Master Batch Records (MBRs), i.e. general manufacturing instructions, are an important component of Manufacturing Execution Systems (MES). They form the basis for the description of pharmaceutical manufacturing processes, which later lead to batch records containing all pharmaceutically relevant data.
Designing MBRs in an MES on the basis of the physical process flows is therefore the essence of digital transformation in manufacturing. It's a crucial translation exercise at the interface of the material process steps on the shopfloor and the IT solution. This means MBR design is where the magic happens!
We have established know-how and expertise in MBR design using Werum PAS-X MES for many major customers. During MES PAS-X projects, MBR Design is always a game-changer: It can be a bottleneck and massively delay the going live, or, if done right, it will drive efficiency and greatly reduce the number of hiccups when the system is live.
Here, our MBR design expert and Senior Consultant Dr. Thomas Utschig explains the approach to successful MBR design projects.
Efficient PAS-X MBR Design
written by Dr. Thomas Utschig (MBA), Senior Consultant Manufacturing IT, FrontWell Solutions
MBR design can be accomplished in an efficient manner only, if a certain basis has already been established, as the following example from a biopharma MES (manufacturing execution system) project emphasizes.
MES for a new biopharma production facility
For a newly constructed production facility for a new group of products, the implementation of a MES (PAS-X) was commissioned, with the PMO in charge. For the task to support the MBR design, I was hired as external PAS-X consultant.
Well used features of PAS-X
Right after the onboarding and obligatory company as well as system specific trainings, I found that some features of PAS-X were already prepared to an extent that modelling could commence. For simple operations, that appear in the same sequence frequently, it is advisable to create a library element (LE), that contains this sequence. E.g. for draining a process vessel, the following process steps can be conveniently aggregated into a LE ‘Drain process tank’: 1) apply pressure, 2) open and (after a delay) close the bottom valve, 3) vent the vessel.
(1) Library elements
LEs are customized building blocks of a set of (repeated) process steps, automated decisions, etc., that are built in as references, additionally with the option of individual parametrization in each referenced case. They have several advantages:
- The MBR/ESP (equipment specification) design appears much more concise and well- arranged.
- The review of all MBR/ESP designs is eased, as any LE must be reviewed only once – upon its release – while if implemented as reference, only the individual parametrizations must be reviewed, respectively.
- Modelling new MBRs/ESPs is sped up significantly by using these existing building blocks.
- The future MES maintenance is greatly eased, as any necessary change (e.g. a changed command name due to an update in the distributed control system (DCS) must be done only in one central dataset (the corresponding LE) in the library and is automatically implemented into all references in all MBRs/ESPs.
LEs should be consequently utilized right from the beginning, wherever possible. Model- ling, reviewing and releasing them takes some effort upfront but makes the work – and related workflows – afterwards much more efficient, therefore it is highly recommended to use them.
(2) Generic MBRs
Since this facility was planned for a group of similar products, it was a good and obvious decision to utilize the GMBR (generic MBR) concept to increase the efficiency of MES implementation, as the GMBR functionality can be used for a group of target materials (TM) with similar production processes.
All TMs within one group have the same GMBR. The differing values for each TM are modelled as parameters into the GMBR design. Afterwards, those parameter values are entered into TM-specific parameter value lists (PVLs).
From a GMBR together with a PVL, a parameterized GMBR (PMBR) can be generated, that will produce one specific TM. For a number of n TMs within one group: GMBR + PVLn = PMBRn
Great care must be taken if any incoming material or target material of an MBR are handled by the ERP. The same spelling of material IDs in both systems and the same order in lists handled by both systems are mandatory and a frequent source of initial malfunctions.
Issues
The main issues I faced during this project were that (1) the master data (MD) and state diagrams (SDs) had not all been created and allocated, yet, and (2) the process flow was nowhere documented in a comprehensive manner and furthermore, it even had not been finalized in all details, which has a quite severe impact on the digitalization process, as pointed out below.
(1) Master data & state diagrams
To create an easy-to-use set of MD, it is necessary to choose a standard format that is compatible with all interfaces. Requirements regarding ERP-inherited objects must be considered as well as those for command names of the DCS. To ensure consistency during MD creation or modification, a documentation policy is recommended. Data sets must be complete at the beginning of the modelling phase, to be usable. For any piece of equipment (EQ), a unique EQ ID needs to be defined, that belongs to an EQ type which must be defined prior to the EQ ID, for which an EQ class must be selected in the first place.
Finally, a SD needs to be assigned to the EQ type, to ensure GMP - conformity during operation. In that all foreseen transitions between different states (like ‘clean’, ‘in use’, ‘in cleaning’, ‘in quarantine’, etc.) are defined, as well as all allowed transitions from any source states to target states, respectively.
(2) Process flow not finalized
Without a comprehensive documentation of the process flow, MBR design is vague task and frequent reworks are unavoidable, making it overall quite inefficient. In this project, at the beginning of my assignment, the process details were known to such limited extent yet, that even the DCS operations at the interface to the various process equipment were not all functional, and several iterations of reconfigurations including new parametrizations had been necessary.
Too few and therefore overloaded process and automation engineers in reach were the reason for interruptions of the modelling workflow, that caused a lower efficiency due to frequent switching between open ends.
(3) Other issues: High-level parametrization options initially not fully available
High-level parametrization options for MES – actually features of the GxP - accredited platform PAS-X – should never be ruled out like it was the case in this project firstly:
Supposedly, in the late planning phase a restraint must have appeared at the intersection of operations, quality, and IT departments within the project management triangle (scope, time, cost). This led to ruling out the high-level parameters CX and CA from the scope of risk assessment for ESP, while in parallel they were chosen to be implemented and there- fore assessed for MBR (master batch records) right from the start.
CX: characteristic – editable within ESP/MBR
CA: custom attribute – editable in the master data but fixed within ESP/MBR
During the project realization, the need for these features in combination with ESP appeared. Fortunately, the extra effort for the risk assessment for ESP is marginal, as the structure and features are very similar to the ones of MBR – and even reduced by material related functionalities, but it certainly caused a major interruption and time delay.
Long term effects had clearly not been considered in the decision of ruling out CX and CA for ESP, as the maintenance cost of software linearly increases with the number of data- sets to be maintained during their life cycle management (LCM). However, this maintenance cost can be decreased exponentially by introducing high-level parameters like CX and CA in PAS-X.
Thinking of later necessary modifications due to e.g. capacity increase, debottlenecking, or simply the physical equipment life cycle, foreseeable performance-level decreases were neglected: Without CA, an insignificant or noncritical hardware change due to introduction of a later released equipment model or a new equipment supplier cannot be implemented without increasing the equipment type number, triggering a document LCM-workflow and thus follow-up costs.
(4) Other issues: Incomplete Run Types – ESP/MBR matrices
Often, the MES is clustered hierarchically into several run types with increasing scrutiny concerning compliance (and therefore increasing effort in reviewing by extended circles of reviewing departments), like
(a) commissioning/test,
(b) engineering/qualification,
(c) GMP/mass production.
In the matrix of run types vs. ESP/MBR, better all possibilities should be enabled. My experience showed that ruling out one or more options will greatly hinder a smooth MES implementation afterwards, once the demand for a ruled out run type – ESP/MBR combi- nation appears.
Learning & Resolution
Thorough upfront planning plus preparing different built-in functionalities presuppose a smooth and efficient implementation of any chosen MES. Consistent MD creation is a time- consuming task, as is the work on SDs. Both should be completed upon starting the modelling work, in order to prevent tedious rework. Also, the preparation of LEs is better done upfront, at least to some extent, so that LE utilization becomes feasible as a practical tool for a good designing experience.
Without a fixed and described in detail process flow, efficient MBR design is simply not possible.
Continuous support of subject matter experts (SME) from all key areas right from the beginning is a prerequisite for success. Most important SMEs are from process engineering, automation, infrastructure IT, and MES, not to forget those in charge of quality. Ideally, they should be available already from the later planning phase, when first more detailed decisions must be taken. This helps in avoiding strategic errors in excluding built-in features or functionalities upfront, even if they may appear superfluous for a perceived simple process to be digitalized, as shown above.
Dr. Thomas Utschig (MBA), is a Senior Consultant Manufacturing IT at FrontWell Solutions GmbH where he is leading on consulting in Manufacturing IT, MES implementations and MBR design with focus on the Pharma & Life Sciences sector.
Efficient PAS-X MBR design.