Cement Manufacturers Association (CMA)

18 External Standard (G-factor) Method: Calibrate the machine once with a pure standard. No mixing needed. Accuracy: ±0.5–2 %. Good for low-amorphous samples or research. Rietveld-PONKCS: Treats the broad “hump” from amorphous material as special fake phases. It can separate different types of glass (useful in LC³). Accuracy: ±1–2 %. Easy to automate and best for modern blended cements. ISM and PONKCS are mostly used in plants because they balance speed and accuracy. Table 2 Accuracy of Rietveld QXRD variants for industrial clinker Simple Plant Implementation Steps 1. Install a modern benchtop XRD machine with fast detector and automatic sample changer. 2. Take fresh clinker samples from the cooler and grind them finely (<10 µm). 3. Choose the best option (ISM with ZnO, G- factor, or PONKCS). 4. Press the powder into a pellet for consistent results. 5. Run a 5–15-minute scan. 6. Use ready-made software templates to automatically calculate phases and amorphous content. 7. Send results to the plant computer system and show them next to Bogue numbers on the control screen. 8. Adjust kiln or cooler settings when differences appear. 9. Check accuracy every three months and recalibrate the model once a year. The full process takes only 15–25 minutes per sample and gives repeatable results within ±0.5–1 % for main minerals. 3.4 Machine-Learning / AI Models AI models learn from thousands of real plant readings (oxides + temperatures + cooling data) combined with XRD results. A 2025 study achieved only 1.24 % error for alite — far better than Bogue. Implementation: Collect 1–2 years of plant data, train the model, install it in the control system for real-time predictions, and update it every few months. 3.5 Emerging Synergies: The True Digital Bogue The most powerful approach combines real- time XRD (Rietveld or PONKCS), thermodynamic modelling and AI. These “digital twins” can predict not only mineral amounts but also future strength. AI often shows that cooling rate is the most important factor. This system turns mineralogy from a slow lab check into instant, predictive control — helping plants reduce waste, save fuel and develop new low-CO 2 cements faster. Variant Major-phase error Amorphous error Repeatability Reproducibility Best Use Conventional Rietveld ±3–8 % Not measured ±1–2 % ±3–5 % Only for rough checks Internal Standard (ISM) ±1–3 % ±1–2.5 % ±0.5–1 % ±1–2 % Daily plant QC External (G-factor) ±0.5–2 % ±1–2 % ±0.5 % ±1–2 % Research, low amorphous PONKCS ±1–2 % ±1–2 % ±0.5–1 % ±1–2 % LC³ and blended cements