Pfizer: Right First Time, Twice

April 21, 2005
Two black belt teams at Pfizer's Kalamazoo plant confronted challenges and doubters head on, and both solved troubling issues within the calendar year.
TEAM OF THE YEAR FINALIST, SMALLER-SCALE PROJECTS:PFIZER'S KALAMAZOO TRI BLENDER AND PROCESS IMPROVEMENT TEAMSPrior to the start of Pfizer, Inc.&rsquos Right First Time quality initiative, manufacturing teams throughout the company often faced what Jill E. June, director of Right First Time at the Kalamazoo, Mich. plant, calls &ldquoblack hole issues&rdquo &mdash chronic problems that seemed to vacuum up resources and clever ideas and yet never get resolved.Last year, under the Right First Time umbrella, two black holes were destroyed. One was a chronic cleaning validation problem for a key product&rsquos tri blenders. The other involved chronic low yields and increasing impurity levels for an API process. Two Six Sigma black belt teams were put together to address the issues. A year later, both can boast that their missions have been accomplished.TRI BLENDER TRIALS, TRIBULATIONS AND TRIUMPHSTri blenders for one key product at Kalamazoo are cleaned about 400 times per year. Repeated attempts to validate the cleaning process, however, had failed. The product involves a hard-to-clean suspension, meaning the tri blender cleaning could not be done safely with solvents. Instead, the equipment had to be taken down, and parts cleaned by hand, put into an automated washer, then manually unloaded and dried.Every single cleaning required a 100% inspection on the part of the cleaning validation lab, consuming about 25% of the lab&rsquos time and resources. Part cleaning had a high failure rate, necessitating extensive recleaning. Tabitha Bratt, the cleaning validation lab manager and a Six Sigma black belt in training, was chosen to head up an eight-person cross-functional team to address the tri blender troubles once and for all. Bratt&rsquos team included several line operators and operations mechanics that had been with Pfizer for 15 or 20 years.As a process engineer at the plant, Bratt had often had moments of inspiration about how to fix the tri blender problems. Many operators had proposed solutions as well, yet no program or teams existed to carry them out. Right First Time presented that opportunity. As part of her black belt training, Bratt needed a project to work on, so the tri blender project was the perfect fit.The team began by meeting twice a week. They conducted several walk-throughs in the shop and mapped every step in the process. Each team member had action items to address, and compared notes during meetings. They began to develop a list of root causes.Importantly, they listened to the team members who worked on the floor. One operator had been frustrated with the cleaning of one particular small part. &ldquoWhy don&rsquot we just throw the thing away?&rdquo he wondered. That&rsquos what they did, and went to using disposable ones.Eventually, the team drafted new operating procedures and videotaped the cleaning process for computer-based training for all operators. In October, the process was finally validated. There was a lot of excitement on the floor, Bratt says, and a little apprehension on the part of the operators. They, not the lab, now owned the validation process.The key was getting the operators involved from the start, says June. &ldquoYou have to put people in positions of responsibility who understand what the problem is,&rdquo she says. &ldquoThese guys were living it, breathing it, and seeing it every day.&rdquoThe experience &ldquoopened up a new way of working out problems,&rdquo says Tim Wang, a process technologist on the team. &ldquoIt caused the operators to be more accepting of change and more willing to participate in driving change.&rdquoHYPOTHESIZING AN APIA process in the API plant had seen yield declines and cycle time increases for three years prior to Right First Time. High levels of an impurity required a time-consuming and expensive recrystallization step. Costs increased to the point that it appeared they would have to be passed on to the product&rsquos one customer.Cameron Clark, a production chemist, had been on the case for months, doing initial trending and looking into old batch records to determine what had changed. He had some intuitions about what the problem was, but neither the time nor tools to dig deeper.Still, Clark was not enthused when a black belt was &ldquothrown at the problem.&rdquo Once he met black belt Sarah Blech and realized that there was a firm company commitment and ample resources behind the effort, he began to come around.The Process Improvement team was cross-functional, with representatives from the engineering and chemistry departments (including Clark), in addition to one experienced line operator. The team got to work and held brainstorming sessions. Blech began testing hypotheses, collecting and analyzing data and taking results to the team. This prompted further lab comparisons. The team would brainstorm again, look at causes and effects, and draw fishbone diagrams of the process. In time, they put together a master sample and data collection plan to make sure no stone was unturned. Eventually, they narrowed their focus to a few specific areas of the process.One roadblock was that lab tests failed to agree with results found on the plant floor. &ldquoManagement felt the problem was operational in nature,&rdquo Clark recalls. The team suspected otherwise. They felt one of the problems might relate to materials of construction. Despite management&rsquos skepticism, they persisted in investigating their hunch. One suggestion was that a metal part being used in the lab was made from a slightly different alloy than that on the floor. &ldquoWe went down to the maintenance shop, picked up a couple bolts and gave them to Cameron in lab,&rdquo Blech recalls. Lo and behold, when the lab used the bolts, the process started behaving like it did in the plant.There were many potential root causes of the problem, June says. What&rsquos important is that without the cross-functional expertise of the team, perhaps only one or two would have been addressed.The operators &ldquowere suffering along with us,&rdquo says Rich Greter, a production engineer in the plant, &ldquoand as they saw we were making strides toward success, they took some ownership in that. That led to further willingness to participate in the project.&rdquoThere was a &ldquoEureka!&rdquo moment when results from the first &ldquonew&rdquo batches started coming in. &ldquoEach day we seemed to get better and better news in terms of yield and quality,&rdquo Greter says. &ldquoWe finally came to the realization that we had latched on to something.&rdquo