Supply chain planning and scheduling, from someone who has done it.

I’ve implemented linear-programming-based advanced planning and scheduling (APS) applications at more than 75 sites since 1989. An APS implementation is not to be taken lightly: it usually requires re-engineering the planning process and higher skill levels from planners. These are the strongest signs it will pay out:

• Tightly coupled make/pack or feeding/consuming operations.
• Scheduling difficulty beyond the scope of a single planner.
• Constrained production.
• Highly deployed inventory.
• Rescheduling required within the day.
• Production and shipping that must be coordinated within the day. Otherwise you may plan to ship at 10 am something that won’t be produced until 2 pm.
• Complex spreadsheets used to plan production or distribution.

To gain productivity and create better plans with an advanced planning and scheduling system, you must do three things: precisely schedule the minimum horizon, allow the system freedom to plan, and focus on the real constraints.

Schedule the minimum horizon

With less capable systems, planners spend time building a long-term plan for capacity projections and supplier planning. An APS can create that plan automatically from the rules and constraints you configure. Manually adjusting it adds no value, because it will change the next time you plan. If the longer horizon doesn’t look right, revisit the constraints and planning rules rather than adjusting the plan by hand.

Allow the system the freedom to plan

If everything is a constraint, no automated plan can satisfy them all. Planners end up adjusting the plan by hand to make it feasible, which means breaking constraints anyway, and then they conclude the system saves them nothing. Break the cycle by identifying the few constraints that truly cannot be violated, using weighted penalty costs for everything else, and letting the system find the low-cost plan.

Focus on the real constraints

If planners routinely violate something to get a feasible plan, it isn’t a real constraint.

The number of possible production sequences grows factorially with the number of products on a line: 4 products have 24 possible sequences, 8 have 40,320, and 15 have about 1.3 trillion. The planner’s job is to find the one that minimizes setup cost while protecting customer service. These factors determine how much an advanced planning system is worth:

• Planners per value stream. The ideal is one planner across the whole horizon; benchmark companies use no more than two.
• Capacity-to-demand ratio. Tight capacity leaves no room to run the wrong product.
• Levels of production. The best schedule at one level is rarely optimal, or even feasible, at another.
• Storage between levels. Less buffer time means more blocking and starvation.
• Shared operations and resources. Shared lines, crews, and washout stations need visibility and coordination.
• Planning cycle frequency. A shorter cycle cuts response time and safety stock, but without better tools it adds planner workload.
• Reaction time. The time to recognize the plan is off track, replan, produce, and ship drives the inventory you have to carry.

At P&G we didn’t have enough qualified instructors to support a global OMP+ implementation. Our answer was a set of eLearning modules that let students learn the basics of the software at their own pace, and completing them was required before attending instructor-led sessions.

Instructor-led training time fell by 50%, and students retained more. Classroom time went to the difficult topics that need interaction, and the approach let us offer remote classes to anyone, anywhere in the world.