Shortening the Retort Hold on a Low-Acid Canned Soup
A cannery packs a low-acid soup (pH 6.2, aw 0.98) in 400 g cans and processes them in a saturated-steam retort at 121.1 °C. The scheduled process, filed fifteen years ago, specifies a 25-minute hold after come-up; the original validation gave a cold-point F0 of 3.6 min with z = 10 °C and a reference D121.1 of 0.21 min for Clostridium botulinum. To lift throughput 20 %, production proposes shortening the hold to 20 minutes. A calibrated data logger placed in the slowest-heating can during a 20-minute trial recorded a cold-point history integrating to F0 = 2.3 min. Cooling water quality, the vent schedule and the can specification are unchanged. The retort's mercury-in-glass thermometer reads 1.0 °C higher than the calibrated reference at 121 °C, and the recorder chart has not been recalibrated for two years.
1. Determine the minimum acceptable F0 and judge whether the 20-minute hold can be approved.
Read the worked answer
The target is a 12-decimal reduction of C. botulinum spores: F0 = 12 × D121.1 = 12 × 0.21 = 2.52 min.
The 20-minute trial delivered F0 = 2.3 min, which is 0.22 min short, about 8.7 % below target. It cannot be approved on lethality grounds alone.
The 25-minute process delivered 3.6 min, a margin of 43 % over the 12D target, which is the sort of margin a filed process should carry because it absorbs can-to-can variation in heat penetration, come-up drift and raw-material changes. The proposed 20-minute process has no margin at all: it would fail on any batch where the slowest can heats slightly more slowly than the trial can.
The comparison also depends on the trial being representative. A single logger run in one can does not establish the slowest-heating can in the retort load; the correct practice is a heat-penetration study across the load with replicate cans, and the filed process should be based on the slowest can with a statistical margin. On the evidence given, the answer is no.
2. Quantify the effect of the thermometer bias and assess whether raising the setpoint can rescue the 20-minute schedule.
Read the worked answer
With z = 10 °C, lethality scales as 10^((T − 121.1)/10). A thermometer reading 1.0 °C high means operators actually run the retort at 120.1 °C when the gauge shows 121.1 °C, so every minute delivers 10^(−0.1) = 0.794 of the intended lethality. The filed 25-minute process therefore delivers about 3.6 × 0.794 = 2.86 min in routine production rather than 3.6 min, still above 2.52 but with the margin cut from 43 % to about 13 %. That is a finding about historical production, not just about the proposal, and the batches made since the last calibration should be assessed.
Correcting the bias alone restores 3.6 min. To make the 20-minute process equivalent to 2.52 min, the process would have to yield a factor of 2.52/2.3 = 1.096 over the trial, requiring a temperature increase of 10 log10(1.096) = 0.40 °C above the trial condition. On the biased gauge that means a setpoint near 122.5 °C.
Technically that closes the arithmetic gap, but it would be an unvalidated process change at zero margin, with poorer product quality and no heat-penetration data at the new temperature. A 1.0 °C bias also exceeds the ±0.5 °C accuracy expected of a retort mercury-in-glass thermometer, so the instrument is out of tolerance regardless.
3. What should the cannery do before any change to the scheduled process is filed?
Read the worked answer
First, retain the 25-minute hold. It is the only validated process in hand, and its margin is the safety system.
Second, fix the instrumentation: replace or recalibrate the mercury-in-glass thermometer against a certified reference, recalibrate the recorder, and institute a daily check against a traceable standard. The 1.0 °C bias is a control failure, not a rounding error, and it must be documented in the deviation record.
Third, assess exposure: identify all product processed since the last calibration, estimate the delivered F0 using the measured bias and the recorded temperature histories, and decide on release, rework or destruction. The 13 % residual margin suggests most product is probably acceptable, but that judgement must rest on actual records, not on assumption.
Fourth, revalidate properly if throughput must increase: a heat-penetration study in the slowest-heating can across the load, replicate runs, and a filed process set at a target F0 of about 3.0 min to preserve a margin, rather than 2.52 min exactly. Alternative throughput gains - larger retort loads, improved come-up control, or a different container - should be evaluated before any reduction in lethality is considered. Costing must include the recall exposure, which dwarfs a 20 % throughput gain.