In one of those fortuitous hallway conversations, I discovered that Emerson’s Bill Robertson had written a project management article for Pharmaceutical Manufacturing magazine. Bill is a certified project management professional (PMP) and professional engineer (PE).
His article, Vertical Slicing for Customer Acceptance Tests, describes a different methodology for efficiently executing automation projects than has typically been done. Bill describes a pharmaceutical batch process for a set of bioreactors with clean-in-place (CIP), fill, stir, heat, and dump operations. Typically, each of these units will have separate detailed design specifications from which the automation system configuration and customer acceptance testing (CAT) is performed.
It seems intuitive that the most efficient way would be to execute the engineering effort in parallel on all of these units. Bill describes this process:
The project team will start marching down all these units simultaneously–all the CIP, then the fills, the stirs, and so on. There is parallel activity going on, but engineer A, B and C may not conduct their work in exactly the same way.
These inconsistencies between units get uncovered in the CAT phase, and often result in considerable rework. This rework may need to be applied to all of the units. It’s the discovery and rework that causes project delays, tension among the project team, and communication breakdowns. This is amplified in large projects with global teams executing portions of the projects from different countries.
Bill describes a counterintuitive approach, vertical slicing. Instead of executing the designs for all of the units in parallel, Bill describes the approach:
…a “vertical slice”. The “vertical” term was used in reference to the ISA88 batch architecture standard. The “slice” taken was a single unit, a bio-reactor, with the most number of steps. Components of the selected unit included I/O, Control Modules, Equipment Modules, Phases, Operations, Procedures and Unit Procedure.
We picked the most complex one because we wanted to be sure that we captured all the complexities, variations of configuration, and the sequencing that may come up on the project.
By taking one of the most complex units from start to finish, it allowed the pharmaceutical manufacturer project team members to review the design and execution, involve the operations team to incorporate their improvement ideas, and foster good communications among all of the team members, even when separated by great geographical distances.
The end result is a good blueprint for the rest of the units, or “85% of the execution” as Bill describes it. The inefficiency of executing a single vertical slice from start to finish is offset by the overall reduction in project risk. Bill notes:
There are huge savings in avoiding problems and making deadlines though it’s hard to put a price tag on them.
As the project progressed to other units, this same vertical slice approach was used. Bill summarizes how this process led to a more efficient project:
The vertical slice concept forces the customer and the vendor to talk earlier, agreeing to what the procedures are and what the programming should look like. This allows for efficient execution and predictability in the outcome of the product acceptance test.
This approach has been very successfully applied in new product development processes. It makes sense that even if counterintuitive–it will work for most process automation projects as it did for this one.