Common sense in sand sampling and reduction.
Throughout history, Americans--particularly American businessmen--have identified and tackled basic problems with huge arsenals of technology, productive hours and money. In other words, overkill.
As radical as a minimalist solution to an engineering or manufacturing problem in the foundry may sound, the tired adage of "getting back to the basics" can and does become workable if properly applied.
AFS Molding and Core Aggregates Committee (4-J) is responsible for the maintenance of 23 separate test procedures in the Mold and Core Handbook. After completing the last revision, the committee became racked with questions about those test procedures. Were the results responsive to the industry in which they will be used? The application of new statistical techniques and gage studies on the procedures and devices specified might expose shortcomings overlooked in the past.
When these questions were aired, they were suddenly followed by a multitude of concerns not previously raised.
The problem became where to begin in this plethora of procedures--all critical to some facet of the foundry industry. Would it be particle size? Acid Demand Value (ADV) determination? Bulk density? Temperature? How about clay determination and by which method?
As the ever-growing wish list of new work projects was compiled, a basic question kept surfacing. What do all of these procedures have in common? A simple, basic answer--all procedures were sensitive to sample variation. As Brian Kaye of Laurentian University said, "All too often, excessive effort goes into attempts to improve experimental precision in the characterization procedure when data variability is generated by unsatisfactory sampling techniques."
The questions regarding sampling and sample reduction grew steadily:
* What if the written procedures themselves did not produce the statistically representative samples they were intended to produce?
* What if the sampling devices and procedures prescribed actually exacerbated sampling variation?
* Would a new technician picking up the handbook for the first time be able to select the proper sampling and sample reduction techniques from the titles listed? Would a seasoned technician fare any better?
The Process Evolves
Applying some of the tenets of basic quality system management, the committee approached the sampling and sample reduction dilemma head-on, but viewed it as a "process."
Author Harry I. Forsha defines the key elements of a process as:
* Inputs: things provided at the beginning of a process, to which the process is applied.
* Actions: the things that are done to the inputs to change them.
* Outputs: the results of the actions, which then usually feed some other process.
These elements clearly exist in a sampling and sample reduction process. The best way to specify and illustrate the inputs, actions and outputs of the process is through the development of a basic process flowchart.
Getting the Picture
The symbology of the process flowchart is quite simple. Square boxes represent actions or process steps, diamonds are decision points in the process, and ovals or circles denote beginning or end points of a process or process branch. The symbols are interconnected with lines to show the direction of the process flow.
Other processes or subprocesses can feed into or out of the basic flowchart at any point. While software is available that allows linking to spreadsheet, word processing or project management software packages, inexpensive templates or even hand-drawn charts are just as effective.
Making It Work
How then is the flowchart integrated into the development and clarification of AFS sampling and sample reduction procedures? There is a "wet" side and a "dry" side to the flowchart to take into consideration the basic forms sand can take in the field.
The numbers shown in parentheses on the flowchart refer to the existing AFS procedure number. If more than one procedure is shown in a process step, it indicates that several documents may apply, but that they are shown in order of preference. In procedures where open parentheses are shown, a new procedure is being undertaken and will be plugged in when completed.
Step 1 in the flowchart departs from existing logic in material sampling where the sampling itself is the goal. The approach is to develop a test plan prior to sampling and sample reduction.
The test plan is less complex than it sounds and is a classic example of "working smart or working hard." Before a manager sends a technician out to the railcar or pile with the official sampling coffee can, the process of testing must be well understood by all.
Developing the Test Plan
The basic premise is that the technician involved with the sampling and reduction must know how much sample he/she will require downstream for subsequent testing. Which samples can be reused for other tests? Which tests are destructive and will render the sample worthless for other procedures?
If you wish to determine moisture content, loss on ignition, particle size and 25 micrometers of clay on the lot of material, can you really perform them on the same sample? If not, how many samples will you need and of what approximate sample mass?
Besides the actual mass of sand required for testing, the order in which the testing is performed must be choreographed. If, for example, a moisture test is critical to your operation and the technician doesn't exercise proper care in obtaining or storing the sample before testing, the resulting erroneous test data may have a negative impact on your sand system downstream.
Another departure from previous logic and practice is a move back to basic engineering in obtaining the actual samples from the lot. Steps 2 and 3 are decision diamonds that ask you to determine if the sand can be sampled while in free fall; or if the sand is bagged. If neither condition can be met, the flowchart leads to a "stop" condition in step 4.
The presence of segregation of sand in a static pile precludes the ability to obtain representative samples from the pile. While the segregation conditions are far worse in dry sand, even damp sand can exhibit sufficient segregation during stacking to introduce sampling error into your process. Segregation is thought to be a problem of particle size measurement only.
Unfortunately, it also compounds problems in other basic foundry tests--whether you are performing bulk density, clay determination or wettability for example.
The particle dynamics in material systems increase the concentration of fine particles in the center of the pile, while larger particles tend to migrate toward the surface.
"Fine particles in a heap standing on a vibrating surface are subject to two segregation mechanisms--percolation and walking," Kaye said.
"Percolation is the movement of fines down through the structure of the heap. Walking occurs when a relatively large fine particle, originally in a horizontal position, tilts under vibration so that smaller fine particles near its shoulder pour underneath it and consolidate the new position."
"A later random elevation on the other end of the larger fine particle results in the elevation of the fine particle relative to its original position. Over a period of time the net effect of many vibrations and powder within the heap is to walk the larger fine particles to surface of the heap."
To illustrate, think of the distribution of salt in your popcorn by the time you get from the kitchen to the TV. Would a sample taken from the top of the bowl represent the popcorn to salt ratio of the whole?
Terry Allen, in his book Particle Size Measurement, makes a nontechnical but valid statement on sampling from a static pile: "There is only one sound piece of advice to give regarding sampling from a heap--Don't! Never!" He also refers to the "Golden Rules of Sampling." They are:
* A powder should be sampled when in motion.
* The whole of a stream of powder should be taken for many short increments of time in preference to part of the stream being taken for the whole of the time.
The committee took these points to heart in dealing with the dilemma experienced by foundries with sand systems that use static sand piles. Again, after attempting to come up with a high-tech solution chock full of complex statistics, a low-tech, basic concept dawned on the committee (with the help of some outside consultations with process and mining engineers).
How did the static pile get there? It had to be conveyed, stacked, blown, shoveled or otherwise moved from supplier to user. While in motion, was there some point of free fall at which samples could be taken? If so, your decision diamond path in the flowchart would take the affirmative path and the process could continue.
You say you didn't think of that before you stacked it and now you still have an unsampled pile? Will you ever use all of that sand again or will only a portion of it have to be moved from the pile to the sand system? Sound familiar? Again, your decision diamond number 2 in the flowchart would take the affirmative path and the process could continue.
Our Process Continues
The committee has a full plate of work to do on the procedures and equipment just within this flowchart, not to mention the balance of the procedures left to evaluate. Within the flowchart, the process steps showing multiple options of applicable documents are all ripe for study for precision and bias, as well as reproducibility and repeatability studies.
The use of the process flowchart helped us reduce the clutter or noise in the measurement system and get down to basic roots of the process. We were able to work through the process logically and expose deficiencies in the system, such as the need for drying procedure, separate procedures for gate vs. nongate splitters, and for evaluating the relative precision of the different pieces of equipment.
Will this commonsense approach be used in the review of other AFS procedures? No question. The group development of the flowchart is an outstanding consensus-building device and will prove useful as we tackle other committee concerns.
|Printer friendly Cite/link Email Feedback|
|Date:||Mar 1, 1994|
|Previous Article:||Benchmarking the nobake binder systems.|
|Next Article:||Using SPC in the coreroom.|