You may have a scheduling module within your ERP system, and it’s good at what it does. But it’s not clairvoyant.

In most cases, it gets inputs from the sales staff and looks at what’s in house now, and if there isn’t a sufficient quantity in stock, then schedules a run when current runs are complete or when there’s space in the schedule. With any luck, it checks on available ingredients, and if they’re not in stock alerts the appropriate person to order what’s needed to make the product. Then, it passes along the order information to Operations, who laboriously copies the information into an Excel production schedule file—or worse yet, writes it on a clipboard.

But your operation makes both foods with and without allergens, and you have to deal with seasonal changes to orders, and fulfill large, customized orders for a big box store, for which you have an ongoing contract. And in today’s tight labor market you do not always have the right people at the right time when you need them, and equipment downtime is always a factor. How do you schedule in this chaotic environment?

There ought to be a better way. That’s what three professors from the University of Colorado thought. In the early 1990s, James Kelly, Manuel Laguna and Fred Glover were combining their diverse backgrounds and expertise in mathematics, engineering and artificial intelligence to pioneer the development of new methods for complex optimization.

At the time, many companies were looking to optimize their processes, but optimizing with goals and constraints were often all too complex to be expressed in classical mathematical programming formulations. Then, throw in the presence of uncertainty and any hope of optimizing a process with traditional methods goes out the window.

The three professors recognized that it might be possible to optimize a process by modeling with simulation. Simulation is ideal for handling conditions in several areas—including production, distribution, energy systems, power transmission, workforce planning, environmental analysis, engineering design, and many others.

This relatively new field of optimization and decision science, called meta-heuristics, uses the sophisticated problem-solving methods the three professors had developed to replace traditional problem-solving methods that had proven to be inefficient and inadequate.

This was the birth of OptTek Systems, Inc. and the OptQuest optimization system, the company’s first core software technology. Besides providing the OptQuest optimization system, OptTek also produces OptPro for solving complex production scheduling problems and OptForce for workforce planning and optimization plus specialized government and military applications.

OptTek, which presented at FE’s recent FA&M Conference, has applications in use at companies such as GE, IBM, Rockwell Automation, Oracle and several others. I asked Dr. Marco Better, vice president of analytics services, OptPro & OptForce, (and speaker at FA&M) to explain how OptPro can be applied in the food and beverage industry for solving the kinds of production scheduling problems that crop up.

FE: OptTek has worked with some pretty large corporations in diverse fields over the years. Who are some of the food and beverage companies you’ve helped?

Dr. Marco Better: I cannot disclose the name of some of the companies as they are still active engagements and we have non-disclosure agreements in place. However, I can say that we have worked with a large global liquid foods processing and packaging equipment company; a large South American animal protein processing company; and we have proposals out to several US-based food and beverage enterprises.

FE: As I said earlier in the introduction, food and beverage companies are prone to all sorts of production scheduling problems. If you’re making electronic components, they don’t spoil if they’re not sold right away. But with food, scheduling is crucial to prevent waste and financial losses—not to mention food safety issues. What are some of the key production scheduling issues you’ve dealt with in your experiences with food and beverage companies?

Better: Production scheduling is centered around three main concepts: (1) How long do you produce a particular SKU; (2) What resources (lines, machines, people) do you assign its production; and (3) How do you sequence production of your different batches. In food and beverage, all of these become absolutely critical.

Let’s suppose you work on a weekly basis. Ideally, in order to maximize the capacity utilization of your equipment, you would like to produce your weekly planned production of a particular SKU all at once, before changing over to the next SKU. This would minimize changeovers, cleanings and setups (which can be rather costly if they involve allergens, for example); however, if you do so you will run into problems with perishability and storage capacity, not to mention that your customers may expect more frequent deliveries of that SKU than just once a week.

In addition, especially with liquids, you need to store your work-in-process in dedicated tanks, and products cannot be mixed. Thus, you need to try to utilize your storage capacity as efficiently as possible. So, the size of your production batches, the assignment of those batches to production lines, and the sequence in which those batches are produced become critical.

FE: How is OptPro better at scheduling and optimization than production/scheduling modules in an ERP and/or MES application?

Better: Your ERP scheduling module may tell you what you need to produce that week to satisfy demand, but only a mathematical approach can tell you where, when and how much to produce of each SKU such that plant capacity and throughput are maximized while inventory limits are met, and operating costs are minimized, without sacrificing your delivery commitments to customers.

FE: Before going further, can OptPro be used in conjunction with ERP or MES? If yes, what is involved in the integration?

Better: OptPro is a scheduling optimization engine. It can be integrated into any ERP or MES system, but that requires building custom interfaces. We typically do not build those, because the client wants the interfaces to maintain the look-and-feel of their own tools. What we do build are the data input and output layers.

The data input layer is designed to take a flat file (a .csv file, for example) that contains data that has been extracted from the ERP/MES system and read it. Then, the scheduling optimization engine processes the data and transforms it into an optimal production schedule. Once the schedule is ready, the data output layer transforms it into a flat file that can then be read by the ERP/MES system and turned into specific instructions at each workstation in the plant, such that operators can know what assets to use to process the different SKUs, for how long and in what sequence.

FE: What, exactly, is optimal production scheduling?

Better: Optimal production scheduling is a methodology that uses mathematical modeling and algorithms to obtain the best possible production schedule. In this sense, we refer to true optimization in the mathematical sense, as opposed to the way the term “optimization” is frequently misused to merely mean improvement.

In the mathematical sense, true optimization really means finding THE best solution—the best schedule—given an objective and a set of restrictions.

Examples of objectives are maximizing production throughput, minimizing operating costs, maximizing on-time shipments, etc. In other words, these are the goals you are aiming for. Restrictions, on the other hand, are constraints that limit your ability to reach those goals. Changeovers and setups, cleanings and maintenance, inventory concerns and storage capacity, limited resources, etc. are all examples of these.

Optimal production scheduling considers these goals (alone or in combination) and these restrictions in order to find the best possible schedule that optimizes the goals by exploiting the restrictions.

FE: Who should use optimal scheduling—other than everyone?

Better: Companies ought to invest in production scheduling optimization technology when their manufacturing operations exhibit one or more of the following characteristics:

  • High complexity
  • Inefficient setups/changeovers
  • Multiple products share common resources, in terms of infrastructure or labor, or both
  • Production costs represent a large proportion of the cost of goods
  • Want to increase production without making additional capital expenditures
  • Experience high product losses and waste

We have found that many medium and large operations in the food and beverage industry typically fall into one or more of these categories. Dairy, animal protein, and milling operations are just a few prime examples.

FE: So how does OptPro work in the real world of food production, which involves non-allergens and allergens, short shelf-lives, cleaning/CIP cycles, holiday/seasonal crunch times, ingredient suppliers who crap out at the last minute, warehouse/logistics scheduling, unplanned machine downtime, employee sick-outs and the like? Did I leave anything out?

Better: You hit the nail on the head! The real world is much more complex than any academic exercise. When I last checked, a Google search of “production scheduling optimization” yielded more than one million scholarly articles. However, on the one hand, there is very little evidence that companies are applying any of the methods and techniques espoused in those articles; and, on the other hand, those articles typically analyze a narrow sliver of a production “problem”, and ignore many of the realities you mention.

OptPro is designed to apply advanced techniques and methods rooted in classical science, but with an eye towards real-world situations and complexities. That is why we are not a one-size-fits-all tool that you can just buy off-the-shelf. We provide a highly configurable solution that represents the reality of a client’s unique situation. Based on a foundation of more than 30 years of research and 25 years of commercialization, we have developed a solution that combines classical optimization with simulation to provide a comprehensive approach. The approach basically creates a “digital twin” of a client’s factory, with all of its physical and logical restrictions, and applies the appropriate optimization methodologies to address their principal objectives.

This approach allows the client to optimize multiple objectives simultaneously, looking at optimal trade-offs between, for example, maximizing throughput and minimizing operational costs, all while complying with inventory restrictions, shelf-life considerations, and changeover requirements. But the approach is also fast enough that it allows operations to be responsive to last-minute changes and disruptions such as a supplier crapping out on you, an emergency order coming in, or a machine breakdown. Our methodology can be employed to quickly reoptimize the production schedule around the disruption, without changing the schedule so much that it makes it impracticable. This is something our clients really appreciate!

FE: At FA&M, you described a lot of mathematics that is used inside of OptPro. At a food company, what academic credentials/qualifications are needed to use this software? Can existing staff use it successfully? What do you need to know before employing OptPro?

Better: We are very proud of the technology that runs within OptPro, so we like to explain the basic concepts of our methodology to our clients during the implementation phases; but it is not necessary for an operator, a planner, or a scheduler to understand the detailed math behind OptPro in order to be able to use it.

The sophisticated mathematics in OptPro occur automatically when the user runs the software. Assuming we have captured all the intricacies of the client’s operation successfully during the initial implementation and configuration of the solution, then the software should run correctly without much input from the user.

FE: What are the inputs a food processor supplies to OptPro?

Better: The inputs OptPro needs can be classified into two basic types: (1) a one-time input of the plant configuration, with a list of the assets (equipment and labor) involved in production, and the sequence of steps, processing times, and changeover/setup times required to process each SKU at each asset; and, (2) a periodic input of the customer demand (i.e., customer order quantities and due dates) and the current status of the plant so that the scheduling software can know what it needs to schedule.

FE: How does OptPro output knowledge to a company’s production/scheduling system?

Better: OptPro outputs a detailed schedule to a client’s database. The schedule includes all the information necessary to transform the data into operating instructions at each workstation of the plant.

The information provided includes, at a minimum:

  • What SKU, when and how much of it to produce (SKU batch ID) on which asset;
  • The sequence of production runs on each asset;
  • The changeovers required between each production run on each asset;
  • When to perform scheduled maintenance on each asset.

FE: What are some real-world results a food company could expect to realize after employing Opt-Pro?

Better: The results vary widely from one engagement to another, depending on multiple factors, such as the primary objectives of the company, and the severity of the restrictions on things like personnel availability, inventory constraints, shelf life, etc.

Depending on the business problem (objective or objectives) being addressed and timing horizon, we have had improvements in productivity (defined as maximum throughput, percentage on-time shipments, minimum cost, etc.) ranging from a few percentage points to nearly 50% for a given SKU, line, and/or total plant.

FE: I notice that OptTek has a workforce optimization tool called OptForce, and it really wasn’t covered in your FA&M presentation. Is this a tool that can be used in conjunction with OptPro in terms of scheduling employees, or is it really geared more towards HR functions—like finding and retaining employees or long-range planning/putting together a workforce?

Better: OptForce is geared towards more long-term, strategic planning of talent, so we offer it entirely independently of OptPro. However, many of our OptPro clients are ideal candidates for OptForce, so once we get “in the door” and the client has had a chance to experience our high-quality delivery with one product, we are always interested in cross-selling the other.

FE: I’ve visited food plants where finding skilled technicians and engineers is a real problem because of shortages and/or competition with other manufacturers in the area. Can OptForce help in any way with this issue?

Better: Yes, OptForce could help. OptForce looks at optimally allocating your budget among retention and recruitment practices to attract and retain the talent required to support corporate financial and operational goals.

If you have a particular company in mind, we would love the opportunity to contact them and provide a free demo!

FE: Besides hard-to-find technical people, machine operators and laborers can be equally difficult to attract and retain. Can OptForce help?

Better: Again, OptForce can be applied to any and all roles within an organization, so yes, machine operators, maintenance technicians, plant workers are all within the scope.

FE: Can OptForce help in managing employees when moving to or building a new plant in a new location? How can it help?

Better: This is a good use case for OptForce. When a company is thinking of shifting or expanding its workforce, OptForce can look at talent availability in new markets, compensation levels, and onboarding and training programs to help attract and retain employees.

FE: What didn’t I ask that you feel is important?

Better: I just want to reiterate the importance of optimal production scheduling. In a complex operation, it is impossible for even the most experienced schedulers to create an optimal production plan. Thus, many companies conclude that they are at capacity, and end up investing large amounts of capital to expand their current operation with new equipment or new facilities, when they may have been able to achieve similar productivity improvements by optimizing their schedules.

I believe there is a gap in ERP-MES systems that OptPro can fill to help companies achieve these productivity improvements without additional capital expenditures. As we like to say to our clients: if you’re not optimizing your operations you are leaving money on the table!

About Dr. Marco Better

Dr. Marco Better manages all activities related to the implementation, delivery and service of Opt-Tek’s production scheduling and workforce planning solutions. Better obtained his Ph.D. in Operations Research from the Leeds School of Business of the University of Colorado at Boulder. He holds a B.S. in industrial engineering and an M.B.A. Better has over 25 years of professional work experience in the automobile, banking, and telecommunications industries, both in the U.S. and in Latin America. Dr. Better teaches courses in International Business and Business Process Analytics at the Leeds School of Business. His current interests lie in the application of optimization, simulation and predictive analytics technology to solve complex problems in industry.

For more information visit OptTek at