Integration of multi-criteria decision making and random forest regression for evaluating and predicting the optimal performance of elephant dung pellets enhanced with thermochemical by-product additives

Elephant dung (ED) is an abundant yet severely underutilized biomass resource due to its exceptionally high ash content and weak structural integrity when densified. This study establishes an integrated multi-domain decision-oriented framework for ED-based biomass pellet optimization. The framework quantitatively links pellet mechanics, fuel chemistry, heating value, and combustion kinetics within a unified strategy using thermochemical by-product additives, including bio-oil (BO), carbonization tar (CBT), and pyroligneous acid (PRA), with palm oil (PO) as a reference binder (5–15 wt%). All formulations exhibited statistically dominant additive effects across physical, proximate, ultimate, thermal, and combustion properties (ANOVA, P < 0.001; η 2 and ω 2 ≥ 0.99). Additive incorporation transformed ED from a low-grade residue into a technically viable upgraded solid fuel, increasing fixed carbon from 11.6% to 21.7%, reducing ash from 26.7% to 10.3%, and achieving pellet densities up to 1112 kg/m 3 with dimensional stability approaching 98.5%. Higher heating value peaked at 18.60 MJ/kg for 5% CBT, while 15% BO delivered superior operational performance, combining exceptional moisture durability (water resistance at 78.8%), rapid ignition (29 s), and high burning rate (0.9 g/min). A hybrid multi-criteria decision-making framework integrating the criteria importance through an intercriteria correlation weighted sum model with random forest regression produced a robust composite suitability index (R 2 = 0 . 913 , MAE = 0.068, RMSE = 0.076). This work advances solid biofuel development from empirical trial-and-error toward predictive, data-driven fuel engineering. Overall optimization identified 15% BO-modified pellets as the highest-performing formulation, with 5% CBT as a cost-efficient energy-dense alternative for circular bioenergy systems. • BO additive shows highest stability, fastest ignition, and top C i ranking. • CBT achieves balanced performance with strong energy yield and low cost. • Thermochemical by-products restore wood-like composition and fuel quality. • Machine learning validates Ci-based selection of high-efficiency additives. • Optimization analysis suggests BO–CBT synergy for superior pellet performance.