Home/Insights/Recovery boilers
Insights · Recovery boilers · Failure analysis
Where recovery boilers leak
Every incident reported to the Black Liquor Recovery Boiler Advisory Committee between 2001 and spring 2026, 1,103 of them, mapped onto the boiler by location and root cause. Click around. Then read what the numbers suggest you do about it.
The map
Click a region of the boiler
Gas flows left to right: furnace, screen, superheater, boiler bank, economizer. Regions are shaded by incident count. Switch to quality-control defects to see where workmanship, not service, is the cause.
Incidents by boiler location, 2001–2026
All locations
Whole boiler, 2001–2026
Not tied to one location
Table view: every location and cause
By year
No sustained decline in twenty-five years
Reported incidents ran at about 44 a year from 2001 to 2012 and about 43 a year from 2013 to 2025. The quality-control share held near one in five throughout. Reporting is voluntary and membership shifts, so read the counts as a floor, not a rate.
Reported incidents per year
* 2026 covers the spring meeting only. Hover or tab across the columns for the values.
Table view: incidents per year
What the numbers say
Four things to take back to the mill
The record is public. What it is telling you about your own inspection and repair program is less obvious.
The economizer is where workmanship shows up
Forty-one percent of every incident in the record is in the economizer, and so are 149 of the 215 quality-control failures. One economizer leak in three traces to a weld or fabrication defect. In the superheater it is one in twelve; in the boiler bank, one in seven. The economizer carries more field welds, more header-to-tube joints and more retube work than any other section, and the record shows it. That is where the weld inspection budget belongs: tube-to-header welds, field welds from retubes, and every repair weld before the unit goes back on line.
Economizer 452 incidentsQC share 33%Superheater QC share 8%Boiler bank QC share 14%
One leak in five was built in, not worn in
Weld and fabrication defects plus material, manufacturing and design defects account for 19.5 percent of the record, and that share has barely moved: 19 percent across 2001 to 2025, never below 8 percent in a year, never above 34. These are not failures of the operating environment. They are failures of fabrication, installation and repair quality control, and another thickness survey will not find them before they leak. Welding procedure, welder qualification and examination of the finished weld will.
Weld / fabrication 208Material / mfg / design 7Share 2001–2025 19%
Aging mechanisms are still the majority, and they are what inspection is for
Corrosion, thinning and erosion (204), fatigue (141) and stress-assisted corrosion (129) together make up 43 percent of the record. These are the mechanisms an in-service inspection program exists to find, and they reward a program that knows where to look. Stress-assisted corrosion in particular sits at attachment and membrane welds in the economizer and boiler bank, where a thickness reading on the tube crown tells you nothing. Inspection scope should be written from the mechanism, not from the drawing.
Corrosion / thinning / erosion 204Fatigue 141Stress-assisted corrosion 129Combined 43%
One in eleven was never explained
One hundred incidents carry a root cause of unknown or undetermined. A leak without a cause is a leak that comes back, usually in the same place, often in the same outage cycle. The metal, the deposits and the operating history almost always hold the answer if someone is tasked to read them before the tube section goes in the scrap bin. Independent failure analysis is what turns a repair into a fix.
Unknown / undetermined 100Share 9%
Method and caveats
How the record was built
Every incident appendix in 45 BLRBAC semi-annual meeting-minutes files, spring 2001 through spring 2026, was extracted into one dataset: 1,103 incidents, each with its reported location and root cause preserved verbatim. Incidents were then grouped by boiler region and classified by primary failure mechanism.
Classification is position-based. A fatigue crack at a weld counts as fatigue (the weld is the location). Weld porosity, lack of fusion, slag, or a poor field weld counts as a weld defect (the weld is the cause). Quality-control defects are the sum of weld / fabrication defects and material / manufacturing / design defects.
Totals were validated against BLRBAC’s own per-meeting tallies. Location grouping is roughly 85 to 90 percent concordant with BLRBAC’s own categories; root-cause classification from free text carries a few percent of uncertainty. Read 19.5 percent as “about one in five.” This was built for analysis, not certification. Mill names are in the source minutes and are deliberately not reproduced here.
Source
- Black Liquor Recovery Boiler Advisory Committee (BLRBAC), semi-annual meeting minutes, incident and emergency shutdown appendices, 2001 to spring 2026
- 45 meetings. Reporting is voluntary and limited to member mills, so counts are a floor
- Analysis: Robert Vesak, Trust Integrity Management Inc., 2026
Root-cause categories used
- Weld / fabrication defect and material / mfg / design defect (together: quality control)
- Corrosion / thinning / erosion
- Fatigue; thermal stress; stress-assisted corrosion (SAC)
- Overheating / flow; mechanical damage; gasket / seal
- Operator / procedure; no leak (ESP only); other; unknown / undetermined