Prediction of the Wall Factor of Arbitrary Particle Settling through Various Fluid Media in a Cylindrical Tube Using Artificial Intelligence

Table 3

Fit of the experimental results to the predicted values of the comprehensive model.

Correlated	Sum of; squared error; (𝑄)	rms^a; deviation; (devrms)	Sum of squared; relative error; (𝑄rel)	Sum of; relative error; (𝛿)	Correlation; coefficient; (𝛾)	Range analysis: predicted data points; within specified range; of experimental results
						±15%	±10%	±7.5%	±5%
Training sets
Sphere	1.8282	0.1062	33.6267	27.854	0.4843	75.93	59.88	47.53	30.25
Cylinder	0.119	0.0251	0.7264	7.093	0.974	97.36	95.24	87.3	76.72
Rectangular prism	0.0507	0.0386	0.3259	2.1079	0.9792	91.18	85.29	79.41	61.77
All particles	1.9978	0.072	34.679	37.0549	0.8773	87.79	79.48	69.87	55.84
Test sets
Sphere	0.9038	0.1294	19.2326	11.7315	0.4675	72.22	66.67	51.85	35.19
Cylinder	0.0402	0.0255	0.2236	2.5431	0.9669	96.77	90.32	83.87	72.58
Rectangular prism	0.01124	0.0306	0.0634	0.6657	0.7792	91.67	91.67	83.33	58.33
All particles	0.9552	0.08639	19.5205	14.9403	0.7804	85.94	80.47	70.31	55.47

arms indicates root-mean square.