EPCAM: A Strategy to Enable Manufacturing-Process Control Transformation

The healthcare industry lags significantly behind the electronic, chemical, and automotive industries in process quality. Process sigma in healthcare is around 2.0 (achieving success about 85% of the time), while other industries' process sigma is as high as 6 or 7 (achieving success more than 99.9999% of the time). Healthcare companies have relied on inspecting quality into the products at enormous cost, yet defective product still finds its way into the marketplace too frequently. Because newer devices and pharmaceuticals are becoming more complex, it is a mathematical certainty that 2-sigma manufacturing processes will fail.

AUTHORS

The healthcare industry argues that process-improvement efforts are hampered by regulations and that they must spend as much as 50% of each improvement program on validation. These claims are true, but new regulations allow manufacturing processes to be validated according to a science-based mechanistic understanding of process control. Once this validation is accomplished, the requirement for ongoing validation efforts is greatly reduced. Ongoing process-improvement efforts can proceed with minimal revalidation.

The US Food and Drug Administration is actively encouraging manufacturers to adopt process analytical technology (PAT) and quality by design (QbD). Both strategies require a process-control management method and a technical infrastructure for ongoing operations.

Enterprise process control and management (EPCAM) is a strategy for achieving this increase in process understanding. EPCAM is a macromanagement process executed through an integrated information solution and supported by associated systems and data, including machine-process controls through programmable logic controls (PLCs), supervisory control and data-acquisition systems (SCADA), manufacturing execution systems (MES), middleware, statistical process control (SPC), laboratory information management systems (LIMS), adaptive process control (APC), and fault detection and classification (FDC). EPCAM uses Six Sigma as the fundamental macro management process to supervise manufacturing-process control across the entire value chain for any given element, including raw materials, suppliers, and subcontractors.

EPCAM has largely been overlooked for three major reasons. First, process control is typically executed manually and is primarily focused on stand-alone unit operations. Second, data integration across operations is lacking. Third, the software to draw the relationships between process parameters and product attributes is not widely known or commercially available. Some examples of these issues include the following:

• Large automated equipment trains have their own control systems

• Manufacturing is typically split into functional areas such as raw-material preparation, mixing, compression, machining, coating, and sterilizing, with control autonomy in each area

• Control remains focused on an operation-by-operation basis versus an end-to-end, product end-state outcome view

• Manufacturing uses specialized control systems that addresses specific needs

• Process knowledge is kept by the worker and may not be documented or shared among peers

• Product test data (percent active, color, tablet hardness, and weight) reside in systems such as LIMS, while process data may be held in an MES or recorded manually.

The requirements of manufacturing in the regulated life-sciences and healthcare environments often necessitate that manufacturing-process improvement take second priority to validation and responding to nonconformances. As a result, process control in many manufacturing operations is not well understood.

Many companies turn to Six Sigma as the preferred method for process improvement, then apply it problem by problem. Six Sigma is well suited for improving business processes following the methodology of define, measure, analyze, improve, and control (DMAIC). The problem with most DMAIC efforts is the amount of time and resources needed to collect the relevant process data and analyze them with the corresponding product data.

For example, a plant site of about 350 people adopted the Six Sigma process hierarchy with the intention of continually improving the Sigma of the critical processes and the overall process. The idea was sound, but significant practical problems derailed the implementation. The first problem was that the effort required about 30 full-time employees to track and maintain the data. The second problem was uncertainty about what to do with 10,000 SPC charts after they had been gathered.

EPCAM solves this process-execution and data-management problem by:

• Establishing DMAIC as the macro process-control management method across the enterprise

• Integrating process and product data through DMAIC. The data are collected to enable the correlation of process parameters to product attributes. This technique eliminates the considerable manual effort required by a typical DMAIC project of data collection and analysis.

• Providing the adaptive process-control applications that analyze product and process data, correlate the data, and optimize performance. This step provides a mechanistic process understanding.

• Providing the automation to run operations based upon the above process understanding in real time.

• Identifying process variations early and determining the optimum parameters throughout an entire process chain, as opposed to focusing on a single operation.

Thus, EPCAM allows process control to be executed across the enterprise in an efficient, reliable, and effective manner.

EPCAM can be achieved in a controlled, phased approach. There are five logical levels or phases in moving from the typical environment to the EPCAM end state.

The fifth and lowest level is the automation of the batch record (i.e., electronic batch records). The fourth level is MES. The third level is point-to-point process control using PAT. The second level is adaptive process control (APC), which links process control across multiple operations. The first and highest level is FDC, which enables control of processes and the environment. These levels will be explained in more detail in the "EPCAM solution and value" section. Achieving EPCAM can seem like a daunting journey, but the process can be completed in steps, building the organizational capability and trust required along the way.

The typical manufacturing environment in a life-sciences or healthcare company

Figure 1 depicts the current and desired states of most manufacturers in the life-science and healthcare industries.

Traditionally, supply chains in the pharmaceutical industry have not been expected to be efficient. They have merely been expected not to cause problems. In general, the industry's margins have been sufficiently high, and lost sales received more attention than inventory control. Because the industry is regulated, making a significant process change often requires revalidation. Compliance and responding to nonconformances have always been the first priorities from a process perspective.

As a result, process quality today in healthcare is about 2 sigma, and process capability is around 1. Many processes are less than 1 sigma, and the cost to serve is relatively high. EPCAM provides the opportunity to improve process quality, process capability, and cost to serve. EPCAM goals for the pharmaceutical industry are levels of at least 4 sigma for the manufacturing processes and process-capability (CpK) of critical operations of 2.5 sigma.

It would be disastrous for product of 2-sigma quality to be released to patients. Recalls would be extensive, adverse events would abound, the industry would lose credibility, and patients would be at increased risk. For this reason, the industry has instituted robust inspection processes that raise product quality to approximately 6 sigma (see Figure 1). Most companies continue to invest in inspection to improve quality even more (see the short red arrow in Figure 1).

Figure 1: The business case for enterprise process control and management (EPCAM). (AUTHORS)

The authors believe, however, that the right course is to invest in improving process sigma and process capability (see the long red arrow in Figure 1) while keeping the inspection capability constant. Thus, product quality for the patient would increase to 7 sigma and attain the benefits of lowered cost and reliable supply. The volume under the curve in Figure 1 can be used as a relative indicator of the cost of quality.

Many developments in the industry indicate that this is the proper course:

• Products are becoming more complex. Supply chains that remain at a low process sigma and low process capability would result in high yield losses, unreliable supply, and increased likelihood that out-of-specification product reaches the patient. Product complexity comes in the form of biologic drugs, drug–device combinations, and time-released products.

• FDA is encouraging improved process control and acknowledging that some of its regulations have been obstacles to this goal. The agency is encouraging manufacturers to adopt PAT and decreasing approval times for new products filed with this type of process control.

• Mapping the human genome means that the ability to determine drug efficacy based upon genetic makeup will decrease the number of blockbuster drugs. Industry supply chains will need to make more products at lower volumes. Process capabilities must improve for companies to efficiently produce this mix of products.

• Because products' complexity is increasing and their side effects are growing more acute, the industry and FDA are becoming increasingly concerned about spikes in dosing. These concerns are inspiring the development of active delivery mechanisms that increase the demands on the supply chain's process capability.

Improving process capability significantly can be a daunting task, but current technologies make this project more feasible and faster to complete. In the past, regulators' conservative approach discouraged innovation, but today they are willing to assist and encourage transformation. This support is based on the belief that increased scientific understanding and applied technologies reduce variation and risk.

EPCAM integrates information

EPCAM integrates information through a combination of open applications, integration software, and adaptive process-control software. Integrated information enables the macro management process to improve manufacturing processes. The following results occur when EPCAM is fully implemented in both manufacturing and product-development environments:

• CpK improves from the typical level of 0.7–1 to > 2

• Process sigma improves from roughly 2 to about 4.5

• Costs of compliance and nonconformances decrease sharply, and the possibility of recalls is greatly minimized

• The reliability and speed of new product launches are greatly improved

• Manufacturing throughput is increased.

Many EPCAM software components support open standards and are not proprietary. EPCAM works with most installations of commercially available software (MES, ERP, and LIMS). Two proprietary applications, APC and FDC, will be discussed in more detail. Companies want to achieve the level of quality that EPCAM process control failitates, but they are uncertain about how to make the transition from the current environment to the targeted EPCAM end state. To make this change, the authors have developed a phased approach that progresses through five logical levels of control (see Table I).

Table I: Process-control hierarchy.

The achievement of each level improves quality and control and leads to business benefits such as greater yields and better products. In addition, organizational capability (e.g., process skills, experimentation skills, the number of experienced people) grows with the attainment of each level. When companies reach levels 4 and 5, their organizational capability and process-control skills must keep pace with their technological advancement. It is absolutely critical to build the organization's process-control expertise, discipline, and the culture of control in both the manufacturing, product-development, and information-technology organizations as progress is made.

Table II: Transformation demands from the current state to the desired state using EPCAM.

The five levels of control

Level 5 is the level of electronic batch records. At this level, the management of manufacturing control is improved by typically linking it to ERP and LIMS electronically. This link guards against mistakes in manual transcription. Conformance to standard operating procedures (SOPs) is increased. Better conformance reduces paperwork errors, which, in turn, ensures that correct manufacturing sequences are followed and reduces manufacturing mistakes. Paperwork is also reduced at this level, and nonconformances resulting from paperwork are significantly reduced. Process capability, however, typically does not improve, and process sigma improves only slightly. This level is a good stage at which to introduce technology to the shop floor, gain experience within operations, and reconcile and improve data.

MES characterizes Level 4. This level provides automated machine instructions that establish the process parameters in which the equipment operates. MES typically links ERP directly to the shop floor. This link reduces operator and transcription errors. Processes are standardized, and operating conditions are measured and recorded. SCADA regulatory requirements are satisfied. Messages automatically alert operators to out-of-specification operating conditions. Quality and compliance are increased by additional standardization and mistake-proofing. At this level, process capability may increase slightly if the process parameters correlated to product attributes are known. Even if this is not the immediate result, the standardization of MES provides the controlled environment that enables future process improvement.

Level 3 is marked by point-to-point process control. This level of automated process control allows users to modify product attributes by adjusting process parameters. For example, the amount of active/in.² in a layer (patch) to be applied to a substrate may be controlled by adjusting the line speed of the coating operation. The thickness of the layer is measured after coating, and the line speed may be increased or decreased to achieve the targeted thickness. This control is an example of PAT that FDA has been encouraging during the past several years. At this level, process capability is increased for the selected operation raising it to more than 1.0 sigma. This measurement means that, most of the time, the product will be made within specification but will vary within the specifications. Quality and compliance increase, thus providing significant business results such as improved yields, reduced waste, increased capacity, and fewer errors.

Level 2 is APC. APC is the application of process control to the entire manufacturing process. The manufacturing process can be within one plant or shared among multiple plants. The process can include suppliers and contract manufacturers. APC allows operators to adjust a process based upon known data from previous operations, the condition of raw materials to be used in the process, or the known condition of the equipment tool. As in Level 3, where the thickness of the active coating on a patch is controlled by line speed, APC achieves the exact concentration of active using data about the concentration of the active in the coating batch. APC thus attains the correct thickness level based on batch concentration. APC automatically adjusts from batch to batch.

In addition, the uniformity of thickness across the patch surface is important, and the uniformity of the coating is based on the batch's viscosity, surface tension, and temperature. APC automatically adjusts machine parameters such as coating-bar vacuum, web tension, and width between coating rollers to obtain the precise uniform coating based on the batch characteristics.

A key requirement of Level 4 is the integration of data for APC. This integration should be planned ahead of time through network design. The data elements that are related through various algorithms include metrology data (which are related to products attributes such as thickness, active concentration, and color), machine parameters (e.g., current, vacuum, and motor temperature), process parameters generated by machine parameters (e.g., solution temperature, pressure, line speed, machining speeds, and feeds), logistics data (which identify the material batch and location), event data (such as material consumed and steps completed within an equipment unit), and other local data from the tool bus such as sensor data.

APC uses various algorithms to determine the relationship between the product attributes and the machine and process variables. Based on these relationships, APC controls multiple process steps to achieve the desired product attributes. APC operates in feed-forward and feedback modes as required. Analysis is performed from unit-operations and multivariate perspectives. Because of its multivariate capability, applying APC improves the process performance to more than 2.0 CpK.

Process capability increases to more than 2.0 with this type of control. Process capability can rise to the point where nonconformances become an annual event rather than a daily occurrence. This control is called adaptive because process and product data are fed forward to the process-control management system. APC instructs controllers so that the current processing step is adapted according to the known condition of the materials. This flexibility further reduces variability within specifications and increases process capability.

In addition to quality and compliance, benefits include reduced product variability, which ensures safety and efficacy in drugs and devices. FDA is increasingly concerned about the need for constant active dosing levels in patients. This level of process control will reduce product variability to help ensure constant dosing.

Level 5, FDC, adds the remaining elements to the process-management picture. It brings control over the environment. It measures and controls elements that are largely constant but drift over time or vary because of events such as normal wear, abnormal wear, maintenance, or software upgrades. The following situations describe examples of what FDC controls:

Example I. A manufacturer lost control of its chemical sterilization process. Process parameters on SOPs were being followed. Bacteria were found on product tests as many as four days later. Several clean-up and sterilization procedures were conducted. For six weeks, no cause or solution could be found. A large percentage of the manufacturer's product underwent sterilization, which caused significant supply-chain delivery problems. Employees ultimately found that a different type of oil was being used in one of the vacuum pumps. When the oil was replaced with the normal oil, the problem was eliminated.

The residual gas analyzer that provides environmental control could have detected the change in air composition before the operation started a second sterilization batch. In this case, only one batch would have been affected. FDC's classification capability would also have instructed the operator or mechanics where to look for the problem.

Example II. A manufacturer lost several batches of a sold-out product because of low active-release rates. During manufacturing, the product was a capsule with a hole that allowed the active to enter the patient's body over a period of time. The capsule was held in place by a spring-loaded clamp and laser drilled. During routine maintenance, the standard springs were not available, so stronger springs were used. The stronger springs increased the clamping pressure and forced the active out of the capsule during the drilling operation. This deviation caused the low release rates. FDC could have detected the increased clamping pressure and alerted the operator and the mechanic of the problem before the first batch was run, thereby saving several batches of the sold-out product.

FDC detects machine failure, machine drift, and environmental changes long before procedures or routine checks would identify such conditions. FDC detects failures through monitoring machine parameters and analyzing them in relation to each other. For example, suppose a machine running in a partial-vacuum condition demonstrates pressure within specification, but the vacuum pump is pulling more electrical current than it normally does. Suppose the pressure varies within the chamber more than normal. FDC would detect this anomaly and send a message to the operator to check the seal or the pressure plate. This action would occur well before an out-of-specification situation could occur.

FDC enables excellent process control and anticipates problems before they occur. FDC allows the manufacturer to achieve the perfect or "golden model," the conditions that produce the product with the right attributes all the time with minimal variation. The conditions include the processing parameters such as temperature, pH, conductivity, pressures, speeds, and product attributes from the previous operation. The conditions also include environmental factors such as the manufacturing air and machine statistics such amp usage, pressures, and speeds. The process, environmental, and machine conditions are recordeded to achieve the golden model. The golden model can then be attained every day with FDC and APC.

FDC brings process control to higher than CpK 2.5, a level at which operators can adjust or shut down processes well before problems or nonconformances occur. If a problem is likely, FDC directs the operator to the cause to correct it. The effect of this achievement on a life-sciences or healthcare company is great. The cost of compliance plummets. Technical resources are directed to continuing process improvements, which no longer require revalidation if a QbD strategy is followed with the FDA, or product-development activities instead of addressing nonconformances. FDA audits can be forestalled by providing regulators access to process systems and results. Additional benefits include a more frequent on-time product launches (see Figure 2).

Figure 2: The deployment of process analytical technology helps manufacturers achieve adaptive control and automatic fault correction. MES is manufacturing execution system, WIP is wash in place, and PAT is process analytical technology. (AUTHORS)

The business-process model and closed-loop systems architecture for EPCAM combine the process orientation of Six Sigma and advanced manufacturing technologies. EPCAM was developed by a joint task force comprising IBM's Research and Development, Software Development, Integrated Supply Chain, Engineering and Technical Services, and Global Business Services experts. The conceptual architecture shown in Figure 2 delivers 8.0-sigma process capability at IBM's semiconductor fabrication facility in East Fishkill, New York.

Commercially available ERP and MES software systems are foundational layers of EPCAM needed to move from 2-sigma performance up the process-control hierarchy. Middleware moves data between the processing centers and the APC and FDC systems.

EPCAM can be built incrementally, focusing on the most critical areas of the operation, based on the individual business's priority. The end state should be envisioned before the process begins to ensure that the integration will be achieved and the business value will be realized.

Organization

The transformation required to implement EPCAM is a cultural change enabled by advanced tools that reduce risk and shorten the process. The process includes organizational challenges such as:

• Skills development

• Changes in job roles and organizational structures

• Changes in mindset and culture

• Modification of policies and procedures

• Data sharing

• Establishing the primacy of science.

The EPCAM architecture may integrate existing stand-alone tools, or employ new technologies that require extensive training and support. Advanced technologies often are already in place, and the organizational comfort and technology maturity are sometimes high. Even when an organization is technologically advanced, individuals learn at different rates. Supervisers must consider this variable when they design and execute the learning and knowledge-transfer process.

New tools, increased levels of automation and integration, and process changes, will lead to changes in job roles. A process orientation, focused on the outcome of an integrated series of activities, is one of the most significant benefits and one of the largest role-based changes of the EPCAM strategy. Tasks, measurements, and reinforcements should be realigned to maintain a sustained focus on the entire process.

Once the mindset change from an individual operation focus to an end-to-end orientation is accomplished, other elements of the operating culture must be addressed. Fear commonly accompanies the deployment of technologies. Individuals experience anxiety when they are asked to maintain their performance while learning new methods and tools. Personnel may become uncertain when asked for the first time to focus on the science and chemistry of the operation rather than on managing the equipment. Reducing reliance on intuition (often perceived as skill developed over time) and increasing scientific understanding may be difficult for some employees. Abandoning established approaches is challenging, particularly for employees who have maintained high performance using those methods.

Perhaps the most critical cultural change is trusting the data, process, policies, and procedures to produce the batch, and trusting the technologies to integrate and capture the information. Without this trust, the benefits of EPCAM are quickly lost. One of IBM's clients referred to the skills-development process as "upskilling," and employed a rigorous change-management program to create adequate awareness, obtain acceptance, and provide training and support.

Figure 3 shows a transformation "road map" featuring one progression from the current state to the EPCAM capability in a logical and feasible timeframe. The figure depicts when the effort should begin and when EPCAM capability should be achieved to meet the future needs of the supply chain.

Figure 3: A road map for progressing from the current state to enterprise process control and management capability. APC is adaptive process controls, ERP is enterprise resource planning, FDC is fault detection and classification, LIMS is laboratory information management system, MES is manufacturing execution system, PAT is process analytical technology, PLM is product life cycle management, RFID is radio-frequency identification, SC is supply chain, and TTS is target treatment solutions. (AUTHORS)

The process for implementing EPCAM

Organizational capabilities must be built along with tools and technology. Technology is great enabler of process understanding but it requires a management team with vision and a commitment to success. The management team must include dedicated people with process skill to use tools and improve process capability. Many people employed today in the life-science industry have math and science backgrounds. Many of these employees can be trained to focus on process improvement.

The first step in establishing EPCAM is a foundational layer of process-management training that includes lean and Six Sigma, implements ERP, and sets the stage for a strategic change.

The second step is attaining point-to-point process control, deploying radio-frequency identification internally to track products, and implementing electronic batch records. Experience with process control accumulates at this stage. The organization should undergo training to foster the right skills and behaviors.

The third step is the integration of the existing ERP, LIMS, MES, PLM, PLCs, sensors, and actuators to the APC system. Integration puts the foundational data in place and facilitates the change from point-to-point control to end-to-end process management. The company should set a target for the percent of end-to-end processes to be within APC control. A good initial goal is 30–50%. The new process-improvement organization should regularly perform advanced analysis using the data and APC capabilities to continually improve process capability. All new products in development should use the APC capabilities.

The next step is implementing FDC and continuing to expand the end-to-end process coverage within APC. FDC captures the environmental features that affect the product and brings stability and standardization at the highest level of process control. Expanding control to trading partners such as raw-material suppliers and subcontractors should be part of this step. Product development should be fully integrated with the manufacturing systems at this step as well.

An incremental progression is possible. The project is a large undertaking that will take more than five years, but each step creates great business value. The quality of the product and the process is enhanced at each step. Risk is lowered for the manufacturer, and safety is increased for the patient.

EPCAM and the extended enterprise

Once established internally as an operating model, EPCAM can be extended to strategic business partners as well. IBM has pioneered a concept called the "Control Tower" to monitor, integrate, and manage contract-manufacturing operations, thus crossing internal operational boundaries. The rationale is that an operational network is more effective than a vertically integrated business model. IBM significantly improved its return on net assets and its supply-chain agility by partnering with fewer strategic suppliers. Technology is a key enabler that provides visibility, operational integration, and lower overall supply variability. The next-generation Control Tower establishes joint working teams that address specific product, technical, and business problems, thus improving process control. For example, today IBM's external development partners, IBM Research and Development, and its Manufacturing Management teams are located in the same facility. This organization solves the company's most difficult challenges, including developing new products faster, gaining market share, improving quality, and solving compliance problems.

Summary

The technological advancements, management science, and regulatory authorities' receptivity to innovation that reduces variability, uncertainty, and risk have created a fertile environment for change. The opportunity and need for major process-control changes in the life-sciences and healthcare industries are unprecedented. The current favorable regulatory environment and the emergence of breakthrough technologies that facilitate Six Sigma process control at the manufacturing-enterprise level make this an opportune time for manufacturers to pursue improvement.

The need for improvement is great and will only grow. Products are becoming more complex, many products will likely be produced at low volumes, and various bundling options such as drugs and devices are being considered.

The decision about whether to pursue EPCAM is one of the most important that a company will make. It should not be a quick decision, rather the decision should be made as part of key strategic initiative that the organization, including manufacturing, product development, finance, research, and human resources, accepts. The decision affects all employees and changes the way work is done. Roles and responsibilities change, new skills are developed, and unneeded activities are abandoned.

The process-control transformation can be achieved though the five-level process-control hierarchy model previously described.

EPCAM brings the following advantages:

• Improved patient safety

• Reduced cost of goods sold and cost of quality

• Increased supply reliability even with low volumes and complex products such as biologics

• Timely launch of new products.

The complexity of new products, market demands, and the need for speedy new-product launches are all addressed by EPCAM.

James M. Prendergast* is a global supply-chain leader, Michael Passow is a manager of advanced technology development and adaptive process control, Mike Ricci is a global supply-chain leader, and Srinivas S. Dagalur is a lead application architect at IBM, 525 Lincoln Dr. W., 5 Greentree Center, Third floor, Marlton, NJ 08053-3422, tel. 877.227.9248, james.prendergast@us.ibm.comDennis Bell is an associate partner and Limuel Sagadraca is a managing consultant at IBM Business Consulting Services.

*To whom all correspondence should be addressed.

Submitted: Oct. 18, 2007. Accepted: Jan. 7, 2008.