Q: For shelf-life materials that have temperature and humidity requirements for storage, do you have to use the same temperature and humidity controls after the material is removed from storage and used in assembly areas?
Sun Valley, CA
A: The general answer is yes, but there are several things to consider. For shelf-life material, if any environmental conditions are critical, the storage area is assumed to be controlled for those conditions. A controller can create and maintain specified temperatures and relative humidity (RH) levels, and a data logger can monitor these factors at scheduled intervals to verify the conditions were constant.
Monitoring equipment must be calibrated and traceable to the International System of Units through the U.S. Metric Program of the National Institute of Standards and Technology. A laboratory that is accredited to the ISO/IEC 17025 standard can help your organization meet this metrological traceability requirement.1
If your storage area is not controlled after the material is removed—depending on the volume of storage—it may be more difficult to achieve the controlled temperature and RH conditions after another batch of environmentally sensitive material is acquired for storage.
During an audit assessment, it’s your organization’s responsibility to prove that required environmental conditions were maintained at all times, including time in storage. Without time-stamped data that’s logged with a traceable instrument, it’s difficult to prove environmental conditions were maintained, including a user’s intention to not maintain the storage conditions when material was not stored. This also includes the amount of time needed to reach a specified temperature and RH condition because you also must consider factors in stabilization from ambient conditions.
It also is prudent to include records for time and date of storage, and any removals of sensitive material that can be compared with the temperature and RH data logs for compliance verifications.
Dilip A. Shah
President, E = mc3 Solutions
- International Organization for Standardization and International Electrotechnical Commission, ISO/IEC 17025:2005—General requirements for the competence of testing and calibration laboratories.
ISO 9001 revision and AS9100
Q: How will ISO 9001’s 2015 revision affect AS9100, because this standard already incorporates all current ISO 9001 requirements?
San Lorenzo, CA
A: This question is best suited for—and dependent on—your third-party registrar, and what I provide can be only directional: The final answer must come from your "trusted advisor" for your quality system. With that said, a couple major differences are highlighted on the International Organization for Standardization’s website, such as the area of structure and risk-based thinking.1
The new structure has moved from five sections to seven. This allows ISO 9001:2015 to be more user friendly for certified organizations that are using multiple management systems.
The other main difference is how risk-based thinking is applied. Risk-based thinking was always part of ISO 9001. But previously, the idea was that risk-based thinking was part of what organizations did intrinsically. Now, however, the expectation is that all risks must be considered. After reading several articles on the topic, I walked away feeling like a failure mode and effects analysis event will be a part of the documentation of the entire supply chain in an organization.
This is a brief answer to a broad question, and I want to emphasize that you should consult with your third-party registrar. It is well-suited to help you understand your specific implications for these two quality systems.
Certified quality engineer
- International Organization for Standardization, "ISO 9001: Moving from ISO
9001:2008 to ISO 9001:2015," ISO.org,
Safe and statistically sound
Q: I work at a seafood production facility that processes individually quick-frozen sea scallops. We measure shell defects (shell pieces attached to the scallops) by drawing a sample from a continuous-flow conveyor belt, counting the shell defects and rejecting the lot if we find defects in two consecutive samples.
Considering this process—which isn’t statistically sound—what is the best method of control that could be applied to the process to ensure we can take action in a timely manner? Visualize our process as one in which the scallops resemble thousands of marbles travelling on a conveyor belt, and a few might have shell pieces on them.
Lockeport, Nova Scotia
A: While the process described is not statistically derived, it raises three questions that must be answered before instituting a new process:
- How large is the sample you’re currently gathering and possibly rejecting?
- Based on past data, what percentage of your samples are you rejecting, or what percentage of defective products do you think you have?
- What is the population size? This could be the approximate number of scallops on the belt per unit of time—such as 15 minutes, one hour, four hours or 24 hours—and how many samples are taken per time unit?
Using sampling for attributes based on the Handbook of Sampling for Auditing and Accounting (McGraw-Hill, 1974) and a statistical sample size calculator,1 we were able to create the following example:
If your production facility processes 5,000 pounds of scallops per hour and there are, on average, 20 scallops per pound (or 100,000 scallops processed per hour), and you believe 2% or less of your scallops have shells (are defective):
- You want to be 95% sure that the occurrence of shells is less than ±2%.
- You want the precision of your estimate to be ±2%.;
You would need a random sample of 188 scallops per hour to be 95% certain that you are estimating the true incidence of shells with ±2% precision.
If 188 out of 100,000 are too many tests to conduct or begin taking too long, there are many ways to personalize the process to your facility to minimize loss. The test also would be based on a batch size that’s specific to the facility if different processing steps have varying failure rates.
Taking a random sample from a continuous conveyor over a fixed time also can be daunting. We would recommend starting at a random place in the batch and taking a systematic sample from that point on. Statistically speaking, this is a well-documented technique.
Pharmacogenomics doctoral student
University of California-San Francisco
Professor of biostatistics
University of California-San Francisco
- "Statistical Sample Size Calculator," Cognet QC Systems, cogentqc.com/tools-resources/statistical-calculator.
"Skaginn Completes Frozen Scallop Processing Line for Eastern Fisheries," Undercurrent News, May 6, 2015, http://tinyurl.com/scallop-processing.
"Scallop Program," Tichon Seafood Corp., http://tinyurl.com/scallop-program.