TY - JOUR T1 - Use of Control Charts and Regression Analysis in Cement Manufacturing in South Africa AU - Boer, Pieter Henk AU - Munapo, Elias AU - Mpeta, Kolentino JO - Journal of Engineering and Applied Sciences VL - 13 IS - 7 SP - 1732 EP - 1744 PY - 2018 DA - 2001/08/19 SN - 1816-949x DO - jeasci.2018.1732.1744 UR - https://makhillpublications.co/view-article.php?doi=jeasci.2018.1732.1744 KW - deleterious material KW -cement KW -CUSUM KW -Shewart KW -process capability KW -control AB - Continued monitoring of the mean strength of cement in manufacturing is essential. Cement strength is best measured after 7 and 28 days and has to meet the minimum standards as outlined in the South African Bureau of Standards. Early detection of low cement strength is important as continued production would result in large quantities of unusable cement and wastage to the environment. It is not known whether an

or Cumulative Sum (CUSUM) control chart will be feasible for cement manufacturing in South Africa. It is also not known from a South African context how 28 days mean strength could be predicted from variables associated with 7 days mean strength. Accurate predictions of this nature could safeguard against waiting for 28 days. Data was obtained from a cement factory in the North-West Province of South Africa from 17 Sept., 2015 to 31 May, 2016. This time-frame provided a data-set of 117 data points. Ethical approval was obtained from the North West University Ethical Committee (NWU-00433-16-S9). Data was analysed for normality and as descriptive statistics were calculated. Data was further assessed with x-Shewart and CUSUM control charts. Pearson correlation analysis was then applied on all variables (physical, chemical and strength). Stepwise linear regression analysis was used for the prediction model. A p<0.05 was considered statistically significant. All data was normally distributed with no missing values. The results of the current study demonstrated that it is feasible to use the

and CUSUM control charts. Out-of-control data points were readily detected by both

and CUSUM control charts. The two control charts reported similar findings. It is also shown that the process meets certain requirements or specifications as set out by the manufacturer. Lastly, 7 days strength and TiO₂ were the two independent variables that significantly predict 28 days with a coefficient of determination R² = 0.35 and SEE of 6.5 kPa (p<0.05). Effective use of control charts can lower the amount of chemical waste. The results of the current study demonstrated that it is feasible to use the

and CUSUM control chart to monitor and detect out-of-control values for cement strength (7 and 28 days) in cement manufacturing in South Africa. This would enable process engineers to timeously implement the necessary remedial actions plans to avoid waste which will eventually reach the environment. Moreover, 28 days mean strength can be significantly detected by early strength (7 days strength) and TiO to provide an additional 2 "warning system" for the early detection of a process running out-of-control. ER -