Periodic re-Validation, optimization of procedure

i’m Marco from Italy, and i’m actually doing a internship (before master graduation) in a big pharma company with the aim of finding new kind of procedure to revalidating equipment as autoclaves, oven, SIP station and so on.

i’m new in the pharma world and i would like to know your opinion and suggestion on a basic issue.

What i can see from available data is that all the tests performed in validation that involve temperature measurement (with TC: kaye, expertlog …) are totally far from the measure i get from the process probes, more randomize, less stable …
so i’m asking my self which is the convenience of perform a validation measuring values that are so far from the ones that normally control the process during production.
is that the same in your companies? do you get reliable data from validation tc measurement of the system?

just two basic example, to let you better understand the problem:

1) tank sterilization:
if i watch to the data i can see:
- in the same run: validation probes measure values with difference of 50% from position to another
- same probes, different run (time), validation probes measure different value, from 5% to 100% difference between two cycle
- process probes VS validation probes: if i look at a validation probe that is close to a process probe, i get difference (in F0) of 50-80%

2) Autoclaves (canister sterilization):
i have evidence that autoclaves is performing well (with thermomapping)
i’m actually monitoring 11 canister over 17, by put a TC and BI inside the loaded canister.
since the probability of failure belong all to the canister assembly, why i should monitor just 11 canister and not all?
moreover, where is the meaning to monitor the canister load when the problem own just to human action (canister assembling) ?

finally, my observation suggests that validation are extremely innacurate because the human hand that put the probes, load the canister, prepare the bi and so on…

where i’m wrong?

thanks and sorry for my bad english!


Dear Marco,

In my experience, when you validate a process that has to do with temperatures, the validation gives you the parameters for normal process, so you should not be having many difference between validation and normal process.
When validating autoclaves and ovens for example you have to validate the biggest load and the lowest load so that all the materials between these two quantities will be surely sterilized and included in the validation.
Also materials have to be placed inside the equipment in the same way each time, this can be reached by training operators but also having load graphics that can be consulted whenever the instrument is going to be used.
You have to be very careful when placing thermocouples on the load as sometimes they can be touching for example a stainless steel material and so temperature will be much higher than we would like to. In this case you would need to invalidate your run and begin again.
The idea of monitoring just one part of the materials inside the chamber is to avoid waisting BI for example with all materials, assuming that you have already ran a heat performance chamber (alone, with no material) which will provide you assurance that heat transfer is the same inside the equipment. You select at random the places to be monitored, providing a rational of choosing each one of them, don’t forget to include the coolest point of your equipment which normally is close to the drainage.
I hope this helps you a little bit.

thanks for your answer!

the main issue i want to expose is on the actual convenience to perform period revalidation of equipment like autoclaves/oven/tank SIP where i have already performed the OQ,IQ,PQ sucessfully and i can monitor (better and with more reliability) the functionality and efficacy of sterilization by the process probes.

what’s you opinion on this point?

Ciao Marco,

If you are still seeing some variability in your data as you revalidate you may also want to consider how you are inserting your thermocouples and BIs into the system if you are not using a wireless system. This can have a great effect on how the system behaves compared to normal production runs.

For tanks/other SIP work you might want to look into purchasing Smart gaskets and clamps - sanitary gasket/clamp systems which allow you to insert wires and obtain a good seal on the system prior to initiation of a run. Once the wires are in, you can attach BIs to the wires using zip ties as long as they are far enough from the thermocouple tips that they won’t interfere with data collection.

For autoclaves there are several pass through adaptors that allow you to feed the thermocouples into the pressure vessel and obtain an integral seal.

Also you have to consider only the data that you obtain during the “exsposure phase” or “sterile hold time” of the cycle when assessing the point to point data. It is normal - especially in large tanks - for the temperature to have a large point to point variance during the “air evacuation pulses” or “vacuum pulses” phase of the cycle. (You will see the temperature drop and rise during this time but it will never be a uniform process in something with a more complex shape such as a tank.) The temperatures will become more uniform once steam is charged into the system.

In regards to Fo analysis, you should analyze during the exposure phase. Depending on the system, it is possible to have a large variance from point to point. You must make sure that the “slowest to heat spot” or lowest Fo meets your minimum requirement (usually 15 minutes) for the exposure time.

Hopefully some of this helps you.


Periodic revalidation has become something of an issue of late. My personal belief is that revalidation should be based on change (relocation, major maintenance, system upgrades, etc) or on trend observations (process temperature monitoring being a good example); and not performed periodically for no obvious reason. There are obvious exceptions to this rule, such as in HVAC where specific periods are defined in the various standards. In my opinion, most calibration procedures are based on validation/pharmacopoeial procedures in any case, so every scheduled calibration performed in that fashion tends to be an OQ. This must be documented as such in some or other review document (Refer PICs GMP, Annex 15 on Qualification).

Processes are bit different in that all contributing variables must be tightly controlled. Any evidence that indicates a process moving out of control is evidence of a problem within the process that must be addressed. Once found, this would either have to be engineered out (change control - assessed by team of experts, requalify affected kit if required, then revalidate affected process if required) or managed out (change control - update sop’s, note in QMS). Therefore, in my opinion, for processes one should always consider periodic verifications based on the original validations. Should any of these indicate a problem, revalidation of the process should be undertaken…