Industry numbers, with their methods
· Erman Demirel · 6 min read
Summary
These figures cannot be added together; they come from different definitions, countries and years. Next to each item are its year, sample and how far it can be trusted. WeldTracker's own result figures are not published yet.
Loss on an industrial site is not in one place. The eight items below were each measured in published studies, or noted as not measurable. The figures cannot be summed. Each item ends with its evidence grade.
1. Rework
The cost of rework ranges from 0.4% to 12%, depending on how it is defined.
- 12.4%: CII study published in 1987, 9 industrial projects. Old and small sample.
- 0.38% (before completion) and 0.76% (including after): the Love study, summarized by ASCE in 2026. The same study reports that project records understate true cost by about 300%.
What it means: the newest study does not say "rework is low"; it says "records are incomplete". A loss that is not recorded cannot be measured.
Evidence grade: measured (the ASCE summary was read; the paper itself was not accessible).
2. Searching for information and resolving disputes
In the 2018 FMI and PlanGrid survey, about 600 respondents reported spending more than 14 hours a week on non-productive work: searching for project data (~5.5 h), resolving disputes (~5 h), fixing defective work (~4 h).
Limit: survey-based, respondents mostly US commercial construction. Not directly applicable to industrial plants.
Evidence grade: industry estimate.
3. Information transfer
A 2004 NIST study estimated the cost of inadequate interoperability in US capital facilities at USD 15.8 billion a year (2002 data; 105 interviews with 70 organizations). Of this USD 15.8 billion, USD 10.6 billion falls on facility owners and operators. Looked at by life-cycle phase, USD 9.1 billion falls in the operations and maintenance phase. The two figures are different slices of the same total; they do not add up. The authors consider the figure conservative.
Limit: 22-year-old data, US, interview-based.
Evidence grade: industry estimate.
4. Productivity
According to the McKinsey Global Institute's 2017 report, construction productivity grew about 1% a year for roughly twenty years, against 2.8% for the global economy and 3.6% for manufacturing in the same period.
Evidence grade: industry estimate (reputable source, method in the report).
5. Weld repair
According to TWI's industry survey, the average weld repair rate in oil-and-gas and power work is 1-3% (for pipes, "repaired welds ÷ total welds"). Local peaks 25%, exceptional 55%.
Limit: TWI 2011 survey, sample size not stated; the rate was calculated by two different methods (repaired weld length ÷ total length, or number of repaired welds ÷ total number). For piping systems, all materials together 3.0%. Peaks are mostly in root passes, fillet welds and hard-to-reach places. It is not the same as the first-radiograph rejection rate.
Cost: published per-repair cost figures mostly do not state their scope. A customer case page from an equipment manufacturer says "USD 800-1,000 per test" and does not say what it includes (marketing page, not audited).
Evidence grade: industry survey (TWI, 2011). No audited unit figure for cost; USD 1,000 per repair is one manufacturer's example.
6. Missing documents and data at handover
We could not find a measured rate for "what percentage of documents is lost at handover". The concrete finding is data defects: in a CFIHOS application with Petrobras data, not all mandatory attributes could be mapped (some fields missing, some empty, different hierarchy). An estimate attributed to FIATECH: owners spend 2-4% of project cost correcting O&M information; this is an example multiplication, not an audited average.
Evidence grade: case/application paper and industry estimate. No generalizable rate.
7. Material traceability
There is no established "wrong material rate". What exists: a 1995 paper estimates unintended substitution in about 1-3% of installed alloy components (based on "documented experience", not a controlled sample). In documented incidents (BP Texas City, CSB 2006) the problem was not the certificate: during maintenance, similar-looking elbows were mixed up, a carbon steel elbow was fitted to a high-temperature line, and no material verification was done after installation.
In a refinery consultancy case page (Pinnacle Reliability, April 2022), more than 420,000 procurement records from 17 sites were reviewed; in about 1.8% of records PMI appeared necessary but the PMI flag was missing, narrowed to ~5,000 records by expert review. This is a missing-control-flag rate, not a wrong-metal rate; no physical PMI result is given.
Evidence grade: 1995 technical paper (estimate), CSB incident bulletin, consultancy case page. Not combined.
8. Hydrotest and test packages
We could not find a combined rate for failed hydrotests or returned packages on B31.3 plant projects. The one verified case: two leaks in a newly installed 12" 304 stainless line at hydrotest, caused by microbiologically influenced corrosion (MIC), not a weld defect; the total number of joints tested is not given, so no rate follows.
In a composite (GRE/GRV) underground gas line case (NDE 2019, CP-312), leaks occurred at a joint near the ultrasonic valve pit and at a tee; the causes listed were non-compliance with manufacturer recommendations, insufficient support in the pit, and a tee manufacturing defect; supply of the large-diameter tee delayed the project by 6 months. Not metallic B31.3.
In a lean planning study at an NGL plant (SPE, 2017), 75.6% of packages were completed and approved by the time panels arrived; this is a completion rate, not a rejection rate.
Evidence grade: two published cases (composite line, 304 stainless) and one lean planning study. Return before test and leak at test are measured separately.
What record do these losses become in WeldTracker
This table does not say "this much saved". It only shows which item is put on record.
| Loss item | Visible in WeldTracker |
|---|---|
| Weld rejection, rework | Weld log, NDE record, non-conformance record: which weld, who made it, which material batch |
| Searching for information | Work preparation (Workfront), document control (DCC), QR part passport |
| Revision disputes | Isometric revision tracking, red-mark board, conflict log |
| Material uncertainty | Heat number tracking, off-cuts, warehouse, cold stamp, QR |
| Test readiness | Test package (hydrotest), line readiness status, punch list |
| Information loss at handover | As-built package, databook, CFIHOS export (partial coverage) |
What we do not claim
- We do not yet publish WeldTracker's own result figures. When measured, we will publish them with their method.
- The figures above come from industry studies; they are not WeldTracker's effect.
- Ranges are wide because definitions differ. Giving a single "loss percentage" would not be right.
- For three areas we found no combined rate: missing documents at handover, wrong material rate, hydrotest/test package returns. We have not yet found Türkiye-specific data either.
Sources
- NIST GCR 04-867 (2004): nvlpubs.nist.gov
- CII, Costs of Quality Deviations in Design and Construction: construction-institute.org
- ASCE, "How much does field rework in construction actually cost?" (22 Jan 2026): asce.org
- Love, Quantifying the Costs of Field Rework in Construction: doi.org
- McKinsey Global Institute, Reinventing construction (2017): mckinsey.com
- FMI/PlanGrid (2018), Construction Dive summary: constructiondive.com
- TWI, industry survey of weld repair statistics (2011)
- Pinnacle Reliability, refinery procurement records case page (April 2022)
- Gunasekaran, NDE 2019, CP-312; Ibrahim, SPE 2017
- CSB, BP Texas City report (2006)
Related articles
Rework and scattered data, with their sources
Published studies on the cost of rework and how much of it comes from project data, with their years, samples and limits.
Oct 8, 2026 · 6 min read