Industrial control of volatile organic compound (VOC) emissions from medium-density fibreboard (MDF) production remains constrained by a shortage of compound-resolved evidence from full-scale plants, where wood furnish, amino resin chemistry, heat transfer, gas flow, and wet gas cleaning act simultaneously. Here, we analysed more than 20,000 synchronized operating records from a full-scale single-stage flash-tube MDF dryer at an industrial SWISS KRONO production line and linked total VOC (TVOC) measurements from flame ionization detection with Fourier-transform infrared speciation on the cleaned stack. Five compounds—α-pinene, 3-carene, limonene, methanol, and formaldehyde—accounted for more than 80% of the resolved VOC signal. Process–state contrasts showed that higher digester residence time, discharge screw speed, adhesive amount, urea amount, dryer inlet temperature, and scrubber–water temperature increased one or more representative compounds, whereas higher hardwood share, additional flue-gas supply, and higher scrubber–water pH decreased them. Limonene, methanol, and formaldehyde were substantially more process-sensitive than α-pinene. An exploratory decorrelation step further showed that a drying/throughput domain explained about half of the variability of the screened process space. The study therefore identifies the small set of compounds and operating domains that most strongly govern the cleaned dryer-stack signature and provides a mechanistically grounded prioritization framework for follow-up causal experiments, source apportionment, and emission-mitigation design in industrial MDF manufacture. Unlike product or chamber emission studies, this work links the compound-resolved FTIR/FID chemistry of the final cleaned industrial stack with synchronized production variables; it therefore addresses a scale-integration gap by transforming routine compliance-type exhaust monitoring into a process-diagnostic framework for ranking emission sources, abatement-sensitive variables, and mitigation experiments.
Nedić et al. (Mon,) studied this question.