Real-time tracking of renewable carbon content with AI-aided approaches during co-processing of biofeedstocks

Decarbonization of the oil refining industry is essential for reducing carbon emissions and mitigating climate change. Co-processing bio feed at existing oil refineries is a promising strategy for achieving this goal. However, accurately quantifying the renewable carbon content of co-processed fuels can be challenging due to the complex process involved. Currently, it can only be achieved through expensive offline 14C measurements. To address this issue, with high-quality and large-scale commercial data, our study proposes a novel approach that utilizes data-driven methods to build inferential sensors, which can estimate the real-time renewable content of biofuel products. We have collected over 1,000,000 co-processing data points from refineries under different bio feed co-processing ratios and operational conditions—the largest dataset of its kind to our knowledge We use interpretable deep neural networks to select model inputs, then apply robust linear regression and bootstrapping techniques to estimate renewable content and confidence interval. Our method has been validated with four previous 14C measurements during co-processing at the fluid catalytic cracker. This novel methods provides a practical solution for the industry and policymakers to quantify renewable carbon content and accelerate the transition to a more sustainable energy system. •This study is innovative in using artificial intelligence for high accuracy (average error rate below 4%), real-time renewable carbon content tracking.•We collected over 1,000,000 co-processing data points under different bio feed co-processing ratios and operational conditions—the largest dataset of its kind to our knowledge.•The efficacy of the approach is validated through comparison with AMS 14C laboratory measurements.•The proposed method already offered substantial economic benefits, aiding both refineries in cost reduction and governments in policy formulation for renewable energy.