11 replies to this topic

[jcieslowski]

I'm currently struggling to wrap my brain around a verification on a cooking process in my facility.  We have a pressure cooker to cook beans.  FDA requires that beans are cooked at 200°F for 3 minutes to render safe to eat.  Our cooker (which is pressurized), cooks at a minimum of 230°F.  All of our programs have dwell times of 9 minutes plus.  There is a digital readout of the temperature and an analogue thermometer dial on the side of the cooker.  

 

Weekly, as part of a weekly walkthrough, the QA manager (me) documents the analogue temperature against the digital readout to ensure they're the same (always have been so far).  Monthly, the QA manager (still me) compares the analogue thermometer when the machine is off (ambient) to a certified lollypop style thermometer (we just get a new one each year as opposed to re-calibrating / re-certifying).

 

Recently, my FDA inspector challenged (I'm a little annoyed because she said it was just a discussion point for us but then wrote it up as a violation) that just because the calibrated thermometer matches the analogue thermometer at ambient conditions does not mean it would match it at temperature (don't think I agree on this point) and that comparing the digital and analogue at temperature doesn't ensure that either is right (which I sort of agree with this part - it's all dependent on the first thing).

 

Anyway, if someone could just tell me that I'm not completely crazy for thinking this is a bit much, I'd appreciate that but what I REALLY am looking for is ideas on how to address this.

 

I can't put anything into the cooker.  It's got a corkscrew, some screens, and a water fed chute that I'm afraid of getting anything stuck in.   The cooker is at pressure, so I can't measure the inside from the outside.   I tried using a temp gun to find the hottest part of the metal but it's double walled so I'm seeing a 20° drop from the thermometers to the outside of the cooker.  

 

Anyone experience anything similar and what do y'all do?

[TimG]

"Recently, my FDA inspector challenged (I'm a little annoyed because she said it was just a discussion point for us but then wrote it up as a violation)"

Huh..did the same thing to me in my last visit. Wrote up a bunch of 'talking points' as violations...

 

Anyway, we have ports right after our inline therms that we can take a high temp sample of to match our annually replaced NIST calib therm. I would agree that ambient temps wouldn't work for accuracy checks, and you would want to check accuracy in the range of what your process therm is working at.

 

It helps that our temp thresholds are for processing only (quality), not food safety. 

[GMO]

I'm currently struggling to wrap my brain around a verification on a cooking process in my facility.  We have a pressure cooker to cook beans.  FDA requires that beans are cooked at 200°F for 3 minutes to render safe to eat.  Our cooker (which is pressurized), cooks at a minimum of 230°F.  All of our programs have dwell times of 9 minutes plus.  There is a digital readout of the temperature and an analogue thermometer dial on the side of the cooker.  

 

Weekly, as part of a weekly walkthrough, the QA manager (me) documents the analogue temperature against the digital readout to ensure they're the same (always have been so far).  Monthly, the QA manager (still me) compares the analogue thermometer when the machine is off (ambient) to a certified lollypop style thermometer (we just get a new one each year as opposed to re-calibrating / re-certifying).

 

Recently, my FDA inspector challenged (I'm a little annoyed because she said it was just a discussion point for us but then wrote it up as a violation) that just because the calibrated thermometer matches the analogue thermometer at ambient conditions does not mean it would match it at temperature (don't think I agree on this point) and that comparing the digital and analogue at temperature doesn't ensure that either is right (which I sort of agree with this part - it's all dependent on the first thing).

 

 

Sorry. But I agree with the inspector. Albeit I don't agree on raising something after they've left site.

When you're using a thermocouple for a CCP, or any measuring device really, you need to measure it over the range you want it to be accurate over. Not just at a single point and not even multiple points far away from the temperature you need to be sure it's at so the food is safe.

 

So if we go back to your cooker. It cooks at 230F for 9 minutes.

 

So your dwell time is 9 minutes, you'd want to be sure that's accurate and your temperature is 230F.

 

You have a digital readout of the temperature. Therefore there is at least one temperature probe in the system. Calibrate it. And if you need someone to come in to do that for you, why would you skimp on that out of all things you have in your process? How they'd do it? I guess they may be able to detach it and test it in a heat block? Unsure. But ask an engineer who knows what they're doing.

 

If you did find a way to do it yourself, the old phrase is it must be tested against something which is "calibrated and traceable to national standards". So if your probe you're testing it against is calibrated against a reference probe which is calibrated to national standards (it would be UKAS in the UK, not sure in the US) then fine. But you'd still need to test more than one data point spanning the range you're interested in.

 

Lastly, I'm not sure on your cooker. But even pressurised canning systems have dataloggers specifically designed to be used within them. These kinds of things. Although if you're cooking the beans in a free flowing process, I'm wondering if you could attach the second in some way to a part of the process so it's not clattering around inside?

 

Canning logger for thermal food treatment in retorts and sterilizers

EBI 11-T230 Robust logger for sample control

 

Just a thought. But calibrating the existing probes will probably be easier.  Also if you do have two probes in the system (and the digital and analogue read out are genuinely linked to two different probes) that's helpful as well to monitor there's no drift between calibrations. I'd get my team members checking that and have a limit how far apart they're allowed to be before you stop.

[dsglinski]

I'd agree that temperature measuring devices can (and often do) have variation in their accuracy/correction factors across large temperature ranges. 

 

Coming from more of a laboratory background, I would consider the option of using a datalogger purpose made for high pressure, high temperature, waterproof use. I've used HOBO Dataloggers from Onset for temperature stability studies, validation and verification studies of laboratory autoclave cycles. It's likely this sort of instrumentation would allow you to go above the bare minimum in showing your time at temperature in your pressure cooker.

 

For example, I've used this model before: https://www.onsetcom...loggers/u12-015

[jcieslowski]

"Just a thought. But calibrating the existing probes will probably be easier."  - I think this is the answer!   Thanks, now to find someone who can do that.

 

Much obliged. 

[Tony-C]

"Just a thought. But calibrating the existing probes will probably be easier."  - I think this is the answer!   Thanks, now to find someone who can do that.

 

Much obliged. 

 

Hi jcieslowski,

 

I’m of the view that all process thermometers should be calibrated anyway?

 

I have come across this scenario previously supplying USAF according to the PMO standard and we used to remove dual thermometer probes (for auto-divert) whilst they were still operational and test them in a water bath against the NPL calibrated thermometer. Something you could look at as you could test up to around 210F with boiling water.

 

Kind regards,

 

Tony

[GMO]

 

I’m of the view that all process thermometers should be calibrated anyway?

 

I have come across this scenario previously supplying USAF according to the PMO standard and we used to remove dual thermometer probes (for auto-divert) whilst they were still operational and test them in a water bath against the NPL calibrated thermometer. Something you could look at as you could test up to around 210F with boiling water.

 

Yep I agree. I brought it up because I thought if records were available of this, they would have shown them. Still, worth checking Engineering aren't already doing it.

 

You've made me think though Tony. Is there the potential to do a more frequent verification of the probes as well? If you can't easily remove the probes (although it would be great if you could) If this is a flow cooking system for beans, could you run water through that flow at cold and hot temperatures and measure it with a handheld probe, ideally if there's a recirculation mode so everything is stabilised? (I'm not sure if the temperature is the same throughout.) Could be an option if you can do it safely with hot water and with cold water.

[matthewcc]

I didn't see it mentioned, but have you considered a thermolabel such as this?

 

https://paperthermom...abel-200-f-93-c

[jcieslowski]

Ok, everyone, here's an update just for you all.

 

Turns out the thermometer can be REMOVED with one wrench.  It just unscrews.  (has to be unpressurized).

 

So I just waited til we were down, took it out, and tested it against my NIST certified / calibrated thermometer in a pot of boiling water.

 

The hardest part is that the thermometer stem is very short so some care was needed to not burn the heck out myself and try to read the dial against the steam.

 

The only inconvenience is that we need to be down and that means after hours  / overnight work.

 

It ALSO turns out that new ones come certified for 1 year.  This one is 6 years old and still 'calibrated'.   

 

My boss (bosses, amirite), says that we should save the money and just validate ourselves with the check annually.   

 

I know the real answer is risk-based but what do you all think should be the frequency of action on this sucker?  I have trouble defining a risk because the cooker doesn't run below 230 and the critical limit is 200.

 

Just to recap, here's the controls / checks:

 

1. Analogue vs digital compare daily by QA tech

2. Weekly 'CCP check' comparing analogue vs. digital by QA Manager

3. Monthly analogue vs. NIST thermometer at ambient conditions (not really helpful - maybe remove)

4. ??? Vs. NIST in boiling water  (not hard but need to be down)

5. Yearly NIST calibrated thermometer installed (if my boss goes for it.  It's not an expensive thermometer by any means)

 

Thoughts?

[TimG]

Uhh..well how much is the new in line thermometer? (sorry, bosses gotta boss...)

[jcieslowski]

Uhh..well how much is the new in line thermometer? (sorry, bosses gotta boss...)

 

Right!  $130

[jfrey123]

I used to work for a company that sterilized spices and other stuff in a large dry steam chamber, we had to do thermoprobe verifications there too.  Snipping a couple of jcieslowski's items from post #9 in bold to respond:

 

So I just waited til we were down, took it out, and tested it against my NIST certified / calibrated thermometer in a pot of boiling water.

This is what I used to do.  The auditors will really like you if you also test them in an ice bath and record the high/low against your certified and calibrated thermometer.  Bonus points:  research what temperature water boils at your facility's altitude.  My town being at 4400ft, water boils at 204.  Had an auditor try and tell me my certified thermometer used to verify the probes must be broken because we targeted 204F on the form, so I had to educate him that lower pressure at higher altitudes affects boiling temperature.  Luckily also had it referenced in the SOP because he wanted to write me up regardless...

 

My boss (bosses, amirite), says that we should save the money and just validate ourselves with the check annually.

Annually feels like a long time, but you're documenting that check of the analog vs digital during processing on a daily basis so that means if one of them started to fail then you would know it.  And the likelihood of them both failing at the same time is nil.  Our steam chamber had 6 in the walls and I think 4 that went into product.  We successfully defended monthly checks, also knowing that a faulty probe wouldn't read low or high, it would merely stop reading and there was sufficient redundancy to know the treatment was still effective.

 

I know the real answer is risk-based but what do you all think should be the frequency of action on this sucker?  I have trouble defining a risk because the cooker doesn't run below 230 and the critical limit is 200.

I absolutely love that you've written the critical limit to 200F knowing your system operates at 230F:  this means should a probe be discovered reading 10F low, you still met your critical limit.  So if analog during a run says 230F, and then the digital shows 220F, you've still proven you're exceeding your critical limit.  But such a deviation should trigger a spot check after the run to figure out which probe is off and replacement of the faulty before continuing.

 

I don't like that you're calling it a daily QA tech check of digital vs analog with a "weekly" CCP check...  Your daily checks should be more frequent as the CCP check, even for each batch going through that 9 minute cycle.  Then monthly you could do the pull apart test of digital and analog against your calibrated standard thermometer.

 

3. Monthly analogue vs. NIST thermometer at ambient conditions (not really helpful - maybe remove)

Replace with monthly check of the analog and digital against your calibrated stand alone thermometer in boiling water (with also the ice bath water if you want to be extra cool like I said).

 

5. Yearly NIST calibrated thermometer installed (if my boss goes for it.  It's not an expensive thermometer by any means)

Replacing this probe annually is nice as it keeps it within the "certification" period, but I've defended continuing to use a thermometer as long as it proves accurate against my calibrated stand alone.  I'll admit I'm about 10 years out of practice on this, however, so I'm open to letting someone suggest that isn't a good practice anymore.