Latest Community Postings
Recent Blog Entries
Ankur's Tech Blog
Community Admin Blog
Energy Efficient Hot and Cold Water
Electrical Process Tomography
Biodiesel: The Road Ahead
Methanol Plant Capacity Enhancement
Plate and Frame Heat Exchangers: Preliminary Design
Compressor Surging Under Control
Plant and Equipment Wellness, Part 1: Observing Variability
Share this topic:
To understand variability and why it is a problem there is a simple tabletop game to play that is a great introduction to the variability within processes.
In Figure 1.1 two lines are drawn crossing at 90o with a 2mm circle drawn around their intersection. The game is to sit at a table and drop a pen into the two millimetre diameter circle from a height of around 300 mm (one foot). Getting a hit within the circle is the outcome required from this process. Repeat the targeting and drop process at least thirty times. After each drop measure the position of the new mark to an accuracy of half a millimetre. Record the horizontal distance from the vertical line (the x distance) and the vertical distance from the horizontal line (the y distance) in a table like that of Table 1.1.
Observe the average and spread, of the X and Y results. In Table 1.1 no hits are within the two millimetre circle; some are on or near the edge while most are well away. Even though great effort was made to control the process, the results were across a wide band of outcomes. This same problem occurs in all business and operations processes. The outcomes of a process are spread across a range of results. That is variability. Variability becomes a problem for a business when the results from a process are not consistently within their required boundaries.
If the aim of the game is to have every pen-drop fall inside the 2mm circle, then we have a very poor process for achieving that outcome. To get better results requires changing the process. The game can be repeated using a different process. The results in Table 1.2 were from a process where the pen was dropped after aiming it at the circle from above, much like dropping a bomb from an aeroplane using targeting sights.
The results of the second attempt to play the cross-hair game using a modified process are better; the Y values are virtually the same as the X. The averages of the modified process indicate that the hits were closer to the intersection than those of the first process. There was less spread. But the second process is still not suitable for meeting the requirements of consistently getting the pen within the circle from a height of 300mm. It is very unlikely that any process using human hands to drop a pen within a two mm circle can be controlled sufficiently accurately. If the requirement is to be met it will not be done by dropping the pen with human hands.
To get the pen consistently within the circle requires the creation of a process that removes the variability caused by the human hand. A number of devices have been proposed. These include a long, tapered funnel to guide the pen onto the target, a vee-shaped slide to direct the pen into the circle and a robot with a steady manipulator to drop the pen.
One answer that was jokingly suggested was to open the circle up to 50mm diameter and then everything will be on target. The suggestion totally defeats the purpose of having a process that delivers accurate results. But unfortunately it is the solution that many businesses select. They chose to widen the target and accept any result, good, mediocre or disastrous, rather than improve their processes. A business that does not purse excellence in their activities will not last.
Examples of processes with inherent high variability are those that at some point:
· require decisions
· require choices
· are done without exacting training
· have no standards
· have inadequate procedures
· lack correct information
· are based on opinion
· involve emotion
· have multiple ways it can be done
· are not measured
· have high rates of equipment failure
· involve interpretation of data
When such situations arise in a process the chance of variability rises because the process contains varying degrees of randomness and uncertainty. Ill-defined, inexact processes and those with poor monitoring and control are at risk of being impacted badly by any change in performance of their critical success factors. This is particularly the case in sales and marketing, finance, human resources, administration, engineering, design, customer service, production, manufacturing, dispatch, after-sales service and maintenance.
Variability can be identified in processes by charting or graphing process parameters and process outputs over a period of time. Such charts are called run charts and are used to locate the times that the process did not produce the required result.
If you want immediate control over a process then track the process variables, those factors that influence the result, in real-time so they are observed as they change. If the change is bad you have time to react and correct it before too much damage is done. If you want pre-emptive control of a process then trend the variables of the process inputs before they enter the process. By being sure that the inputs into a process are correct and right you can be more certain that the process they feed into will be better behaved. If you only want to know how well a process performed then monitor its final output, the product from the process. Unfortunately monitoring the final output puts you into the position of asking what happened when something goes wrong. Just like the company in Example 1.1 who had no idea what had changed to cause a spate of raw material stock-outs. By tracing the replenishment process on a run chart it was possible to highlight process fluctuations and then identify the underlying causes of their problem.
By: Mike Sondalini, Enterprise Asset Management Columnist for Cheresources.com
Send this Page to a Friend
|Advertise with Us|
|The Chemical Engineers' Resource Page, 1442 Goswick Ridge
Road, Midlothian, VA 23114, Email
All material is Copyright © 2008, www.cheresources.com