blog

Video Q&A: Guidance on warehouse mapping sensor placement

Sensor placement in warehouse mapping study
Life Science

See below for an edited transcript of this video. Here is another brief video question and answer session from our warehouse mapping webinar. In this webinar we suggest a cost-effective way of mapping your warehouses continuously. Senior Regulatory Expert Paul Daniel answers questions we did not have time for during the webinar. Enjoy and feel free to leave further comments below.

[00:00:06] Good day and welcome to another Q&A video with Vaisala Senior Regulatory Expert Paul Daniel. Today we're going to cover some of the questions that we received from our webinar: “Continuous Mapping: Better Data, Better Compliance.”

[00:00:30] Our focus in that webinar was warehouse mapping. We were very grateful to have over two hundred people in that webinar, but we didn't get a chance to answer all the questions. So we're doing it today. *See an earlier Q&A video from that webinar.

[00:00:54] Question one: “Why stack the sensors vertically, on top of each other? And can I use stacks of two if the room is not very tall?”

[00:01:04] Paul: In the webinar I recommended starting with stacks of three. That's a stack of sensors with one low, one middle, and one high, and then distributed through the warehouse. This attendee is asking why we should stack sensors vertically. Also, can we use shorter stack stacks [in mapping studies]? This is a perfect question for a thought experiment; we only have three choices when it comes to sensor placement. Option one: distribute sensors randomly throughout the warehouse. This sounds like fun to me, but I guarantee your quality department will hate it.

Option two: we can place sensors only at the places that we think are going to provide interesting data. Again, this sounds like fun, but that's not going to be good enough. Finally, option three: we distribute the sensors evenly and geometrically throughout the warehouse covering the entire space.

As fun as it sounds, nobody does random [sensor placement in] mapping. We map with evenly distributed sensors. That is option three and occasionally, we put extra sensors in places of interest. If you distribute sensors evenly in space, geometrically distributed, you will inevitably end up with stacks because rooms tend to be square. So with sensor in an upper corner and one in a lower corner you automatically have to stack sensors.

[00:02:23] Let's look at the guidance. USP Chapter 1079 Good Distribution Practices states that sensors should be placed in three planes in three dimensions.

A temperature mapping study should be designed to assess temperature uniformity and stability over time and cross a three-dimensional space. Completing a three-dimensional temperature profile should be achieved by measuring points at not less than three dimensional planes in each direction/axis—top-to-bottom, left-to-right, front-to-back, where product will be present.

That's a confusing way to say stacks of three. The only way to meet these criteria is to use three stacks of sensors, one on top of the other. If you don't have a stack of at least three sensors, you can't meet that requirement.

Another guidance source for this is the WTO, the World Health Organization. "Temperature mapping of storage areas Technical supplement to WHO Technical Report Series, No. 961, 2011" The WTO also recommends three sensors in a stack if your ceiling is less than 3.6 meters. They go up to five sensors in the stack when your space is 6m or more. I think five sensors in a stack is sort of overkill. I've never seen anyone use that many sensors in a controlled space. For a warehouse I recommend stacks of three in general. To our question asker, if the ceiling is low, say less than 8 feet, or 2.5 m, then place sensors in stacks of two.

[00:03:28] Next question: “What lateral distance between sets of stacks is appropriate?” You said 20 m maximum lateral distance. Where does this number come from?”

[00:03:52] Paul: In the webinar, I recommended a 20 meter distance laterally (sideways) between those stacks of three. I have not found a guidance or regulation yet that gives us a specific number for placing sensors like this. It is just not there. That 20-meter limit is just a rule of thumb that I came up with myself and it is based entirely on my own experiences and warehouse mapping. In practice it works. It is a distance that gives reasonable confidence to most people, including auditors. But, it doesn't hit that point of overkill where we begin to feel like there are too many sensors in space.

Remember that 20 meters [lateral distance] is for ambient warehouses. We sometimes map more extreme environments, say large cold rooms and large freezer rooms. For those, I would decrease the distance to 8 meters. It is a bit of a sliding scale determined by the size of the space, the temperature dynamics, the product stored, and the data that we get from mapping studies. But for a warehouse storing ambient products, I found that 20 meters is a good maximum lateral distance between stacks of mapping sensors. Given the huge variety of floor plans you might run into, even that parameter should be adapted to the particular situation at hand.

[00:05:08] Question three: “How many mapping sensors for 3 x 3 meter room?” Is there any guideline we can use as a reference for?”

[00:05:33] Paul: Yes, we actually have a good guidance for that. The threshold is not so much the floor plan, but the volume and the volume threshold we need to be worried about is 20 m3 of volume so that 3 x 3 meter room would probably be right there at a limit of 20 m3. Any rectangular place that is less than 20 m3 volume is going to need 16 sensors, with one sensor in the middle and a sensor in each corner. Since there are eight corners, we have a total of nine. Then we would put one sensor in the middle of each side. A room has six sides and that takes six sensors. That makes 15 so far. The last sensor, number 16, we would place adjacent to the control center. The outline for this kind of sensor placement is detailed in the ISPE’s “Good Practice Guide for controlled temperature chambers”.

For a webinar on this topic, See: “Mapping Made Easy

Comment

Ana Yagudaev

Oct 1, 2020
Hi Paul
Is it practical to use the sacks of three sensors spaced 20 m apart laterally in the big warehouses?
We have a warehouse 350' X 250' x 25 with 40 double-sided racks.
We used 75 sensors to map it, alternating the staking every other rack. One rack, we used double staking: low and top of each end of the rack, and next rack we position the sensor in the geometrical middle of the rack. Also, we placed sensors on the racks affected by HVAC discharge. Do you consider it is acceptable?
How many sensors will you think we will need to perform three-stack mapping?

Paul Daniel

Oct 5, 2020
Hi Ana. I think your approach is practical and acceptable. It aligns with my advice which is based on guidance and personal experience. I recommend 3-stacks as only place to start, as it will satisfy many, but I personally think this is overkill. This is why I present the strategy to “remove alternating sensors” based off the ISPE diagrams. The end result of this practice, is to have sensor rows that alternate from double stacks (hi-low) to single sensors (middle) all the way down each row. If you want to send me a diagram of your facility, I could give you a more specific answer.

Gerry Gutierrez

Oct 1, 2020
Paul, the updated "Temperature mapping of storage areas Technical supplement to WHO Technical Report Series, No. 961, 2015" does have the 20 to 30 meters between sensors for large warehouses. Great video on Q&A from previous webinar.

Best Regards

Gerry

Paul Daniel

Oct 5, 2020
Thanks, Gerry! This is a great reminder to read documents carefully. The WHO document, in the main section, recommends locating mapping sensors “…every 5-10 metres” and thi sis followed by a footnote. In this case the footnote is EXTREMELY important, as it says at the bottom of that page “ “In very large facilities, this can be up to 20 or 30 metres.” For the examples in the webinar, we were dealing with a 4000 square meter facility, which I would say qualifies as a very large facility. Now we have a documented source for the 20 meter recommendation. And you may note, they even allow up to 30 meters!

Marius

Oct 6, 2020
Paul, According to a "large warehouse" situation. From which size does the warehouse becomes "large"?

Paul Daniel

Oct 17, 2020
I don’t have a well-defined definition for large. There is no regulation or guidance that will clearly define the correlation between the size of a storage area, and the resulting sensor densities required for mapping. The ISPE did it for small chambers, but their guidance doesn’t cover spaces over 20 cubic meters total volume, so that doesn’t help us with warehouses.

But just for fun, let’s unpack this concept. Maybe, we should drop the word “warehouse” here, because that already implies a large size space. Let’s just call them “storage areas”, which is what WHO calls them in the title of Supplement 8. Technically, we could call a warehouse a large storage area that is dedicated to managing incoming and outgoing goods. If the WHO basic guidance is telling us to place our sensors stacks 5-10 meters apart, then maybe we can guess the starting size for applying this guidance? Our smallest size area, with three stacks of sensors down each side, at 5m apart, would give us a space that was 10m on a side, or 100 square meters. Maybe we could call this a SMALL storage area?

Let’s say we have a space that is big enough to move our sensors stack to 10 m apart, and we have 4 stacks down each side… Now we have a space that is 30m on each side, or 900 square meters. Maybe we could call this MEDIUM storage area?

As I hope you can tell, I am just trying to extrapolate from the little information we have. In other words, I am making it up with some informed guesses. There are no exact numbers in any of this, as we don't get this information detailed for us in the guidance. I think the best we can do is apply a standard of reasonable confidence. Reason is logic. So all mappers should ask themselves, at what density of sensors will it be logical for me to have confidence that I am collecting data with enough resolution to understand the temperature dynamics of the warehouse space?

I am sort of saying that it might not be worth it to worry about whether your warehouse is large or small. Just ask yourself if sensors placed 10m apart will accurately represent the space being mapped? Are you confident you won’t miss a hot spot? If your answer is yes, move the sensors to 15m apart and ask yourself the same question. Repeat until you find that point of logical confidence.

Also, there are some logistics of warehouse mapping to be considered. Mapping sensors need to be attached to something, and usually it is racks and support columns. Your choices of sensor location will often simply be predicated by the location of the existing racks and columns. So you simply pick a density of sensors that aligns with the available areas (racks and columns) to place sensors.

And don’t worry about getting it wrong. Auditors would rather see a mapping that can be improved, than see no mapping at all.

Add new comment