Integrated Strategic Planning: New Horizons for ...€¦ · Integrated Strategic Planning...

30 PHARMACEUTICAL ENGINEERING • JANUARY/FEBRUARY 2001

Integrated Strategic Planning

Integrated Strategic Planning:New Horizons for Infrastructure,Image, Function, and Business onthe Pharmaceutical Campus

Integrated Strategic Planning:New Horizons for Infrastructure,Image, Function, and Business onthe Pharmaceutical Campus

M

by Gary D. Anderson, PhD, and David M. Jonik, PE

This articledescribes SiteMaster Planningin the context ofcorporatestrategicplanning. Thevariousdimensions ofphysical planningare seen asmeans ofadvancingcorporate goals.Moreover, thesuccessful plan isone thatintegratesfacilities, process,site design, andinfrastructure inrecognition oftheir interrelatedimpacts onachievingstrategic goals.

Overview: The Relationship of theBusiness Plan, Master Planning and

Site Operations

Master planning is a reflection of corpo-rate vision. The success of any masterplan is dependent upon a planner’s

ability to translate the business plan, as devel-oped by corporate management, into a planningvision to guide future physical development.Physical development has many dimensions –from the purely functional to the purely aes-thetic. A good plan uses all of these dimensionsas tools to advance corporate goals. Therefore,master planning on pharmaceutical campusesneeds to address facilities, plant processes, sitelayout and design, and infrastructure. Site de-sign and architecture can enhance image; func-tional relationships – whether in facilities orprocess layout – can promote efficiencies ofoperation, and infrastructure planning can helpto realize economies in the procurement of en-ergy and utility services and in waste handling.To best utilize the site, and to derive optimalbenefit from all of these components, each mustbe developed in conjunction with the others. Bytaking an integrative, strategic approach to allaspects of planning, the best results can beobtained.

What do we mean by integrative? Figure 1shows an integrative approach in relating theoverall planning process to the ultimate out-come of site design and operations. The left sideof the diagram represents the various compo-nents of the planning process. The diagrambegins on the left-hand side with the corporatebusiness plan. The business plan sets the over-all direction for the company and may addresseverything from corporate image to productrange and marketing. It is one of the maindeterminants of requirements relative to pro-cesses and facilities. These requirements are

identified and satisfied through facility plan-ning and process planning, indicated by the twoboxes to the right of the corporate business plan.In turn, the results of facilities planning andprocess planning help to determine infrastruc-ture requirements, which are addressed throughinfrastructure planning, the next box to theright. (The infrastructure planning process isshown in more detail in Figure 2, and will bediscussed further below.) Together, the threetypes of planning (process, facility, and infra-structure) are encompassed by Site Master Plan-ning.

An important outcome of the overall sitemaster planning process is the degree of successwith which the operation of the site fulfillsbusiness plan objectives. Site operation is showndiagrammatically on the right side of the dia-gram (and is portrayed in more detail in Figure3).

An integrative process recognizes that allthree of the site planning activities – facility,process, and infrastructure – are interrelated.All contribute to fulfilling the objectives of thecorporate business plan, and all affect the ulti-mate success of site operations. None possessesabsolute primacy over the others. For this rea-son, they should be considered concurrently.

Many large pharmaceutical companies haveestablished campuses that meet the needs ofadministration, research, and manufacturingthrough physical design and planning. Due bothto the specialized nature of company productsand processes, and the importance of the public’sperception of the company, pharmaceutical com-panies must present a modern, well-organizedand inviting image not only in their internalfacilities, but also in their overall outward ap-pearance. In other words, a successful facilitymust address aesthetic as well as functionalissues.

Reprinted from The Official Journal of ISPE

PHARMACEUTICAL ENGINEERING® January/February, 2001 Vol. 21 No. 1

©Copyright ISPE 2001

JANUARY/FEBRUARY 2001 • PHARMACEUTICAL ENGINEERING 31


Problems Associated withEvolution and Change

Sometimes, however, the past history of a pharmaceutical sitemakes this objective difficult to achieve. The configuration ofa pharmaceutical company’s campus is often the result ofdecades of sequential growth and adaptation to ever-changingcommercial, technological, organizational, and regulatory en-vironments. The result of this continual adaptation presentsproblems not only for master planning, but also for day-to-daytechnical issues which are driven by corporate policy, supplyparameters, regulatory issues, and other factors.

Facilities and elements of utility systems that were notdesigned to serve the exact functions they are called on toperform today exist side-by-side with state-of-the-art ele-ments, some of which depend on antiquated services for sup-port. The configuration and spatial relationships among re-lated or unrelated facilities and utilities are likely quitedifferent than how they would be designed today to servepurpose-built facilities.

This situation is not unusual. It is the natural outcome forany successfully self-sustaining organization with a respectfor the careful use of resources and a standing commitment tothe community. On the other hand, this situation does make itmore challenging to improve, campus-wide, the overall ap-pearance of the campus, as well as the configuration of pro-cesses and the complex infrastructure that has evolved tosupport both facilities and processes.

Moreover, change never ends. Just as the evolution ofcurrent infrastructure often results from past needs for adapt-ing to change, any renovated infrastructure also will likely becalled upon to support a continually changing campus in thefuture.

Applying the Strategic Planning Processto Site Master Planning

What is the role of strategic planning in site master planning?One challenge is that – just as the evolution of current facilitiesand infrastructure on a campus may have resulted from acontinual need to adapt to change – there is every reason tobelieve that renovated or new development will likewise becalled upon to support a continually changing campus. Toaccommodate and support these changes, strategic planningprovides an on-going process, in contrast to a “one-shot” devel-opment plan. Although the creation of development plansrepresents important events in the strategic process, any givendevelopment plan is likely to have a limited useful life span,simply because the context in which it was created is subject tochange. Strategic planning continually monitors the factorsthat should have an impact on a successful plan, and makesadjustments when these factors change. What are these vari-able factors? They include the corporate business plan, whichis itself strategic in nature. They include the state of technol-ogy, the cost of operations, and environmental legislation thatdefines acceptable levels of environmental impact. They in-clude, especially in a deregulated economic environment, thecost of purchasing energy and utility services. In the case ofimage or aesthetics, they may even include fashion.

Figure 2 represents the Strategic Planning Process. Thisprocess could be applied to the overall site master planningprocess or to any aspect of it (process, facilities, circulation,utilities), but here it has been specially adapted to reflectinfrastructure planning. Even as shown, it is applicable tovirtually any utility or infrastructure system. By itself, the

column of boxes in the center of the diagram represents aperfectly acceptable approach to developing a “one-shot” plan.Generally, the actions represented by the boxes proceed from (1)gaining an understanding of the thing to be planned and theexternal parameters that affect it, through (2) developingalternative planning and design responses, and (3) evaluatingthose alternatives relative to cost and other pre-establishedobjectives (such as might be derived from the corporate busi-ness plan), culminating in (4) the selection and development ofa preferred alternative.

There are several features, however, that distinguish thisstrategic planning process from the “one-shot” approach. Mostof these features are represented by activities in boxes thatdiverge from the central column. First, near the lower third ofthe column are two activities that interact: Develop AlternativeDesign Responses and Develop Alternative Operating and Man-agement Responses. This reflects the recognition that planningis more than merely the manipulation of physical features.Strategic master planning recognizes that there is interplaybetween operations and design and that there are often possi-bilities to adjust both sides of the equation. For example, achange in the way a function is performed may obviate the needfor a physical planning or design change. If adjusting thefunction is not detrimental to its operation or outcome, this canmean a savings in cost achieved by foregoing new constructionor renovation. Strategic site master planning should not acceptcurrent procedures as givens.

Another feature that distinguishes strategic planning isillustrated in the diagram by several of the activities that areshown to the left of the central activity column. In general, thesecollateral activities represent a recognition of the impacts offactors that are completely exogenous to the system beingplanned. In some cases, these factors are even located off-siteand may include market conditions or environmental exigen-cies as noted above.

Yet another characteristic that distinguishes the strategicapproach from the one-shot approach is demonstrated by thefeedback loops which are shown in the diagram. These operateat various levels. As shown in the diagram, one such feedbackloop demonstrates how the determination of costs can lead toa re-evaluation of alternatives. (The arrow from the Cost boxgoes against the general flow of the other activities and returnsto the Evaluate Alternatives box above it.) In fact, this is only oneof several such feedback loops that operate throughout anyplanning process: decisions are continually re-evaluated inlight of new information that is developed as planning activi-ties proceed.

Of utmost importance is a more macro-scale feedback loopthat leads from the final activity box (Select Alternative andRefine Recommendations) back to near the top of the process(Analyze System Operation). This arrow represents the con-tinual monitoring of parameters as mentioned above. If thesystem or the external parameters that affect it have notchanged, revisiting the process should not yield any new plan-ning needs. On the other hand, a change that occurs anywherealong the process may indicate the need for planning action,depending on how that change is interpreted and acted upon inthe numerous evaluation steps. Instituting an iterative ap-proach, such as this, helps to ensure that site design andoperation are fine-tuned to the environmental, cultural, andbusiness contexts in which they exist.

As mentioned above, this strategic process can apply to anyaspect of site master planning. Take planning for image and




aesthetics: Is post-modern architecture “out” anddeconstructionist “in”? Maybe it’s time to re-evaluate yourimage relative to the cost of updating it – taking into account,of course, exogenous system elements such as the State HistoricPreservation Officer (SHPO) and the design guidelines in affectwithin the local planning jurisdiction.

Special Problems and Solutions Related toInfrastructure Planning: Planning-Level

ModelingInfrastructure represents a special challenge in the contextstrategic planning. That is the nearly unmanageable degree ofdetail required to examine each special case and to proposesolutions to every shortcoming involving an energy system orinfrastructure across an entire campus. Monitoring these com-plex systems is complicated enough, but what happens whenchanging parameters suggest a change in design or operation?Evaluating the effects of proposed adjustments and improve-ments not only on the system itself, but on all of the othersystems addressed by site master planning, can border on theimpossible, especially on large, multi-functional campuses. Forcampus-wide planning purposes, one means of dealing with thiscomplexity is to limit the level of concern to the study of majorelements of infrastructure systems - those lines that connect

facilities or major lines that feed secondaries from on- or off-sitesources. Even at this level, however, the number of elements andlinkages and the ranges of operating parameters for all of thepertinent systems is daunting. For this reason, it is helpful toformulate models of the systems to evaluate how proposedimprovements will affect their performance under differentplanning assumptions.

Figure 1 shows such a model in very broad outline under thePlant Design and Operation side of the diagram (the right-handside). In this diagram, the Infrastructure Planning activity, fromthe left-hand, Planning, side of the diagram feeds directly intothe rectangle of a similar color labeled Infrastructure Design andOperation, on the right-hand Design and Operation side. Belowthe Infrastructure Design and Operation box is a wide arrowpointing downward and labeled On-Site Processes and Con-sumption. The concept here is that site infrastructure supportson-site processes, whether they be manufacturing or other typesof consumption (e.g. fuel consumed in heating or cooling anadministration building). Superimposed over these two sys-tems is an arrow labeled Essential Services. Essential Servicesis in fact a component of the overall bundle of infrastructuresystems, but it carries out the specialized function of controllingand orchestrating the other infrastructure systems, even to thepoint – in well planned and engineered sites – of controlling

Continued on page 34.Figure 1. Multi-system, campus-wide infrastructure model - business context.




Figure 2. Strategic planning process for a typical infrastructure system.




some functions within processes and facilities.Feeding into these three on-site components (Infrastructure,

Processes, and Essential Services) are Supply Parameters,indicated by the broad arrow at the top of the column. SupplyParameters comprise the cost and availability of energy andutility services, as well as the actual operation and configura-tion of off-site supply systems and the manner in which theyinterface physically with on-site systems. Flowing from On-SiteProcesses and Consumption are Waste Parameters, shown in thediagram by the broad arrow. Waste from on-site processes andconsumption includes waste water, sewerage and run-off, solidwaste, heat, noise, particulate matter, and numerous chemicaldischarges. Most waste eventually returns to the Environment.The Essential Services function (or arrow in the diagram), inaddition to overlying and penetrating the Operation and Con-sumption functions, spans from the Supply Parameters to theWaste Parameters. It is this control system that can adjust day-to-day or minute-to-minute operations and can adjust to chang-ing supply and waste requirements or opportunities in order toimprove on-site efficiencies and economies relative to the all-encompassing, off-site macro-context. It can be seen from thissimple diagram that Essential Services plays a critical role inmaintaining the sustainability of the campus.

Figure 3 represents a conceptual model to describe in moredetail, but still greatly simplified, the various systems thatcomprise an Infrastructure Master Plan for a pharmaceuticalcampus. The diagram is arranged along the same lines and

reflects the same general relationships as those in the Designand Operation side of Figure 1. In Figure 3 Off-Site Systems(Providers and Regulators) are located at the top of the diagramand correspond to Supply Parameters in Figure 1. In a strategicplanning process, these systems are continually monitored inorder to adjust in a timely manner to changes that may repre-sent opportunities or problems (e.g. changes in price structuredue to deregulation). As in Figure 1, the lower portion of Figure3 represents consumption, waste, and environment. In Figure3, the central horizontal band represents the On-Site Infrastruc-ture Systems. On pharmaceutical campuses, this is a technicalarea which can influence and be influenced by integrated stra-tegic planning. The horizontal band cuts across three verticalcolumns which represent three main categories of infrastruc-ture systems: Essential Services, Utility Supply Systems, andWaste Removal Systems.

In Figure 3, all of the infrastructure systems are shownconcurrently. This simple model shows that many of thesystems are interrelated even across the lines of the majorcategories. The implication is that external influences on onesystem may very likely have wide-reaching impacts on othersas well.

Integrated strategic planning, as applied to site infrastruc-ture planning, focuses on the white boxes in the diagram thatappear on the band labeled On-Site Systems, as well as on thelines that connect the various activities, processes, and orga-nizations represented by the other boxes above and below the

Figure 3. Conceptual model - multi-system, campus-wide model.




On-Site Systems band. What goes on inside the shaded boxes isusually not subject to planning recommendations developedthrough the process of campus master planning, and must beaccepted largely as given, but variable. The contents of thosewill determine the parameters within which the on-site infra-structure systems must operate. These parts of the model,external to the systems themselves, can be treated as “blackboxes.” (Of course, recommendations concerning further studyof these “black boxes” may be included in the master plan if itbecomes clear that any of them has an impact on the infrastruc-ture that might be improved with adjustments to internaloperations.)

Summary: The Benefits of Integrated StrategicMaster Planning for Pharmaceutical Campuses

How does a concern for integrated and strategic master plan-ning affect what we normally think of as site master planning?An integrated strategic master plan will present the pharma-ceutical company with opportunities derived from the macro-environment which can be capitalized upon with the assis-tance of operational models. These models may be formulatedfrom detailed information, but are simplified in order tohighlight important relationships rather than minute elemen-tal descriptions. Using these approaches, the total masterplanning effort will address facilities, siting issues, interrela-tionships with the surrounding community, transportation,site material, vehicular and pedestrian circulation, and thesolutions to many of the utility-related problems, some ofwhich arise through the gradual evolution of the campus.

The result will be a site master plan that not only deals withaesthetics and basic functional layout, but which also can:

• improve operational efficiency• ensure balanced development• set the stage for further development• capitalize on the changing macro-context rather than fall

victim to it• enhance sustainability• advance the objectives of the corporate business plan

An Illustrative ExampleSTV Incorporated’s master plan for a Belvidere, NJ, vitaminproduction facility provides a real-world example of the inte-grative strategic approach to pharmaceutical campus plan-ning. Activities at the site included administration, research,production, and distribution. STV was asked to develop aconventional physical campus plan for the 500± acre site on theDelaware River.

However, before beginning to explore opportunities for thephysical arrangement of facilities on the site, STV assessed theutility infrastructure at the macro level. Our appraisal of theinterrelationships among power and steam generation, con-sumption, and the impact of deregulation led to the recognitionthat a new cogeneration plant might be brought on line. Also,our analysis of environmental regulations and required com-pliance at the plant led to recommendations for a differentapproach to stormwater management procedures, calling forlimited containment and treatment of “first flush” stormwater.This ultimately saved capital funds that would otherwise havebeen required to increase the wastewater treatment facility,while still meeting strict environmental guidelines.

It was only after these basic functional issues were ad-dressed that the more traditional components of campus mas-

ter planning proceeded – with an analysis of functional relation-ships among activities, project growth in various organiza-tional units, and attention to the image that managementwished to address at the site. Melded into these considerationswere significant new spatial determinants – the potentialcogeneration plant and the land required for stormwater man-agement – which would not have come to light without integrat-ing considerations that fell outside the normal concern of aphysical campus plan. The result was a physical developmentplan that was more sustainable in that it would be less vulner-able to major changes resulting from inevitable engineeringdevelopments.

References1. Binder, S. 1992. Strategic Corporate Facilities Manage-

ment. New York: McGraw-Hill, 1992.

2. Bumble, S. 2000. Computer Simulated Plant Design forWaste Minimization/Pollution Prevention. Boca Raton, FL:Lewis Publishers, 2000.

3. Burnham, J. M. 1994. Integrative Facilities Management.Burr Ridge, Ill.: Irwin Professional Pub., 1994.

4. Christ, C. (ed.) 1999. Production-Integrated Environmen-tal Protection and Waste Management in the ChemicalIndustry. New York: Wiley-VCH, 1999.

5. Cole, G. 1998. Pharmaceutical Production Facilities: De-sign and Applications. London: Taylor & Francis, 1998.

6. Dober, R. P. 1996. Campus Planning. Ann Arbor: SCUP,1996.

7. Froment, G. F. (ed.) 1979. Large Chemical Plants: EfficientEnergy Utilization, Plant Design and Analysis, Processes,Feedstocks; Proceedings of the 4th International Symposiumon Large Chemical Plants. Amsterdam; New York: ElsevierScientific Pub. Co., 1979.

Presentation by B. Frank, "Energy Saving in ChemicalProcesses."

8. Goumain, P. 1989. High Technology Workplaces: Integrat-ing Technology, Management, and Design for ProductiveWork Environments. New York: Van Nostrand Reinhold,1989.

9. Ingalls, L. 1999. Scenario planning for effective facilitiesdecisions. FM Data Monthly (online edition); April 1999.

10. Mecklenburgh, J. C. 1973. Plant Layout; A Guide to theLayout of Process Plant and Sites. Aylesbury: L. Hill (forthe Institution of Chemical Engineers; Great Britain),1973.

11. O’Mara, M. A. 1999. Strategy and Place: Managing Corpo-rate Real Estate and Facilities for Competitive Advantage.New York: Free Press, 1999.

12. Reeve, J. R. and M. B. Smith 1995. Planning for MasterPlanning. Alexandria, Va.: Association of Higher EducationFacilities Officers, 1995.


13. Schmertz, M. F. 1972. Campus Planning and Design. NewYork: McGraw-Hill, 1972.

14. Sievert, R. W. 1998. Total Productive Facilities Management:A Comprehensive Program to Achieve Business Goals byOptimizing Facility Resources, Implement Best Practicesthrough Benchmarking, Evaluation and Project Manage-ment, Increase Your Value to the Organization. Kingston,Mass.: R.S. Means Co., 1998.

15. Turpin, J. R. 1998. Staying afloat while the current shifts.Engineered Systems Vol. 15, No. 12, pp. 28–30; December1998.

16. Wrenall, W. and Lee, Q. (eds) 1994. Handbook of Commer-cial and Industrial Facilities Management. New York:McGraw-Hill, 1994.

About the AuthorsGary D. Anderson is a registered architect and a member ofthe American Institute of Certified Planners. He received a BAdegree and Master of Urban Design degree from the Universityof Southern California. He was awarded a Diplom in SocialScience from the University of Stockholm (Sweden), and a PhDfrom the Department of Geography and Environmental Engi-neering, Whiting School of Engineering, The Johns HopkinsUniversity. He has held positions in several architecture,planning, and engineering firms, including Gruen Associates(Los Angeles), RTKL Associates (Baltimore), and STV Incorpo-

rated (Baltimore). He also has held positions in the publicsector, including the Department of Community Developmentin Prince George’s County, Maryland and the Office of Planningin the Maryland State University system. Dr. Anderson hastaught and lectured in architecture and planning at severaluniversities, both in the United States and abroad, and he iscurrently an adjunct faculty member at The Johns HopkinsUniversity, Berman Institute of Real Estate, where he teaches“Design Issues in Development.” He holds the position ofprogram director on the executive committee of the BaltimoreDistrict Council of the Urban Land Institute, and he is amember of the board of directors of 1000 Friends of Maryland,a managed-growth advocacy group. Dr. Anderson currentlyholds the position of Vice President and Director of Planning atSTV Incorporated.

STV Inc., 21 Governors Ct., Baltimore, MD 21217.

David M. Jonik, PE holds a BS and an MS in chemicalengineering from Villanova University, Villanova, PA. He servesan array of science and technology clients in industry andgovernment and has prepared program implementation strat-egies for DOE, Roche Vitamins, Merck, and others. His currentposition is Vice President of UAI Associates, Inc., in Reading,PA. His primary responsibility is the continued development ofUAI’s biotechnology, pharmaceutical, and microelectronics ini-tiatives. Jonik has competed in many long distance swimmingevents raising money for the March of Dimes and Alzheimerresearch, and he has presented career development seminars atseveral local schools.

UAI Associates, 6 Commerce Dr., Reading, PA 19607.



Capability Study

Application of a Capability Studyto a Syringe Filling OperationPart 2 of 2: Protocol Execution, Data Collection,Analysis and Discussion

Application of a Capability Studyto a Syringe Filling OperationPart 2 of 2: Protocol Execution, Data Collection,Analysis and Discussion

P

In Part 1 of thisarticle, wediscussed theutility ofcapability studiesin factoryacceptancetesting,validationtesting, anddocumentationto meet bothproduction andregulatoryrequirements.We closed Part 1with a preparedprotocol. In Part2, we execute theprotocol, verifymodel validity,analyze the data,and report on thefindings. In doingso, we illustratehow data arecollected,transformed intoinformationabout theequipment orprocess, and thisinformation itselfinto productiveknowledge.

by Peter A. Hugunin

Figure 1. Sampling andserpentine fill pattern.

On-Site Adjustments

Prior to execution of the machine capabil-ity study all installation and most op-erational qualification attachments were

executed. Hose type and identification was re-corded; pump type and serial numbers wererecorded; information on fill needles verified ortaken; the machine and trays leveled; the op-eration of the machine checked, and the indi-vidual syringes weighed and tagged (see scatterdiagram in Appendix 1 for information on sy-ringe variability). In spite of these and otherpreparations, some field adjustments were nec-essary to make the testing more meaningfuland successful. These changes, most naturally,resulted first from a discussion with the plannedequipment operator (the day before departure);

and next from the kick-off meeting with the OEMengineers and technicians just minutes beforestarting data collection. Three field modifica-tions resulted:

First: The manufacturer’s operator had iden-tified break time as meaningful to the process,and although a process as opposed to a machinevariable — it was a manual operation for theprocess under study — it was considered veryimportant and useful information. Conse-quently, during execution we introduced a pro-cess variable: Time between loading of Tray(TL). We decided to vary this time between TL1and TL2 to test what affect it might have on fillvolume and to remove these samples from fur-ther statistical analysis as necessary.

Second: The OEM technicians and engineersinformed us that wecould not operate at topspeed. Speed must beset at something otherthan maximum. Re-portedly it could notbe guaranteed that thefill needles would befully retracted fromthe syringes beforemovement began. Thismight result in brokenor bent fill needles.Also, a problem withvibration might resultat top speed given thesmall fill volume wehad selected for test-ing. Rather than riskboth damaging theequipment and abort-ing the study it wasdecided to operate atsome lower setting anddiscuss the issue ofthroughput at some fu-ture date.





Capability Study

Third: This machine has two fill lines each with separatehoses, pump, and needle. In two machine strokes, four syringesare filled, and although the data entry table was designed toaccommodate two fill lines, the serpentine filling pattern, right-left-right movement, was not anticipated - Figure 1. This mademaintenance of the time-ordered sequence as well as plannedsample size somewhat difficult, but not impossible to hold. Wehad not planned for the serpentine pattern of fill in setting upour original data entry table, consequently we set up anintermediate table for data entry (which was kept on file) andthen the final table - Appendix 2. This complicated not onlydata entry, but the pre-weighing and tagging of the syringes aswell.

Due to the variation in individual syringes each was weightedprior to fill and after fill. Rubber gloves were worn during datacollection at all times and forceps were used in the handling ofall syringes in our samples. Filling was performed in anuncontrolled industrial environment at the equipmentmanufacture’s site in Schwabish Hall, Germany.

Data AnalysisTime-Ordered Sequence of all data and OutliersThe pattern of outliers observed in the line graph of allindividual observation reveals the effect time between loading(TL) had on fill volume and the critical importance of control-ling this process variable - Figure 2. As noted above, this

process variable has no bearing upon the machine capability asloading of the trays is a manual operation. Thus the followingoriginal data rows were eliminated from any further statisticalanalysis: 7, 13, 19, and 25.

When the plotted data were shown to other members of theexecution team, on the day following data collection, two teammembers (one from the vendor and one from the owner)informed the remainder of our group of an additional, un-planned, operator adjustment which occurred after the datacollection had already started. This “adjustment” resulted intwo data rows containing observations far outside the toler-ance limits and more than six standard deviations from themean. These points were eliminated from further statisticalanalysis as outliers with a known cause unrelated to themachine capability. Thus, original data rows 1, 2, and 3 wereeliminated from further statistical analysis leaving the re-vised data table (Appendix 2) for final data analysis of themachine capability with 23 rows of data as opposed to theoriginally planned 30.

A visual examination of this first line graph already sug-gested some difference between the line-pump-needle configu-rations - Figure 2.

Goodness-of-Fit TestingWith the final data set decided upon our next step was to viewthe empirical data distribution and form a judgement on model

Figure 2. Line graph (all data).



Capability Study

fit, i.e. was the analytical approach valid, and if not, whatadjustments if any were to be made. The empirical datadistribution may be viewed in a frequency histogram or acumulative frequency diagram with K-S type confidence bandsbuilt about it.1 The author finds the frequency histogram moreuseful from a visual and graphical perspective. A visual exami-nation of the frequency histogram suggests again the differ-

=ence between Line 1 and Line 2 and suggests, given that x is0.20068 ml that the combined data are something other thannormally distributed, e.g. right skewed suggesting a chi-squareddistribution - Figure 3. It is not required that the data matchthe model distribution perfectly.1,2 (See discussion of modelvalidity below.)

In this project, all data analysis was performed using anExcel spreadsheet and available statistical tables. The as-sumption of normal distribution was tested using the chi-squared analysis of residuals method described by Conover.3

This method permits the estimation of population parametersfrom the sample data with a loss of one degree of freedom foreach parameter estimated. Contingency table sizes of 4 and 8were built with the following results:

The overall fill population (line 1 plus line 2) is not normallydistributed. Line 2 is normally distributed but line 1 is not.Line 2 is skewed, as is line 1; however, in testing for goodnessof fit using the Chi-Squared test at both classifications of 4 and8 the null hypothesis (population is normally distributed) isaccepted.

The Issue of Sample Size and Data “Grouping”With the addition of the process variable TL the subgroups(S1,S2,S3, S4,S5) of sample size ng=4 was reduced from (ri ) (cj) =24 to (ri ) (cj) = 20. The calculated size for the subgroup analysiswas 22, but was based upon an estimated population varianceof 9.0-3; however, since the actual population variation was atleast 10 times less (8.9-4), samples of size 20 proved adequate.With subgroup S1, where data rows 1, 2, and 3 where dropped,we were not so fortunate and consequently data from S1 wereonly included in the grouped analysis - Appendix 3 and Appen-dix 4.

An initial review of these data indicated an apparentdifference in line-pump-needle configurations which was laterconfirmed with a simple sign test from which we concludedwith more than 99.99% confident that there is a differencebetween the two lines - Figure 3. This difference also isapparent from the 99% CIs built about their respective means- Appendix 4. As a result of this analysis we further grouped thedata based upon one sample of size 92, and two samples eachof size 46 to conduct additional analysis. Given the cleardifference between line 1 and line 2, what possible justificationcould there be for grouping the two lines together? First, eachof the two lines must be individually capable of operating withinthe specified tolerances; and second, the noted statistical varia-tion is not an economic or ethical issue as both “lines” are wellwithin the tolerance limits specified.

Appendix 1. Scatter diagram and analysis of syringe weight data.



Capability Study

Figure 3. Line graph (revised data).

population distribution? More than 99% of the samples fallbetween –1.22 and + 2.05 Std. Dev. from the mean on thenormal distribution curve, and it is fully 10.5 Std. Dev. to theUTL from the mean and 6.11 Std. Dev. to 0.2061 ml. For thesereasons, there are no serious concerns. Thus, seen the normaldistribution is an elegant, simple, and adequate model of ourmachine’s operation.

An additional prospective may be gained by viewing the datain light of Tscheyscheff’s Theorem. From line one’s center (0.2004ml) to the Upper Tolerance Limit (UTL ,0.2100 ml), it isapproximately 9.6 Std. Deviations from the empirical distribu-tion - Figure 4. No matter what the distribution no more than1.1% should fall outside this limit on the high end and 0.9% onthe low end. And, from line two’s center (0.2010 ml) to the UpperTolerance Limit (ULT , 0.2100 ml), it is approximately 15 Std.Deviations. No matter what the distribution no more than 0.4%should fall outside this limit on the high end and 0.3% on the lowend. No matter what the distribution no more than 1% will beoutside the tolerance limits.

The Machine CapabilityGiven the nature of the machine design (two needles/two fill perstroke) and the population distribution, the author felt someadjustment toward the conservative side of error was in order.From sample to sample, our point estimates of the population

To maintain a meaningful and valid analysis we varied fromthe test protocol as follows:

1. The short term machine capability as defined, and indicatedon the data entry table, would not be calculated for subgroupone (S1).

2. The long term machine capability, defined as production ofat least 12,000 units would be calculated based upon thenew rc matrix. r = 24, c = 4; and, for each line separately.

upon the set-point nor was the process normally distributed,consequently an additional factor of conservative estimationwas added in that the author prefers the use of kSm as opposed_to either 3R/d2 or 3Sm in calculating the Cmk - Appendix 3 andAppendix 4.

Subgroup S1 was validly eliminated from a separate shortterm, small sample, analysis because of its size — only 12 —however, it could not by any means be eliminated from otherstatistical analysis since, unlike the outliers which whereclearly attributed to known events not related to the machine’snormal operation, there was no explanation for r3c1’s variationfrom the mean. This data point was, without explanation, atleast five standard deviations from the mean and hence not achance occurrence. It deserves some consideration if not inves-tigation. Cmk compensates for this by looking at both sides. A



Capability Study

Figure 4. Frequency histogram (revised data): empirical data distribution.

The validation team must not permit itself to become slaveto the servant, but instead consider all relevant issues as theyarise, whenever they arise, and remain flexible to the highestextent possible. Thus viewed the qualification/validation teamis an open system responding and adopting to a living environ-ment. Good managers, validation engineers, and teams under-stand, accept, and act accordingly.

Comments on Line Difference, Height of Bulk Fill Fluid,and Machine Set-pointThe lack of adequate fill in both lines after introduction of theprocess variable TL was attributed primarily to “back-siphon-ing.” This was tested ad-hoc, no data recorded, and foundsubstantially true and might therefore be improved or perhapseliminated through changing the height of the bulk fill containerrelative to the filling needle heads. Kinks in lines may have beenanother contributing factor to this major difference betweenlines themselves at these observed points.

As stated above from the application of a simple sign test,we are able to conclude with more than 99.99% confidence thatthere is a difference between the two lines. Such a differencealso was suggested by a review of the data (Figures 2 and 3),and confidence intervals built about the line means - Appendix4. In short, the population is not homogenous, and assignablecause is at work here. It may be the result of natural variationin the inside diameter of the fill needles, some other machinevariable, or even design, but it is neither of economic norethical concern to the process or product of this study.

very cautious estimate of the Cmk for this machine and setpointis: Cmk ≥ 2.

Closing Discussion and Limitations of ProtocolGains and Losses from Field Modifications to PlanA test protocol is a guideline and tool to assist in the equipment(or process) testing and the generation of a document packageto support manufacturing and regulatory requirements. It is asubstantial document having many official signatures andthere will always be some apprehension when making fieldmodifications to approved test plans. An experienced teamleader or validation engineer will discuss proposed changeswith other team members most familiar with the technicalaspects and implications first and next consider the organiza-tional and regulatory aspects prior to making any changes. Hadour team not responded adequately to the environment (factualand judgmental information) the time-ordered sequence ofsampling would have been lost; equipment damage would likelyhave resulted in a significant delay if not cancellation of thetesting altogether, and information important to the processand use of the equipment may not have been captured in a timelymanner. With respect to meeting the protocol’s stated objec-tives, nothing was lost. Even the fact that the maximum speedwas not tested is not a failure as this is better discussed withthe equipment manufacturer as a separate item, and if need be,either the maximum setting changed, or additional testingperformed.



Capability Study

Note, that both fill lines were somewhat off the target of 0.20ml suggesting the possibility that the machine was set on datafrom one of the lines only; therefore, one possible improvementinvolving absolutely no costs would be to set the target basedupon data from both. Again, this has no negative effect upon themachine’s ability to produce within the tolerance limits.

Setting In-Process Control LimitsAlthough setting in-process control limits was not a statedobjective or acceptance criteria in our test plan (see the testprotocol in Part 1 of this two part article), it may be seen as alogical conclusion to our data analysis since such control limitswed well with promises of the qualification/validation philoso-phy of reduced inspection costs and improved control5 and maybe based upon the collected data. In moving from capabilitytesting to in-process control, there is a shift in the statisticalanalysis from the population to the sample distribution accom-plished without losing or sacrificing the requirement that allindividual syringe fills fall within the previously defined toler-ance limits. The objective of in-process control is to identify andcorrect assignable causes of variation, which may, if left un-checked, lead to product non-conformance. In-process controllimits are based upon the sampling distribution which willalways be normally distributed. These control limits should notbe too narrow. We do not want manufacturing and maintenancegroups over-responding to “ghost data” or tampering with a

Subsample ID

(Syringe ID)Orig

inal

row

nr.

Rev

ised

row

nr.

c1:

Line

1

c2:

Line

2

c3:

Line

1

c4:

Line

2

Xba

rr

Xm

ax

Xm

in

Rr

r4 r1 0.1997 0.2002 0.1997 0.2005 Subsample 1 0.2000 0.2005 0.1997 0.0008r5 r2 0.1998 0.2003 0.1997 0.2008 S1 0.2002 0.2008 0.1997 0.0011r6 r3 0.2061 0.2008 0.2003 0.2010 0001 - 0024 0.2021 0.2061 0.2003 0.0058

r8 r4 0.1999 0.2005 0.1999 0.2003 0.2002 0.2005 0.1999 0.0006r9 r5 0.1998 0.2006 0.2007 0.2024 0.2009 0.2024 0.1998 0.0026r10 r6 0.2012 0.2020 0.2003 0.2021 Subsample 2 0.2014 0.2021 0.2003 0.0018r11 r7 0.2001 0.2018 0.2000 0.2019 S2 0.2010 0.2019 0.2000 0.0019r12 r8 0.2019 0.2025 0.2022 0.2024 3001 - 3024 0.2023 0.2025 0.2019 0.0006




Notes:=

1. In both Analysis of Data Summary Tables (Appendix 3 and Appendix 4) x has been used as the estimator for population mean (µ).

2. Observation r3c1 = 0.2061 and is 5.73 to 6.12 standard deviations form the mean depending upon whether one uses the grouped or theline data. It appears clear that this is the only point of concern. It seems not to be a random occurrence. Even when we consider that thepopulation is chi-square distributed and not normal… in fact it becomes of increased concern.

Appendix 2. Data summary table: (r23

c4: n

g = 4).

functional system. Nor do we want these limits too wide result-ing in an untimely response to actual shifts in the machine’soperational characteristics.

In addition to the shift in distribution functions, there is anecessary shift from the machine to the process capability. Inour case, this shift may only be accomplished with certainassumptions and we therefore introduce the concept of hypo-thetical process capability into our analysis. If we assume thatprocess variables such as TL, raw materials, and others (Fig-ure 1, Part 1: Partial Fishbone) are under control or do notcontribute to variation in fill, then the calculated Cmk may beviewed as the process capability and in-process control limitsbuilt accordingly.1

For a product with substantially the same viscosity, density,and setpoint/fill volume as our test fluid, these limits may beset as follows:6,7,8

1. Since we have variable data on a ratio scale we recommend_ _building an x-Chart and R-Chart. The x-Chart will detectchanges in the machine’s aim and the R-Chart will pick upproblems with individual syringes.

2. The most economic and appropriate basis for setting theselimits is information from ng = 2 (Appendix 4) and theempirical data - Figure 4. We recognize that considerationof operational and process aspects such as the number of



Capability Study

= _ _ = _ =n x R Sm R/d2 NPL = x ± 3 R/d2 k NPL = x ± k Sm Cmk Notes

_3(Sm) 3(R/d2) k Sm

_Grouped 92 0.20068 0.00124 8.864-4 6.018-4 ±1.805-3 3.1 ±2.748-3 3.51 5.16 3.39 Sm > R/d2

full ricj 0.1989 to 0.2025 0.1980 to 0.2035matrix

S1 12 — — — — — — — — — — Sample sizetoo small forthis analysis

_S2 20 0.20113 0.00150 9.989-4 7.285-4 ±2.19-3 4.17 ±4.17-3 2.96 4.06 2.13 Sm > R/d2

0.1989 to 0.2033 0.1969 to 0.2053_

S3 20 0.20077 0.00088 3.813-4 4.274-4 ±1.282-3 4.17 ±1.59-3 8.07 7.20 5.81 Sm > R/d2

0.1994 to 0.2021 0.1992 to 0.2024_

S4 20 0.20014 0.00074 3.468-4 3.594-4 ±1.078-3 4.17 ±1.45-3 9.48 9.15 6.82 Sm < R/d2

0.1990 to 0.2012 0.1986 to 0.2015_

S5 20 0.20066 0.00104 4.418-4 5.051-4 ±1.515-3 4.17 ±1.84-3 7.05 6.16 5.07 Sm < R/d2

0.1991 to 0.2022 0.1988 to 0.2025

0.20043 < µ.99 < 0.20093 (based upon n=92)Notes:

_ = _1. Sx = 9.241-5. It is Sm2/sqr root of n. x ± 2.66 (sx). T995,60 used to calculate 99% C.I. on µ. T-Distribution used instead of Normal for two

reasons: a. pop. is not normally distributed._

2. This operation is slightly off center. As a consequence the Cmk is a somewhat better summary statistic than Cm.. Sm ≅ R/d2 and is anunbiased estimate of the population variance.

3. ng = 4, d2 = 2.059, A2 = 0.729, D4 = 2.282, and D3 = 0(8)._

4. In each case kSm is a more conservative unbiased estimator than 3 R/d2. Vendor software packages sometimes use Sm, without anyadjustments, to calculate the CpK.

5. Building of Control Limits (ng = 4):_ = _

_ = _ _ _ _ _UCLx = x + A2R and LCLx = x + A2R UCLR = D4R and LCLR = D3R

= 0.20068 + (0.729)(0.00124) = 0.20068 - (0.729)(0.00124) = 2.282(0.00124) = (0) (0.00124)= 0.20158 = 0.19978 = 2.2897-3 = 0

Appendix 3. Analysis of data summary table: (r23c4: ng = 4).

= _ _ = _ =n x R Sm R/d2 NPL = x ± 3 R/d2 k NPL = x ± k Sm Cmk Notes

_3(Sm) 3(R/d2) k Sm

_Line 1 46 0.20037 1.0-3 — ±3.500-3 3.21 4.22 2.75 Sm > R/d2

0.1969 to 0.20390.20068 0.00086 7.61-4 3.5 _

Line 2 46 0.20100 6.0-4 — ±2.100-3 5.00 3.94 4.29 Sm < R/d2

0.1989 to 0.2031

Line 1: 0.20022 < µ.99 < 0.20052Line 2: 0.20091 < µ.99 < 0.20109

Notes:_ _1. sx1 = 1.474-4 sx2 = 8.847-5

2. ng = 2, d2 = 1.128, A2 = 1.88, D4 = 3.268, and D3 = 0(8).

3. Building of Control Limits (ng = 2):_ = _

_ = _ _ _ _ _UCLx = x + A2R and LCLx = x - A2R UCLR = D4R and LCLR = D3R

= 0.20068 + (1.88)(0.00086) = 0.20068 - (1.88)(0.00086) = 3.268(0.00086) = (0) (0.00086)= 0.20230 = 0.19906 = 2.81-3 = 0

4. A2, D4, and D3 are constants for using the average range to find control limits for subgroup averages and subgroup ranges (8).

See Part 1 of this article for additional terms and abbreviations.

Appendix 4. Analysis of data summary table: (r46c2: ng = 2).



Capability Study

undetected bad units the operation can afford prior to correc-tion or implications of break-time may drive management tomore frequent sampling until additional information isobtained and confidence can be built.

3. Sampling for in-process control must be rational, i.e. occur-ring at planned intervals with time-ordered-sequence main-tained. Samples of size 2, one from each of the “lines,”should be drawn about every 10,000 units.

_ _4. The calculated limits are: UCLx = 0.20230 ml and UCLx =0.19906 ml. And UCLR = 0.00281 ml - Appendix 4. Statisti-cally, it is a long way from these control limits to thetolerance limits. This is a very favorable position for manu-facturing.

5. One may or may not employ narrow-limit gauging,8-9 e.g.any individual observation greater than 0.2042 ml (morethan +4 std. Dev. from mean) or less than 0.1995 ml (morethan - 2 Std. Dev. from mean) should trigger an investiga-tion.

ConclusionsWe have seen in this case problem that assignable causes tovariation may be, and undoubtedly frequently are present,which are of neither economic nor of ethical concern. We haveadditionally seen that the model assumption of normal distri-bution may be, and undoubtedly frequently is, violated with-out invalidating its use to the actual application.

From a theoretical and practical perspective, we have tiedthe purchase and sales agreement as well as manufacturingspecifications, through the capability study design, execution,and analysis to the mission of validation as well as to thefulfillment of manufacturing and regulatory affairs require-ments. And, finally, we have explained and demonstrated aclear and meaningful distinction between machine and processcapability as well as between short term and long term studies.

References1. Conover, W.J., “Statistics of the Kolmogorov-Smirnov Type,”

Practical Nonparametric Statistics. 2nd Edition NY NewYork: John Wiley & Sons, 1980, pp. 344 - 385.

2. Wheeler, Donald J., and David S. Chambers, “The Whys andWherefores of Control Charts,” Understanding StatisticalProcess Control. 2nd Edition Knoxville, Tenn.: SPC Press Inc.,1992, pp.55 - 88.

3. Conover, W.J., “Contingency Tables,” Practical Nonpara-metric Statistics. 2nd Edition NY New York: John Wiley &Sons, 1980, pp. 143 - 208.

4. Juran, J.M. and Frank M. Gryna Jr., “Manufacture – Statis-tical Aids,” Quality Planning and Analysis (From ProductDevelopment through Use). 2nd Edition, NY New York: McGraw-Hill Publishing Company, 1980, pp.283-313.

5. FDA, Guideline on General Principles of Process Validation,May 1987, p.6

6. Ott, Ellis R., and Edward G. Schilling, “On ImplementingStatistical Process Control,” Process Quality Control, 2nd.Edition, New York NY: McGraw-Hill, Inc., 1990, pp. 171 -219.

7. Ott, Ellis R., and Edward G. Schilling, “Ideas from TimeSequences of Observations,” Process Quality Control, 2nd.Edition, New York NY: McGraw-Hill, Inc., 1990, pp. 44 - 77.

8. Juran, J.M. and Frank M. Gryna Jr., Appendix Table I_(Factors for x and R Control Charts; factors for estimatings and R), Quality Planning and Analysis (From ProductDevelopment through Use). 2nd Edition, NY New York: McGraw-Hill Publishing Company, 1980, p. 611.

9. Ott, Ellis R., and Edward G. Schilling, “Narrow-LimitGauging in Process Control,” Process Quality Control, 2nd.Edition, New York NY: McGraw-Hill, Inc., 1990, pp. 154 -170.

About the AuthorPeter A. Hugunin is currently a Validation Engineer atWatson Laboratories in Corona, CA. He has held field serviceengineering and sales engineering positions prior to enteringa validation career. His validation experience has ranged fromdesigning and implementing quality systems and master vali-dation plans, to qualification as well as validation work atseveral cGMP facilities. He holds a BS from SUNY Plattsburgand an MBA from Syracuse University. He has been a memberof various professional organizations including ASQC, PDA,and ISPE.

Watson Laboratories, 311 Bonnie Circle, Corona, CA91720.



Batch Management

Building the Perfect BatchBuilding the Perfect Batch

Tby William N. Gracely, PE, Alan Karner, and Richard E. Parapar

This articleaddresses oneapproach toimplement theS88 standardsoftwaredevelopmentmethodology forbuilding a batchcontrol systemallowing users torealize manyadditionalbenefits byapplying thisstandard.

Introduction

T he ISA S88.01 Standard for Batch Con-trol provides an excellent model forimplementing enhanced control capa-

bilities in industrial applications. Process op-erations and equipment functionality are clearlyrepresented as configurable recipes, phases,units, and control modules in the control sys-tem. This architectural structure supports ahigh level of capability and flexibility. Plantequipment becomes more responsive to bothinternal and external influences such as abnor-mal process conditions, operator commands,equipment alarms, material substitutions, andproduction changes.

Many Batch Management packages providethe necessary tools to create and execute recipeprocedures. The process of recipe assembly firstrequires that a process model be defined whichdescribes the capabilities of the process equip-ment. These characteristics subsequently dic-

Figure 1. Components of batchmanagement and control logiclayers.

tate the assembly of recipes in terms of units,phases, and associated parameters. In this way,Batch Management provides an importantimplementation layer of the S88 model.

But in a highly automated manufacturingfacility, the Batch Management layer is liter-ally just the tip of the iceberg. Full implementa-tion of an S88-based control system involvesdesigning, programming, and testing hundreds,if not thousands, of distinct software compo-nents - Figure 1.

Unfortunately, definition of the Batch Man-agement process model does little to advancedevelopment of units, phases, and control mod-ules in the control system. Often, the complex-ity of implementing the S88 architecture pre-sents a significant engineering challenge forthe inexperienced control system developer.

This article discusses a development meth-odology that addresses many of the difficultiesassociated with implementing S88-based con-

trol applications. Itwas first successfullyapplied in 1996 for de-velopment of Genen-tech’s Clinical Manu-facturing Facility(CMF) in South SanFrancisco, California.This experience dem-onstrated that by ap-plying a well-struc-tured software devel-opment approach,batch control systemscan be built in a costeffective manner withhigher quality andlower risk.

CMF ProjectOverview

Genentech first ap-plied the S(P)88 batchmodel for the design ofits manufacturing con-trol systems in early1993. In the three sub-sequent years, dozensof PLC-based, unit-centric S88 control sys-





Batch Management

tems were successfully developed for chromatography, centrifu-gal, and Tangential Flow Filtration (TFF) protein recoveryapplications.

Early in 1996, the decision was made that the CMF controlapplication also would be based on the S88 Standard and thatall major process operations in the facility would be recipedriven. This would be the company’s first attempt at applyingS88 for a large-scale multi-product, multi-train application.

CMF Process ApplicationThe new plant consisted of six fully instrumented fermenters(80L, 400L, 2kL, and 12kL scales) and three TFF mediatransfer skids (400L, 2kL, and 12kL scales). It included fullyintegrated Clean-In-Place (CIP) and Steam-In-Place (SIP)systems capable of servicing any vessel or transfer line in theplant - Figure 2.

The CMF Application required batch recipe procedures tobe developed for the following automated process operations:

• CIP and SIP of all fixed tanks, portable tanks, and transferlines

• Sanitization of TFF transfer units• Fermentation media batching• Cell culture growth operations (e.g., inoculation, feeds,

perturbations, sampling)• TFF media exchange, straight pressure, and solera media

transfers• Cell product harvest at both the 2kL and 12kL fermenter

vessel scales

More than 47 separate processing units were identified in thefacility’s equipment infrastructure. They required more than

200 phase logic and 300 control modules to support the processrequirements of the facility. Each of these components neededto have detailed specifications written. Control logic had to beprogrammed and extensively tested. Detailed process graphicand operator interface displays were necessary to supportoperator supervision and manual operation of the plant. Testprocedures had to be prepared so that subsequent projectactivities such as integrated testing, plant start-up, and vali-dation could begin.

CMF System ArchitectureThe CMF control system was based on an open distributedcontrol system (DCS) platform with integration of severalPLCs. Steam temperature RTD monitoring was implementedin the PLCs and fully integrated with the control logic execut-ing in the DCS. The RBATCH Batch Management packagewas used for recipe management and execution. Data gener-ated by the control system was collected and archived by a datahistorian application running on a separate, dedicated server.Information from the historian allowed users to evaluate theoperating characteristics of the facility’s process operationsand equipment, as well as overall product quality.

CMF Development RequirementsPreliminary estimates indicated that more than 20 man-yearswould be required for designing, developing, and testing theCMF control software. Not surprisingly, this greatly exceededthe project schedule, which allowed only 12 months for theimplementation of the control system. The engineering teamrealized it needed to try and shorten the development cycle bycreating a process modeling tool that could capitalize on thehighly modular structure of the S88 architecture. It was

Figure 2. CMF process architecture.



Batch Management

Figure 3. Genentech clinical manufacturing facility - project database components.

believed that using this tool would result in more efficientdevelopment and, hopefully, deployment of the control soft-ware. The major requirements of the new process modelingtool were to:

• Assist the design team in defining the functional character-istics of the CMF process requirements in terms of S88-based components. The tool would help manage the S88architecture’s inherent complexity by tracking the manyunits, phases, and control modules, as well as the relation-ships between these objects in the CMF control application.

• Provide a ‘single point of entry’ for process modeling andconfiguration data. The information would be verified atthe point of entry, automatically updated, archived, andpresented in the appropriate format to the various projectpersonnel requiring access (instrument technicians, con-trol programmers, Quality Assurance test personnel).Change control would efficiently ensure that users alwaysreferenced the most current information.

• Facilitate automatic generation of test procedures andreports. Standardized test forms would be developed that,when populated with data from the process model, wouldhelp streamline formal system testing and validation.

• Generate the thousands of tag specifications needed toconfigure the data historian’s scanner interface to the DCS.This would allow the data historian’s batch model to accu-rately track the DCS control application without extensiveamounts of redundant (and error prone) data entry.

• Support the ability to automatically configure and distrib-ute control logic in the DCS. Process model data, includingreferences to phase logic and control module templates inmaster libraries, would enable the DCS Control Configura-tion application to propagate control logic to the DCScontrollers.

This last requirement was clearly the most ambitious to fulfill.If successful, it would significantly reduce the amount of timeneeded for software development. The “hands-off” distributionof control logic throughout the DCS would also greatly enhancethe overall quality of the finished system by eliminating manyopportunities for programming, copy and paste, and data entryerrors.

CMF Process Model DatabaseThe CMF engineers determined that a PC-based relationaldatabase application, like Claris FileMaker Pro or MicrosoftAccess, would satisfy the requirements for the process model-ing tool. The “Process Model Database” would serve as thecentral S88 software design repository. When integrated withDCS configuration application, it also would support auto-matic creation of the DCS control logic - Figure 3.

CMF Process ModelAn important activity in the design of the CMF Process Modelwas identifying common process requirements and equipmentcharacteristics in the plant. From these common requirementsand characteristics, detailed specifications for typical softwaremodules were written representing entire classes of opera-tions and equipment in the process model. For example, the



Batch Management

Figure 4. Control module typical tags.

CMT Tag Description

CMF01001 Fermenter dO2 ControllerCMF01002 Fermenter pH ControllerCMF01003 Fermenter Agitator ControllerCMF01004 Fermenter Overlay Air Flow ControllerCMF01005 Valve Supervisor (16 valve max)CMF01006 Fermenter Backpressure ControllerCMF01007 Addition Pump w/ calculated feed totalCMF01008 Fermenter Temperature ControllerCMF01009 SIP Temperature/Pressure ControllerCMF01010 Fermenter Volume IndicatorCMF01011 Analog Input IndicatorCMF01012 Proximity Jumper Supervisor (16 max)CMF01013 Discrete Valve ControllerCMF01014 TFF Membrane TMP CalculationCMF01015 TFF Feed Flow ControllerCMF01016 TFF Filtrate Pump ControllerCMF01017 Analog Indicator w/ Alarm DelayCMF01018 TFF Recycle Tank Agitator ControllerCMF01019 TFF Recycle Tank Level ControllerCMF01020 TFF Recycle/CIP Pressure ControllerCMF01021 TFF Recycle/CIP Temp ControllerCMF01022 TFF Chemical Dist Pump ControllerCMF01023 Media Supply ControllerCMF01024 Discrete Device w/ HOA StationCMF01025 Harvest Filter dP Control ModuleCMF01026 CIP Recirc Tank Level ControllerCMF01027 DIW Tank Level ControllerCMF01028 CIP Supply Temperature ControllerCMF01029 CIP Supply Flow/Pressure ControllerCMF01030 CIP Chemical Makeup ControllerCMF01031 Discrete Switch IndicatorCMF01032 ESTOP MonitorCMF01033 Plant Powerloss MonitorCMF01034 Analog SelectorCMF01035 Plant Alarm AnnunciatorCMF01703 Unit/Equipment Module Supervisor

PLT Tag Description

PLF01002 CIP4 Skid Server SequencePLF01004 Fermenter Class CIP SequencePLF01005 Fermenter Pneumatic TestPLF01007 Fermenter SIP SequencePLF01009 Fermenter Antifoam TransferPLF01010 Fermenter Addition TransferPLF01011 Fermenter Set EnvironmentPLF01012 Fermenter OUR PerturbationPLF01013 Fermenter Addition MonitorPLF01014 Fermenter Sample MonitorPLF01015 Fermenter Culture MonitorPLF01018C 2kL/12kL Fermenter InoculationPLF01022A Fermenter Heat Kill – FillPLF01022B Fermenter Heat Kill – HeatPLF01022C Fermenter Heat Kill – DrainPLF01026A Fermenter Cool Filter (w/ Integrity Test)PLF01026B TFF Cool Filter (w/ Integrity Test)PLF01027 Fermenter Media BatchPLF01030A Fermenter A Line Flush Media FilterPLF01032 TFF Transfer Line CIP SequencePLF01034 TFF DrainPLF01035 TFF Single Pass FlushPLF01038 TFF Membrane Pressure Hold TestPLF01039 TFF SIP SequencePLF01041 TFF Media Exchange TransferPLF01042 TFF Base Clean SequencePLF01043 TFF Acid Clean SequencePLF01045 TFF Straight TransferPLF01047 TFF PBS Flush (Harvest)PLF01048 TFF Harvest TransferPLF01052 TFF Tank PW RinsePLF01053 TFF Tank Flush ConnectPLF01054 TFF Tank Flush DisconnectPLF01055 TFF Tank CIP SequencePLF01056 Harvest Filter CIP SequencePLF01061 Message Board Prompt PhasePLF01062 Transfer Panel Jumper SetupPLF01100 Resource Allocation/Deallocation

Figure 5. Phase logic typical tags.

control logic in a Pump Control Module served as a templatefrom which all similar instances of pump control logic in theplant were reproduced.

While many of the major equipment units in the CMF weredetermined to be similar, if not identical, the decision wasmade to only develop “typicals” for phases and control mod-ules. The relatively few number of unit instances, compared tophases and control modules, would not result in a significantsavings in the programming effort. A decision to support UnitTypicals may be made at some future date. For the CMFProject, Unit Instances were defined directly in terms of PhaseLogic and Control Modules Instances, and by association theirtypicals.

The CMF Application was normalized into its base func-tional components yielding 59 distinct Phase Logic Typicalsand 42 Control Module Typicals out of the 244 and 312 totalinstances, respectively.

CMF Control ModulesControl Modules offer an object-oriented interface to I/O andthe means by which signal conditioning and alarm monitoringare accomplished, as well as the user interface for enteringlocal setpoints or running devices in manual. For example,every valve is issued open/close commands through its controlmodule interface. Although the plant employs both air-to-open

and air-to-close valves, operators don’t need to consider whetherto ‘force on’ or ‘force off’ the valve output signal when exercisingmanual control.

CMF Phase LogicOne of the core design principles of the CMF Project was thebelief that recipes should provide the primary interface forprocess scientists and engineers to configure and executeprocess operations via the control system. The intent was tomake plant users more responsible for knowing how to buildand execute their own recipes. Phase logic in the controlsystem, therefore, serves as one of the fundamental buildingblocks process scientists and engineers use to build recipeprocedures in the Batch Management package.

If recipes are to be user configurable, then recipe phasesmust be defined at a level that makes sense from a processperspective. Phases should represent, in of themselves, minorprocess operations in the facility. Examples of process-ori-ented phases include “Fermenter SIP,” “Fermenter PneumaticTest,” and “TFF Harvest Transfer.” These phases are definedto be broad in scope, and include the complete sequence ofequipment control required to accomplish significant process-ing steps. Several advantages are realized using this phasedesign approach.



Batch Management

Figure 7. Recipe tags.

Recipe Tag Description

RCP01001 400L/2kL/12kL Fermenter Media Batch RecipeRCP01002 80L Fermenter Media Batch RecipeRCP01003 400L/2kL/12kL Fermenter Media Line CIP RecipeRCP01005 400L/2kL/12kL Fermenter Vessel CIP RecipeRCP01006 80L Fermenter Vessel CIP RecipeRCP01009 Portable Tank CIP RecipeRCP01010 TFF Recycle Tank CIP RecipeRCP01012 2kL/12kL Fermentation Process RecipeRCP01014 80L/400L Fermentation Process RecipeRCP01015 400L/2kLTFF Skid Sanitization RecipeRCP01016 2kL/12kL Fermentation Vessel SIP RecipeRCP01017 80L/400L Fermentation Vessel SIP RecipeRCP01018 2kL TFF Harvest Transfer RecipeRCP01019 400L/2kL TFF Media Exchange Transfer RecipeRCP01020 400L/2kL TFF Straight Transfer Recipe

• For recipe configuration, larger process phases appear morefamiliar and are easier to configure for users. Many equip-ment and operation details don’t need to be made visible atthe recipe level. For example, the “Fermenter Media Batch”phase allows the user to specify only the major processparameters associated with the batching of a fermentervessel (e.g., media selection, temperature, transfer volume)without having to assemble the segments of the mediatransfer sequence using many small phases. Control con-figuration parameters are implemented at each of thecontrol module, phase logic, and unit levels, but are notmade visible at the recipe level unless there is an explicitrequirement to change the value from one recipe executionto the next. This approach minimizes the number of recipeparameters and makes phases easier to configure by theuser.

• More sophisticated holding and restart logic can be imple-mented within a larger phase. Combining many smallphases in the recipe is more difficult, where limitations inthe SFC programming language become problematic. Alarger phase, responsible for all the activities in a majorprocess step, is thoroughly familiar with the Equipmentand Control Modules that are used in the step. It knows thecorrect start-up and shutdown sequence, and is bettersuited for implementing sophisticated exception and hold-ing logic than at the recipe level.

For the CMF application, the implementation of large phases,representing complete segments of major process operations,resulted in smaller recipes. These recipes were easier for usersto configure in the Batch Management package, while allow-ing control programmers to implement more sophisticatedcontrol sequences in the controllers.

CMF RecipesThe implementation of large, class-based phases naturallyleads to class-based recipes. By developing recipe templateswith large phases that describe only the major process steps,

the same recipe template can be applied to a wide range ofequipment sizes (e.g., 80L, 400L, 2kL, and 12kL).

For example, a single fermenter class SIP recipe success-fully addresses the sanitizing requirements of all four 2 kL and12 kL fermenters in the plant. The same master recipe is usedfor each vessel, and is made up of only three phase typicals -Figure 6.

From the user’s perspective, the SIP recipe is very easy toconfigure and execute because of its class-based design. TheSIP Phase Logic in each unit is pre-configured with the param-eter values specific to each equipment instance. The recipeensures that the SIP phases execute in the proper order. Thisgreatly reduces the overall number of recipes required for theCMF Application - Figure 7.

Recipes made up of fewer, larger phases have a significantadvantage in that the majority of the process sequencing logicexecutes in the controller, and not in the Batch Managementpackage. The Structured Function Chart (SFC) representationof recipes is often misapplied for implementing detailed se-quence programming at the recipe level. This means thatrecipe execution by the Batch Management package, whichtypically resides outside the controller, is responsible forcoordination of phase starts/stops in a synchronous manner. Inany case, interlocks between concurrent phases must be imple-mented to ensure this happens correctly. Also, larger phasesresiding in the controller can be executed in manual mode atthe Unit level, outside the context of a recipe. This is advanta-geous during system start-up, as well as during normal opera-tions.

CMF Units and Equipment ModulesThe CMF Process Model recognizes the existence of both Unitsand Equipment Modules. Equipment Modules are an impor-tant layer of the S88 model because they allow units to sharecommon resources with other units. As a result, Units aresmaller and easier to develop. They become more open to class-based normalization and more reproducible in the processmodel.

Equipment Modules are functionally equivalent to Unitsthat have only Control Modules and no Phases. This constructallows greater flexibility of equipment grouping in the plant.Control Modules are bundled into a single use, shared equip-ment resource that can be acquired as needed by Units.

At the recipe’s direction, a Phase in the Unit books an

Figure 6. Fermenter SIP recipe example.



Batch Management

Equipment Module for the same batch procedure as the Unit isbooked for. This is accomplished in conjunction with the BatchManagement package where equipment resource managementis performed.

Once the Equipment Module is successfully acquired, PhaseLogic in the active Unit can manipulate the Control Modulesin the remote Equipment Module. The Equipment Moduleprovides a physical extension of the unit supervision logic inthe Unit - Figure 8.

In the CMF Application, a common control logic templatewas applied for all instances of Units and Equipment Modulesin the process model. The Unit/Equipment Module changes itscharacteristics depending on whether it is acquired by BatchManagement for recipe phase execution, or by another Unitrequesting access to its Control Modules. As an EquipmentModule, it provides a valuable service to the acquiring Unit bycoordinating command functions and alarm notification be-tween its Control Modules and the Unit’s Phase Logic.

An example in the CMF Application of a dual-use Unit/Equipment Module is the TFF Recycle Tank. During mediatransfer operations between fermenters, the tank functions asa subordinate Equipment Module to the TFF Media Exchange

Unit. Later, after being released by the transfer operation, thetank is acquired as a Unit for the CIP Recipe and executes CIPPhase Logic to perform its cleaning sequence - Figure 9.

CMF Process Model DatabaseThe CMF Process Model Database was constructed to serve asthe central S88 software design repository and support auto-matic creation of the DCS control logic. The overall schema ofthe Process Model Database is shown in Figure 10.

The first step in creating the database was to define the datalayout, or schema, that identifies the significant objects tomanage. Based on our knowledge of S88, we knew that objects

Figure 8. Equipment module acquisition.

Figure 9. Unit/equipment module - TFF recycle tank.

Unit Tag Description Function

CIP4 CIP System UNITF1228 2kL Harvest Filters EMT1215 400L Fermenter UNITT1215A 400L Fermenter Media Line UNIT/EMT1215E 400L Fermenter Inocaulum Line EMT1215JM 400L Fermenter Harvest & Recycle Lines EMT1215N 400L Fermenter CIP/SIP Header EMT1228 2kL TFF Recycle Tank UNIT/EMU1221 400L TFF Media Exchange Skid UNITX1225 TFF Acid Distribution Subsystem EMX12kL 12kL Media Transfer Subsystem EM

Figure 10. CMF process model database schema.



Batch Management

Figure 12. Example CM typical to CM instance propagation.

Figure 11. Relationship between control module instances and control moduletypicals.

such as units, phases, control modules, and I/O signals wouldbe needed. As the schema in Figure 10 illustrates, UnitInstances are composed of Phase Logic and Control ModuleInstances.

I/O Instances and ReferencesI/O signals are represented in the database by objects called“I/O Instance” objects. Every signal to/from a device or piece ofequipment is described using attributes that describe itsfunction and logically associate it with a specific ControlModule Instance (CMI).

I/O information includes the Control Processor (CP) andField Bus Module (FBM) through which the signal is accessed,a detailed description of the signal or device, the high and lowsignal range, engineering units, and in the case of discreteoutputs, whether the signal should be inverted (as for air-to-close devices). Other attributes describe the cabinet, rack, andI/O card where the point is connected to the control system.With this information, detailed wiring tables and calibrationtest forms are created to help ensure that all signals areproperly landed and configured.

In some cases, it is necessary for more than one controlmodule to read the data value of an input signal. An exampleof this is a pressure transmitter that multiple pressure control-



Batch Management

Instances are derived. A relationship similar to that betweencontrol modules and Control Module Parameters applies forboth Phase Logic Instances (PLI) and Phase Logic Typicals(PLT). Phase parameter objects are called PLI Parameters andPLT Parameters, respectively. Again, default values are de-fined in the library template, and then modified in the instanceas necessary.

Recipe TypicalsRecipe Typicals represent the set of phases required to conductprocess operations in the plant, like sanitization of a mediatransfer unit or batching of a fermenter vessel. Recipe typicalscorrespond to Master Recipes in the S88 model. It is possibleto define the characteristics of recipes with Phase Typicals. Forprocess modeling purposes, it is sufficient to view recipes asconsisting of only two parts: the header and the procedure. Therecipe’s header identifies the master recipe by name, descrip-tion, version number/date, and related information. To keepthings simple, the procedure is made up of only phases and notoperations. A recipe’s procedure can be described by a list of thePhase Typicals referenced in the recipe. This interpretation ofrecipes, while limited, supports the indication in the recipe ofchanges to component phases or associated parameters.

Data objects called “Recipe Elements” are analogous to PLTElements and CMT Elements. They describe the relationshipbetween a Recipe Typical and a specific Phase Typical that therecipe expects to execute in the unit - Figure 13. This associationallows the Process Model Database to identify PLT Parametersthat have their visibility set at the “recipe” level. A list of theparameter configuration requirements can be constructed foreach master recipe.

CMF ResultsUse of the CMF Process Model Database to design and configurethe control application was instrumental in the successfulcompletion of the CMF Project. It took approximately twomonths to set up and populate the database with informationabout I/O signal instances and the unit/phase/control modulebreakdown of the CMF process model. An additional month wasneeded to integrate export data from the database with the DCSconfiguration application. Once the database’s ASCII-basedexport files were properly formatted, data transfer to theconfiguration application worked seamlessly and reliablythroughout the project lifecycle. The control software itselfrequired approximately 12 months of detailed Phase Logic andControl Module design and programming.

The final software was extensively tested and verified to bebug free, a significant accomplishment in light of the complexityof the application. The DCS/PLC control system proved capableof executing sophisticated process operations with minimalinteraction with plant personnel. Plant operators were consis-

lers read for their process variable. Since the I/O Instance canonly “belong” to one Control Module, a data object called an“I/O Reference” is defined. An I/O Reference represents theconnection from a Control Module to an I/O Instance belongingto another Control Module. Only input signals are linked byreferences.

Control Module Typicals, Elementsand Parameters

Given that I/O Instances are the components of Control ModulesInstances, an equivalent object is required to represent thecomponents of Control Module Typicals (CMTs). These arecalled the “Elements” of a Control Module Typical. Each CMTElement defines a functional signal interface for a ControlModule Typical that an I/O signal will fulfill in a Control ModuleInstance. In the case of the Pump Control Module, the discreteinput signal indicating that the pump is active is representedby the “Pump Running” Element in the Pump Control ModuleTypical. Either an I/O Instance or I/O Reference can be associ-ated with a CMT Element.

In addition to I/O, Control Modules require configurationparameters to define the runtime behavior of the control logic.Continuing the example, a Pump Control Module has a param-eter that defines how long the control logic should wait afterstarting the pump before enabling its “Loss of Running” alarm.A default value for the parameter is defined in the ControlModule Typical, but can be adjusted for each Control ModuleInstance to filter out nuisance alarms. These parameters asso-ciated with Control Module Instances (CMI) and Control Mod-ule Typicals (CMT) are called CMI Parameters and CMTParameters, respectively - Figure 11.

The ability to associate I/O signals and parameter valueswith the functional elements of template-based logic is thefoundation for automatically populating controllers with fullyconfigured control logic. Information in the Process Model Da-tabase is exported to the DCS configuration application thatmaps the I/O signal designations and control parameter valuesonto templates residing in its software library. This allows theconfiguration application to automatically create control mod-ule instances in the designated controllers - Figure 12.

Phase Logic TypicalsPhase Logic Typicals are the templates from which Phase

Figure 13. Relationship between recipe typicals and phase logic typicals.

Figure 14. CMF software development schedule.



Batch Management

Editor's Note

This paper was originally presented at the 2000 World BatchForum, Atlantic City, NJ. It is reproduced/presented here byarrangement with the World Batch Forum which holds thecopyright.

World Batch Forum may be contacted at [email protected].

tently impressed with the capability and reliability of the newcontrol system and pre-licensing ‘mock’ runs resulted in excel-lent product yields for the plant scientists.

Several project goals were achieved using the Process ModelDatabase as a tool with which to develop the S88-based controlapplication.

Compression of theSoftware Development Cycle

Because of S88’s inherent modular structure, and the ability ofthe Process Model Database to effectively track all the softwarecomponents, it was possible to significantly compress thesoftware development schedule by performing developmenttasks in a tightly staggered, parallel manner - Figure 14.

Control system developers were assigned to different soft-ware engineering teams responsible for developing specifica-tions, programming control modules and phase logic, buildingHMI displays, simulation and wet testing, project documenta-tion, and systems management. Their efforts were coordinatedthrough the Process Model Database and well documented,object-oriented interfaces minimized programming interac-tions between the different groups.

Increased Responsiveness to Changesin the Requirements

The automated software development mechanism proved veryresponsive to late process and equipment modifications to theplant’s design. In cases where equipment had to be added at thelast minute, it was possible to quickly add the instance defini-tion to the Process Model Database and automatically generatenew control logic from library typicals.

In other situations, the functionality of existing ControlModules needed to be augmented (e.g., implementing a newcontrol algorithm for all Dissolved Oxygen Controllers). ControlModule Typical logic was able to be quickly modified, tested,and redistributed to all instances of the Control Module in theplant. By maintaining the instance-specific configuration pa-rameters in the database, the updated Control Module In-stances were deployed and fully functional with no manualprogramming required. This helped to ensure that the qualifi-cation status of tested modules was never compromised.

Improve Efficiency of Validation ProcessComprehensive document and version control ensured thatvaluable time and resources were not wasted at any point inthe project schedule. Development and test personnel relied onthe Process Model Database to ensure they always workedwith the most current software designs.

The test and validation effort was greatly accelerated.Testing of I/O installation began while Control Module designwas not yet complete, and before Phase Logic design evenstarted. Once I/O was verified, testing continued for ControlModules, then Phase Logic, and finally entire Recipe proce-dures. Automatic generation of test forms permitted the quali-fication effort to closely track design and development. Effec-tive change management and version control was the key factorin allowing software testing to overlap the design and develop-ment efforts for the project.

ConclusionImplementation of the process modeling tool, which capital-ized on the S88 Batch Standard’s modular architecture, provedto be instrumental in creating a flexible, capable control

system of high quality, in a cost efficient manner, while mini-mizing the overall risk of project failure.

References1. ANSI/ISA-88-1995- Batch Control Part 1: Models and Ter-

minology, American National Standards Institute/Instru-mentation, Systems and Automation Society.

About the AuthorsWilliam N. Gracely, PE, has worked for The Foxboro Com-pany since 1981. Prior to this, he worked for five years in salesat Taylor Instruments and two years as a design engineer at anew photographic film manufacturing facility with Polychrome.His responsibilities at Foxboro have included work in interna-tional industry sales, corporate engineering liaison, food pro-cessing applications specialist, and his current assignment asa principal technical consultant based in San Ramon, CA. Hisinvolvement in batch control began in 1983 as a member of thedevelopment team for Foxboro’s Easybatch product. He hasworked on batch projects in the biotechnology, pharmaceuti-cal, food, and chemical industries. Gracely is a registeredControl Systems Engineer in the state of California.

The Foxboro Co., 2000 Crow Canyon Pl., San Ramon, CA94583.

Alan Karner is an Automation Manager - Standards andProductivity Tools - with APV, Rosemont, IL, and is a graduateof Rensselaer Polytechnic Institute (Troy, NY) with a BS inmechanical engineering. Karner’s experience includes servingas an engineering officer in the US Navy nuclear submarineservice. He subsequently spent more than 10 years with TheFoxboro Co. as an automation training developer and principalapplications engineer, where he developed custom controlapplications for batch industry clients. He is currently respon-sible for design, development, and implementation of automa-tion productivity tools that produce high quality, reusablesolutions for the food and pharmaceutical industries.

APV, Bristol Park, Foxboro, MA 02035.

Richard Parapar is a Principal Systems Engineer withGenentech, Inc., S. San Francisco, CA., and is a graduate ofRensselaer Polytechnic Institute (Troy, NY) with a BS incomputer and systems engineering. As a Systems Architect inOperations Systems, he is currently involved in the integra-tion of Genentech’s ERP, MES, EDMS, LIMS, and DCS appli-cations to assemble a comprehensive, synergistic batch manu-facturing environment for Manufacturing, Quality Control,and Quality Assurance. Parapar has more than 20 years ofexperience designing and implementing advanced automationsystems for the biotech, food processing, chemical/refining,and high-purity gas industries.

Genentech Inc., 1 DNA Way, South San Francisco, CA94080.



Pharmaceutical Terminology

I

Applied Terminology for thePharmaceutical IndustryApplied Terminology for thePharmaceutical Industryby Michelle M. Gonzalez

Introduction

In a world where expediency in the communication of ideas and concepts is the ever soconsuming issue, it has become necessary to

identify organizations, procedures, techniques,and many other daily qualifiers not by theircomplete descriptions, but rather by their acro-nyms. Increasingly, these abbreviations are partof our every day endeavors regardless of whatfield of work we may be involved in. The prob-lem is that we find ourselves either hearing orusing a particular combination of letters thathave a precise meaning, but we do not know themeaning or we may have forgotten it. Findingthese meanings may be sometimes challengingsince they are usually referenced on a verylimited basis and in many different places.

In the Biotechnology and Pharmaceuticalindustries this particular issue becomes morecomplicated due to the globalization of proce-dures and regulations in addition to the dailyscientific advances involving research, testing,new drug production, new treatments for dis-eases, and advanced means of drug application.

This acronym list is not meant to be a uniqueor complete guide useful to every individualinvolved in this industry, but rather as a “livingdocument” that will be periodically updated toprovide a handy reference to scientists, engi-neers, designers, technicians, owners, contrac-tors and other individuals as they apply to thisenergetic and still evolving industry. Commentsand/or suggestions for widening the scope ofthis document will be welcome, and should bedirected to the author or the ISPE Communica-tions Department.

Section I - Abbreviations and Acronyms

- A -AAAS American Association for the Ad-

vancement of ScienceAABC Associated Air Balance CouncilAAALAC American Association for the Ac-

creditation of Laboratory AnimalCare

AAMA American Architectural Manufac-turers Association

AAMI Association for the Advancement ofMedical Instrumentation

AAPS American Association of Pharma-ceutical Scientists

AASHTO American Association of State High-way & Transportation Officials

ABC Association of Biotechnology Com-panies

ABPI Association of British Pharmaceu-tical Industries

ABRF Association of Biomolecular Re-source Facilities

ABS Acrylonitrile Butadiene StyreneACDP Advisory Committee on Dangerous

Pathogens (United Kingdom)ACGIH American Conference of Govern-

mental Industrial HygienistsACGM Advisory Committee on Genetic

Manipulation (United Kingdom)ACI American Concrete InstituteACI Alloy Casting InstituteACIL American Council of Independent

LaboratoriesACP Acyl Carrier ProteinACPA American Concrete Pipe Associa-

tionACS American Chemical SocietyACTH Adrenocorticotropic HormoneADA Americans with Disabilities ActADAMHA Alcohol, Drug Abuse, and Mental

Health AdministrationADC Air Diffusion CouncilADME Absorption, Distribution, Metabo-

lism, and EliminationADP Adenosine DiophosphateADR Adverse Drug ReactionADSE Atmospheric Dust Spot Efficiency

(Filter test)ADSL Asymmetric Digital Subscriber LineAFM Atomic Force MicroscopyAGA American Gas AssociationAGS American Glovebox SocietyAHU Air Handling UnitAIA American Institute of ArchitectsA.I.A. American Insurance AssociationAIChE American Institute of Chemical

EngineersAIDS Acquired Immune Deficiency Syn-

dromeAISC American Institute of Steel Con-

structionAISI American Iron and Steel InstituteAITC American Institute of Timber Con-

structionALSC American Lumber Standards Com-

mittee

These acronymswere compiled tohelp engineersunderstand basicterminology as itapplies to therelatively newBiotechnologyindustry, as wellas the moreestablishedPharmaceuticalindustry.

The completeglossary will beavailable onISPE’s Web site.It will includedefinitions ofterms used inbiology,chemistry,HVAC,manufacturingprocesses,medicine,materials,metallurgy,regulatoryconcerns, watertreatment, andwelding.






ALUS Automatic Loading Unloading SystemAMCA Air Movement and Control AssociationANDA Abbreviated New Drug ApplicationANSI American National Standards InstituteAOAC The Association of Official Analytical ChemistsAPA American Plywood AssociationAPCI Atmospheric Pressure Chemical IonizationAPI Active Pharmaceutical IngredientAPI American Petroleum InstituteARI Air Conditioning and Refrigeration InstituteARV Avian ReovirusASC Adhesive and Sealant CouncilASCB The American Society for Cell BiologyASCII American Standard Code for Information Inter-

changeASHRAE American Society of Heating, Refrigerating & Air-

Conditioning EngineersASME American Society of Mechanical EngineersASPE American Society of Plumbing EngineersASSE American Society of Sanitary EngineeringASTM American Society for Testing and MaterialsATCC American Type Culture CollectionATP Adenosine TriphosphateATSDR Agency for Toxic Substances and Disease RegistryAW Arc WeldingAWS American Welding SocietyAWWA American Water Works Association

- B -BAC Bacterial Artificial ChromosomeBAS Building Automation SystemBASIC Beginner’s All-purpose Symbolic Instruction CodeBAT Best Available Control Technology for existing

direct dischargers (EPA Regulations)BCT Best Conventional Control Technology for existing

direct dischargersBEV Bovine EnterovirusBEVS Vaculovirus Expression Vector SystemBFS Blow/Fill/SealBGA Institute für Arzneimittel des

Bundergesungdheitsamtes (German Health Au-thority)

BGM Buffalo Green MonkeyBHK Baby Hamster Kidney cellsBIA Brick Institute of AmericaBIOS Basic Input Output SystemBL Biosafety LaboratoryBLA Biologics License ApplicationBMA British Medical AssociationBMS Building Management SystemBOCA Building Officials and Code AdministratorsBOD Bases of DesignBOD Biochemical Oxygen DemandBOD Biological Oxygen DemandBP British PharmacopoeiaBPC Bulk Pharmaceutical ChemicalsBPD Biocidal Products DirectiveBPE Bioprocessing Equipment (ASME National Stan-

dard)BPT Best Practicable Control Technology for existing

direct dischargers (EPA Regulations)BSC Biological Safety CabinetBSCC Biotechnology Science Coordinating Committee

BSE Bovine Serum AlbuminBSE Bovine Spongiform Encephalopathy (“Mad Cow”

disease)BSI British Standards InstituteBSL Biosafety LevelBSO Biological Safety Officer (NIH Guidelines)BVD Bovine Viral DiarrheaBVDV Bovine Viral Diarrhea Virus

- C -CA Cellulose AcetateCAA Clean Air ActCAB Cellulose Acetate ButyrateCABO Council of American Building OfficialsCAMU Corrective Action Management UnitCAP Cellulose Acetate PropionateCARB Center for Advanced Research in BiotechnologyCBER Center for Biologics Evaluation and ResearchCCAR Closed Cycle Air RefrigerationCCCT Critical Crevice Corrosion TemperatureCCL Commodity Control ListCDA Copper Development AssociationCDC Centers for Disease ControlCDER Center for Drug Evaluation and ResearchCDI Continuous DeionizationCDRH Center for Devices and Radiological HealthCE Corps of Engineers (U.S. Department of the Army)CEFIC European Council of Chemical Industries Federa-

tionCEN Comité Européan des Normes (European Commit-

tee for Standardization)CERCLA Comprehensive Environmental Response Compen-

sation, and Liability ActCF Cresol-FormaldehydeCFC ChlorofluorocarbonCFR Code of Federal RegulationsCFU Colony Forming UnitCGA Compressed Gas AssociationCGAP Cancer Genome Anatomy ProjectcGLP current Good Laboratory PracticecGMP current Good Manufacturing Practice (FDA Regu-

lations)CHO Chinese Hamster OvaryCIA Chemical Industries Association (UK)CIMS Computer Integrated Manufacturing SystemCIP Clean-In-PlaceCISC Complex Instruction Set ComputerCISPI Cast Iron Soil Pipe InstituteCLEC Cross-Linked Enzyme CrystalsCLL Constant Level Loading (Lyophilizer)CLND Chemiluminescent Nitrogen DetectionCMC Carboxymethyl CelluloseCMC Chemistry, Manufacturing, and ControlsCMMS Computerized Maintenance Management SystemCMO Contract Manufacturing OrganizationCMU Cementitious Masonry UnitCN Cellulose NitrateCOD Chemical Oxygen DemandCOP Clean Out of PlaceCOSHH Control of Substances Hazardous to HealthCP Cellulose PropionateCP Cyclic PolarizationCPCT Critical Pitting Corrosion Temperature




CPG Compliance Policy Guides (US-FDA)CPMP Committee on Proprietary Medicinal ProductsCPSC Consumer Product Safety CommissionCPU Central Processing UnitCPVC Chlorinated Polyvinyl ChlorideCR Chloroprene Rubber (Neoprene®)CRO Clinical Research OrganizationCRO Contract Research OrganizationCRSI Concrete Reinforcing Steel InstituteCS CaseinCSA Canadian Standards AssociationCSCC Chloride Stress Corrosion CrackingCSI Construction Specifications InstituteCSM Chlorine Sulphonyl Polyethylene (Hypalon®)CSO Contract Service OrganizationCSRS Cooperative State Research Service (USDA)CSVC Computer System Validation Committee (PhRMA)CTI Ceramic Tile InstituteCVMP Committee on Veterinary Medical ProductsCVTR Constant Volume Terminal Reheat (HVAC)CWA Clean Water Act

- D -DCIC Dual-Column Ion ChromatographyDCS Distributed Control SystemDDC Direct Digital ControlDDS Detailed Design SpecificationDDT Dichloro Diphenyl TrichloroethaneDE Diatomaceous EarthDEA Drug Enforcement AdministrationDEL Design Exposure LimitDFISA Dairy and Food Industries Supply Association (E-

3-A Standards)DG Directorate General (UK)DHI Door and Hardware InstituteDHSS Department of Health and Social SecurityDIA Drug Information AssociationDIC Dairy Industry CommitteeDIN Deutsche Institut f?r NormungDIW Deionized WaterDMA Direct Memory AccessDNA Deoxyribonucleic AcidDNAPLS Dense, Non-Aqueous Phase LiquidsDOC Department of CommerceDOE Department of EnergyDOP Dioctyl PhthalateDOP Dispersed Oil ParticulateDOT Department of TransportationDQ Design QualificationDRAM Dynamic Random Access MemoryDRR Division of Research ResourcesDSC Differential Scanning CalorimetryDSL Digital Subscriber LineDURIP Defense University Research Initiative Program

- E -EBRS Electronic Batch Record SystemsEC European Community (guidelines for GMP manu-

facturing)EC Ethyl CelluloseECACC European Collection of Cell CulturesECLs Established Cell LinesECTFE Ethylene Chlorotrifluoroethylene (Halar®)

EDF Environmental Defense FundEDMS Electronic Document Management SolutionsEFPIA European Federation of Pharmaceutical Indus-

tries AssociationsEFTA European Free Trade AssociationEIA Electronic Industries AssociationEIR Environmental Impact ReportEJMA Expansion Joint Manufacturer’s AssociationELA Establishment Licensing ApplicationELGs Effluent Limitations GuidelinesELISA Enzyme-Linked Immunosorbent AssayELSD Evaporative Light Scattering DetectionEM Electron MicroscopyEMEA European Agency for Evaluation of Medicinal Prod-

uctsEMF Electromagnetic Force or Electromotive ForceEMS Eosinophilia-Myalgia SyndromeEOQ European Organization for QualityEP Epoxide, epoxyEP European PharmacopeiaEPA Environmental Protection AgencyEPCRA Emergency Planning and Community Right-to-

know ActEPDM Ethylene-Propylene-Diene (Nordel®)EPO ErythropoietinEPS Encapsulated PostscriptERW Electric Resistance-Welded (pipe)ERW Endotoxin Reduced WaterESACT European Society for Animal Cell TechnologyESCA Electron Spectroscopy for Chemical AnalysisEST Expressed Sequence TagETFE Ethylene Tetrafluoroethylene (Tefzel®)EU European Union

- F -FAA Federal Aviation Administration (U.S. Depart-

ment of Transportation)FACA Federal Advisory Committee ActFAT Factory Acceptance TestingFCC Federal Communications CommissionFCI Fluid Controls InstituteFDIS Final Draft International StandardFEP Fluorinated Ethylene PropyleneFHA Federal Housing AdministrationFASEB Federation of American Societies for Experimen-

tal BiologyFDA Food and Drug AdministrationFDAMA Food and Drug Administration Modernization ActFDLI Food and Drug Law InstituteFEP Fluorinated Ethylene Propylene (Teflon®)FIBC Flexible Intermediate-Bulk ContainerFIFRA Federal Insecticide, Fungicide, and Rodenticide

Act (EPA Regulations)FIP Federal Implementation ProgramFISH Fluorescent In Situ HybridizationFM Factory Mutual (Insurance Underwriters)FPM Fluorine Rubber (Viton®)FRP Fiber-Reinforced PlasticFRP Fiberglass-Reinforced PlasticFRS Functional Requirement SpecificationFTIR Fourier Transform Infrared (spectroscopy)FTP File Transfer ProtocolFWPCA Federal Water Pollution Control Act




- G -GA Gypsum AssociationGAMP Good Automated Manufacturing PracticeGAO General Accounting OfficeGCLP Good Control Laboratory PracticeGCP Good Clinical PracticesGDP Good Distribution PracticesGEP Good Engineering PracticeGILSP Good Industrial Large Scale PracticeGLP Good Laboratory PracticesGLSP Good Large-Scale PracticeGMAW Gas Metal-Arc WeldingGMO Genetically Modified OrganismGMP Good Manufacturing Practice (FDA Regulations)GPM Gallons Per MinuteGPS Global Positioning SystemGRP Glass Reinforced PlasticGSA General Services AdministrationGTAW Gas Tungsten-Arc Welding

- H -HAPs Hazardous Air PollutantsHAZ Heat Affected ZoneHAZAN Hazard AnalysisHAZOP Hazard and OperabilityHAZWOPER Hazardous Waste Operations and Emergency

Response (OSHA)HCFC HydrochlorofluorocarbonHCT High Containment TransferHDPE High Density PolyethyleneHDS Hydrostatic Design StressHEL Human Embryonal Lung cellsHEPA High Efficiency Particulate Air (Filtration)HEW Health, Education, and WelfareHFC HydrofluorocarbonHFP HexafluoropropyleneHHS Health and Human ServicesHI Hydraulic InstituteHIMA Health Industries Manufacturers AssociationHIV Human Immunodeficiency VirusHMR Hazardous Materials RegulationsHMTA Hazardous Materials Transportation ActHPB Health Protection Bureau (Canadian equivalent of

FDA)HMW-HDPE High Molecular-Weight High Density

PolyehtyleneHPLC High Pressure Liquid ChromatographyHSA Human Serum AlbuminHTF Heat Transfer FluidHTML Hyper Text Markup LanguageHTS High-Throughput ScreeningHTTP Hyper Text Transport ProtocolHVAC Heating, Ventilation, and Air ConditioningHVI Home Ventilating InstituteHWAC Hazardous Waste Action CoalitionHWM Hazardous Waste ManagementHWTC Hazardous Waste Treatment Council

- I -IAFIS International Association of Food Industry Suppli-

ersIAMFES International Association of Milk, Food, and Envi-

ronmental Sanitarians (E-3-A Standards)IAPI Institute of American Poultry Industries (E-3-A

Standards)IAPMO Informational Association of Plumbing and Me-

chanical OfficialsIBA Industrial Biotechnology AssociationIBC Institutional Biosafety Committee (NIH Guide-

lines)IBCs Intermediate Bulk ContainersIBRV Infectious Bovine Rhinotracheitis VirusICBO International Conference of Building OfficialsICH International Conference on HarmonizationICLAS International Council on Laboratory Animal Sci-

enceIEEE Institute of Electrical and Electronic Engineers,

Inc.IES International Electrophoresis SocietyIEST Institute of Environmental Sciences and TestingIFN InterferonIFPMA International Federation of Pharmaceutical Manu-

facturers AssociationIIR Isobutene Isoprene (butyl) RubberIMB Irish Medicines BoardIMP Investigational Medicinal ProductIND Investigational New DrugINN Investigational Nonproprietary Name (committee

of WHO)IOM Institute of MedicineIPEA International Pharmaceutical Excipients Audit-

ingIPEC International Pharmaceutical Excipients CouncilIQ Installation QualificationIRB Institutional Review BoardISA Instrument Society of AmericaISDN Integrated Service Digital NetworkISO International Organization for StandardizationISP Internet Service ProviderISPE International Society for Pharmaceutical Engi-

neeringITA International Trade AdministrationITG Inspection Technical Guides (FDA)ITIC International Toxicology Information CenterIUCLID International Uniform Chemical information Da-

tabaseIUPAC International Union of Pure and Applied Chemis-

tryIVD In Vitro Diagnostic

- J -JIT Just In TimeJPEG Joint Photographic Experts Group

- K -Kb KilobaseKB KilobitKHz Kilohertz

- L -LAER Lowest Achievable Emission RatesLAF Laminar Air FlowLAL Limulus Amoebocyte LysateLAN Local Area Network




LAT Loading Accumulation TableLCM Lymphocytic Choriomeningitis virusLDPE Low Density PolyethyleneLDRs Land Disposal RestrictionsLEL Lower Explosive LimitLHM Liquid Handling Module (Chromatography)LIMS Laboratory Information Management SystemLLDPE Linear Low-Density PolyethyleneLPM Liters Per MinuteLUST Leaking Underground Storage TanksLVP Large Volume ParenteralLYO Lyophilizer (freeze dryer)

- M -MAb Monoclonal AntibodyMACT Maximum Achievable Control Technology for ex-

isting major sources of HAPsMACT Maximum Achievable Control Technology for new

major sources of HAPsMAIT Minimum Auto-Ignition TemperatureMAP Mouse Antibody ProductionMBMA Metal Building Manufacturer’s AssociationMB MegabitMCA Medicines Control Agency, (British equivalent of

FDA inspectors)MCAA Mechanical Contractors Association of AmericaMCB Master Cell BankMCC Motor Control CenterMCLs Maximum Contaminant LevelsMCLGs Maximum Contaminant Level GoalsMCMV Mouse CytomegalovirusMEIR Master Environmental Impact ReportMF Melamine FormaldehydeMF Micro FiltrationMHz MegahertzMIE Minimum Ignition EnergyMIL Military Standardization Document (U.S. Depart-

ment of Defense)MIPS Millions of Instructions Per SecondMKT Mean Kinetic TemperatureML Manufacturer’s LicenseMMPs Matrix MetalloproteinasesMMS Maintenance Management SystemMoAb Monoclonal AntibodyMPW Medical Pathological WasteMRA Mutual Recognition AgreementMSDS Material Safety Data SheetMSS Manufacturers Standardization SocietyMTV Mouse Thymic VirusMVM Minute Virus of Mice

- N -NAAQS National Ambient Air Quality Standards (EPA

Regulations)NAD Nicotinamide Adenine DinucleotideNADA New Animal Drug ApplicationNARMS National Antimicrobial Resistance Monitoring Sys-

temNAS National Academy of SciencesNBC National Building CodeNBE New Biological EntityNBR Nitrile (Butadiene) RubberNBS National Bureau of Standards (U.S. Department of

Commerce)NCCLS National Committee for Clinical Laboratory Stan-

dardsNCE New Chemical EntityNCI National Cancer InstituteNCMA National Concrete Masonry AssociationNDA New Drug ApplicationNDE New Drug EntityNDR Nondispersive Infrared AnalysisNEBB National Environmental Balancing BureauNEC National Electrical CodeNECA National Electrical Contractors AssociationNEMA National Electrical Manufacturers AssociationNEPA National Environment Policy ActNESHAPs National Emission Standard for Hazardous Air

PollutantsNF National FormularyNFPA National Fire Protection AssociationNHLA National Hardwood Lumber AssociationNIH National Institute of HealthNIOSH National Institute for Occupational Safety and

HealthNIR Near Infrared (spectroscopy)NIST National Institute of Standards and TechnologyNLM National Library of MedicineNME New Molecular EntityNMRS Nuclear Magnetic Resonance SpectroscopyNORMs Natural Occurring Radioactive MaterialsNPDES National Pollutant Discharge Elimination System

(EPA)NPDWR National Primary Drinking Water Regulations

(FDA)NPL National Priorities ListNPS Nominal Pipe SizeNPT National Pipe ThreadNSF National Science FoundationNSPS New Source Performance Standards for new direct

dischargers (EPA Regulations)NTIS National Technical Information ServiceNTU Nephelometric Turbidity Unit

- O -OAC Oxygen-Arc CuttingOC Oxygen CuttingOCPSF Organic Chemicals, Plastics, and Synthetic FiberOCR Optical Character RecognitionOEL Occupant Exposure LimitOEL Operator Exposure LevelOEM Original Equipment ManufacturerOHER Office of Health and Environmental Research

(DOE)OMB Office of Management and BudgetOOS Out Of SpecificationOPP Office of Pesticides ProgramsOPTS Office of Pesticides and Toxic SubstancesOQ Operating QualificationORA Office of Regulatory AffairsORDA Office of Recombinant DNA Activities (NIH Guide-

lines)ORO Office of Regional OperationsOSD Oral Solid DosageOSHA Occupational Safety and Health AdministrationOSPRA Oil Spill Prevention and Response Act




OTC Over The Counter (Medicine)

- P -PA PolyamidePAB Pharmaceutical Affairs BureauPAC Plasma Arc CuttingPAGE Polyacrylamide Gel ElectrophoresisPAI Pre Approval InspectionPAO PolyalphaolefinPAR Proven Acceptable RangePB PolybutylenePB – ECL Performance Based – Exposure Control LimitsPB – OEL Performance Based – Occupational Exposure Lim-

itsPC PolycarbonatePCI Prestressed Concrete InstitutePCR Polymerase Chain ReactionPCTFE Polychlorotrifluoroethylene (Kel-F®)PDA Parenteral Drug AssociationPDAP Polydiallyl PhthalatePDF Portable Document Format (Adobe®)PDI Plumbing and Drainage InstitutePDUFA Prescription Drug User Fee ActPE PolyethylenePEC Chlorinated PolyethylenePEEK Polyaryl Ether Ether KetonePEG Polyethylene GlycolPEI Porcelain Enamel InstitutePEL Permissible Exposure LimitsPETP Polyethylene TerephthalatePF Phenol-FormaldehydePFA Perfluoroalkoxy resin (Teflon®)PFD Process Flow DiagramPhRMA Pharmaceutical Research and Manufacturers of

America (formerly PMA)PHS Public Health ServicePI Principal Investigator (NIH Guidelines)PIB PolyisobutylenePIC Pharmaceutical Inspection Convention (Europe)PICS Pharmaceutical Inspection Cooperation SchemePID Piping and Instrumentation DiagramPL Product License (for a biological)PLA Product License ApplicationPLC Programmable Logic ControllerPM Preventive MaintenancePMA Pharmaceutical Manufacturers Association (see

PhRMA)PMMA Polymethyl MethacrylatePMP Preventative Maintenance ProgramPOM Polyoxymethylene (Kematal®)POTW Publicly Owned Treatment WorksPOU Point Of UsePP Personnel Protection (insulation)PP PolypropylenePPE Personal Protection EquipmentPPMV Parts Per Million VolumePPS Polyphenylene SulfidePPVE PerfluoropropylvinyletherPQ Performance QualificationPREN Pitting Resistance Equivalent NumberPS PolystyrenePSD Prevention of Significant Deterioration of air qual-

ity permit (EPA Regulations)

PSES Pretreatment Standards for Existing Sources (EPARegulations)

PSNS Pretreatment Standards for New Sources (EPARegulations)

PTFE Polytetrafluoroethylene (Teflon®)PTO Patent and Trademark OfficePUR PolyurethanePVAC Polyvinyl AcetatePVAL Polyvinyl AlcoholPVB Polyvinyl ButyralPVC Polyvinyl ChloridePVCA Polyvinyl Chloride AcetatePVDC Polyvinylidene ChloridePVDF Polyvinylidene Fluoride (Kynar®, Sygef®)PVF Polyvinyl FluoridePVK Polyvinyl CarbazolPVM Pneumonia Virus of MicePW Purified Water

- Q -QA Quality Assurance (organization)QC Quality Control (organization)QP Qualified PersonQSIT Quality Systems Inspection Technique (used in

medical devices)

- R -Ra Arithmetic Average RoughnessRAC Recombinant DNA Advisory Committee (NIH

Guidelines)RACT Reasonably Available Control TechnologyRAPS Regulatory Affairs Professionals SocietyRCRA Resource Conservation and Recovery Act (EPA

Regulations)rDNA recombinant DNARF Radio FrequencyRFB Rotary Fluidized-BedRFI Radio Frequency InterferenceRFLP Restriction Fragment Length PolymorphismRH Relative HumidityRISC Reduced Instruction Set ComputerRMP Risk Management PlanningRMS Root Mean SquareRNA Ribonucleic AcidRO Reverse OsmosisRSD Relative Standard DeviationRTP Rapid Transfer PortRTP Reinforced Thermoset Plastic

- S -SAL Sterility Assurance LevelSAN Styrene-AcrylonitrileSAT Site Acceptance TestingSAW Submerged Arc WeldingSB Styrene-ButadieneSBC Southern Building Code or Standard Building CodeSBCCI The Southern Building Code Congress Interna-

tionalSBV Split Butterfly ValveSCADA Supervisory Control And Data AcquisitionSCAQMD South Coast Air Quality Management DistrictSCIC Single-Column Ion ChromatographySCID Severe Combined Immunodeficiency (bubble-boy

Continued on page 52.




syndrome)SCSI Small Computer Systems InterfaceSDI Steel Deck InstituteSDLC System Development Life CycleSDP Sterile Drug ProductSEM Scanning Electron MicroscopySGML Standardized General Markup LanguageS-HTP Secure Hypertext Transfer ProtocolSi SiliconeSI Système Internationale (system of measurement)SIM Society for Industrial MicrobiologySIP Steam In PlaceSIP Sterilize In PlaceSISPQ Strength, Identity, Safety, Purity, or QualitySMACNA Sheet Metal and Air Conditioning Contractors

National AssociationSMAW Shielded Metal-Arc WeldingSMO Site Management OrganizationSMTP Simple Mail Transfer ProtocolSOCMA Synthetic Organic Chemical Manufacturers Asso-

ciationSOP Standard Operating ProcedureSPC Statistical Process ControlSPCC Spill Prevention and Countermeasures ControlSQL Structured Query LanguageSSPMA Sump and Sewage Pump Manufacturer’s Associa-

tionSTL Safety Toxic LevelSTS Sequence Tagged SiteSUPAC Scale-Up and Post-Approval ChangesSVP Small Volume Parenteral

- T -TAGMK Tertiary cultures of African Green Monkey Kidney

cellsTCA Tissue Culture AssociationTCLP Toxicity Characteristic Leaching ProcedureTCM Tissue Culture MediumTCP/IP Transmission Control Protocol/Internet ProtocolTDD Trans-dermal Drug Delivery (product)TDS Total Dissolved SolidsTFE Tetrafluoroethylene (Teflon®)TG ThermogravimetryTIG Tungsten Inert Gas (welding process)TIMA Thermal Insulation Manufacturers AssociationTIS Total Ionized SolidsTLV Threshold Limit ValueTNKase TenecplaseTNF Tumor Necrosis FactorTNT Tumor Necrosis TherapyTOC Total Organic CarbonTOP Turn Over PackageTPA Tissue Plasminogen ActivatorTRI Toxics Release InventoryTSCA Toxic Substances Control Act (EPA Regulations)TWA Time-Weighted Average

- U -UAT Unloading Accumulation TableUF Ultra FiltrationUF Urea-FormaldehydeUFAS Uniform Federal Accessibility StandardsUHMWPE Ultrahigh-Molecular-Weight Polyethylene

UL Underwriters Laboratories (Insurance Underwrit-ers)

ULDPE Ultra Low-Density PolyethyleneULPA Ultra Low Penetration Air filtersUNS Unified Numbering System (Metallurgy)UP Unsaturated PolyesterUPS Uninterruptible Power SupplyURS User Requirement SpecificationUSAN United States Adopted NamesUSB Universal Serial BusUSDA United States Department of Agriculture (E-3-A

Standards)USP United States PharmacopeiaUSPHS United States Public Health Service (E-3-A Stan-

dards)

- V -VAC Volts, Alternating CurrentVAV Variable Air VolumeVEAs Vasopermeation Enhancement AgentsVCM Vinyl Chloride MonomerVCT Vinyl Composition TileVDC Vinylidene ChlorideVFD Variable Frequency Drive (speed)VHP Vaporized Hydrogen PeroxideVMD Veterinary Medicines Directorate (UK)VOC Volatile Organic CompoundVPHP Vapor Phase Hydrogen PeroxideVRAM Video Random Access MemoryVTAs Vascular Targeting Agents

- W -WAN Wide Area NetworkWCB Working Cell BankWFI Water For InjectionWHO World Health OrganizationWIP Lab Work In Progress Laboratory

- X -XPS X-ray Photoelectron Spectroscopy

- Y -YAC Yeast Artificial Chromosome

About the AuthorMichelle M. Gonzalez is the Engineering/Quality Managerat Fluor Daniel, South San Francisco, CA. She has nearly 35years of experience in facilities design and engineering. Sincerelocating to the US in 1965, she has held positions of increas-ing responsibility in mechanical engineering with firms suchas Shell Oil, Kaiser Engineers, Bechtel Corporation, and FluorDaniel. For the last 15 years, she has focused her professionalexpertise in the pharmaceutical and biotechnology industries.Gonzalez holds an MS in architecture from the PontificiaUniversidad Javeriana, in Bogota, Colombia. She is a lecturerat the Stanford School of Engineering, member of the ISPEBaseline® Biotech Guide Task Team, and ASME BPE subcom-mittees on dimensions and tolerances, and surface finishes.She also has been a speaker and active member of ISPE for thelast 10 years.

Fluor Daniel, 395 Oyster Point Blvd., Suite 321, South SanFrancisco, CA 94080. E-mail: [email protected]



Leak Testing

Leak Testing of Isolator Half-Suitsand Waste ContainersLeak Testing of Isolator Half-Suitsand Waste Containers

Wby Thomas P. Burns

This articleillustrates the useof a speciallydesigned inflationplate and amodifiedpressure can toammonia testisolator half-suitsfor ammonialeaks using a newmethod insteadof the traditional‘upside down’method. It alsodiscusses thecreation of aspecial air inletconnector inconjunction withthe modifiedpressure can thatwill allowammonia testingof isolator wastecontainers.

Introduction

W ith an increased use of isolators inmanufacturing operations and ste-rility testing laboratories, there is

an urgency to ensure integrity of these unitsand their components. While there has been aneffort to improve the methods used to test theintegrity of isolator canopies and gloves, therehas been minimal noted work on the testing ofhalf-suits and waste containers. Historically,leak testing these two components has beeninconvenient at best. The current standardmethods for testing half-suits and waste con-tainers are as follows:

1. For half-suits, one suit is installed in theisolator upside down (head on the floor), thena bottle of ammonium hydroxide is openedinside the isolator. When the isolator is satu-rated with the ammonia vapor, the invertedsuit is checked with a yellow ‘ammonia test’cloth for leaks. (These special cloths changecolor from yellow to green when exposed toammonia vapor. They are frequently used asa tool to search for holes/leaks in isolatorcanopies.) The problems here are (a) the suitto be tested is in an inconvenient location, (b)the suit is not completely extended, (c) theinverted suit is difficult to purge of ammoniavapor, (d) only one suit at a time can betested, and (e) the workstation is unavail-able for cleaning, etc. while testing is beingperformed. To test a second suit, the ammo-nia vapor must be purged from the isolator,the inverted suit placed into the correct posi-

tion, the second suit inverted, and the ammo-nia steps repeated. This is very time consum-ing, since it takes an hour or more to purgethe ammonia from the isolator and the half-suit.

2. For waste containers, a bottle of ammoniumhydroxide is opened inside the workstationisolator. When the isolator is saturated withammonia vapor, a waste container is con-nected to the isolator via Double Port Trans-fer Entry (DPTE) - a type of sterile door forisolators, and the waste container lid re-moved. The container is then checked with ayellow test cloth for leaks. The problems hereare (a) the waste container is in an inconve-nient location under the isolator, (b) it is notcompletely extended, (c) the container is dif-ficult to purge of ammonia vapor, and (d) theworkstation is unavailable (must be non-sterile) during leak testing.

Phase I: Air Pressure Decay/Visual Inspection

The first attempt to improve these processeswas to find a way to fully inflate the suits andthe waste containers.

Half-SuitsAn inflation plate was created for the suitswhich included an attachment ring similar tothe one in the isolator and an inlet valve - Figure1. A half-suit is suspended by the shoulderhooks above the inflation plate, and then at-tached to the plate with a standard half-suit

Figure 1. Inflation plate for half-suits. Note inlet valve on rightwith 'quick disconnect'connector.





Leak Testing

rubber band and clamp. Since leaks are frequently discoverednear the base of the suit skirt, the suit should be clamped to theplate exposing as much of the skirt as possible. (In the isolator,the suit is slid to the bottom of the clamping ring beforeattaching the rubber band and clamping band. On the inflationplate, the suit is slid on, the rubber band attached, and the suitslid up to the top of the clamping ring before attaching theclamping band. An air hose connects the benchtop air inlet tothe inflation plate (for ease of use, it is recommended to use‘quick disconnects’ for all connections). The air valve on thebenchtop is opened (30-40 psi, 2-3 kg/cm2), and the valveopened on the inflation plate. The suit is fully inflated, but notoverly stressed, and all valves are closed. This allows thoroughvisual inspection of the entire suit.

Originally, a pressure gauge was to be used to monitorpressure decay, but since the relative pressure inside the suitis quite small, the pressure was simply gauged by the heightof the suit arms. Drooping arms indicated a leak in the suit.

It was discovered that inflating the suits made them easierto clean. Before using the inflation plate, the suits werecleaned by laying them across a table or hanging them by theshoulder hooks, and wiping them down with cloths. An inflatedsuit is much easier to clean and dry since folds and creases areminimized. Using a mild soap solution to wash the suits alsocan reveal leaks since holes in the suit will cause bubbles as thesoap solution is passed over them.

Deflating the suit is achieved by simply unhooking theclamping band.

Waste ContainersAn air inlet connector was created using a sterilization inletcap and a short piece of pressure hose - Figure 2. This air inletconnector is attached to the sterilization inlet port of a clean,empty waste container. An air hose with ‘quick disconnect’connectors is then used to attach the benchtop air inlet to theair inlet connector. The waste container is inflated, beingcareful not to overinflate. The fully extended waste containeris now easy to visually inspect for holes.

The pressure decay in the waste container can be measuredby watching the air inlet connector. If there is a hole in thecontainer, the connector will droop.

To deflate the waste container, the air hose is disconnectedfrom the air inlet connector.

Phase II: Ammonia Leak TestingAlthough utilizing these inflation methods for visual leak

Figure 2. Air inlet connector for waste containers.

testing greatly increased the chances of finding holes, there wasstill a desire to find a way to ammonia test the half-suits andwaste containers. The first step in the process was modificationof a pressure can. A small pressure can was obtained, and thedip tube was removed. The intent of this process is to transferammonium hydroxide vapor, not liquid. Next, ‘quick disconnect’connections were added so the pressure can could be connectedand disconnected easily from the hoses - Figure 3. A second airhose also is required, and it also should contain the ‘quickdisconnect’ connections.

Ammonia Testing Half-SuitsThe biggest concern with ammonia testing the suits was beingable to exhaust the ammonia so an analyst could climb into thesuit and hook it back into the isolator. An attempt was madeto exhaust the ammonia through another valve in the base ofthe inflation plate, but the suit did not deflate very efficiently(this extra ‘outlet’ valve may be seen on the left side of theinflation plate in Figure 1). This problem was solved by cuttingthe fingers off a suit glove, and using a PVC pipe connector andhose clamps, attaching it to a piece of 2" (50 mm) general use‘flat’ rubber hose - Figure 4. The other end of the rubber hoseis attached to the isolator exhaust manifold via a valve.

Additionally, to ensure that the ammonia vapor circulatedthroughout the suit, a piece of tubing was attached to theinside of the inflation plate inlet. A 4-way tubing connector wasused, allowing the insertion of one piece of tubing down thenon-exhaust arm, one in the helmet, and one in the bottom ofthe inflation plate.

To ammonia test a half-suit, a suit was suspended from theshoulder hooks above the inflation plate. A small piece ofyellow test cloth was tied to the inside framework of the helmet(so the color change can be observed). One of the ends of thecirculating tubing was secured in the helmet (it was tied to thehelmet framework with the yellow test cloth), a second piece oftubing was placed into the non-exhaust arm of the suit, and thethird piece of tubing was allowed to dangle in the bottom of theinflation plate. The suit was attached to the inflation plate asstated above to allow maximum exposure of the skirt. The‘other’ glove was removed from the suit, and replaced with theexhaust glove/tubing. The pressure can was placed into a fumehood and the lid removed. A small amount of fresh ammoniumhydroxide (at room temperature) was poured into a smallbottle. This container was placed into the pressure can, and thelid replaced. (It should be noted that originally the ammoniumhydroxide was poured directly into the pressure can, but theresulting post-leak testing cleanup was difficult. Placing theliquid into a removable bottle made cleanup and disposal mucheasier.) One hose connected the benchtop air inlet to the inletside of the pressure can. A second hose connected the outletside of the pressure can to the inlet valve of the inflation plate.The exhaust manifold valve (where the glove/tubing is at-tached to the manifold) was opened. The benchtop air inlet andthe inflation plate inlet were both opened. Air now circulatedthrough the pressure can, past the ammonium hydroxide, andentered the suit. When the yellow test cloth began to turngreen, the exhaust manifold valve was closed so the pressurecould build up in the suit. When the suit was at the desiredinflation, the inflation plate valve was closed (the benchtop airinlet valve would likewise have accomplished the same effect).The inflated suit is shown in Figure 4. The suit was washedwith a mild soap solution, rinsed, and dried thoroughly. Ayellow test cloth was used to inspect the suit for leaks, and



Leak Testing

Figure 3. Pressure can with dip tube removed and 'quick disconnect' connectionsadded.

Figure 4. Inflated half-suit. Note the yellow test cloth in the helmet, and close-up ofmodified glove/exhaust line and close-up of 'high' clamping of the suit to theinflation plate.

holes were patched as necessary. (Patching a leak is performedeasily with the suit inflated, as the pressure inside the suitexerts resistance against the patch; the small amount of aircoming out of the hole usually does not interfere with theadhesion of the glue). Now, the ammonia vapor had to becleared out of the suit. First, the exhaust manifold valve wasopened to release the pressure from the suit. The benchtop airinlet valve was confirmed closed, and any residual pressurewas relieved from the pressure can by using the pressure reliefvalve. The air line was disconnected from the benchtop inletline, and the air line from the outlet of the pressure can was

removed and attached to the benchtop air inlet. To contain anyresidual vapor, the free end of the air hose was attached to theoutlet of the pressure can to make a closed loop. Now thebenchtop air valve and the inflation plate inlet valve wereopened, and the exhaust manifold valve was confirmed to stillbe open. The air flow was adjusted at the benchtop air inlet (theinflation plate valve would accomplish the same task) to keepthe suit approximately halfway inflated, and air was allowedto purge the suit until the test cloth in the helmet changed fromdark green to light yellow. When this occurred (approximately30-60 minutes), the air inlet valve was closed and the suit wasallowed to deflate. The exhaust manifold valve was closed, theglove/tubing was removed, and the suit glove was replaced.The half-suit was removed from the inflation plate, and thecirculation tubing and the small piece of test cloth wereremoved. By holding the shoulder hooks, the suit was trans-ferred to the isolator, and the suit was hung in the isolator(because of the residual ammonium hydroxide vapors, it is bestnot to enter the suit at this time). The half-suit air supply linewas attached to the suit, and the air was turned on 100%. Theair was permitted to circulate for at least 30 minutes beforeconnecting the half-suit to the isolator attachment ring. Byusing this method, virtually no residual ammonia vapor hasbeen found in the suit.

Ammonia Testing Waste ContainersA clean waste container was obtained and connected to thesterilization outlet line. The sterilization outlet valve wasopened. The air inlet connector was attached to the inlet of thewaste container. The top plate of the waste container wasremoved, and a small piece of yellow test cloth was tied to the

Figure 5. Inflated waste container. Air inlet connector is on the left and thesterilization exhaust is on the right.



Leak Testing

inlet tube near the top (so it could be seen through the plasticsleeve), and the plate replaced. The pressure can (with ammo-nium hydroxide and associated air lines) was prepared asmentioned above, and the air line from the outlet of thepressure can was attached to the waste container air inletconnector. The benchtop air inlet valve was opened, and whenthe test cloth inside the waste container began to turn green,the sterilization outlet valve was closed. When the wastecontainer reached the desired inflation, the benchtop air inletvalve was closed. (Caution - the pressure can contains someresidual pressure and will continue to inflate the waste con-tainer for a few seconds. It may be advantageous to close the airline before complete inflation, then adjust as needed later.)The inflated waste container is shown in Figure 5. A yellow testcloth was used to inspect the waste container for leaks, andholes were patched as necessary. Now, the ammonia vapor hadto be cleared out of the waste container. First, the sterilizationexhaust valve was opened to release the pressure from thewaste container. The benchtop air inlet valve was confirmedclosed, and any residual pressure was relieved from the pres-sure can by using the pressure relief valve. The air line wasdisconnected from the benchtop inlet line, and the air line fromthe outlet of the pressure can was removed and attached to thebenchtop air inlet. To contain any residual vapor, the free endof the air hose was attached to the outlet of the pressure can tomake a closed loop. Now the benchtop air valve was opened,and the sterilization outlet valve was confirmed to still beopen. The air flow was adjusted at the benchtop to keep thewaste container approximately halfway inflated, and air wasallowed to purge the container until the test cloth in the sleevechanged from dark green to light yellow. When this occurred(approximately 30 minutes), the benchtop air inlet was closedand the waste container was allowed to deflate. The wastecontainer lid was removed, the small piece of test cloth wasremoved, and the lid replaced. The sterilization outlet valve was

The biggest concern with ammonia testingthe suits was being able to exhaust theammonia so an analyst could climb into

the suit and hook it back into the isolator.

““ ““

closed, the inlet and outlet connections were removed from thewaste container, and the waste container inlet and outlet capsreplaced.

SummaryHalf-suits and waste containers are critical components ofisolators, and thorough leak testing has not always been easilyperformed. This method of leak testing half-suits and wastecontainers is a relatively quick, efficient way to find holes. Itonly takes one hole to cause a sterility breach, easily justifyingthe extra effort to find and/or prevent it. See Figure 6 for theoverview diagram.

About the AuthorThomas P. Burns is the Lead Scientist of the Sterility Labora-tory for Eli Lilly and Company in Indianapolis, IN. In more than15 years at Lilly, he has been involved with various areas ofparenteral product testing, including the chemical and biologi-cal laboratories, technical services, and sterilization valida-tion. Burns is a graduate of the University of Pittsburgh atJohnstown with a BS in chemistry. He can be contacted at 1-317/276-9207 or [email protected].

Eli Lilly & Co., Lilly Corporate Center, DC 2817, Indianapo-lis, IN 46285.

Figure 6. Overview of ammonia leak testing of half-suits and waste containers.


Integrated Strategic Planning: New Horizons for ...€¦ · Integrated Strategic Planning...

Documents

Transcript of Integrated Strategic Planning: New Horizons for ...€¦ · Integrated Strategic Planning...