Evaluation of the thermal properties of Jatropha curcas L. kernels using near-infrared spectroscopy

• NIRS was used to evaluate thermal property of Jatropha kernels for the first time.
• NIR models for thermal conductivity at 40 °C and 100 °C were recommended.
• The models developed are useful for the drying process of Jatropha kernels.
• Bond vibration of oil, cellulose and hemicellulose of kernels effected the prediction.

The determination of the thermal properties, including thermal diffusivity, thermal conductivity, and specific heat, of Jatropha curcas L. kernels was conducted using near-infrared (NIR) spectroscopy. A total of 100 samples of whole kernels from green, yellow and black fruits and oven dried kernels were scanned using a Fourier transform NIR spectrometry over the range of 1,250,000–400,000 m−1. Models correlating the spectral data and the thermal properties measured by a reference method were developed by partial least squares regression and validated by test set validation. The results showed that for thermal diffusivity, thermal conductivity at 40 °C and 100 °C and specific heat at 40 °C and 100 °C, the coefficients of determination (R2) were 0.5968, 0.7592, 0.7509, 0.4211 and 0.6396%, respectively; the root mean square errors of prediction (RMSEP) were 1.1 × 10−6 m2 s−1, 0.0169 W m−1 °C−1, 0.0685 W m−1 °C−1, 5.88 kJ kg−1 °C−1 and 15.8 kJ kg−1 °C−1; the biases were −2.52 × 10−7 m2 s−1, 2.85 × 10−3 W m−1 °C−1, 2.52 × 10−2 W m−1 °C−1, 1.83 kJ kg−1 °C−1 and 4.69 kJ kg−1 °C−1; and the ratios of prediction to deviation (RPD) were 1.57, 2.04, 1.98, 1.28 and 1.64, respectively. These show the possibility of using NIR spectroscopy as an alternative method to estimate the properties of Jatropha kernels, especially the thermal conductivity at 40 °C and 100 °C.