Continuous Improvement: Which Methodology Is Best?

Summary

Talk to a consultant of any continuous improvement (CI) methodology, and they will indicate (or come right out and state) that their methodology is better at achieving improvements than any of the others. In consulting with many companies over the years (and reading countless posts on industry websites), I am surprised by the number of companies that define one of these methods as being better than another (in general) or that a CI method of choice is the only one that will ever be required.

Lean is better than the theory of constraint, ISO-9000 is better than Six Sigma and the comparisons go on. But which methodology is better than the others?

The answer is “it depends.”

What it depends on is determined by what you are trying to achieve. In other words, what are the goals manufacturing (or any business really) needs to improve, and how do you want to improve them?

The more common methodologies used in manufacturing include Lean, Six Sigma, ISO-9000 and Theory of Constraints. Each of these methodologies has been considerably successful, and there is ample money to be made in consulting for each. But there have also been some failures with each of them. In my research, I found that these failures are primarily due to improperly implementing the methods or selecting the wrong methodology for what they were trying to accomplish.

A general understanding of these methodologies is required before the aspect of “which is better” can be explored in detail. The following is a brief review of each methodology, looking at the background and the premise by which each methodology works. The descriptions of the following methods are highly simplified but should provide enough explanation for comparison.
 

Understanding lean

The purpose of using the Lean methodology is to determine small and ongoing incremental waste reduction and, therefore, process improvements (as a result of reducing waste) to increase the effectiveness of the manufacturing process and to move toward a production line setup of “one-piece-flow” with no wait times or buffers between operations. As with Six Sigma, the Lean initiative provides the methodology and tools needed for analysis as a foundation for the company’s CI program. The primary concepts of Lean can be summarized into two key goals, voice of the customer and supply the customer on demand.

Voice of the customer. Analyze the processes the company uses to deliver its product or services. Determine the steps within those processes that add value (from the customer’s perspective) and then systematically remove (or reduce) any steps that do not add value and would be considered a waste. The main driver of this concept is to ask the customer, “Is this step something they are willing to pay for?” If it is not, figure out how to remove or minimize it. As a result, any aspect of the company’s operation that does not add value would be considered waste (wasted labor or wasted material). Action would then be taken to eliminate or reduce this waste using predefined methods and tools from the Lean framework.

Example: In manufacturing, the only reason for doing inspections is because of the manufacturer’s inability to produce without errors. As a result, just about any inspection operation would be considered a waste. However, exceptions like aerospace, where customers will pay for inspections, are possible. The Lean initiative would then require that the manufacturer determine the need for inspection (errors being observed) and determine process changes that will prevent those errors. Now that controlled and managed processes are preventing errors, operations can reduce the number of inspections performed to find those errors.

Supply the customer on demand. The product should be available only when the customer asks for it and only in the quantity the customer requests. If the customer has not ordered the product, there is no reason to be working on it. The result of this approach is that, theoretically, a product should not be released to work in process (WIP) unless it is directly linked to a customer order, and it should be made to satisfy the customer order as quickly and reliably as possible. The target process would not need buffering, and there would be zero waste from scrap, rework/repair or excess inventory.

To move toward this goal, the company removes the need for buffering WIP, for example, having a balanced line with all operations having approximately the same cycle time and staying within the estimated takt time needed to deliver the product when expected. The company also removes the possibility of lost time because of equipment failure and manages line changeover to new products, which includes equipment setup times that target single-minute changeovers. If equipment is not needed for a product run, it is to sit idle, be available for maintenance or be used by a different product during that time. At no time should a product be worked on anywhere in the system if it is not directly tied to a customer order.

Example: Under this concept, any production schedule used to move product to finished goods inventory that is not already associated with a customer order would be considered waste. Also, under this concept, a company should reduce the need for maintaining finished goods inventory unless a customer demands it.

When examined from a manufacturing perspective, an issue with Lean is that it is arguably not suitable for some industries. For example, when looking at a seasonal product like lawn chairs, a single company is not likely to have the facilities to provide the market with enough product to satisfy an on-demand manufacturing delivery for an entire season. Companies that make lawn chairs must estimate the demand well in advance and start production well before a single order is received. In this scenario, using some Lean concepts would probably not be as fruitful. It could be argued that sales could locate a different market that would provide a more consistent product flow, but as lawn chairs are a highly local market, this is not likely to be of much benefit to the manufacturing floor.
 

Understanding Six Sigma

The purpose of using the Six Sigma methodology is to identify process improvements that will invoke a major step in the effectiveness of the manufacturing process. The primary concept of Six Sigma is to use a rigorous set of process management and analysis tools to drive all the characteristics of a process (and product performance) as close to the target characteristics as possible.

For example, a curing oven may have an acceptable temperature range. With Six Sigma, it is not sufficient to maintain the temperature anywhere within the acceptable range. Six Sigma would require understanding the correct target temperature and would perform a series of analyses to determine why the oven temperature drifts away from the target temperature. After identifying the causes, an effort will be made to determine which cause has the greatest impact on the drift. Then, a change would be made to the oven or process (depending on the cause) to minimize the drift as much as economically possible.

Finally, once the process was normalized (implemented as part of normal production without unexpected deviations), it would be reanalyzed to ensure that the implemented change achieved the expected performance improvement.

As Six Sigma looks at process capability from a target perspective, thereby driving toward achieving Six Sigma capability (i.e., a quality level of 3.4 defects per million opportunities), it is frequently used to determine and fix issues to achieve jumps in process capability. Whereas with Lean, the expectation is to achieve smaller incremental steps in removing process waste.
 

Understanding ISO-9000

Where Lean and Six Sigma are focused on analyzing and improving the actual processes, ISO-9000 is a management methodology. The purpose of ISO-9000 is to provide a management model for companies to use to establish a program for managing CI itself. Based on the concept of “say what you do and do what you say,” ISO-9000 looks to have a company progressively define all the processes within manufacturing and use process auditing to ensure that these processes are followed. In addition, ISO-9000 requires that the effectiveness of defined processes have a means of being measured (it does not explain what the measurements are to be) and that, according to specified measurements, the processes are improving over time.

ISO-9000 requires a company to develop business and management-level processes to ensure improvements are made continuously. It does not define a specific analysis and improvement methodology; however, it does require one.

Understanding Theory of Constraints

hen looking at the Theory of Constraints (TOC) strictly from a manufacturing perspective, it is another operational methodology with a framework of procedures and tools used to help operations clarify the goals of a process, define the limiting factors of that process (called the constraint) and then take two distinct and concurrent directions: Maximize the throughput of the constraint through various actions; and make process improvements that increase the throughput of the constraint.

The primary driver of the TOC is that within any process (or system), there is always something that is a constraint or a limiting factor. The constraint within a manufacturing process defines the throughput of the entire process.

The key understanding of the TOC is that regardless of the activities that occur elsewhere in the process for improvement, improving the throughput of the constraining operation is the only way to improve the throughput of a line. Furthermore, if time is lost at the constraining operation from downtime or improper scheduling, that time is not recoverable. As a result, effort should be made to schedule material to the floor and any preceding operations to the constraint to ensure that the constraining operation is scheduled and loaded to its most efficient capability.

The flow rate through the constraint should be the driver of all other activity (referred to as the drumbeat to which production follows). Any preceding operations should be scheduled to ensure enough of a buffer in front of the constraint so it can continue if other operations are shut down (e.g., due to scheduling demands).
Additionally, only enough material that can be processed through the constraint in a scheduled period should be pulled onto the floor. This process is like a rope pulling the material as needed. All of this has given rise to the term drum-buffer-rope to describe TOC scheduling concepts.

If there is a need to improve the production line throughput and/or the revenue generated from a production line, action must be taken to improve the constraint.

Now that we have a basic understanding of these methodologies, which is better? The answer is “none of them.” It depends on what the company is trying to achieve. Additionally, they will likely need to use each methodology at different times within a company’s improvement cycles.

Further reading on continuous improvement

For more discussions on manufacturing analysis topics, see the ISA book that Vokey coauthored with Tom Seubert: “Manufacturing Execution Systems: An Operations Management Approach,” Second Edition. Vokey’s latest ISA book, “CoE: The Key to Data-Driven Manufacturing,” is also available.
 

Using the methodologies

Using the following manufacturing scenario, I will show how each method—and combinations of methods—can be used to address throughput and cost issues. The details of the actual product or the particulars of the production line are not required. For the sake of attempting to be brief, many of the concepts presented have been highly simplified, and the operations discussed are not representative of any industry.

Here’s the scenario: A discrete manufacturing company has a production line with a takt time of 60 seconds to meet delivery requirements (i.e., one production unit is required off the line every 60 seconds to meet scheduled customer demands). The actual throughput for the line varies from a product every 55 to 70 seconds, with the line averaging 68 seconds (the latter two values being well above the required takt time). The production line has a first-pass yield (FPY) of 75% (i.e., 25% of products require extra work to be usable). It also has a scrap rate of 10% (10% of production cannot be used and is written off), meaning that 90 out of 100 units processed by manufacturing will be sold (the 10% scrap is part of the 25% that requires extra work). This could make it harder for the company to achieve a market price point.

Applying ISO-9000. The first issue to be looked at is the effectiveness of the company’s CI program. In this case, assume that the company has an ad hoc CI program based on reviewing customer complaints and fixing the problems the customers complain about. Historically (according to operations management training programs) only about 1 in 8 customers will complain about a problem. Most will just take their business to another manufacturer (that is a lot of lost customers before a problem is recognized or even corrected), which makes for an ineffective strategy for CI management. Another concern about the ad hoc strategy is that many of the issues that cause the line to miss its takt time go unresolved because the customer only sees the products that go out the door, and most problems causing variation in meeting takt time are only visible internally.

To help start the manufacturing floor on the right path to properly managing its CI program, the company should implement ISO-9000. This would require defining the processes (including a means of measuring the processes) and ensuring that these processes are being followed. Each of the other methodologies also requires measuring the manufacturing processes in one manner or another, but ISO-9000 is used to define a program of CI management. Although there is a requirement to measure and improve the manufacturing processes, ISO-9000 does not explain how the processes are to be analyzed, only that they are. In addition, process auditing ensures that processes are defined and followed (this is process standardization).

Through analysis, operations will map out the process of making a product, specifically define each step and train production operators to follow the defined process. This will reduce the amount of common cause variation within the process, making it easier to determine the reason for special cause variation. The production line improves process capability by determining and correcting special cause variation.

Standardization may or may not improve the process capability. If standardizing may not actually improve the process, then why standardize? One of the hardest types of problems to fix within a process is those that happen intermittently. When a process has many factors causing variation, there is considerable fluctuation in the performance measurement results. Certain variances that could benefit or deteriorate a process may occur in one production run and not occur for the next couple of production runs. This makes it very hard to analyze the analysis results. After processes have been standardized, there are fewer variances to track, fewer interactions between variances and—most importantly—more consistent results. This makes it much easier to determine and control the root cause of each variance.

The company has implemented ISO-9000, standardized most of the manufacturing-level processes and is now certified. Actions are being taken to define, monitor and correct process deviations and reduce process variation. The company is well on its way to properly managing its CI program (this is highly simplified). What does the company do next? As a result of the process standardization that was implemented with ISO-9000, the company has probably seen some improvement in process capability. Although it is not a certainty, it is not uncommon to see improvements in some process key performance indicators (KPIs). Looking back at the scenario above, the company had an FPY of 75%. Achieving a 10% improvement would bring the FPY up to 82.5%.

However, from the perspective of manufacturing effectiveness (efficiency in particular), although FPY is important, it is not the only concern. Improving manufacturing efficiency also means improving the cost of manufacturing, which is frequently measured by “cost per unit of production.” Lean can be implemented to address this issue.

To understand the aspect of the “cost of manufacturing,” here is a simplified example of the cost of the scenario above. Assume that there are five production operations in manufacturing the product, a total direct labor cost of $50 and a total material cost of $25 (both per unit of production). Each operation consumes approximately 20% of the total cost of labor and material, which brings the example scenario to a total manufacturing cost of $75 per unit of production and consumes $15 of that cost per operation (again, the costing is highly simplified).

Applying Lean. Having documented all the manufacturing processes as part of their ISO-9000 certification and used ISO-9000’s requirement for internal process auditing to ensure they are being followed, the operations staff now has better visibility of all the process operations and will likely have found some operations (or substeps) that were not expected. Lean’s focus is to reduce waste within a process to make production more cost-efficient. Hence, the company decided to implement Lean.

In this case, the processes will be analyzed to determine which operations are required (according to the customer) and which are redundant or unnecessary. An example of an unnecessary step might be an “end-of-line” inspection operation that was implemented to ensure that a once-common mistake is no longer happening. This inspection may have been implemented while reviewing the overall process to find the actual cause of the mistake. Then, after the process was updated to ensure the mistake can’t happen, the inspection was not removed. This review of operations in Lean is called value engineering, and the methods used in value engineering are specifically defined within the Lean methodology.

Assume that the company has now performed several value engineering events and has improved the effectiveness of the processes. As part of the value engineering, the manufacturing engineers would reevaluate the “end-of-line” inspection. Recognizing that under the Lean mindset, inspections, in general, are non-value-add (they are only used because of a lack of capability in manufacturing quality), one activity would be to determine whether the inspection is still needed.

If the problem still exists, the engineers will determine the root cause and update the process to reduce occurrences or ensure that it can’t happen. The inspection operation is then revisited, and if it is no longer needed, it can be formally removed. In many companies, these end-of-line inspection operations are frequently continued long after the problem has been fixed because of a “that’s the process” mindset. Assume that the problem was properly closed and the inspection operation was removed from the process. Looking at the cost of manufacturing, we can recognize the improvements.

In the original process of five operations and $15 consumption per operation, the manufacturing cost can be updated to a total cost of approximately $60 ($75 - $15 = $60). Keep in mind that this is a simplified example.

Also, through these Lean analysis events, let’s say changes have achieved an average operational cycle time of 63 seconds (ranging from 58 to 65 seconds) and improved FPY to 85%. In addition, the company is still using ISO-9000 to define and improve managing the CI program overall.

Once these Lean events have stabilized and the processes have become well understood, it is time to engage in more in-depth analysis to determine what is causing the remaining fallout. The methodology used from this point on depends highly on the kind of problems that are found. There may be process-related issues that are detrimental to the product quality.

An example is the damage in electronics manufacturing that used to happen due to static electric shocks to the product from people handling the production unit without static control. Using proper static control procedures is now common throughout the electronics manufacturing industry. For these types of problems, it is common to use Lean methods to determine what is causing the waste resulting from additional need for repair and rework.

Applying Six Sigma. Other types of problems may result from process parameters that drift because of instability in the controls or wear from extended use of the equipment. Although some Lean methods may still apply, it may be better to resolve these issues using Six Sigma.

The process will be analyzed using Six Sigma methods to determine the normal distribution of the major factors that cause variances. With Six Sigma, there is a specific target value for process/product performance and a need to understand which parts of the product or process (including equipment settings) have the most influence in maintaining that target as well as process parameters that cause changes from one production run to the next (known as “Gage R&R”). By controlling the parameters that cause variation within a production run and from one production run to the next, Six Sigma tries to ensure a process is both repeatable and reproducible (the R&R part from above). This may involve tighter control of parameters, updates in equipment or changes in the product (material use or design influences).

Six Sigma’s DMAIC (define, measure, analyze, improve, control) analysis process is specifically developed to investigate these problems. From this point on, process management and a company’s CI program should isolate specific problems and perform in-depth analysis into the root cause of these problems.

In this case, I want to emphasize the importance of investigating specific problems. At any one time, a production line should be focused on determining, isolating and resolving one issue at a time. Many times, a production line will try to resolve multiple issues on seemingly disparate problems only to have either the analysis or the solution testing of one problem interfere with the analysis or solution testing of another problem. This can cause skewed analysis or testing results and skewed or downright wrong decisions being made in CI management.

But why should a production line engage in this level of analysis? An easy solution to achieving the required takt time would be simply increasing the amount of people or equipment needed to get product out the door. The problem with increasing the number of people or equipment is that it adds cost to the process of making a product and the potential of creating more variation (possibly making the situation even worse). Reducing variances using methods like Lean and/or Six Sigma is the only way to ensure that production costs are reduced.

Applying the Theory of Constraints. In Lean, one of the primary directions that consultants give is to use a line set up based on “one-piece flow,” as discussed in the Lean overview earlier. This requires all the operations for a particular product to be available within a production line and that all operations be in close physical proximity to allow for efficiently moving production units between operations one unit at a time. In addition, all operations in the line setup must be very close in cycle times. The concern over this kind of setup is that it assumes that production volumes are high enough to need continuous production or that a production line can sit idle during low-demand periods.

An alternate concept could be to find several low-volume products that share many production operations and resources, which allows them to be manufactured on the same production line with minimal line change effort. This would enable a one-piece-flow setup, and switching from one product to another for scheduling would be easier due to the similarities in production processes.

In many production environments, this is not always possible. When these production line configuration requirements do not apply, Lean’s one-piece-flow setups cannot be used effectively.

When multiple production lines share only a couple of high-cost resources, or if operations within a single production line have significant differences in cycle times, there will always be one constraining operation. In this case, TOC is the best line management methodology. The key driver of line management is to recognize and properly manage the constraining operation. If time is lost at the constraining operation (e.g., due to equipment failure that causes the constraining operation to stop), this time is permanently lost and cannot be recovered.

When a production line uses TOC, the CI initiatives must address the priority of what must be improved and the goal for improvement. If the goal is to reduce waste or cost, Lean may be the preferred methodology. However, if the goal is to improve the line throughput or increase revenue, TOC must come into the picture. A couple of improvement strategies that can be used are: Reduce the cycle time of the constraining operation (increasing the number of production units that the constraining operation can process); or reduce the failure rate of the products that go through the constraining operation (increasing the rate of production units that successfully complete the constraining operations and all operations after it).

It is important to recognize that using the above two strategies on a nonconstraining operation will not improve throughput regardless of the line setup. Even in a one-piece-flow environment, this concern may still hold true as there will still be a marginal constraining operation (the operation with a cycle time closest to the planned takt time).

As part of an initiative to decrease the cycle time of a one-piece-flow environment, there needs to be a review of the original takt time the line was designed to achieve. The review would likely result in redistributing the activities in each operation and possibly adding or removing operations to achieve the takt time. It is important to understand that in one-piece flow, you can increase the number of operations (thereby increasing the full process cycle time) and maintain the takt time if each operation’s cycle time is kept at or below the takt time. It will take longer for the first unit to come off the production line, but the takt time will be maintained for each additional production unit.

Using TOC provides focus on the critical operation that determines production throughput and the overall production schedules. It also focuses process improvement and product improvement on problems that affect that critical operation’s throughput.
 

Final thoughts

Returning to the original question, “Which CI methodology is best?” There is no simple answer. In looking at CI from a production perspective, it is important to understand the improvement goal (what you are trying to improve.) Only when you know this can you determine which CI methodology will be best for that specific goal. Each methodology can contribute to implementing an efficient, cost-effective and profitable manufacturing process.

ISA-9000 is used to create an effective CI program, an essential component of a successful manufacturing company.

Lean analysis improves the cost of manufacturing by determining which operations are required and which are redundant or unnecessary. Lean also directs focus to CI procedures for selecting improvements that affect production costs. Lean analysis can be a key component of an effective CI program.

Six Sigma analysis identifies the parameters that have the greatest effect on variation within a production run and from one production run to the next. This method is used to ensure that a process delivers output as close to the target as possible, which helps ensure that the process is both repeatable and reproducible. This also is a key to an effective CI program.

Theory of Constraints (TOC) focuses on critical operations that determine production throughput and overall production schedules. It also focuses process and product improvement on problems that affect throughput.

_{This feature originally appeared in the November 2024 issue of AUTOMATION 2024.}