H. Jurgen Spanier

The accuracy of integrity testers that operate according to different measuring principles was evaluated by comparing their results with those obtained using a reference (weighing) method. Prior to the investigation, both instruments under examination were tested for correct function and accuracy in accordance with the manufacturer's data. The water flow test (WFT) was selected as the filter testing procedure because this type of test measures very low flow rates and thereby places the greatest demands on the accuracy of the measurement. In the case of diffusion and/or forward flow measurement (when comparable filter areas are used), the expected measured values are a magnitude higher than the values obtained using the WFT.	

The tests were done at a pressure of 2500 mbar. The flow quantities were varied, as were the net volumes. This made it possible to test whether the accuracy of the measurement was actually dependent on the net volume or not.	

The flow rates were set from 2 mL/h (equivalent to a WFT of 0.033 mL/min) up to 100 mL/h (equivalent to a WFT of 1.66 mL/min). Here, the pre-volume of 0 mL (and 57 mL, including tubing) varied up to 500 mL (and 557 mL, including tubing).	

All weighed values were converted to water volumes using the density that corresponded to the water temperature. The measured values were stated "per minute" to enable direct comparison with flow rate readings. The displays on both instruments were limited to readings with two decimal points. (The display on the balance used went up to three decimal points.) The error limits were defined as 65% tolerance, but had to be within the resolution of the instruments (0.01 mL/min).		

Figure 2:  Comparative measurements of flow measurements (FM) and pressure measurements (PM).			

The two filter integrity testers were connected one after the other to a power source and the measuring tubing was connected to the reference volume vessel. The exact gas volume setting (net volume) was adjusted by means of the water filling level. The flow controller was used to connect the outlet of the reference volume to a volumetric flask.			

Once the testing apparatus was set up as illustrated in Figure 1, the tests were done with the two measuring instruments (see sidebar "Measuring instruments used") and then evaluated according the test matrix shown in Table I.	

Analysis and interpretation of the results

In summary, both instruments yielded results within the tolerance limits (Figure 2). The minimal spread of the results was attributed to the limited resolution of the measuring instruments (two decimal points). The results showed that both test methods possess the necessary degree of accuracy required for the conventional areas of applications in which the filters are used.		

Reference flow 100 mL/h (1.666 mL/min). The reference flow setting of 100 mL/h was equivalent to a water flow of six 10 in. filter elements. The pressure drop measurement (PD) performed with a net volume of 0 mL yielded a measurement that was below the tolerance curve (Figure 3). This combination of very high flow and a lack of any net volume will practically never occur when testing the integrity of six filter cartridges.			

Figure 3:  Comparative measurements of flow measurements (FM) and pressure measurements (PM). Reference flow 100 mL/h (1.66 mL/min).			

Reference flow 50 mL/h (0.833 mL/min). The reference flow selected was equivalent to the water flow rate of three 10 in. filter elements. All results of the two measuring instruments were within the defined tolerance ranges and independent of the net volume setting (Figure 4).				

Figure 4: Comparative measurements of flow measurements (FM) and pressure measurements (PM). Reference flow 50 mL/h (0.833 mL/min).			

Reference flow 20 mL/h (0.333 mL/min). The reference flow of 20 mL/h was equivalent to the water flow rate of a 10 in. filter element. Here too, virtually all results for the two measuring instruments were within the tolerance range (Figure 5).	

Reference flow 10 mL/h (0.166 mL/min). This reference flow was equivalent to the WFT value of a 5 in. filter cartridge. All results for the two instruments were within the tolerance range, independent of the net volume setting (Figure 6).				

Figure 5:  Comparative measurements of flow measurements (FM) and pressure measurements (PM). Reference flow 20 mL/h (0.333 mL/min).			

Reference flow 5 mL/h (0.083 mL/min). The reference flow setting of 5 mL/h was equivalent to the flux rate for water of a filter cartridge with 2000 cm2 (Mini, Junior filter series). Here, both devices produced deviations outside of the tolerance range at large net volumes (Figure 7). These deviations were attributed to the fact that the drop in pressure during the pressure drop measurement was too low. This pressure drop could no longer show a sufficiently accurate resolution because of the resolution limits of the pressure sensor. With the flow measurement method, the accuracy was similarly limited by the resolution of the pressure sensor. In this example, the pressure had to be kept constant during the measuring phase. Generally, it is not advisable to measure a low flow rate with too large a net volume, but because this combination rarely, if ever, occurs in practice, the deviations in the two test methods can be ignored.				

Figure 6:  Comparative measurements of flow measurements (FM) and pressure measurements (PM). Reference flow 10 mL/h (0.166 mL/min).			

Reference flow ,5 mL/h (,0.083 mL/min). These results are of no practical interest because the measured flow rates belong to filter sizes that cannot be reliably determined with the water flow test. This is attributed to the display accuracy of the integrity testers used. The magnitude of the tolerance range does not leave enough leeway to expect that exact readings will be displayed (Figure 8).	

As a general rule, the smaller the flow to be measured, the smaller the given net volume must be. When the net volume is small enough, both the pressure drop measurement and the flow measurement method yield results that have practical relevance.	

It is in the interests of both manufacturers and users that these test devices be used in the production sector to help minimize risk (that is, contamination resulting from defective filters). The investigation described here compared two test methods available on the market with regard to their practical applicability. The reference points selected were directly connected with the filter assemblies most commonly used in the pharmaceutical industry. The water flow test results, which were less than 5 mL/h, indicated that both test methods have potential for improvement. 					

Figure 7:  Comparative measurements of flow measurements (FM) and pressure measurements (PM). Reference flow 5 mL/h (0.083 mL/min).						

The current debate regarding fundamentally increasing the accuracy, therefore, appears to be an academic one, having little relevance in terms of everyday use. When establishing the integrity test limit values for the respective filter, confounding variables, such as temperature and quality of the wetting medium, should always be factored in within a defined range. Moreover, the tolerance of both the integrity test method and of the instruments is accounted for within the definition of the limit values. It is additionally recommended that the instruments be qualified under the actual conditions that prevail during the user's application. It is also important to ensure that the specifications stated by the manufacturer (accuracy, reproducibility and error recognition) are adhered to during the specific application. 					

Figure 8:  Comparative measurements of flow measurements (FM) and pressure measurements (PM). Reference flow ,5 mL/h (,0.083 mL/min).			

The two test methods yielded results within the defined tolerance range of 5% of the reference value. There were no significant differences between the pressure drop measurement and the flow measurement method. Both test methods should be used under the same limitations. Particularly when measuring water flow, it should be borne in mind that the gas volume required has to be accommodated for by the filter units to be tested. This limitation applies to the flow measurement method just as equally as it does to the pressure drop measurement. In general, filter integrity testers should be suitably qualified for the applications in question. Some manufacturers of these instruments offer the appropriate validation support. Therefore, the user's ultimate reason for choosing a measuring method is not based so much on the accuracy of the method, but more importantly, his choice is influenced by factors such as validatibility, user friendliness, operator guidance features and service friendliness.	

Articles

Testing sterilizing grade filters using integrity testers has become a standard method in biopharmaceutical production and quality assurance. In accordance with international regulations and recommendations, these filters should be tested before and after filtration. For these applications, a variety of automatic integrity testers is available. Currently, there are two groups of devices that are used to conduct validated testing procedures, such as the bubble point test, the diffusion test, and the water intrusion test (WIT) and water flow test (WFT). Whereas one group of devices relies on the principle of flow measurement, the other group is based on pressure drop measurements. The following report compares the accuracy of the two test methods, using the WFT as a reference.