Engineering question
Why can average billet length be correct while individual CCM billets remain inconsistent, and what evidence should be trended?
Accuracy and repeatability describe different problems. A stable 30 mm long result can often be corrected with a parameter; alternating long and short billets usually indicate variable contact, timing or machine response. Treating both with repeated scale-factor edits makes the system difficult to diagnose and can worsen production variation.
A useful review follows the event chain: strand movement, encoder pulses, high-speed count, target crossing, captured count, permissives, cut command, actuator movement, material separation and final inspection. Each transition should have an observable value or timestamp so that variation can be assigned to a part of the process.
Calculation basis
Formulas and units
Length error
error = measured billet length − target length
Keep the sign convention consistent in reports and compensation logic.
Repeatability spread
range = maximum error − minimum error
Range is easy to use on shift data; standard deviation is better for a stable, larger sample.
Delay-equivalent length
Ldelay = casting speed × effective delay
Convert units consistently; at 1.2 m/min, each 100 ms is about 2 mm of strand travel.
Worked example
Apply the formula
At 1.2 m/min, five billets show errors of +18, +21, +19, +20 and +22 mm.
- 1Mean error = +20 mm.
- 2Range = 22 − 18 = 4 mm.
- 3The small spread with a stable positive offset suggests compensation or reference-point error rather than poor repeatability.
Result: Correct the identified constant offset only after confirming the measurement basis; do not retune the geometric scale from this five-billet result alone.
Open Encoder Length & Pulse CalculatorEvidence to collect
- Raw and scaled counter at target crossing
- Hardware capture value and PLC task timestamp
- Casting speed at command and separation
- Shear/gas-cutter ready, command, motion and completion timestamps
- Measuring-roll pressure or contact state
- Billet inspection result, temperature basis, strand and heat reference
Error-pattern diagnosis
A percentage error across different target lengths points toward encoder scale. A fixed offset across lengths points toward reference position or delay. Speed-dependent error points toward variable delay, slip or a compensation law. Random spikes point toward lost counts, poor contact, asynchronous reset, intermittent permissives or mechanical cutting behaviour.
- Compare error versus target length.
- Compare error versus casting speed.
- Compare by strand, shift, wheel condition and cutting device.
- Do not combine data from different hot/cold inspection methods.
Common mistakes
- Changing calibration after every billet
- Averaging two strands with different mechanics
- Using the HMI display timestamp as a control-event timestamp
- Ignoring measurement-system repeatability
Troubleshooting checks
- Stable offset: inspect reference point and cutting delay.
- Percentage error: inspect scale chain and effective wheel diameter.
- Speed-correlated error: inspect delay compensation and wheel slip.
- One-strand-only issue: compare encoder mounting, I/O channel and cutting mechanics between strands.
Assumptions
- Actual billet length is measured consistently
- Casting speed is available at useful resolution
- Command and feedback timestamps share a reliable time base
Limitations
- Statistical evidence does not replace mechanical inspection
- Thermal contraction and crop practice depend on plant-specific conditions
- No universal acceptance tolerance is implied
Topic cluster