M&SOM Review

Biopharmaceuticals constitute around 40% of the clinical development pipeline (Walsh 2018). These therapeutics are next-generation drugs produced by biomanufacturing technologies. Biomanufacturing uses live organisms (e.g., bacteria) during the production process, and hence the resulting active ingredients are highly complex compared with conventional drugs. For example, a biopharmaceutical molecule might contain 25,000 atoms whereas an aspirin molecule consists of only 21 atoms. In this setting, we focus on the biomanufacturing of engineered proteins. These proteins are custom-designed and manufactured as part of biopharmaceutical research and development. 

Protein manufacturing operations can be broadly classified into two main steps: upstream fermentation and downstream purification. Upstream fermentation is often carried out in stainless steel vessels. The final outcome of fermentation is a batch mixture that contains the target protein along with several unwanted impurities. After fermentation, the batch proceeds with a series of chromatography operations, as part of downstream purification. The main purpose of downstream purification is to eliminate unwanted impurities. For example, the final batch often needs to be free of impurities if the potential end users are humans. For this purpose, chromatography is used as a critical technique to eliminate unwanted impurities. 

How Much to Waste?

The complexity of protein manufacturing operations leads to unique challenges and trade-offs. For example, the final product should meet certain production requirements that are specified by the end user or application. These production requirements often consist of a yield requirement (i.e., the desired amount of target protein) and a purity requirement (i.e., the minimum acceptable quality). However, meeting both of these requirements could be challenging in common practice because of purity—yield trade-offs. More specifically, the biomanufacturer needs to lose some amount of the target protein at each chromatography step in order to improve the batch purity. As Peter A. Leland, a protein purification expert states, "we are constantly playing a game between purity and yield'' (Aldevron 2016). If too much protein is lost, then the yield requirement might not be achieved. In contrast, if too little protein is lost, then the purity requirement might not be achieved.

Peter A. Leland (one of the co-authors) leads a protein purification team. Photo courtesy of Aldevron Madison.

Although the purity—yield trade-off is inevitable, it is possible to control it by simultaneously optimizing three operating decisions: (1) the amount of target protein produced in upstream fermentation, (2) the choice of chromatography technique, and (3) pooling window (i.e., chromatography operating policy) used in each chromatography step. However, optimizing these three layers of inter-dependent decisions could be challenging for practitioners owing to several factors:

  • Multiple impurities: A batch often consists of multiple different impurity types. Each impurity exhibits a distinct purity—yield trade-off on different chromatography techniques. Therefore, the biomanufacturer needs to carefully exploit the unique purity—yield trade-off associated with each impurity on each available chromatography technique.  
  • Multiple inter-dependent steps: Protein purification consists of multiple chromatography steps in series. Therefore, a suboptimal decision made at an earlier step has a magnifying effect on subsequent steps.  
  • Randomness: Because of the underlying biological and chemical dynamics, the amount of target protein and impurities obtained at a chromatography step involves randomness. 
  • Engineered proteins: Each order represents an engineer-to-order protein, and hence, operating decisions need to be customized for each order. 
  • Starting batch: The condition of the starting material obtained from fermentation is one of the critical factors for success. For example, the starting material could involve "too little" target protein, such that the final purity and yield requirements can never be achieved. In contrast, producing "too much" protein in fermentation might alleviate the purity—yield trade-off but it also leads to higher fermentation operating costs.

These challenges are addressed in a research article (Martagan et al. 2019) recently published in Manufacturing & Service Operations Management. The article builds a stochastic optimization model to help biomanufacturers decide how much to waste. The "waste" in this setting corresponds to the amount of protein produced in excess of the yield requirement, and the amount of protein lost at each chromatography step in order to improve the batch purity. The article presents a finite-horizon Markov decision model, and provides an industry case study to help operations managers understand how going beyond the current practice makes a difference. The case study quantifies the potential gains in the expected profit and operational efficiency through optimizing the three layers of operating decisions. The managerial insights derived from the case study indicate that

  • The percentage improvement (in the expected profit) obtained from simultaneously optimizing the three operating decisions is almost double that which can be achieved by optimizing a single operating decision alone.
  • For low-revenue, low-penalty projects, purification optimization mostly dominates the potential room for improvement in current practice.
  • For high-revenue, high-penalty projects, it can be effective to deliberately increase the upstream production amount to alleviate downstream purity-yield trade-off.  
  • Simultaneous optimization of the three operating decisions can achieve up to 48% higher profit compared to operating policies used in common practice. 

The project has been conducted through three years of close collaboration with Aldevron, a small and medium-sized (SME) biomanufacturer specializing in protein manufacturing. Results obtained from the case study are encouraging to support the competitiveness of SMEs like Aldevron. Nevertheless, the potential impact of this framework extends beyond SMEs. As Tom Foti, the vice president of Aldevron states "we are producing 50-liter cultures here, but our clients [large pharmaceutical companies] are dealing with 5,000-liter cultures. If we can build optimization models here, and demonstrate the feasibility of how it works, our clients could also do that. If they can reduce health care costs, that could directly pass to the patients'' (Aldevron 2016). 

An earlier version of this work received the first prize in the 2016 POMS Applied Research Challenge (click here for a video). Operations Research methodologies have not yet been widely used in the biomanufacturing industry. However, as more biomanufacturers embrace OR methodologies, we believe that it will significantly help the industry reduce drug development costs and lead times. 


 Aldevron (2016) Biomanufacturing with Aldevron, UW-Madison Industrial and Systems Engineering. https://www.youtube.com/watch?v=_haDU6NviOg; accessed August 12, 2019. 

 Martagan, T., A. Krisnamurthy, P. A. Leland (2019) Managing trade-offs in protein manufacturing: How Much to Waste? Manufacturing & Service Operations Management 22(2):330-345, https://pubsonline.informs.org/doi/10.1287/msom.2018.0740.

 Walsh, G. (2018) Biopharmaceutical benchmarks 2018. Nature Biotechnology 36:1136-1145.


Executive committees are a key decision-making mechanism in the vast majority of organizations. They are typically comprised of senior managers, who hold responsibility for the approval or termination of the company’s strategic initiatives. In that light, they play a critical role in any company’s strategy execution. A key strategic dimension often examined in such committees is the innovation output of the company that determines its future competitiveness as captured by its R&D/innovation pipeline of projects. Committees often called ‘innovation councils’, ‘R&D project committees’, or ‘Early Stage Product Committee’ undertake such responsibility to select the right projects that should receive funding support to continue towards completion. By design, participants in such committees span different functions and product lines of the organization.

View Full Post »

The cronut fad started in May of 2013. Soon after Dominique Ansel’s small bakery in SoHo began turning out its hybrid croissant-donuts, the lines to purchase them grew interminably long, sometimes reaching to one hundred sugar-starved sybarites. In August of this year, a similar situation occurred at Popeye’s in Gainesville, Fla., where a line stretched around the corner of the building by 10 a.m., full of people hungry for the restaurant’s famous chicken sandwich.

View Full Post »

With the growing concern for the environment, manufacturing facilities across the world have come under increased regulatory scrutiny. For instance, in the United States, the Environmental Protection Agency (EPA) monitors compliance for 44 programs authorized by statutes, such as the Clean Water Act, Clean Air Act, Toxic Substances Control Act, and so forth. These programs often use facility inspections to confirm compliance with environmental regulation. To illustrate, the EPA conducted over 187,000 facility inspections, from 2009 to 2014, just toward ensuring compliance with the Clean Water Act.

View Full Post »

Governments of developing countries have long been using local content requirements (LCRs) to regulate foreign firms operating within their boundaries. LCRs are provisions, usually under a law or regulation, requiring foreign firms to meet a minimum threshold of goods or services that must be purchased locally.

View Full Post »

Assembly supply chains are ubiquitous in manufacturing. Firms such as Toyota, Apple, and Boeing procure various components from different suppliers to assemble automobiles, smartphones, and airplanes. Such large original equipment manufacturers (OEMs) typically possess dominant market power and can extract most profits out of suppliers.

View Full Post »

Cities around the world are experiencing two socioeconomic transformations: First, the boom of the so-called sharing economy engage people to be more collaborative in sharing access to goods and services and less obsessed with ownership. PWC projects that the market size of the sharing economy will reach $335 billion by 2025. One of the most profoundly impacted sectors is transportation.

View Full Post »

In 2015, it was estimated that the Indian information technology (IT)-enabled services industry generated revenue of $146 billion, up from $105 million in 1989. A compound annual growth rate of over 30% for greater than 25 years is a remarkable industrial success story.

View Full Post »

Many firms provide ancillary services in addition to a main service to enhance the experience of consumers. Consumers need to purchase the main service in order to consume the ancillary service. For example, when taking a flight, consumers may need to transport bags or have a meal during the flight; when staying at a hotel, consumers may have breakfast or need internet connection. Ancillary revenue has become an important source of revenue generation in several industries.

View Full Post »