Abstract:
Investigating engine performance and emissions under varying
conditions and for different fuels is a costly and time-
consuming exercise. This may also require sophisticated
equipment which may not be readily available. In this study,
two Analytical Neural Network (ANN) models were developed
to predict diesel engine performance and emissions
respectively, when fuelled by Distilled Tyre Pyrolysis Oil
(DTPO). The models were based on back propagation learning
algorithm. The data used to train and test the ANN was
experimentally collected from a single cylinder four stroke
diesel engine operated at speed ranging from 800 to 3500 rpm.
The fuel blends contained 0 – 80% Distilled Tyre Pyrolysis Oil
(DTPO) in diesel fuel. The fuel blends and engine speed were
the input variables for each network. The performance of the
model was evaluated by comparing experimental and ANN
predicted results. The coefficient of determination (R2) was
found to be 0.9831, 0.9977, 0.9852, 0.9836, 0.9961, 0.9921 and
0.997 for Torque, Power, Brake Specific Fuel Consumption,
Peak pressure, HC, NOx and CO respectively. The Mean
Square Error between the measured and simulated values was
0.00396 for the engine performance model and 0.000163 for
the emissions model. It can be concluded that the engine
performance and emissions of a Diesel Engine running on
DTPO and its blends with diesel fuel can be reliably predicted
using Artificial Neural Network.