Automated quality control of diesel using near-infrared spectroscopy

Significance of the topic

Quality control of diesel fuel requires monitoring multiple physicochemical parameters that directly affect engine performance, emissions, safety and regulatory compliance. Conventional reference methods for cetane number/index, flash point, density, distillation (T95), and kinematic viscosity are accurate but time- and resource-intensive because they require different instruments, reagents and sample preparation steps. Developing a rapid, reagent-free, multi-parameter screening method based on near-infrared (NIR) spectroscopy can greatly increase laboratory throughput, reduce costs and lower hazards associated with flammable samples, while providing results that are fit-for-purpose for routine QC and process control.

Objectives and overview of the study

This application note evaluates the feasibility of using the OMNIS NIR Analyzer Liquid combined with automated sample handling (OMNIS Sample Robot M – NIR) to predict key diesel fuel quality parameters simultaneously. Target analytes were cetane number, cetane index, density, flash point, T95 (distillation cut at 95% recovery), and kinematic viscosity at 40 °C. The study demonstrates method speed, reproducibility under controlled temperature, model performance against reference methods and the operational advantage of full automation for routine workflows.

Methodology

Diesel samples were measured in transmission mode across the NIR spectral range 1000–2250 nm. Measurements used disposable borosilicate vials with an 8 mm optical pathlength and a controlled sample temperature of 30 °C to ensure spectral reproducibility and to follow applicable ASTM guidance for NIR measurements. The OMNIS Sample Robot M – NIR performed unattended sample presentation for up to three racks (supporting many vials), and closed vials minimized flammability risk. OMNIS Software handled data acquisition and multivariate calibration model development. The dataset was split with 75% of samples used for calibration and 25% retained as an external validation set. Model performance was evaluated using R2P (prediction coefficient), SEC, SECV and SEP as figures of merit.

Used instrumentation

Key hardware and software used in the study:

• OMNIS NIR Analyzer Liquid — transmission-mode NIR spectrometer for liquids (1000–2250 nm) with built-in sample temperature control (25–80 °C).
• OMNIS Sample Robot M – NIR — automated sample handler supporting up to three racks and disposable vials to enable high-throughput unattended operation.
• Disposable borosilicate vials, 8 mm pathlength (closed vials with stoppers) for safe transmission measurements.
• OMNIS Software (Quant Development license) — for data acquisition, model calibration and prediction deployment.
• OMNIS stand-alone software license option (2 instrument licenses) was noted for deployment scenarios.

Main results and discussion

All target parameters produced robust calibration models with strong agreement between NIR-predicted values and reference methods. Key performance metrics from the external validation are summarized below (units indicated):

• Cetane number: R2P = 0.922; SEC = 0.61; SECV = 0.65; SEP = 0.67.
• Cetane index: R2P = 0.903; SEC = 0.79; SECV = 0.80; SEP = 1.06.
• T95 (°C): R2P = 0.895; SEC = 3.42 °C; SECV = 3.69 °C; SEP = 3.75 °C.
• Viscosity (cSt): R2P = 0.962; SEC = 0.06 cSt; SECV = 0.06 cSt; SEP = 0.07 cSt.
• Flash point (°C): R2P = 0.973; SEC = 1.24 °C; SECV = 1.58 °C; SEP = 1.90 °C.
• Density (g·cm−3): R2P = 0.985; SEC = 0.0012 g·cm−3; SECV = 0.0014 g·cm−3; SEP = 0.0016 g·cm−3.

Discussion points:

• High R2P values (especially for density, flash point and viscosity) indicate very good predictive power for routine screening and process control. Density and viscosity, which have strong, direct influences on NIR absorbance via overtone and combination bands, showed the best precision.
• Cetane-related predictions (number and index) also showed strong correlations, although with somewhat larger prediction errors compared with physical properties; this reflects the indirect relationship between compositional spectral features and combustion-related properties.
• T95 predictions achieved useful precision for many QC applications, but larger uncertainty (≈3.7 °C SEP) suggests confirmatory distillation tests remain necessary for regulatory or forensic decisions requiring high accuracy.
• Using closed disposable vials and automated sample handling reduces operator exposure and flammability risk while enabling fast measurement cycles (measurements in seconds).

Benefits and practical applications

Practical advantages of the NIR approach demonstrated in this work include:

• Multi-parameter screening from a single, reagent-free spectral measurement, reducing instrument and labor demands.
• Rapid analysis times (spectra acquired in seconds) enabling high sample throughput in production or QC labs.
• Automated sample handling with closed vials enhances safety and allows 24/7 unattended operation when required.
• Pre-calibrated models (OMNIS pre-calibration for diesel referenced) allow immediate deployment and reduce model development time.
• Good analytical precision for density, viscosity and flash point supports process control and release testing; cetane and T95 predictions can serve as a fast screening tool to flag out-of-spec samples for follow-up with reference methods.

Future trends and applications

Near-term and longer-term opportunities to extend this approach include:

• Expanding calibration sets with broader feedstock variability and seasonal samples to improve robustness across fuel sources and additives.
• Implementing transfer learning and model update strategies to maintain prediction accuracy across instruments and sites.
• Combining NIR screening with orthogonal rapid assays (e.g., portable density or refractive index probes) in hybrid laboratory workflows to improve decision confidence.
• Embedding predictive models in process analytical technology (PAT) frameworks for on-line or at-line monitoring within fuel blending and production lines.
• Regulatory acceptance pathways that define when NIR-based screening can replace or reduce frequency of time-consuming reference tests for routine QC.

Conclusion

The application note demonstrates that transmission-mode NIR spectroscopy using the OMNIS NIR Analyzer Liquid, combined with automated sample handling, provides a fast, reproducible and safe method for simultaneous prediction of multiple diesel quality parameters. Model performance supports the method as an effective screening tool and high-throughput QC solution; some parameters (density, viscosity, flash point) achieve sufficiently low prediction errors for many routine applications, while cetane-related properties and T95 are suitable for rapid screening with confirmatory testing as required. The approach reduces analysis time, reagent usage and operator exposure, improving laboratory efficiency.

References

Metrohm. Automated quality control of diesel using near-infrared spectroscopy. Application Note AN-NIR-080. OMNIS NIR Analyzer Liquid and OMNIS Sample Robot M – NIR. Metrohm AG.