Live from WBF 2008--Getting an organization ready for ISA95

March 25, 2008
Jack Greene, from Alkermes Inc., was first up after lunch in Session II, chaired by Scott Sommer of Jacobs Engineering Group.

Getting an Organization Ready for S95 – Designing and Implementing Plant Models within Enterprise Applications

Overview:

The S95 Business Case The S95 Architecture Getting Ready for S95 Building the Model The CMMS Model The Microbiology Data Tracking Model The Di...

Jack Greene, from Alkermes Inc., was first up after lunch in Session II, chaired by Scott Sommer of Jacobs Engineering Group.

Getting an Organization Ready for S95 – Designing and Implementing Plant Models within Enterprise Applications

Overview:

The S95 Business Case The S95 Architecture Getting Ready for S95 Building the Model The CMMS Model The Microbiology Data Tracking Model The Discrepancy/CAPA Tracking Model The ERP/MRP Model Conclusions The S95 Business Case: There is a need for the Transaction-Based Enterprise Business Applications (e.g. – ERP, MRP, CMMS, EDS, LIMS) to transfer data among each other and to trigger changes of state based upon the Real-Time Control Applications (e.g. – SCADA, PLC, DCS). A cold room temperature alarm triggers.  What should happen? CMMS – A Work Order opens to examine/repair the cold room MRP – All Raw Material lots in that cold room transition from Released to Quarantined Discrepancy Tracking – An Investigation is opened LIMS – Material testing is initiated How is this workflow implemented in traditional application architecture models? The S95 Architecture Case: As the need for data transfer grows – so does the number of connections For N Business Applications you will eventually build 2N Database Connections 10 applications leads to 1024 connections This complexity makes application upgrades complicated as everything is interdependent Greatly increases long term costs How is this workflow implemented in the S95 model? As the need for data transfer grows – so does the number of connections But here, N Business Applications you will only eventually build 2N Database Connections 10 applications leads to 20 connections This reduction in complexity lessens the dependencies and makes the system easier to support Greatly reduces long term costs Electronic systems can be used to streamline and optimize business processes but… There is no magic Data Associations need to be built and models need to be developed and implemented. Getting Ready for S95 Example – A batch is to be made on Bulk Process Cell 3 (PC3), MES could answer: Are all of the PC3 instruments calibrated? Are there any open work orders on PC3? Is there pending Calibration/Maintenance work on Utility systems that PC3 needs? Are there any open PC3 Deviations/CAPA’s? Are all of the PC3 users trained on the required procedures? Is there a lot restriction open on PC3? Are there any Microbiology Alert/Actions pending on the PC3 suite ? Take Home Lesson #1 If the native application that holds critical data cannot bracket it by Area/Process Cell/Unit, then you cannot extract Area/Process Cell/Unit specific data Take Home Lesson #2 If the native application cannot extract Area/Process Cell/Unit specific data - than neither can an MES In order for an MES to reach its potential, each of the applications it uses must be aware of the S88/S95 plant model and have queries in place to facilitate MES functions. Implementing a S88/S95 model into business applications can take place long before an MES is built and enhances their use – even without MES. Building the S88/S95 Model Alkermes chose the S88/S95 plant model and looked at Business applications in terms of the model. We launched a core team with the SME of each business application. For each application we asked “How would we want to use this data?” and “What is the deepest S88/S95 layer that we need to go to get that data?” For the most part – almost no users needed to extract data from their apps more granular than the unit level (and many only need it to the Process Cell level) What open deviations are there for: The Plant All Bulk Manufacturing Bulk Process Cell 3 Rinse Vessel on PC3 The exceptions to this rule were the SCADA system and the CMMS system. In both of these cases, granularity was required to the lowest level available in the model.

CMMS

Microbiology Data Tracking

Discrepancy/CAPA Tracking

ERM/MRP

Application

Vendor

Brand

IBM

Maximo

Compliance Software

Solutions Corporation MIMS

Sparta Systems

TrackWise

Oracle

OPM

Number of

Model

Levels

Allowed

Unlimited

Three –

software limitation

Three - (limited by

our UI

Implementation

not the software

Three - limited by our

Implementation

model

Number of

Parents a

Child May

Have

One

One

One

One

Top Level in

the S88/S95

Model

Enterprise

Site - Each plant has

its own instance

Enterprise

Enterprise

The CMMS Systems Model Of all systems in this presentation the CMMS required the most level of granularity – So we started here. Opportunity – The application allows an almost unlimited number of model layers Challenge – a child object can only have one parent within a hierarchy, but in reality, a child has many parents

Example – a use point valve on a Bulk Process Vessel has the following parents:

It is a Child of Process Cell 3 It is a Child of the USP Water Distribution System It is a Child of the Utility PLC It is a Child of the Electrical Distribution System

Solution – Implement multiple hierarchies within CMMS so that a child can have as many parents as required:

Site Primary Hierarchy – usual S88/S95 Site Building Utility Hierarchy Site Process Utility Hierarchy Site Controls Hierarchy Site Power Hierarchy 

Site Primary Hierarchy

Area

Bulk Manufacturing

Filling

QC Labs

Other Labs

Product Storage Areas

Rooms With People

Mechanical and Prep Spaces

Process Cell

Each Bulk Line

Each Fill Line

Chemistry, Microbiology, Stability

Each Group

Each Major Storage Location

Each Zone/Floor

Each Zone/Floor

Site Building Utility Hierarchy

Area

HVAC

Steam

Chilled Water

Hot Water

Process Cell

Each AHU

Each Boiler, Each Pressure of Steam

Each Chiller, Each Cooling Tower, Each Loop

Each Heat Transfer Skid, Each Loop

Site Process Utility Hierarchy

Area

USP Water

Process Solvent

Process Waste

Aqueous Waste

Pure Steam

Compressed Air

Nitrogen

Process Cell

Generation, Storage, Distribution

Loading, Storage, Distribution

Collection, Storage, Unloading

Collection, Storage, Treatment, Discharge

Generation, Distribution

Generation, Drying, Instrument, Product

Storage, High Pressure, Low Pressure

The Microbiology Data Tracking Model Interviews with Microbiology users revealed that they only require the model to exist down to the Process Cell Level Challenge – The application can only have three model layers Challenge – Some rooms impact more than one Process Cell Example – a Gowning room is used to access Two Bulk Process Cells (PC2 and PC3) and a Sterile Equipment Room is used to Store Equipment for all Plant Process Cells.  How can you extract if there is an alert/action for PC3: Ask the application the status of PC3 suite rooms Ask the application the status of the shared PC2/PC3 rooms Ask the application the status of the global impact rooms The answer is the sum of the three answers Model – Implement multiple top level hierarchies broken out by zone so that it is possible to formulate a query for any one Process Cell or Clean utility.  This leads to the following classes Process Cell Impact EM – Rooms that impact exactly one Process Cell Process Cell X/Y Impact EM – Rooms that impact Process Cells X and Y Global Impact EM – Rooms that impact all Process Cells Routine Impact EM – Rooms that impact no Process Cells (only themselves)

Microbiology data Tracking Hierarchy

Area

Each Bulk Line

Each Fill Line

Routine EM

Global Impact

Utilities

Rooms

Each Room on that Line

Each Room on that Line

Each Room In this Category

Each Room In this Category

Each Monitored Utility

Sites

Each Room Sample Site

Each Room Sample Site

Each Room Sample Site

Each Room Sample Site

Each Utility Use Point

The Discrepancy/CAPA Tracking Model Interviews with the QA and Manufacturing users revealed that they only require the model to exist down to the Unit Level Challenge – The application does not use a hierarchal layer model Opportunity – It has an open API to create custom fields and pulldowns and map them to fields  Design Model User Interface (UI) Design Model – Implement multiple Plant Areas as Radio Buttons on the UI.  When one is selected, a Pulldown appears so users can assign the Process Cell. When one is selected, a Pulldown appears so users can assign the Unit. Database Design Model – When the Area/Process Cell/Unit assignments are made, write the results into three custom fields for each record.

Discrepancy/CAPA Tracking Hierarchy

Area

Bulk Manufacturing

Filling

QC Labs

Product Storage Areas

Building Utilities

Process Utilities

Process Cell

Each Bulk Line

Each Fill Line

Chemistry, Microbiology, Stability

RT Quarantined RT Released Cold Quarantined Cold Released

HVAC, Steam, Chilled Water, Hot Water

USP Water, Process Solvent, Waste, Air, Nitrogen

Unit

Each Vessel

Each Major Component is a Unit  

Chromatography Systems Computerized Lab Instruments Pipettes Meters Balances SCADA Monitored Equipment

Each Warehouse is a Unit

Each Major Component is a Unit

Each Major Component is a Unit

The ERP/MRP Model Interviews with the ERP/MRP Users revealed that they only require the model to exist down to the Process Cell Level Opportunity – The application has a strong hierarchal model that supports Site, Area and Process Cell for Manufacturing Operations and Has Built in designations for each Warehouse  

ERP/MRP Hierarchy

Area

Bulk Manufacturing

Filling

Product Storage Areas

Process Cell

Each Bulk Line

Each Fill Line

Each Warehouse

Conclusions! With the implementation of the S88/S95 model into these applications: Users can now ask these systems a host of critical questions that affect the day to day operations of manufacturing plants As the model gets implemented in each new system, the effort required to decide the best model implementation gets easier and easier for systems like documentation, training and change control With the implementation of the S88/S95 model into these applications: Even without an in-place MES, these enhancements increases the value of and utility of these applications and with this structure in place, downstream MES implementation will go far smoother Acknowledgements: Don Ellwanger – Manager, Metrology Services SME on CMMS Tricia Vail – QC Microbiology Analyst II and SME on Microbiology Data Tracking Jinna Penachio – Manager, Quality Information Services and SME on Discrepancy/CAPA Tracking Jennifer Strong – Manager, Materials Planning and SME on the ERP/MRP Inventory Control Modules Doug Brenner – Information & Technology Manager for Superior Controls and SME on SCADA Integration Tom Harvey – Vice-President of IT (and My Supervisor) for his support throughout this project