Skip to content
#explainable ai Open access

HCTGS v47 The Heat Refinery.Would you build your next AI campus where land, power, cooling, water and the permit could be secured before the first spade?

Oct 2026 · Zenodo (CERN European Organization for Nuclear Research) · 11 references
Thermodynamic and Exergetic Analyses of Power and Cooling Systems

Abstract

Abstract An oil refinery splits crude oil by boiling point and sells every fraction to the buyer who pays most for it; HCTGS (Hydro-Core Tower Gravity Systems) v47 applies the same principle to heat and joins the two halves of the series into one refinery: the thermal mineral and alloy cascade, which runs from 2,200 °C down to 350 °C, and the HCTGS tower, which uses the same heat again and again, from steam temperature down to near ambient, and turns it into fresh water and power on the way, with the sea, the air or the salt pit as the cold end, so that every stage takes its slice, passes the rest down and sells each kelvin once at its highest value. For new readers a Part 0 explains the tower, its path of the water, its pressure chain and its operating points, and treats the gain output ratio (GOR) as a design value that depends on heating-steam temperature, sink temperature, exchanger area and therefore on capital budget: tables give ideal, design and robust values for 70, 90 and 110 °C heating steam against 35 and 25 °C sinks (for example 17 effects and GOR 14.4 at 90 °C against a 35 °C sink, with a 10 K colder sink raising the design value by 12–40 %), together with delivery levels by technology and build form; condensation runs up to 99 % on a mist of the tower’s own distillate, and the water reaches farms and towns by U-pipeline and gravity from a hill cistern 50–100 m below the crown. The first contribution moves hydrogen to where it is made best: a solid oxide electrolyser at 700–850 °C fed with clean steam from the 350 °C stage and stack heat from the 1,000 °C stage, with a design target of 35–38 kWh of electricity per kilogram of hydrogen against 51–52 kWh for PEM and about 40 kWh for solid oxide electrolysis with external steam; a 100 MW module yields 21–22 kt of hydrogen and 21–23 t of oxygen per hour for the oxy-flames of the foundry, and the bromide-mediated paired electrosynthesis published by the Technion in August 2026 is added as an option that makes hydrogen together with an epoxide while the tower’s own bromide circulates as the mediator. The second contribution is a catalogue of every heat source the refinery can use, from electric arc, concentrated solar and geothermal fluids to biogas, industrial off-heat and the AI cluster, matched to the stage it feeds — including flue gas condensation at neighbouring plants, which gives about 140 MW of heat and 3,900 m³ of combustion water a day per GW of natural-gas fuel — with a site matrix for Iceland, the Gulf, North Africa, Chile, Central Europe and East Africa, three heat sets that close the 2.1–2.3 GW one tower needs for its full 1 million m³ a day (the heat of a 1 GW AI cluster at 55 °C makes up to a quarter of the water, solar thermal, geothermal fluid, industrial off-heat and the alloy cascade the rest), and a store ladder from −21 °C to 1,000 °C built largely from the cascade’s own salts: particle beds, a chloride melt of its own MgCl₂, KCl and NaCl, the NC-MX-1 rock bed and the NC-SPT-1 salt pit, whose heat-exchange capacity is sized here by loop metres. The edition then opens the alloy shelf: fifteen alloy families from the sea’s magnesium, lithium, calcium and boron combined with regional aluminium, zinc, titanium, zirconium and rare earths, an Australian route in which can-grade Al–Mg alloy replaces plastic bottles, a Moroccan route to magnesium phosphate cement, and the energy floor of magnesia by pyrohydrolysis at about 4.2 MWh per tonne, of which about 2.6 MWh return to the cascade at 80–110 °C. The foundry at the fence keeps the value on site, since a tonne of metal leaves as one tonne of finished alloy instead of 4–8 t of salt, and compact induction and oxy-flame furnaces feed their off-heat back into water; the scale is set by tower count, from 4.7 Mt of magnesium a year for ten towers to the gigatonne range for a full fleet, on a time axis of 10–50 years in which construction, transport, medicine and finally space become the large buyers, as aluminium once moved from a treasure beside the crown jewels to a metal for everyone. The third contribution places the AI cluster in the first build: a 10–50 MW brain that runs the refinery from commissioning, then a gigawatt partner that builds on a permit-ready plot inside the HCTGS field — zoned, assessed at plan level and connected in advance, because power, cooling, water and the environmental envelope stay within the field, on the pattern of the EU’s Net-Zero Acceleration Valleys — brings capital at the start, shares in the mineral margin and gains about USD 105 million a year in cooling value, while the tower turns its heat into 154,000–278,000 m³ of water a day, 3.6–6.4 times what an evaporatively cooled cluster of the same size would draw; cooling runs on four physically different paths, each N+1, and a neighbours’ share of water, power and compute, the renewal of the surrounding pipe network and new urban centres turn the cluster into a guest that leaves its region richer. Pension funds are proposed as first owners with a 20–30 % mandate, the fleet is planned with 150–300 % surplus in two tranches, and reinvested yield lets each cluster finance the next. Twelve work packages are matched to the industries that build them, with a pilot path that measures the inland modules in Graz and Styria and builds the first coastal tower on the northern Adriatic. The plant starts simple — an AI brain and the first water train in years 1–2 — and grows into the full refinery from year 4; a maturity table gives the readiness level of every module, and the economics separate the tower’s own asset from the partners’ plants: the tower breaks even at about USD 1.19/m³ on water and power alone and at about USD 0.60/m³ with magnesia and the AI cooling share. A closing Part VI turns the distillate into water by recipe: a soft Network Base of about 7 °dH for the mains and twenty-one mineral-rich recipes for people, herds, horses, pets, fish farms and fields, each with balanced ions, at its calcite balance point and within drinking-water limits, dosed at node stations, adapted waterworks and the bottling hall, with table water as the legal path, cascade magnesium as sacrificial anodes and ceramic gradient skins for intake, brine and heat-exchanger parts. This is a concept-stage open prior-art publication; all technical figures are design values calculated with steam tables and stated with their assumptions, to be measured in the pilot. CC BY-NC-ND 4.0.

View source

Similar papers

#artificial intelligence Conference Open access Apr 2020

ECCOLA - a Method for Implementing Ethically Aligned AI Systems

The method, ECCOLA, is presented, which aims at making the high-level AI ethics principles more practical, making it possible for developers to more easily implement them in practice.

Ville Vakkuri, Kai-Kristian Kemell, P. Abrahamsson · 64 citations · ⚡6
#computer vision Review Apr 2024

AI-powered Code Review with LLMs: Early Results

The goal is to not only refine the accuracy of the LLM-based tool but also to underscore its potential in streamlining the software development lifecycle through proactive code improvement and education.

Z. Rasheed, Malik Abdul Sami, Muhammad Waseem et al. · 62 citations · ⚡3
#computer vision Open access Mar 2024

LLM-based agents for automating the enhancement of user story quality: An early report

The use of large language models to automatically improve the user story quality in Austrian Post Group IT agile teams is explored, with a reference model for an Autonomous LLM-based Agent System developed and implemented at the company.

Zheying Zhang, M. Rayhan, Tomas Herda et al. · 48 citations · ⚡4
#computer vision Review Mar 2024

System for systematic literature review using multiple AI agents: Concept and an empirical evaluation

This paper introduces a novel multi-AI-agent system designed to fully automate SLRs, and demonstrates how it substantially reduces the time and effort traditionally required for SLRs while maintaining comprehensiveness and precision.

Abdul Malik Sami, Z. Rasheed, Kai-Kristian Kemell et al. · 44 citations · ⚡2
#computer vision Feb 2024

Can Large Language Models Serve as Data Analysts? A Multi-Agent Assisted Approach for Qualitative Data Analysis

The proposed LLM-based multi-agent system automates qualitative data analysis process, creating opportunities for researchers and practitioners, and future improvements focus on enhancing multilingual performance and integrating continuous expert feedback.

Z. Rasheed, Muhammad Waseem, Aakash Ahmad et al. · 41 citations
#artificial intelligence Conference Open access Jun 2018

The Key Concepts of Ethics of Artificial Intelligence

It is suggested that the focus on finding keywords is the first step in guiding and providing direction for future research in the AI ethics field.

Ville Vakkuri, P. Abrahamsson · 39 citations · ⚡2

Related blog posts

MIT News · Artificial Intelligence Oct 15, 2026

AI model decodes the language cells use to communicate

A new computational approach reveals how cells communicate to assemble into tissues. This could accelerate stem cell engineering, improve disease models, and inform therapies.

GPT-Lab Oct 14, 2026

Building trustworthy AI: The role of AI harnesses, intent contracts and evals

AI is moving beyond impressive demos. After World Summit AI 2026 in Amsterdam, GPT-Lab's Muhammad Waseem reflects on what organizations need to make AI agents reliable, responsible and genuinely useful in everyday work. The post Building trustworthy AI: The role of AI harnesses, intent contracts and evals appeared first on GPT-Lab.

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.