Before using the mean range control chart, some appropriate preparations must be made:
The first item of preparation is to determine the characteristics of the drawing.
To determine the characteristics of the mean range control chart (Xbar-R), we should first consider the customer's needs, such as the SPC key control items explicitly required by the customer in the drawings, and give priority to the key quality characteristics as the mean range control chart (Xbar-R). We should also consider the current potential problems, such as areas with outstanding quality problems, and give priority to them. In addition, when there is obvious correlation between characteristics, we can consider controlling one or several of these related characteristics, which can improve efficiency. For example, in the stamping process, there is a certain relationship between many dimensions, because the die determines the proportion between these dimensions, so we don't need to control all the dimensions, just select some features for analysis.
The second preparation is to determine the measurement system.
It is necessary to analyze the measurement system before drawing the mean range control chart (Xbar-R). Because the variation of the measurement system is too great, the difference between the measured value and the real value of the product will be too great, which will affect our decision-making on the production process. Therefore, first determine whether the variation of the measuring system is acceptable, then determine the corresponding measuring personnel, instruments and equipment, and use standardized measuring procedures to measure the products, so as to ensure that the variation of the measuring system is minimized.
Next, the drawing steps of mean range control chart (Xbar-R) are described in detail.
A.
gather data
A. 1
Select subgroup size, frequency and data.
A. Subgroup size
In the initial research of the process, the sub-group is generally composed of 4 to 5 parts produced continuously, which only represents the parts produced by a single tool, punch and cavity (that is, a single process flow).
The first key step of measurement control chart is to determine "reasonable grouping", which will determine the effect and efficiency of control chart. You can learn more details through SPC control chart training class _ lesson 7 _ control chart sampling number (sample).
B. Subgroup frequency
Subgroup frequency refers to the sampling interval of every two groups of samples, and its purpose is to check the changes of the process after a period of time. The potential causes of these changes may be shift change, operator change, temperature rise trend, material batch and other reasons. The sampling frequency varies greatly with the processing technology of products. For example, many hardware processing industries usually sample a group of samples around 1 hour, and some electronic enterprises such as PCB enterprises sample a group of samples every 20 minutes.
C. the size of the number of subgroups
The number of subgroups should meet two principles. From the process, collecting more subgroups can ensure that the main cause of deterioration has a chance. Generally speaking, 25 or more subgroups containing 100 or more single-value readings can be used to test stability well. Most enterprises adopt the model of 25 groups of samples.
A.2
Establish control chart and record original data.
The mean range control chart (Xbar-R) usually draws Xbar at the top of the R chart, followed by data columns. The numerical values of Xbar graph and R graph are ordinate, and the subgroups in time series are abscissa. Data values, ranges and average points should be vertically aligned. Fill in the corresponding fields with each group's single reading and identification code.
A.3
Calculate the average (x) and range (r) of each group.
The characteristic quantities drawn on the control chart are the sample mean (x) and sample range (r) of each subgroup. They reflect the mean and variance of the whole process respectively.
A.4
Select the scale of the control chart.
For the X graph, the difference between the maximum and minimum values of the scale values on the coordinate should be at least twice the difference between the maximum and minimum values of the subgroup average (x). For the R graph, the difference between the scale values from the lowest value 0 to the maximum value is twice the maximum range (r) encountered in the initial stage.
Answer 5
Draw the average value and range on the control chart.
B.
Calculation control limit
For the calculation related to mean range control chart (Xbar-R), please refer to SPC Control Chart Training Course _ Lesson 65438 +05 _ Calculation related to mean range control chart (Xbar-R).
C. process control interpretation
If the variability between each part of the process and the process mean remains at the existing level (estimated by R and X respectively), the range (R) and the mean (X) of a single subgroup will change randomly independently, but they rarely exceed the control limit. Moreover, there will be no obviously different graphs and trends from random changes in the data. The purpose of analyzing control charts is to identify any evidence that process variability or process mean is not at a constant level.
For the analysis of mean range control chart (Xbar-R), please refer to SPC control chart training class _ Lesson 16 _ How to analyze mean range control chart (Xbar-R).
D.
Process capability interpretation
Before explaining the process capability, it is necessary to determine whether the current process meets the following conditions:
The process is in a statistically stable state;
The process and each measured value obey normal distribution;
Engineering and other specifications accurately represent the needs of customers;
The design target value is located in the center of the specification;
The measurement deviation is relatively small.
If it has been determined that a process is under statistical control, there is still the question of whether the process can meet the needs of customers, which is why we should explain the process capability. Before explaining process capability, we must first calculate it.
D. 1
Standard deviation of calculation process
D.2
Computational processing power
For the calculation of process standard standard deviation and process capability, please refer to SPC Control Chart Training Course _ Lesson 15 _ Calculation of Mean Range Control Chart (Xbar-R).
D.3
Evaluate process capability
Usually, we use CPK process capability index to evaluate process capability. Generally speaking, CPK is above 1.33, which shows that the process capability is still very good. When evaluating process capability, we also need to understand the needs of customers. In the auto parts industry and some electronic industries, many customers require suppliers to provide products with key quality characteristics with CPK above 1.67.
Dingsi
Improve process capability
If the process capability index CPK is not ideal, you can improve the process performance by reducing the changes caused by common causes or adjusting the process average to near the target value. Or, you can change the product tolerance and adjust the product specifications to make them consistent with the process performance. Of course, this premise is that you have the right to modify the design.
Generally speaking, in order to improve the process capability, reducing the variation caused by common causes is the most powerful countermeasure. Attention must be directly focused on the system, that is, the fundamental factors that cause process variability, such as machine performance, consistency of input materials, basic methods of process operation, training methods or working environment. Generally speaking, these improved methods will exceed the capabilities of operators or their field managers. On the contrary, these appeals need management intervention to be effective.
D.5
Draw the control chart and analyze the revised process.
After the process change, when all the unstable factors in the change period are solved, the new process capacity should be evaluated as the basis for future operation control limits. Usually, the changed data of 25 subgroups are enough to establish new control limits.