Guided Wave Radar at Genentech

The bioprocess container with bag was transported to a Mettler Toledo Vertex floor scale (+/- 0.1-kg accuracy; calibration range: 0.0-800.0 kg), which recorded the tare weight. The bag was slowly filled with purified water through the bottom inlet line. The GWR probe was calibrated by recording the GWR output current (mA) and the corresponding floor scale weight using a total of 17 strapping points. The bag film surrounding the GWR probe was manipulated during each measurement to eliminate creases and folds. A picture of the bioprocess container with filled bag is shown in Figure 3.

Water Fill Studies:

A 500-L gamma-irradiated Sartorius-Stedim bag was installed in the bioprocess container and tared on the floor scale. While resting on the floor scale, the bag was slowly filled with purified water to 475L and subsequently drained. GWR and floor-scale volumes were recorded every 25L while filling and draining the bag. This procedure was repeated three times for a total of four runs (Runs #1-4) and a fresh bag was installed after each drain. The floor scale provided a stable, level surface so variability due to floor slope was not a factor.

Figure 3. Container with filled bioprocess bag

During Run #1, the bag was not manipulated or touched, except during the initial fill of the first 25L. GWR readings during the fill and drain for Run #1 are plotted in Figure 4.

During Runs #2-4, the bag film was manipulated between each reading to eliminate creases and folds in the area surrounding the GWR probe, which otherwise could have contributed to error in the GWR reading. Average GWR readings for these three runs, including 95% confidence bars, are plotted in Figures 5 and 6.

Cell Culture Harvest Studies:

The GWR system was also tested in pilot plant operations where we performed eleven fills with Chinese Hamster Ovary harvested cell culture fluid (HCCF), each time using a fresh bag.  The fill volumes varied from 165.4L to 445.1L. After each fill, the bioprocess container was transported to the floor scale and the net weight of the liquid was recorded, taking into account the tare weight of the bioprocess container and empty bag. The volume of liquid measured by GWR was also recorded.  The bag film surrounding the GWR probe was not manipulated in any way prior to recording the volume.  The results were tabulated and the percent deviation between the floor scale and GWR measurements was calculated (Table 2).

The data points were also plotted next to the water fill study average for Runs #2-4 and are shown in Figure 7. The conversion of floor-scale weight to volume was based on an HCCF density of 1.017 kg/L.

Results

Water Fill Studies:

Figure 4. GWR readings for Run #1 (fill & drain)—no bag manipulation

Figure 5. Average GWR readings while filling for Runs #2-4

Figure 6. Average GWR readings while draining for Runs #2-4

Figure 7. Cell culture harvest data points (Runs #1-11) 

Cell Culture Harvest Studies:

Table 2. Results from GWR vs. floor-scale comparison study

Click image for full-size table.

Discussion

Water Fill Studies:

The following observations were made from the water fill studies:

• Variability in GWR volume measurement was greater when the bag film was not manipulated, such as Run #1, compared to Runs #2-4 where it was manipulated.  Without tugging on the corners of the bioprocess bag during the fill, the bag does not always fill uniformly, especially at fill volumes less than 250 L. Also, by not adjusting the bag film surrounding the GWR probe, the film would trap pockets of water as it folded over, leading to erroneous signals. If the fold extended upward, the liquid level at the GWR probe was above the bulk liquid level and the GWR reading was artificially high (Figure 8). If the fold extended downward, the GWR reading was artificially low.  The best way to mitigate this issue was to manually manipulate the bag film to remove the fold.

  Figure 8. Bag with upward fold

  This was done for Runs #2-4 and the results from Figures 4 and 5 indicate improved GWR accuracy.
• The GWR accuracy improved as the fill volume increased, likely due to greater static head pressure inside the bag. The increased pressure naturally eliminated major creases and folds and helped draw the bag film tightly against the inside walls of the bioprocess container and GWR probe, leading to extremely accurate and consistent GWR readings between 250-475L as seen in Figures 5 and 6.
• It was difficult to measure volumes under 50L accurately using GWR, especially during the fill. With a low static head pressure, the bag did not press tightly against the GWR probe or conform to any consistent shape, thus GWR liquid measurements within this range were typically inconsistent.
• Some of the largest deviations in GWR accuracy occurred within the range of 150L-200L. Since the GWR reading was consistently low for all four runs, accuracy may be improved by recalibrating within this range or incorporating additional strapping points.
• GWR consistency improved with the presence of an air pocket above the liquid surrounding the GWR probe (Figure 9). The air pocket prevented liquid from drawing into the upward fold and impacting the GWR signal. Under these conditions, the liquid level at the probe was fully representative of the bulk liquid level.
• Overall, the GWR was accurate and consistent as seen by the 95% confidence bars in Figures 5 and 6. GWR appears to be a reliable method for liquid volume measurement as long as the calibration is done carefully.