Who Misread the Figure? UFC’s Gas Pressure Chart and Spall Thresholds Against Tests

1. Two last questions

After SIG-06 two questions remained: for peak gas pressure in a chamber the site’s engine gave a value 4.3% below UFC 3-340-02’s own worked example; which is right? And how well do UFC’s spall and breach thresholds agree with real stand-off tests?

2. Who misread Figure 2-152?

UFC Example 2A-7 reads 2,650 psi from Figure 2-152 at W/V = 1.0 lb/ft³; the engine gives 2,537 psi. Independent sources do not settle it: the 177-test fit of Baker et al. (1983) gives about 2,060 psi and Lloyd’s Register’s naval rule about 2,405 psi. Both lie below the figure, which is a conservative design curve. The figure itself decided.

The figure was re-measured, calibrated from its gridlines: each logarithmic decade line was identified from the spacing pattern (the 1→2 gap is the widest in every decade). This also explained the first attempt’s error: on the pressure axis the 10,000 psi line is not the frame edge but lies inside it. With correct calibration the new measurement reproduces the engine within 0.1–0.3% over W/V = 0.002–4 (2,539.5 and 2,537.5 psi at W/V = 1). The printed line itself spans 2,474–2,597 psi there; 2,650 psi lies outside the line. Conclusion: the engine is faithful to the figure; UFC’s example read its own figure about 4% high. The error is on the conservative side, so the engine was not changed. One caveat remains: we have the 2008 edition, and Change 1 (2014) extended the figure to lower W/V.

3. How well does UFC match tests?

Holgado et al. (2024) showed on a 463-test dataset that the UFC §4-55 thresholds hit only 52–65% per damage class; since row-level data were not published, this serves as a performance envelope. Mohotti et al. (2024) compared their 13 tests with UFC and published what UFC predicted for each. This allowed comparing the site’s engine with an independent implementation: with a surface burst and the charge over the 225 mm band beam, the engine reproduces all four of the paper’s UFC predictions (breach for contact S14; spall for S21, S23 and S42) for any C-4 TNT equivalence from 1.0 to 1.37. That means “consistent”, not “verified”; changing the input configuration changes the result.

In the same four tests UFC’s prediction is more severe than observed: S14 showed major spall, the other three no damage. For high-strength (55–61 MPa) concrete UFC stays on the safe side.

4. Conclusion

• Peak gas pressure: the engine is faithful to the printed figure; the example’s 2,650 psi is a conservative misreading.
• Spall and breach: the engine implements UFC §4-55 exactly and is consistent with an independent implementation; UFC itself matches tests moderately and is on the safe side in these examples.
• v636 — tested against McVay (1988): the report’s Appendix D table (public) was read from page images; 204 rows passing two internal checks against the printed R/W^1/3 and T/W^1/3 were saved as a dataset. With a surface burst the engine scores 60.4% on the 101 classifiable tests — inside the 52–65% Holgado et al. (2024) found for UFC on 463 tests. 16 of 21 breach tests are right; 14 of 20 no-damage tests are predicted damaged (conservative); but 11 spall tests are called “no damage” and 5 breaches only “spall” (unsafe). 81 tests fall outside Ψ = 0.5–14, where the engine does not decide.