politecnico di milano · 2018-06-06 · biotechnology biotechnology is the application of scienti c...

POLITECNICO DI MILANO

Scuola di Ingegneria Industriale e dell'Informazione

Corso di Laurea Magistrale in Ingegneria Chimica

Modelling analysis and comparison between single-use system and stainless steel equipment in a biopharmaceutical manufacturing

Relatore: Prof. Francesco MAESTRI Correlatore: Ing. Marco MAGNI

Tesi di Laurea di: Alberto AGAZZI Matr. 853338

Anno Accademico 2016 - 2017

Abstract

In the past decade, biotechnology, applied to pharmaceutic processes, have undergonemultiple innovations, not always followed by changes, potentially resulting in signi�cantimprovements in e�ciency. The constant evolution of disposables conquered a big paren-thesis in this debate, claiming to lead to simpler operations, eliminating cleaning andsterilization steps, as well as their validation, thus reducing costs and time of operationper batch. This already resulted in concrete investments and realizations in the recentpast. However, this process had not come without skepticism, mainly related to productquality. Assessed the goodness of the product from literature [1], the present thesis shallevaluate the bene�ts of a full disposable plant (addressed as SU, single-use) for a mono-clonal antibody production, following a generic model of exploitation of a mammalian cell(e.g. Chinese Hamster Ovary, CHO) [2]. The comparison with a common stainless-steelmodel (addressed as SS) will be conducted through a batch process simulation carriedwith SuperPro Designer ©.

A canonical structure for the mAb plant has been followed. The main feature thatneed to be speci�ed is the discontinuity nature of the process, except for some not-so-rarecases of semi continuous behavior, exploiting the perfusion technique [3].

First �eld of comparison is the schedule of the process. The focus is switched betweenplant level and section one, analyzing, for example, both maximum number of batchesand equipment utilization factors. The software allows a graphical output to easily com-prehend the issues and deal with con�icts.

Second �eld of comparison is the economic one. Investment cost and operating costare assessed, allowing a thorough evaluation. Both are discorporated and analyzed, sub-divided into plant section components and di�erent nature components, in order to beable to properly identify the reasons of the di�erences.

Third and last �eld of comparison is the environmental one. A basic life cycle assess-ment has been carried out, comparing mainly the di�erent impact generated by plasticwaste (for the SU facility) and greater WFI utilization (for the SS one).

The purpose of this elaborate, a part from the unique and not secondary results of thiscase study, is to highlight the importance of such simulation tools in every process earlyphase of development. Furthermore, revamping or general investment decision makingthrough what-if analysis would become focused and legitimated.

Sommario

Nello scorso decennio, la biotecnologia, applicata a processi farmaceutici, ha assistito amolte innovazioni, potenzialmente risultanti in incrementi di e�cienza. La costante evo-luzione dei disposable ha conquistato una importante parentesi nel dibattito, a�ermandodi condurre ad operazioni più semplici, ad esempio eliminando procedure di lavaggio esterilizzazione e relative validazioni, riducendo costi e tempi operativi per batch. Ciò haportato a concreti investimenti nel recente passato. Ad ogni modo, questo mutamen-to è stato accolto con non poco scetticismo, nonostante la forte spinta da parte dellaletteratura, principalmente dovuto a preoccupazioni sull'intaccamento della qualità delprodotto.

La tesi si propone di valutare la bontà di un impianto single-use per la produzionedi anticorpi monoclonali, seguendo un modello generico di cellule mammifere. Il con-fronto con un comune impianto stainless steel verrà condotto attraverso un software disimulazione, SuperPro Designer ©, specializzato in processi discontinui.

Primo ambito di confronto è la schedula di processo. L'analisi svarierà dal livello del-l'impianto a quello delle singole sezioni, passando dal numero totale di batch annuo ai fat-tori di utilizzo delle unità. Il software garantisce un output gra�co facile da comprendere,ottimo strumento per l'ottimizzazione.

Secondo ambito di confronto è quello economico. Sono stati calcolati costi di investi-mento ed operativi, entrambi scissi e nuovamente analizzati, suddivisi per sezione o pernatura, per garantire completezza e solidità, permettendo di identi�care appropriatamentele ragioni delle di�erenze.

Terzo ambito di confronto è l'impatto ambientale. Attraverso un basilare life cycleassessment, sono state comparate le principali di�erenze, generate dai ri�uti plastici dauna parte, e dai maggior utilizzo di WFI dall'altra.

Lo scopo dell'elaborato, a parte i risultati delle sezioni appena descritte, è quello disottolineare l'importanza dei tool di simulazione ad ogni fase di sviluppo del progetto.Infatti, qualsiasi investimento, anche di revamping e ottimizzazione attraverso what-ifanalysis, migliorerebbe in ogni aspetto.

Contents

List of Figures 4

List of Tables 6

I Introduction 8

1 Biotechnology 9

1.1 Red Biotechnology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

2 Steel or Plastic 13

2.1 Advantages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

2.2 Disadvantages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

2.2.1 Extractables and Leachables . . . . . . . . . . . . . . . . . . . . . . 15

2.2.2 Risks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

2.2.3 Wastes and Environmental Impact . . . . . . . . . . . . . . . . . . 18

2.2.4 Scaling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

2.3 Lifecycle Assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

2.3.1 Key Assumptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

2.3.2 Preliminary Results . . . . . . . . . . . . . . . . . . . . . . . . . . 25

2.3.3 Midpoint Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

2.3.4 Endpoints Results . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

2.3.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

2.4 Case studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

3 Economics 30

3.1 Strategy Proposal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

3.2 Case Studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

3.2.1 GE Healthcare . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

3.2.2 A typical MAb economic model . . . . . . . . . . . . . . . . . . . . 36

3.2.3 Simbiopharma Approach . . . . . . . . . . . . . . . . . . . . . . . 38

3.2.4 Biomanufacturing Comparison . . . . . . . . . . . . . . . . . . . . 41

4 Simulation 46

1

II Process Analysis 51

5 Overview 52

5.1 Scheduling Terminology . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53

6 Basis of Design 57

6.1 Production . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57

6.2 USP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57

6.3 Harvest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62

6.4 DSP Pre-viral . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

6.4.1 Chromatography Skids . . . . . . . . . . . . . . . . . . . . . . . . . 63

6.4.2 Intermediate Containers . . . . . . . . . . . . . . . . . . . . . . . . 65

6.4.3 Viral Filtration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65

6.5 DSP Post-viral . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65

6.5.1 UF/DF Skid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65

6.6 Media Preparation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67

6.7 Bu�er Preparation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68

7 Process Model Output 70

7.1 USP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73

7.2 Harvest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74

7.3 DSP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76

7.3.1 DSP Pre-viral . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76

7.3.2 DSP Post-viral . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77

7.4 CIP skids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79

7.5 Media Preparation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80

7.6 Bu�er Preparation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82

III Results 84

8 Cost Estimation 85

8.1 Investment Costs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86

8.1.1 General De�nitions . . . . . . . . . . . . . . . . . . . . . . . . . . . 86

8.1.2 SuperPro De�nitions . . . . . . . . . . . . . . . . . . . . . . . . . . 87

8.1.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93

8.2 Operating Costs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95

8.2.1 De�nitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97

8.2.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107

8.3 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113

9 Environmental Impact 115

2

10 General Comments 119

10.1 SU �uid waste treatment . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11910.2 SU hidden maintenance costs . . . . . . . . . . . . . . . . . . . . . . . . . 11910.3 SU footprint advantages . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11910.4 SchedulePro® . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120

Bibliography 121

3

List of Figures

1.1 The number of biotech companies in the last 15 years has been continuouslyincreasing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

1.2 Nature of active companies . . . . . . . . . . . . . . . . . . . . . . . . . . 10

2.1 CDMO's perspective of bene�ts and limits of disposable . . . . . . . . . . 14

2.2 A recent review of biomanufacturers and CMOs shows that the risk ofleachable materials entering drug products is the highest on a list of end-user concerns [1] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

2.3 Decision tree for risk analysis of extractables and leachables [1] . . . . . . 16

2.4 Further comparison of some parameters between SU and SS . . . . . . . . 18

2.5 Circles of pros and cons of Single Use Technology . . . . . . . . . . . . . . 22

2.6 Attempt to balance the factors . . . . . . . . . . . . . . . . . . . . . . . . 24

2.7 Midpoint impacts for the production of monoclonal antibody in a full pro-cess train at 2000L working volume scale with assumed mAb titre of 6 g/L.Traditional impacts are normalized to 100%. Single-use impacts are ex-pressed relative to traditional impacts within each impact category. . . . . 26

3.1 Attempt to balance the factors . . . . . . . . . . . . . . . . . . . . . . . . 30

3.2 Photo supported SS-SU comparison . . . . . . . . . . . . . . . . . . . . . . 31

3.3 Production schedules for a multi-product facility using either (A) stainlesssteel equipment or (B) single-use equipment . . . . . . . . . . . . . . . . . 34

3.4 Annual cost of goods per gram (COG/g) on a cost category basis for thestainless steel (SS) pilot plant, disposables based (DISP) pilot plant, andhybrid (HYB) pilot plant, each assumed to produce the same yield perbatch. The costs are relative to the baseline stainless steel case. . . . . . . 39

3.5 Annual direct cost of goods per gram (COG/g) on a task basis for thestainless steel (SS) pilot plant, disposables-based (DISP) pilot plant, andhybrid (HYB) pilot plant, each assumed to produce the same yield perbatch. The most expensive tasks are the CIP's in the stainless steel caseand the capture chromatography steps in the disposable and hybrid plants. 40

5.1 Sketch of the SU process grabbed from SuperPro. USP: black; DSP Pre-Viral: red; DSP Post-Viral: orange; Media: green; Bu�er: olive . . . . . . 54

5.2 Example of the chart analysis . . . . . . . . . . . . . . . . . . . . . . . . . 56

4

7.1 SS single batch schedule . . . . . . . . . . . . . . . . . . . . . . . . . . . . 717.2 SU single batch schedule . . . . . . . . . . . . . . . . . . . . . . . . . . . . 717.3 SS one year schedule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 727.4 SU one year schedule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 727.5 SS USP multiple batches schedule. Production Bioreactor BR-101 is the

Bottleneck: unit procedure with the longest duration. Facility utilizationfactor = 90%. Inoculum steps not included in the chart. . . . . . . . . . . 73

7.6 SU USP multiple batches schedule. Production Bioreactor DBS-101 is theBottleneck: unit procedure with the longest duration. Facility utilizationfactor = 90%. Inoculum steps not included in the chart. . . . . . . . . . . 73

7.7 SS Harvest schedule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 757.8 SU Harvest schedule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 757.9 SS DSP Pre Viral schedule . . . . . . . . . . . . . . . . . . . . . . . . . . . 767.10 SU DSP Pre Viral schedule . . . . . . . . . . . . . . . . . . . . . . . . . . 767.11 SS DSP Post Viral schedule . . . . . . . . . . . . . . . . . . . . . . . . . . 777.12 SU DSP Post Viral schedule . . . . . . . . . . . . . . . . . . . . . . . . . . 787.13 CIP skids schedule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 797.14 SS Media Preparation schedule for one batch . . . . . . . . . . . . . . . . 807.15 SU Media Preparation schedule for one batch . . . . . . . . . . . . . . . . 807.16 SS Media Preparation schedule for multiple batches . . . . . . . . . . . . . 817.17 SU Media Preparation schedule for multiple batches . . . . . . . . . . . . 817.18 SS Bu�er Preparation schedule for multiple batches . . . . . . . . . . . . . 827.19 SU Bu�er Preparation schedule for multiple batches . . . . . . . . . . . . 82

8.1 IC comparison. Refer to Table 8.2 for the categories. . . . . . . . . . . . . 958.2 Depreciation visual summary of a �ctitious case . . . . . . . . . . . . . . . 1048.3 SS generic operating month of labor requirement . . . . . . . . . . . . . . 1098.4 SU generic operating month of labor requirement . . . . . . . . . . . . . . 1098.5 SS OpEx Pie Chart . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1118.6 SU OpEx Pie Chart . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111

5

List of Tables

1.1 Italian overview of biotech industry . . . . . . . . . . . . . . . . . . . . . . 101.2 Italian focus on red biotech industry . . . . . . . . . . . . . . . . . . . . . 11

2.1 SU systems disposal techniques available . . . . . . . . . . . . . . . . . . . 20

3.1 Utilities consumptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 443.2 Investment cost comparison . . . . . . . . . . . . . . . . . . . . . . . . . . 443.3 Operating cost comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

6.1 Bu�er List . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 686.2 Bu�er Preparation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69

7.1 Recipe Scheduling Information . . . . . . . . . . . . . . . . . . . . . . . . 707.2 USP Utilization Factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . 747.3 USP Utilization Factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . 747.4 DSP Pre Viral Utilization Factors . . . . . . . . . . . . . . . . . . . . . . . 777.5 DSP Post Viral Utilization Factors . . . . . . . . . . . . . . . . . . . . . . 787.6 CIP skids Utilization Factors . . . . . . . . . . . . . . . . . . . . . . . . . 797.7 Media Preparation Equipment Utilization Factors . . . . . . . . . . . . . . 817.8 Bu�er Preparation Equipment Utilization Factors . . . . . . . . . . . . . . 83

8.1 Itemized Fixed Capital Cost Estimation . . . . . . . . . . . . . . . . . . . 888.2 Fixed Capital Estimate Summary, prices in e . . . . . . . . . . . . . . . . 938.3 Equipment Purchase Cost: Section Speci�cation . . . . . . . . . . . . . . . 948.4 Operating cost items and ranges . . . . . . . . . . . . . . . . . . . . . . . 968.5 Annual Operating Cost - Process Summary . . . . . . . . . . . . . . . . . 1078.6 Consumables Cost - SU Process Summary . . . . . . . . . . . . . . . . . . 1108.7 Material and Utilities Detailed Costs . . . . . . . . . . . . . . . . . . . . . 1128.8 Cost Estimation Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . 1138.9 Arbitrary score table for evaluation purpose . . . . . . . . . . . . . . . . . 114

9.1 Bag Weights Estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1179.2 LCA results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118

6

Part I

Introduction

8

Chapter 1

Biotechnology

Biotechnology is the application of scienti�c and engineering principles to the productionand the exploitation of products and/or microorganisms through the usage of biologicalagents. The applications fall into: health (red biotech), food, feed and agriculture (greenbiotech), environment and sustainability (white biotech) [4].

To better understand the importance of this sector, some graphs and tables are takenfrom a 2016 report regarding Italy as example, being this research developed in thiscountry [5].

Figure 1.1: The number of biotech companies in the last 15 years has been continuouslyincreasing

Italian biotech industry encounters a continuous development, achieved through bothacademic and industrial research, and the companies' will is to translate innovation intovaluable products.

Biotechnology is subdivided in categories, usually associated with colors as in Figure 1.2:red is related to health-care, white to chemistry, green to agriculture. GPTA stays forGenomica, Proteomica e Tecnologie Abilitanti, meaning Genomic, Proteomic and Key

9

1 � Biotechnology

Figure 1.2: Nature of active companies

Enabling Technologies. The last ones are called such because developing innovating so-lutions and technology improvements. Anyway, the picture underlines the prevalence ofbiotechnologies applied to human health, topic where our case study will be focused on.

Almost the same pattern is reproducible for R&D and sales. Regarding this lastinformation, the 75% of the total sales is under the red biotech and is foreseen to increase,overall, between 12.8% to 18.1% within two years. The R&D is mainly focused on health:investments with 90% and labor with 79%. Breaking down these values, the 53% ofdeveloping projects regard preclinic phase, again accordingly to the case study that willdevelop in this elaborate. Following the trend, 16% of the projects focus monoclonalantibodies, being the second voice behind generic small molecules, with 32%.

BIOTECH All companies R&D Biotech R&D Italian asset

Companies 489 256 240Sales [ke] 9 440 916 3 836 558 838 867

Investments [ke] 1 855 187 419 748 194 592Labor 9229 4054 2921

R&D Labor 3670 2517 1699

Table 1.1: Italian overview of biotech industry

Results shown in Table 1.1 and Table 1.2 re�ect the sequencing of human genome andthe comprehension of illnesses' genetics and molecularities, allowing approaches more andmore selective and focused. Personalized medicine has become a new paradigm, increasinge�ectiveness.

Further statistics from Assobiotec [6] assess that Italy is placing third in Europe, with250 billion e after Germany (327 billion e) and France (285 billion e). The weight ofbio-economy on the overall production is even better, placing Italy in the second place(8.1%) behind only Spain (10.1%). It is important to precise that this data paints

10

1 � Biotechnology

RED BIOTECH All companies R&D Biotech R&D Italian asset

Companies 261 131 117Sales [ke] 7 131 284 3 663 551 667 880

Investments R&D [ke] 455 902 358 966 164 219Total Labor 49 995 8968 4264Biotech Labor 6566 3221 2107R&D Labor 7524 2211 1347

Biotech R&D Labor 2911 2073 1261

Table 1.2: Italian focus on red biotech industry

the picture from a higher perspective. Looking in detail to Italian red biotech, the sliceinterested is around 2%, being the main part taken by agriculture, food and manufacture.The results' order of magnitude is consistent with previous one.

1.1 Red Biotechnology

An increasing number of therapeutic candidates, including monoclonal antibodies, bio-therapeutic proteins, and vaccines, are currently entering early-stage process development.At the same time, biologics are being introduced onto the market or have recently beenintroduced. In this competitive market, time-to-market, cost-e�ectiveness, and manu-facturing �exibility are key issues that all must be achieved while maintaining productquality.

Bioengineering is pushing towards high cell densities, high productivity, cost-e�ectiveprocess design and speed to reach market introduction.

Indeed, for a moderately successful drug (one with annual sales of 350 million �) eachday's delay to market incurs a loss of 1 million � (e.g., Clemento, 1999). Historically,attractive returns have made companies emphasize speed to market rather than focus onimproving process economics (Sadana and Beelaram, 1994).

Another reason for optimization is biologics' price: in the U.S. alone, biologics accountfor 40% of prescription drug spending, despite only 2% of the population using biologicdrugs [7]. The cost of these medications is quickly rising from expensive to simply unaf-fordable. This global issue will likely become more problematic with the steady growth ofthe world population and the advent of new biologic therapies. Some strategic measureshave been implemented to mediate the rising prices such as a cost-e�ectiveness analysismethodology which appraises health interventions and selects technologies based on thereturns expected for a �nancial investment, therefore urging pharmaceutical companiesto provide biologics at competitive rates [8]. Many governments have passed legislatureaimed at reducing expenditures on follow-on biologics such as biosimilars [9] which de-creases the costs of a product entering the market by shortening the approval pathway,ensuring compulsory licensing and encouraging data sharing. These strategic initiativesare expected to moderate the high costs of new biologics by encouraging a competitivemarket. Despite these strategic initiatives, drug prices are reaching as high as 50 000 �per treatment [10]. At the same time, innovation manufacturers (bringing a new drug

11

1 � Biotechnology

to market) are battling low pro�t margins. The average total cost of launching a noveldrug was 3 billion � between 2004 and 2009, but the R&D portion of that total rose frombetween 18% and 23% to 34%. Expenses exceeded sales of novel drugs in this �ve-yearperiod [11] [2].

The rising market for generics or biosimilars are supported by most regulatory healthauthorities because they reduce the costs placed on national health infrastructures [12].Biosimilars are intended to be priced lower than the innovator product, thus creating ademand for more cost-e�ective production [13]. It is important to keep in mind that,since dealing with health, increasing pressure to reduce product costs must come whilemaintaining product quality. That is why the introduction of such a promising innovationlike disposables is being so controversial.

Our case study focuses monoclonal antibodies (MAb), large protein molecules used totreat a wide variety of illnesses, such as rheumatoid arthritis, psoriasis, Chron's disease,transplant rejection, and a variety of cancers. They constitute the fastest growing segmentin the biopharmaceutical industry. More than 20 MAbs and fusion proteins are approvedfor sale in the USA and Europe and approximately 200 MAbs are in clinical trials for awide variety of indications. The market is growing more than 15% per year, was valuedat 85.4 billion � in 2015 and is expected to reach a value of 138.6 billion � by 2024 [14][15].

12

Chapter 2

Steel or Plastic

Decisions regarding whether and when to use single-use (SU) (disposable) devices orstainless steel (SS) equipment for biopharmaceutical manufacturing have been discussedfor more than a decade. To date, no argument in terms of safety, cost-e�ectiveness, oroperational e�ciency is fully convincing to choose one technology platform or the otherfor all applications. Changing to or implementing single-use bioreactors will need to beginprior to manufacturing � this may alter time to market as well as the �nancial picture [16].

Nowadays, more and more companies are investing in building and revamping plantsfollowing this road, but sometimes the awareness is not as high as it should be, and thechoice is driven mainly by a wave of optimism. In other terms, the technology is de�nitelya valid alternative, but the e�ective advantages still lay under a dim veil.

Traditionally, stirred tanks, glass vessels, and stainless-steel tanks have been used atlaboratory and pilot scales for process development and production of research grade,toxicological, and Phase I clinical material. Stainless steel tanks dominate large-scalemanufacture (>1000L) of bio-therapeutics. However, �xed plant equipment is costly,requiring long lead times for installation and quali�cation. There is also a high burdenfrom validation e�orts related to sterility and cleaning, as well as for maintenance. Therisk for cross contamination in standard steel or glass equipment leads to strict rules forcleaning and cleaning validation [17].

Single-use technology is of speci�c interest when producing material for early clinicaltrials to avoid a capital investment early, when the �nal production bioreactor volume,as well as the future of the molecule is unknown. Further advantages are encounteredwhen building a new facility, because the need for utilities might be reduced in a fullydisposable environment, therefore reducing start up time, installation costs, and campaignturnaround [18]. Single-use bioreactors o�er the advantage of fast installation, lower in-vestments in infrastructure, and a signi�cant decrease of validation costs.

Biopharmaceutical companies often do not have in-use data to make strategic manu-facturing decisions. Contract Development and Manufacturing Organizations (CDMOs)have recently focused this aspect.

The most critical reasons to embrace single-use technologies are to reduce capitalinvestment in facility and equipment and to increase speed to clinical trials. Its burdenis also spread out and redistributed more evenly over the life of the plant [19] [20]. This

13

2 � Steel or Plastic

feature makes disposables very appealing for contexts of limited capital availability, likepublic research centers. New stainless-steel�based facilities can take up to four years tobuild, validate, and become fully functional, which requires signi�cant capital investmentand consequently a higher payback time.

The speed of product changeover is another factor to consider. CDMOs with highlydynamic and frequently changing product portfolios bene�t from disposables that allowfaster transitions between di�erent production campaigns. That leads to better facilityuse, faster returns on investment, and minimal unproductive downtimes for cleaning.

Figure 2.1: CDMO's perspective of bene�ts and limits of disposable

2.1 Advantages

Some advantages [21] are collected, maybe repeated, in the following list:

� increases in batch success rate

� elimination of potential cross contamination

� more rapid changeover between campaigns

� reduced time for a new facility to become operational

� reductions in water and waste water requirements

� eliminations of clean-in-place (CIP) and steam-in-place (SIP) and their skids andvalidation

� regulatory robustness

� ease of use

� lower cost of operation

14


� lower investments

� �exible production capacity (can be increased quickly)

� late stage development (scale-down experiments)

� rapid production at di�erent scales.

Furthermore, since disposable bags do not require SIP or CIP steps, in general, mediaand bu�er preparation cycle times for the disposables option are shorter than the stainless-steel case [22]. Moreover, since no autoclaves nor steaming and cleaning skids are needed,the limited infrastructure required is also a key advantage for single-use equipment.

2.2 Disadvantages

2.2.1 Extractables and Leachables

Figure 2.2: A recent review of biomanufacturers and CMOs shows that the risk of leachablematerials entering drug products is the highest on a list of end-user concerns [1]

Accurate risk assessment requires consideration of quality issues for many single-usematerials. Speci�cally, biomanufacturers should consider variability of polymeric materi-als and related leaching or extracting of small compounds from polymers into products.Leachables are chemicals that migrate from the product contact surface into the pro-cess �uid (e.g., bu�er, water, or process intermediate) under normal exposure conditions,

15


Figure 2.3: Decision tree for risk analysis of extractables and leachables [1]

whereas extractables are chemicals that can be removed from the product contact sur-faces using appropriate solvents under extreme exposure conditions to facilitate theiridenti�cation and quantitation. The extractable pro�les released from bioreactor bagsunder extreme temperature and solvent conditions are typically characterized by vendors.Leachables are typically a subset of the extractables and are expected to be released atvery low levels under normal conditions [23].

In particular, SU tools are often manufactured from plastic derivatives. There havebeen concerns raised regarding the e�ect of extractable and leachable compounds, suchas antioxidants, plasticizers, and curing agents, from these plastic disposable technologieson the quality of the �nal product. For example, an extensive study by Amgen reportsthat bis(2,4-di-tert-butylphenyl)phosphate (bDtBPP), a cytotoxic compound (i.e., toxicto cells), leaches from certain SU bags and impedes cell growth under many cell cultureconditions [24]. The Bio-Process Systems Alliance (BPSA) has issued a comprehensivereport outlining recommendations for extractables and leachables testing from SU equip-ment [25] [2].

16


Other detailed studies of this kind are reachable in literature. One of those comparedwater and ethanol e�ects at, mainly, two di�erent timings: initial �ush and 12-monthsample. Results claimed by Ding et al. are various:

� the initial system-�ush sample shows higher values of extractables, suggesting aninitial pre-�ush, whenever possible, to remove a substantial proportion of extractablematerial

� in general, water extractables, most common in biotech processes, are substantiallylower than ethanol extractables for organic compounds

� in most cases, no signi�cant increase in extractables is seen after storage for up to12 months

� pH is usually in compliance with US Pharmacopoeia requirements

� the Nonvolatile Residues (NVR) show a time-dependent gradual increase in ex-tractables

� for volatile and semi volatile compounds, no detectable peaks are found in watersamples, while several compounds are identi�able in the ethanol extracts, with someadditionalities after 12 months, but anyway with concentration < 5 ppm.

Furthermore, during process validation, if such thorough data from all components ofsingle-use system are available and the test conditions represent a worst case to processones, then little or no further extractables testing may be required. Otherwise, a process-and product-speci�c extractables validation study may need to be performed on the wholesingle-use system for determining potential leachables and assess any toxicity or impacton quality of a �nal drug product [1].

Primary focus on patient safety, manufacturers must ensure that leachables which oc-cur during processes will be removed from �nal products. So, the closer to �nal producta single-use material is used, the higher the risk of leachables contamination and insu�-cient elimination. Leachables from upstream processing have a rather low signi�cance topatient safety because many downstream puri�cation sequences have a highly su�cientcapability of removing them.

2.2.2 Risks

One reoccurring issue is the risk of leakage that could destroy a production batch. Such asituation can arise from improperly welded seams of bags, from punctures during handling,and from loose connections. Therefore, suppliers and end users need to provide a highlevel of quality control. Regarding handling, a labor related risk factor must be included.Using a single-use bioreactor is di�erent than using a stainless-steel bioreactor, which canmake operators uncomfortable when they �rst start to use single use. The operator willneed to learn to handle and install the single-use bag correctly � all relatively easy, butvery di�erent than working with stainless steel systems [26].

17


Another risk is the potential binding between media and plastic. Literature stud-ies highlight the potential of the extracellular environment, which here is the polymericsurface of the bags, to modify medium composition and to emphasize the importanceof medium formulation strategies, including those used in the delivery of hydrophobiccomponents. It is noted, however, that the level of loss is very dependent on the speci�csystem including the composition of the culture medium used [27] [28]. In our experience,most CHO cell lines do not have any issues in being operated in single-use bioreactorsystems [23].

Figure 2.4: Further comparison of some parameters between SU and SS

2.2.3 Wastes and Environmental Impact

On top of the validation concern regarding leachable and extractable materials, the useof plastic SU products also triggers the debate over their potential environmental impact.The growing implementation of SU products is accompanied by increasing concern aboutthe adverse environmental e�ects from the regular, large-volume disposal of plastic waste.Some suppliers have assessed the possibility of recycling but concluded that the likelihoodis very limited due to mixed plastic content and possible requirement of pre-treating bio-hazardous materials; so, traditional methods of disposing SU plastics through land�ll andincineration will remain as the standard. While plastic waste management is a drawbackfor SU technologies, several studies have shown that the environmental bene�t of using SUtechnology due to reduced consumption of energy, water, and cleaning agents compared tomulti-use stainless steel facilities outweighs the disadvantage and can signi�cantly reducethe overall carbon footprint [29] [30] [2].

Single-use systems do not require such intensive sanitization e�orts; their environmen-tal footprints are more a result of their plastic content. Plastics are essentially para�n-

18


like chemicals with repeating chains of CH2 molecules. Given this chemical nature, plas-tics are indeed fuels. The preferred disposal method for such fuels is incineration, with orwithout energy recapture. Environmental impact should account for facility size, waterconsumption, energy use, and carbon emissions from all steps, including even the datafrom mining the iron ore to steel manufacture and the drug manufacturing facility, thediesel consumed in transporting plastic as well as the transportation of waste plastic tothe incineration facility, and the incineration of the plastic itself, and eventually employ-ees driving to work, which is evidently a more thorough approach than the simple carbonfootprint. Furthermore, it is instructive, even if it is not correct to generalize, to see thatby far the largest contributor to carbon footprint of all categories in this facility is workersdriving to work, with a percentage of 80% [29].

Nevertheless, one way to reduce the impact of such waste could be to convert backpart of the 32.6GJ/ton of energy stored in plastic in waste-to-energy incineration facil-ities, not necessarily solving the issue of carbon footprint. The ultimate solution mightreside in recycling these disposable products, requiring further development on innovativetransformation methods. Some other interesting future directions with respect to single-use bioreactors could be the development of systems for perfusion process applications,as well as more insights on leachables [31] [32].

Therefore, single-use bioprocess systems can provide a range of environmental bene-�ts beyond those of stainless steel systems. Although single-use systems may generateadditional solid waste, bene�ts include reduction in the amount of water, chemicals, andenergy required for cleaning and sanitizing as well as avoiding the labor-intensive cleaningprocesses required with stainless steel systems [23] [33].

In most cases, single use devices can be disposed of by the same means currentlyused for process �lters, disposable laboratory supplies and other solid wastes. However, athorough overview and comparison of the disposal options is presented in Table 2.1, wherethe estimated capital and disposal costs increase according to the presentation order.

Untreated : One industry player land�lls an untreated component because its systemdoes not require prior treatment. Another company whose products do not require land�llpretreatment uses biohazard bags before disposal in solid waste trash. Treated: Basedon the application, some companies �kill� the used system or component with a dose ofchlorine dioxide or other deactivator and then dispose of the item in a land�ll. Thisoption is more expensive than disposing of an untreated component because it requiresextra steps before land�lling. It does, however, allow for product to be land�lled after usewithout other cleaning and decontamination steps.

Grind and Autoclave: Some items are pretreated and shredded before land�lling. Thisoption is appealing because it may be accepted as safe in some cases and reduces land�llvolume compared with unshredded product. Additional discussions are ongoing regardinguse of other hospital waste treatments such as autoclaving, thereby making a single-use system or component suitable for disposal in a standard municipal waste incineratoror land�ll (if allowed). Some companies dispose of their used components or systemsinto a grinder-autoclave currently used at many hospitals. Land�ll technology continuesto advance with the development of renewable energy facilities. Biomethane and otherland�ll gases are environmentally sustainable fuels that can provide alternative energy

19


Option Advantages Disadvantages

Land�ll, untreated Lowest operating cost, nocapital cost

Not an option for haz-ardous waste; perceived asenvironmentally unfriendly

Land�ll, treated Inexpensive, no capital cost Perceived as environmen-tally unfriendly

Grind, autoclave, andland�ll

Generally accepted as safe,reduces land�ll volume

Signi�cant capital cost, re-quires extra handling

Recycling Environmentally appealing Impractical for mixed ma-terials

Incinerate Generally accepted as safe May be legally restrictedand costly

Incinerate with generationof steam or electricity(cogeneration)

Most environmentally be-nign, some return on in-vestment

May be legally restricted,and presents the highestcapital cost

Pyrolysis Produces usable pure dieselfuel; fuel produced burnsmore cleanly than that pro-duced from a re�nery

New technology � few op-tions available; subpar e�-ciency

Table 2.1: SU systems disposal techniques available

sources for future generations. Because single-use systems contain insigni�cant amountsof biomass/organic material, it is unlikely that they will be a source of sustainable fuelswith present technology.

Recycling : Recycling is viewed as an environmentally appealing option because it en-tails collecting materials that would otherwise be considered waste, sorting and processingrecyclables into raw materials, and manufacturing raw materials into new products. Re-cycling is a desirable option especially for products made from a homogenous material.However, many single-use bioprocess systems are made of multiple materials or layers in-cluding polyethylene (PE), polypropylene (PP), ethylene vinyl alcohol, or nylon. It wouldbe more feasible to recycle the silicone tubing that is often attached to bag systems, butthat would require segregating tubing and bag, the logistics of which are not simple. Inaddition, recycling may require pretreatment of biohazardous single-use materials, de-pending on the process step for which the disposable equipment was used. Recyclinggenerally is not an option for such products because they require extensive e�orts to sepa-rate the components into homogeneous components. As a result, most single-use systemsand components are not amenable to recycling e�orts. The combination of these factorsmakes it extremely di�cult to develop an economically viable case for recycling disposabletechnologies in their current con�guration.

Incineration: According to the US Environmental Protection Agency, �incineration

20


is a widely-accepted waste treatment option with many bene�ts. Combustion reducesthe volume of waste that must be disposed in land�lls, and can reduce the toxicity ofwaste�. Incineration is a method of disposal that is used in many countries, and somecompanies incinerate as part of their standard disposal policy. In the European Union,several directives speci�cally address the issue of waste incineration and disposal:

� Directive 2000/76/EC of the European Parliament and the Council on the inciner-ation of waste.

� Directive 94/67/EC of the Council of the European Communities excludes inciner-ators for infectious clinical waste unless rendered hazardous according to Directive91/689/EEC of the Council of the European Communities on hazardous waste.

Cogeneration is a process in which a facility uses its waste energy to produce heator electricity. Cogeneration is considered more environmentally friendly than exhaustingincinerator heat and emissions directly up a smokestack. One bioprocess company sendsall its waste to a facility that incinerates and uses it to generate electricity for a majorcity. Another company uses a waste heat boiler to make low-pressure steam. Althoughthere are wide variations, the heat value of mixed plastics waste is estimated to be about15.000 to 20.000BTUs/lb (34.890 to 46.520 kJ/kg), which compares favorably to coal at9.000 to 12.000BTUs/lb (20.934 to 27.912 kJ/kg). Cogeneration is more widely appliedin Europe and Asia than in the United States. In the United States, this process is beinginstalled increasingly at universities, hospitals, and housing complexes for which boilersand chillers can serve multiple large buildings.

Pyrolysis: Pyrolysis is a method for converting oil from plastics such as PE, PP, andpolystyrene (PS) that can be used as fuel for internal combustion engines, generators,boilers, and industrial burners. Plastics are separated into oil, gas, and char residue bybeing heated in a pyrolysis chamber. Gas �owing through a catalytic converter is con-verted into distillate fraction by the catalytic cracking process (enzymatically breaking thecomplex molecules down into simpler ones). The distillate is cooled as it passes through acondenser and then is collected in a recovery tank. From the recovery tank, the productis run through a centrifuge to remove contaminates. The clean distillate then is pumpedto a storage tank. About 950mL of oil can be recovered from 1 kg of certain types ofplastics. A comparison of the distillate produced by pyrolysis and regular diesel showsgood similarity between the fuels, with the advantage that distillate from pyrolysis burnscleaner.

Although this discussion of options for disposal illustrates the main attributes of di�er-ent methods, standards for environmental practices vary by location, country, and regionof the world regarding handling biohazardous wastes, plastic wastes, and waste �uid. Forexample, one European company classi�es waste based on its potential hazard and selectscorrespondingly its disposal method.

Appropriate disposal of single-use systems and components can be part of a processthat is environmentally friendly by reducing overall energy consumption as well as chemicaland cleaning demands associated with traditional stainless-steel systems [34].

21


Figure 2.5: Circles of pros and cons of Single Use Technology

2.2.4 Scaling

SU tend to be limited in scale. Those size limits are mainly attributable to the achievableoxygen transfer. Low oxygen-transfer coe�cients prohibit most microbial applicationswith an increased oxygen demand. So, disposable bioreactors are primarily used formammalian-cell cultivation, for which mass transfer of oxygen seems su�cient to sup-port high cell densities. To enhance oxygen transfer, if needed, stirred bioreactor (micro-)spargers are applied and air may be enriched with oxygen. The easiest solution of improv-ing the stirring is unfeasible: even if it is believed that shear stress due to agitation hasbeen over-estimated to damage cells, shear may result in nonlethal physiological responsesthat can alter product quality [35].

However, during the past 15 years, advanced cell-line engineering and process devel-opment have resulted in more productive cell cultures: cell-culture titers in fed-batchprocesses have increased from 0.05 to over 10 g/L today, allowing the use of smaller scalebioreactors, thus alleviating this just spawned problem.

There is less doubt nowadays, however, that a combination of disposable and reusableequipment is bene�cial for process economics in both new and existing manufacturingplants. Such hybrid models can be applied with di�erent degrees of disposable integration,and that degree generally depends mainly on process scale and the biotech manufacturers'experience with disposable technology. Mid-size plants will probably bene�t the mostfrom the use of this hybrid model, but data from real case studies are not yet available to

22


validate this hypothesis.Considering the scale problems and any ways of unraveling them, single-use approach

can be applied only downstream, even if leachable contamination is increased, in chro-matography and �ltration. Gradual implementation of single-use systems in a stainless-steel plant results in a hybrid approach, using, for example, as just stated, disposablesonly in the puri�cation train. For the same reason, disposable rocking type bioreactorshave found increasing popularity in the �eld of cell expansion operations, where scale islimited by microorganism needs.

A path followable to achieve desired productivity without increasing too much units'dimensions is adopting a continuous perfusion culture rather than the more widespreadfed-batch reaction. This consists in a continuous cell culture and fermentation, withharvest collection at regular intervals. As anticipated, due to the higher productivity ofperfusion, production levels like those of the fed-batch mode can be obtained with smallerfermentation volumes. However, the high medium turnover required for the continuousoperation of perfusion requires the use of several medium and harvest storage vesselsfor each production fermenter. A large-scale perfusion operation also involves frequentpuri�cation runs with intensive bu�er preparation and storage operations. In general,all this leads to a smaller, but more sophisticated, manufacturing facility compared to atypical biotech plant [19].

2.3 Lifecycle Assessment

Lifecycle Assessment (LCA), or Lifecycle Cost Analysis (LCCA) is an internationallyrecognized discipline that can be used to examine products and processes from an envi-ronmental perspective across the full lifecycle of a product or process, from raw materialextraction and re�ning through manufacturing, use, and end-of-life disposal or recycling.

Adoption of single-use process technology reduces or eliminates the need for exten-sive cleaning and steam sterilization between batches but can introduce environmentalimpacts related to the manufacturing, use, and disposal of consumable materials. Thisenvironmental study of single-use process technology is the �rst to o�er a comprehensiveexamination of environmental impacts across the full process train.

An extensive case study of a process producing monoclonal antibodies (mAbs) atclinical and commercial manufacturing scales has been conducted by GE Healthcare incollaboration with BioPharm Services Ltd., developer of Biosolve Process. The life-cycleassessment models were developed in SimaPro Analyst® version 7.2.4 life-cycle assess-ment software.

2.3.1 Key Assumptions

To maintain sterility, traditional durable equipment must be cleaned and steamed in place(CIP/SIP) between each batch. This requires a large amount of process water, waterfor injection (WFI), acids, and bases. The energy and supporting equipment requiredare all considered in this analysis. Single-use components that contact media do notrequire rigorous cleaning and sterilization, but instead are pre-sterilized by o�site Cobalt-60 irradiation.

23


Figure 2.6: Attempt to balance the factors

The traditional durable equipment is nominally assumed to have 10-year lifetimes,after which 25% of the equipment is re-used while the remainder is either recycled (90%)or land�lled (10%). The single-use process trains contain components that are designedto be used once and then discarded. The exceptions are the replacement of single-usechromatography columns, which are typically reused for several batches depending on thenumber of cycles per batch. In this case, a recommended usable life for a ReadyToProcessCapto S 2.5 chromatography column is 20− 50 cycles.

Sensitivity analyses are used to identify key parameters in the model that tend to a�ectthe model's response more than others. It is also important to understand the potentialfor the model to yield di�erent conclusions in response to the choice of parameter values.In addition, uncertainty assessment was considered, in particular when the sample set wassmall.

Regarding treatment at end-of-life, for single-use components such as cellbags, �lters,

24


and connectors, disposal was assumed to occur by hazardous waste incineration withoutwaste heat recovery. Non-hazardous waste was sent to land�ll or wastewater treatment.Process water was assumed to be used once without recovery.

Use-phase electricity was assumed to be from an average US grid mix. Selection ofan average European electricity grid mix exhibits lower environmental impacts but doesnot lead to any discernable shift of relative magnitudes between single-use and traditionalprocess technology.

The fuel mix for generation of WFI was composed of di�erent ratios of fuel oil, naturalgas, and electricity. The default mixture was equally weighted for fuel oil and natural gasat 45% each while electricity was weighted at 10%.

2.3.2 Preliminary Results

The �rst results reported focus on global warming potential (i.e., greenhouse gas emis-sions), cumulative energy demand (i.e., embodied energy), and water usage. They arecategorized by life-cycle stage. The supply chain phase includes materials and manufac-turing of all process equipment and consumables required to support a 10-batch mAbproduction campaign. The use phase includes all impacts that occur during mAb pro-duction, including cleaning and sterilization of traditional durable equipment betweenbatches. The end-of-life phase includes the disposal of consumables and the disposal,re-use, or recycling of allocated portions of durable components.

Cumulative energy demand (CED) and global warming potential (GWP) results arevery similar because all of the GWP is related to energy production and consumption. Thesingle-use process train exhibits 38% lower GWP and CED during use phase comparedto a traditional durable process train, which there reaches a peak of 350 000 kgCO2eq

and slightly less than 6000GJeq. The most substantial di�erences are related to thesupport CIP/SIP system: water savings are realized during the use phase for single-use process technology due to the reduction or elimination of cleaning and sterilizationbetween batches. Other phases are neglected because supply-chain impacts represent<11% of the life-cycle CED impact and <5% of the life GWP impact. Environmentalimpacts from the end-of-life stage represent <1% of overall life cycle impacts.

Shift to single-use process technology can result in substantial reductions in cumu-lative energy demand, global warming potential, and water usage for the production ofmonoclonal antibodies, in addition to improving �exibility and productivity. Althoughsingle-use process technology introduces a need for the production, distribution, and dis-posal of single-use components, this approach also reduces or eliminates the need for largequantities of steam, process water, and water for injection.

2.3.3 Midpoint Results

For the following sections, the ReCiPe midpoint and endpoint impact assessment methodhas been used. The ReCiPe method is one of the most recent and comprehensive impactassessment methods available to LCA practitioners. The method addresses a wide rangeof environmental concerns at the midpoint level and then aggregates the 18 midpointsinto a set of three endpoint damage categories. Midpoint impacts re�ect the impact at a

25


common midpoint in the cause-e�ect chain, while endpoint impacts assess the measure ofthe damage caused by that stress or at the end of the cause-e�ect chain.

ClimateChange

OzoneDepletion

HumanToxicity

Photochemicaloxidant

Formation

ParticulateMatter

Formation

IonisingRadiationTerrestrial

Acidi�-cation

FreshwaterEutroph-ication

MarineEutroph-ication

TerrestrialEcotoxicity

FreshwaterEcotoxicity

MarineEcotoxicity

AgriculturalLand

Occupation UrbanLand

Occupation

NaturalLand

Transfor-mation

WaterDepletion

MetalDepletion

FossilDepletion

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

SSSU

Figure 2.7: Midpoint impacts for the production of monoclonal antibody in a full pro-cess train at 2000L working volume scale with assumed mAb titre of 6 g/L. Traditionalimpacts are normalized to 100%. Single-use impacts are expressed relative to traditionalimpacts within each impact category.

The single-use approach exhibits in Figure 2.7 lower environmental impact in all 18midpoint impact categories studied.

26


2.3.4 Endpoints Results

The endpoints considered are human health, ecosystems, and resources. Human healthis expressed in units of Disability Adjusted Life Years (DALYs) and covers fate, expo-sure, e�ect, and damage analyses. Climate change on human health, ozone depletion,human toxicity, photochemical oxidant formation, particulate matter formation, and ion-izing radiation are the impact categories associated with human health. The ecosystemsdamage category is expressed in terms of the percentage of species that disappear in acertain area due to the environmental load. The units are expressed in species per yr(PDF m2 yr). The category indicators for ecosystems are: Climate change on ecosys-tems, terrestrial acidi�cation, freshwater eutrophication, terrestrial ecotoxicity, freshwaterecotoxicity, agricultural land occupation, urban land occupation, and natural land trans-formation. Finally, resources are expressed as the cost to society due to extraction of theresource.

Data shows clearly that the traditional durable process technology has greater po-tential impact across all damage categories, production scales, and almost all of the unitoperations than the single use alternative.

Again, the majority of the life cycle environmental impacts occur during the Usephase, in which the single-use process train exhibits 39% lower impact on ecosystems,38% lower impact on human health, and 40% less impact on resource compared to atraditional durable process train.

In the ecosystem damage category, the impact of climate change on ecosystems showsthe greatest contribution. The human health damage category is mostly impacted fromclimate change in human health, particulate matter formation, and human toxicity. Fi-nally, the resources damage category is dominated by impacts from fossil depletion.

Coherently, the dominant impact categories for each damage category are related toenergy generation and use. This result is due to the large amount of energy requiredfor operation of equipment, generation of water-for-injection (WFI), and clean steamassociated with traditional Clean-in-place/Steam-in-place support processes.

This study has shown that a shift from traditional durable process technology tosingle-use process technology can result in substantial reductions in a broad range ofenvironmental impact categories. For the conditions explored in this study, the single-useprocess technology exhibits lower environmental impact compared to traditional durableprocess technology in all environmental impact categories studied at each of the productionscales under consideration. Although single-use process technology introduces a need forthe production, distribution, and disposal of single-use components, this approach alsoreduces or eliminates the need for large quantities of steam, process water, and WFI.

2.3.5 Conclusions

As sustainability concerns become more prevalent in business decision-making, the insightspresented in this study can be helpful in planning or retro-�tting either single-use ortraditional biopharmaceutical manufacturing facilities, as well as providing guidance forenvironmentally conscious product and process development for the biopharmaceuticalmanufacturing industry [36] [30] [37].

27


2.4 Case studies

Stainless steel is still a forced choice at some scales, but concrete examples of fully ded-icated and discrete, disposable manufacturing train are reachable [38]. It has been esti-mated that designing a new production facility based on single-use systems can reducecapital costs by up to 40% compared to a conventional hard-piped facility [19]. The adop-tion of single-use technologies has been growing year on year since 2004 and disposablesare predicted to reach a 20% share of the bioprocess technologies market within the next3−5 years. Already, up to 90% of manufacturers and contract research organizations usedisposable �lters and tubing at some stage of the development pipeline, whereas 77% usedisposable bioreactors and 58% use disposable membrane adsorbers [21].

More recent data, extracted from a 2014 survey conducted by BioPlan Associates,65.6% of clinical scale manufacturers and 42% of commercial scale manufacturers havecited implementation of SU bioreactors for new facilities as a major factor that haveresulted in improvements in their bioprocessing. According to the same report, the SUupstream bioprocessing market is expected to grow by >320% in �ve years [2] [39].

Biosolve analysis, an alternative to SuperPro, shows that operating costs per gramof mAb for a SU facility compared to a stainless-steel facility are 22% lower and this isprimarily due to less labor, utilities, maintenance, and waste [40].

In a case study presented at Cambridge Healthtech Institute's PepTalk conference,annual savings of 250.000 � in WFI (water for injection) generation costs and 60.000 � inlabor time for set-up and cleaning stainless steel tanks were realized in a clinical plantthat was retro�tted to implement disposable bulk freeze containers and bu�er hold bags[41].

When looking at DSP improvements, chromatography resins make up a large portionof the costs so changing a polishing step from resin chromatography to a SU membranecolumn can signi�cantly reduce expenses, as demonstrated by a Biosolve Process modelthat evaluates commercial mAb production. In this study of 1000L and 5000L scalesand various mAb titers, the unit operation cost is 19% to 33% lower for the membraneprocess and bu�er volume is decreased by up to 55% [42].

In a whitepaper published by Biopharm Services Limited, modeling software showsthat building a large-scale mAb process with thirty 2000L disposable bioreactors has atotal capital investment of 250M� which is a signi�cant reduction from the 352M� capitalinvestment required for a stainless-steel facility with the same capacity [43].

In another case study that compares the costs for a SU versus MU 2 × 1000L newfacility, the SU facility saves 11 Me annually in capital investment and only costs 1Me more in operating costs. This model shows that signi�cant CapEx reductions resultfrom lower engineering costs (decreased by 83%) and instrumentation costs (decreasedby 37%) and the increased running cost was in�uenced mainly by a higher consumablescost (increased by 51%) [44].

SU operations are smaller and can be collapsed through advanced systems, allowinga space required reduced by 45%, labor by 21%, electricity by 30%, CO2 emissions by25.5%. Water usage is reduced by 87%, chemicals (caustics and acids) required for vesselscleaning are eliminated for an overall reduction of more than 95% of cleaning materials.Capital cost drops by 67%. An overall combination of these improvements causes a 32%

28


reduction of the CoG to 104.8 $/gram of mAb, reaching for brand new plants valuesaround 85 $/gram [2] [29].

In conclusion, some biopharmaceutical facilities use a mixture of stainless steel andsingle-use systems, few use fully disposable systems. But what is the optimal mix of bothwhen starting a campaign to expand manufacturing capacities? A thorough businessanalysis must be performed in all cases. Computer-aided process design and simulationtools facilitate analysis and evaluation of process alternatives and assist scientists andengineers in their decision-making process.

29

Chapter 3

Economics

As foreseeable, pro�t is the parameter commonly chosen for optimization. Di�erent ap-proaches are followable, resulting in di�erent functions: cost of goods (CoG) analysis,life-cycle cost (LCC) analysis (or net present cost analysis). Cost of goods manufacturedis based on the amount of work-in process completed. This work-in process includescosts of direct materials put into production plus direct labor and overhead. CoG is animportant consideration in evaluating single-use systems and conventional stainless-steeldesigns, as shown in Figure 3.1.

Figure 3.1: Attempt to balance the factors

LCC analysis recognizes the total cost associated with a facility, including constructioncapital and operating costs for a green�eld facility, assuming 10− 20 years of operation.

Lifecycle costs applied to single-use systems vary depending on how these are imple-mented, and also with the location of the facility. Capital, labor, and utility costs varygreatly between the United States, Europe, and Asia. A single-use application that re-duces labor hours by 20% may make economic sense in one location but not in another.Similarly, capital savings associated with single-use systems will depend on the geographiclocation of the facility, and whether the equipment is sourced locally or imported.

30

3 � Economics

Generally, less complicated single use systems such as simple storage bags for mediaand bu�er have more favorable lifecycle economics than for more complex applications.This is because the low replacement cost of single-use components compare favorably withthe cost to maintain and clean stainless-steel equipment. Lifecycle economics for morecomplicated single-use applications such as bioreactors and fermenters is less clear due tothe high cost of the single-use components, but, properly conducted, helps determine theoptimum mix.

3.1 Strategy Proposal

Furthermore, a strategy for process economy calculations has been proposed by the Gen-eral Electric Company [45], which follows the following steps:

� De�ne scope/objectives

� Collect input data � identify di�erences and similarities

� Make assumptions

� Identify cost categories to investigate

� Make calculations

� Analyze outcome

In our case study, the objective would be the estimation of the batch production costof both stainless steel and single use. For the second point, the achievement will not beso smooth and granted. An extensive review on the economics of mAb manufacturing ishindered by the lack of peer reviewed articles because many manufacturers do not sharethis sensitive information, except at conferences [2].

However, literature is generous regarding the theoretical di�erences between the twocases, as already vastly shown previously, and summoned again for easy recalling in Figure3.2.

Figure 3.2: Photo supported SS-SU comparison

31

3 � Economics

General assumptions could follow such trail:

� Number of available days/year (e.g. 7920 h/y)

� Cost of labor (e.g. 150USD/hr/man)

� Number of labor shifts performed per day (e.g. 3 shifts/d)

� Batch failure rate (e.g. zero)

� Depreciation model (e.g. linear, spread over a 5-year period)

Observe that, when using a plant simulator, most of these choices are automaticallymade by the software and, since it is pursued a comparison, usually there are no reasonsto alter them. Consequently, unit operations with identical needs can be excluded fromthe model, if not necessary for the completion of the recipe. For example:

� Seed train procedure in shaker �asks

� Type and amount of medium components

� Minor hardware such as scales and tube welders

� Minor disposables such as pump tubing, syringe �lters, vials and similar

The choice of cost categories develops parallelly to what explained in the initial partof this section. Again, here are proposed some choices made by GE in the ongoing casestudy:

� Capital investments

� Installation and operation quali�cations (IQ/OQ), performance quali�cation (PQ),and cleaning validation

� Production related costs:

� Preparations prior to fermentation

� Fermentation process in the production facility

� Disposables, chemicals, water for injection (WFI), steam and similar

� Annual requali�cation and maintenance

After the whole setup explanation, it is useful to report some results:

� single-use equipment enables higher throughput in both types of facilities

� doubled production capacity enabled in multi-product facilities with single-use equip-ment

� stainless steel cost is higher for capital investment, quali�cations, annual mainte-nance and requali�cation, while single-use cost is higher for consumables (dispos-ables, facility media, chemicals, etc.)

32

3 � Economics

� Single-use equipment is advantageous if facility utilization rate is low or when a highproduction capacity is needed, while stainless steel equipment is advantageous atmid-facility utilization rates and when capacity is not a limiting factor

� Savings reach various �elds:

� Facility costs

� Footprint

� Facility build-out time

� Equipment cost

� Labor cost

� Cycle turnover time

� Water, chemicals and energy

Wise suggestion is to start early with process economy, in parallel with process de-velopment. A rational strategy for process economy calculations, integrated in processdevelopment, can be revolutionary.

3.2 Case Studies

3.2.1 GE Healthcare

Literature o�ers multiple case studies that responded promptly to the emerging technol-ogy. Most of them are �nanced or even conducted by the parts directly involved. There-fore, an example, which not casually follows the just proposed pattern, indeed broughtagain by GE Healthcare, comes in handy [46]. Observe that this is a bacterial fermenta-tion, therefore not every assumption here unraveled can be extended to our case study.The speci�c objectives for the investigation were the following:

� Investigation of e�ects of the equipment choice on the production capacity.

� Estimation of batch production cost for processes in which either stainless steel orsingle-use equipment was used.

� Understanding of how the equipment strategy a�ects the total annual cost at dif-ferent batch throughputs.

� Assessment of the pro�t opportunity for the di�erent equipment strategies.

The following assumptions were made:

� 300 days per year available for fermentation. The remaining time is dedicated forannual maintenance. Capital investments (including 10% interest) and quali�ca-tion costs are spread over the number of batches that can be produced during thedepreciation period (10 years) of the equipment.

� A cost of 100USD/man− hour.

33

3 � Economics

Figure 3.3: Production schedules for a multi-product facility using either (A) stainlesssteel equipment or (B) single-use equipment

� The fermentation is assumed to run over night with two operators present to monitorthe process (same for all scenarios). Labor is performed in two shifts: one from 6am to 2 pm and one from 2 pm to 10 pm.

� The batch failure rate was not considered.

� For single-product facilities, it was assumed only one product was produced and theproduction capacity of 300 days was utilized to 100%.

� For multi-product facilities, it was assumed the facility could produce di�erent prod-ucts and the production capacity of 300 days was utilized to 100%. Each productwas assumed to be produced in campaigns of �ve batches.

To avoid tedious repetitions, the results are anticipated:

� Batch produced with single-use fermentation equipment will take 33% less time tocomplete compared with when using stainless steel equipment.

34

3 � Economics

� For the described multi-product facility, the production capacity can be doubledwith single-use equipment compared with stainless steel equipment. In Figure 3.3the time dedicated to maintenance (red) is equal in both scenarios, whereas thetime dedicated to cleaning and associated analyses (yellow) is less with single-useequipment. The time needed for carry-over calculations, reporting, and qualityassurance (QA) approval (blue), included in the stainless-steel scenario, is omittedin the single-use scenario. If �ve batches are harvested (green) each campaign, 67batches can be harvested annually with stainless steel equipment versus 135 annualbatches with single-use equipment.

� The total cost per batch was calculated within six main categories:

� Capital investments

� Quali�cation

� Annual maintenance and requali�cation

� Production preparations

� Production (fermentation)

� Consumables (disposables, facility media and chemicals)

The total cost per batch is higher for the single-use processes: 29% higher in thesingle-product facility and 25% higher in the multi-product facility. The higherbatch cost with single-use equipment is due to the increased cost for consumables.However, the capital investments, quali�cation costs, and annual maintenance costsare higher for stainless steel, which can be expected as a stainless-steel facilityincludes a larger amount of �xed infrastructure in comparison with a single-usefacility.

� At low utilization rates, data show that a single-use strategy is more bene�cial froman annual cost perspective, with costs reduced in a range of 10% to 27%. The mainreason for the lower cost for single-use equipment is less time spent on equipmentquali�cation, since the annual cost for maintenance of stainless steel equipment wasshown to be 21 times higher than the corresponding cost for single-use equipment,as maintenance cost remains constant regardless of equipment utilization rate. Asthe utilization rate increases, however, the di�erence between the stainless steel andthe single use strategies is leveled out, becoming even favorable for stainless-steelscenario, until the capacity becomes a limiting factor. Indeed, the shorter batchcycle time of the single use case allows increasing the number of batches producedper year and therefore increasing productivity.

� The category that drives most of the cost for the single-use scenarios is the costfor consumables including the disposable fermenter bag and the mixing and sterile�ltration consumables. However, when looking at the �xed costs, including capitalinvestment, annual maintenance, and quali�cation costs, the cost burden is higherfor stainless steel equipment. The �xed costs will remain whether the facility is inuse or not, whereas the variable consumable costs only occur when the facility is inuse for production.

35

3 � Economics

� The pro�t opportunity is higher for the single-use alternative. The main reason forthis outcome is the increased batch throughput, which bene�ts essentially outnum-ber the slightly higher production cost per batch for the single-use scenario.

In comparison with a stainless-steel strategy, this study shows that the annual pro-duction capacity can be increased with up to 100% with a single-use strategy due to afaster batch changeover procedure. In other words, a substantial amount of time can besaved by using single use equipment instead of stainless steel equipment. The increasedproduction capacity with single-use equipment means that a de�ned number of batchescan be produced in shorter time, for example, in a manufacturing campaign or duringprocess development. The increased batch throughput, along with process �exibility andreduced cross contamination, also generates a greater pro�t opportunity, which bene�tscan outnumber the higher production cost per batch associated with the single-use al-ternative. With the decreased �nancial risk with single-use equipment, the business casebecomes more agile in comparison with stainless steel equipment associated with higher�xed costs. All these aspects are even more pronounced for a multi-product facility.

The appeal of these results can �nd application in the possibility to conduct moreexperiments, as well as allowing poor therapeutic candidates to be eliminated. However,the publisher cared for precising that these results are slaved to a single case study, sincedi�erent assumptions and conditions could have a signi�cant impact on the outcome.

3.2.2 A typical MAb economic model

To enable broad, global access to life-saving biopharmaceutical products, our industry isfacing signi�cant pressure to reduce the overall cost of manufacturing and enable localmanufacturing where possible. Development of high-titer, high-yield processes, has pres-sured to a shift in the industry's approach to facility design and construction. Biophar-maceutical production facilities must be �exible, cost e�ective, and readily constructedwith minimal capital investment and construction time lines [40].

Here we present an economic model for a typical monoclonal antibody (MAb) platformproduction process along with a detailed comparison of the operating costs for that processin facilities designed and built using either conventional reusable stainless-steel equipmentor single-use technologies to the maximum extent possible.

The estimation of the operating costs for the two model facilities includes materialsused, in the form of the cost of cell culture media, bu�er components, process water(PW), and water for injection (WFI). Prices are based on catalog prices for standard,commercially available versions with anticipated volume discounts.

For consumables, the average price for each item is used, based on standard pricingfrom di�erent vendors, again allowing for volume discounts. And it is used the amortizedcost of chromatography media, considering the number of cycles for which each columnwould be used before being repacked.

Labor costs are based on average US labor rates for operators, supervisors, and qualityassurance and control (QA/QC) sta� in the New England area.

Maintenance costs include stocking spare parts and labor associated with equipmentupkeep as well as the overall cost of maintaining a facility in a clean and GMP-readycondition.

36

3 � Economics

Utility costs include electricity to operate a facility as well as the cost of natural gasrequired for heating and steam generation.

All costs associated with handling and disposal of contaminated and uncontaminatedaqueous and solid waste are included under waste, along with the cost associated withdisposing of plastic bags and other single-use components.

The cost advantage of using disposable bioreactors is evident in about 22% savingsin total operating cost per batch for a single-use facility compared with a stainless-steelfacility. Comparing detailed costs for di�erent components of the overall manufacturingcosts for these two model facilities, those associated with materials, labor, and mainte-nance/utilities/waste for the single-use facility are lower than those for the stainless-steelfacility, but the consumables cost is higher.

Material cost savings for the single-use facility are driven primarily by savings asso-ciated with eliminating bioreactor cleaning and sterilization. A single-use facility alsoeliminates PW, WFI, and chemicals such as phosphoric acid and sodium hydroxide forcleaning stainless steel bioreactors, which generates a signi�cant cost savings. That comesprimarily from the di�erent size and types of PW and WFI systems needed for each fa-cility. In the single-use facility, a relatively small but rapid water generation source isrequired rather than the slower but much larger system for the stainless-steel facility, andthe water system must produce less water overall. But if the single-use facility were torun at its maximum capacity, it would require a faster water-generation rate than wouldthe stainless-steel facility because of the shorter duration of each batch. Therefore, thePW and WFI systems in the single-use facility would be more expensive per liter of waterproduced.

The consumables cost di�erence for these two model facilities is entirely attributableto the cost of disposable bags. Single-use systems for media and bu�er preparation andstorage � in addition to the disposable bioreactor bag systems used in cell culture for eachbatch of monoclonal antibody � yield about 40 000 � higher consumable costs per batchfor the single-use facility. If the single-use facility runs at maximum capacity (limited toone failed batch per year), then the cost of disposable bag systems per gram of productproduced would be lower.

The lower maintenance/utilities/waste costs for the single-use facility are primarilyattributable to elimination of maintenance costs associated with stainless steel bioreactors.Also, about 20% of the cost savings in this category come from reduced use of clean steamand power for cleaning and sterilizing those bioreactors. Of note is the almost negligiblecost di�erence for waste disposal between these two model facilities. That is because thecost of disposing larger amounts of solid waste associated with a single-use facility is o�setby the disposal cost of the larger quantities of liquid waste generated by the stainless-steelfacility.

Those results are consistent with a comprehensive study of the environmental impactof single-use systems, which showed signi�cant reductions in energy demand and wateruse for media preparation and bu�er preparation as well as cell culture in a single-usefacility compared with a stainless-steel facility [47] [36]. The conclusion is that a single-use facility is about 50% less energy intensive due to signi�cantly lower consumptionof energy otherwise required to heat large volumes of water for cleaning and sterilizing

37

3 � Economics

stainless steel equipment [48]. In their study, the cost of plastic bag disposal in a single-usefacility was o�set by the energy recovery gained through their incineration.

As noted above, we restricted productivity of the single-use facility in our model to thesame number of batches per year as the stainless-steel facility. However, a single-use facil-ity with the same size bioreactors as a stainless-steel facility can produce a higher numberof batches each year. Time saved by eliminating bioreactor cleaning and sterilization �along with shorter turn-around times � allows for completion of 20 batches/year in amodel single-use facility, for a total production rate of 40 kgMAb/year. This large num-ber of batches e�ectively lowers the operating CoG for the single-use facility to 170 �/gwhen it operates at full capacity � a much larger cost savings for the use of disposablebioreactors and other such systems.

One key potential bene�t derived from implementing single-use technologies in bioman-ufacturing operations is a reduction in labor requirements for supporting ongoing GMP.Such reductions come primarily from lowered requirements for equipment cleaning be-tween production batches and associated ongoing cleaning validation/veri�cation. Thosereduced cleaning requirements can not only lower headcount for direct manufacturingoperations, but they may signi�cantly reduce sta�ng requirements for QA/QC as well.Also, shorter change-over times and fewer hard-piped utilities and systems in a facilityincorporating single use technologies will lower maintenance and metrology costs, reduc-ing sta�ng requirements for those areas as well. A single-use facility requires about 15%fewer manufacturing sta� for drug substance manufacturing and about 12% fewer QA/QCsta�, resulting in about 13% lower total headcount than a stainless-steel facility. To bethorough, if the single-use facility were to run at full capacity, then its total headcountwould be slightly higher, but so would be the total amount of product per year.

Sta�ng reductions are consistent with other reports of reduced labor requirements forfacilities based on single-use technologies. For example, labor savings are estimated fortechnology transfer and ongoing production of a MAb product in a single-use facility at56% for technology transfer and 17% for ongoing production compared with conventionalstainless-steel facilities. Furthermore, a recent presentation showed that about 75 laborhours/batch could be saved using disposable prepacked chromatography columns insteadof reusable, self-packed columns. Those prepacked columns reduced the total cost of eachchromatography step some 20 000 � [49] [50] [51] [52] [53].

3.2.3 Simbiopharma Approach

The case study addresses whether start-up companies should invest in a stainless-steelpilot plant or use disposable equipment to produce early phase clinical trial material.Simbiopharma, a prototype tool for simulation, captures both the technical and busi-ness aspects of biopharmaceutical manufacture within a single tool that permits manu-facturing alternatives to be evaluated in terms of cost, time, yield, project throughput,resource utilization, and risk. Incorporating the e�ects of risk would help to enhance thequality of decision-making within a company. Uncertainties a�ecting the manufacture ofbiopharmaceutical candidates are technical and market-related. Examples of such tech-nical uncertainties include the product titer during fermentation, the puri�cation yield,and the duration of the manufacturing tasks. Market uncertainties can be characterized

38

3 � Economics

by �uctuating clinical trial demands for material. However, this aspect will be neglectedsince evading from our goal and being not necessary for the sake of the example.

The use of disposable equipment is attractive since it allows for a lower initial capitalburden and a potentially more balanced spread of costs over a plant's life, particularlywhere the risk of failure during clinical trials is higher. Disposable processing componentsmay become even more important as personalized medicines become more common tomatch treatment to individual genomes.

The assumptions will be neglected, since are quite common. The only note comesfrom the application �eld, which in this case is restricted to the clinical trial materialpreparation.

The method for calculating the �xed capital investment required was based on mul-tiplying the total equipment purchase cost by a Lang factor. For the stainless-steel case,a Lang factor of 7 was assumed; the multiplying factor suggested for a disposables plantwas 4; for the hybrid case, a value of 5 was used, based on the ratio of stainless steel todisposable equipment. The cost inputs were determined from literature or vendor sources.

Figure 3.4: Annual cost of goods per gram (COG/g) on a cost category basis for thestainless steel (SS) pilot plant, disposables based (DISP) pilot plant, and hybrid (HYB)pilot plant, each assumed to produce the same yield per batch. The costs are relative tothe baseline stainless steel case.

From the tool, it is possible to view the operating costs on a category basis such asmaterials, utilities, sta�, �xed overheads, and depreciation charges. Figure 3.4 shows thecost of goods per gram in the �rst year of operation relative to the stainless-steel case.Comparing the total annual cost of goods per gram for the three cases, the disposablesand hybrid plants are seen to o�er signi�cant reductions (30% and 19%, respectively)in operating costs relative to the conventional stainless-steel case. This is due to thehigher number of projects completed relative to the conventional case, hence increasingthe annual gram output, as well as associated drops in certain cost categories. Examiningthe stainless-steel case indicates that the COG/g is dominated by the �xed overhead costs(maintenance, local tax, insurance and general utilities) and depreciation charges, whichrepresent 67% of the cost. These costs are proportional to the capital investment required,which is highest in the stainless-steel case. By contrast, in the disposables and hybrid

39

3 � Economics

cases these �xed costs fall by at least a third of the conventional value. The sta� andutilities costs are lower in the disposables and hybrid cases owing to the smaller numberof ancillary operations such as cleaning and sterilizing equipment.

Figure 3.5: Annual direct cost of goods per gram (COG/g) on a task basis for the stainlesssteel (SS) pilot plant, disposables-based (DISP) pilot plant, and hybrid (HYB) pilot plant,each assumed to produce the same yield per batch. The most expensive tasks are the CIP'sin the stainless steel case and the capture chromatography steps in the disposable andhybrid plants.

Figure 3.5 shows the direct cost of goods per gram for each of the manufacturing tasksfor a pipeline of projects that were completed within the �rst year of plant operation.This re�ects the cost of the direct materials, utilities, and sta� allocated to each task,thereby providing the capability to view where these resource costs are concentrated. Inthis hypothetical case, the cleaning-in-place (CIP) procedures in the stainless-steel caseand the capture chromatography steps in the disposable and hybrid cases consume themost resources and are hence the most expensive. For the disposable units, the majorcontributor is the Protein A chromatography. This implies that the process developmentteam must have good processing reasons to defend the use of such an expensive step ina disposable manner. However, some companies use their matrices in a product-speci�cmanner to reduce the cleaning validation studies required to demonstrate the number ofcycles that the material can be used with con�dence with the performance not deteriorat-ing. In such cases there would be no di�erence in the cost of the Protein A chromatographystep between the stainless-steel case and the disposable option. From �gure, it is appar-ent that the disposable and hybrid options have signi�cantly higher direct costs than thestainless-steel case, owing to the higher material demands. However, since the indirectcosts are proportional to the capital investment, the stainless-steel case has signi�cantlyhigher indirect costs and hence higher overall operating costs as illustrated earlier [20].

40

3 � Economics

3.2.4 Biomanufacturing Comparison

Two technologies are used in biomanufacturing: single-use disposable bag technologyversus traditional �xed stainless-steel vessels. The study was based on a commerciallyrelevant monoclonal antibody process, run at 2000L scale, exploiting information availablein the public domain, using discrete event modeling. This allows simple what-if analysis,available on-line to the user through a friendly interface. The objective is to minimize thecapital investment required for a speci�c throughput.

The major concerns for biological manufacturers are:

Safety Today the main GMP (Good Manufacturing Practice) de�ciency reported frombiopharmaceutical plant audits is linked to cross contamination, which represents15% of total de�ciencies

Lack of �exibility plants are designed for speci�c product processes

Maintenance Biologicals plants are extremely complex, requiring expensive mainte-nance

Long construction time Average construction time is 2− 3 years or more followed byextensive validation - the commissioning phase may take several months. Companiesmay lose the battle to get their drug on the market ahead of the competition

Lack of production capacity in relation to the number of biologicals in the pipelinethe industry is facing a serious capacity shortage problem over the next 5 years

Capital intensive some of the sums spent on new biologicals facilities represent morethan 1 billion e

Process costs One of the biggest costs in biomanufacturing is the cost of transfer ofsterile �uids (such as product and reagents) through di�erent process steps locatedin di�erent parts of the facility: traditionally, the logistics of �uid transfer hasbeen handled through product piping, stainless steel vessels, routing manifolds andvalves. All this equipment must be cleaned and sterilized. Equipment validationis required before re-use. As an example, a typical CIP (Clean In Place) cycle forvessels ranging between 100L to 1000L can take between 90 to 150 minutes. Ifthe process is classed as totally sterile, vessels need to go through an additionalSIP (Steam in Place) cycle, which with steaming, vessel cooling and hydrophobic�lter integrity testing may take an additional 3 hours, possibly more. Traditionalequipment preparation and validation for sterile �uid handling is extremely timeconsuming; meaning production capacity is not being optimized.

Single use technology basically consists in substituting vessels with bags. The bagsystems are provided pre-assembled, sterile and pyrogen-free, ready for process-use. Thesystems can be customized according to customer speci�cations, designed for space sav-ing and ease of maneuverability around the facility. The bag systems are manufacturedaccording to GMP and are tested according to US and European Pharmacopoeias, bio-compatibility testing is carried out according to ISO 10993 − 1, Biological Evaluation

41

3 � Economics

of Medical Devices. Chemical compatibility testing of bags and solutions is carried outaccording to ASTM D1239− 98, Standard test method for resistance of plastics �lms toextraction by chemicals.

The impact of disposable in the process ca be outlined as:

Capital Investment

Equipment Reduced requirement for vessels for media, bu�er and product hold, reducednumber and size of CIP skids. Reduced utility systems capacity. Containers for thebags and tube welders will be required.

Reduced �oor area in the facility The requirement for contained areas is reducedthrough use of tube fusing systems to make aseptic connections from bag to bag.

Validation Requirements IQ/OQ/PQ time required for disposable technology in com-missioning phase of a new facility is considerably less than time required for tradi-tional equipment.

Operation

Productivity Disposable bag technology allows immediate equipment turnaround withno cleaning, sterilization and revalidation time required and no downtime whilstvessels are cleaned.

Utility Consumption With disposable technology, there is no cleaning; so, a reducedrequirement for utilities such as WFI, PW and clean steam (only required for speci�cprocess equipment) and reduced requirement for CIP chemicals.

Maintenance Stainless steel vessels require regular maintenance, which is not requiredwith bag technology.

Validation Reduced annual validation requirements required for CIP & SIP.

Labor There is reduced labor requirement, as is shown in the model.

Flexibility and e�ciency in process The size of the bags can be adapted to a varietyof batch sizes matching make up of bu�ers to requirements with no waste. Bu�erscan be available just in time.

Safety The bag technology functions in a closed circuit (no need for an air-vent �lter)moving towards the total-containment concept that the FDA prefers. As a single-usesystem bag technology eliminates the risk of cross contamination.

Some limitations can be associated to bag technology. The main are: a maximumvolume boundary which is quite low and the mixing issues, related to mass and heattransfer, increasing with the volume.

42

3 � Economics

The method of comparison for this case study is a process simulation tool, whoseobjectives are:

� Optimization of capital requirements by maximizing asset utilization

� Measuring production rates

� Optimizing production

Quantitative data was generated for:

� High quality utility systems, water and steam

� Assessing size of utility generator

� Assessing storage capacity requirements

� Calculating consumptions

� Measuring production rates

� Evaluating CIP Requirements

� Flexible bag model

� Number of the di�erent bag types that are used per batch

� The maximum number or bags in use at any one time

� The size of the utility systems required to support bag operation

� Fixed vessel model

� Identify the size and number of �xed vessels required to run the plant

� The size of the utility systems supporting �xed vessel operation

At the end of a simulated operation the process simulation model automatically trans-fers key performance data to the COG spreadsheet: capital requirements (for CIP, utilitysystems, number and size of hold vessels, number of disposable bag containers); materialconsumptions per batch (all critical utility consumptions, CIP chemical usage, number ofdisposable bags); production rate.

Furthermore, a priced equipment list for the major items is generated from an internaldatabase of costs built up from recent projects (both US and European) and vendorinformation. Capital estimations are obtained through proper Lang factors; material onesinclude requirements for consumables, indirects and materials themselves. Labor headcount is evaluated within the COG model in the process simulation, where it is availablethe needed information regarding the allocation time.

Table 3.1 shows that the reduced requirement for CIP (Clean in Place) resulted inreduced consumption of utilities and chemicals per batch by the disposables sub model,however o�set by an increased consumption of plastic disposable bags. These tangible

43

3 � Economics

Material Usage/Batch Bags Vessel Vessel - Bags

WFI 19 500L 160 500L 141 000LPuri�ed Water 34 500L 240 000L 205 500LSteam 33 kg 181 kg 148 kgDil. caustic for CIP 450L 2580L 2130LDil. acid for CIP 450L 2580L 2130L

Table 3.1: Utilities consumptions

di�erences have economic and environmental implications both for new and existing in-stallations:

� CIP skid requirement (to match the reduced CIP requirement)

� Number of bag holders required containing in process disposable bags

� Speci�c bag processing equipment (i.e. tube welders)

� Reduced �oor area requirements (stacking bag systems minimizes �oor area requiredto hold and process bu�ers)

Capital Estimate

Stainless Steel Vessel Sub Model 24 970 532 eDisposable Bag Sub Model 19 670 355 eOverall Capital Saving 21.2 %

Table 3.2: Investment cost comparison

Table 3.2 shows a signi�cant reduction in the capital requirement for the single usedisposable bag production line compared to that required for the one based on stainlesssteel vessels. This is a result of a reduced requirement for CIP by the disposable bagsub model. For example, the WFI generation capacity required for the disposable bagsub model is around 400L/hour however the stainless-steel vessel sub model requires1500L/hour, saving around 5 Me.

The question remains as to whether this reduction in capital results in a reduced costof goods for the monoclonal antibody. In this analysis, it does. As shown in Table 3.3,capital is factored in an amortized charge: this is based upon an 8-year plant life and15% cost of capital and 5% residual value in the plant after the 8 years. In addition,for the disposable bag sub model there are higher consumable costs resulting from baguse. These are o�set by utility savings and savings in quality costs. In this analysis, itestimated that the quality head count would reduce by 5 for the disposable sub modelcase, resulting from a reduced ongoing cost requirement for Clean In Place and Steam inPlace annual validation and reduced paper work. The overall impact is a saving of 9% onthe cost of goods resulting from a new plant installation based on disposable bags whencompared to one based on stainless steel vessels.

44

3 � Economics

Cost of Goods e/batch Vessels Bags Vessels - Bags

Fermentation materials 26 048 26 048 0Fermentation consumables 46 069 48 214 −2144Puri�cation materials 23 294 23 294 0Puri�cation consumables 29 422 41 440 −12 018Engineering spares 2402 2402 1Sub total 127 236 141 397 −14 162Direct fermentation 41 707 41 695 11Direct puri�cation 8604 8604 0Plant overhead 38 620 38 610 10QA/QC 56 304 45 043 11 261Sub total 145 234 133 952 11 282Indirect materials 9056 9054 2Utilities 33 750 26 579 7171Sub total 42 806 35 633 7173Capital Charge 210 806 166 016 44 790Total 526 082 476 998 49 084

Table 3.3: Operating cost comparison

One of the concerns about the use of disposable bag technology is the environmentalimpact of the consumption of the single use bags. The results, though, assured alsoregarding this �eld:

Disposable sub model requires approximately 100 disposable plastic bags plus asso-ciated tubing per batch amounting to around 200 kg of plastic waste. In manyfacilities, this material either goes o� site for incineration or into land�ll.

Fixed vessel sub model requires an additional 141 tonnes of Water for Injection gen-erated from 204 tonnes of Puri�ed Water. In addition, about 4.2 tonnes of diluteCIP chemicals are needed (equivalent to about 100L of 40% caustic and 5L of 80%phosphoric acid). Overall these additional materials will require treatment beforedischarge.

By balancing, the increased consumption in plastic for the bags is o�set by largereductions in water and CIP chemical requirements that also reduce capital requirementfor new facilities.

Summarizing, for a new installation disposable bag technologies will result in reducedcapital requirements of around 20%, which in turn result in an 8% reduction in the costof goods [54].

45

Chapter 4

Simulation

Process simulators are software tools that enable the user to readily represent and analyzeintegrated processes. Minimum requirements for a biochemical process simulator are theability to handle batch as well as continuous processes and to model the unit operationsthat are speci�c to bioprocessing.

Most biochemical processes operate in batch or semi-continuous mode, in contrastto continuous operations, which are typical of the petrochemical and other industriesthat handle large throughputs. Batch process manufacturing is practiced in industriesthat produce low-volume, high-value products such as pharmaceuticals, �ne chemicals,biochemicals, as just stated, food, consumer products, etc. Most batch manufacturingfacilities are multiproduct plants that produce a variety of products. In continuous op-erations, a piece of equipment performs the same action all the time (which is consistentwith the notion of unit operations). In batch processing, on the other hand, a piece ofequipment goes through a cycle of operations. For instance, a typical chromatographycycle includes: equilibration, loading, washing, elution and regeneration.

The �rst step in preparing the building of a simulation model is always the collectionof information about the process. Engineers rely on draft versions of process descriptions,�ow diagrams, and batch sheets from past runs, which contain information on materialinputs, operating conditions, etc. Reasonable assumptions and approximations are thenmade for missing data. Rough estimates used at the starting point can be countercheckedand adjusted accordingly once completed the simulation. The following information isrequired for modeling a batch process:

� Processing steps, their durations, performance parameters and sequencing

� Equipment available to processing steps

� Materials consumed by or generated by the process

� Other resource requirements (e.g., labor, utilities, etc.)

It is a good practice to build the model step-by-step, checking and verifying the resultsof the evolving model before additional complexity is added.

46

4 � Simulation

The registration of materials (pure components and mixtures) is usually the �rst prac-tical step in building the model. The �owsheet is developed by putting together the re-quired unit operations (which will sometimes be referred as unit procedures) and joiningthem with material �ow streams. Operations are then added to unit procedures and theiroperating conditions and performance parameters are speci�ed.

As deducible, unit procedures contain individual tasks called operations. A unit pro-cedure is the recipe of a processing step that describes the sequence of actions requiredto complete that step. The program displays the operations that are available for theselected unit and the operator must select the proper ones according to the project. Thesigni�cance of the unit procedure is that it enables the user to describe and model thevarious activities of batch processing steps in detail. Furthermore, for every operationwithin a unit procedure, SuperPro, the name of the simulator selected for our elaborate,includes a mathematical model that performs material and energy balance calculations.Based on the material balances, it performs equipment-sizing calculations reconciled witheach speci�ed demand and appropriate for all operations. A vessel, for example, is sizedso that it is large enough that it will not be over�lled during any operation, but it is nolarger than necessary, to minimize capital costs. To accomplish so, before any simulationcalculations can be done, the user must initialize the various operations by specifyingoperating conditions and performance parameters through appropriate dialog windows(for example: operating conditions, labor, scheduling). Through scheduling, one speci�esthe start/end time of the operation relative to the start/end of another operation in thesame procedure, or relative to an operation in another procedure. After the correct ini-tialization of the operations, the simulator performs material and energy balances for theentire process and estimates the required sizes of equipment. Alternatively, the user cande�ne the sizes and SuperPro can output the productivity of the plant. Optionally, thesimulator may be used to carry out cost analysis and economic evaluation calculations.

The outputs of batch process simulators include the following:

� Visual representation of the entire process

� Material and energy balances

� Sizing of equipment and utilities

� Estimation of capital and operating costs

� Process scheduling and cycle time analysis

� Throughput analysis

� Environmental impact assessment

As shown, not-so-intuitive tasks that can be handled by process simulators includeprocess scheduling, environmental impact assessment, debottlenecking, and throughputanalysis. In throughput analysis and debottlenecking, the engineer analyzes the capacityand time utilization of equipment and resources (e.g., utilities, labor, raw materials) andtries to identify opportunities for increasing throughput with the minimum possible capitalinvestment. Once achieved a satisfying model, consequent to proper validation, the user

47

4 � Simulation

may begin experimenting on the computer with alternative process setups and operatingconditions, leading to potential cost and time reductions. A review of the results by anexperienced engineer can play the role of validation, and it will be the case for our work[55].

Scheduling and cycle time analysis in the context of a simulator is fully process-drivenand the impact of process changes can be analyzed in a matter of seconds. For instance,the impact of an increase in batch size (which a�ects the duration of charge, transfer,�ltration, distillation, and other scale-dependent operations) on the recipe cycle time andthe maximum number of batches can be seen immediately. Simulation tools that allowusers to describe their processes in detail, and to quickly perform what-if analyses, can beextremely useful. For instance, if the process already operates at its maximum possiblebatch size (based on the equipment capacities), the only way to increase production is byreducing the process cycle time and thus increasing the number of batches per year. Thecycle time can be reduced through process changes or by addition of extra equipment.However, major process changes in regulated industries such as pharmaceuticals usuallyrequire regulatory approval and are avoided in practice. As a result, addition of extraequipment may be required to achieve cycle time reduction and accommodate the requiredproductivity.

A project can be developed using a platform technology approach that aims to stan-dardize the number and the sequence of the production steps as well as the media andbu�er solution used, which heavily impact these processes. All process parameters thata�ect product quality (e.g., bed height of chromatography puri�cation steps) can be �xedby the end of process development. Such process parameters can hereafter not be alteredduring the scope of this project. Instead, the focus can be on engineering parameters thata�ect capital cost and capacity (e.g., number and size of vessel for bu�er preparation andstorage, requirement for transfer line, in-line dilution, cleaning and steaming skids).

Using simulators, complex variables like labor cost versus capital investment trade-o�, since usually inversely dependent, as a function of the shifts choice, can be takeninto consideration. Moreover, storage of multipurpose raw materials, like WFI (water forinjection), consisting of a still generating distilled water, a surge tank, and a circulatingloop for delivery, can be reasonably dimensioned, in a compromise between tank and pumpcapacity, and corrected online. Sizing of bio-waste treatment systems can be handled ina similar way.

Biotechnology process development scientists have a short time window to optimizethe process for a promising new molecule. Similarly, engineering teams face challengeswithin the design and construction of new production lines and facilities required for man-ufacturing newly developed pharmaceutical products. The challenges of both groups canbe lessened using appropriate computer aids, such as process simulators and productionscheduling tools [14].

At the early stages of idea generation, process simulation is primarily used for screeningand evaluating potential projects in order to determine which ones to move forward with.The �owsheets put together during process synthesis must be analyzed and comparedon the basis of capital investment, manufacturing cost, environmental impact, and othercriteria in order to decide which ideas to consider further. Referred to the developing

48

4 � Simulation

topic, the alternatives, at various scales, regard disposable items and their spread into theplant (for example, if limited to bu�er and media storage).

In process development, simulation tools are used to evaluate alternative processingscenarios from an economic, cycle time reduction, and environmental point of view. Cost-of-goods analysis facilitates identi�cation of the critical steps and aspects of a process,and this information is used to guide subsequent research and development work. Capitalcost estimation facilitates decisions related to in-house manufacturing versus outsourcing.Again, returning to the classical example, capital and operating costs analysis can assessthe impact of single-use systems on the demand for cleaning materials, cleaning-in-place(CIP) skids, labor and utilities.

Environmental impact assessment is an activity closely related to process design andcost estimation. Biochemical plants generate a wide range of liquid, solid, and gaseouswaste streams that require treatment prior to discharge. The cost associated with wastetreatment and disposal has skyrocketed in recent years due to increasingly stricter environ-mental regulations. This cost can be reduced through minimization of waste generation atthe source. However, generation of waste from a chemical or biochemical process is depen-dent upon the process design and the way the process is operated. Thus, reducing wastein an industrial process requires intimate knowledge of the process technology, in contrastto waste treatment which essentially is an add-on at the end of the process. In addition,minimization of waste generation must be considered by process engineers at the earlystages of process development. Once a process has undergone signi�cant development it isdi�cult and costly to make major changes. Furthermore, regulatory constraints that areunique to the pharmaceutical industry restrict process modi�cations once clinical e�cacyof the drug is established. These are only some of the reasons that process synthesis mustbe considered not only during, but before, the selection of unit operations for individualsteps.

When a process is ready to move from development to manufacturing, process simu-lation facilitates technology transfer and process �tting.

Sensitivity analyses are greatly facilitated by process simulation tools as well. Theobjective of such studies is to evaluate the impact of critical parameters on various keyperformance indicators (KPIs), such as production cost, cycle times, and plant through-put.

The following features can be enumerated among the qualities of a process simulator[56]:

� Documentation and process understanding

� Calculation of material and energy balances

� Sizing of equipment and utilities

� Cost-of-goods analysis

� Process scheduling

� Cycle time analysis and debottlenecking

49

4 � Simulation

� Resource tracking as a function of time

� Environmental impact assessment

Batch process simulation and production scheduling tools can play an important rolethroughout the life-cycle of product development and commercialization. In process devel-opment, process simulation tools are becoming increasingly useful as a mean to analyze,communicate and document process changes. During the transition from developmentto manufacturing, they facilitate technology transfer and process �tting. Productionscheduling tools play a valuable role in manufacturing as well. They are used to gener-ate production schedules based on the accurate estimation of plant capacity and resourceconstraints, thus minimizing late orders and reducing inventories. Such tools also facili-tate capacity analysis and debottlenecking tasks. Finally, they are useful for performingongoing tracking and updating of the manufacturing schedule. The batch process in-dustries have begun making signi�cant use of process simulation and scheduling tools.Increasingly, universities are incorporating the use of such tools in their curricula. In thefuture, we can expect to see increased use of these technologies and tighter integrationwith other enabling IT technologies, such as supply chain tools, manufacturing executionsystems (MES), batch process control systems, process analytical technology (PAT), etc.The result will be more robust and e�cient manufacturing processes [57].

The completed detailed model, which constitutes at all e�ects a virtual plant, can behanded over to the operations team to help in preparing the personnel for the start-upof the plant and its routine production schedule. Furthermore, a computationally lightermodel can be used to:

� Plan the activities during the start-up of the new production facility

� Analyze the bottlenecks at full production capacity

� Analyze and schedule changeovers in a multiproduct facility (change from one pro-cess to another on a production line)

� Consider the impact of equipment maintenance on production schedule

� Analyze the in�uence of a failure or delay of one step on the following steps of abatch and on the scheduling of subsequent batches

� Understand interdependencies between shared areas and production lines

As a matter of fact, future practitioners should be aware that process simulation toolsmust be used as early as possible to achieve more successful synergies.

50

Part II

Process Analysis

51

Chapter 5

Overview

To develop the process, it has been chosen to exploit SuperPro Designer by Intelligen®.SuperPro Designer is a comprehensive process simulator that facilitates modeling, costanalysis, debottlenecking, cycle time reduction, and environmental impact assessment ofintegrated biochemical, bio-fuel, �ne chemical, pharmaceutical (bulk & �ne), food, con-sumer product, mineral processing, water puri�cation, wastewater treatment, and relatedprocesses. Its development was initiated at the Massachusetts Institute of Technology(MIT). SuperPro is already in use at more than 500 companies and 900 universitiesaround the globe (including 18 of the top 20 pharmaceutical companies and 9 of the top10 biopharmaceutical companies).

The upstream portion of this platform process begins with cell expansion from small-scale bioreactors, called seed bioreactors, followed by product production and accumula-tion in a 2000L bioreactor. Usually, for mammalian cell cultivation, the scaling factoris located around 5 : 1, since a too sudden and radical change of environment could af-fect cell growth and product concentration and quality. After cell growth and proteinproduction in the 2000L bioreactor, its contents are clari�ed through a combination ofcentrifugation and membrane �ltration unit operations to yield a clari�ed cell cultureharvest containing crude MAb product. For the single-use case, pressure obtained in thecentrifuge are not sustainable for the plastic materials. Therefore, depth �ltration is usedinstead. That product is then puri�ed from the clari�ed harvest using a three-columnpuri�cation process including protein A a�nity chromatography, cation-exchange (CEX)chromatography, and anion-exchange (AEX) chromatography. A low-pH virus inactiva-tion step follows the protein A column. A nano�ltration step follows chromatography foradditional viral clearance, after which puri�ed MAb product is concentrated and formu-lated by a combination of ultra�ltration and dia�ltration (UF/DF) to produce the �nalpuri�ed and formulated bulk product.

In the single use facility, the whole USP train and all the preparation and storage ofmedia and bu�er are disposable. Each disposable plastic bag used in those applications isheld in a portable, stainless steel system designed to hold bags of a desired volume. Sys-tems used for product and intermediate storage were also equipped with disposable mixingsystems (where appropriate) to keep the product well mixed. Single-use mixing systemstypically involve a magnetic or mechanical external drive with a disposable impeller inside

52

5 � Overview

the bag.

The process model is based on the mAb platform process. The production suite iscomposed of:

� N. 1 upstream section, equipped with 1 seed train, 2 production bioreactors (2×2K)and support equipment (media preparation)

� N. 1 media preparation and storage section, equipped with preparation and storageequipment.

� N. 1 harvest section, equipped with centrifuge/depth �ltration package, clari�cation�lter, process bag/tank.

� N. 1 downstream section, equipped with puri�cation equipment (mainly processbags/tanks, chromatography skids, Nano�ltration and UF)

� N. 1 bu�er preparation and storage section, equipped with preparation and storageequipment.

5.1 Scheduling Terminology

The terminology refers to the language used in the process model study:

SU single use technology

SS stainless-steel technology

Recipe Cycle Time the time between the start of two consecutive batches. It is alwayssmaller or equal to the Recipe Batch Time and larger or equal to the MinimumRecipe Cycle Time (see below). If a batch is started right after the previous oneis ended (but not before) then the cycle time equals the batch time. If a batch isstarted before the previous batch is ended (more typical) then the cycle time is lessthan the batch time. Note that there is a constraint as to how soon we can start abatch (while the previous is still in progress).

Minimum Recipe Cycle Time the minimum time possible between the start of twoconsecutive batches. It is equal to the longest Equipment Cycle Time amongst allequipment resources (excluding those that are ignored by the scheduling calcula-tions).

Target Cycle Time it is the cycle time to be matched in order to grant productiontarget

Maximum Number of Batches the maximum Number of Batches possible to be pro-cessed in a calendar year (without violating the available annual operating time).This Number of Batches can be achieved when the process operates under the Min-imum Cycle Time (or the Cycle Time Slack is 0.0).

53

5 � Overview

Figure 5.1: Sketch of the SU process grabbed from SuperPro.USP: black; DSP Pre-Viral: red; DSP Post-Viral: orange; Media: green; Bu�er: olive

54

5 � Overview

Target Number of Batches it is the number of batches/year based upon calculatedcycle time

Equipment Busy Time the total time that the equipment resource is active (busy)with the execution of one or more procedures hosted by it during a single batch.If the equipment resource is hosting only one procedure, then the equipment busytime is the same as the procedure time of the hosted procedure. If the equipmentresource hosts multiple procedures, then the equipment busy time is the sum of allprocedure times hosted by the equipment resource.

Equipment Idle Time the total time that the equipment resource is doing nothing (itis not hosting a procedure) during a batch. It is the sum of equipment waiting timeand equipment unoccupied time. Note that the sum of equipment busy time andequipment idle time equals the recipe cycle time (since the equipment can be either�busy� or �idle").

Equipment Occupancy Time The time spanned from the start of the �rst hostedprocedure by an equipment resource to the end of the last hosted procedure by thesame equipment resource during a single batch. Note that this may not be equalto the sum of procedure times (if multiple procedures are hosted) as there may besome waiting time between the hosted procedure engagements. If the equipmenthosts only one procedure, then the equipment occupancy time equals the proceduretime of the hosted procedure.

Equipment Unoccupied (or Available) Time The time that the equipment is notconsidered occupied during a batch (it simply waits until it is engaged by the nextbatch). It is the time between the end of the last procedure that is hosted by theequipment resource in one batch and the start of the �rst procedure that is hostedby the same equipment resource in the next batch. Note that the sum of equipmentoccupancy time and equipment unoccupied time equals the recipe cycle time.

Equipment utilization factor ratio between equipment occupancy time and recipe cy-cle time. It is expressed as percentage.

Facility utilization factor ratio between target number and max number of batches. Itis expressed as percentage. Corresponds to bottleneck equipment utilization factorProduct residence time: it is the time that the product spends into each processroom

Size Equipment Factor allows identifying the underused equipment in terms of capac-ity utilization and represents the percentage of equipment capacity that is utilizedduring a certain unit procedure

Stagger mode same sequences of operations on di�erent equipment, for example DB-410(Production Bioreactor 1), STG1 -> DB-420 (Production Bioreactor 2)

55

5 � Overview

Figure 5.2: Example of the chart analysis

56

Chapter 6

Basis of Design

The scope of this chapter is to describe the basis of design used to build the process modeland to present the outputs from SuperPro® software.

6.1 Production

The basis of design in terms of production target, annual operating time and shifts patternare:

� 48 working weeks per year

� 24 hours per 7 days of working time

� 90% facility utilization factor

� 5 g/L product concentration output from bioreactor

� 50 g/L �nal product concentration

6.2 USP

It is useful to introduce some de�nitions:

Cell Culture the process of growing cells outside of the organism (e.g. mammal, insector plant) from which they are harvested. This is accomplished by providing the cellswith water, nutrients (e.g. sugars, salts, proteins, vitamins, minerals and essentialgrowth factors like insulin and hormones) oxygen and the environmental conditions(e.g. temperature and pH.)

Bioreactor a vessel with associated control system in which a cell culture occurs.

Seed Bioreactor The �seed bioreactor� is sterilized and �lled with media, and the �in-oculum� (that was prepared in the lab) is added. Seed Bioreactor steps are repeatedat stepwise larger scales to grow the quantity of cells needed for the �ProductionBioreactor.�

57

6 � Basis of Design

Production Bioreactor The contents of the last �Seed Bioreactor� are transferred tothe (pre-sterilized, larger) Production Bioreactor. The Production Bioreactor growsthe cells to the �nal quantity needed to produce the desired product.

Scale-up ratio refers to the volume and cell count increase from one cell culture stepto the next. The scale-up ratio for cell culture is typically limited to 5 : 1 (muchsmaller than what is commonly seen in fermentation.) This lower growth ratio andslower doubling time (compared to fermentation) increases the preparation time ofinoculum in the seed lab. In addition, the lower growth ratio almost always necessi-tates the use of multiple �Seed Bioreactor� steps before moving to the �ProductionBioreactor.�

Bioreactor System enables a �seed� or �production� cell culture process to occur underaseptic conditions. It allows for the aseptic addition of media and inoculum, it mustinclude a container large enough to hold the �nal solution volume, it must have theability to control and monitor environmental conditions such as temperature andpH and it must allow for and facilitate the addition and adsorption of gasses asnecessary.

The Bioreactor System itself deserves an extensive focus:

Bioreactor Vessel sanitary pressure vessel with a dimpled or half-pipe coil jacket andba�es. L/D ratios are usually governed by gas transfer requirements and are typi-cally between 1 : 1 and 2 : 1. The bioreactor vessel typically includes sparge tubesbelow the lower agitator impeller for gas addition.

Bioreactor Agitator Typically center-mounted and often bottom-mounted. Since cellsmust be handled more gently than the microorganisms found in fermentation appli-cations, special �low shear� agitators are often used. As a result of the comparativelyslow cellular metabolism, air sparge requirements are low and this results in lowerRPM and horsepower requirements (for cell culture bioreactors, compared to biore-actors). Bottom-mounted impellers require wetrunning double mechanical seals.Clean steam condensate is usually used as the seal �ush �uid.

Temperature Control Unit The TCU provides temperature control of the bioreactorcontents. The heat load in bioreactors is typically much lower than in fermentorsdue to the lower growth rate. Heat input may be required to maintain 37 ◦C formammalian cells and lower for insect cell culture. The TCU often uses plant steamfor heating and chilled water for cooling. Single �uid systems may be used incertain circumstances. The main components of the TCU system are a pump, heatexchanger(s), piping, instrumentation, and control valves.

Chemical Addition Tanks Acid and base are sometimes required for pH adjustmentand �antifoam� solutions are sometimes also required. The bioreactor system in-cludes small addition tanks, which are sanitary jacketed pressure vessels. It is oftenmore economical to replace these tanks with single use bags.

58


(Optional) Chemical Addition Pumps Chemical additions are often made by pres-sure transfers, however peristaltic pumps or electronic diaphragm metering pumpmay also be used. Pumps are required if single use bags are used for the additions.

(Optional) Sanitary Heat Exchangers May be required to keep the exhaust fromcondensing on the vent �lter or to cool / condense the vent stream beyond the vent�lter. A heat exchanger may also be required to condense clean steam to �ushagitator mechanical seals.

Gas / Vent Filters Sub micron-rated sanitary cartridge �lters on gas addition andvent/exhaust lines are used to maintain sterility during operation of the bioreactor.The vent �lter may be steam jacketed or electrically traced to prevent condensationfrom blinding �lters.

Gas manifold system Required for metering in the proper gas ratio to maintain dis-solved oxygen and pH.

Sampling Device A sampling device is typically provided for on-line sampling duringthe cell culture process. This sampling device is typically on the side of the bioreac-tor, and is equipped to provide both a sterile sample as well as maintain the sterileboundary of the bioreactor system.

Instrumentation such as pH, conductivity, temperature, pressure, O2/CO2 concentra-tion, dissolved oxygen, level and/or load cells, o�-gas analysis, gas �ow meters

The Bioreactor System works following these steps:

� Bioreactors are typically steam sterilized empty prior to charging media. Dependingon the design, supplies of supplemental media and/or bu�er may be connectedto the bioreactor and the connections sterilized with the empty bioreactor vessel.Otherwise, they can be connected during the cell culture by steam docking or asepticconnectors.

� The seed bioreactor vessel is brought to the culture temperature.

� Inoculum (the initial charge of microorganisms) is sterile transferred into the seedbioreactor and the culture begins.

� During the cell culture, media (including �nutrients� for cell growth), gases (e.g.air, oxygen, nitrogen, carbon dioxide) and other solutions (e.g. media, bu�er, acid,base, anti-foam) may be added into the bioreactor for advancement and/or controlof the process. Cell culture is typically very sensitive to temperature. Temperatureis maintained using a temperature control module (including a circulation pumpand one or more heat exchangers) providing a tempered heat transfer liquid to thebioreactor vessel jacket. The heat duty is much lower than fermentors, and heatinput may be required to maintain 37 ◦C.

� Each cell culture step may take days or even weeks to complete (Each seed step istypically 3 days, the production cell culture can be 10 to 16 days).

59


� The �Seed Bioreactor� step is completed when desired cell density is reached. (Thismay be veri�ed by sampling.) The �Production Bioreactor� (or next �Seed Bioreac-tor� scale up size) is sterilized and the contents of the Seed Bioreactor are transferredto the Production Bioreactor via a pump or pressure transfer.

� The cell culture process described above is repeated in additional seed bioreactorsas necessary and then in the production bioreactor.

� At the end of the production cell culture the contents of the production bioreactor aretransferred to a �Harvest� skid, which provides for initial separation of the productfrom waste materials.

The stirred single use bioreactor works very much like their stainless steel counterpartsbut without the extensive automation required for CIP/SIP. Probes and stirring systemare typically inserted into a plastic sleeve. At the conclusion of the culture, the bagcontents are transferred to the subsequent harvest operation and the bag assembly isdisposed of.

The �rst upstream step in the SU facility is a WAVE bioreactor, which is a furtheralternative in the disposable panorama. WAVE bioreactors feature a bag partially �lledwith cell culture medium and mounted on a tray that can be rocked to provide agitationand gas transfer. Oxygen enters the culture from the headspace above the culture medium.Parameters such as the rocking angle, rocking rate, and bag �ll ratio can also in�uencethe mass transfer rates in WAVE systems, which are now available at volumes of up to500 liters and have become established as part of the seed expansion steps for most cellculture-based production operations. However, rocker technology becomes less practicalas the scale increases. They are compatible with a wide variety of mammalian, insect,and plant cell lines.This system goes under the rocking style bioreactor, as anticipated, and works followingthese steps:

� The (pre-sterilized) bag is half �lled with media

� The bag is in�ated and placed on the rocker table

� The rocking motion is initiated and the heating elements in the table bring themedia to a preset temperature

� Inoculum is added to the bag

� During the cell culture (which may last several weeks,) the temperature is controlled(via the heating table) and gasses are added to the bag as necessary

� The gentle wave motion facilitates gas transfer by �entrainment� while impartingvery little shear force to the cells

� Samples are taken as necessary during the culture

� At the end of the culture, the bag contents are transferred to the subsequent harvestoperation

60


� The bag assembly is disposed of

The following cell culture steps can be subsided to the stirred tank bioreactors cate-gory. The majority of cell culture operations for large-scale biopharmaceutical productioninvolve the use of stirred-tank bioreactors. The key development in this area was the con-cept of culturing cells in an integral plastic bag that could be mounted within a cylindricalframe to support the bag. Hyclone was the �rst market entrant with its Single Use Biore-actor (SUB) system, which used a top driven impeller for mixing and agitation. Xcellerexfollowed with the XDR disposable stirred tank bioreactors featuring magnetically-coupledbottom-driven agitators. A wide variety of small-volume and large-volume single-usestirred-tank bioreactors are now available. Early challenges with e�cient mixing and aer-ation have been addressed and the disposable systems now appear to be equivalent tostainless steel stirred-tank systems for CHO cell cultures.

Knowing the scale up ratio for the kind of substances we are dealing with, it is requiredto consider three seed/growth steps before the bioreactor.

From literature, it is known the cell culture duration range of feasibility and it waschosen a cell culture duration of 14 days as arbitrary mean value.

The main di�erence between stainless steel and single use systems, regarding schedule,is the SIP and CIP procedures for the �rst one and a setup time for moving and preparingthe bag for the latter. This statement can be extended to all tanks and bags, respectively.In any option, the worst case is always the stainless steel, from this perspective.

The setup consists in di�erent operations leading to the proper installation of the baginto its holder. A list of some possible explicative procedures of the worst-case scenario(2000L bioreactor) is presented:

1. Prepare the holder: (optional) lock the wheels and open the door

2. Insert the bag exploiting a lifting device

3. Mount the drain port adapter after aligning the bag at the center

4. Raise the bag and connect it to the top port

5. Connect auxiliaries: inlet air tubing, �lters, sparger, exhaust, pressure sensor

6. In�ate the bag, adjusting concurrently its position

7. Close the door

8. Install sensors: sampling port, (optical) pH, (optical) DO, (nozzle) temperature

It is good habit to perform a visual inspection after critical steps. Be cautious toavoid damage that could lead to leakage at any time. Almost all connections are engagedthrough manually closed tri-clamp. If sparger mixing is not su�cient, a mechanical agi-tator, with all due complication, must be considered. It could be operating from bottom(blade disk agitator) or from top (blade segment impeller).

Furthermore, since the item is single-use, in the setup time is also considered theuninstalling:

61


1. Deactivate and switch o� all the electrical attachments: pumps, TCU, stirrer, etc.

2. Close the drain clamp

3. Disconnect all probes, �lters, �exibles

4. Open and remove the top tri-clamp

5. Open the holder door

6. Lower the bag

7. Start removing the bag exploiting a lifting device

6.3 Harvest

This section consists in cell membrane disruption and �rst rough separation of product.More commonly for biotech applications, �rst separation is a �two phase� one, involving

a solid phase (cells or cell debris) and a liquid phase (cell culture broth). Indeed, acentrifuge is a machine that uses centrifugal force to separate substances of di�erentdensities.

Furthermore, clari�cation is a separation process designed to recover �pure� liquidwhile leaving solids (and some small amount of liquid) behind.

Applications are the following:

Clari�cation Removal of cells, debris, other solids from fermentation broth where theproduct is extracellular.

Concentration/Cell Separation Removal of media where the product is intracellular.

The following issues are critical in selection of a particular type of centrifuge:

Shear In cell culture separation, the cells must be handled gently to avoid breaking thecell wall membrane. This is described as a �shear sensitive� application.

�Sterility and Cleanability� Many centrifuge designs originally developed in non-sanitaryapplications are inherently di�cult (if not impossible) to clean and/or sterilize.

The �ltration skid falls under the depth �ltration category. Its membrane has a �brous,granular or sintered matrix that produces a random porous structure. Particles becometrapped in the tortuous network of �ow channels. These are sometimes used as pre-�ltersbecause of their relatively high �dirt-holding� capacity.

As anticipated in process analysis overview, the centrifuge is exploited only in the SScase, while is substituted from a two-stage �ltration in the SU one. Centrifuges performa robust clari�cation process. This is a common technique used to harvest large-scalecell culture vessels because it combines low running costs with uncomplicated processdevelopment and operating robustness. Centrifuges remove a considerable proportion ofcells and cell debris that can foul downstream �lters and chromatographic steps, lead-ing to unacceptable pressure drops and reduction in overall performance. An optimized

62


centrifugation process minimizes cell lysis � and related generation of additional celldebris or release of intracellular impurities and proteases � and maximizes sedimenta-tion of sub-micron particles and product yield. Centrifugation with disk-stack bowls is awell-established technology for clarifying cell-culture broths. This process, however, canbe challenging in biopharmaceutical streams. Cell-culture solutions have a broad rangeof particle sizes. When those solutions are homogenized, the particles can be small andhighly hydrated. Such particles can have a density that is very close to that of the bulkliquid. Furthermore, due to cell-shearing and/or presence of nonviable cells, intracellu-lar components such as nucleic acids can signi�cantly increase harvest stream viscosity.Centrifugation is a preferred method for clarifying cell cultures from batch sizes > 2000L.Nonetheless, centrifuge centrate still does not have the requisite clarity for downstreamprocessing. In addition, current upstream processes are delivering cell densities of > 150million viable cells per milliliter and monoclonal antibody (MAb) titers of > 25g/L. So,the limitations of centrifugation processes for high cell-density harvests are again be-coming a challenge. Hence, invariably, depth �ltration is used after the centrifuge step,mandatory before loading the �rst chromatography step downstream. When only depth�ltration (also referred to as depth micro�ltration or pre�ltration) is used at this step,it is often conducted in two stages. First, a �lter with an open-pore structure removescells and cell debris. Then a �lter with a tighter pore structure removes colloidal matter.Depth �ltration is highly preferable because of its low investment cost, good scalability,easy handling, and reproducible clari�cation e�ciency [58][59].

However, regarding schedule, being the operations consecutive in both cases, the over-all time required is comparable, with or without centrifuge.

Note that the volume needed by the collection equipment in the SU case is muchhigher, since the double �ltration step require a consistent �ushing stream for productrecovery.

6.4 DSP Pre-viral

From the process simulation, the equipment for the DSP Pre-viral are sized as indicatedbelow:

6.4.1 Chromatography Skids

Chromatography is the separation of molecules based on di�erence in structure and/orcomposition. The process typically involves a stationary adsorbing medium and a mobilemedium. In biotech, the mobile medium is always liquid and can be Low Pressure LiquidChromatography (LPLC) or High-Performance Liquid Chromatography (HPLC).

Di�erent chromatography techniques (a�nity, ion exchange, hydrophobic interaction)have been widely implemented in biologics puri�cation in the format of resin columns.The majority of functional ligands are grafted within internal pores of polymeric chro-matography resins. In order to interact with the ligands, molecules have to take long andrestricted di�usion pathways which signi�cantly hinder mass transfer and limit �ow rate.In the attempt to increase productivity, resin columns are sized by volumetric �ow rateinstead of capacity, resulting in over-sized unit operations (up to 2m in diameter with

63


10 cm to 20 cm bed height) that require vast investment in large columns (especially forcostly Protein A resins formAb capture), associated hardware, supporting systems, andfacilities. To be cost e�ective, these large columns need to be amortized over many cyclesand batches, which increase the oversight of quality and regulatory groups for consis-tent processing. Besides the economic issues, large columns can su�er from scale-relatedpacking problems including hysteresis, edge e�ects, and resin compression.

In our case study, the following techniques are exploited:

A�nity chromatography (AC) substances of interest are selectively and reversiblyadsorbed (bound) to a bed material, while unbound substances are washed away.Protein A falls under this category, characterized by a very speci�c and thereforeexpensive resin.

Ion Exchange Chromatography (IEX) separation of molecules based on electricalcharge. In particular, the second chromatography step sees an anionic exchange(AEX), while the third and last step a cationic exchange (CEX).

The chromatography system is composed of:

Chromatography column :

� Column tube: cylindrical pressured vessel shell made of acrylic, glass or stain-less steel, depending on the application and design pressure

� Top and bottom end plates

� Bed supports (with liquid distribution design and a screen)

� Integral valves (design vary from vendor to vendor, and depending on theconnection types)

Column packing system :

� Column packing tank with agitator and vent �lter

� Column packing pump

� Solids addition equipment as necessary

Liquid Handling system :

� Feed pumps

� Instrumentation including UV, conductivity and pH control

� Bubble trap (to remove all gas from the �uid, before it enters the column,a�ecting bed uniformity)

� Control panel

The chromatography system works following these steps:

1. The chromatography column is packed (�lled) with resin (a solid substance such assilica, with a tight particle size tolerance).

64


2. The chromatography column includes top and bottom end plates which includescreens of a pore size to trap the resin while allowing the liquid to �ow through.

3. The top end plate is lowered to compress the resin bed

4. The operations included are then usually: equilibration, load, strip/wash, elution,�ush, regeneration, cleaning, storage. Note that the last two operations are notcycled.

Calculations associated to chromatography are complex. Being the choice of the kindof separation technique the starting point, it is obtainable a binding capacity, relatedto the resin exploited. Then, it is necessary to know the target concentration, in orderto de�ne the volume of resin needed and therefore the combination of number of cyclesversus column bed volume.

6.4.2 Intermediate Containers

Tanks or bags are introduced among puri�cation steps to avoid useless occupation time.Furthermore, in two consecutive ones, a procedure of viral inactivation through pH regu-lation is exploited. In the �rst equipment a bu�er is added such that pH reaches valuesthat do not allow virus survival. Consequently, another bu�er is added next to recoverthe initial pH.

6.4.3 Viral Filtration

It is part of the �Dead End,� �Dead Ended Filtration,� �Through Flow� or �Normal FlowFiltration (NFF)� � Fluid is directed toward the membrane. Particles too large to passthrough the pores of the membrane accumulate at the membrane surface. This is typicallya single pass process. A �lter cake will often form on the �lter membrane. �Sterile�ltration� (gas or liquid) and tank venting are examples of dead end �ltration. Dead end�ltration is also used for bulk clari�cation (e.g. through depth �lters and plate and frame�lters). The pore sizing is small enough (oom: nm, nano�lter) that eventual viruses areretained and, therefore, with this puri�cation step ends the pre-viral downstream section.

6.5 DSP Post-viral

From the process simulation, the equipment for the DSP Post-viral are sized as indicatedbelow:

6.5.1 UF/DF Skid

Ultra�ltration (UF) is a tangential �ow �ltration process where extremely small parti-cles and dissolved molecules are removed from a �uid by passage across a microporousmedium. Rather than using pore size, ultra�lters are typically rated according to �molec-ular weight cut-o�� (MWCO). While the primary basis for separation is molecule size,molecule shape and charge play a role as well. Substances with molecular weights ranging

65


from (approximately) 1000 to 1 000 000 molecular weight are retained by ultra�lter mem-branes, while salts and water pass through. Colloidal and particulate matter can also beretained.

In �Tangential Flow Filtration� (TFF) or �Cross Flow Filtration�, �uid is pumped �tan-gentially� parallel to the surface of the membrane. Particles too large to pass through thepores of the membrane are swept along by the tangential �ow. The retentate is typicallyreturn to a feed tank and recirculated back through the �lter. The cross �ow prevents thebuildup of a signi�cant �lter cake on the membrane and helps prevent membrane fouling.Thus, di�cult to �lter solutions (�slimy�) may be �ltered using TFF. It requires:

� Feed Tank - Sanitary stainless steel vessel or SU bag, typically with agitator andvent �lter.

� Recirculation Pump - Typically a sanitary rotary lobe pump with a variable speeddrive.

� Instrumentation - including tank level (and/or load cells) temperature, �ow, andtrans-membrane pressure (TMP) �across� the membrane.

� Clean-In-Place and Steam-In-Place (Components are designed to allow the feed tankand piping to be cleaned and sterilized in place. Clean-In-Place solution and cleansteam are provided to the system.)

� Control Panel

In a UFDF skid one or more concentration operations are coupled with dia�ltration.

Dia�ltration (DF) is the process of adding �uid (referred to as �dia�ltration bu�er�)during the �ltration process which is typically added at a rate equal to the permeaterate so as to maintain constant volume. Large molecules (such as proteins) areretained by the �lter, while small molecules pass through the �lter membrane. Inprocesses where the product is in the retentate, dia�ltration washes small moleculecomponents out of the product pool into the �ltrate, thereby exchanging bu�ersand reducing the concentration of undesirable species. When the product is in the�ltrate, dia�ltration washes the product through the membrane into a collectionvessel.

Concentration implies that bu�er is NOT added to replace the permeate during theprocess.

The process steps are:

1. Fill feed tank

2. Circulate �uid while controlling trans-membrane pressure (TMP control) or cross�ow rate and permeate rate

3. (Optional) Initial concentration step (meaning �uid is recirculated without replacingthe permeate)

66


4. (Typically) Replace permeate by adding �dia�ltration bu�er� to the feed tank

5. Continue the �dia�ltration� until a predetermined volume of �dia�ltration bu�er�has been added

6. (Optional) Begin �concentration� by continuing the circulation without adding �di-a�ltration bu�er.�

7. When the �nal dia�ltration volume or concentration volume has been reached, trans-fer �batch� to collection tank if product is in the retentate. If product is in thepermeate, collect permeate in collection tank and at end of process send feed tankto drain.

8. After processing, the membranes are typically cleaned using dilute solutions of caus-tic (0.5 to 1N).

9. After cleaning, the membranes may be left in a solution of storage bu�er to minimizeopportunity for microbial growth. This may be dilute caustic (0.1 to 0.5N).

10. Prior to use, the storage bu�er is �ushed from the membranes using typically WFI.A normal water permeability test (NWPT) may be conducted to determine if themembrane has been fouled.

In our case study, both concentration steps, one before and one after dia�ltration, areexploited.

6.6 Media Preparation

Media and Nutrients are prepared in SS or SU equipment and stored in SS or SU equip-ment.

The water quality required for the Media Preparation is Puri�ed Water (PW).Two types of media were exploited: the �rst one for the seed steps, while the second

one only for the bioreactor. In the latter, also four di�erent nutrients were prepared andintroduced distributed during cell growth in arbitrary intervals.

The process steps for the SS case are:

1. SIP

2. WFI addition

3. Powder Load

4. Mix

5. Transfer to tank

6. CIP

The overall duration ranges between 10 to 14 hrs, directly dependent on tank dimension.The process steps for the SU case are:

67


1. Set up

2. WFI addition

3. Powder Load

4. Mix

5. Transfer to bag

The overall duration ranges between 5 to 8 hrs, directly dependent on bag dimension.

6.7 Bu�er Preparation

Bu�er solutions are prepared in the indicated SS or SU equipment and stored in SS orSU equipment.

The water quality required for the Bu�er Preparation is Water-for-Injection (WFI).

Table 6.1 shows the bu�er list coupled with the operations associated.

Name Description of Puri�cation Step

ProtA

B1 EquilibrationB2 WashB3 ElutionB4 StripB1 Flush

BRegProtA RegenerationBCleanProtA Cleaning

VIB5 VIB6 pH Regulation

CEX

B7 Equilibration 1B8 Equilibration 2B9 StripB10 ElutionBReg Regeneration

AEX

B11 EquilibrationB11 WashB12 ElutionB13 StripBReg Regeneration

NF B12 Viral Filtration

UF/DFB12 EquilibrationB14 Dia�ltration

Table 6.1: Bu�er List

68


NaOH solutions are used for sanitization, cleaning and storage. As assumption, in-linedilution is not included as starting condition, even if simulated for NaOH.

Table 6.2 shows the equipment sizes studied for the bu�er preparation.

Case N. of Equipment & Working Volume

SS1× 4000L2× 1500L1× 100L

SU1× 3000L2× 2000L1× 100L

Table 6.2: Bu�er Preparation

It is important to remember and stress that SU bag dimensions that are exploitableare very limited in number compared to the SS o�er.

The process steps for a generic bu�er preparation for the SS case are:

1. SIP

2. WFI addition

3. Powder Load

4. Mix

5. Transfer to tank

6. CIP

The overall duration, in this case, is stable around 12 hrs, independent of tank dimension.The process steps for the SU case are:

1. Set up

2. WFI addition

3. Powder Load

4. Mix

5. Transfer to bag

The overall duration, again, as for the SS case, is independent of bag dimension. It �oatsaround 5 hrs

69

Chapter 7

Process Model Output

Table 7.1 summarizes the main production data and bottleneck for the entire facility.

Description SS SU

Bottleneck Production Bioreactor Production BioreactorMinimum Recipe Cycle time 7.40 days - 178 hrs 7.29 days - 175 hrsRecipe Cycle time 8.14 days - 195 hrs 8.02 days - 193 hrsMaximum Number of Batches 41 batches/year 43 batches/yearFacility Utilization Factor 90% 90%

Table 7.1: Recipe Scheduling Information

To enable a better comparison, it has been decided to set the productivity of the plantequal for both cases. Therefore, the product obtained per year will be the same, whilethe equipment utilization factors will change to adapt to this assumption.

General scheduling outputs will follow, showing di�erent equipment on the y-axis andtime line, on hourly and daily scale, on the x-axis. Each color identi�es a distinct batch,while each colored rectangle de�nes a procedure (occupying an equipment), which, inturn, is subdivided in sections, assessed by vertical black lines, that are called operations.

70

7 � Process Model Output

Process equipment occupancy for one batch schedule, SS and SU respectively

Figure 7.1: SS single batch schedule

Figure 7.2: SU single batch schedule

71


Process equipment occupancy for one year batches schedule, SS and SU respectively

Figure 7.3: SS one year schedule

Figure 7.4: SU one year schedule

72


7.1 USP

The following chart shows the equipment occupancy of USP, SS and SU respectively

Figure 7.5: SS USP multiple batches schedule.Production Bioreactor BR-101 is the Bottleneck: unit procedure with the longest duration.Facility utilization factor = 90%.Inoculum steps not included in the chart.

Figure 7.6: SU USP multiple batches schedule.Production Bioreactor DBS-101 is the Bottleneck: unit procedure with the longest dura-tion.Facility utilization factor = 90%.Inoculum steps not included in the chart.

73


Table 7.2 summarizes the equipment utilization rates.

Description Equipment Utilization Factor [%]

SS

Seed Bioreactor SBR-101 58.13Seed Bioreactor SBR-102 55.05Seed Bioreactor SBR-103 55.82Production Bioreactor BR-101 90.90

SU

Wave Bioreactor RBS-101 52.98Seed Bioreactor DSBS-101 54.02Seed Bioreactor DSBS-102 54.80Production Bioreactor DBS-101 90.90

Table 7.2: USP Utilization Factors

The bioreactors are the equipment with the highest occupancy rate (∼ 90%) and thebottleneck for all overall process. Delays related with operations in this equipment maya�ect production schedule. Delays in production bioreactors operations may be dampenedwithout a�ecting production schedule. Delays up to 17 h may be tolerated.This is due to the initial assumptions of leaving a∼ 10%margin, which is a wise procedure,foreseeing risk factors of failures and delays. However, it is useful to keep in mind that thebottleneck procedure, and its related ones, should be modi�ed as few as possible beingthe master of the overall process and determinant factor for the overall productivity.Therefore, when undergoing rescheduling activities, it is good habit to consider �exibleshifts, advances or delays, for the auxiliary procedures, like media and bu�er preparationtrains, and, more in general, for equipment with lower equipment utilization factor.

7.2 Harvest

Following charts will show the equipment occupancy of the harvest section, SS and SUrespectively.



SSCentrifuge 11.27Clari�cation 5.38Clari�ed Tank 15.24

SUDepth Filter 5.74Clari�cation 2.34Clari�ed Bag 14.71

Table 7.3: USP Utilization Factors

All equipment has low utilization factors. More than su�cient spare time is available

74


Figure 7.7: SS Harvest schedule

Figure 7.8: SU Harvest schedule

between consecutive batches. It is possible to have �ve di�erent concurrent staggeredproducts puri�cations, due to the fact of the low occupancy time.

The �rst di�erence between SS and SU facility is highlighted. As already discussed,centrifugation is not suitable for disposable technology, but the alternative �ltration showsa lower equipment utilization factor. This is a common line that will follow for most ofthe procedures, assessing the greater �exibility of the single-use facility.However, there may arise a question regarding why not choosing �ltration also for tradi-tional facility, then. The answer is not certain, since the operating cost of the changingof the cartridges of the �lter is somehow balanced by the heavy energy request, electrical,of the centrifuge. Furthermore, the �lter is at borderline operating margin, since, at this�ow-rates, the risk of packing is not negligible.

75


7.3 DSP

7.3.1 DSP Pre-viral

The following charts show the equipment occupancy of Pre-viral DSP, SS and SU respec-tively.

Figure 7.9: SS DSP Pre Viral schedule

Figure 7.10: SU DSP Pre Viral schedule

76




SS

ProtA Chrom 13.28VI Tank 14.86pH Tank 14.39AEX Chrom 10.21AEX Collection Tank 15.43CEX Chrom 7.65CEX Collection Tank 9.51Viral Filter 6.86

SU

ProtA Chrom 15.34VI Bag 12.57pH Bag 8.62AEX Chrom 10.35AEX Collection Bag 12.79CEX Chrom 7.73CEX Collection Bag 6.79Viral Filter 4.62

Table 7.4: DSP Pre Viral Utilization Factors

All equipment has low utilization factors. More than su�cient spare time is availablebetween consecutive batches. It is possible to have �ve di�erent concurrent staggeredproducts puri�cations, due to the fact of low occupancy time.

7.3.2 DSP Post-viral

The following charts show the equipment occupancy of Post-viral DSP, SS and SU respec-tively.

Figure 7.11: SS DSP Post Viral schedule


77


Figure 7.12: SU DSP Post Viral schedule


SS

UFDF Tank 8.47UFDF 10.57Final Filter 10.50Final Tank 13.57

SU

UFDF Bag 5.48UFDF 2.93Final Filter 1.04Final Bag 1.56

Table 7.5: DSP Post Viral Utilization Factors

All equipment has low utilization factors. More than su�cient spare time is availablebetween consecutive batches. It is possible to have six di�erent concurrent staggeredproducts puri�cations, due to the fact of the low occupancy time.

A comment regarding the previous section has been skipped, since no heavy di�erencewas identi�able, being the equipment of interest not much a�ected by CIP operations(e.g. chromatography columns).

After the viral �ltration, the equipment utilization factors assume completely di�erentvalues. Analyzing �gures of Section section 7.3.2 on the preceding page, it is easilydenotable how the SS facility has the need of delaying the washing operations for eachequipment, being very short and fast sequenced. The white space before the last operationof the last three procedures is not actual busy time, but the equipment is waiting for CIPand cannot be utilized, therefore is occupied. This e�ect can snowball easily, as showedby the �nal storage, which has an equipment utilization factor roughly ten times higherthan the correspondent twin of the single-use facility.

78


7.4 CIP skids

The following chart shows the equipment occupancy of CIP skids, of course only for theSS case.

Figure 7.13: CIP skids schedule



CIP skid USP 9.47CIP skid DSP Pre Viral 8.96CIP skid DSP Post Viral 5.12CIP skid Media 16.64CIP skid Bu�er-1 24.33CIP skid Bu�er-2 24.33

Table 7.6: CIP skids Utilization Factors

CIP skids number is chosen following two criteria: segregation and equipment uti-lization factor. The �rst reason is preventing cross contamination between sections ofdi�erent sanitary classi�cation, while the second one regards schedule. The maximumEUF allowed for a cleaning-in-place skid is around 35% (which is precautionary), mainlydue to the fact that a cleaning could fail and it is not conceivable to have a delay causedby a non process unit. Only bu�er section needed a secondary skid for this reason.

79


7.5 Media Preparation

The following charts show the schedule of media preparation equipment and bioreactorsfor one batch, SS and SU respectively.

Figure 7.14: SS Media Preparation schedule for one batch

Figure 7.15: SU Media Preparation schedule for one batch

There are 8 preparations for each batch. The choice of the subdivision was driven bytaking on consideration the minimum working volume of 25%. Furthermore, it is clearthe exploitation of holding units, since the preparation procedures do not coincide withthe respective utilizations.

The following charts show the schedule of media preparation equipment for multiplebatches, SS and SU respectively.

Due to the fact that the utilization factors are very low, the schedule can be organizedin order to prepare the media solutions before each batch: there is spare time betweenconsecutive batches. Furthermore, the utilization across batches allows better schedulingand risk handling.

80


Figure 7.16: SS Media Preparation schedule for multiple batches

Figure 7.17: SU Media Preparation schedule for multiple batches



SSMedia Prep 200L 29.4Media Prep 1100L 20.0

SUMedia Prep 500L 13.0Media Prep 1000L 7.8

Table 7.7: Media Preparation Equipment Utilization Factors

All equipment has low utilization factors. More than su�cient spare time is availablebetween consecutive batches. It is possible to have three di�erent concurrent staggeredproducts preparations, due to the fact of the low occupancy time.

81


7.6 Bu�er Preparation

Bu�er preparation usually allows stocking the solution in cold rooms in order to haveready to use the materials for multiple batches by anticipating preparation. However,SuperPro Designer does not allow such feature, forcing to simulate a preparation for eachbatch. In other words, the bu�er solutions are prepared each batch and used before everyindividual process step of DSP. Anyway, this is the scheduling worst case scenario forboth cases, even if it is a�ecting more the stainless-steel case, due to the SIP and CIPskid utilizations. It is needed to keep it under consideration.

The following charts show the schedule of bu�er preparation equipment for multiplebatches, SS and SU respectively.

Figure 7.18: SS Bu�er Preparation schedule for multiple batches

Figure 7.19: SU Bu�er Preparation schedule for multiple batches


All equipment has low utilization factors. More than su�cient spare time is availablebetween consecutive batches. It is possible to have two di�erent concurrent staggeredproducts preparations, due to the fact of the low occupancy time.

It is distinguishable by eye that SU schedule is way lighter than SS one, having amedium higher idle time between consecutive batches. The reason is again the steamingand washing required by the SS facility. In particular, for vessels where no complicateor long operations take place, this factor impacts heavily on the equipment utilization

82



SS

Bu�er Prep 100L 12.8Bu�er Prep 1500L_1 37.4Bu�er Prep 1500L_2 29.4Bu�er Prep 4000L 36.9

SU

Bu�er Prep 100L 5.2Bu�er Prep 2000L_1 21.8Bu�er Prep 2000L_2 18.7Bu�er Prep 3000L 18.7

Table 7.8: Bu�er Preparation Equipment Utilization Factors

factors. On the other hand, it is useful to foresee that the other face of the medal willappear later on in the elaborate where the consumables variable will enter the scene.

As in a mass balance, all that enters should somewhere exit and this aspect is partic-ularly true in our comparison, giving it particular interest and complexity. By thinkingoutside the box, every easy conclusion that may have been made, has been retreated byother considerations. Further analysis is required to achieve a thorough and satisfyinganswer.

83

Part III

Results

84

Chapter 8

Cost Estimation

The preliminary economic evaluation of a project for manufacturing a biological prod-uct usually involves the estimation of capital investment, estimation of operating costs,and analysis of pro�tability. Since in our case the main goal is a thorough comparisonand since the output selling price is not given with certainty, the last part will not betaken in consideration. Mainly it has been exploited an important feature of SuperProDesigner, that facilitates the estimation of capital and operating costs and also performspreliminary economic evaluation of manufacturing and environmental processes. The keyeconomic, �nancial, operating and market input parameters are analyzed by followingcommon approaches that are available in the literature in order to determine four majoreconomic aspects of an investment. To actively interact with this systemic evaluation, itis possible to tune some economic information, collected in literature and reviewed by anexpert. default cost data and calculation options are provided for the bottom four eco-nomic information categories (e.g., unit costs of labor, power and consumables, equipmentpurchase cost, capital and operating cost factors with respect to purchase cost, informa-tion regarding time valuation, �nancing and production level, etc.). However, the defaultparameters may not be suitable for a particular project.

Material-related information This includes the di�erent unit costs (e.g., purchaseprice, selling price, waste treatment/disposal cost) of pure components and stockmixtures.

Stream-related information This includes the classi�cation of input and output streamsinto di�erent categories (e.g., raw materials, revenues, wastes, etc.).

Operation-Related information This includes the unit costs of labor and utilities(heat transfer agents and power) for each operation.

Equipment-related information This includes the capital and operating costs for eachequipment (e.g., purchase cost, installation, maintenance, consumables, etc).

Section-Related information This includes cost factors that are used to determine thecapital and operating costs for each process section.

85

8 � Cost Estimation

Process-related information This includes economic evaluation parameters that arespeci�ed at the process level, such as time valuation, �nancing, production level andadditional operating cost information for the entire project.

8.1 Investment Costs

8.1.1 General De�nitions

Capital cost estimation can be classi�ed di�erentiating through level of detail.

1. Detailed estimate

2. De�nitive estimate

3. Preliminary estimate

4. Study estimate

5. Order of magnitude estimate

Number one is the most accurate assess, as deducible by its name, leading to a precisionrange of ±6%.

Order-of-magnitude estimate also called "Ratio or Feasibility Estimate": this typeof estimate typically relies on cost information for a complete process taken frompreviously built plants. This cost information is then adjusted using appropriatescaling factors, for capacity, and for in�ation, to provide the estimated capital cost.

Study estimate also called "Major Equipment or Factored Estimate": this type of es-timate uses a list of the major equipment present in the process. This includes allpumps, compressors and turbines, columns and vessels, �red heaters, and exchang-ers. Each piece of equipment is roughly sized and the approximate cost determined.The total cost of equipment is then factored to give the estimated capital cost.

Preliminary design estimate also called "Scope Estimate": this type of estimate re-quires a sizing of equipment more accurate than the one used in the Study Estimate.In addition, approximate layout of equipment is made with estimates of piping, in-strumentation, and electrical requirements. Utilities are estimated.

De�nitive estimate also called "Project Control Estimate": this type of estimate re-quires preliminary speci�cations for all the equipment, utilities, instrumentation,electrical, and o�-sites.

Detailed estimate also called "Firm or Contractor's Estimate": this type of estimaterequires complete engineering of the process and all related o�-sites and utilities.Vendor quotes for all expensive items will have to be obtained. At the end of adetailed estimate, the plant is ready for construction.

86


First rough estimates are made to compare alternatives' feasibility, upgrading the com-parison to the most pro�table in greater detail. It is di�cult, given these de�nitions, toclassify our case study in a speci�c class, �tting in some terms to one, and in others tothe previous or the following one [60].

8.1.2 SuperPro De�nitions

The Total Capital Investment refers to the �xed costs that are associated with a process.This is calculated as the sum of the following cost items over all sections of a process:

1. Direct Fixed Capital

2. Working Capital

3. Startup and Validation Cost

4. Up-Front R&D Cost

5. Up-Front Royalties

Direct Fixed Capital (DFC)

The DFC refers to the �xed assets of an investment, such as plant and equipment. Itis calculated at the process section level as the sum of direct, indirect and miscellaneouscosts that are associated with a plant� s capital investment. The direct costs include costelements that are directly related to an investment, such as the cost of equipment, processpiping, instrumentation, buildings, facilities, etc. The indirect costs include costs that areindirectly related to an investment, such as the costs of engineering and construction. Ad-ditional costs such as the contractor's fee and contingencies are included in miscellaneouscosts. By default, the DFC is estimated using cost correlations to estimate the purchasecost of all major process equipment and cost factors with respect to purchase cost togenerate estimates for all other cost elements.The DFC for small biotechnology facilities is usually in the range of 30 to 60 million�, whereas for large facilities it is in the range of 100 to 250 million �. For preliminarydesign purposes, usually, the various items of DFC are estimated based on the total equip-ment purchase cost (PC) using several multipliers that can be found in traditional processdesign textbooks and technical literature (Peters and Timmerhaus, 1991; Ulrich, 1984;Valle-Riestra, 1983; Garrett, 1989; Seider et al., 1999; Douglas, 1988), as shown in Table8.1 [55].

Total Plant Direct Cost (TPDC) The Total Plant Direct Cost (TPDC) is the sumof the following direct cost items:

Equipment Purchase Cost (PC) this is the vendor's selling price of major equipment.It excludes items such as taxes, insurance, delivery and installation. It is alsoknown as the free-on-board (FOB) cost. For a preliminary economic analysis, thepurchase cost of equipment is typically estimated based on cost correlations. Super-Pro Designer provides correlations for estimating the purchase cost of major listed

87


Cost Item Average Multiplier Range of Values

Total Plant Direct Cost (TPDC)Equipment Purchase Cost (PC)Installation 0.50× PC 0.2− 1.5Process Piping 0.40× PC 0.3− 0.6Instrumentation 0.35× PC 0.2− 0.6Insulation 0.03× PC 0.01− 0.05Electrical 0.15× PC 0.1− 0.2Buildings 0.45× PC 0.1− 2.0Yard Improvement 0.15× PC 0.05− 0.2Auxiliary Facility 0.50× PC 0.2− 1.0

Total Plant Indirect Cost (TPIC)Engineering 0.25× TPDC 0.2− 0.3Construction 0.35× TPDC 0.3− 0.4

Total Plant Cost (TPC = TPDC + TPIC)Contractor's Fee 0.05× TPC 0.03− 0.08Contingency 0.10× TPC 0.07− 0.15

Direct Fixed Capital (DFC = TPC + 12 + 13)

Table 8.1: Itemized Fixed Capital Cost Estimation

(modeled) equipment. The user may also provide his/her own cost values or costcorrelations for all listed (modeled) equipment. In SuperPro Designer, PC is calcu-lated at the section level. For each section, the user may also specify the purchasecost of unlisted (overlooked) equipment as a factor of the section's PC. Generally, asection's PC will be the sum of the purchase costs of listed and unlisted equipmentfor that section.

Installation Cost this cost item refers to the in-place construction of equipment at thenew plant site and includes the cost of foundations, slabs, supports, and local equip-ment services. For a preliminary economic analysis, the installation cost of listed(modeled) equipment can be estimated by multiplying the corresponding purchasecost by a suitable factor. In SuperPro Designer, the installation cost is calculatedat the section level. For each section, the user may also specify the installation costof unlisted (overlooked) equipment as a factor of the corresponding purchase cost ofunlisted equipment for that section. Generally, a section's installation cost will bethe sum of the installation costs of listed and unlisted equipment for that section.In general, equipment delivered mounted on skids has a lower installation cost.

Process Piping Cost this cost item incorporates the cost of process �uid piping thatconnects the equipment, as well as connections to the main utility headers andvents. Included are valves, piping supports, insulation, and other items associatedwith equipment piping. For a preliminary economic analysis, this cost is typically

88


estimated by multiplying PC by a suitable factor. In SuperPro Designer, this costis calculated at the section level as a factor of the section's PC.

Instrumentation Cost this cost item includes the costs of transmitters and controllers(with all required wiring and tubing for installation), �eld and control room terminalpanels, alarms and annunciators, indicating instruments both in the �eld and inthe control room, on-stream analyzers, control computers and local data-processingunits, and control room display graphics. For a preliminary economic analysis, thiscost is typically estimated by multiplying PC by a suitable factor. In SuperProDesigner, this cost is calculated at the section level as a factor of the section's PC.

Insulation Cost this cost item includes the cost of insulation and painting, which isusually included in the cost of installation and piping. In low temperature plants,however, insulation cost can become unusually high. An insulation surcharge isrecommended for such plants. For a preliminary economic analysis, this cost istypically estimated by multiplying PC by a suitable factor. In SuperPro Designer,this cost is calculated at the section level as a factor of the section's PC.

Electrical Cost this cost item refers to the cost of electrical facilities. These includesbattery limits substations and transmission lines, motor switch gear and controlcenters, emergency power supplies, wiring and conduit, bus bars, and area lighting.Separate equipment estimation is required for electrolytic installations. For a pre-liminary economic analysis, this cost is typically estimated by multiplying PC by asuitable factor. In SuperPro Designer, this cost is calculated at the section level asa factor of the section's PC.

Buildings Cost this cost item includes the cost of process towers, subsidiary concreteslabs, stairways and catwalks (not equipment-speci�c), control rooms and otherbattery limits buildings (e.g., change rooms, cafeteria, furnished o�ces, warehouses,etc.). It also incorporates the costs for non-electric building services as well as fora variety of safety-related items. For a preliminary economic analysis, this cost istypically estimated by multiplying PC by a suitable factor. In SuperPro Designer,this cost is calculated at the section level as a factor of the section's PC. For moreaccurate estimation of building costs, it is necessary to estimate the process arearequired based on the footprint of the equipment and the space required around theequipment for safe and e�cient operation and maintenance. Therefore, appropriateunit cost per area are provided as a function of type and classi�cation.

Yard Improvement Cost this cost item refers to the costs of excavation, site grading,roads, fences, railroad spur lines, �re hydrants, parking spaces, and others. For apreliminary economic analysis, this cost is typically estimated by multiplying PC bya suitable factor. In SuperPro Designer, this cost is calculated at the section levelas a factor of the section's PC.

Auxiliary Facilities Cost this cost item includes the cost of satellite process-orientedservice facilities that are vital to the proper operation of the battery limits plant.An example of an auxiliary facility is a steam plant. For a preliminary economic

89


analysis, this cost is typically estimated by multiplying PC by a suitable factor.In SuperPro Designer, this cost is calculated at the section level as a factor of thesection's PC.

Total Plant Indirect Cost (TPIC) The Total Plant Indirect Cost (TPIC) is the sumof the following indirect cost items:

Engineering this cost item includes the preparation of design books that document thewhole process (e.g., the design of equipment, speci�cation sheets for equipment,instruments, auxiliaries, etc., the design of control logic and computer software,the preparation of drawings) and other engineering-related costs. For a preliminaryeconomic analysis, this cost is typically estimated by multiplying TPDC by a suitablefactor. In SuperPro Designer, this cost is calculated at the section level as a factorof the section's total direct cost.

Construction this cost item includes the costs associated with the organization of thetotal construction e�ort. They do not include the cost of construction labor. Thisis incorporated in direct cost items that involve construction. For a preliminaryeconomic analysis, this cost is typically estimated by multiplying TPDC by a suitablefactor. In SuperPro Designer, this cost is calculated at the section level as a factorof the section's total direct cost.

The sum of TPDC and TPIC is denoted as Total Plant Cost (TPC).

Contractor's Fee and Contingency Costs (CFC) The following additional costsare also considered:

Contractor's Fee this is the contractor's pro�t. It should be added even if a corporationdoes its own construction, because the construction division is expected to show apro�t. For a preliminary economic analysis, this cost is typically estimated bymultiplying TPC by a suitable factor. In SuperPro Designer, this cost is calculatedat the section level as a factor of the section's total direct and indirect costs.

Contingency the more speculative a process is, the more likely it is that key elementshave been overlooked during the project's early stages. This cost attempts to com-pensate for missing elements. However, even advanced-stage estimates will include acontingency to account for unexpected problems during construction, such as strikes,delays, and unusually high price �uctuations. For a preliminary economic analysis,this cost is typically estimated by multiplying TPC by a suitable factor. In Super-Pro Designer, this cost is calculated at the section level as a factor of the section'stotal direct and indirect costs.

Based on the above de�nitions, the total DFC of an investment is calculated as the sumof TPC and CFC.

90


PC Insight An equipment resource may represent multiple equipment units. It mayinclude a number of units (N) that are operated in parallel (i.e., simultaneously), extrasets of parallel units (M) that are operated in staggered mode (i.e., out of phase) anda number of standby units (K). Consequently, an equipment may represent a total ofN(1 +M) +K units. The purchase cost of an equipment resource will be equal to thecorresponding purchase cost of a single unit of that type times the total number of unitsthat it represents. For batch processes, only a fraction of that cost will be charged to eachsection. This is calculated as the fraction of total batch time that the equipment is in useby procedures contained in that section. This time factor is calculated by the program aspart of the simulation.The equipment purchase cost can be estimated from vendor quotations, published data,company data compiled from previous projects, and by using process simulators andother computer aids. Vendor quotations are time-consuming to obtain and are thereforeusually avoided for preliminary cost estimates. Di�erent speci�cation options are availablein SuperPro for the purchase cost of a single unit.

1. set by the user

2. estimated based on a built-in model

3. estimated based on a user-de�ned model

By default, the purchase cost of equipment is estimated based on built-in cost correlationsaccording to the second option. SuperPro Designer provides correlations for estimatingthe purchase cost of all major listed equipment. If the third option is selected, customcost correlations can be speci�ed based on the following power law:

PC0 = C0

(Q

Q0

)αwhere:

� C0 is the base cost

� Q0 is the base capacity or size

� α is the exponent of the power law

This law, also called scaling law, is oftentimes useful when cost data for one or twoequipment sizes is available, but the cost for a di�erent size piece of equipment must beestimated. If more data is available, an interpolation trend can be exploited for moreaccuracy. The mathematical form of this law explains why cost versus size data tend tofall on a straight line. The value of the exponent (α) in the equation ranges between0.5 and 1 with an average value for vessels of around 0.6. Indeed, the scaling law isalso known as the 0 .6 rule, which is just under 2/3, the ratio of surface to volume forvessels. According to this rule, when the size of a vessel doubles, its cost will increase ofapproximately 52%. This reasoning is ofter referred as economy of scale.To account for the time value of money, you must also specify the reference year for which

91


the cost is valid. If the purchase cost is set by user, you may also specify a reference yearor let the cost be �xed (independent of the year of the analysis). The price of equipmentchanges with time due to in�ation and other market conditions. That change in priceis captured by the Chemical Engineering Plant Cost Index that is published monthly byChemical Engineering magazine, and built-in in the simulator. For years not available,an in�ation rate is used.

PC1 = PC0

(I

I0

)where index I is used to update equipment cost data accordingly. By default, the year ofanalysis for a new project is the present year and the in�ation rate is 4%.Another factor that a�ects equipment purchase cost is the material of construction. Astainless steel unit is more expensive than a plastic one of the same size. Moreover,similarly, a stainless steel tank costs 2.5 to 3 times as much as a carbon steel tank of thesame size. However, in bioprocessing most of the equipment is made of stainless steel orplastic for GMP (Good Manufacturing Practice) reasons and selection of material is lessof a problem, with respect to chemical processes. Other factors material related includethe �nishing of metal surface and the instrumentation provided to the equipment, whichare the major causes for the wide range of the multipliers for bioreactors.

Working Capital

The working capital represents tied-up funds required to operate the business. It includesthe investment in raw materials, consumables, labor, utilities, waste treatment/disposal,etc. In SuperPro Designer, the working capital is speci�ed at the section level. For eachsection, this is estimated as the sum of major operational costs covered for a certain oper-ating period; these include the costs for labor, raw materials, utilities (i.e., heating/coolingagents and power), waste treatment, and miscellaneous costs. The user speci�es the num-ber of days that labor, raw materials, utilities and waste treatment costs are covered.The cost per section of each of the above items is calculated by multiplying the speci�ednumber of days by the corresponding daily cost of that item. The working capital of asection is calculated as the sum of all the above cost items. The timespan is usually inthe order of magnitude of months and can be assessed around 10 to 20% of the DFC.

Startup and Validation Cost

The startup and validation cost includes pre-opening, one-time expenditures incurred toprepare a new plant for operation. In SuperPro Designer, the startup and validationcost is speci�ed at the section level. For each section, this is estimated as a percentageof DFC, commonly 5 to 10%, which can represent a signi�cant capital investment for apharmaceutical plant.The startup and validation cost may optionally be depreciated. Depreciation is a measureof the decrease in the value of equipment and improvements over a period of time.

92


Up-Front R&D Cost

The up-front R&D cost accounts for the cost of research & development required before aproduct is manufactured. In SuperPro Designer, this is speci�ed at the section level andis preset to zero if the user does not set otherwise.

Up-Front Royalties

The up-front royalties account for the payments made for use of assets, resources, patents,etc. prior to the initiation of a project. In SuperPro Designer, this cost is speci�ed at thesection level and is preset to zero if the user does not set otherwise.

8.1.3 Results

Total Plant Cost (TPC) SS SU

1. Equipment Purchase Cost 17 990 000 7 539 0002. Installation 1 079 000 226 0003. Process Piping 7 196 000 1 508 0004. Instrumentation 10 794 000 1 885 0005. Insulation 540 000 75 0006. Electrical 1 799 000 754 000

TPC 39 398 000 11 987 000

Contingency (CFC) SS SU

7. Contingency 9 272 000 1 199 000

CFC 9 272 000 1 199 000

Direct Fixed Capital Cost (DFC = TPC+CFC)

DFC 43 338 000 13 186 000

Table 8.2: Fixed Capital Estimate Summary, prices in e

Calculation have been made following these assumptions:

� Installation, Process Piping, Instrumentation, Insulation and Electrical Cost are in-cluded in the Equipment Purchase Cost. Indeed, it has been considered a lumpedcost for skids that are ready-to-use, taken from similar projects already commis-sioned and/or concluded.

� All the just cited categories shall be, however, further examined for the overallplant. Therefore they will be di�erently evaluated with respect to what showed inthe previous section.

93


� Piping and Insulation include only connections among the skids and the distributionlines of the clean utilities.

� Installation considers the move in of the skid, its installation and its �nal hook up.

� Instrumentation, has been revised including also the Automation requirements. Per-centages of 30% each for the SS facility have been considered (for a total of 60% ofthe EPC), while a total lumped factor of 25% has been assumed for the SU one.

� Electrical considers only the process related equipment and, therefore, is the onlypercentage which sets at 10% for both facilities. No essential di�erences have beenidenti�ed to justify a modi�cation on this behalf.

� Buildings, Yard Improvement and Auxiliary Facilities are neglected, since it is evad-ing the purpose of this study. Note that, however, stainless steel structures, poten-tially needed by some units, are included in the EPC. The main reason is the strictrelation with the footprint of the plant, which is not assessable at this stage ofadvance.

� Engineering and Construction, as well as Contractor's Fee, have been neglected,since they would have spoiled con�dentiality of both parties. Furthermore, thesevoices would be useful more to the investor, while from the process point of view,which this thesis purpose is to analyze, are negligible.

� Contingency, has been kept untouched, since considered realistic.

� As a consequence to the previous assumption: the Total Plant Indirect Cost (TPIC)is null and the Total Plant Direct Cost (TPDC) coincides with the Total Plant Cost(TPC).

The DFC of the SU plant is equal to 30% of the SS one, meaning a saving regardingthe CapEx of slightly more than 30 million e. The equipment purchase cost, aroundwhich all the other categories are evaluated, is 2.4 times greater for the SS case comparedto the SU one, with a di�erence around 10 million e, as clearly denotable from Graph8.1.

SectionEPC [e]

SS SU

USP 2 370 000 1 440 000DSP Pre-Viral 4 490 000 2 712 000DSP Post-Viral 1 421 000 894 000Media 1 615 000 161 000Bu�er 4 496 000 824 000

Table 8.3: Equipment Purchase Cost: Section Speci�cation

As understandable by Table 8.3, the main di�erence is brought by the vessels incomparison to the holding skids, impacting evidently in Media and Bu�er sections. In the

94


1 2 3 4 5 6 7

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

·107Cost(e)

SS SU

Figure 8.1: IC comparison. Refer to Table 8.2 for the categories.

stainless steel case the weight of this equipment is around 45%, while for the single useasset only 16%, with an absolute di�erence around 7 million e.

As expected, the SU case is convenient from all perspective up to this point. Plasticis cheaper than steel, leading to immediate savings and a faster entrance in the market.The stainless steel has a reputation of long term choice and the con�dence given by allthe years that have passed until nowadays. The last word in the economic panorama hasto come from the OpEx, which can con�rm or reverse the previous statement.

8.2 Operating Costs

The operating cost to run a biochemical plant in the sum of all expenses associated withraw materials, consumables, labor, heating/cooling utilities and power, waste disposal,overhead, as well as additional operational costs. Dividing the annual operating cost bythe annual production rate yields to the unit production cost (in $/kg). Biotechnology isa unique industry when it comes to the range in unit production cost. There are productsthat cost less than 1.0 $/kg, like citric acid and others that cost more than 10 million $/kgto make, like a therapeutic monoclonal antibody.

Operating costs are of di�erent nature: direct or indirect, �xed or variable. Fixed costs

95


are those that are incurred regardless of volume of product output, like depreciation, whichis an equipment-dependent cost. A variable cost, instead, is raw materials'. Most othercosts have a �xed and a variable component. Di�erently from IC, operating costs ranges,

Cost Item Type of Cost Range of values (% of total)

Raw Materials Direct 10− 80Labor Direct 20− 50Consumables Direct 1− 50Lab/QC/QA Direct 2− 50Waste Disposal Direct 1− 20Utilities Direct 1− 30Equipment-Dependent Indirect 10− 70Miscellaneous Indirect 0− 20

Table 8.4: Operating cost items and ranges

as shown in Table 8.4, are too wide to be averaged. Therefore a proper dedicated detailedanalysis must be pursued.More speci�cally, the annual operating cost (AOC) is calculated as the sum of the followingcost items:

� Materials cost

� Consumables cost

� Labor-dependent cost

� Utilities (heating/cooling utilities and power) cost

� Waste treatment/disposal cost

� Facility-dependent cost

� Laboratory/QC/QA cost

� Transportation cost

� Miscellaneous costs

� Advertising/selling costs

� Running royalties

� Failed product disposal cost

If a project includes streams that are classi�ed as credit or generated power that is recycledor heat that is recovered from operations, then a net annual operation cost (AOC) iscalculated by subtracting these credits and/or savings from the AOC.

96


8.2.1 De�nitions

Material Cost

This is the total cost of all bulk materials (pure components and stock mixtures) anddiscrete entities that are utilized as raw materials in a process. These may include bulkmaterials and/or discrete entities contained in process input streams that are either clas-si�ed as `raw material' or `cleaning agent' streams, and bulk materials that are used asheat transfer agents in process operations. Biopharmaceutical examples are: cell culturemedia, recovery bu�ers and cleaning materials. Water for Injection (WFI), which will bea big issue of discussion, costs 0.05− 0.2 $/kg, 100 to 500 times as much as city water.The annual cost of each material is calculated by multiplying the corresponding unit cost(i.e., purchasing price) by the corresponding annual amount that is utilized in a process.The user can specify the unit costs of materials, either pure components or (prede�ned)stock mixtures, whereas the corresponding annual amounts are automatically calculatedby the program as part of the simulation. For completion, the user can classify the streamsas raw material or cleaning agent.The annual cost of a heat transfer agent may include a lumped cost that is included inthe Utilities cost category and a material-based cost that is include in the Materials costcategory. The material-based cost is calculated based on the annual amount and unit costof the associated bulk material that is consumed for producing that agent.

Consumables Cost

Some equipment require the use of at least one consumable, that may be used up, fouled,or otherwise damaged during processing. For example, a chromatography column requiresthe use of a resin. This cost element includes the costs of periodically replaced materials,such as membranes, chromatography resins, activated carbon, and other materials whichmay be required for the operation of process equipment. More easily, all single-use equip-ment, for de�nition, exploits consumables with replacement frequency of 1 item/batch.The annual cost of a consumable type utilized by an equipment unit is calculated by mul-tiplying the corresponding unit cost (expressed as purchase cost per consumable amount)by the corresponding annual amount consumed:

AnnualCost = UnitCost×AnnualAmount

The annual amount consumed is calculated by multiplying the consumable amount peruse by the annual number of replacements:

AnnualAmount = AmountPerUse×AnnualReplacements

The consumable amount per use is calculated by the multiplying the consumption rate (ex-pressed as consumable amount per consumption basis) by the consumption basis (numberof equipment unit or equipment size):

AmountPerUse = ConsumptionRate× ConsumptionBasis

97


Finally, the annual number of replacements is calculated by multiplying the consumablelife (or replacement frequency) expressed per operating basis (operating cycles or hours)by the equipment's annual operating basis:

AnnualReplacements = ReplacementFrequency ×AnnualOperatingBasis

The user can specify the purchase cost, the consumption rate (per consumption basis), andreplacement frequency (per operating basis). The annual amount and cost of consumablesare calculated by the program as part of the simulation. The user can add consumablesin the Consumables Databank, through which it is also possible to view and edit purchasecost and life (or replacement frequency).

Labor-Dependent Cost

This cost includes all labor-dependent operating costs except those for laboratory analyses,quality control and quality analyses, which is included in the Laboratory/QC/QA cost.This is estimated based on the total number of the various operations as a function of time.It is important to underline how labor requirement in a batch manufacturing facility varieswith time. However, in our case study, the situation is simpli�ed since it is a single productfacility and since no shifts are considered. Another factor in�uencing the evaluation isthe level of automation of the plant. SU systems require usually more manual operationslabor demanding with respect to SS counterparts. For example, 2 to 3 operators areneeded to set up a SUB, while in a highly automated SS facility, a single operator fromcontrol room can handle up to 6 di�erent reactors. In general, a typical biotech companythat deals with high-value products will allocate at least one operator to each processingstep during its operation. This condition is the preset one of the simulator.In SuperPro Designer, the labor-dependent cost is calculated at the section level. Morespeci�cally, a total labor cost (TLC) is calculated for each section as the sum of the laborcosts of the di�erent labor types (i.e., operator, supervisor) that may be required for thatsection. The labor cost of each labor type is calculated by multiplying the correspondinglabor demand per type (LDT) by the corresponding labor rate (i.e., unit cost) per type(LRT).For each section, the LDT may include an itemized estimate (operating labor as de�nedin the process on a step-by-step basis), and a lumped estimate (additional labor de�nedon a lumped-time basis). By default, only the itemized estimate is considered. For eachlabor type, the itemized estimate is calculated as the sum of individual labor demandsby all operations of that section. An estimate of direct labor demand (e�ective worktime devoted to process-related activities expressed in labor-hours per operating cycle orper operating hour) can be speci�ed for each added labor type. The actual (total) labordemand is calculated by dividing the direct demand by the direct time utilization factorof that labor type.The LRT of a section can be calculated using either lumped labor rate estimates for alllabor types, or detailed labor rate estimates for all labor types. By default, the secondoption is selected. For each labor type, a detailed labor rate is estimated by adjusting thebasic rate (i.e., the basic unit labor cost) for the following additional costs:

98


Fringe Bene�ts this refers to expenditures that are paid out by the company to covervarious bene�ts which are not included in the basic labor rate.

Supervision this refers to the salaries of non-operational sta� engaged in supervision ofoperational and clerical sta�.

Operating Supplies this includes everyday items required to keep the plant in properrunning condition, as well as clothing, tools, and protective devices for operators.

Administration this refers to the cost of non-process-related administrative and secre-tarial support.

The above costs are speci�ed as factors of the basic rate. The detailed labor rate of alabor type is estimated as:

DetailedLaborRate = BasicRate (1 +Benefits+ Supervision+ Supplies+Admin.)

Utilities Cost

This is the total cost of heating/cooling utilities (i.e., heat transfer agents) and power uti-lized in a process. USA examples, in terms of unit cost, are: electricity around 0.1 $/kWh,heating steam around 4 - 8 $/1000kg, clean steam (generated using puri�ed water) around10 - 50 $/1000kg, refrigerants around 0.05 - 0.1 $/1000kcal of heat removed.Utilities Cost is the sum of the following costs:

� the cost of heat transfer agents utilized in every process operation

� the cost of power utilized in every process operation (like electricity)

� the cost of additional power that may be required for each section; this includes thepower consumption for unlisted equipment, support operations (e.g., night lighting),or other purposes that are not directly associated with the execution of any speci�coperation.

Note that there are two kinds of utilities (heat transfer agents and power) associated withan operation: utilities that are essential for an operation model (e.g., the heating/coolingrequired to achieve a temperature speci�cation of a stream, or the power required to drivean equipment) and auxiliary utilities that may be optionally speci�ed for an operation. Foressential utilities, the mass �ow rate of a heat transfer agent and the power consumptionof a power type can be either set or calculated as part of the simulation. For auxiliaryutilities, these are always set by the user. The user also speci�es the unit costs of heattransfer agents and/or power types utilized in a process and the amounts of additionalpower that may be required for each section. The annual amount and cost of each utilityare calculated by the program as part of the simulation. The annual cost of a power typeconsumed by process operations is calculated based on the annual amount and purchasingprice of that power type.It is needed to clarify the relation with the Material Cost: the annual cost of a heattransfer agent may include a lumped cost that is included in the Utilities cost category

99


and a material-based cost that is include in the Materials cost category. The lumped costis calculated based on the annual amount of that agent and a lumped unit cost speci�edfor that agent (either on a mass or energy basis). Furthermore, sometimes, puri�ed water,used for bu�er or media and equipment cleaning, is classi�ed as a utility and not as a rawmaterial, thus disrupting the balance towards this section.

Waste Treatment/Disposal Cost

This includes the cost of treatment or disposal of certain process output streams thatcorrespond to wastes (e.g., undesirable by-products, solvents, etc.). Wastes are typicallyclassi�ed as solid, aqueous, organic, or gaseous (emissions). Depending on the phase,the complexity of the facility, and the nature of the waste, the treatment cost can varysubstantially. For a waste stream, you can either specify directly the unit cost of wastetreatment/disposal, or allow it to be calculated based on the corresponding cost associatedwith each component present in that stream and the stream's composition. The annualamount and cost of each waste stream are calculated by the program as part of thesimulation.

Facility-Dependent Cost

This accounts for additional costs related to the use of a facility. In cases of new (green-�eld) designs, where no prior experience on the use of equipment exists, this is typicallycalculated as the sum of the costs associated with equipment maintenance, depreciationof the �xed capital cost, and miscellaneous costs such as insurance, local (property) taxesand possibly other overhead-type of factory expenses. For existing multi-product facili-ties, however, which are usually operated in batch, the estimation of maintenance- anddepreciation-related expenses and the allocation of these expenses among di�erent projectsmay not be straightforward. Therefore, it is usually more convenient for such facilitiesto calculate facility-related costs based on operating parameters. Optionally, in SuperProDesigner, both approaches can be used together.Generally, the facility-dependent cost of a section may include the following estimates:

� an estimate based on equipment usage/availability rates

� an estimate based on a lumped facility availability rate

� an estimate based on the production rate of the process

� an estimate based on capital investment parameters (i.e., maintenance, depreciationand miscellaneous costs)

Estimation based on Equipment Usage/Availability Rates This estimate of thefacility-dependent cost is calculated as the sum of individual equipment contributions tothis cost. Each contribution may be viewed as a rental fee for the use of the correspondingequipment, which is calculated either by multiplying the equipment usage rate (the costingrate based on usage) by the hours that the corresponding equipment is actually used by asection (usage basis), or by multiplying the equipment availability rate (the costing rate

100


based on availability) by the hours that the corresponding equipment is reserved for asection (availability basis).Optionally, the user may exclude some of the equipment utilized in the modeling (e.g.mixers, splitters, etc.) so that they do not arti�cially in�ate the overall equipment usage-or availability-dependent cost. The equipment usage and availability hours are calculatedby the program as part of the simulation.

Estimation based on Facility Availability Rate Instead of tallying up the equip-ment usage or availability hours for each equipment, one may utilize a �at rate for theentire facility. Using this approach, a facility-dependent cost is calculated by multiplyingthe speci�ed facility availability rate by the hours that the facility is available. The latterare calculated by the program as part of the simulation.It is possible to use site data for the facility availability rate of a process section. It maybe convenient to store distinct rates behind model database sites and allocate the sitewith the most appropriate rate to the relevant section(s) within the recipe.

Estimate Based on Production Rate The facility-dependent cost may be estimatedbased on the unit production reference rate speci�ed from the Main Product/Revenuestream and �ow basis or from the unit rate reference �ow. It is calculated by multiplyingthe speci�ed unit cost by the unit production cost reference rate, which is calculated bythe program during simulation.Furthermore, the Ùnit Reference Rate' (or Flow) is used to convert a calculated annualcost into a per-unit cost (e.g., cost per kg of raw materials or products). It correspondsto the total �ow or a component �ow in a selected Ùnit Reference Stream'. If theÙnit Reference Stream' is the same as the `Main Product/Revenue Stream', the ÙnitReference Rate' will correspond to the `Main Product/Revenue Rate'. Optionally, the`Main Product/Revenue Rate' may be discounted by the product failure rate.However, these options are invalid since it has already been explained that no pro�tabilityanalysis will be conducted, therefore no main product stream has been de�ned.

Estimation based on Capital Investment Parameters This method will use thepurchase cost of equipment as reference for computing (indirectly) the facility-dependentoperating cost. This may include the following costs:

� maintenance

� depreciation

� miscellaneous costs (insurance costs, local taxes and factory expenses)

Maintenance Cost The maintenance cost accounts for the maintenance of theequipment and the facility in general. It can be estimated either using equipment-speci�cmultipliers, or as a percentage of the section's DFC that is assigned to this project (usually10%). Note that if the section's DFC is set by user, the �rst option will not be available.If the �rst option is selected, the maintenance cost is calculated as the sum of individual

101


equipment maintenance costs. The maintenance cost of each equipment is calculated bymultiplying its purchase cost (the fraction that is assigned to this project) by a suitablemaintenance factor. For equipment resources that are shared by multiple sections, themaintenance cost is distributed to the various sections based on time utilization. Morespeci�cally, the maintenance cost of an equipment that is allocated to a particular sectionis calculated by multiplying the total maintenance cost of that equipment by the fractionof total utilized time that this equipment is being utilized by unit procedures in thatsection. The latter is calculated by the program as part of the simulation.

Depreciation Depreciation is an income tax deduction that represents a �xed capi-tal loss which is mostly due to equipment wear out and obsolescence. It may be consideredas a time-dependent operating cost, spread over a prede�ned depreciation period. For eachsection, SuperPro Designer depreciates the fraction of DFC that is assigned to this projectand has not been depreciated already minus its salvage value1 at the end of the projectlifetime. The user also has the option to depreciate the startup and validation cost.In the general case, the total depreciable amount, dtot, of a section's assets over the entiredepreciation period is calculated as:

dtot =

N∑1

dk = B − S

where:B = fp × UDFC + Cs

S = fs × fp × UDFC

and:

� dk is the depreciable amount of a section's assets in year k

� N is the depreciation (recovery) period

� B is the cost basis of a section's assets (the cost right before the project starts)

� S is the salvage value of the section's assets at the end of the depreciation period

� fp is the fraction of a section's DFC that is assigned to this project

� UDFC is the undepreciated DFC of a section (i.e., the fraction of a section's DFCthat has not been depreciated already)

� Cs is the startup & validation cost of a section

� fs is the salvage fraction of the entire DFC

Three classical methods are available for the calculation of the annual depreciation of asection's assets, namely:

1Salvage Value (Residual Value) is the value of (pro�t obtainable by) the plant at the end of its lifetime

102


� the straight-line method

� the declining balance method

� the sum-of-the-years-digit method

The straight line method assumes a constant annual depreciation which is calculated foryear k as follows:

dk =dtotN

The declining balance method assumes a constant depreciation rate and, therefore, de-creasing annual depreciable amounts. Based on this method, the annual depreciation foryear k is calculated based on the following equations:

dk = R× (BV )k−1

where:(BV )k = B(1−R)k

R = 1− f1Ns

and:

� R is the depreciation rate

� (BV )k is the book value (i.e., the amount that has not been depreciated) of asection's assets in year k.

The sum-of-the years-digit method also assumes decreasing annual depreciable amounts.Based on this method, the annual depreciation for year k is calculated as:

dk =NkdtotSY D

Nk = N − k + 1

SY D =N(N + 1)

2

where:

� Nk is the remaining depreciable life at the beginning of year k, and

� SY D is the sum-of-years digits.

The annual depreciation of a section's assets, which contributes to the facility-dependentcosts of a section, is calculated based on the straight-line method. This method de�nesa constant depreciation equal to the initial investment divided by the expected lifetime,expressed in years (usually 10 for preliminary cost estimates), of the plant. However,USA government allows corporations to depreciate equipment in 5−7 years and buildingsin 25 − 30. These ranges communicate a huge variability, not even considering di�erentlegislations around the globe. Land is never depreciated.

103


The undepreciated DFC of a section can be calculated either based on the undepreciatedpurchase cost of equipment, or based on the speci�ed percentage of a section's DFCassigned to this project that has already been depreciated. Note that the �rst option isonly available if the DFC is not set by the user. In this case, the undepreciated DFC iscalculated similarly to the DFC except that the purchase cost of listed equipment is nowcalculated as the sum of undepreciated equipment purchase costs. For each equipment,the undepreciated purchase cost is determined by subtracting the fraction of the purchasecost that has already been depreciated from the purchase cost. The fraction, fp, of DFCthat is assigned to the project is speci�ed, either directly or on a unit-by-unit basis.

0 2 4 6 8 10

20

40

60

80

100

Y ears

PercentageUndepreciated

Straight Line Declining Balance Sum-of-the-Years-Digit

Figure 8.2: Depreciation visual summary of a �ctitious case

Miscellaneous Facility-Dependent Costs Miscellaneous costs include the follow-ing individual costs:

Insurance insurance rates depend to a considerable extent upon the maintenance of asafe plant in good repair condition. A value of 0.5− 1 % of DFC is appropriate formost bioprocessing facilities. The processing of �ammable, explosive, or dangerouslytoxic materials usually results in higher insurance rates. The local (property) tax isusually 2− 5 % of DFC.

Local Taxes these refer to local property taxes (not income taxes).

104


Factory Expenses these refer to the overhead cost incurred by the operation of non-process-oriented facilities and organizations, such as accounting, payroll, �re pro-tection, security, cafeteria, etc. A value of 5− 10 % of DFC is a safe guess for thesecosts.

Each of the above cost items is speci�ed as a percentage of the DFC. It is possible to usesite data for the miscellaneous facility-dependent costs of a process section. It may beconvenient to store distinct sets of these factors behind model database sites and allocatethe site with the most appropriate factors to the relevant section(s) within your recipe.

Laboratory / QC / QA Cost

This accounts for the cost of o�-line analysis, quality control (QC) and quality assurance(QA) costs. Chemical analysis and physical property characterization from raw materialsto �nal product is a vital part of chemical operations. This cost is usually 10 − 20% ofthe operating labor cost. However, for certain biopharmaceuticals that require a largenumber of very expensive assays, this cost can be as high as the operating labor. For suchcases, it is important to account for the number and frequency of the various assays indetail. Changes in lot size that can reduce the frequency of analysis can have a majorimpact on the bottom line. For our case study, this goes beyond our purposes.In SuperPro Designer, this cost is estimated for each section. It may include a lumpedestimate calculated as a percentage of a section's total labor cost (TLC), and a detailedestimate calculated as the sum of the costs of di�erent tests carried out and of a �xed costfor QA activities; in that case, the user speci�es detailed information about the numberand unit cost of the various assays along with a �xed cost for QA activities.

Transportation Cost

This accounts for the cost of long-distance transportation of raw materials and productsby sea, land, and air. Transportation operations are the only process steps that cancontribute to transportation cost. The following operations are available:

� Transport by Truck (Bulk Flow)

� Transport by Truck (Discrete Flow)

� Transport by Train

� Transport by Sea

� Transport by Air

The primary objective of transportation operations is to account for and estimate theshipping cost associated with the transportation of raw materials and �nished productsof a manufacturing facility. Through the cost-related tab of a transportation operation'sdialog, the user speci�es the following cost factors:

� �xed cost (per shipment)

105


� quantity dependent cost (i.e., cost per shipping quantity)

� quantity and distance dependent cost (i.e., cost per shipping quantity and shippingdistance)

The annual transportation cost (ATC) is estimated using the following equation:

ATC = s0 (C0 + s1C1 + s1s2C2)

where:

� s0 is the number of shipments per year,

� C0 is the �xed cost,

� s1 is the quantity per shipment,

� C1 is the quantity-dependent cost,

� s2 is the shipping distance, and

� C2 is the quantity- and distance-dependent cost.

For units that transport bulk material as well as discrete entities, the above equation isapplied twice.

Miscellaneous Operating Costs

This cost element accounts for:

� on-going R&D expenses,

� process validation expenses, and

� other overhead-type expenses that are not covered by other cost categories.

By default, this cost item is zero. The relevant speci�cation parameters can be modi�ed.For each section, the process validation expenses are speci�ed as a �xed cost (per yearor per batch). Each of the on-going R&D and other expenses categories may include a�xed cost term (per year or per batch), and a variable cost, which is speci�ed as cost perkg of main product. For the �rst term, the conversion between annual cost and cost perbatch is based on the speci�ed or calculated annual number of batches for this project.For the second term, an annual cost is calculated by multiplying the variable cost bythe annual `Main Product/Revenue' rate. Optionally, the `Main Product/Revenue' ratemay be discounted by the main product failure rate. The latter, however, again, willnot be taken into consideration. For each section, the annual miscellaneous costs arecalculated as the sum of annual on-going R&D expenses, process validation expenses andother expenses. The total miscellaneous costs for the project are calculated as the sum ofmiscellaneous costs for each section over all sections.

106


Advertising and Selling Costs

This is the cost that is associated with the activities of the sales department. It mayinclude a �xed annual cost, and a variable cost, which is speci�ed as cost per kg of mainproduct. For the second term, an annual cost is calculated by multiplying the variablecost by the annual `Main Product/Revenue' rate. Optionally, the `Main Product/Revenue'rate may be discounted by the main product failure rate. The latter, however, again, willnot be taken into consideration.

Running Royalties

If the process, any part of the process, or any equipment used in the process are coveredby a patent not assigned to the corporation undertaking the new project, permission touse the teachings of the patent must be negotiated, and some form of royalties is usuallyrequired. The licensing agreement usually calls for a �at charge per unit of product orelse a percentage on the sales dollar. In SuperPro Designer, the user speci�es the runningroyalty expenses as cost per kg of main product. Note that the default value for this costis zero. The annual running royalty expenses are determined by multiplying the speci�edcost by the annual `Main Product/Revenue' rate. Optionally, the `Main Product/Revenue'rate may be discounted by the main product failure rate. This, however, again, will notbe taken into consideration.

Failed Product Disposal Cost

This is the cost associated with the disposal or o�-site recycling of scrapped product. InSuperPro Designer, the user speci�es the disposal cost per kg of main product scrapped,and the main product failure rate as percent of main product. Note that the defaultvalues for these parameters are zero. An annual failed product disposal cost is calculatedby multiplying the corresponding disposal cost per kg of main product scrapped by the`Main Product/Revenue' rate and the main product failure rate. This, however, again,will not be taken into consideration.

8.2.2 Results

SS SU

Cost Item e % e %

Raw Materials 98 000 1.04 55 000 0.46Labor-Dependent 5 736 000 61.11 4 247 000 35.91Facility-Dependent 1 799 000 19.16 754 000 6.38Consumables 1 733 000 18.46 6 770 000 57.25Utilities 22 000 0.23 0 0.00

TOTAL 9 388 000 100.00 11 826 000 100.00

Table 8.5: Annual Operating Cost - Process Summary

107


Compared to capital investment, operating costs give a wider opportunity of discus-sion.

Facility-Dependent For brief recall, the Facility-Dependent Cost is a function of:equipment usage rate, maintenance, depreciation, insurance, local taxes, accounting, pay-roll, �re protection, security, cafeteria, etc. For details, please refer to section 8.2.1 onpage 100. It is automatically evaluated by the software, mainly as a function of the DFC.For assumptions similar to the ones stated for the CapEx, giving importance to processrelated parameters, only maintenance has been considered. Furthermore, following theline of thought, the correlation with capital investment goes straight to the equipmentpurchase cost, being more strictly related to the di�erent units. A value of 10% hasbeen exploited. The snowball e�ect, caused by this system of calculation of SuperProdesigner, operates in a way that the most expensive plant is also the most onerous torun. Facility dependent cost, corrected in such a way, for the SS is 1.8 million e, whichis 1 million e higher than the one of the SU. Without the just stated correction it wouldhave almost covered all the OpEx of the disposable case.Neglecting miscellaneous factory expenses is a big statement, but is also conservative.This voice can easily overturn the overall estimation if overestimated. It has been thoughtwiser to limit the extent to parameters (like maintenance), well known from a processionalperspective (proper of this elaborate), evading from more �nancial parameters (like de-preciation). Even if taking a similar way is limiting from the investor point of view, itis the aim of the thesis to be as transparent and as above criticism as possible, at theevident price of limiting the extent of the work.

Labor Labor result is the �rst big statement against literature claims. The numbersobtained are absolutely comparable, instead of certifying a de�nite advantage to the SUfacility. The large majority of the operations, indeed, is the same and it is reasonable toa�rm that the same number of operator is required. Anyway, even if some distinctionshould have been made, generally the SS facility is (potentially) more automated thanthe SU one, as already stated at section 8.2.1 on page 98.The main di�erence is brought by the operations of SIP&CIP (for the SS facility) andSETUP (for the SU facility). Even if the setup of the SU units, described at section 6.2on page 61, is the most impacting operation regarding labor (it is the only one which canneed two operators simultaneously), it is generally briefer than the respective SIP&CIPoperations. This statement is assessed by Graphs 8.3 and 8.4: the red line is the instan-taneous request of labor by the process, while the blue line is the day averaged quantity;the ordinate variable is Number of Operators, while the abscissa is the process time-line.

� SS has maximum peak of 6 operators versus the 10 of the SU facility. This is becausemultiple setup operations occur contemporaneously.

� SS has a daily average value oscillating from 1 to 3 operators, with a likely estimate of2 operators as an overall average. SU daily average is more variable, 0−5 operators,resulting in a similar guess of 2 operators equivalent average.

108


Figure 8.3: SS generic operating month of labor requirement

Figure 8.4: SU generic operating month of labor requirement

109


It is deducible that SU is exploited heavily during shorter periods, because of generallylower equipment occupancy factors and no need for schedule arrangements during projectphase. Indeed, SS facility is more distributed also due to the need of prepare with largeadvance some bu�ers, because of equipment and CIP skids availability. Therefore, aschedule adjustment under this focus of the SU process should lead to similar graphicoutput.The equal estimate of overall average operators required is proof of the same economicoutput of labor dependent operating cost.

Observe that SuperPro does not consider labor shifts, nor labor segregation (an op-erator cannot freely operate through the whole plant, due to sanitary and classi�cationrestrictions). Further speci�c analysis could generate problems regarding the more vari-able situation of the SU facility, which is not applicable in reality. For example, it is notconceivable to have nobody in the plant, nor to have an operator coming for a single setupoperation and pay only for that. However, the economic results obtained at this stage arejudged suitable.

Consumable Units Cost (e) Annual Amount Annual Cost (e) %

DEF Cartridge 940 195 item 183 300 2.71PBA Chrom Resin 1410 1179L 1 662 778 24.56Membrane 376 4m2 1655 0.02Mixer Bag 2000L 2000 507 item 1 014 000 14.98Storage 1000L 1100 195 item 214 500 3.17Storage 500L 400 117 item 46 800 0.69Storage 3000L 1900 78 item 148 200 2.19Mixer Bag 1000L 1500 117 item 175 500 2.59Storage 2000L 1500 741 item 1 111 500 16.42Mixer Bag 3000L 2100 468 item 982 800 14.52Storage 100L 200 117 item 23 400 0.35Mixer Bag 100L 900 78 item 70 200 1.04Wave Bioreactor Bag 3200 39 item 124 800 1.84Seed 500L 5500 39 item 214 500 3.17Bioreactor 2000L 9600 39 item 374 400 5.53Seed 200L 4300 39 item 167 700 2.48Mixer Bag 500L 1100 195 item 214 500 3.17Storage 200L 200 195 item 39 000 0.58

TOTAL 6 769 533 100.00

Table 8.6: Consumables Cost - SU Process Summary

Consumables In Table 8.6 only SU consumables are reported. The correspondent de-tailed table of the SS case is similar to the �rst three rows of the SU case, where the maindi�erence is brought by the additional �ltration step present in the SU facility instead ofthe centrifuge in the harvest section. The di�erence expected, as announced by literature

110


and common sense, is clear: 6.8 million e versus 1.7 million e, weighting respectively for57% and 18% of the total operative costs. This voice, is impacting as expected, such asthe overall investment cost in favor of the SU facility. In other words, it is economicallysu�cient to justify the SS choice, regarding the operating costs, being the di�erence ofthe two choices enclosed in consumables, assessed roughly around 5 million e. Charts 8.5and 8.6 show this correlation.Furthermore, consumables need to be further considered in the environmental aspect,

61.11%

19.16%

18.46%

1.27%

LaborFacility

Consumables

Materials & Utilities & Others

Figure 8.5: SS OpEx Pie Chart

35.91%6.38%

57.25%

0.46%

LaborFacility

Consumables

Materials & Utilities & Others

Figure 8.6: SU OpEx Pie Chart

where the disposal of plastic can back�re onto the Waste Treatment/Disposal cost, a�ect-ing indirectly the just examined topic. However, a con�dential source assessed a realisticvalue of disposal around 160 e per tonne, which results in a operating cost per year of

111


58 000 e. This result, with respect to the overall value and the e�ective di�erence withSS, will be neglected.

Utility/Material Unit Cost (e) Annual Amount Annual Cost (e) %

SS

Std Power 0.09 5448 kWh 512 2.32Steam 11.28 1895MT 21 379 96.83Cooling Water 0.05 181 MT 8 0.04Chilled Water 0.38 479 MT 180 0.81subTOT 22 079 100.00

PW 0.01 2 200 715 kg 28 610 29.33WFI 0.04 1 969 638 kg 68 937 70.67subTOT 97 547 100.00

TOTAL 119 626

SU

Std Power 0.09 2558 kWh 240 55.53Steam 11.28 1 MT 17 3.86Chilled Water 0.38 468 MT 176 40.61subTOT 433 100.00

PW 0.01 77 922 kg 1011 1.82WFI 0.04 1 555 437 kg 54 442 98.18subTOT 55 453 100.00

TOTAL 55 886

Table 8.7: Material and Utilities Detailed Costs

Materials and Utilities Raw materials, in the strict meaning of the term, have beenneglected. Voices like cells, culture broth and bu�er ingredients can heavily impact oper-ating costs, but these voices are the same for both SS and SU facilities. Therefore, sinceour main goal is a comparison, and since no appreciable di�erences are recorded amongraw materials, we are going to consider only PW and WFI. Furthermore, the discriminantinformation exploited during future calculation will be the di�erence of these quantities(the net amount) between SS and SU cases.

Since in the simulator not everything inserted in Table 8.7 was grouped, it is necessaryto overdo some calculation to further comment. The total costs are: SS - 120 000 e;SU - 56 000 e. Both numbers are extremely low compared to the other outcomes of theoperating costs analysis, as visually displayed in pie charts 8.5 and 8.6. Correlated to therelative OpEx, the percentages of SS and SU stay below 0.5%, even if it was expected tobe much more impacting for the canonic stainless steel facility.Observe that it has been considered that all the bu�er are made of pure WFI and all mediaand nutrients of PW, as already expressed in Part II. All SIP and CIP requirements areaccounted for.

Regarding utilities, the di�erence in the standard power is attributable to one of the

112


rare process di�erences of the plants: the centrifuge. It is responsible for 2400 kWh/yr outof 2900 kWh/yr, total di�erence. Furthermore, there are ancillary costs called GeneralLoad and Electricity per Unlisted Equipment, expressed as percentages of the total utilitiesload (respectively 15% and 5%), which can justify the remaining gap.

8.3 Conclusions

The following assumptions were made:

1. No investment cost for the SIP and CIP skids.

2. No multi-product features or variable product campaigns.

3. No exploitation of the di�erent equipment utilization rate.

4. Material cost evaluation only for PW and WFI.

5. No labor di�erentiation.

6. No bu�er storage agitation considered for SUS.

Points 1, 2 and 3 favor SS facility.Points 4, 5 and 6 favor SU facility.

SS (e) SU (e)

CapEx 112 493 000 52 831 000

OpEx 28 664 000 21 144 000

Table 8.8: Cost Estimation Summary

Deepening Table 8.9, powder handling and bu�er agitation are interlinked and some-how magni�ed by the in-line dilution studied and exploited for the NaOH. Concentratedbu�er require high powder (or pellets) amount, leading to problems for the SUS, not hav-ing an agitation as e�cient as the SS one. Furthermore, in-line dilution does not allowsampling and therefore costs of control instrumentation and risk assessment increase. Ac-curacy of in-line dilution can be a concern. Bu�er storage agitation in SUS is completelyabsent: that is why the score is so unbalanced towards SS.Linked to this aspect, waste management di�erence is brought by plastic disposal. Ingeneral, solid handling in the single use facility is more challenging. Other wastes usuallyare not critical, since we are after all dealing with biotechnology.Water consumption di�erence is caused by SIP & CIP skids. Even if SU facility makeusage of slight more bu�er, due to the additional �ltration step in the harvest section,this aspect is not impacting. However, the disposable items consumption, practically no-ticeable only for the single use case, balances the score.The last voice of the table regards schedule. Even if the maximum batches per year are

113


Legend : 1.Poor - 3.Acceptable - 5.Good - 7.Best

SS SU

CapEx 1 7OpEx 5 7Layout - footprint 1 7Powder handling 7 3Bu�er agitation 7 1PW and WFI consumption 1 7Consumables consumption 7 1Waste management 7 3Flexibility 1 7

Total score 37 43

Table 8.9: Arbitrary score table for evaluation purpose

almost the same, SU facility has lower equipment occupancy factors in almost every otherunit, in other words, where the cell culture is not the determining operation. Therefore,thinking o� the box, the SU choice o�ers more options, like dealing more USP staggeredsections with a single DSP. Furthermore, it is possible to consider a multiproduct facilityexpansion, an increase or decrease of production following the market request, and similarscenarios without having pharaonic shut-down periods and prohibitive investments.

Given the discussion and the results exposed, the economic recommendation is notunilateral. Indeed, CapEx for sure favors the single-use choice as much as Opex favors thestainless steel one. The latter is the data against the current of literature. Before achievingthe results, raw materials cost and utilities were supposed to balance the consumable voice.However, when trying to deepen the results achieved by similar articles, the path was darkand mysterious and the directions poor and misleading. However it would be arrogantand substantially wrong to generalize. Our case study is driven a lot by the very longreaction duration, which, not only bottlenecks the process, but also dilutes the e�ect of theSIP & CIP operations. If we had considered a bacterial reaction in a fermenter, which isusually faster, the number of batches produced per year would be non negligibly favoringthe SU facility. Moreover, by considering a �xed productivity, of course the comparisonis more strict and well founded, but the main feature of the SU facility is deaden. Ifa multi-product facility had been considered, at the price of much more complexity, ofcourse, again the conclusion section of this elaborate would have been di�erent.

In light of all, the �nal decision is, economically speaking, very much case driven.Therefore, probably, the best solution, and the most coward possible, is an hybrid sys-tem. Literature assesses good results being vastly cataloged by many societies trying torevamping their facilities with single-use equipment. Indeed, di�erent sections of the plantmay be more suitable for one choice or the other and this work assesses how a processsimulator could easily help out these kind of decisions.

114

Chapter 9

Environmental Impact

Environment has become more and more appealing during last years. Laws and rulesworldwide are getting stricter in order to achieve a more ethic process building. Industrieshave consequently put a lot of e�ort and money into this �eld, often also in the researchand development sector, trying to optimize the problem with one unspoiled variable. Theresult is that brand-new branches of specialization, a�ecting education and job, havedeveloped, leading inevitably to di�erent philosophies of approach and overall complexity.

The approach chosen for our process has to thank The BioPharm International Guide,which has collected individuals and organizations calculations regarding carbon footprint[29]. This last term addresses a life-cycle assessment, as shown in detail at section 2.3on page 23, that however stops at a rougher level of detail. Of course, once the processprogressively acquires shape, the analysis can go deeper and integrate this work. Inother terms, our comparison will couple the two most signi�cant voices of the respectivefacilities:

� net WFI for the stainless steel one

� consumables for the single use one

Care that a thorough essay would have considered: utilities, materials, consumables, labor,footprint and HVAC, electricity, steelwork.

To solve the problem, the following assumptions are considered:

1. PW is neglected

2. Plastic transportation is neglected

3. Tubings are considered in the bag weights

4. Energy recapture from plastic is neglected

Observe that points 1 and 4 favor the SS case, while points 2 and 3 relieve SU one.Furthermore, plastic bags will be considered as if only composed of polyethylene, whichis not a strong hazard considering the data publicized by Thermo�sher Scienti�c in theSingle-Use BioProcess Containers Brochure, accessible online.

For WFI, the following data are used:

115

9 � Environmental Impact

� Electricity generated by average US mix of coal, gas, and other: 0.66 kgCO2

kWh

� Enthalpy of steam relative to water at 60°F: 3722 kJkg = 1.034 kWh

kg(personal calculation have found this value slightly greater than expected, but noneed for correction has been estimated)

� Boiler e�ciency: 85%

� Energy for chilled water to condense WFI: 17% of the total energy needed to produceWFI.

Calculations follow. First of all, we need to consider the incremental energy neededdue to boiler ine�ciency:

1.034 kWhkg

0.85= 1.216

kWh

kg

where the result is the total energy needed to produce 1 kg of steam.Considering that for multiple e�ect WFI generation, which is the most common

method, only 0.32 kg of steam are required for each kg of WFI, already assessing forthermal e�ciency [61]:

1.216kWh

kg steam× 0.32

kg steam

kg WFI= 0.389

kWh

kg WFI

Note that the power needed to carry on the WFI generation has been neglected, since theoverall value can be approximated to the one needed by steam. Next, we need to add thecondensation term:

0.389kWh

kg WFI+ 0.17× 0.389

kWh

kg WFI= 0.455

kWh

kg WFI

where the result is the total energy considered for WFI LCA assessmentThen, considering the annual WFI amount as the di�erence between SS and SU, and

the electricity conversion, we obtain the net WFI impact.

0.455kWh

kg WFI× 414 201

kg WFI

year= 188 565

kWh

year

188 565kWh

year× 0.66

kg CO2

kWh= 124 453

kg CO2

year

For plastic, the following data are used:

� PE embodied energy1: 81 MJkg PE = 22.5 kWh

kg PE [62]

� Plastic extrusion: 0.25 kWhkg plastic

1Embodied energy is the sum of all the energy required to produce any goods or services, consideredas if that energy was incorporated in the product itself.

116


� Plastic polymerization: 0.15 kWhkg plastic

� Stationary combustion emission factor for PE: 3.83 kg CO2

gallon PE = 1.01 kg CO2

L PE [63]

Further data can be found in the Handbook of Industrial Polyethylene and Technology.The missing information is the total kilograms of plastic used. To obtain it, the weight ofeach single use item has been approved and displayed in Table 9.1.Simply by multiplying the unit weight (Table 9.1) and the annual amount (Table 8.6,

Item Capacity (L) Weight (kg)

Storage Bag

100 0.5200 0.7500 2.51000 52000 10.53000 16

Agitated Bag

100 8.5200 9500 101000 13.52000 203000 20

Bioreactor Bag

25 0.5200 11500 122000 25

Table 9.1: Bag Weights Estimation

section 8.2.2 on page 110) for each kind of consumable, then summing up the data, it isobtainable the annual consumption of plastic.∑

AnnualAmount× UnitWeight = 36 075kg

year

Calculations follow. First, we can evaluate the carbon dioxide equivalents derivingfrom production energy requirements:

22.5kWh

kg PE+ 0.25

kWh

kg plastic+ 0.15

kWh

kg plastic= 23.15

kWh

kg plastic

23.15kWh

kg plastic× 36 075

kg plastic

year= 835 136

kWh

year

835 136kWh

year× 0.66

kg CO2

kWh= 551 190

kg CO2

year

117


Secondarily, we need to consider the disposal. A common solution is to exploit a sterilizingshredder. Through its yearly electricity request, the overall plastic impact would onlyincrease of around 250 kgCO2/year (no calculation detail is inserted due to con�dentiality;if interested, please contact author).The next disposal step considered is burning. Considering a medium LLDPE density, it ispossible to start by converting the stationary combustion emission factor and concludingby considering the annual plastic consumption:

1.01 kg CO2

L PE

0.918 kgL

= 1.1kg CO2

kg plastic

1.1kg CO2

kg plastic× 36 075

kg plastic

year= 39 690

kg CO2

year

Therefore, the total environmental impact of plastic is the sum of the previous:

551 190kg CO2

year+ 250

kg CO2

year+ 39 690

kg CO2

year= 591 130

kg CO2

year

SS SU

124 tons CO2year 591 tons CO2

year

Table 9.2: LCA results

As predictable, the environmental impact of plastic wide exploitation is condemningwith respect to the SU facility. The di�erence is more than 450 tons of CO2 per year,which is a big slice of both total values.

Literature, in the article mainly exploited for the calculations and in similar ones foundon the Internet, following what already stated in previous sections, claimed that raw ma-terials and utilities greater consumption of the stainless steel facility should have balancedthe consumables. Again, coherently with the results obtained from the analysis of the op-erating costs, our elaborate opposes such a statement. It is a must to remember, however,that our numbers are partial and is impossible to further estimate an impact on any �eld.Furthermore, environmental impact depends on local laws and society philosophy.

Anyway, literature generally assesses the light gravity of the contribution of the justanalyzed voices on the overall number, since voices like labor transfer (driving from hometo work) is worth as much, if not even bigger. Of course this should not be a justi�cation,but must make thinking about how much nowadays common habits are sometimes asdangerous as (apparently) innocent.

118

Chapter 10

General Comments

The last chapter of the elaborate is dedicated to the aspects that came up during datacollection and project development, but that, for complexity of the �eld, for lack of data,for evasion from the purpose, for con�dentiality, have not been treated thoroughly.

10.1 SU �uid waste treatment

The lower water output of the SU facility can lead to waste regulations infringement,indirectly operating on other components concentration. Diluting is forbidden, thereforea system of �uid waste treatment should be taken into consideration. This could revertthe results obtained through a super�cial analysis of the process. Some information,however, are obtainable only if the plant placement is already known, since law is locationdependent.

10.2 SU hidden maintenance costs

By default and intuition, usually, SU equipment are associated with no maintenance cost,deceiving through their disposable nature. However, there are non negligible items associ-ated to single-use equipment that can increase a lot this voice. For example, the disposablebioreactor needs auxiliary expenses associated to devices that need quali�cation, calibra-tion, substitution, spares, like the gas mixing station or pumps. This is why SuperProDesigner often considers an overestimate of economic data, for example by consideringunlisted equipment, trying to include hidden and unforeseeable information.

10.3 SU footprint advantages

SUS is granted being a closed environment. Therefore, it implies savings related to clean-liness classi�cation. It has not been considered this aspect, yet, because of complexity.However, HVAC (heating, ventilation and air conditioning) is an impacting factor in theplant realization. Its cost increases with the cleaning requirements of the area and SUSallows an important reduction of this kind of footprint, for example by downgrading clean

119

10 � General Comments

classi�cation of disposable only equipment area.Furthermore, being the SU holding equipment movable, the preventive preparation of�uids and the consequent storage is greatly facilitated. The facility should consider therealization of a cold room where to park procedure in idle time, instead of accomplishingcooling operations under constant stirring, which is the common case for SS facilities.Moreover, equipment connections of all kind, in a SU plant, will be �exible and sometimesnot even all plugged. Therefore, all auxiliary operations, which for example can simplybe an innocent visual monitoring of the unit, are largely facilitated.

10.4 SchedulePro®

An alternative process analysis and consequent process model output of the same recipewith outages was realized, exploiting the other software of Intelligen: SchedulePro. Anoutage is a period where procedure cannot operate, due to facility closure or labor un-availability. In the complementary study, the working day was split into three shifts of 8hours, where the night one, from 10 p.m. to 6 a.m., was set as outage. Furthermore, alsothe weekend was considered unavailable, from 10 p.m. of Friday to 6 a.m. of Monday.Of course, some operations can be set to be able to run through outages, otherwise, justthinking to the main reaction in the bioreactor, which lasts 14 days, it would not bepossible to execute a batch.Further analysis, concerning simulation of real client orders, tracking of inventory, what-ifevents were at disposal and a major awareness has been achieved.

120

Bibliography

[1] W. Ding and M. Jerold. Implementing Single-Use Technology in BiopharmaceuticalManufacturing. BioProcess International, pages 34 � 52, 2008.

[2] R. Jacquemart, M. Vandersluis, M. Zhao, K. Sukhija, N. Sidhu, and J. Stout. ASingle-use Strategy to Enable Manufacturing of A�ordable Biologics. Computationaland Structural Biotechnology Journal, 14:309 � 318, 2016.

[3] S. S. Farid. Process economics of industrial monoclonal antibody manufacture. J.Chromatogr. B, 848:8 � 18, 2007.

[4] C. Cristiani. Microbiologia Industriale. 2016.

[5] F. Assobiotec and Enea. Le imprese di biotecnologie in Italia, Facts and Figures.2016.

[6] Assobiotec. La bioeconomia in Europa. 2017.

[7] S. Miller. The $250 billion potential of biosimilars. Industry Updates 2013. 2016.

[8] S. R. Hill. Cost-e�ectiveness analysis for clinicians. BCM Med, 10(10):1 � 3, 2012.

[9] A. Sarpatwari, J. Avorn, and A. S. Kesselheim. Progress and hurdles for follow-onbiologics. N Engl J Med, 25(372):2380 � 2382, 2015.

[10] D. H. Howard, P. Bach, E. Berndt, and R. M. Conti. Pricing in the market foranticancer drugs. J Econ Perspect, 1(29):139 � 162, 2015.

[11] E. R. Berndt, D. Nass, M. Kleinrock, and M. Aitken. Decline in economic returnsfrom new drugs raises questions about sustaining innovations. Health A�, 2(34):245� 252, 2015.

[12] A. Rathore. Follow-on protein products: Scienti�c issues, development and chal-lenges. Trends Biotechnol., 27:698 � 705, 2009.

[13] U. Gottschalk. The need for innovation in biomanufacturing. Nat. Biotechnol., 30:489� 492, 2012.

[14] A. Toumi, C. Jurgens, C. Jungo, B. A. Maier, V. Papavasileiou, and D. P. Petrides.Design and Optimization of a Large Scale Biopharmaceutical Facility Using ProcessSimulation and Scheduling Tools. Pharmaceutical Engineering, 30(2), 2010.

121

BIBLIOGRAPHY

[15] G. V. Research. Monoclonal Antibodies (mAbs) Market Analysis. 2016.

[16] Rentschler. Stainless steel or Plastic. BPI West, 2017.

[17] N. M. G. Oosterhuis and H. J. van den Berg. How Multipurpose is a DisposableBioreactor? BioPharm International, 24(3), 2011.

[18] A. Ravisé, E. Cameau, G. De Abreu, and G. Pralong. Hybrid and disposable facilitiesfor manufacturing of biopharmaceuticals: Pros and cons. Springer: Advances inBiochemical Engineering / Biotechnology, pages 185 � 219, 2009.

[19] J. M. P. Chanfrau, K. Zorrilla, and E. Chico. The Impact of Disposables on ProjectEconomics in a New Antibody Plant: A Case Study. Biopharma International,22(12), 2009.

[20] S. S. Farid, J. Washbrook, and N. J. Titchener-Hooker. Decision-Support Tool forAssessing Biomanufacturing Strategies under Uncertainty: Stainless Steel versus Dis-posable Equipment for Clinical Trial Material Preparation. Biotechnology Progress,21(2):486 � 497, 2005.

[21] E. Langer. 9Th Annual Report and Survey of Biopharmaceutical ManufacturingCapacity and Production. A Study of Biotherapeutic Developers and Contract Man-ufacturing Organizations. BioPlan Associates, 2012.

[22] Fuji�lm Diosynth Biotechnologies. Process Design for an All Single-Use Manufactur-ing Facility: Scaling Low to High Titer Processes to Fit Standard mAb Equipment.BioProcess International West, 2017.

[23] U. Gottschalk and A. A. Shukla. Single-use disposable technologies for biopharma-ceutical manufacturing. Trends Biotechnol., 31(3):147 � 154, 2013.

[24] M. Hammond, L. Marghitoiu, H. Lee, L. Perez, G. Rogers, Y. Nashed-Samuel, andal. A cytotoxic leachable compound from single-use bioprocess equipment that causespoor cell growth performance. Biotechnol. Prog., 30(2):332 � 337, 2014.

[25] R. Colton. Recommendations for extractables and leachables testing. Bioprocess Int,5(11), 2008.

[26] K. Clapp. The Great Debate: Stainless Steel Versus Single Use. 2016.

[27] G. Altaras. Quantitation of interaction of lipids with polymer surfaces in cell culture.Biotechnol. Bioeng., 96:999 � 1007, 2007.

[28] J. Okonkowski. Cholesterol delivery to NS0 cells: Challenges and solutions in dis-posable linear low-density polyethylene-based bioreactors. Biosci. Bioeng, 103:50 �59, 2007.

[29] A. Sinclair, L. Leveen, M. Monge, J. Lim, and S. Cox. The Environmental Impactof Disposable Technologies. The BioPharm International Guide, pages 4�15, 2008.

122

BIBLIOGRAPHY

[30] M. Pietrzykowski, W. Flanagan, V. Pizzi, A. Brown, A. Sinclair, and M. Monge.An environmental life cycle assessment comparison of single-use and conventionalprocess technology for the production of monoclonal antibodies. Journal of CleanerProduction, 41:150�162, 2013.

[31] A. Desgeorges, H. Kornmann, and E. Cameau. Evaluation of a Single-Use Bioreactorfor the Fed-Batch Production of a Monoclonal Antibody. BioPharm International,Supplements, 2010.

[32] R. Porter and T. Roberts. Energy savings by wastes recycling. Elsevier, 1985.

[33] T. Scheper. Advances in Biochemical Engineering/Biotechnology. Springer, 2009.

[34] Disposals Subcommittee of the Bio-Process Systems Alliance. Guide to Disposal ofSingle-Use Bioprocess Systems. BioProcess International, pages 22 � 28, 2007.

[35] H. Zhang, W. Wang, C. Quan, and S. Fan. Engineering Considerations for ProcessDevelopment in Mammalian Cell Cultivation. Current Pharmaceutical Biotechnology,11:103 � 112, 2010.

[36] M. Pietrzykowski, W. Flanagan, V. Pizzi, A. Brown, A. Sinclair, and M. Monge.An Environmental Life Cycle Assessment Comparing Single-Use and ConventionalProcess Technology. BioPharm International, 2011 - Supplement(8), 2011.

[37] W. Flanagan, M. Pietrzykowski, V. Pizzi, A. Brown, A. Sinclair, and M. Monge.An Environmental Lifecycle Assessment of Single- Use and Conventional ProcessTechnology: Comprehensive Environmental Impacts. BioPharm International, 27(3),2014.

[38] J. Wilson. A Fully Disposable Monoclonal Antibody Manufacturing Train. BioPro-cess International, Supplement:34 � 36, 2006.

[39] E. Langer and R. Rader. Single-use technologies in biopharmaceutical manufacturing:A 10-year review of trends and the future. Eng Life Sci, 14(3):238 � 243, 2014.

[40] H. L. Levine, J. Lilja, R. Stock, A. Gaasvik, H. Hummel, T. C. Ransoho�, and S. D.Jones. Single-Use Technology and Modular Construction. BioProcess International,2013.

[41] A. Goldstein and O. Molina. Implementation strategies and challenges: Single usetechnologies. PepTalk Presentation, 2016.

[42] A. Xenopoulos. A new, integrated, continuous puri�cation process template for mon-oclonal antibodies: Process modeling and cost of goods studies. Biotechnol, 213:42 �53, 2015.

[43] Biopharm Services. Mab manufacturing today and tomorrow. 2014.

[44] R. Eibl and D. Eibl. Single-use technology in biopharmaceutical manufacture. JohnWiley & Sons., 2011.

123

BIBLIOGRAPHY

[45] General Electric Lifesciences. Process Economy for Vaccines. Life Sciences, 2017.

[46] General Electric Lifesciences. Process economy and production capacity using single-use versus stainless steel fermentation equipment. Life Sciences, 2015.

[47] M. Mauter. Environmental Life-Cycle Assessment of Disposable Bioreactors. Bio-Process Int, 8:18 � 28, 2009.

[48] B. Rawlings and H. Pora. Environmental Impact of Single-Use and Reusable Biopro-cess Systems. BioProcess Int, 7:18 � 26, 2009.

[49] G. Hodge. Disposable Components Enable a New Approach to BiopharmaceuticalManufacturing. Biopharm Int, 17:38 � 49, 2004.

[50] G. Hodge. The Economic and Strategic Value of Flexible Manufacturing Capacity.2009.

[51] J. Fromison. Disposables in Clinical Manufacturing. Am. Pharmaceut. Rev., 12:20 �27, 2009.

[52] A. Foulon. Using Disposables in an Antibody Production Process: A Cost-E�ectiveness Study of Technology Transfer Between Two Sites. BioProcess Int, 6:12� 17, 2008.

[53] F. Malcolm. Implementing Disposable Chromatography. Process and TechnologyFit., 2013.

[54] A. Sinclair and M. Monge. Quantitative Economic Evaluation of Single Use Dispos-ables in Bioprocessing. Biopharm Services, 6:1 � 11, 2016.

[55] D. Petrides. Bioprocess Design. Bioseparations Textbook, page 60, 2000.

[56] V. Papavasileiou, C. Siletti, and D. Petrides. Systematic Evaluation of Single-UseSystems Using Process Simulation Tools � A Case Study Involving MAb Production.Biopharm International, 2008.

[57] D. Petrides, D. Charmichael, and C. Siletti. Batch Process Simulation. Design ofBatch Processes, page 30, 2013.

[58] T. P. Obrien, L. A. Brown, D. G. Battersby, A. S. Rudolph, and L. P. Raman.Large-Scale Single-Use Depth Filtration Systems. BioProcess International, 2012.

[59] S. R. Schmidt, S. Wieschalka, and R. Wagner. Single-Use Depth Filters: Applicationin Clarifying Industrial Cell Cultures. BioProcess International, 2017.

[60] S. Moioli. Process Plants. 2017.

[61] M. A. King. Selection Criteria for WFI Production Equipment. CEMAG, 2005.

[62] Università degli Studi di Padova. Materials and the Environment. Elsevier, 2013.

[63] Environmental Protection Agency. Emission Factors for Greenhouse Gas Inventories.2014.

124

Trovo difficile racchiudere tutti i ringraziamenti, espressi e non, che ho vissuto inquesti ormai quasi sei anni. Gioie e dolori, soddisfazioni e difficolta, tutte incastonate neirapporti con le persone con cui ho vissuto un periodo che, certamente, lascia una tracciaindelebile nella mia vita. Il piu sara sicuramente dimenticato, se gia non lo e, eppure,come si suol dire in questi casi, ogni esperienza passata definisce il nostro essere futuro.Percio eccomi, in un soleggiato pomeriggio di primavera appena sbocciata, che immaginodi essere su un sentiero di montagna, giunto ad un punto panoramico, e di voltarmiindietro, con lo sguardo che dal percorso appena trascorso spazia verso l’orizzonte, colsorriso sul volto.

Grazie a Jacobs All’ing. Simona Cucco, che ha reso possibile questa esperienza.All’ing. Marco Magni, che ha pensato a questo progetto ed ha donato generosa disponibi-lita, nonostante la mole di lavoro. All’ing. Angela D’Urso, la cui cordialita ha fornito unsupporto, in termini di tempo e conoscenza, essenziale. A tutti i dipendenti che, volentio nolenti, hanno condiviso con me questi ultimi mesi di carriera universitaria.

Grazie al Politecnico di Milano Al prof. Renato Rota, che ha accettato la re-sponsabilita del progetto con professionale attenzione. Al prof. Francesco Maestri, concui ho maggiormente mantenuto contatto e che ha dimostrato accortezza e disponibilita.Ai compagni di corso, miei amici. Un pensiero a quelli con una marcia in piu, che daqua gia sono passati, a quelli a cui faccio i migliori auguri per arrivarci al piu presto ea quelli che sono al mio fianco. Senza di loro questo traguardo sarebbe stato impossibile,e, ne sono certo, non sarebbe stato altrettanto bello. Compagni di corso che, mi auguro,rimarranno compagni di vita.

Grazie alla mia famiglia Ai miei genitori, che ammiro da tempo per i loro valorie virtu e per come assieme si completino: spero siate orgogliosi. A papa, che, piu o menoconsciamente, e stato stimolo ed esempio per il mio cammino; la sua dedizione al lavoro eincoraggiante, ispiratrice, che infonde speranza per il futuro. A mamma, motore silente ditutta la famiglia, sicuramente e la persona a cui sono stato, sono e saro piu riconoscente.Ad Andrea, quella testa calda di mio fratello; gli auguro di poter raggiungere traguardiambiziosi e che possa trovare pace e serenita.Ai Dubini, Maurizio, Antonella, Matteo, Marco, Anna. Agli Agazzi, Gianluca, Barbara,Marco, Elisa. A nonna Livia e zia Marisa, per tutti. Voi siete le mie radici, la mia casa.Spero questa sia un’occasione di gioia e commozione come fosse anche vostra.

Grazie ai miei amici Agli amici di ieri, di oggi, di domani. A Gabriele. A Simone.Agli amici di Lazzate. Con voi, ho condiviso emozioni, ho disegnato sogni, ho vissuto cioche mi ha condotto a quello che sono.

Grazie a te che hai cercato questa pagina in cerca di un encomio speciale, ma, permia mancanza, non sei soddisfatto. In questa ricorrenza, ti abbraccio con tutto il cuoree ti auguro di potermi empatizzare.

politecnico di milano · 2018-06-06 · biotechnology biotechnology is the application of scienti c...

Documents