every process performed the same way every time. While the chemistry output of a week’s work of castings might be within the specifi cations, having one lot’s chemistry run on the high end and the next run on the low end indicates a poorly controlled process that could be- come out of specifi cation soon. Further, Schorn said the most economical way to run is to produce the same process results repeatedly. Variation results in added cost.
Interpreting the Data How you collect and read your data
can range from the simple to the com- plex. In any case, data from a single
day or hour has to be accumulated in a meaningful way to be compared to prior data. On the simple end, a metalcaster can make a run chart by plotting the data on a bar or line graph to see where movement in process parameters occurs. “A run chart is the most elemental
and initial evaluation,” Schorn said. “It’s used so you can see whether things are stable or not.” To make the information more power-
ful, Schorn said statistical process control (SPC) enables a person to place control limits on the run chart in order to de- termine what is statistically signifi cantly
Is Your Process Under Control? S
tatistical process control (SPC) is a quality control application focused on helping an organization monitor process behavior using statistical methods, according to the American Society for Quality (ASQ). One of
the most-used SPC tools is the control chart, which allows an operation to distinguish between typical variation that will occur in a process no matter what and a variation that is caused by external circumstances and is statisti- cally out of control. Control charts come in many variations but generally feature three horizontal
lines. The line in the center represent the average of the process, the lines above and below that represents the limits within which a process would be considered under statistic control. When data falls outside those limits, there is a strong chance it is due to special causes that may require correction. SPC can be used on various metalcasting processes, as well as casting dimension checks. On its website,
www.asq.org, ASQ has a generic control chart template to
calculate upper and lower limits of a testing point for a graphic representation of a process’ variation, including when the process edges out of control.MC
different than the average of the data. For instance, Schorn said reject
rates may be 2.1% one day, 2.7% the next and 1.9% the next. The tempta- tion is to react to each day’s scrap rate individually. But after looking at reject rates over a longer period of time, the typical variation could turn out to be between 1.5-3%. So, a day with a 7% or 0.5% reject rate would be radically different from the norm and indicate that something signifi - cant happened in the operations to achieve that result. “The power of SPC is it allows you
to know when to chase the move- ment of that indicator,” Schorn said. “You can tell when it’s a true signal of something important.” For measurements that involve a
hard number, such as those performed with the use of a gauge or thermometer, Schorn said meaningful SPC charting requires 20-30 data collection points, which would be taken over several hours or days, depending on the casting volume of the facility. For measure- ments that involve a count, such as scrap rate, SPC charting requires 30 to 40 data collection points to be able to discriminate between typical variation and signifi cant variation. “It is recognized good management technique to get a historical look at what is happening,” Schorn said. “How do we put this data into context?” According to Schorn, a point out-
side the control limit has less than a 1% chance of being in that area without something special going on in the operations, whether it be a change in equipment, workforce, material or technique. The formulas used to calculate
SPC limits can be performed manu- ally or with automatic measurement and calculation tools and software. “If you are going to actually con-
trol the process and just want to get started with SPC, it’s okay to do it by hand on paper,” Schorn said. “There’s a great value in helping people who are actually making the measurements to understand what is being calculated to fi nd those limits.”
MC For More Information
Shown is an example of a control chart used in statistical process control. Line P is the average of the data, LCL is the lower control limit and UCL is the upper control limit. Points beyond the UCL and LCL would be considered out of control.
MODERN CASTING / July 2010
“Determining Preventative Costs in the Foundry Then Reacting,” N. Fox, AFS Transactions 2003 (03-123).
Quality Control Systems in the Foundry, R. Struk, American Foundry Society, 2001.
25
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56 |
Page 57 |
Page 58 |
Page 59 |
Page 60