In recent years, with the rapid development of global high technology and my country's economy, the automotive industry has also experienced transformative innovation and development. New technologies and new business formats such as big data, artificial intelligence, the Internet of Things, and cloud computing are surging. Automakers equipped with various advanced technologies have undoubtedly become the biggest winners when the public buys cars. Even in this good situation, we must be clearly aware that as people's car purchasing standards gradually improve, the domestic automobile industry will face huge challenges. One of the challenges is: how assembly plants and parts suppliers in the automotive supply chain can increase the quality control capabilities of the entire automotive supply chain in response to increasingly fierce competition and integrate with international standards.
As the leading assembly plant in the automotive supply chain, it is faced with how to guide and promote the development of the quality control system throughout the supply chain to jointly achieve the goal of "preventing defects, continuously improving, reducing variation and waste throughout the supply chain", and ultimately reaching consumers. satisfy. For the majority of parts suppliers in the automotive supply chain, one of the many problems they face is how to deal with the different quality control system requirements of different automotive assembly plants.
Judging from the successful experiences of leading companies such as General Motors, Delphi, Toyota, and Honda, data-driven continuous improvement activities are one of the magic weapons for them to maintain their long-term competitiveness. The automotive industry has always attached great importance to data analysis. For example, among the five tool manuals of the early industry standard TS16949 quality system, there are two categories ("Measurement System Analysis" and "Statistical Process Control") that specifically introduce quality from the level of statistical analysis. The role and methods of improvement. The quality control and quality improvement capabilities of JMP statistical analysis software, which has been deeply involved in the automotive industry in recent years, also reflect this feature.
Compared with American and Japanese companies, the gap between my country's automobile industry in this regard is relatively obvious. Fortunately, the rapid development of information technology provides us with opportunities to narrow the gap and improve quality. The increasing popularity of high-end desktop statistical software represented by JMP in domestic automobile companies such as Chery has greatly reduced the data calculation burden of the majority of non-statistical engineering and technical personnel. More importantly, by applying the graphical analysis technology, advanced experimental design and other tools provided by JMP software, companies can use scientific and objective analysis ideas to solve many practical quality problems that have been troubled for a long time. The cost of inferior quality has been doubled, and production efficiency has been significantly improved. .
The following will be combined with a successful case to help everyone understand how to apply statistical techniques in the automotive industry.
Background introduction: The cylinder of a certain automobile engine is assembled by the cylinder block, cylinder head, piston, piston bearing, connecting rod, rod bearing and crankshaft (as shown in Figure 1). The current headache for the cylinder assembly shop is that although the dimensions of each component meet the original specifications, a considerable proportion of the finished products will always be too loose or too tight after assembly, and each batch will cost a lot of extra money. rework costs. Although the company has realized that the size of the parts should be adjusted, it cannot determine: which part or parts will be most effective? To what extent should the tolerance range of this component be adjusted most appropriately? ?
Figure 1 Structural diagram of engine cylinder
First of all, based on the principle of quantitative management, we can define this product quality characteristic as assembly clearance (= cylinder block + cylinder head – piston – piston bearing – connecting rod – rod bearing – crankshaft). The final assembly result cannot be too loose or too tight, which means that the tolerance specification of the assembly gap must be defined within a certain range, such as [0.005, 0.015].
Secondly, let’s look at the situation where each component meets the current specification requirements, that is, the cylinder block ~ 8.50121 ± 0.00468, the cylinder head ~ 0.70921 ± 0.00219, the piston ~ 2.00156 ± 0.0011, the piston bearing ~ 0.25035 ± 0.00114, and the connecting rod ~ 5.00017 What is the quality level of the assembly clearance when the rod bearing is ±0.00327, the rod bearing is ~0.20008±0.00099, and the crankshaft is ~1.74825±0.00249. Through the analogy and graphical functions of JMP's powerful "Predictive Profiler", it is not difficult to find that the defective rate at this time reached 4.96%, and the PPM reached 49,600 (as shown in Figure 2).
Figure 2: Current quality level of assembly clearance
Obviously, the assembly defect rate of nearly 5% cannot meet the requirements of mass production. Where should we make breakthroughs and seek improvements?
In order to solve this problem, we first understand the concept of overall defect rate. Taking the component "piston" as an example, if the length parameter of the piston is inappropriate, it will produce two types of defects: one is that the assembly gap of the finished product is too small due to the parameters being too large, that is, the assembly is too tight; the other is that the parameters are too small The resulting assembly gap of the finished product is too large, that is, it is installed too loosely. These two types of defects cannot be minimized at the same time, but can only reach an overall minimum at a certain location. This position is the lowest point of the red overall defect rate curve in Figure 3. Other parts have similar overall defect rate curves, but the lowest point is different and the steepness of the curve is different.
Figure 3 Schematic diagram of overall defect rate formation
Similarly, JMP software can be used to quickly and easily produce the defect rate map of all parts as shown in Figure 4. Obviously, the average value of all parts has reached the lowest point of its overall defect rate curve, but the steepness of the cylinder is the most obvious, indicating that it is the most sensitive to the defect rate of the assembly gap, and adjusting its specification limit can most significantly improve the finished product. quality level.
Figure 4: Current defect rate chart of each component
Therefore, we decided to reduce the specification limit of the cylinder block from the original 8.50121±0.00468 to 8.50121±0.00234 (that is, the standard deviation is reduced by 1/2). The specification limits of other parts remain unchanged. The improved results are shown in Figure 5. . It is noted that after using the JMP software analogy at this time, the assembly defect rate was reduced to 1.76%, and the PPM was reduced to 17600. Compared with the quality level before improvement, the effect is very significant. Of course, this result is not the most ideal, but we can also continue the analysis just now and find improved component objects step by step to improve and verify until a satisfactory assembly defect rate is obtained.
Figure 5: Improved quality level of assembly clearance
The above case analysis reflects the statistical analysis concept of tolerance design, but it is very simple to implement specific operations using JMP software. It does not require special statistical knowledge to understand, and compared with traditional experimental design, it saves a lot of experiments. The cost of data collection is low, so it is very popular among domestic and foreign automobile companies who are engaged in continuous improvement.
The application of statistical analysis technology in the automotive industry goes far beyond this. The improvement of the overall competitiveness of the automotive industry chain cannot rely solely on statistical technology. But there is no doubt that statistical technology provides a new idea and new method for those companies and employees who lack quantitative thinking to solve practical problems. Moreover, with the help of easy-to-use interactive statistical analysis software like JMP, practitioners no longer need to memorize complicated mathematical formulas, and no longer need to waste precious time on derivation calculations. They only need to focus on the things being analyzed themselves. Just apply exploratory discoveries into practice, which will inevitably accelerate breakthroughs and innovations in China's automobile quality.
