SPC for small to medium foundry's sand system: part II.

SPC for Small to Medium Foundry's Sand System: Part 2

The purpose of statistical process control is not to conduct tests and make charts. It's to gather data and present it in such a way that you will be able to recognize potential problems before they cost you money in poor-quality molds that produce scrap castings.

In Part 1 of this series, we focused on how to start a statistical process control (SPC) program and the benefits that it will bring to your foundry. An effective SPC program requires careful planning, testing and accurate record keeping.

In this section, we'll discuss more about sand testing--what, where, how often and how to test. We'll also look at methods of collecting and storing the data and results of your tests.

What should I measure?

It's necessary to know what variables are really important and control the quality of your sand, and concentrate your testing time. However, there are four primary tests that are most important because they tell you how well your sand is bonded. They are:

Compactibility--Compactibility is considered the most important test. It is a measure of the bond (clay) to water ratio, and in combination with a moisture test, indicates the bond level in your sand.

Moisture--The moisture test measures the demand of water-absorbing materials in your sand. If the moisture content is too low, the action of the clay in bonding your sand will be reduced, as clay needs water to form the clay bond. If there is too much moisture, the excess water will act as a lubricant, causing the sand particles to slip over each other instead of being bound together.

Methylene Blue--This test measures the active clay in the sand, that is, the amount of clay capable of forming a clay-water bond (clay that has been heated to too high a temperature loses its ability to use water in forming a bond and becomes inert dust that should be discarded as fines). Methylene blue is a tricky test to perform correctly. Care must be exercised to make sure it is done the same way each time.

Loss on Ignition (LOI)--The carbonaceous content of the sand is measured by this test. Too much carbonaceous material will interfere with the bond and too little will cause the surface finish of your castings to deteriorate.

If you watch the results that these tests give you, you will be able to see what changes are occurring in your system.

What other tests should I make?

A series of other tests will yield additional information on your system. Depending on the type of castings you pour, the sand mix and system in use, you may find that one or two should be considered primary tests and performed as frequently as the four just mentioned. But you will want to make the test(s) periodically to improve control over your system. These tests are:

Green Compression Strength--This test gives an indication of the control of your sand-making operation. Control is influenced by the amount of fines, muller efficiency, clay and moisture content.

Permeability--Permeability measures the flow of gases through the sand. It's a measure of the fines content, screen distribution and effect of recycled core sands in your system.

Density--Density indicates buildup of fines and dead clay, over-and underbonding and dilution by core sands.

Temperature--This test is much more important than most foundries realize. Sand that is too hot won't bond efficiently. Therefore, you should routinely monitor your sand temperature. Think of this as being nearly as important as the four primary tests discussed earlier.

Muller Efficiency--This test tells you whether your muller is working right. It also will tell you when the bond level is out of control or when the batch size is incorrect, and when the sand has become contaminated.

AFS Clay--This is not the same as the methylene blue test. It will give you an idea of fines and dead clay buildup in your system.

AFS Grain Fineness--This test should be performed on every batch of incoming system sand to assure compliance with your specifications and to avoid problems with off-analysis sand.

Other Tests--From time to time you may find that other tests are necessary to identify special and assignable causes to sand problems. A wide variety of tests can be used, but be sure that you've learned all you can from the standard tests that have been described.

Where in my system should I take samples for these tests?

That depends on your system. You should make a flow chart of your system and use it to decide where to make the tests. There are some guidelines, however.

Your muller operator is responsible for producing good sand. He is the one who should be making the primary tests of moisture and compactibility and he should make them at the muller. LOI and methylene blue tests should be made under controlled conditions in your laboratory.

In addition, it's a good idea to test the sand at the molding machines periodically because that's where the molds are made. Samples for secondary variables may be taken there or at the muller.

The importance of sand temperature cannot be overemphasized. You should know the temperature of the sand as it enters your muller. Hot sand can tell you when your sand system has problems, such as funneling or blockage in the new sand addition hopper. It also can warn you that standard additions of water and bond will not give adequate sand strength.

How often should I makes these tests?

That's going to depend on your system and on the volume of sand that you use. It is possible to overtest. If you are using a batch muller, each muller load should be tested for the primary tests. In a high-production shop with a continuous muller system, you should test at least once per system turnover and at least once an hour. As you set up your SPC system, you will want to test more frequently while you determine your sand system process capability.

The secondary tests can be run less frequently (after you've determined your process capability). For instance, green compression strength, permeability and density might be run four to six times daily, and the muller efficiency measure once or twice a day. AFS grain fineness should be run on each new batch of incoming sand and the AFS clay test perhaps only once or twice a week.

Once you've determined that your process capability is good, you can decrease the frequency of the tests. For instance, if your LOI doesn't vary greatly, cut its testing down to once or twice a shift. But if you get into trouble, you should resume frequent testing until you're sure that the problem is solved.

How should I make these tests?

The AFS Mold & Core Test Handbook gives directions for these tests. In addition, the suppliers of test equipment provide directions.

Read the directions. Follow the directions. Every time. No shortcuts. Ever.

It's absolutely essential that the personnel who make these tests be well trained in the procedures. You will find it well worth the money to send them to a Cast Metals Institute course for training.

Now for the two most important points.

Do the tests the same way each time. The results will be different if different procedures are used. This means that each person who does the test must know the correct way to do them. Periodically you should give them the same sample and see if they get the same results. If they don't, you may have a problem.

Keep your test equipment calibrated. If the equipment gets out of calibration, the data collected is worthless. Follow the manufacturer's instructions for keeping your test equipment well maintained and in correct calibration.

You will be making important decisions based on the results of these tests. If the test results don't tell you accurately what's going on in your sand system, you won't have a chance to make good decisions.

Data Collection & Storage

It sounds simple--all we have to do is follow the directions.

Right. Follow the directions on how to do the tests. With all of the variables that affect sand and casting quality, it's easy to get confused. The reason for an SPC program is to help clear up that confusion. So your tests have to be consistent--you don't want them to become another variable. Use the same equipment, the same (exactly the same) chemical reagents and the same procedures each time. If you don't, you'll get results that can't be compared to previous ones. And that's no help at all.

As for calibration of your test equipment, do it at least once a year. If the equipment is out of calibration, fix it! And if you can't get it fixed, buy a new piece of equipment.

You must have confidence in your test results. Any test that gives a suspicious result (or one that's very different from the previous results) should be repeated, paying particular attention to following directions. Balance pans sometimes stick, dirt falls into a beaker when you're not looking or part of the sample is spilled. These little events cause wrong answers. If your results don't seem right, repeat the test. Chances are that the problem is in the test, not the sand.

As we said, you should make the four major tests, plus any others that you find necessary to control your system, at least once during each cycle of sand through your system. You should test again when you make a change in operating your system (like changing the product mix to one with a very different sand to metal ratio or core sand to green sand ratio).

It's also a good idea to repeat the first series of tests at the beginning of a shift to be sure that your test equipment is working.

Your sand system operator is responsible for producing and delivering consistent quality sand to the molding lines. That means that he's responsible for controlling the quality of the sand and, therefore, has to know how good a job he's doing. The best way for that to happen is to let him make the measurements so that he sees first-hand how well he's succeeding.

This means that he has to be motivated, committed and trained to operating your sand system to make the best sand possible. You can get training help from vendors, and by enrolling your sand operator in CMI courses on sand and SPC. His supervisor should attend, too, but it's the operator who makes the sand, so he's the one who really needs the training.

Of course, you should check up on the test results. Every now and then you should see if the supervisor can duplicate them, and your lab technician should verify a certain percentage of them. Maintaining high performance standards in your SPC program is absolutely essential.

How many tests should I make to get good results?

Part of making a test is deciding on a sample to test. You want that sample to be representative of the whole lot. If you're testing a muller batch, you want to be sure that test results accurately represent the whole batch. But remember: Your sand varies not only from batch to batch but within each batch. Because you want to decrease this variability, you must measure not only the batch-to-batch variations but also the within-batch variation.

To do this, you will need to take more than one sample and average the results. For batch mullers, for instance, you might want to take one sample near the beginning of the discharge, one in the middle and one near the end. For continuous mullers, you may want to pull one sample every minute for three minutes. Each of these samples is then tested and the results are averaged.

After I've made the tests, what do I do with the data?

First, record the data. Write it down on a form especially designed for that test. The form should include the date, time and name of the person who ran the test, as well as the test results. If calculations are part of the test, there should be space for them so that you can check to be sure that they were done correctly.

There should be a place for any comments the operator wants to make (such as, if a new additive has been used, if there has been a power or equipment failure or a line shutdown). Each test should have a separate form.

Every reading should be recorded. If a reading is suspicious and you repeat it, both readings should be written down, with a note as to why the test was repeated.

The object here is to be sure you know what the characteristics of your sand are. You're looking for trends. Even the fact that you have to make repeat readings is telling you that things are getting out of control.

Then find the average value for the the results. To do this, you simply add up all of the data and divide by the number of data points.

You or your operator may be tempted to try and get by on just one sample or data point to save time and calculations. Don't let that happen! The more data points you take, the more you will know about the sand quality and how well you are controlling it.

Statisticians recommend six samples as a minimum if you want to be sure of the accuracy of your measurements. In practice, most foundries find that at least three samples are necessary, although five or six are recommended.

After awhile, you'll find that the process will dictate the number of samples needed. A process that runs smoothly will normally need fewer samples than one that has a tendency to get out of control.

After you have a test value (the average of the values obtained for each of the samples) in which you have confidence, plot it on a control chart (which will be discussed later in this series). Plot the data as soon as you get it so that deviations from your normal values are found quickly.

This isn't a case of running the tests or plotting the data when you get around to it. You want to know as soon as possible how your system is performing so that if it is out of control, you can correct it before it makes scrap castings.

When you start your SPC program, you have to determine your process capability. You're going to be wondering just how many tests to make. Here, it's better to err on the side of overtesting. Since you don't really know what part of your variability in test results is due to common variables or special variables, it's also important to know just what conditions are when the tests are made.

After you have determined your process capability, you can back off on the number of tests (but not the number of samples in each test) and the detail required on the data sheets. But don't cut back too far. The example data sheet contains the minimum information needed.

There also is another set of data you should keep. It is a record of your casting production--what castings were made on which mold lines or machines, when they were made and what the reject and rework rate was. Only with this data can you get an idea of how your sand conditions affect casting scrap and rework. The shrewd foundryman will use the sand control and casting reject rate to tailor his sand system so that it is not a cause of casting scrap.

In starting your SPC system, this data on the quality of the castings produced is very important. If your normal sand conditions are producing poor castings, then your process capability is poor and it must be corrected before you can continue with the SPC program.

Do I need a computer for an SPC program?

No. Although a computer may make some of the job easier for you after your system is established, it's not necessary at the beginning (and, in fact, you may find that it's not necessary at all).

Get your SPC system up and running first. Be sure that it's giving you the information you need. Then, get a computer if you believe that it would benefit you by doing some of the calculations more quickly and accurately. There are many statistical programs for PCs that will help.

But remember that the purpose of the SPC program is to help your operator make better sand. The computer terminal and all of its information should be at his station, as well as in the sand lab.

You will have to save the data. This means saving the original data sheets. You will need these to show your customers that your process is being controlled, and to record for your own information when you had a problem and what you did to fix it. Of course, at some time you may decide to save this data on computer disks but that won't happen until well after you have made SPC work in your plant.

PHOTO : Green compression strength is being tested on a 405 strength machine. This test determines

PHOTO : the control of the sand-making operation: fines, muller efficiency, clay and moisture

PHOTO : content.

PHOTO : Compactibility is considered the most important sand test. Here, a standard threeram

PHOTO : compactibility test is being conducted.