RGB ENGINEERING

BED CORRECTION FEATURE

COMPUTERIZED CROWNER RETROFIT SYSTEM

Roll Shape Comparison - Effect of Bed Correction Curve

Plot #1 on the right shows the Pro-Mic record of a hot mill work roll ground in a 1969 vintage 36" Farrel grinder in a steel mill roll shop. For this test grind, the crowner system was turned off, allowing the machine to cut the natural "bed" shape into the roll. As can be seen, this is a severely distorted convex profile. It would be impossible to compensate for this non-symmetrical shape error using a standard mechanical crown cam.

Plot #2 below is the Pro-Mic shape after grinding the same roll with the RGB Engineering crowner system running with the bed correction feature activated. Since the bed correction is superimposed automatically with the targeted crown curve, the operator simply sets the desired crown, which in this case was "0" for a flat roll.

Bed Correction Curve - Programming Example

The bed correction curve is input as an x-y table of values in the FFP program. The "x" values represent carriage (or table) encoder readings which correspond with marked station readings on the Pro-Mic plot. For this example, the encoder reading at station "0" is 81.84 and at station "20" it is 157.13. As many points as necessary can be selected to properly define the curve. The software has curve smoothing algorithms that will fit the curve to the points selected. The final bed correction curve must be continuous for the entire wheel travel, even though the end sections will probably be outside of the normal grinding area. For these regions, the curve can be extrapolated as shown below. The "y" ordinates are obtained from the Pro-Mic plot at each station using any convenient scale. The final scaling is fixed by the "bed correction factor", input in the Setup menu of the FFG program. This factor is the largest ordinate distance from the highest point on the curve to the baseline and must be in radial measurement units.

As shown above, it is not unusual to have one or two iterations before the final bed correction curve is defined. Follow-up checks can be done periodically to monitor the "flat" roll profiles. Once the technique is established, it is an easy matter to "tweak" the curve to keep the grinder running at optimum performance.

Test Grind - Crown Shape

Using the same bed correction curve from the example above, a test grind was done in this machine to grind a crown shape with the RGB Engineering FFG/FFP/SCA control.  The results are shown in the Pro-Mic plot reproduced below.  The crown height is slightly less than the target due to the fact that this Pro-Mic does not measure all the way to the end of the roll, which is the true end point of the crown.  This crowner retrofit is installed on a 1955 vintage 36" x 16' Farrel roll grinder in a steel mill cold reduction roll shop.  The grinding was done on an 82" face four stand work roll.  As can be seen from the original test grind without the bed correction, it would be impossible to produce this accuracy with a manual crowning system due to the unsymmetrical shape of the bed error.  This would have the effect of shifting the apex of the crown toward the headstock, with a corresponding distorting effect on the overall shape accuracy.

Example of Bed Correction Superposition

These three plots were made using an auxiliary LVDT mounted as shown to measure the actual wheelhead tilt motion. The tests were done on a 1965 vintage Farrel grinder equipped with the RGB Engineering FFG/FFP/SCA crowner system.  Plot "A" on the top shows the wheelhead motion without the bed correction curve activated, for a 70 degree profile shape with a .005" concave setting. The roll end limits "TSH" and "TSF" define the crown axis for the 80" face roll. Plot "B" in the middle was made with the bed correction curve activated, and with a zero crown setting. This curve shows only the inverted bed error, without any crowning contribution.  The TSH and TSF trip points are overlaid on the plot to indicate the portion of the curve (yellow area) that will be superimposed with the crown profile. Plot "C" on the bottom was taken with a .005" concave programmed and with the bed correction curve active. This is simply the algebraic sum of plots "A" and "B". The small end taper (vertical offset between TSH and TSF) will be automatically compensated for when the operator makes normal taper corrections. As can be seen from this example, the working portion of the bed correction curve is automatically selected when the operator sets the TSH and TSF end stops in the FFG setup menu.

RGB/Pro-Mic Integration

RGB Engineering and Pro-Mic Corporation are working together to "automate" this bed correction process.  This is an ongoing development project in an Aluminum mill roll shop, which has a RGB crowner installed on a Cincinnati TT grinder and a Pro-Mic with modified software .  After each final roll shape is skated, the operator initiates a download of the data to a holding file, along with a few key parameters (target crown and end stop settings).  After an extended period, these accumulated  roll shapes are sent to RGB Engineering for processing in a data base program.  Each plot is analyzed and compared to the true target shape, resulting in a "deviation" curve.  These are averaged and further processed to produce the final bed correction curve, which is e-mailed to the customer and then installed in the FFG replacing the original bed curve. 

After sufficient testing, the plan is to establish a hard-wired connection between the Pro-Mic computer and the RGB FFG crowner which will essentially automate the complete bed correction process.  This will then be a form of true "adaptive" control, since the grinder will always be programming itself to continuously optimize the roll shapes, without operator intervention.  This simplified form of adaptive control is a much less expensive option compared with the newer style CNC grinders with in-process calipers.