Can Automation Be a Lean Tool?
Automation that enhances flow is lean; automation that reduces uptime and extends changeover is not lean
By Rick Harris
Chris Harris, DBA
VP of Operations
Harris Lean Systems Inc.
Murrells Inlet, SC
Web site: www.harrisleansystems.com
It's not a question of whether lean is manual or not—effective lean production systems use both manual and automated processes—the task is to determine the appropriate type of automation. Taichi Ohno, the co-inventor of the Toyota Production System, said long ago that there are two sure facts about forecasts: they are almost never correct, and they always change. With this thought in mind, machines need to be developed so that the lean production system can have increments of capacity without a large capital expenditure. In other words, how can automation be developed within the production system to react to changing customer demand?
The first area that needs to be addressed when designing or specifying a machine for lean production is the machine itself. What must be automated to meet customer demand? There really can be no argument that there are some things that must be produced by somewhat automated processes, for example, items with very tight tolerances, small electronic components, parts that are much too heavy to handle manually, or medical products that just cannot be touched by human hands.
The question for engineers in lean production environments is, first: what absolutely must be automated? The next question is: what in the process does not have to be automated? Answering these two questions objectively can sometimes be a challenge, as old paradigms inhibit creativity.
When it's decided which parts of the process must be automated, the different levels of automation need to be considered. Within lean production there are five levels of automation that we normally deal with. Using these levels of automation, we try to develop the right level of automation—the right machine—for the right process.
As different levels of automation need to be considered, discovering how many functions the machine has to perform is also important. Does the automation have to be in one machine, or can it be spread over multiple machines? Can the machine perform only one function, or should it perform many different functions? What are the benefits of each approach?
Discussing automation strategy requires, first, that we understand the five different levels of automation that usually exist in industry.
In the first level, everything is done manually. The operator loads the machine and starts the machine, and the machine cycles. Next the operator unloads the part and manually transfers it to the next production step. An example of this machine arrangement would be a manual press in which the operator loaded the press, pressed the part, unloaded the press, and took the part to the next station.
The second level of automation is when the operator manually loads the machine, the machine automatically cycles, and the operator manually removes the part and takes it to the next station. In level three automation, the operator manually loads the part into the machine, and the part automatically cycles. The part is automatically unloaded from the machine (or at least provides an empty nest for the operator to load the machine again when he or she arrives), and the operator then moves the part.
Level four automation automatically loads the part; it's automatically cycled, automatically unloaded, and then manually transferred to the next process. Finally, level five automation is entirely automatic. The machine is automatically loaded, cycled, and unloaded, and the part is transferred by automation. Though there are some processes that must be done utilizing level five automation, we see many examples of level five automation used in production facilities that do not require level five automation.
There is a great divide between level three and level four automation. This divide represents MONEY in the form of maintenance costs, engineering costs, costs of the machine, etc. When making the jump to level four automation, where the machine is automatically loaded, automatically cycled, automatically unloaded, and manually transferred, cost often increases while flexibility can decrease.
A level-three piece of equipment can and does run with about 95% uptime, while level four will likely run at 70–75% uptime, and level five equipment will likely run with uptime in the 65–70% range. There is a gradual decline in the amount of uptime that a machine will likely have as it becomes more automated. Changeover is also an area in which the change from level three automation to level four or five is impacted. In levels one, two, and three the changeover time is often much less than level four or five. A very desirable machine attribute is that it can changeover in one takt time (customer demand rate) cycle. It tends to be much easier to accomplish this task within the first three levels of automation versus the latter two. Remember, the lower the changeover time, the less inventory the company needs to carry.
Another point that we have noticed over our years of implementation has to do with manpower and machine design. It is not uncommon that when a level five machine is developed to eliminate the need for a production associate, the result is the need to hire a maintenance technician and an engineer to constantly tend to the machine. This tradeoff is often harmful to the company, because the production associate replaced by the level five equipment costs the company much less than the engineer and maintenance technician who must deal with the new equipment.
We've found that level three automation (when we have that option) tends to fit better into a lean production system using production cells, because in many instances operators find it relatively easy to load and transfer the workpiece, thus the expense of going to a level four or five machine is sometimes not worth the result. Also, level three automation tends to have higher uptimes and quicker changeover times than levels four and five. This is because level three tends to employ simpler machines that are dedicated to a single task—which leads us into the next part of lean automation design, machine functionality.
Let's say that a company has developed a new widget, and it has become popular. The demand for this widget seems to be stable, so the company decides to develop a production system to produce the product. They first begin to study the machines that make the widget's parts, and are successful in finding small machines that have a very high uptime and short changeover. Because these machines are dedicated to a single task, the company is able to achieve a high level of uptime and quick changeover.
As the company continues to think about its production system, however, engineers and managers begin to believe that if the machine could do more, they could use less labor to make their widget. They contact the producers of the widget machines, and begin to ask if the discrete widget machines can be combined into one piece of equipment to do multiple other tasks. They ask if one machine can weld, drill, and tap if various robots are used, and if the equipment can then place the widget in a box and seal the box closed. In essence, the widget company has now built their production system around a machine that does many different things. They have taken various pieces of equipment that have high levels of uptime and short changeovers, and folded all of them into one machine that does not have that high level of uptime and short changeover.
This example details the easiest choice of automation for many lean production processes, but maybe not the right choice. This production system is built around a machine. But a lean production system is created to flow material and products. Machines are added to enhance that flow, or because they are absolutely necessary to create the product.
The second reason that this is not the ideal choice of automation for a lean production system is that it hampers the flexibility that the company needs to react to customer demand.
Another factor that usually enters the equation is uptime. Many machines are sold on a guarantee of 85.95% uptime. This specification is common because it is difficult to get capital for a machine that is going to run 65% of the time. However, many facilities receive a piece of equipment expected to run 95% of the time that actually runs about 75% of the time. Why does this happen? There are probably many reasons, but the most likely is because these are highly automated, unique, multifunctional (does many process steps) pieces of equipment. In the example above, the widget company took the simple widget-production machines that had high uptime and short changeover, and combined them until the uptime and changeover suffered. This leads into the heart of automation in a lean production system.
The three problems that the company had with the widget machine were uptime, changeover, and flexibility. There is a way, through looking at different forms of automation, to attack these problems. If the company were to adopt a philosophy of employing many different simple single-purpose machines with the right levels of automation, they would likely have higher uptime, quicker changeovers, and more flexibility.
Lean production systems are not always manual, and lean experts are not opposed to the use of automation. Lean systems are, in fact, well-suited to the right type and level of automation.
Determining the level of automation that will work best in a production system is a good place to begin when designing a production process. It's important to remember that flexibility is a key ingredient when you're trying to compete in a global market. When designing automation, it is important to keep this in mind, and not design machines solely based upon the future forecasted demand, because we know that the forecast is not likely to be correct, and will probably change.
The next shift in thinking required when designing automation for lean production is that the complexity and simplicity of machines must be considered. Complexity is not always the right answer. A facility's flexibility can be greatly hampered by a machine that performs many different functions. In this instance, often a company has built its production system around the machine, and has not developed a system to flow production. We have seen many companies face this predicament, and they end up succeeding, or failing, solely because of the effectiveness of a machine that has low uptime and long changeover times.
Unfortunately, it's our experience that it is difficult to find machine builders that are interested in building these simple pieces of equipment. Machines in today's fiercely competitive global market need to have high uptime and be very flexible, and that requirement calls for machines that are fairly simple. Lean producers throughout the world would likely benefit from a leader in lean-machine building that understood the philosophy behind machine design for lean production systems, and welcomed the opportunity to build them. Too many times, a simple machine design with high uptime and quick changeovers is changed because multifunctional machines promise high efficiency and high uptime. Adopting such equipment can, of course, lead to long lead times, downtime, and frustration for the personnel attempting to implement lean production systems.
There is no doubt that the right level of automation has a place in lean production, but understanding the impact of the various forms of automation and machine design on a lean production system is imperative to creating a flexible, efficient, world-class production system. A lean production system should be designed to flow, and automation should be selected after deciding how best to improve flow and fit into the flow. Lean is NOT manual, but the right type of automation is required.
Simple, Flexible Automation
In this example, the widget maker can react to demand very quickly by adding production associates into the system or taking them out of the system. In its current state, for example, one person could run all three workstations and still meet the customer demand rate. On the other end of the spectrum, if the customer demands more products and the company must produce them faster, say every 10 sec, two more people could be added into the system. This flexibility can be quick, immediate, and fairly inexpensive when compared to a system built around a multifunctional machine, where the only way to increase capacity is to buy another machine.
As an added bonus, the uptime in these three machines or workstations is likely to be high and changeover times quicker. This type of thought process when dealing with machine design gives the company the stability and flexibility that it likely needs to react to changing customer demand.
Automated Widget Production
In this illustration, it looks as though everything is fine. The customer demand rate (takt time) is 30 sec, and the total machine time is 25 sec for one complete widget. However, the demand may not stay stable for this product, and the uptime for this machine is likely not going to be what was originally estimated.
The demand currently can be made, but what if the product becomes more popular and/or the economy booms? What can the company do to meet the increased production requirement? Even though the company can currently meet customer demand, what if demand goes to 24 sec per widget? In this case, the company would have to buy a new machine. This machine is not inexpensive because it is one of a kind—a very sophisticated piece of equipment. Consequently, the company would have to spend a great deal of money on a machine that would likely produce very little volume.
This article was first published in the August 2008 edition of Manufacturing Engineering magazine.