| In many workplaces, material or product must be moved. Many times it is done with an automated material handling system, but most often, it is done by pushing or pulling a cart manually. In order for an employee to push or pull this cart safely throughout the day, push and/or pull forces may be measured. However, not all carts are designed the same and the material or load in the cart may change. Constantly measuring push/pull forces may become tedious. This study determined approximate push forces on two different carts, the Ergo Metro Cart and the Dynapace Lo-Temp Cart, using a statistical model that considers the independent variable, cart load weight. Considerations in selecting the statistical models were best fit of data based on residual analysis, lowest root mean square error and largest R2 value.;For predicting initial push forces for the Ergo Metro Cart, the quadratic model showed the best correlation between cart load weight and predicted push forces with an R2 of 0.8080. The root mean square error for the linear model was slightly less than the quadratic model; however, the quadratic model is more reliable. The sustained push force model involved a reciprocal transformation of the dependent variable. The quadratic model did have a slightly greater R2 value; however, the root mean square error of the reciprocal transformation was the smallest.;Similar to the initial push force model of the Ergo Metro Cart, the quadratic model provided the best fit for initial push forces on the Dynapace Lo-Temp Cart. Compared to other transformations, the root mean square error is the smallest and the R2 value is the highest (0.8710) for the quadratic model. For the sustained push force model on the Dynapace Lo-Temp Cart, the log transformation of the dependent variable, push forces, provided the best fit of data. The root mean square error is also the smallest and the R 2 value (0.9507) is the largest compared to the other transformations. |
