Models for Predicting Flash Point of Penetration-Grade Bitumen Blended with Lighter Petroleum Products

摘要

Models based on Newton's divided difference interpolation polynomial, and curve-fitting by least-square method, are presented for predicting flash point of penetration-grade bitumen blended with kerosene, diesel, and lubricating oil, up to 10% (by volume) of each diluent per 100ml of pure bitumen. The least-square model under-predicts the flash point of pure bitumen in each blend, indicating that errors exist in the experimental data of the three bitumen blends. A modified least-square model which corrects the under-predicted flash point of pure bitumen is also presented. Predictions of Newton's model and the modified least-square model are similar and compare well with experimental data.