Thank you to everyone who attended our training seminars in Woburn, MA last week.
Here's some feedback from an attendee:
This is the second Vaisala seminar I've attended and once again I was impressed with the content and professionalism. The team here at our calibration lab continues to recommend Vaisala to our customers. I look forward to more seminars in the future!
Best Regards,Tom Masterson
Essco Calibration Laboratory
Thank you for being so interactive and engaged! If you missed Part I - read it here.
Modern Rules for an Old Practice: Sensor Placement for GxP Environments
I put my probes at the end of the space I intend to use with my process. If you
intend to use it, you should include it in your mapping. For a refrigerator or incubator that depends on moving air to control temperature, you should leave at least 10% of each lateral dimension empty and free of product.
For a 1 meter wide refrigerator, this would mean to keep all products at least 10 cm from the walls. This is also where you would place the mapping sensors. I also recommend placing your mapping sensors in air only, not glycol or aluminum blocks. Air presents the worst case challenge. Don't worry about buffering the probes until you have data that says you might need to do this.☺
The use of a setpoint implies that the space is being controlled for a given process, and therefore mapping is necessary to demonstrate that the space is actually performing as expected by setting the setpoint.
So in some cases, if an environment does not have a setpoint, then it has no control parameters and may not need to be mapped. One way to think about temperature uniformity is uniformity around a setpoint, such as 5 +/-3C. It is just as valid to think that you are showing temperature uniformity to 2 to 8C. It is the same at the end of the day regardless of the setpoint. If there is a range of set points, my preliminary answer is to map once at each setpoint you intend to use.☺
The webinar addresses this, but in the meantime – the long answer is to map it and see what it looks like in terms of temperature distribution. The quick answer – put sensors in "stacks of three": one near the floor, one as high as you store stuff, and one in between at a medium height. Place these stacks of three throughout your warehouse, but never more than 20m apart. Then Map.☺
My best advice is not to map in sections. In my experience that this creates more problems than it solves. Rent the equipment you need to map it all at once. If you must map in sections the only "recommended" way to do it is make sure your sections correspond to the control zones of your HVAC system.
Doing it in sections is better than not mapping at all, but an auditor's response will be to tell you to repeat it as a single unit.☺
For mapping, most sensors that are available for the purpose have low mass and a fast response time that is suitable for mapping air temperatures. If you have a specific application where you think a specialized sensor is required, please reach out to me individually.☺
Minimum 9 sensors for 2 cubic meters or less, 15 sensors for spaces between 2 and 20 cubic meters. Be prepared to add more sensors to deal with other variables such as shelves, vents, etc. For spaces larger than 20 cubic meters, you really need to analyze your space, but the number of sensors will increase slowly relative to the volume of the space.☺
This webinar is designed for mapping of storage and production spaces. It is not intended for ovens, autoclaves, sterilizers, or lyophilizers. However, some of the concepts are similar.☺
Yes. The refrigerated box is colder, and because of the larger differential between the outside and inside temperatures, it should have a higher density of sensors than might be used for a room temperature area. In cold rooms, I like to have my sensors no more than 8 meters apart, where I will do 20m in room temperature spaces.☺
Yes, almost. However, always remember that every space is different, every process is different, and every item being stored is different. So a distribution that works for one process may be inappropriate for another process. But you can use this as a rule of thumb: Minimum 9 sensors for 2 cubic meters or less, 15 sensors for spaces between 2 and 20 cubic meters.
Be prepared to add more sensors to deal with other variables such as shelves, vents, etc. For spaces larger than 20 cubic meters, you really need to analyze your space, but the number of sensors will increase slowly relative to the volume of the space.☺
That is the thought. However, placing a probe in a buffer hides data. It is better to monitor air temperature to prove your equipment can do the job, and then there are fewer questions.
Until you do a study in air you won't have the data to show that buffering is required. Furthermore, you will also have to evaluate that the poor performance of air temperatures in your environment is not the result of design or maintenance issues.☺
Knowing what is going on in your storage process is a regulatory requirement for every region. Some regions will specify that they want to know what is going on in the worst case locations (hot spot/cold spot). I would treat it as a regulatory requirement if the EMA has already asked you to do this.
This can be interpreted as saying to put a sensor in this location, or it can be interpreted as saying make sure you know what is going on in this "worst-case" location by monitoring somewhere else. The way you guarantee that the worst-case location doesn't move is by practicing good change control and remapping regularly.☺
What about temperature mapping in automated freezers (where the carousel or gantries move)?
The moving parts obviously add a level of complexity. I would look at wireless or battery-powered loggers to allow the parts to move, and I would develop a regular schedule of movement of your carousels that mimics a production situation so you can evaluate the impact of the moving parts. Sounds like fun!☺
This is a big question. But basically, the standard is not in the distribution itself, but in the density of the distribution. You must, of course, change your distribution based on the shape and factors of the space being mapped. If there are shelves they must be considered. However, what kind of shelves matter as well. If the shelves are wire, then I would ignore them.
If the shelves are solid, I would make sure that they can't be easily moved by the user before taking them into account in my mapping. If they are fixed solid shelves, you will likely have to use more sensors, treating each shelf as its own chamber.
However, -80C, despite their fixed shelves, have such simple conduction –based cooling systems, and stay closed so much of the time, that you may find this level of complexity unnecessary. Play around and map a -80 with a whole bunch of sensors and you will see there isn't much variability in those environments. And yes, put a mapping sensor near the controlling sensor – it makes auditors happy even though it doesn't always seem useful.☺
I would recommend that you use one set of devices to control your warehouse and an entirely different set of devices to monitor your warehouse. Your control devices should be placed by HVAC engineers to optimize your system so that it cools and heats efficiently and is properly balanced to the expected loads.
Then map your warehouse and place your monitoring sensors based on the results of your mapping study. If you just place control sensors and use them for monitoring you will only see data that says you are in control – meanwhile, your product will be too cold or too hot.☺
I refer to them, but that isn't the purpose of the webinar. This webinar is to help you learn where to place sensors. Basically, the regulators and auditors want to see that you have mapped, ideally twice during each of the seasonal extremes. They will want to see a nice density of mapping sensors, presented in an executed protocol and report. But the regulations will just tell you to "validate your HVAC system" and "map your warehouse." You won't get what you need to know and do from the regulations.☺
Paul Daniel has worked in the GMP-regulated industries for over 20 years helping manufacturers apply good manufacturing practices in a wide range of qualification projects. His specialties include mapping, monitoring, and computerized systems. At Vaisala, Paul oversees and guides the validation program for the Vaisala viewLinc environmental monitoring system. He serves as a customer advocate to ensure the viewLinc environmental monitoring system matches the demanding requirements of life science and regulated applications. Paul is a graduate of University of California, Berkeley, with a bachelor's degree in biology.