Reflections on Lean Philosophy and the Theory of Constraints

Lean Manufacturing and the Theory of Constraints

Lean manufacturing and the theory of constraints (TOC) may well go hand in hand. Lean manufacturing makes value flow through the factory, for instance, by trying to separate value streams so that they use dedicated resources sized to the same capacity (even if kanbans do not optimize the constraints). TOC takes this idea further by recognizing critical bottlenecks, which are the most overloaded resources that determine the maximum flow rate of production, and making value flow through these bottlenecks. It does this by allowing manufacturers to optimize production through their critical bottleneck in order to meet market demand.

This is Part Seven of a multipart note entitled Lean Manufacturing: A Primer.

For these reasons, a TOC production planning solution might be appropriate for manufacturers with make-to-order (MTO) environments, where demand is volatile and where different product lines share the same resources, resulting in bottlenecks. It could also be used for mixed mode manufacturing. In fact, by offering daily production planning for customer orders received, TOC enables business performance improvements in such environments in terms of lead time or cycle time reductions, increased throughput and sales, service level improvements, and inventory level reductions.

Thus, despite the fact that many people immediately invoke a vision of kanban when lean manufacturing is mentioned, TOC supports a lean philosophy where there is a complex environment. However, where lean planning focuses on the flow and the takt of the flow through the factory, TOC optimizes the flow through the factory by focusing on planning the takt of the flow through the bottleneck. TOC is also consistent with lean manufacturing in that both kanban, which is a part of the just-in-time (JIT) philosophy, and drum-buffer-rope (DBR), which is a part of the TOC philosophy, represent synchronized and pull signal production control approaches.

For an exhaustive discussion of lean manufacturing in previous notes, see Lean Manufacturing: A Primer, Lean Tools and Practices that Eliminate Manufacturing Waste, How to Achieve Lean Manufacturing, Manual versus Information Technology Enabled Lean Manufacturing, Enterprise Resource Planning Vendors Address Lean Manufacturing, and The Theory of Constraints Enters the Lean Manufacturing Arena.

The TOC Vernacular

More similarities between TOC and lean can be extracted by analyzing some TOC definitions. For example, in the TOC lingo, throughput is the rate at which the system generates money through sales. In other words, throughput is production that can be invoiced—only monetized sales generated by the system get counted. Building inventory (just for the sake of stocking up), on the other hand, is not throughput in TOC terms. This is consistent with lean manufacturing's focus on the customer and customer value-adding activities. Another example is TOC's definition of inventory, which includes all investments in procuring materials to meet customer demand, such as raw materials, work-in-process (WIP), finished goods, and scrap. The crucial point, however, is that, according to TOC, inventory is a liability and certainly not an asset. This is consistent with lean manufacturing's focus on eliminating waste. Finally, TOC's definition of operating expenses encompasses all the money the system spends to turn inventory into throughput, such as all employee time, depreciation, etc. Therefore, TOC focuses on increasing throughput, while reducing inventory and lowering operating expenses. A TOC cost and managerial accounting system thus logically accumulates costs and revenues into these three areas.

The TOC accounting system is somewhat similar to activity-based costing (ABC), since it does not create incentives (through allocation of overhead) to build up inventory. It is considered to provide a truer reflection of actual revenues and costs than traditional cost accounting. Since it is closer to a cash flow concept of income, TOC accounting provides a simplified and more accurate form of direct costing, one that subtracts true variable costs (those costs that vary with throughput quantity). Also unlike traditional cost accounting systems, in which the focus is generally placed on reducing costs in all the various accounts, the primary focus of TOC accounting is on aggressively exploiting constraints to make more money for the firm. Similarly, TOC's goal is to maximize throughput on the bottleneck, which is equal to the profit, since, according to Goldratt et al's 1984 blockbuster business novel, The Goal, "an hour lost on the bottleneck is lost forever and an hour saved on a non-bottleneck is a mirage."

TOC Implementation

In practice, TOC is implemented by following the subsequent five straightforward steps.

1. Identify the constraints. This should not be too difficult, since large piles of WIP are very noticeable, and every plant supervisor should know intimately the sore spot or bottleneck within the plant.
   
   
2. Exploit the constraint. One has to maximize the possible amount of work going though the constraint, while ensuring that there is an uninterrupted flow of work coming into the constraint, so that it never has to wait for work (i.e., an inventory buffer is kept in front of the bottleneck to ensure that it is never idle).
   
   
3. Subordinate everything else to the constraint. Since the efficiency at other resources does not really matter, there is no point in upstream work centers producing more work than the constraint can absorb. It is sufficient to provide an indication of the task priority of other non-bottleneck resources, since the utilization of non-bottlenecks is determined by the critical bottleneck.
   
   
4. Elevate the constraint. If possible, increase the capacity of the constraint by offloading some work, subcontracting work, adding more capacity (by buying more machine, adding another shift, etc.), and so on. 5. Repeat the entire process for continuous improvement. This is another similarity with the lean philosophy. It is likely that elevating the constraint will stop it from being a constraint, but a new constraint will come to light. One then has to exploit, subordinate, and elevate this new constraint.

DBR Explained

The DBR process is used within TOC to manage resources in order to maximize throughput. In simplified terms, throughput becomes the critical index of production performance. The barrier to maximum throughput is typically thwarted by a single capacity-constrained resource (CCR), or bottleneck, so the focus is on maximizing utilization of that bottleneck.

The term drum-buffer-rope encapsulates the main concepts of DBR. The drum refers to the rate or pace of production set by the system's constraint. The buffers establish protection against uncertainty (e.g., machine breakdowns, material shortages, labor problems, etc.), so that the system can maximize throughput. The rope is a communication process from the constraint to the gating operation that checks or limits material released into the system to support the constraint (i.e., a sort of a pull system, which is yet another similarity with lean).

In TOC, the constraint is viewed as a drum, and non-constraints are, according to Dr Eliyahu Goldratt, like soldiers in an army who march in unison to the drumbeat—that is all the resources in a plant should perform in unison with the drumbeat set by the constraint. In this regard, one should note that the system constraint may be either internal or external. In fact, Infor reveals that the vast majority of its customers who have implemented the lean and TOC approach have discovered, once the work flow has been corrected, that the market becomes the constraint. Other constraints to throughput include resources, materials, and, most insidiously, management.

Thus, DBR begins by identifying a critical bottleneck, which is the strategic drum or synchronous control point. The drum schedule for the plant, which sets the pace for the entire system, must reconcile customer requirements with the system's constraints. Other resources may be a temporary bottleneck for a short period depending on the order mix. Market pull is scheduled on the drum, and material is released onto the floor at the rate that the drum can operate. This rate is the rope, which consists of the minimum set of instructions to ensure that non-constraint resources are used (and not over-activated or misallocated). Material is consequently released into the system and flows to the buffers in a way that supports the planned overall system throughput. In fact, material release occurs a set buffer time ahead of demand, so that some buffer physical inventory (but not too much) is present at the drum resource to guarantee its performance in order to plan against uncertainty. In TOC, buffers can be either time or material to support throughput or due date performance. They can be maintained at the constraint, convergent points (with a constraint part), divergent points, and shipping points.

The Role of Extended Enterprise Resource Planning Systems in TOC

Enterprise systems come in handy when calculating complex TOC algorithms, such as, for example, defining the planned start and stop time per order down to the minute, or determining the production rate for the entire factory. A system such as Infor's Easy Lean/DBR system can manage internal constraints, time buffers, and replenishment or kanban buffers. Users can thereby execute operations on the bottleneck according to the planned start time. In addition, the priority on each operation and remaining buffer levels can be visualized—the earliest start time of the buffer indicates how realistic the plan is, while the remaining buffer controls execution priority depending on when it is planned on the bottleneck. As with kanbans, the rule of thumb is to start with a large buffer size and keep reducing it until one has a smooth flow, since the smaller the buffer sizes, the shorter the lead times and the faster the production flow.

When it comes to the execution on non-bottleneck resources, this can be done by indicating the remaining buffer in the system using red, yellow, and green buffer flags. Red flags indicate the highest priority tasks that should be focused on, while yellow designates less critical tasks, and green denotes tasks that are in the buffer and thus still in "good shape". Operators use these flags to execute tasks according to the priority level, rather than according to a defined order sequence and specific times as in material requirements planning (MRP) or advanced planning and scheduling (APS) systems. This gives operators more flexibility and the ability to make some decisions about which task to execute next. This can increase the motivation level, and is in tune with lean philosophy's employee empowerment mantra.

Yet another thing that differentiates TOC systems is the fact that, since inventory is only held in front of the critical bottleneck, it is normal for the company to end up with significantly less inventory in a TOC system than when using MRP or JIT. WIP inventory is often lower than those of kanban systems because aggregating the buffers offers the same protection overall, while simultaneously reducing the amount of protection required. Shorter production cycle times also have a similar result.

TOC is also a departure from the heyday of APS technology in the late 1990s, when the hope was that APS-based applications could fully optimize factory resources within a company, or even within a supply chain. In the 1990s, there was widespread belief that some supply chain planning (SCP) software vendors had the magic answer to supply chain problems. This line of thought allowed and even entrenched bad business processes. For instance, the probabilities, statistical variants, and noise generated while attempting to model everything in a system can actually cause bottlenecks, or become one on its own. Moreover, it is certainly much easier to schedule one constraint than to concurrently schedule them all, as sophisticated APS models that attempt to model everything are difficult to set up and almost impossible to understand and maintain. For more information, see Advanced Planning and Optimization Software: Myths, Facts, and User Perceptions and Advanced Planning and Scheduling: A Critical Part of Customer Fulfillment.

TOC Hurdles

TOC is attractive on several levels. For instance, part of TOC's appeal is the fact that it is fairly easy for the planner to use. It is also easy to communicate resource plans, as they are visual (yet another lean concept). High data quality is not an issue either, since reasonably good data quality is only required on the bottleneck, and it does not have to be 100 percent accurate, though it does need to be good enough. As only the lead-times for the bottleneck are focused on, inaccuracies in other lead-time elements are averaged. Moreover, when a bottleneck is fixed (i.e., exploited, subordinated, and elevated) and no throughput improvement is apparent, one knows to look first for a data collection error. In addition, the fact that there is one common plan and each order is prioritized ensures that all departments are working towards the same goals. However, despite all these advantages, there are definite challenges in implementing TOC.

Though the Supply Chain Council (SCC) defined the DBR scheduling technique as a best practice in the Supply Chain Operations Reference (SCOR) version 7 model, and though it is simple to grasp in theory, DBR is counterintuitive to decades of engrained management thinking about optimizing all production resources. The biggest barrier—and one deeply embedded in management thinking—is traditional cost accounting, under which view of manufacturing, all machine and labor resources should be utilized to the maximum. By contrast, as mentioned previously, the TOC view of the world revolves around pacing a plant around a key bottleneck using DBR techniques, even if that means under-utilizing non-bottleneck areas.

DBR or TOC-based software can help determine and monitor the bottleneck center; set and manage the buffers; time the release of work to maintain a smooth flow within the bottleneck and to maximize throughput within the plant; and analyze resources and buffers for continuous improvement. However, this is only a small portion of what has to happen. The software alone cannot convert a plant from a traditional mindset to a TOC way of doing things. TOC, like lean manufacturing, is more about attitude, approach, and business practices than about software. The major thing that has to change is the company culture—it has to commit to making decisions based on throughput information rather than on cost information. To do that, one has to put new measures in place; the software can only help to reinforce these new procedures, disciplines, and measurements.

Moreover, subordinating non-bottleneck feeder plants to the constraint plant occasionally requires equipment to stand idle, and only a stringent TOC education can help management accept this transformation. Some education is also needed for workers, who may sabotage the system for fear of job insecurity ominously suggested to them by reduced buffers and WIP—ironically, the prospect of increasing throughput and sales volume should ensure greater job security. Another mind-set-based hurdle for TOC is the fact that it is often easier to sell a software solution that appears to do more, such as an APS package that optimizes dozens of resources, than a DBR or TOC-based solution that focuses on only one or two resources.

Finally, when implementing TOC and DBR, an enterprise often will conduct a comparative analysis of production for all its manufactured items, only to find that, for many items, the full absorption costing was the exact opposite of what one would expect from TOC in terms of realizing increased throughput. Outside expertise often helps with both the required change in mindset, and the more technical aspects of TOC. Thus, it is highly recommended that system experts and TOC experts are not the same people.

Is There Something Simpler that Might Work?

As you can see, a full-fledged TOC deployment requires certain thinking process tools, which are important in identifying the root problem, identifying and expanding win-win solutions, and developing implementation plans. Full-blown lean manufacturing, on the other hand, requires significant time and diligence to set up, and is not always the best route. A simplified hybrid approach called simplified market pull (SMP) features the best of both TOC and lean manufacturing principles, and, for some companies, is easier and quicker to deploy. It was introduced a few years back by the former Lilly Software (a pioneer in concurrently addressing both material and capacity constraints), and is still promoted by Infor.

With SMP, users do not need to map the whole value stream to significantly reduce WIP and lead time, but they do need to simply stop releasing work to the plant floor until the right time. Under traditional push-style manufacturing, the wishful thinking is that the sooner work orders are released to the floor, the sooner they will come out the delivery end. Of course, what typically happens is that capacity gets allocated to forecasted orders that do not have an actual current demand. SMP, conversely, involves protected pull, in which work is only released to the floor if there is a confirmed customer order. If the customer wants the product immediately, the solution's logic protects it with some inventory buffer, whereas if the customer asks for the order by a certain date, the protection is via time buffering.

Since the factory floor will not be flooded with forecast-driven work, much of the queue time and WIP goes away after a while, and shorter lead times, more accurate deliveries, and even increased actual sales can be accommodated without much complex algorithmic calculation. For this to work, however, another change in thinking, in addition to overcoming the fear of seeing fewer buffers and thinking the work is drying up, is required. Namely, under the traditional push approach, one focuses on the latest possible date to release an order to the floor, while under SMP the principle becomes to not even think about releasing the order before the determined date.

User Recommendations

Manufacturers and distributors of all sizes and across all industries that plan to improve productivity and customer relationships should assess their operations for lean ways to reduce waste and add customer value. To do this, they have to answer basic questions, such as how to buffer operations against instabilities and how to level production. These manufacturers must also be fully aware of whether their system uses actual demand, sales forecasts, or a combination of two in order to populate their master production schedule (MPS). These companies should be sure to leverage existing and new enterprise systems investments towards the lean mission to achieve true success and, ultimately, profitability. In addition, they should understand the components of a complete lean or demand-driven solution, so that they can decide how much functionality they need for their business.

It is also important to take into account the fact that although many enterprise application providers profess lean or demand-driven functionality, most still support pseudo-JIT ways of accommodating mass customization. On the other hand, while enterprise resource planning (ERP) once had a reputation for not supporting lean initiatives, things have changed completely since the advent of real time data collection and monitoring; bills of materials (BOM) and routings management; the ability to map demand accurately and to smooth, plan, and schedule it accordingly; and various operational capabilities, such as real time pull and backflushing. Such functionality has actually placed extended ERP systems in the center of the new lean revolution. Even MRP should not be discounted as useless, since it often can and should be used together with lean or demand-driven practices (e.g., MRP will typically handle planning at the corporate level to determine aggregate loaded capacity, while lean or DBR will deal with the execution). In this regard, user enterprises should understand which products are suited to MRP and which are best for lean or TOC execution. They should select a product line or family that is best suited to lean execution and start their implementation there, rather than wait until the value-stream analysis and mapping are perfect. Successful manufacturers will also start with best physical practices on the shop floor (e.g., a visual system factory) and add technology as they start to expand their implementations.

In today's manufacturing environment, many companies need to support a combination of manufacturing methods, since some products are better suited to traditional resource planning methods (those with high variability of demand or long and unreliable lead times from suppliers) and others are more easily converted to lean manufacturing techniques. JIT or lean works best when items have low variability of demand, when production lines can be dedicated to specific product lines, and when lead time from trading partners is reliable. When different product lines and streams use shared resources, TOC might be needed to consider different product mixes in the round. Poor order shipment performance with late deliveries (within shops whose capacity is wasted on anonymous, forecasted orders, instead of expended on actual orders) might be mitigated by leveraging SMP.

We believe that most organizations require some sort of a hybrid approach to planning and control. For example, an enterprise might use electronic kanbans to pull material through the supply chain and TOC production planning to schedule manufacturing of complex parts, while forecasting is used to generate the high level capacity plan and set up the blanket order or purchase agreements against which material supplies from suppliers are called off. Thus, although most vendors will excel at some particular approach and set of tools, the advantage should be given to those that can cater to various needs (e.g., flow, TOC, MRP, APS, etc.). In addition, vendors that have a business process design tool to optimize the value-adding process, and those that offer a fully integrated total productive maintenance (TPM) solution or analytic solution to support the quest for perfection, should be looked upon favorably.

This concludes Part Seven of a seven part note entitled Lean Manufacturing: A Primer.