Carlos Martinez, Associate Director Manufacturing Sciences & Technology, Sarepta Therapeutics
Carlos Martinez introduced the audience to Sarepta Therapeutics, a global biotechnology company with an objective to engineer precision genetic medicines against rare diseases.
He dove into his talk, and underlined the importance of automated systems and connectivity of data, an area in which the peptide industry has been lagging behind. For example, he argued that paper batch records should be replaced by electronic batch records to improve review and and processing by other systems. Then he explained the evolution of industry 4.0 and the remarkable progression in industry over the last 200-years, which can be classified into four eras. Industry 1.0 is includes mechanization and is associated with the use of steam power, and saw a three-fold increase in terms of production. Industry 2.0 is linked to mass production and the use of electricity in manufacturing. Industry 3.0 involves electronics, automation, and IT, and is notably influenced by the development of the computers. He said that we are in the middle of Industry 4.0, based on notions like smart factory, data modeling, and machine learning. Industry 4.0 is at the convergence of multiple technologies such as robotics, cloud computing, virtual reality, artificial intelligence and so on.
He shared his opinions on strategic enablers to Industry 4.0 underlining four key areas that are needed to be successful. Upper management support is essential since there will be a need for consultancy at several steps. Creation and maintenance of a culture of innovation is another important aspect. With new technologies people with higher skill sets and broader knowledge are required to establish the know-how. A road map is necessary because the implementation can take a long time. He also underlined the importance of the patience because it probably will not be a short project.
He also mentioned major benefits of Industry 4.0 by listing sensor based process control, flexible production lines, increased efficiency and resiliency, and improved employee experience and safety. All of these contribute a safe, effective, and robust manufacturing process. For instance, the ability to predict a failure will not only increase efficiency, but also might save a batch worth a million dollars.
Sarepta started with small steps by bringing the data out of paper batch. This digitalized data became suitable to implement programs enabling the monitoring of the processes. Digitalization also provided transparency across the whole network. Involving others such as quality analysis (QA), quality control (QC), and supply chain allows them to gain momentum. Following that automation, collecting information and connecting everything to a network are essential.
Martinez made some suggestions on automation implementations. He underscored the importance of ISA standards since it is easy and feasible to follow a structure that has been understood and recognized across the industry. Leveraging Bender Libraries such as Rockwell is recommended. Implementation of batch solutions like MES allows someone to query the data by batch number and provides details such as when manufacturing operations occured.
He also shared the pipeline demonstrating the progress at Sarepta. In the first phase, manual data entry system were optimized and a simple transcription of batch record into a database was completed. In the second phase they used Power BI to implement their manufacturing intelligence system. For example, when the yield is low, it is possible to ask Power BI to check the data and give the indicators, influential variables, and so on. They also involved QA and QC operations to the Web interface to collect all kinds of data such as errors in batch records, time that it takes to do QC analysis, identification of bottlenecks and so on. Their power BI dashboard called Theia System is published across the entire organization ensuring visibility, transparency, and ease of communication. Then they have implemented a central repository system, automated their operations, and made some statistical data sets which is going to be followed by modeling studies, which will result in implementation of the predictive models able to predict yield or potential failures.
