Optimizing Manufacturing Policies for COVID-19 and Future Pandemics

Friday, April 24, 2020

As the COVID-19 pandemic continues to evolve under great levels of uncertainty, manufacturing executives are urgently formulating policies to keep their facilities operating while protecting workers from infection. These policies include macro issues, such as which workers are essential, down to micro policies such as how to best manage shared areas like hallways, break rooms and bathrooms. Without rigorous, quantitative tools and support of the new policies they are creating, executives are effectively flying blind. As evidence, in one European survey, about 70 percent of executives from Austria, Germany, and Switzerland said the pandemic is likely to accelerate the pace of their digital transformation.

Rapid Development of an Optimizing Simulation

At leading global manufacturers, executives in operations, quality, and safety have for weeks been developing policies for worker assignment, spacing, flow and other processes. This development typically occurs through virtual meetings, phone calls and screenshares to identify and prioritize issues, build consensus, and define and publicize action steps. Yet the years of hard-won production experience are insufficient to adequately address the high stakes in this Black Swan moment. It is almost a perfect storm of a lack of fast and reliable tests, coupled with possible transmission by asymptomatic carriers—occurring in processes that were never designed with a highly transmittable pandemic in mind.

Princeton Consultants leverages AnyLogic, the leading simulation software for business, to rapidly build custom models. Each model contains the physical layout, work processes, staff, and COVID-19 infection parameters. Different policy suggestions are rigorously evaluated using AnyLogic’s excellent agent-based and discrete-event capabilities. If desired, an AI function can be applied to search through combinations of policy parameters and find the optimal mix that maximizes KPIs, such as production quantity, and minimizes the likelihood of infections.

Screenshot of an optimizing simulation for a food manufacturer’s representative plant.

Optimizing simulations are high-value solutions that require the right team of business and technical experts. Under extreme urgency as during the COVID-19 pandemic, there is a premium on the methodology for rapid deployment of optimizing simulations, which entail both technical skill and the art of knowing what level of detail is sufficient.

A simulation or digital twin is effective for policy creation because its interface lets business executives more easily and comfortably validate results—simulation “speaks their language.” Initially, Princeton Consultants teams construct the critical inputs: the physics of the system (i.e. “constraints” or “rules”), business data (e.g. what SKUs need to be produced and how much of each) and the decisions (i.e. what knobs and levers can be tweaked by the optimizer). Certain aspects of the environment, from the physical layout of the manufacturing plant to the pattern and rates of infectious disease spread, are unchangeable constraints that the optimizing simulation must account for but cannot change through optimization. These can be set for the entirety of the optimization but cannot be changed by the optimizer.

There is also a minimum amount of work that must be accomplished for the business to continue functioning. A complete minimization of infection risk to the workers would result in zero production but halting production is not always a viable option, especially for essential industries like food and beverage. There are also support functions and staff that keep the facility running while not contributing directly to the production and a minimum level of that support must be met. The optimizer in an optimizing simulation works within these physical and business limits to find the best approach across often-competing metrics like maximizing productivity and minimize worker infection risk.

Events in the real world will occur with a lower or higher frequency than the average, so simulations incorporate this randomness seen in real life, generating results that are more realistic and showing the true expected behavior of the system. It takes into account resources and constraints, as well as the way entities interact with each other as time passes, providing more dimensions of reality than can be accomplished with Excel formulas or “gut” approaches. Each run selects a policy of decisions: how many workers per shift, how much to produce of each SKU, which secondary functions will continue and under what limitations to prevent a negative impact on the primary functions of the facility.

The simulation is run many times per policy, with the optimizing algorithm using machine learning and artificial intelligence to iteratively improve the solution. Over time, the optimizer finds the best (“optimal”) policies that consistently yield the best KPIs, while taking into account the physical and business constraints of the facility. Varying the inputs provides an even deeper understanding of sensitivity and robustness. The results of the optimizing simulation quickly provide tangible policies with quantifiable benefits without the need for risky trial-and-error that could endanger lives and hurt the bottom line.

Screenshot of the optimizing simulation results for the food manufacturer, depicting daily production of all and for priority SKUs, and shutdown hours required in the event of worker infections.

Current Work with a Food Manufacturer

For the U.S. division of a global food manufacturer where executives are responding to COVID-19, a Princeton Consultants team performed the following steps in three days:

  • Interviews of Subject Matter Experts (SMEs) to identify key scenarios to be studied and to establish initial simulation objectives.
  • Deep dive into the physical and business constraints including facility layout, specialized business practices, staffing policies and levels, and KPIs.
  • Quick initial iteration of simulation to gain consensus with SMEs and identification of areas that needed fine tuning.
  • Updated the simulation with additional constraints and KPIs. Another review with SMEs to ensure simulation was realistic enough to gain buy-in from operations team, without including unnecessary detail too expensive and time-consuming to include.
  • Presented a prototype optimizing simulation to safety and operations leaders, who confirmed it was a useful tool to determine optimal policies and visualize the implications of implementing and dangers of not implementing specific policies. They believe the tool can greatly aid efforts to educate executives about managing future health crises.

Blending the Human and Technical Sides of Optimization

Most operations and safety executives have not personally used optimizing simulations, so they are relying on the acceleration of traditional policy setting procedures, based on human judgment and interactions, and crisis management. There are analytics leaders at manufacturers who recognize that optimization and simulation tools can help their business colleagues immediately. However, the lack of available internal advanced analytics expertise prevents them from delivering badly needed custom solutions.

Successful development and deployment of optimizing simulations requires expertise in business processes, advanced analytics modeling, and software programming. Since 1981, Princeton Consultants has blended the human and technical sides of optimization to build and implement trusted solutions for many leading industrial businesses. Its proven methodology for rapid solution development and delivery lets its teams respond immediately to executives challenged by the current pandemic.

For manufacturers, COVID-19 requires speed and quality in decision making about operations and worker safety. Optimizing simulations will promptly help executives rigorously assess their options and the impact of the policies they create. In addition to supporting safety and productivity, these nimble, customizable tools will help executives prepare their organizations for future health crises.

We welcome a conversation to explore if Princeton Consultants can be a resource today for your organization. Contact us to set up a call.