Roche Gets to the Core of MES

F. Hoffmann-LaRoche is in the midst of implementing a worldwide MES system based upon a core application that is adopted by local sites with minimal customization. The idea is to establish greater global operational consistency, centralize best practices, reduce errors, and ultimately drive significant cost savings. We spoke with Robert Fretz, who heads up the MES program as part of Roche’s Pharma Global Engineering group, about the challenges and benefits of going global.

PhM: Roche is centralizing MES management and a common standard MES core application is being rolled out to all local sites. What were the key drivers for this?

R.F.: The central approach allows us to collect all experiences and best practices into one system for the benefit of all sites, and at the same time align all business processes to the same high compliance and efficiency standards.

We save money by avoiding parallel developments of site-specific applications. And we have better and more efficient support for MES at smaller sites by providing knowledge from a center of excellence.

PhM: How far along are you in terms of completion of the overall project?

R.F.: We have eight sites using MES, six of which were using PMX [Rockwell Automation’s MES application] already in the past. In 2009, three of these began using the MES Core. Three more sites will move to the MES Core in 2010; by end of 2011 all eight sites will use the common MES Core system.

PhM: Your strategy allows for some (less than 10%) customization at local sites, but no additional functions outside of the core functions. What is the rationale behind this?

R.F.: Customizations create a legacy problem when upgrading to newer system versions, because at least a part of the customizations will have to be re-implemented in the new version, creating cost in terms of development, testing, and validation. Although we are challenged to reach less than 10% customizations by the different type of site operations (solids, parenterals, packaging) to be supported by the same system, having too many customizations will lead to being stuck with an older version as costs for upgrades are getting prohibitively high compared to expected benefits.

We do not allow sites to have additional functions outside the core functions because allowing this—and we had that in the past—leads to all systems being different with consequent loss of functional and support synergies. The process to collect all requirements into a common solution challenges new requirements and allows input from the complete MES community, which will make sure the new function fits and complements existing functions.

PhM: Briefly, how are you gauging the success of implementations worldwide and collecting valuable feedback from local sites?

R.F.: Every MES implementation project has to meet functional, compliance, and financial criteria before it is started. Post implementation reviews measure results against the criteria. Our experience shows that with our core approach subsequent projects can start earlier, are quicker to implement, have fewer problems going on-line, and stabilize in a shorter time. Operations run smoother and smoother, also because users and local support staff can be trained early in existing sites. Project and operational costs of systems are continuously reduced.

PhM: You advise, “Expect considerable effort to connect a diverse collection of already installed equipment.” Can you share a few horror stories or elaborate on this?

R.F.: In the past, designing and installing pharmaceutical production equipment concentrated on the physical output. Data collection in a batch context was not seen as important in a paper-based operation. Diversity of equipment controls was not a problem because equipment was often maintained and serviced by the vendor, especially in smaller sites without a big engineering or maintenance department.

As we go electronic we find that the data are getting as important as the physical product, as we can not release any product without proper documentation of the manufacturing process—today typically managed by MES. It is only natural to collect production data at the source.

Here a number of challenges exist: First, the relevant data must exist in the machine control system. Second, the machine data must also contain recipe, batch, and phase information in order to be later easily grouped when analyzed at batch review time. Often such context data does not exist in older equipment. Third, even if the data exist it can be difficult to access online before the end of a batch, which is needed to automatically capture bad trends or potential deviations as early as possible. Fourth, physical connections to older machine controls are slow or non-standard—if they exist at all.  Fifth, older machine controls are mostly not aware of any standard data access protocols—such as OPC—used today. The combination of the five challenges in a grown environment can make your life difficult and asks for a flexible architecture and can require additional post-processing before presenting the data to MES.

PhM: Have there also been challenges related to site culture and getting people to shift from manual/paper-based documentation to electronic batch records?

R.F.: Yes. First, the operators have to get used to get their instructions step-by-step from a computer screen versus a paper ticket in which they can freely move forward and backward. This “lack of flexibility” perceived initially has now changed to an appreciation of the focus generated by concentrating on the current operation step.