How Scheduling Can Co-Exist with Lean

by Kien Leong

Today, manufacturing excellence is built on execution. Lean manufacturing places emphasis on daily execution to customer demand. It favours a system of visual signals on the factory floor to replace computer planning and paper reports. Does this mean that production scheduling is in conflict with best practices of lean and demand-driven production? Certainly not. Finite scheduling is very different to Master Production Scheduling.  Scheduling is an essential part of integrated planning and a powerful way to model real-world constraints in a supply chain.

Manufacturing Excellence: Planning vs. Execution

Top manufacturing companies have to master two areas of excellence: planning and execution. Planning processes determine what we will do, how and when we will do it; execution processes focus on getting it done. Excellence in execution is well-established and enshrined with principles of lean, demand flow, six sigma and theory of constraints. There are many variants and combinations of these principles and tools. However, they are largely based on common values: customer-centric thinking; continuous elimination of waste; processes linked in a flow; controlled process variation; visual factory management and fast-turning inventory. Excellence in execution is built on a foundation of proven principles that have generated tremendous results for those who have adopted them.

Excellence in planning is less clear-cut. For many companies in manufacturing, the primary tool for planning is MRP. MRP carries a logic that is seriously limited when it comes to planning in a modern manufacturing. MRP likes a forecast and long lead-times on customer orders; it is a “schedule-push” system. When the forecast is unreliable and the customer expects a fast response, MRP fails to perform.

Execution Excellence = End of Planning?

The lean advocate who delivers execution excellence doesn’t like a “schedule-push” approach. It is in conflict with the customer-centric concept: that you should allow your customer to decide what is made and when to make it. This is demand-driven manufacturing: any product, any day, all according to customer demand. So, does it mean that a lean or demand flow company does not need to do any planning and scheduling?

Well as with most things, it is not that simple. Demand-driven manufacturing works to a paradigm that has all materials, work and transactions driven directly to real-time demand. In this paradigm, execution is everything. Get the execution right and you don’t need planning at all. This is great for vision, but the problem is that it ignores the realities of business. If daily demand sometimes exceeds daily capacity, or if suppliers have long lead-times, then you are going to need planning.

When lean believers hear the word “scheduling”, eyes start to roll. To them, “scheduling” means master production scheduling. This is a “schedule-push” method, whereas we want “demand-pull”. Therefore scheduling is evil, a forecast-driven sin.

Not All Scheduling is Evil

If scheduling means Master Production Scheduling to a forecast, then yes it is evil and something we should avoid. However, we need to be specific about what is being scheduled. If we are scheduling forecast requirements in production, then that is one thing. If we are scheduling customer order commitments, then this is entirely different. This is a bit like the old lean mantra: “inventory is evil”. Do we get rid of all inventory just because some of it hides waste and inefficiency? Of course not.

Scheduling is ”the process of deciding how to commit resources between a variety of possible tasks” (Wikipedia). If you have multiple things to do and you cannot do them all at once, you need to schedule. The most flexible, lean factory has to decide the sequence with which sales orders get released into shared resources. Even if you receive orders in the morning and release to production in the afternoon, you will still need to schedule production. Most manufacturers that have a fair degree of capital intensity or resource cost may have to smooth demand over more than a day.

So, scheduling by itself is not an issue. The form of demand is what matters. Are you scheduling an expectation of future demand or a real customer commitment? If it is a forecast expectation then you are push-scheduling. If an order commitment then you are more likely scheduling for demand-pull.

Make-to-Stock and Demand-Driven Scheduling

What about make-to-stock production? Does this mean that all make-to-stock is push-scheduling, the evil kind? No, because there is a third form of demand beyond forecasts and orders. Replenishments are a demand signal that is taken from a movement in stock. If finished goods are consumed below a preset level or to a whole Kanban quantity, a replenishment signal is sent. We can schedule a queue of Kanban quite easily. You can still comply with your visual factory principles and your demand-driven production strategy. The scheduling is there to model the process and ensure that the Kanban or min-max design has enough stock and capacity to meet with demand without shortages.

Modeling, Prediction and Execution

Production scheduling pursues two fundamental objectives: prediction and execution. The most important outcome of a scheduling system is a reliable prediction of what will happen and when orders can be completed. An answer to that golden customer question: “Can you fill my order by my order date?” and “If not, when can you fill it?”. A reliable prediction model enables management to properly size resources and inventory, set policy and make customer service commitments.

Production scheduling systems can also deliver execution: “What should I make next?”. There is still a role for production scheduling in terms of execution. There may be a level of consistency, measurability or optimisation that a manual system cannot deliver. This way, a production scheduling system is able to co-exist nicely with a lean execution system, filling the roles of execution that it does best and deferring to the simple, visual methods for the rest.

The biggest gains from production scheduling in a lean factory are going to arise from meeting order commitments. A customer-driven company makes promises and keeps them. Sometimes, the level of demand or complexity means that the outcome from an execution-led system is not clear. We need scheduling tools that will enable us to make a fast, reliable prediction about capability and control the result to meet our promise.

{ 5 comments… read them below or add one }

Tony Rice September 20, 2010 at 12:52 pm

Very well put. I like to think of the analogy:
When driving a car, you respond to what you see through the windscreen – that is execution.
You also have a GPS – that is scheduling.

John Broughton September 22, 2010 at 3:12 am

We were always told in TOC that you shouldn’t schedule but work everything to the constraint. Nice in theory but sometimes we need more. It does seem to me that there is conflict with the “purist” view and what people actually do.

Kien Leong September 23, 2010 at 4:23 am

Finite scheduling is often put to work with a process flow that has mutiple constraints. The effective constraint moves, depending on the mix of product in the order book. With a single, clearly-identified constraint you may find that capacity planning will suffice. The best practices in drum-buffer-rope, kaizen, line balancing on the factory floor are still required. While you buffer/ improve / transform the process, you still need to take and fill orders. Planning and scheduling give you visibility on current and future capabilities. This is the ability to make due-date adjustments before they become customer promises. To model the process improvement and project a target result.

Louis Marbel September 22, 2011 at 6:10 pm

Kien,
The problem is even more acute in situations such as Quick Service Restaurants (QSR) where you have to react to demand daily and even time of day. Following on your point on FCS (Finite Capacity Scheduling), we can employ a generative approach to FCS using simulation to build a predictive model of resource load against a possible demand distribution. In essence we are executing to test drive possible schedules. On the day of execution we can then just rely on Lean and Kanban line designs with parameters that can be easily controlled and their correlation to real world operational factors well understood through the simulation. I still would like to keep the simplicity of Lean at the execution level.

les bewick January 31, 2012 at 12:52 pm

Reading through some of your articles and working on your template’s, they are obviously well proven techniques, as you comment on MRP Pull and push, i think push has now became i thing of the past, as inventory and high volumes of WIP contribute to bottle necks and add Inventory costs. I have worked on MRP systems ERP with APS and Real time.
They all have buckets of data passing through production with time horizons, Question in your view how would you rate ERP with MRP
thank you

{ 21 trackbacks }

Previous post:

Next post: