Hector Frias, Consulting Director

Every manufacturer plans production. The question is whether the plan survives the shift.

In most Latin American factories, it doesn’t. ERP systems generate plans from demand and routing data and supply, but don’t calculate against real capacity, real material availability or real constraints. A clean bill of materials and a refined backlog still produce outputs that break the moment a mold change runs long, a shipment arrives late or a machine goes down mid-run. Lean tools like TPM and Kanban improve floor visibility, but their visual structures depend on sustained discipline and when the implementation team leaves, that discipline often leaves with it.

Planning and Scheduling Consultores (PSC) launched in Querétaro, Mexico, in 2004 after its team spent years implementing ERP and Lean systems and watching this pattern repeat. The insight was specific: planning modules failed not because the data was wrong but because the tools did not model constraints. PSC built its entire practice around that gap, became one of Asprova APS’s longest-standing partners outside Japan and, in 20 years, has maintained all the companies it implemented as its current installed base.

What a Factory Teaches Fifty Countries

How does PSC apply constraint-based planning in complex global manufacturing operations?

How that practice works is best understood through what it produces at Fogel Centroamericana, a Guatemalan manufacturer of commercial refrigerators. Fogel ships more than 10,000 customized units per month to over 50 countries, synchronizing purchased parts, electrical and mechanical subassemblies and in-house graphic arts for branded display units. Every refrigerator can differ. Every delivery date is a cross-border commitment.

The engagement started where every project begins, but with a proof-of-concept. The team documented Fogel’s objectives, constraints, performance indicators and current conditions across master data, personnel, IT infrastructure and management tools. The output was a formal APS specification sheet, built jointly with Fogel’s teams, mapping where the company stood and where it needed to go.

“We have no preconceived notions about the techniques to be implemented. Our cross-cutting, multidisciplinary approach allows us, together with our clients’ business knowledge, to define the most appropriate planning and scheduling models to achieve key performance indicators,” says Hector Frias, consulting director.

Joint analysis produced a three-horizon planning model for Fogel covering long-term S&OP, medium-term supply chain coordination and short-term shift-level scheduling, connecting strategic targets to what each operator and machine does every shift. Asprova APS, ranked among the world’s top three in advanced scheduling, mathematical programming and constraint programming, provides the engine. More than 3,000 parameters model specific resource conditions, including setup, execution and dismantling times, mold changes, tooling, authorizations, line release and plant maintenance. Nested scheduling commands configure logic by production area, while a proprietary configuration language gives planners direct control.

The system operates as a closed loop. When a deviation occurs, delayed material, equipment failure or power interruption generates a new optimized schedule using the same allocation and dispatch rules as the original. The target holds. The path recalculates.

What Stays After the Project Ends

How does PSC ensure long-term sustainability of planning systems after implementation ends?

From the first phase, PSC establishes a Center of Excellence inside the client’s organization, designed to outlive the project. Internal teams learn to refine models, introduce new optimization rules and extend planning into areas like transportation, predictive maintenance and new production lines without depending on outside consultants for every change. In 20 years, the firm has never encountered a constraint it could not model, but the point it stresses is that clients’ own teams propose and implement many of those rules independently. The planning model belongs to the operation that runs it.

AI models can build optimal production schedules by running millions of rapid trial-and-error simulations using current demand, inventory and promised delivery dates, without prior training. Instead of a random search, they generate candidates intelligently by spotting simulation trends and focusing effort where performance improves fastest. Even one million iterations can run in about two minutes, enabling real-time use. The approach adapts automatically: it may merge lots when inventory is ample and dates are flexible or keep lots separate when inventory is tight.

What measurable improvements result from constraint-based scheduling across diverse manufacturing industries?

Across industries, Foxconn and Panasonic in electronics, Viscofan in food processing, Sisal Floorings Yucatan in textiles, Primetals Technologies and Franz Viegener in metal, Bopisa and Polymer in Packaging, Pentel and Albea in plastic injection and molding implementations share a first-day measure: every production area receives a shift-level schedule detailing order sequencing, material arrivals, priorities and dispatch logic per machine and operator. Scheduling effort drops approximately 90 percent. What schedulers do with the time they recover is where real gains begin. The plan works because the factory shaped it.