Lyophilizer Shelf Temperature Mapping: Part III: Data Analysis
Shelf temperature mapping is a critical component of lyophilizer qualification. The previous articles in this series explored the methods used to record the shelf surface temperatures (Part I) and designing an appropriate study (Part II). Interpreting a shelf temperature mapping study’s result is as critical to assessing uniformity as the temperature measurement, the study design, and the execution.
Interpreting the results begins with proper data analysis. Acceptance criteria should be based on scientifically sound principles that guide data analysis, interpreting the results and drawing justifiable conclusions.
Data analysis considerations include data collection rate, analysis period, and approach to analyzing the results by location across all the shelves. The data collection rate directs the frequency of recording the temperatures for the different locations for the temperature probes on the shelf. The study duration must be long enough for the shelf to reach the necessary temperature. It must also include time for the shelf temperature to stabilize, and for analyzing the data collected after the shelf temperature stabilizes.
Once the collection rate, stabilization time, and data analysis duration have been defined, temperature data at each of the locations can be analyzed. This evaluation encompasses the average, median, minimum, and maximum temperatures at each shelf location over the data analysis interval. The results of the shelf surface temperature analysis by location are used to assess overall temperature uniformity across all shelves.
Data Collection and Analysis
Before analyzing and interpreting data from shelf temperature mapping studies, a number of data collection parameters should be defined. The collection rate should be set to capture the normal periodic fluctuations of the shelf inlet temperature. The shelf inlet temperature should be reviewed at different collection rates to evaluate the appropriate interval for the test.
Figure 1A shows a comparison of the shelf inlet temperature at 1-minute, 2-minute, and 5-minute collection intervals. At the 5-minute collection rate, the inlet temperature’s fluctuation is lost. At a 2-minute collection rate, the periodic nature of the shelf inlet is also somewhat smoothed, but most of the high and low temperatures are evident. For this lyophilizer, it appears the 1-minute collection rate best represents the fluctuations the shelf will be exposed to.
The data analysis period should be long enough to capture multiple periods of the shelf inlet temperature’s normal fluctuations. In Figure 1B, the differences between the shelf inlet temperature and the setpoint at –50°C, 0°C, and 50°C with a collection rate of 1 minute are displayed. Although the 50°C control is extremely tight, with no fluctuations in temperature, the 0°C period is about 6 minutes long and the –50°C period is about 15 to 17 minutes long. Given this variability in the control of the shelf inlet temperature at different setpoints, it is reasonable to recommend a data analysis period of 60 minutes.
After the collection rate and the data analysis period have been defined, the time to reach steady state should be determined to ensure the shelf mapping temperatures recorded are in dynamic equilibrium. The previous articles in this series identified a five-location-per-shelf mapping pattern, including each shelf’s four corners and center. This will be the default pattern for all monitoring locations and analysis explored in this article.
The data analysis period should not start until all dryer locations have been equilibrated and have stopped changing. This may be different for different locations in the dryer (see Figure 2). In this example, after the shelf inlet temperature has equilibrated, it required up to 120 minutes for all of the shelf’s locations to stabilize. In this case, it would be appropriate to define the first 120 minutes as the stabilization period. The stabilization period may be different at the different test temperatures. Therefore, the worst-case temperature should be used to establish the stabilization period. In this case, the setpoint of –50°C took the longest to stabilize.
Figure 1: A) Shelf inlet temperature with 1-minute, 2-minute, and 5-minute collection rates and B) shelf inlet variability at 50°C, 0°C, and –50°C with a data collection at 1-minute interval.
Figure 2: Shelf location temperatures for three hours post shelf inlet temperature steady state.
Analysis of Shelf Temperature Over Test Duration
The shelf temperature mapping’s results reflect a very narrow range of temperatures for the duration of the study. A presentation of the results when evaluating the minimum (MIN), maximum (MAX), and average (AVG) of the 20 locations monitored when averaged for the data analysis period (last 60 minutes out of a 180-minute hold period) shows that the greatest variation is 1.4°C at the setpoint of –50°C, ranging from –49.6° to –48.2°C (see Table 1 and Figure 3).
Table 1: Comparison of single timepoint with test duration in one run over four shelves with the average over the entire test duration.
| Shelf Number | 1 | 2 | 3 | 4 | All Shelves |
|---|---|---|---|---|---|
| –50°C | |||||
| AVG of 5 locations at 60 min | –49.2 | –49.2 | –49.2 | –48.9 | –49.1 |
| MIN of 5 locations at 60 min | –49.6 | –49.6 | –49.5 | –49.4 | –49.6 |
| MAX of 5 locations at 60 min | –48.7 | –48.6 | –48.8 | –48.3 | –48.3 |
| MAX–MIN at 60 min | 0.9 | 1.0 | 0.7 | 1.1 | 1.3 |
| Average over test duration | –49.2 | –49.2 | –49.2 | –48.9 | –49.1 |
| MIN over test duration | –49.6 | –49.6 | –49.5 | –49.6 | –49.6 |
| MAX over test duration | –48.2 | –48.3 | –48.7 | –48.2 | –48.2 |
| MAX–MIN over test duration | 1.4 | 1.3 | 0.8 | 1.4 | 1.4 |
| 0°C | |||||
| AVG of 5 locations at 60 min | –0.3 | –0.3 | –0.3 | –0.2 | –0.3 |
| MIN of 5 locations at 60 min | –0.5 | –0.5 | –0.6 | –0.6 | –0.6 |
| MAX of 5 locations at 60 min | 0.3 | 0.5 | 0.3 | 0.4 | 0.5 |
| MAX–MIN at 60 min | 0.8 | 1.0 | 0.9 | 1.0 | 1.1 |
| Average over test duration | –0.3 | –0.3 | –0.3 | –0.3 | –0.3 |
| MIN over test duration | –0.7 | –0.6 | –0.7 | –0.8 | –0.8 |
| MAX over test duration | 0.3 | 0.5 | 0.3 | 0.4 | 0.5 |
| MAX–MIN over test duration | 1.0 | 1.1 | 1.0 | 1.2 | 1.3 |
| 50°C | |||||
| AVG of 5 locations at 60 min | 48.1 | 48.2 | 48.2 | 48.1 | 48.1 |
| MIN of 5 locations at 60 min | 47.5 | 47.8 | 47.3 | 47.3 | 47.3 |
| MAX of 5 locations at 60 min | 48.7 | 49.0 | 48.8 | 48.7 | 49.0 |
| MAX–MIN at 60 min | 1.2 | 1.2 | 1.5 | 1.4 | 1.7 |
| Average over test duration | 48.0 | 48.2 | 48.2 | 48.1 | 48.1 |
| MIN over test duration | 47.3 | 47.7 | 46.9 | 47.2 | 46.9 |
| MAX over test duration | 48.7 | 49.0 | 48.8 | 48.7 | 49.0 |
| MAX–MIN over test duration | 1.4 | 1.3 | 1.9 | 1.5 | 2.1 |
In general, before drawing any meaningful conclusions, shelf temperature mapping analysis should be done retrospectively to ensure the stabilization period was appropriate. Depending on the collection rate and the analysis period, the data analysis can become quite cumbersome. A quick check of the data can be completed at a certain time within the data analysis period (for example, at the end of the 60-minute hold) and may provide comparable results (see Table 1). Although this cursory analysis may provide some useful insight into how the shelf temperature mapping is progressing, it is recommended to complete the full analysis of the entire defined test duration after running the test.
Figure 3: Measured shelf temperatures over the 60-minute collection period as a function of shelf number at A) –50°C, B) 0°C, and C) 50°C.
Evaluation of Location-Specific Shelf Temperature Over Data Analysis Period
Once the collection rate, stabilization time, and data analysis duration have been defined, the analysis of the temperature data at each location can be completed. There are multiple ways to analyze the shelf temperature mapping study data for each shelf’s five locations during the study. There are also multiple ways to define the acceptance criteria. In general, the factors to consider include evaluating the minimum (MIN), maximum (MAX), median (MED), and average (AVG) temperature values. It is important to recognize that the average and median are of limited use for reflecting any extremes in the results. However, it is useful to compare them with each other. When the calculated average and median are close to each other, the minimum and maximum temperatures are generally not outliers or exerting undue influence on the average temperature. If the median and the average differ widely, the minimum and maximum temperatures should be investigated.
It may be that data analysis duration or stabilization period were inappropriate, and the location of interest had not yet fully stabilized and was still changing when data analysis started. This analysis should be completed for each location individually for the full 60-minute data analysis duration (see Table 2). In this case, the difference between the median and the average is zero for every location at each temperature analyzed and the largest difference between the maximum and the minimum temperature is 0.3°C. The tightness of the data at each location over 60 minutes signals that the stabilization period and the data analysis duration are appropriate.
Table 2: Shelf temperature by location at –50°C, 0°C, and 50°C over the 60-minute test duration. (Number = shelf number, where 1 = top, 4 = bottom, LF= left front, RF = right front, C = center, LR = left rear, and RR = right rear. Red highlighted locations are the maximum temperatures. Blue highlighted locations are the minimum temperatures at each setpoint.)
| –50°C Setpoint | 0°C Setpoint | 50°C Setpoint | ||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Avg | Med | Max | Min | Max–min | Avg–med | Avg | Med | Max | Min | Max–min | Avg–med | Avg | Med | Max | Min | Max–min | Avg–med | |
| 1LF | –49.5 | –49.5 | –49.4 | –49.5 | 0.2 | 0.0 | –0.5 | –0.5 | –0.3 | –0.6 | 0.3 | 0.0 | 48.0 | 48.0 | 48.1 | 48.0 | –0.1 | 0.0 |
| 1RF | –49.3 | –49.3 | –49.2 | –49.3 | 0.1 | 0.0 | –0.3 | –0.3 | –0.2 | –0.3 | 0.1 | 0.0 | 48.2 | 48.2 | 48.2 | 48.1 | –0.1 | 0.0 |
| 1C | –49.5 | –49.5 | –49.5 | –49.6 | 0.1 | 0.0 | –0.5 | –0.5 | –0.4 | –0.6 | 0.1 | 0.0 | 48.1 | 48.1 | 48.1 | 48.1 | –0.1 | 0.0 |
| 1LR | –49.4 | –49.5 | –49.3 | –49.5 | 0.2 | 0.0 | –0.5 | –0.5 | –0.4 | –0.6 | 0.2 | 0.0 | 47.8 | 47.8 | 47.9 | 47.8 | –0.1 | 0.0 |
| 1RR | –49.1 | –49.1 | –49.0 | –49.2 | 0.2 | 0.0 | –0.4 | –0.4 | –0.3 | –0.4 | 0.1 | 0.0 | 47.9 | 47.9 | 47.9 | 47.8 | –0.1 | 0.0 |
| 2LF | –49.3 | –49.3 | –49.2 | –49.3 | 0.1 | 0.0 | –0.3 | –0.3 | –0.2 | –0.4 | 0.2 | 0.0 | 48.1 | 48.1 | 48.2 | 48.1 | –0.1 | 0.0 |
| 2RF | –49.4 | –49.4 | –49.3 | –49.5 | 0.1 | 0.0 | –0.3 | –0.3 | –0.2 | –0.4 | 0.2 | 0.0 | 48.4 | 48.4 | 48.4 | 48.3 | –0.1 | 0.0 |
| 2C | –49.6 | –49.6 | –49.5 | –49.6 | 0.1 | 0.0 | –0.5 | –0.5 | –0.4 | –0.5 | 0.1 | 0.0 | 48.2 | 48.2 | 48.2 | 48.1 | –0.1 | 0.0 |
| 2LR | –49.2 | –49.2 | –49.1 | –49.3 | 0.2 | 0.0 | –0.4 | –0.4 | –0.2 | –0.5 | 0.3 | 0.0 | 48.0 | 48.0 | 48.0 | 47.9 | –0.1 | 0.0 |
| 2RR | –49.3 | –49.3 | –49.2 | –49.4 | 0.2 | 0.0 | –0.5 | –0.5 | –0.5 | –0.6 | 0.1 | 0.0 | 47.8 | 47.8 | 47.9 | 47.8 | –0.1 | 0.0 |
| 3LF | –49.5 | –49.5 | –49.4 | –49.5 | 0.1 | 0.0 | –0.3 | –0.3 | –0.1 | –0.4 | 0.3 | 0.0 | 48.4 | 48.4 | 48.4 | 48.3 | –0.1 | 0.0 |
| 3RF | –49.1 | –49.2 | –49.0 | –49.2 | 0.2 | 0.0 | –0.4 | –0.4 | –0.3 | –0.5 | 0.2 | 0.0 | 47.8 | 47.8 | 47.9 | 47.8 | –0.1 | 0.0 |
| 3C | –49.4 | –49.4 | –49.3 | –49.5 | 0.1 | 0.0 | –0.5 | –0.5 | –0.4 | –0.5 | 0.1 | 0.0 | 48.2 | 48.2 | 48.2 | 48.2 | –0.1 | 0.0 |
| 3LR | –49.4 | –49.4 | –49.3 | –49.5 | 0.2 | 0.0 | –0.5 | –0.6 | –0.5 | –0.7 | 0.2 | 0.0 | 47.6 | 47.6 | 47.7 | 47.6 | –0.1 | 0.0 |
| 3RR | –49.2 | –49.2 | –49.1 | –49.3 | 0.2 | 0.0 | –0.6 | –0.6 | –0.5 | –0.7 | 0.2 | 0.0 | 47.6 | 47.6 | 47.7 | 47.6 | –0.1 | 0.0 |
| 4LF | –49.0 | –49.0 | –48.9 | –49.1 | 0.2 | 0.0 | –0.5 | –0.5 | –0.4 | –0.6 | 0.2 | 0.0 | 47.7 | 47.7 | 47.8 | 47.6 | –0.1 | 0.0 |
| 4RF | –49.1 | –49.1 | –49.0 | –49.2 | 0.2 | 0.0 | –0.4 | –0.4 | –0.3 | –0.5 | 0.2 | 0.0 | 48.0 | 48.0 | 48.0 | 47.9 | –0.1 | 0.0 |
| 4C | –48.4 | –48.5 | –48.2 | –48.6 | 0.3 | 0.0 | –0.7 | –0.7 | –0.6 | –0.8 | 0.3 | 0.0 | 47.3 | 47.3 | 47.4 | 47.2 | –0.2 | 0.0 |
| 4LR | –49.0 | –49.0 | –48.9 | –49.1 | 0.3 | 0.0 | –0.5 | –0.5 | –0.4 | –0.6 | 0.3 | 0.0 | 47.9 | 47.9 | 48.0 | 47.8 | –0.1 | 0.0 |
| 4RR | –48.7 | –48.7 | –48.6 | –48.8 | 0.2 | 0.0 | –0.5 | –0.5 | –0.4 | –0.6 | 0.1 | 0.0 | 47.4 | 47.4 | 47.5 | 47.4 | –0.1 | 0.0 |
| All Locations | –49.2 | –49.3 | –48.2 | –49.6 | 1.4 | 0.1 | –0.4 | –0.5 | –0.1 | –0.8 | 0.7 | 0.1 | 47.9 | 47.9 | 48.4 | 47.2 | 1.2 | 0.0 |
| Shelf Inlet | –49.7 | –49.6 | –49.0 | –51.0 | 2.0 | –0.1 | –0.2 | 0.0 | 0.3 | –1.5 | 1.8 | –0.2 | 49.4 | 49.4 | 49.4 | 49.3 | 0.1 | 0.0 |
Table 3: Shelf temperature by shelf at -50°C, 0°C, and 50°C over the 60-minute test duration.
| Target Temperature | Calculation | Shelf 1 | Shelf 2 | Shelf 3 | Shelf 4 |
|---|---|---|---|---|---|
| -50°C | AVG | -49.4 | -49.4 | -49.3 | -48.9 |
| MIN | -49.6 | -49.6 | -49.5 | -49.2 | |
| MAX | -49.0 | -49.1 | -49.0 | -48.2 | |
| MED | -49.4 | -49.3 | -49.4 | -490 | |
| AVG–MED | 0.1 | 0.0 | 0.0 | 0.1 | |
| MAX–MIN | 0.6 | 0.5 | 0.5 | 0.9 | |
| 0°C | AVG | -0.4 | -0.4 | -0.5 | -0.5 |
| MIN | -0.6 | -0.6 | -0.7 | -0.8 | |
| MAX | -0.2 | -0.2 | -0.1 | -0.3 | |
| MED | -0.4 | -0.4 | -0.5 | -0.5 | |
| AVG–MED | 0.0 | 0.0 | 0.0 | 0.0 | |
| MAX–MIN | 0.4 | 0.4 | 0.6 | 0.5 | |
| 50°C | AVG | 48.0 | 48.1 | 47.9 | 47.7 |
| MIN | 47.8 | 47.8 | 47.6 | 47.2 | |
| MAX | 48.2 | 48.4 | 48.4 | 48.0 | |
| MED | 48.0 | 48.1 | 47.8 | 47.7 | |
| AVG–MED | 0.0 | 0.0 | 0.1 | 0.0 | |
| MAX–MIN | 0.4 | 0.6 | 0.8 | 0.9 |
In addition to the shelf-by-shelf analysis, a single location on the shelf analysis can be completed (see Table 4). This analysis can show variability across the edges and the center among the different shelves. In this case, the center location was the most variable across the different shelves at the –50°C and 50°C target temperatures. At these extreme temperatures, the center location had a difference of up to 1.4°C and 1.1°C, respectively. Armed with this data, the lyophilization scientist can design a process robust enough to accommodate this dryer’s inherent temperature variation.
Run-to-Run Variability
It is recommended that the shelf mapping studies be completed as a Factory Acceptance, Site Acceptance, and Operational Qualification activity. This is also done for any change control verification after major maintenance activities or exchanging the heat transfer fluid. To correctly interpret the results after maintenance, it is necessary to establish a baseline variability of the shelf temperature mapping results. The same temperatures and locations should be assessed over two to four runs. These runs can be performed back to back within the same week, or they can be performed at different intervals throughout the year.
Table 4: Shelf temperature by location at –50°C, 0°C, and 50°C over the 60-minute test duration.
| Target Temperature | Calculation | Center | Left Front | Left Rear | Right Front | Right Rear |
|---|---|---|---|---|---|---|
| –50°C | AVG | –49.2 | –49.3 | –49.3 | –49.2 | –49.1 |
| MIN | –49.6 | –49.5 | –49.5 | –49.5 | –49.4 | |
| MAX | –48.2 | –48.9 | –48.9 | –49.0 | –48.6 | |
| MED | –49.5 | –49.3 | –49.3 | –49.2 | –49.1 | |
| AVG–MED | 0.2 | 0.0 | 0.0 | 0.0 | 0.1 | |
| MAX–MIN | 1.4 | 0.6 | 0.6 | 0.5 | 0.8 | |
| 0°C | AVG | –0.5 | –0.4 | –0.5 | –0.3 | –0.5 |
| MIN | –0.8 | –0.6 | –0.7 | –0.5 | –0.7 | |
| MAX | –0.4 | –0.1 | –0.2 | –0.2 | –0.3 | |
| MED | –0.5 | –0.4 | –0.5 | –0.3 | –0.5 | |
| AVG–MED | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | |
| MAX–MIN | 0.4 | 0.5 | 0.4 | 0.3 | 0.4 | |
| 50°C | AVG | 47.9 | 48.1 | 47.8 | 48.1 | 47.7 |
| MIN | 47.2 | 47.6 | 47.6 | 47.8 | 47.4 | |
| MAX | 48.2 | 48.4 | 48.0 | 48.4 | 47.9 | |
| MED | 48.1 | 48.1 | 47.8 | 48.1 | 47.7 | |
| AVG–MED | –0.2 | 0.0 | 0.0 | 0.0 | 0.0 | |
| MAX–MIN | 1.1 | 0.8 | 0.4 | 0.6 | 0.6 |
Figure 4 shows that run 1 had slightly higher temperatures across the three temperatures measured at each shelf location. Although the average and median temperatures were higher for run 1 and there were some outlier locations at certain temperatures in run 1, 2, and 3, the total magnitude of the range of temperatures at each location over four runs for the 60-minute data analysis period is 2°C or less at each setpoint (see Table 5). Conclusions about the reproducibility of the shelf temperature mapping data can be drawn only after multiple runs at each shelf temperature setpoint.
Figure 4: Variability of shelf temperature, five locations per shelf, over four runs at A) –50°C, B) 0°C, and C) 50°C, where X = mean, inner line = median, top and bottom of the box are the upper and lower quartiles, and the whiskers are the MIN and MAX.
Table 5: Shelf temperature by run at –50°C, 0°C, and 50°C over the 60-minute test duration.
| Shelf Setpoint | Run 1 | Run 2 | Run 3 | Run 4 | Total | |
|---|---|---|---|---|---|---|
| –50°C | AVG | –48.8 | –49.2 | –49.2 | –49.2 | –49.1 |
| MIN | –49.3 | –49.6 | –49.6 | –49.6 | –49.6 | |
| MAX | –48.2 | –48.8 | –48.3 | –48.2 | –48.2 | |
| MAX–MIN | 1.0 | 0.8 | 1.3 | 1.4 | 1.4 | |
| 0°C | AVG | 0.0 | –0.4 | –0.4 | –0.4 | –0.3 |
| MIN | –0.5 | –0.6 | –0.7 | –0.8 | –0.8 | |
| MAX | 0.5 | –0.2 | –0.1 | –0.1 | 0.5 | |
| MAX–MIN | 1.0 | 0.5 | 0.6 | 0.7 | 1.3 | |
| 50°C | AVG | 48.5 | 48.0 | 48.0 | 47.9 | 48.1 |
| MIN | 47.7 | 46.9 | 47.2 | 47.2 | 46.9 | |
| MAX | 49.0 | 48.8 | 48.5 | 48.4 | 49.0 | |
| MAX–MIN | 1.3 | 1.9 | 1.3 | 1.2 | 2.1 |
The variability of the shelf temperature (MIN to MAX temperatures) can vary widely based on the aggregation of the locations being measured. The narrowest temperature variation, a range of 0.1°C to 0.3°C, is observed when comparing the minimum and maximum across a single location over the test duration (see Table 2). The temperature variation expands to 0.9°C when calculating the minimum and maximum of the five locations across each shelf for the study’s 60-minute duration (see Table 3). Evaluating all the temperatures as a single dataset population of results yields the more accurate reflection of the temperature distribution. When considering all raw data, the minimum to maximum temperatures is 1.4°C (see Table 2). Finally, the largest variation is observed when considering run-to-run variability, with the minimum to maximum range of 2.1°C (see Table 5).
Identifying the Test Reference Temperature
The results of each assessment described previously could be compared with the shelf inlet temperature (shelf inlet) at each collect interval, the average shelf inlet temperature over the test interval (average shelf inlet), or actual target setpoint intended for the study (setpoint). The shelf temperature setpoint, the shelf inlet measured temperature at a single time point, and the average temperature measured over the test duration may be as much as 0.9°C different (see Table 6). Careful consideration of the data, method of analysis, and interpretation and understanding of the results are critical to reaching a sound conclusion. Evaluation of the test results for the maximum, minimum, and average, relative to different references, is compared in Table 7.
Table 6: Shelf inlet setpoint and measured temperatures at a single time and over the test duration.
| Shelf Setpoint (A) | Shelf In at 60 Min (B) | Average Shelf In Over Test Duration (C) |
|---|---|---|
| –50.0 | –49.1 | –49.7 |
| 0.0 | 0.0 | –0.2 |
| 50.0 | 49.3 | 49.4 |
Comparing the MIN and MAX shelf temperature across all locations in the dryer to the shelf inlet temperature at any one point may be misleading. For example, at the 60-minute time point at a target temperature of –50°C, the minimum temperature measured across all locations is lower than the shelf inlet by 0.4°C. But when comparing the average shelf inlet temperature across the test duration to the shelf temperatures across all locations for the full test duration, the minimum temperature achieved is 0.1°C above the average shelf inlet temperature. Evaluating only a single snapshot in time may yield erroneous conclusions about shelf temperature control.
Table 7: Difference between shelf surface temperatures across all locations and reference temperature.
| -50°C | 0°C | 50°C | ||
|---|---|---|---|---|
| Shelf surface for test duration – Shelf setpoint (Table 7, A) | AVG | 0.8 | –0.4 | –2.1 |
| MIN | 0.4 | –0.8 | –2.8 | |
| MAX | 1.8 | –0.1 | –1.6 | |
| Shelf surface at 60 minutes – Shelf in at 60 minutes (Table 7, B) | AVG | –0.1 | 0.0 | 0.0 |
| MIN | –0.4 | –0.2 | –0.5 | |
| MAX | 0.7 | 0.1 | 0.5 | |
| Shelf surface for test duration – Average shelf in for test duration (Table 7, C) | AVG | 0.5 | –0.2 | –1.5 |
| MIN | 0.1 | –0.6 | –2.2 | |
| MAX | 1.5 | 0.1 | –1.0 | |
Also, a cumulative calibration offset for the probes and the resistance temperature detector measuring the shelf inlet comes into effect. When comparing run-to-run variability in the same dryer, comparing the shelf temperature with the average shelf inlet temperature as a reference will yield the most reliable assessment of any changes to the heat flux through the shelf surface, no matter how shelf inlet temperatures are calibrated or controlled. But the greatest range of the MIN, MAX, and AVG is observed when comparing the temperatures to the target setpoint. This assessment is mainly valuable when comparing the range of different lyophilizers or when the lyophilizer control system changes. Comparison of the shelf temperature to the setpoint becomes extremely important when evaluating how well different equipment performs and in transferring a product and process from one lyophilizer to another: It reflects the actual shelf temperature the product will be exposed to during the process.
Using only the average temperatures for presenting shelf surface data can be greatly misleading. Though the average value can offer “dampening” of extremes in results, it does not allow for detecting any extreme values or reveal the character of the entire data population. Therefore, it cannot show the actual shelf temperature variation across all the locations.
Conclusion
Shelf temperature mapping or uniformity studies identify the proper data collection rate, the analysis interval or period, and the approach to analyzing the results by location across all the shelves. Sufficient frequency of recording the temperatures for the various locations for the temperature probes on the shelf allows for capturing any oscillation in the shelf surface temperature. The study must be long enough for the shelf to reach the desired temperature and include ample time for the shelf temperature to stabilize and for collecting and analyzing data after the shelves stabilize.
Evaluating the average, median, minimum, and maximum temperatures at each shelf location over the data analysis interval will identify the temperature range across the shelves relative to the target setpoint and any location where the difference is greatest. The homogeneity of each location on the shelf, and shelf to shelf, can then be evaluated to determine whether any dryer locations behave differently than any other dryer locations. These shelf temperature mapping studies should be completed multiple times to evaluate any inherent variability of the temperature measurements.
Comparing the test results with the target setpoint yields the greatest range and difference across the shelves to the reference. This approach pays off particularly when comparing the lyophilizer’s performance with the equipment specifications and other lyophilizers’ performance. The significance reveals itself when comparing the performance and results of different lyophilizers when transferring a product from one lyophilizer to another. It is therefore essential to evaluate the actual shelf temperature achieved relative to the intended critical process parameter (CPP) when considering using the same CPP to process the same product in different lyophilizers.
Adequate control of the shelf temperature that the product experiences and a CPP for the process are essential. If there is variation in the shelf surface temperature relative to the target CPP, and the variation is within an acceptable range, then the product will likely be exposed to the same temperature profile when processed in different lyophilizers. This would signal that the resulting product would exhibit the same finished-product attributes when processed in the lyophilizer under evaluation, if all other influencing factors are comparable.
The most important conclusion of the studies is that with the right measurement method, a suitable study design, and rigorous data analysis, shelf temperature uniformity studies provide insight into the basic functioning of the heat transfer system. Such an approach is also a key foundation for achieving success when pursuing more challenging and rigorous lyophilizer performance measurements.
