) are upper-bound arithmetic, not a vehicle. Do not reprint them as performance.
Centralized economic, regulatory, and professional power repeatedly creates or intensifies problems that already have simple, low-tech, decentralized answers — then obstructs those answers. OpenRoot’s counter is not a complaint. It is dependency-free tooling:
- Benefit measured at the recipient.
- Lowest node first (N07).
- Unnecessary suffering is the error signal.
- Cooperation voluntary.
- Knowledge forkable and offline.
Work is measured in joules. Verified physical work mints ACRE claims. Two independent validators. Replicating a known node in an already-validated climate earns 0 new knowledge mint.
Building the first node in a new climate zone, fixing a documented flaw, shipping a new tool, or writing a new skill doc is mintable. Copying node #47 in a climate already validated is real work and zero new-knowledge mint.
ACRE token deployment is conceptual. No pre-mine. No airdrop. Spec: tokens/ACRE_SPECIFICATION.md.
Release: v1.2.1-popw-ledger
Bounty board is a map of unmet needs, not a live payroll.
| Node | Role |
|---|---|
| Samsung A15 + Termux | Governor, file bus, light inference, ACRE claims |
| OptiPlex 3060 | nomic-embed-text :11434 · qwen2.5-coder :8080 · FTS5-first SQLite RAG |
| Syncthing | Phone ↔ box. No unique-ID theatre when folders are Up to Date |
Related public trees: openroot · openroot-foundation · openroot-thesis · wisdom-scaffold · agape-une / une · black-locust-rmh · agape-primitives
Phone-first live paths (absolute):
/data/data/com.termux/files/home/openroot/storage/emulated/0/openroot- Box:
/home/jesse/openroot - Calc:
/data/data/com.termux/files/home/aerocement/aerocement_calc/
- Never claim greater than 100% thermodynamic efficiency.
- 21-day wet cure — non-negotiable (N13).
- Tunnel FILLED SOLID — never lined or walled.
- Two separate tanks — never combined.
- Desiccant at intake only.
- AeroDisk ≠ labyrinth ≠ RMH. Name the organ you are building.
- Latent heat is transport. Do not add it to solar input.
- Service-count ≠ First Law. Do not sum heat+cold+work against one watt of sun and call it efficiency.
- No patents. Ever.
- Failures are data — document honestly.
- Serve the least first.
Days 1–3. Survey south aperture. Mark 5–15° tilt. Excavate labyrinth trench ~10 ft. Pour porous floor. Sump at low point. Start 21-day clock. Days 4–24. Walls and baffles of wet-capable AeroCement. Capillary / drip feed. Desiccant housing. Ferrocement cold tank + copper coil. R-19 between hot and cold ducts. Backfill with no voids. Cast AeroDisk / panel faces on the same clock. Days 4–14 concurrent. Frame panels. Series stack if you need chimney height. Connect: panel → desiccant → Stirling hot → labyrinth → cold coil → return. Days 25–35. Stirling on the real ΔT, belt, optional alternator / TEG. Measure RPM, torque, ΔT_hot, ΔT_cold, airflow. No brochure η. Days 25–50. Dome pad. Treated cardboard 1v (or specified frequency). Flanged click. AeroCement or ferrocement skin. 21-day cure per lift.
Instrument before you advertise. Cheap sensors beat another manifesto.
Mesh spine (same thought, not a second manifesto): SPINE.md. All cascade numbers are MODEL until H-003.
Sample AeroCement blocks exist. No instrumented Node Zero has been poured.
Workshop offer: materials-at-cost in southeast Missouri for anyone who will measure and share results.
- Zenodo: https://doi.org/10.5281/zenodo.20549528
- IPFS CID:
QmcMjnAVN9FbQ77VbwMPMCteb93U7W4REdZmZbPqoMBE4F
Skills: library/kai-sandbox/skills/ · wiki
Workflows: workflow/
Handbook: OPENROOT_HANDBOOK.md · wiki
There shall come a time when the earth is weeping and the animals are suffering, and from all corners of the earth shall come a tribe of all colors, classes, and creeds, and through their actions they shall make the earth green again. — Hopi prophecy
MANIFEST.md— what this tree isCLAIMS.md— every load-bearing number, gradedINTEGRATION.md— sister trees and live pathsINTEGRATION_CHECKLIST.md— four boxes
SimpleX: https://smp9.simplex.im/a#vklZrSjZTQdgXBqW_sLK1h5FeajDoa7wTaSWGSw62Sw
| Solana | 3fF26gcj1ednMUASxJxo1dt5rQ2ZegXbH7k4ynJazerk | | Bitcoin | bc1qq69dze04yul5cl5lgv3hakg4scxfzq3swje6ey | | Ethereum | 0x8eA4dBF495ef2Ab6E4371C75060390563b79c138 |
The project succeeds when it no longer needs Jesse McMillen.
One pour. One node. One warrior at a time.
CC-BY-SA 4.0 (Hardware) | GPL v3 (Software) | No Patents. Ever.