Temperature Mapping Procedure For Stability Chambers With 9 15 27 Probe Points

A chamber holds its setpoint at the point its control sensor reads. Nowhere is it guaranteed to be exactly that everywhere else. Air moves through a chamber in currents. A spot near the fan runs to one side of the setpoint, a spot in a still corner to the other. The shelf by the door takes the room’s air each time it opens. The top of a tall chamber can sit a degree above the bottom. None of this shows on the single control reading, which holds steady while the corners drift around it. A study does not take place at the control sensor. It takes place in the trays of product spread through the working space. Each tray is held at the temperature of its own spot, not the one the controller reports. If a corner runs warm and a sample sits in it, that sample ages faster than the study believes. The data is then wrong for a reason no one can see. Mapping exists because the control point is a promise about one place. A study needs a promise about all of them. The gap between the two promises is not large in a good chamber, perhaps a degree or two from the warmest spot to the coldest. A stability study is sensitive to exactly that. It runs for months or years and measures a slow change in a product. A degree of unseen warmth in a corner, sustained across that time, is enough to age a sample faster than the data accounts for. Mapping closes the gap by measuring it. It does not make the chamber more even. It shows how even the chamber already is, and where it is least so, which is the knowledge a study needs to trust its own numbers.

□□□How many sensors, and where they go

The number of sensors a mapping uses follows the size of the chamber. The pattern follows its shape. A small chamber, a reach-in or an incubator up to about two cubic metres, is mapped with nine sensors, set out as a three-by-three grid that captures the spread across the space. A larger chamber, up to about twenty cubic metres, takes fifteen. A walk-in room is mapped with twenty-seven, a three-by-three-by-three cube of sensors that reaches into the volume in all three directions. To each of these a maker adds one more sensor at the chamber’s own control and monitoring probe, so the map can be tied back to the reading the chamber reports day to day. The count is not arbitrary. It comes from long-standing guidance, the temperature-mapping annex of the World Health Organization’s Technical Report Series 961. It scales with volume because a bigger space has more room for a gradient to develop unseen. Where the sensors go matters no less than how many. The points reach the extremes of the space, its corners, its centre and its faces, so no part of the volume is left unsampled. A nine-point grid lays a three-by-three pattern across one representative plane. The larger sets reach through the depth as well, the fifteen placing sensors at the eight corners of the box, its geometric centre and the centre of each of its six faces. The sensors sit at least ten centimetres clear of the walls and the floor, in the air the product will occupy, away from a surface that runs to its own temperature. To this regular grid a maker adds sensors at the places risk is known to gather, beside the door that opens to the room, near the inlet where conditioned air enters, by the fan. These are the spots a uniform grid might step over and a real product might sit in. A sensor at the control probe closes the loop, letting the map show how the controller’s one reading relates to the spread around it. The result is a cloud of calibrated points filling the space, enough of them, and placed well enough, that no warm or cold pocket can hide between them. A mapping with too few sensors, or sensors in the easy places, proves only that the chamber is even where it was easy to check.

The control sensor proves a point. A study has to trust the room.

□□□Why one sensor cannot speak for the chamber

The reason a grid is needed at all is that a chamber is never the same temperature throughout. Conditioned air enters at one place and returns at another. Between the two it warms or cools as it crosses the load. The result is a field of small differences, a spread of a degree or two between the warmest and the coolest spot even in a well-built chamber. The control sensor sits in one fixed place and reads only that place. It can hold the setpoint to a tenth of a degree and tell a maker nothing about the corner two metres away that runs a degree warm. A study fills the chamber with product. The product sits in the spread, away from the control point. A tray in a warm corner is held warmer than the controller reports. Over the months of a study that difference ages it faster than the data assumes. The single reading describes only the spot it sits in. Mapping replaces that silence with a measured picture of the field, so the spread becomes a measured fact. The differences have causes a mapping can predict. Warm air rises, so the top of a tall chamber often runs above the bottom. The door is a hole into the room, so the shelf beside it swings each time it opens and never holds the centre’s steadiness. The air inlet blows conditioned air that has not yet mixed, so the spot in front of it can run to the far side of the setpoint. A mapping does not have to guess where the trouble is. It places sensors where these causes point, and lets the data confirm or deny each one. What it cannot do is assume the chamber is even and check only the middle, because the middle is the one place almost certain to be fine.

□□□The sensors that do the reading

The sensors that make a map are not the chamber’s own. They are an independent set, brought in and placed on the grid, each calibrated against a reference that traces to a national standard before the mapping and checked again after, so the map rests on readings of known accuracy. They are usually small data loggers or fine thermocouples, chosen to read quickly and to sit in the air without disturbing it. A sensor with a slow response would smear the swings the map is meant to catch, so the type and its time constant are set in the protocol. Where a sensor goes in or near the product, it is sometimes set in a buffer that mimics the thermal lag of the product itself, so the map reads what a vial of liquid feels, a slower swing than the bare air shows. The sensors log together, on one clock, so a warm moment at one corner can be lined up against the same moment everywhere else. The reason for bringing in a separate, calibrated set is that the map must not depend on the chamber it is judging, since a chamber cannot serve as its own proof of evenness.

□□□Empty first, then loaded

A chamber is mapped twice, empty and loaded. The empty map measures the chamber on its own, the spread its air handling can hold with nothing in the way. It is the cleaner picture. It shows whether the chamber is capable. The loaded map measures what a study will see. Product fills the space and blocks the air that carries the heat, so a chamber that is even when empty can grow warm and cold pockets once it is packed. The loaded map is made with the chamber filled the way it will run, at the density and the layout a real study uses, because a map made on an empty chamber says nothing about a full one. Where a study will run the chamber near its limit, packed tight, the loaded map is made that way, at the worst case the protocol allows. A chamber passes mapping when both maps, empty and loaded, hold their spread inside the tolerance. Between them, the empty map proves the machine and the loaded map proves the study. How the chamber is loaded for that second map is itself part of the protocol. Product is packed at the density a study will use, in the same containers and on the same shelves, because the air finds its way around real boxes differently than around an even, idealised load. A chamber loaded loosely maps better than it will run. A map made that way flatters the chamber. The worst case the protocol allows, the tightest packing and the fullest shelves, is the one that has to pass, because a study run at that load is the one the map has to stand behind. A chamber that holds its spread loaded to the worst case holds it for everything lighter.

□□□A full day, and a few hard tests

A map is not a snapshot. The sensors log together for a stretch of time, at least a full day once the chamber has settled, so the map catches the slow swings a chamber makes as it cycles and as the room around it changes through a day and a night. A reading taken once tells nothing about how a spot drifts over hours. The mapping period also carries the hard tests. The door is opened and held, the way it will be during sampling, to see how far the chamber strays and how fast it recovers. The power is cut and restored, to measure how the space rides through an outage and whether any corner climbs out of tolerance before the chamber catches up. These are not failures to keep out of the map. They are the events the study will meet, captured while the sensors are in place to record them. A map that runs only through a calm, closed, fully powered day proves the chamber under the one condition it will rarely be in.

□□□The worst spot decides

When the mapping is done, the data is read for the spread. The warmest sensor and the coldest sensor, at every moment and across the run, mark the range the chamber holds. The difference between them is the gradient. It has to sit inside the tolerance the study allows. The map names the hot spot and the cold spot by position, the actual corners where the extremes are found. That naming is the point of the exercise. The chamber’s own control and monitoring sensors are then placed at the worst spot the map found, away from the comfortable middle. A monitoring sensor at the hottest corner reports the worst a product sees, so a study judged on it is judged on its hardest case. Put the monitoring sensor in the easy centre and a warm corner goes unwatched, its product over-aged and unreported. The map also draws the line around the qualified space. Where a corner runs outside the tolerance and cannot be fixed, it is excluded and marked as space no product may occupy. The study is then held to the volume the map proved even. The mapping turns a chamber from a box assumed uniform into a space with a measured spread and a known worst point.

□□□Where mapping sits in qualification

Mapping is one step inside the larger work of qualifying a chamber. The empty map belongs to the chamber’s operational qualification, the stage that proves it runs to specification. The loaded map belongs to its performance qualification, the stage that proves it holds the condition in real use. How that qualification is structured, and the records it produces, is the subject of the compliance work. What mapping contributes to it is the evidence that the space is even, the one thing a stack of qualification documents cannot assert without a grid of sensors behind it. A chamber can be installed correctly and run to its setpoint and still fail its mapping, if its air handling leaves a corner warm. Mapping is the step that replaces an assumption of evenness with a measurement of it. The two maps fall in two different stages for a reason. Operational qualification asks whether the chamber works as built. The empty map answers it on the chamber’s own terms. Performance qualification asks whether it works for the job. The loaded map answers it under the load a study brings. A chamber that passes the first and fails the second is a sound machine the study will still defeat. The order of the two maps catches that before any product is committed to it. Mapping is the measured part of both stages, the part no amount of paperwork can stand in for.

□□□How often to map again

A map is not made once and trusted forever. A chamber is mapped again on a schedule, commonly every two or three years, and again whenever something changes that could move its air. A new shelf layout, a repaired fan, a relocated chamber, a change to the load, any of these can shift where the hot and cold spots fall. A map made before the change no longer describes the chamber after it. The requalification interval comes from the same guidance that sets the method. It rests on a plain idea, that a chamber drifts as it ages and its proof has to be refreshed before the drift can mask a fault. A chamber running on a five-year-old map, through two repairs and a move, is trusted on a picture that may no longer be true. Some changes are obvious triggers, a move to a new room, a replaced compressor or fan, a different shelf arrangement. Others are less obvious, the slow fouling of a coil, the wear of a door seal, a control reset to a new offset. The scheduled re-map exists to catch the hidden ones, the drift no single event announces. Between scheduled maps, the chamber’s continuous monitoring at the worst spot stands watch. A trend in that record can call for a re-map before the interval is up. The map and the monitor work together, a thorough look taken now and then backed by a lighter check that never stops.

□□□Mapping the humidity too

A stability chamber holds humidity together with temperature, so the humidity is mapped too. Relative humidity varies through a chamber the way temperature does. Because it depends on temperature, a warm corner is also a drier one. The humidity sensors go in alongside the temperature sensors, on the same grid. The map reports both spreads together. A chamber that holds its temperature evenly can still carry a humidity gradient, near a port where dry air leaks in or above a water tray that wets the air unevenly. For a stability study, where the humidity is part of the condition the product is held at, the humidity map matters no less than the temperature one. Both have to hold their spread inside tolerance before the chamber is trusted with a study. The link between the two makes the humidity map more than a duplicate of the temperature one. Because a warmer spot holds a lower relative humidity at the same water content, the chamber’s hot corner and its dry corner are often the same place. The two maps reinforce each other in naming the worst spot. Where they disagree, near a leak or a tray, the map has found something a single measurement would miss. A stability condition is a pair, a temperature and a humidity held together. A chamber is only as qualified as the weaker of its two maps.

□□□The map becomes a report

The mapping ends as a document. The report records where every sensor was placed, with a diagram of the grid. It carries the raw data from each one across the mapping period. From that data it draws the numbers that matter, the warmest and coldest readings, the gradient between them, how each spot moved through the day, and how the chamber behaved through the door opening and the power cut. It states the conclusion plainly, the spread the chamber held, the position of the worst spot, the volume that qualifies as even, and where the control and monitoring sensors should sit. This report is what an auditor reads years later to see that the chamber was proven even before the study trusted it. A study’s data rests on it, because the temperature each sample was held at is only as certain as the map that proved the sample’s corner matched the controller’s reading. The diagram in the report is not a decoration. An auditor checking a study years later uses it to see exactly where each sensor sat, so the spread can be tied to real positions in a real chamber, each value pinned to a place. The report also states what it could not prove, any corner excluded from the qualified space, any caveat on the load it was made under. A map that omits its weak spots is worse than none, because it offers a confidence it has not measured. The honest report draws the boundary of what the chamber was shown to do, and marks clearly what lies outside it.

□□□The same method, at any size

The same method maps anything that has to hold a temperature, scaled to its size. A cold store or a warehouse is mapped with a grid spread through its racks, with more sensors for its greater volume and its greater chance of a gradient between the door and the back wall. A refrigerated truck is mapped along its length, where the air from a single front unit has to reach the rear doors. A pharmacy refrigerator is mapped with the nine-point grid of a small chamber. A chamber that proves a medical device against its shelf life is mapped to the same rules as one that proves a drug, under whatever standard the device follows. The principle holds at every scale, that the point a system controls from is not the entire space it has to keep. Only a grid of sensors can show the difference.

□□□What the map gives a study

What mapping gives a study is confidence that the number on the controller means what it should everywhere a product sits. It turns the chamber from a box trusted on one reading into a space with a measured temperature and humidity spread, a named worst spot, and a control sensor placed to watch it. A study run in a mapped chamber stands on proof that every tray of product was held at a temperature the data accounts for, where a study in an unmapped one rests on the hope that the corners matched the middle. For a study that runs for years on the evidence the chamber produces, that hope is not enough. Mapping is the difference between a chamber assumed even and a chamber shown to be.

Envsin stability and environmental test chambers, mapped and qualified for even temperature and humidity across the full working space.