Magazine Article | January 31, 2013

One Solution For Managing Pharma Supply Chain Risk

Source: Life Science Leader

By Pedram Alaedini, president and CEO of Primapax Group, and Birnur ÖzbaƟ, Ph.D., program director of Laboratory for Port Security at CAIT, Rutgers University.

In most industries, changes to manufacturing processes or delivery modes are usually internal decisions that can be easily and quickly implemented. However, our industry is highly regulated, and any modifications that could potentially affect product safety or efficacy require expensive and lengthy qualifications and validations in addition to meeting rigorous and lengthy regulatory and approval processes.

Risk management is one way to help safeguard the quality and supply of product to customers and ultimately the end user. It helps anticipate dangers and control risk through an ongoing process of risk awareness, reduction or acceptance, and review. It also helps justify improvement and investment where needed and prevents both potential problems for customers and loss of business.

As part of a risk management approach, a simulation model for supply chain management can be utilized where it is too expensive or risky to do live tests. Simulation provides a relatively inexpensive, risk-free way to test changes ranging from a simple revision to an existing production line or redesign of an entire supply chain.

What is simulation modeling?

Simulation is a tool for managing and accelerating change which provides more than an answer: It shows how the answer was derived and allows you to generate explanations for decisions.

A simulation model is a mathematical representation of a system or process that includes key inputs which affect it and the corresponding outputs that are affected by it. For example, a model can calculate the impact of uncertain inputs and decisions one makes on outcomes that are deemed important, such as manufacturing costs, quality matrices, investment returns, and inventory or safety stock levels. A simulation model includes inputs which are changed by the user and a set of relationships between elements of the modeled system in addition to the model outputs that summarize the behavior of the system for different inputs.

Simulation modeling can be utilized in many areas of the life sciences supply chain to measure and improve outcomes. These include reducing manufacturing costs, product portfolio analysis, network modeling, quality and compliance level measurement and improvement, facility and process design, and customer satisfaction levels. It also helps to analyze and identify poorly performing links in the chain by comparing them to best practices.

A Pharma Case study
A midsize pharmaceutical company with one manufacturing facility in the U.S. expects approval of its new drug — a tablet — in about 12 months. Marketing projections for this product are one million tablets per month, equal to 1,000 kg of bulk finished product.

To prepare for the new product launch, the supply chain group decided to take a systematic approach to determine the required systems, personnel, and procedures to ensure successful launch and uninterrupted supply of product to market. As part of the exercise, the team assigned one of its existing production lines in addition to dedicated manufacturing and quality control personnel exclusively to this new product. The team’s overall objectives were:

  • ensure no product backorder for longer than seven days at any time;
  • keep cost of goods and inventories at the lowest possible level without jeopardizing product supply or quality.

To achieve these objectives, a discrete event simulation model is developed to gain an understanding of the supply chain processes. To simplify the model, the team limited the supply chain to only include the manufacturing line, quality control and assurance, warehouse, order processing, and the three main raw material vendors.

The model involves several parameters that allow the user to test a variety of risky or resource-intensive scenarios. These parameters include product demand, production and inventory plans, lead times, analytical testing durations, inventory levels, and production times. As an example, as part of this simulation exercise the validated batch size is changed in the model, and its effect on the entire system/performance measures was observed. In addition, the inventory policy was modified in different ways, and its effect on required production rate and budget were analyzed. Obviously in real life, such modifications would require significant investments and, in many instances, lengthy regulatory approval processes.

Upon completion of the model, the simulation is run 25 times, each representing the same one-year period. Replication of the simulation provides a complete statistical description of the model variables since there are many uncertain parameters in the model. In other words, it represents a good sample of all possible events that may occur. The results are obtained through averaging the results of 25 replications. Thus, the simulation model provides not just an average value, but also 95% confidence interval and minimum and maximum values. 

Supply Chain Backlogs
The model showed that the current supply chain policy would result in an average of 16 backlogs with 12 of them lasting longer than seven days. On the manufacturing side, production was disrupted 39 times on average because existing raw material inventory did not meet the requirements. The average finished product inventory was 450 kg throughout the year with $12,620 in holding costs. Similarly the average raw material inventory was 439 kg, 198 kg, and 69 kg for API, and two different excipients respectively with the total inventory holding cost of $22,375 over a one-year period.

Statistics also showed at least one QC test failed for a total of 140 kg of the finished product, 20 kg of excipient 1 and 196 kg of excipient 2, but never for API over an average one-year period in this particular simulation. During this simulation exercise, the company produced 40 batches of the new product at the cost of $4,814,400 and shipped 48 shipments resulting in $18,135,000 in transfer pricing gains. It is important to note that in this model manufacturing batch sizes and lot shipments are of different size, and the simulation started with 750 kg of initial inventory.

As seen, base case results show that the initial plan would not meet the supply chain objectives. The product backlog is quite high, and the production plan is often disrupted by raw material scarcity. The team assigned to this project decided to determine and present several possible scenarios where the objectives of the assignment could be fulfilled. The following scenarios were chosen as the ones to control supply and cost of product in addition to inventory levels of both raw material and finished product.

Scenario 1: Production duration is shortened to 4±1 days instead of 5±1 days. As a result, production cost is decreased to $106,000. Although this scenario required overtime payment to operators, the indirect costs were accrued over a shorter time period, hence reducing the total cost.

Scenario 2: Changing the production plan and producing in batches of 600 kg every two weeks instead of the original weekly 300 kg batches. This can be achieved by validating a larger batch size, placing products on stability studies, and filing a supplement with the regulatory agencies. In this scenario, production cost increased to $150,000 per batch.

Scenario 3: In the base case, the average excipients 1 and 2 inventory levels seemed too low compared with production requirements. Therefore through negotiations with the vendors, delivery lead times for excipients were reduced from 1.5±0.5 months to 30 to 45 days.

This scenario analysis showed that shortening of the production time (scenario 1) with other things the same — ceteris paribus — is not useful since production is disrupted resulting in a back-order situation, again primarily due to the shortage of raw material. However, in this scenario, production costs are considerably reduced. In scenario 2, although changing the production plan decreases production disruption by 57% and production cost is decreased by 38%, the backlog problem still persists.

Scenario 3 shows that the most significant bottleneck in the system is raw material availability, especially of excipients. When the upper limit of lead time is decreased by 15 days, backlog of more than 7 days completely disappears and less than 7-day backlogs are rarely observed, hence overall total production disruption is decreased by 46%. Although in this scenario the average inventory level for finished product and the excipients in addition to inventory holding costs increase, this is compensated by the increase in total earnings.

This case study presents only three distinct scenarios. In a real simulation analysis, many scenarios and combinations of them are usually run to achieve better results.

Simulation Models Are The Future
Life sciences supply chain management is a complex and risky process because of the level of uncertainty at its multitude of stages. Historically, high operating margins in the life sciences industry have supported a “better-safe-than-sorry” approach to supply chain activities and, in particular, production and inventory planning. And many companies knowingly overstock excessively to avoid back-order situations. However, now there are significant pressures to reduce working capital without disrupting high service levels or risking compliance issues. In addition, there is intense pressure to reduce costs and improve efficiencies in the fiercely competitive life sciences industry.

Computer simulation, since it can be applied to operational problems that are too complex or difficult to model and solve analytically, is an especially effective tool to help analyze supply chain and logistical issues and help control and improve the systems.