Getting Started with PREDICT

Getting Started with PREDICT

Determine if PREDICT will be effective

PREDICT is based on the assumption that the chemical composition controls the tensile strength of the iron. In most cases, this is a valid assumption; however, there are exceptions. Erratic machining of tensile bars, variations in inoculation/nucleation, and significant temperature differences are just a few of the operational variables that can mask the effect of the chemistry.

In order to make sure the chemistry is playing the significant role in controlling the tensile strength, a multiple regression analysis of the chemistry and tensile strength is performed. Ideally, we like to have at least 50 sets of data with chemistry and resulting tensile properties; however, it can be accomplished with less than 50. (The degree of confidence is reduced the further below 50.)

The multiple regression analysis shows what portion of the variations in strength being experienced can be attributed to the chemistry. If the analysis reveals that the chemistry is not the controlling factor, we will notify the foundry of the lack of correlation and provide suggestions of what variables should be examined in order to gain better control.

Fill Out Operational Questionnaire for PREDICT

Assuming the multiple regression analysis has indicated chemistry is the controlling factor, predict is modified for the specific operation in question. In order to make the program as easy to use as possible it=s necessary to learn about the operation. In order to gain that knowledge a questionnaire has been developed. The questionnaire consists of eight questions that follow.

Grade

Basically, all we need in response to this question is how the alloy in question is designated in your operation. This becomes more important when PREDICT is used to provide predictions for more than one grade of metal.

Specifications & targets

In most cases, tensile specifications require minimum results only. If that information is not obvious from the information already provided, we will need to know that. Using the specification minimum for a target is a very poor idea. It will result in many failures to meet the specification. We will establish a target at something above the minimum. While we perform the multiple regression analysis, we will also determined the variability currently being experienced. Our normal recommendation is to target two standard deviations above the minimum. We will inform you of our recommendation at the time we request you fill out the form.

Depending upon the degree of control being shown currently, this may be a very high target. If you feel targeting that high will have a very detrimental effect on the machinability of your castings, you may certainly choose a different target. (One thing to keep in mind in regard to the target is that if PREDICT is successful, the standard deviation will be reduced and we may be able to lower the target in the future and still maintain the same degree of certainty of meeting the specification.)

List the order in which elements are reported

Providing us with a list of the elements reported by your spectrometer and the order in which it reports them allows us to prepare your version of PREDICT so that the elements appear in that sequence. This makes data entry easier.

List elements not to be considered

Depending upon how you plan to use PREDICT, you may be using other tools to control some of the elements. As an example, if you're using thermal analysis frequently you may be using that to control your carbon and silicon. If that's the case, PREDICT treats those elements differently than those that are not controlled.

List elements that are to be controlled to a specific target

There are many instances when foundries are required, or decide, to maintain an element at a specific percentage. An example of this would be when a customer requires a certain percentage of nickel in a specific alloy. Another example would be when a foundry decides to use a copper and molybdenum alloy combination to achieve class 40 iron. It isn't unusual in that situation to specify a given level of copper and allow the molybdenum to vary as it is needed to obtain the desired strength. PREDICT will make the calculations to recommend the necessary additions to achieve those specific targets as well as the additions needed to achieve the tensile properties.

What element is to be used to achieve tensile control

Different foundries will use different elements to achieve tensile control. In fact, many foundries will use different elements to achieve control of different irons. In order to write the program to fit your operation, we need to know what you want to use to control this iron.

What are the compositions of the alloy is to be used

While copper and nickel are normally used in commercially pure grades, chromium and molybdenum are usually used as alloys. In order to make valid recommendations, we need to know what percentage of the alloys you're going to use is the desired element.

Other information you want to track

Naturally, there is information you want to keep track of with the chemistries and tensile properties. Almost everybody wants the date and time. Others want things like furnace number, ladle number, time from inoculation to pouring sample, time from pouring to shakeout, hardness of the test bar as measured at the laboratory, hardness of the test bar measured at the foundry, and the list goes on. In order to make the program most valuable to you, we need to know what you want. (A word of caution here -- make sure that the information you want to track will be available to the person who will be entering the tensile and chemistry information when that is available. Delaying entering tensile and chemical information while waiting for other information diminishes the effectiveness of PREDICT.)

Modifying the Program

With the information from the questionnaire, PREDICT can then be modified to fit the specific application. Depending upon the completeness of the questionnaire, this can take as little time as a week.

Testing the Program

If corresponding base chemistry, final chemistry, and tensile results were not previously available, collection of that information should start as soon as possible. Before the program is turned over to the foundry, it must be tested. That information can be mailed, faxed, or emailed to Lobenhofer Consulting. Ideally, at least 10 sets of complete information will be processed to assure proper operation.

Installing the Program

The program is then sent to the foundry. This usually happens by sending the program and required files on a 3 1/2-inch floppy; however, if the foundry has the capabilities, the information can be transmitted via email.

Once the information is received it should be copied into a directory on a hard drive created specifically for this. If the foundry personnel are not familiar with this process, Lobenhofer Consulting will take them through the process over the telephone.