Shell temperature monitoring: points, frequency, thresholds and false alarms
The usual mistake in shell temperature monitoring is watching absolute values: on the same line, a reading can move 20-30 C between a winter morning and a summer afternoon. This note gives a copyable measuring-point layout, a weekly and per-shift frequency rule, three threshold bands from watch to shutdown, and six false alarms that get blamed on the lining.
The conclusion first: read the spread and the trend, not the absolute number
Conclusion first: the useful information in a shell temperature reading is the spread across points on the same cross-section, and the rate at which a single point rises over time. A single absolute temperature tells you almost nothing about lining health.
The reason is that shell temperature is the sum of four or five factors: ambient temperature, wind speed, the condition of the insulation layer, the emissivity of the painted surface, and only last the thermal resistance of the lining. On the same 1 t/h line, the same point can read 20-30 C apart between a winter morning and a summer afternoon (site-dependent). A windward point is cooled down and a sun-facing point is pushed up, each by a double-digit margin. Judging remaining lining thickness from such a number is a division where neither the numerator nor the denominator is known.
One more point has to be stated plainly: shell temperature cannot be used to back-calculate remaining lining thickness. Thermal resistance is roughly proportional to thickness, but how much the shell temperature rises also depends on the absolute thickness, the state of the insulation and the surface heat transfer coefficient. The same temperature rise corresponds to very different amounts of wear on different furnaces. Shell temperature supports relative comparison only: a point against its own history, and a point against its neighbours on the same section.
The lining accounts for 8-15% of total equipment investment (in-house figure), with a design life of 3-5 years on a rotary kiln and 5-8 years on a shaft furnace. Once a shell glows red and forces a stop, relining takes 10-15 days on a rotary kiln or 15-20 days on a shaft furnace, before counting cooling and re-heating. The entire value of shell monitoring is moving those 10-20 days from unplanned to planned. Lining structure and insulation practice are covered on the furnace lining and shaft furnace lining pages; this note is only about how to measure and how to judge.
Measuring points: a fixed number matters more than a large number
| Location | Sections | Points per section | Layout requirement |
|---|---|---|---|
| Carbonization shell (450-550 C) | 2-3 | 4 (90 deg apart) | Keep 300 mm clear of welds, manholes and burner openings; stamp an ID at each point |
| Activation shell (850-1100 C) | 3 (front / middle / rear) | 4 | This is where red shells appear; IDs must match the equipment drawing |
| Feed end and discharge end | 1 each | 4 | Ends carry structural bridges; use readings for comparison within one furnace only |
| Shaft furnace body | 4-6 levels | 4 per level | Equal spacing between levels; add separate points at the combustion chamber and discharge |
| Afterburner and hot duct (850-1100 C) | 2 | 4 | Watch circumferential seams, expansion joints and support legs |
Frequency and records: a reading without ambient temperature cannot be compared
- Establish the baseline: before commissioning a new lining and within 72 hours of finishing the dry-out after relining, measure every point and file it. Without this baseline, no later trend analysis is possible; the dry-out sequence itself is described in the 72-hour dry-out curve.
- Cold-state re-measurement: at every planned stop, once the furnace is cool enough to approach, run a full set. Pairing cold and hot readings separates a lining problem from an insulation problem.
- Routine frequency: once a week in steady operation, same weekday and same time window, avoiding midday sun and the transient right after start-up.
- Escalation: when the cross-section spread enters the watch band, or a single point rises week on week for two consecutive weeks, switch to once per shift for three days.
- Full sets before and after every relining, used as the new baseline. Archive the old baseline, never overwrite it.
- Every record must carry: date and time, ambient temperature, wind speed, instrument model, emissivity setting and measuring distance. Miss any one of these and the data cannot be compared across seasons.
- Suggested record fields: point ID / three readings / section maximum and minimum / spread / week-on-week change / weather / action taken. One row per point; seven years of data still fits in a few hundred rows.
Thresholds: three bands, from watch to shutdown
| Criterion | Empirical threshold (adjust per site) | Meaning and action |
|---|---|---|
| Circumferential spread on one section | <=20 C normal; 20-40 C watch; >40 C abnormal | At watch, escalate frequency and re-measure in place; at abnormal, schedule borescope or thickness survey alongside process data |
| Week-on-week rise at one point | <=5 C normal; 5-15 C watch; >15 C abnormal | Rate exposes problems earlier than absolute value; two consecutive weekly rises enter the watch band |
| Absolute shell temperature | Per design documents; without one, use 'not above ambient +50 C' as a rough screen | Check insulation and paint first, lining second; a dull red shell (roughly above 500 C) is an emergency |
| Spread between axially adjacent points | >40 C treated as abnormal | Usually a structural bridge at a manhole, weld or support, not lining damage |
What to do after a hot spot shows up
Step 1: re-measure in place and rule out false alarms first
Measure the same location three times from different angles, then once more with a second instrument, recording ambient temperature, wind speed and sun direction. A good share of hot spots disappear at this step; do not escalate before you have done it.
Step 2: widen to the whole section
Measure all four points on that section and add one section forward and one back. A single high point with normal neighbours looks like local spalling or an open brick joint; a whole ring running hot points to failed insulation or a process over-temperature.
Step 3: pull the history
Look at the last 4-8 weeks for that point and decide whether the rise is a step or a ramp. A step usually means local spalling or cracking; a ramp usually means even wear or insulation degrading year by year. The two need completely different intervention windows.
Step 4: check the process side
Confirm whether the activation zone has been running at the top of its 850-1100 C band, whether the furnace has cycled on and off, whether feed moisture has swung, whether there is local coke build-up. Process excursions usually leave a mark on shell temperature before any instrument alarm.
Step 5: verify at a planned stop
Use the relining window for a borescope, thickness survey or local brick removal: 10-15 days on a rotary kiln, 15-20 days on a shaft furnace. Do not water-cool the shell from outside or weld a patch over it during operation; that converts a plannable repair into an unplanned outage.
Step 6: reset the baseline
After the repair, re-measure every point, adopt the new data as the next baseline, and mark the repaired area on the equipment drawing so the next round does not re-investigate the same point.
Six false alarms that get the lining blamed
1) An afternoon sun-facing side reads 15-20 C above the shaded side and is called a hot spot. 2) A windward point is cooled and hides the real temperature. 3) Emissivity is left at a default instead of the 0.8-0.95 range typical for oxidised steel, shifting every reading. 4) Measuring distance exceeds the instrument D:S ratio, so the spot is larger than the target and readings are averaged. 5) A surface thermocouple has debonded or its lead has broken, dropping the reading to ambient. 6) Damaged insulation or flaking paint is treated as a lining problem. All six sit outside the furnace, and fixing them costs far less than removing brick.
Related notes
Send the raw material, target tonnage and finished spec. You get a configuration, a utility schedule and a budget range — not a brochure.