Journal of China University of Science and Technology Vol.57-2013.10 Improving the Failure Rate of Holding Pressure of Gas Pipeline by 6σ Method 品管控制氣體管路保壓失敗率之研究 1 鐘啟榮 2 李世文 3 陳德請 Chung-Chi Jung 1, Shih-Wen Lee2, Der-Chin Chen 3 1 逢甲大學資訊電機碩士在職專班 2 中華科技大學電機工程系助理教授 3 逢甲大學電機工程系教授 1,3 Department of Electrical Engineering, Feng Chia University 2 Department of Electrical Engineering, China University of Science and Technology 摘要 本研究利用 SIPOC 模組建立實驗流程 魚骨圖與矩陣圖來找出管路保壓失敗的原因, 及應用 DMAIC 手法改善管路保壓失敗率, 訂出最佳更換鋼瓶 SOP 及工具選擇 因氣體壓力會受溫度影響, 使用給呂薩克定律來計算出正確壓力值, 以確保管路保壓壓力的正確性, 並進一步使用六標準差分析實驗數據可靠性及直線迴歸分析証實 SOP 流程的改善效果 此方法經實驗測試, 可以使氣體管路保壓失敗率由 30.5% 降低至 8.3% 關鍵字 : 六標準差 迴歸分析 ABSTRACT This research applies SIPOC (suppliers, inputs, process, outputs, and customers) model to build the experimental processes, the fishbone diagram as well as the matrix diagram to find the causes of failure and applies DMAIC method to improve the failure rate of pipeline, respectively. Then, the optimal SOP of the replacement of cylinders and the selection of the tools are established. The gas pressure of pipeline is affected by temperature, and the Gay-Lussac's law is used to calculate the correct pressure value in order to ensure the validity of the holding pressure in the pipeline. Furthermore, Six Sigma tools are applied to analyze the reliability of the data and linear regression analysis, to assure the improvement of SOP. This method is tested by experiment, results show the failure rate of the holding pressure of gas pipeline can be reduced from 33
Improving the Failure Rate of Holding Pressure of Gas Pipeline by 6σ Method 30.5% to 8.3%. Keywords: SIPOC, DMAIC, 6 Sigma, SOP, regression analysis 1. INTRODUCTION So far, a variety of dangerous gases in the process of the panel plant are used (such as SiH4, PH3, NH3, CL2... etc.) [1]. Gas has to be replaced when it is used up. Now, the gas cylinder is not replaced through the automatic operation, but through manual work. Replacement of gas cylinder has some certain risks (such as leakages of the poisonous and flammable gases, etc.) [2]. Main factors affecting those risks include the way of operation, the tools to be used, and the surrounding temperature. 2. THE PURPOSES After running out of special gas, it is necessary to do the holding pressure test between the pipe and the gas cylinder while changing the cylinder. There would be generating the risk of leakage if the test is not executed. The pressure of gas will be affected by the temperature within one hour of the holding pressure. During the time, it is hard to judge consistency of pipeline pressure. However, the methods and the tools to change the gas cylinder would be one of the factors for holding pressure test successfully. In this research, the SOP are improved and more efficient. 3. THE PROCEURES AND METHODS The method to be used is SIPOC model [3] which defines the process of a special gas cylinder is provided on the spot. Also, the analytical results of the fishbone diagram [4] and the matrix diagram are used to find out the factors which cause the failure to hold the pressure of pipeline after changing the gas cylinder. Then, DMAIC [5] is applied to improve process. Finally, Gay-Lussac's law, six standard deviation [6] and regression analysis are used to calculate the differences and relevance of the failure rate. 3.1 Introduction 6 Sigma DMAIC DMAIC is a kind of structural and data-based process of improvement on the 34
Journal of China University of Science and Technology Vol.57-2013.10 problems [7] [8]. There are five steps described below: 1. Definition: After replacing the cylinder, there is no leakage at the junction of gas pipeline. 2. Measurement: Why does it often fail when gas pipeline is holding pressure? 3. Analysis: If there is no problem about the pipeline, there may be improper operation procedure and the tools used, or the defective material. 4. Improvement: Build up a set of SOP set and proper tools, and change the supplier. 5. Control: Ask the operators to follow up the SOP procedures. 3.2 The procedure of holding pressure in the gas pipeline The test of leakage of the holding pressure is due to the joint points between the cylinder and the pipeline. Figure 3.1 shows the flowchart of the holding pressure in the pipeline. Holding pressure: providing about 1500psi by the high-pressure to the junction points between cylinder and pipeline Cylinder cabinet (PLC + pressure regulator) New cylind er Joint points between cylinder and pipeline High pressure N 2 cylinders After providing the pressure, shut down High pressure N 2 immediately. Low pressure N 2 cylinders Low pressure N 2 provides positive pressure during the replacement of the cylinder. Figure 3.1 Flowchart of the holding pressure in the pipeline 3.3 The Standard Operation Procedures The standard operation procedures (SOP) of replacing the cylinder are described as follows: 1. Make sure the value of the exhaust has reached the requirement. (Above 0.25 inch for H 2 O). 35
Improving the Failure Rate of Holding Pressure of Gas Pipeline by 6σ Method 2. Make sure to replace the gas cylinder cabinets, and then finish the operation of Pre-Purge. 3. Make sure the same name tag of the old cylinder as that of the cylinder cabinet. 4. Wear breathing apparatus (BA) and carry He gas detector, portable gas detectors, torque wrenches, gaskets. It is more necessary to wear fire suit, fire gloves and latex gloves while replacing SiH 4 cylinder. 5. Confirm that the lights on the panel of the gas cabinet are in the normal conditions. 6. Confirm that the pressure of high cylinder He is 1400 psi; the pressures of low pressure and the purge N 2 cylinders are 100 psi. 7. Confirm that the pressure of low pressure transducer (LPT) in the old cylinder is -10 psi. 8. Confirm that the pressure of high pressure transducer (HPT) in the old cylinder is around 110 psi. 9. Confirm that the pneumatic or manual valve of the old cylinder, needing to remove the gas first, is locked and tear off the label of in use. 10. Confirm that the red sleeve in the device of the protection from being removed, so-called Auto Guard, is not retracted. 11. Remove the connector of the cylinder (the instantaneous value after removing the cap of SiH 4 cylinder is 5 ppm). 12. Lock up the caps of the pigtailed connector in the cylinder, and move to the temporary area/frame. 13. Check the exterior of cylinder in good shape. 14. Place the cylinder in a normal position and align it with the pigtail s connector. Check the error is less than 1cm. 15. Remove the film on cylinder while on the shelf and use portable tester to check the every part, including the bottle of the cylinder, and make sure there is no leakage, then record the reading in ppm. 16. Confirm that the new cylinder has torn off the label of "charge the gas tank " or "filling the gas tank". 17. Remove the gaskets and check whether there are any scratches on the surfaces or not. (if they are normal, marked V ; if abnormal, marked X ). 18. Before taking off the gaskets and make sure there are no scratches on the surfaces, loosen the cap of connector with pigtail, replace the gasket, and then 36
Journal of China University of Science and Technology Vol.57-2013.10 use hand to tighten up the connector. After the supervisor confirms HPT is more than 60 psi, then lock up with a torque wrench tight twice. 19. Complete the automatic packing procedures, record the pressure in the primary transducer when the procedures of holding pressure begin. 20. Record the pressure in the primary transducer when the procedures of holding pressure finish, and compare with the value recorded in item 19. When the error between them is within 3 psi, it is normal. 3.4 Applications of fishbone and matrix diagrams At present, the failure rate of holding pressure in the gas pipeline is extremely high. So, it is the goal to find solutions and reduce the failure rate. The fishbone diagrams are applied and the factors are collected to analyze by the matrix data analysis chart [9]. Score each factor by its importance. The higher score means it is more important. Finally, choose the highest eight scores to experiment and analyze [10]. Figure 3-2 shows the fishbone diagram of gas pipeline holding pressure; Table 3-1 shows the results of matrix data analysis. Figure 3-2 Fishbone diagram of gas pipeline holding pressure Table 3-1 Results of the matrix data analysis 37
Improving the Failure Rate of Holding Pressure of Gas Pipeline by 6σ Method Item Various factors 1 2 3 4 5 Failure to SOP process operation Temperature effect Latex gloves dirty Gasket Positioning is not installed The use of a magnifying glass Security (Prevent gas leakage) Feasibility (Effectively reduce the leakage rate) Timeliness (Persistency) Total Weight ratio ( confirm the importance of the individual experiment) 6 5 5 9 8 8 128 5 6 6 90 4 6 6 122 7 6 6 96 4 6 6 117 6 Uneven ground 8 4 4 108 7 Pounds of torque wrench 8 6 5 103 Table 3-1 Results of the matrix data analysis (continued) 8 Gasket brands 7 5 5 92 9 Gas retention period has 2 3 2 37 expired 10 Dust 3 1 1 28 11 No regular maintenance of 2 2 2 32 equipment 12 Cylinder itself abnormal joints 3 1 1 28 13 Parameter setting error 6 1 1 46 14 Pigtail Connector 5 1 1 40 Specifications 15 Cylinder Head Specifications 5 1 1 40 3.5 Application of DMAIC approach 38
Journal of China University of Science and Technology Vol.57-2013.10 packing [11] [12]. The DMAIC approach is applied to improve the failure rate of pipeline (1) The phase of Define After replacing the cylinder, there is no leakage at the junction of gas pipeline. (2) The phase of Measure Why does the holding pressure of gas pipeline fail so often? If there are defects of the cylinder when purchasing, ask thee suppliers to replace new one in order to reduce the problem of poor quality of the cylinders. However, the main reasons of the failure are the workers do not follow up the SOP or the SOP is incorrect. Those are the issues to be revised now. (3) The phase of Analyze Digressing from the problem of cylinder, the other is the problem occurred in the worker s operation according to the current SOP. In order to verify the pressure in one hour without leakage and change with the temperature, Gay-Lussac's law [5] can be applied, in a given volume, the pressure is proportional to the temperature. The higher the temperature is, the greater the gas pressure is. P 1/ T 1 = P 2 /T 2 (3-1) It can make sure that the pressures before and after the holding pressure test are not the same. This problem is affected by temperature. There is really no leak. In Table 3-2, the gas under the tests is SiH 4 and there are the notations of the columns described below: A: Temperature before regulation ( ). B: Pressure after regulation (psi). C: Temperature after holding pressure ( ). D: Pressure affected by Temperature after holding pressure (psi), (Pressure is computed by using P 1 /T 1 = P 2 /T 2.) E: Pressure after holding pressure (psi) F: Pass (Y) or Fail (N). G: The difference between C and A. (G=C-A) H: The difference between D and E. (H=D-E) The pressure after regulation of the first example in the tests is 1424.5 psi, the 39
Improving the Failure Rate of Holding Pressure of Gas Pipeline by 6σ Method temperatures before and after the test are around 29 degrees. According to the eq.(3-1), P 1 /T 1 = P 2 /T 2, then P 2 = 1424.5 (psi); the calculated value(in column D) is almost the same as the value of P 2 = 1424.5; However, the actual recorded value(in column E) is 1420.2 (psi), and the difference value of the pressures between them is -4.3 (=1420.2-1424.5). From above computation, there is a little leakage in the pipeline. If the setting error in a company is within ± 3 psi, the holding pressure test is Fail. Table 3-2 Temperature effect of holding pressure item A B C D E F G H 1 29 1424.5 29 1424.5 1420.2 N 0 4.3 2 30 1450 30 1450 1446.1 N 0 3.9 3 29 1430.3 29 1430.3 1426.8 N 0 3.5 4 31 1421.4 31 1421.4 1421 Y 0 0.4 5 31 1407.5 31 1407.5 1403.2 N 0 4.3 6 31 1429.8 31 1429.8 1429.7 Y 0 0.1 7 31 1434.8 31 1434.8 1433.9 Y 0 0.9 8 32 1427.6 32 1427.6 1427.6 Y 0 0 9 29 1463.7 29 1463.7 1462.9 Y 0 0.8 10 28 1436.7 28 1436.7 1436 Y 0 0.7 11 31 1446.1 31 1446.1 1445.1 Y 0 1 12 29 1575.3 29 1575.3 1575.1 Y 0 0.2 13 29 1569.6 29 1569.6 1564.2 N 0 5.4 14 31 1570.1 31 1570.1 1563.1 N 0 7 15 29 1470 30 1474.8676 1466.9 N 1 7.96755 16 28.5 1416.1 28.5 1416.1 1413.2 Y 0 2.9 17 28.6 1417.2 28.5 1416.7301 1414.5 Y -0.1 2.230106 18 28 1503.2 28.1 1503.6994 1501.6 Y 0.1 2.099402 19 28.2 1490 28 1489.0106 1488.2 Y -0.2 0.810624 20 28.2 1485.6 28 1484.6135 1481.2 N -0.2 3.413546 21 28 1482 28.1 1482.4924 1479.9 Y 0.1 2.592359 22 28.1 1477.1 28.1 1477.1 1475.1 Y 0 2 23 28.5 1450 28.6 1450.4809 1449 Y 0.1 1.480929 24 28.6 1433.1 28.6 1433.1 1430 N 0 3.1 40
Journal of China University of Science and Technology Vol.57-2013.10 Table 3-2 Temperature effect of holding pressure (continued) 25 28.6 1428.1 28.5 1427.6265 1425.3 Y -0.1 2.326492 26 28 1482.8 28 1482.8 1480 Y 0 2.8 27 28.1 1482.8 28.1 1482.8 1480 Y 0 2.8 28 28 1470 28 1470 1468 Y 0 2 29 28.2 1465.5 28.2 1465.5 1462 N 0 3.5 30 28.1 1460.1 28.1 1460.1 1457.3 Y 0 2.8 31 28.5 1458.1 28.5 1458.1 1455.6 Y 0 2.5 32 28 1550.1 28 1550.1 1545 N 0 5.1 33 28 1548.6 28 1548.6 1545.9 Y 0 2.7 34 28.3 1542.3 28 1540.7644 1540 Y -0.3 0.764354 35 28.5 1530.5 28.2 1528.9771 1528.1 Y -0.3 0.877114 36 27.5 1470.1 27.5 1470.1 1468.5 Y 0 1.6 From the calculations, there are totally 36 times of the test, 25 times of Pass and 11 times of Fail. Through the conversion of Table 3-2, there are still 7 times of Fail in the 19 times of test. The 4 th and 18 th procedures in the SOP are the focal points to be discussed. In the 4 th procedure, although there are safe apparatuses to wear, there are still no tools to check the source of pollution. Therefore, it will cause to increase the failure rate severely. As to the 18 th procedure, the problem about whether the gaskets are properly aligned is still mostly considered. (4) The phase of Improve: As to the 4th procedure in the SOP, some inspection tools are required, such as, level instrument, flashlight, magnifying glass, clean cloth inspection tools, etc. The horizontal ruler can check out the tilt angle with the ground when the cylinder is connected. Use the flashlight and magnifying glass to examine the interior of the pigtail on which there are dirty points or scratches, and then remove them with a clean cloth after taking off the gasket. As for the 8 th procedure in the SOP, replace latex gloves before taking gasket. Make sure the gaskets to be replaced have no scratches. Then, loosen the cap of the pigtailed connector to change the gaskets. After changing gaskets, connect it with torque wrench tightly twice after the pressure at the high pressure transducer is in access of 60 psi. The newly added tools and their applications are denoted as follows in Table 3-3. (5) The phase of Control 41
Improving the Failure Rate of Holding Pressure of Gas Pipeline by 6σ Method On the phase of Control, there are many concerns about the operations of replacing cylinder, data analysis about the holding pressure of the pipeline, modifications of the SOP, setting up Check Lists to comply with the operation requirements and holding the regular program about the education and training for the operators. Table 3-3 The applications of the new tools Tool level instrument flashlight, magnifying glass cleaning clothes latex gloves Application Observe the tilt angle between the connector of the cylinder and the ground. Observe interior dirt or scratches in the pigtail. Wipe the inside to prevent dirt in the pigtail Prevent the pollution before taking the gloves to replace the gaskets. 4. IMPROVEMENTS AND EXPERIMENTAL RESULTS Gay-Lussac's law is applied to calculate the holding pressure which is affected by temperature. And check whether the pressure is the same as that before the holding test. In order to eliminate the effect of the temperature, the effectiveness of improvement of SOP is still valid. In the 4 th procedure of SOP, there are, originally, level instrument is not used to observe the displacement of the connector. As well, magnifying glass and flashlight are not used to observe the damage of the pigtail, and also, the clean cloth is not used to wipe the pigtail after removing gasket. In this step, it is easily to cause contamination. And only one small particle source is easily to bring into a leakage. Replacement of latex gloves is a necessary operation, although there are a lot of people may think it is not necessary. Even if there is a small source of pollution in the attached gasket, it will cause the leakage. And new latex gloves are also the sources of pollution. Lock the cylinder and the pigtail of the connector with hands and let the hands not leave the connector is to avoid the misplacement caused by the relaxation of hands after finishing the alignment. Experiments are executed through step-by-step analysis of the process. In SOP, the procedures from the 3 rd to the 8 th are improved one by one. The matrix analysis is statistically applied to each improvement to assure the effectiveness after computing the 42
Journal of China University of Science and Technology Vol.57-2013.10 failure rate. The improvements of procedures and the experimental results are recorded. Six-Sigma analysis is used to calculate the reliability of each improvement and linear regression analysis is applied to the relationship of failure rate in the holding pressure. After 36 times of experiments, if the data of failure rate is within the standard deviation and there are no abnormal effects, it represents the data can be adapted. By Six-Sigma analysis and linear regression analysis to compute the percentage of failure rate in holding pressure, it reduces 3.37% for each improvement. Table 4-1 shows the values of means and standard deviations before and after improvements by means of Six Sigma analysis. Table 4-2 shows the information between the times of SOP improvement and failure rate. Figure 4-1 shows the plots of the regression line and the residual values. By the above analysis of the improvement of SOP, there are only three times of failure in the 36 times of replacement of the cylinders. The improvements in SOP are valid for the procedures in the replacement of cylinders. Finally, the defective ratio is applied to compare the results before and after improvement. The ratio is calculated as follows: defective ratio = Number of the defective products/total number of the products 100% Before the improvements, there are 11 times of failure in the 36 times of tests, defective ratio = (11/36) 100% = 30.5 %. After the improvements, there are 3 times of failure in the 36 times of tests, defective ratio = (3/36) 100% = 8.3 %. From above computations, failure rate after improvements of the DMAIC in SOP, ugh has been significantly reduced from 30.5% to 8.3%. Table 4-1 Means and standard deviations of the pressure difference before and after improvements by Six-Sigma analysis Case A(*) B(*) C(*) D(*) E(*) F(*) G(*) parameters Mean 2.5240 2.2054 1.5748 1.5082 1.5832 1.3971 1.3748 σ 1.8606 1.9886 1.6783 1.3687 1.2155 1.3495 1.0290 Mean +3 σ 8.1059 8.1712 6.6099 5.6143 5.2295 5.4456 4.4617 Mean-3σ -3.0580-3.7604-3.4602-2.5980-2.0632-2.6515-1.7121 *: 43
Improving the Failure Rate of Holding Pressure of Gas Pipeline by 6σ Method A: Before improvement. B: After improvement of the 3 rd item in SOP. C: After improvements of the 3 rd and the 4 th items in SOP. D: After improvements of the 3 rd ~ the 5 th items in SOP. E: After improvements of the 3 rd ~ the 6 th items in SOP. F: After improvements of the 3 rd ~ the 7 th items in SOP. G: After improvements of the 3 rd ~ the 8 th items in SOP. Table 4-2 Relationship between failure rate (%) and the number of improvements number of improvements failure rate in 36 times of test (%) Estimated value (%) Residuals (%) 0 1 2 3 4 5 6 30.55 25 19.44 16.66 13.88 13.88 8.33 28.369 24.9954 21.6218 18.2482 14.8746 11.501 8.1274 2.181 0.0046-2.1818-1.5882-0.9946 2.379 0.2026 SOP process improvement dwell gas pipeline failure rate regression analysis chart Experiment 36 times the failure rate(%)y 35 30 25 20 15 10 5 0 0 1 2 3 y = -3.3736x + 28.369 4 5 0 1 2 3 4 5 6 7 Process Improvement times (times)x 6 Figure 4-1 Regression line 5. CONCLUSION In this study, experimental processes by SIPOC model are established. And the fish bone diagram and matrix diagram are also used to identify the causes of failure in the holding pressure of pipeline. DMAIC is applied to improve failure rate in SOP, and also Six-Sigma analysis and regression analysis are used to estimate the relationship between the improvements in SOP and failure rate. Because the gas pressure will be affected by 44
Journal of China University of Science and Technology Vol.57-2013.10 temperature, Gay-Lussac's law is used to calculate the correct pressure value and finally use defective rate to confirm the achievements in SOP improvements. In the past, we get lessons from the failure of SOP and discuss the contents about how to improve it. Special gas, such as, SiH 4, is a kind of high degree of dangerous material, as long as there is an explosion or leakage, it will cause very significant casualties and damages to the environment. Now, the stringent SOP processes are built in order to avoid the accidents. From the experiences of operations, the SOP processes are learned and modified gradually. So we could achieve the ultimate goal of no gas leaks. Although the SOP process has been set, it does not represent one hundred percent of safe procedures. The operators still need to continue and seek improvements to reach zero percent of leak rate. The failure rate was significantly reduced from 30.5% and 8.3% when comparing it with the procedures before and after improvements, As long as the operator obeys the SOP, the risk can be reduced, the failure rate is also decreased and the cylinder on-line time can be shortened. In order to achieve the goal of simplicity and security in the same time, the current SOP procedures still have room to be improved. 6. REFERENCES [1] Boppana V. Ramabrahmam and G. Swaminathan, 2000, Disaster management plan for chemical process industries. Case study: investigation of release of chlorine to atmosphere, Journal of Loss Prevention in the process industries, vol. 13, pp.57-6. [2] Ko Chun Mou, and Guang Hann Chen, Risk management in semiconductor industry, Semiconductor Manufacturing, Technology Workshop Proceedings, 2004. [3] SIPOC (Suppliers, Inputs, Process, Outputs, Customers) Diagram". Milwaukee, Wisconsin: American Society for Quality, 2012-07-03. [4] Randy Rothwell. successful industrial innovation: Critical Factors for the 1990s. R&D management, 1992. [5] Eckes, G.., Making Six Sigma last: managing the balance between cultural and technical change, New York: John Wiley, 2002. [6] Mast, Jeroen De., Quality Improvement from the Viewpoint of Statistical Method, Quality and Reliability Engineering International, Vol. 16, 2003, pp.301-311. [7] Harold, Dave, Jan.," Design for Six Sigma Capability "Control Engineering", 1999, pp.62~70. [8] Andrew, T. &; Barton, R., &; Okafor, C. C., Applying lean six sigma in a small 45
Improving the Failure Rate of Holding Pressure of Gas Pipeline by 6σ Method engineering company - a model for change, Manufacturing Technology Management, Vol. 20(1), 2009, pp17. [9] Saaty, T. L., The Analytic Hierarchy Process, McGraw-Hill, New York, 1980. [10] Hammer, M. Process management and the future of Six Sigma. MIT Sloan Management Review, Vol. 43(2), 2002, pp26~32. [11] Martin, W. F.(2007). Quality models: Selecting the best model to deliver results. Physician Executive, Vol. 33(3), 2007, pp24~31. [12] Saaty, T. L., The Analytic Hierarchy Process, McGraw-Hill, New York, 1980. 46