#15 수행 영향 분석하기 - Analyzing What's Affecting Performance
In This Chapter
▶ Finding out what's at fault
▶ Using data to prove the point
▶ Introducing the maths of Lean Six Sigma
Whether you manage a day-by-day operation or are involved in DMAIC improvement project, you need to understand what factors can affect performance, especially if you encounter problems in meeting your customers' requirements. In this chapter, I introduce a selection of tools and techniques to help you identify the 'guilty parties'. We focus on how and how well the work gets done - the process and data.
Generating your list of suspects
To find the guilty party, you generate a list of possible causes, check out each possible causes and gradually narrow down the list.
Creating a cause and effect diagram
The fishbone, or cause and effect, diagram was developed by Dr. Ishikawa and provides a useful way of grouping and presenting ideas arising from a brainstorming session.
The head of the fish contains a question that describes the effect you are investigating (make sure you choose a narrowly focused question, or you will end up with whalebone!) "What are the possible causes of delays in delivering customer orders?" or "Why are there so many errors in our invoices?" You can group the possible brainstormed causes under whatever headings you choose. In the below figure, we use the traditional headings of People, Material, Equipment, Method, and Environment. You may find these headings useful in prompting ideas during the brainstorming session, but be aware that they can also inhibit more lateral thinking.
The team comes up with their idea on the possible causes, writing the ideas on the sticky notes so that you can move them around easily during the subsequent sorting process.
Place your major cause headings on the left-hand side of the diagram, forming the main 'bones' of the fish. The brainstormed ideas from the smaller bones. For each possible cause, ask the question 'why do we think this is a possible cause?' and list the responses as smaller bones coming off the main cause. You may have to ask 'why?' several times to identify the probable reason, though you might still need to validate this with the data.
Investigating an interrelationship diagram
Using an interrelationship diagram helps you identify the key drivers behind the effect that you are investigating in your fishbone diagram. We covered the interrelationship diagram in the previous chapter (#2 린 식스시그마의 원리 이해 - Principles), where we show how you can use it with an affinity diagram - a really useful way of helping you get to the root cause, the key driver of the problem you're addressing.
Investigating the suspects and getting the facts
Managing by fact is vital, so validating the possible causes highlighted by the interrelationship diagram, is the next step. To validate the causes, you may need to observe the process and go to the Gemba (#2 린 식스시그마의 원리 이해 - Principles), or check out the data to see whether they confirm your suspicions.
A SIPOC diagram (#3 당신의 고객은 누구인가? - Identifying Your Customers) provides an ideal framework to help you think about all your process measures and now you need to pull together a set of X measures, if you don't already have them. A range of X variables will be coming into your process - the input variables. These input variables affect the performance of the Ys, and may include the volume of activities, for example, the number and type of new orders. The input variables may well concern the performance of your suppliers, too, perhaps in terms of the level of accuracy, completeness and timeliness of the various items being sent to you.
A range of X variables will exist in the process itself - the in-process variables. Here, your deployment flowchart or Value Stream Map(#8 VSM(Value Stream Map) 만들기 - Constructing a value stream map) can help you highlight the potential Xs, including activity and cycle times, level of rework, the availability of people, or machine downtime, for example. Again these Xs will affect your performance. As you identify the X measures you need, you're building a balance of measures to help you manage your process.
Providing your point
When you think you know the cause of the problem in your process, you may need to provide some evidence to back it up. For example, your boss may think she knows the answer, but you may find something different as the result of your careful analysis of the facts.
The below figure shows a simple matrix to show how the various snippets of evidence match against the suspects:
- Remember that correlation may not mean causation.
- One hundred percent certainty is impossible.
- More analysis is almost always possible.
- Fear of being wrong.
- Beware of analysis paralysis.
Using a scatter diagram (scatter plot) can help you strengthen your case. A scatter diagram helps you identify whether a potential relationship or correlation exists between two variables and enables you to give a value to and quantify that relationship. The variables are the cause and effect - X and Y. You can use this method to verify potential root causes of a problem or, for example, to validate the relationship between your input and in-process measures against your output measures.
The dependent Y variable is always plotted on the vertical axis; the independent X variable is plotted on the horizontal axis.
In the below figure, the first one shows a relationship that seems to exist between speed and error rate - the faster we do it, the more errors we get.
The second example shows a negative correlation - the values of Y decrease as the values of X increase and, in doing so, appear to confirm that our theory investment in training leads to reduced error rates.
In the third example in the below figure no correlation exists, so our theory doesn't hold. We need to make sure that the data has been segmented. (For segmenting data, please see the previous chapter: #3 당신의 고객은 누구인가? - Identifying Your Customers)
Assessing your effectiveness
Several lean measures are available to help you understand your performance and the scale of improvement needed, including Takt time and overall process effectiveness, and overall equipment effectiveness.
Taking Takt time into account
Takt time tells you how quickly you need to take action in relation to customer demand. Takt is a German for a precise interval of time such as a musical meter. It serves as the rhythm or beat of the process - the frequency at which a product or service must be completed in order to meet customer needs.
Here you can find the formula of Takt time below.
- Takt Time(TT) = The available work time (per shift)/ The number of customer orders (per shift)
The available time is dependent on how many resources are available. It represents the number of working hours in a day or shift. If a widget factory operates 480 minutes per day and customers demand 100 widgets per day. Takt time is 288 seconds, or 4.8 minutes, as shown in the below figure. If the demand is 240 widgets, the Takt time would be two minutes.
Similarly, if the customer wants two new products per month, Takt time is two weeks.
Recognizing the effect of rework is important because it effectively reduces the Takt time in direct proportion. So, imagine that in the example, a 10 percent error rate exists in the first-pass output of the work, though this is picked up and corrected. In effect, this increases the number of customer requests from 100 to 110, the available minutes are unchanged at 480 minutes, but the impact on Takt time is to effectively make it shorter, at 4.36 minutes. Takt time will effectively be shorter still if we have second-pass corrections to deal with. So, the actual Takt time might be still 480, but the rework means you have less time than that in practice.
For example, using the formula:
- Takt Time(TT) = The available work time (per shift)/ The number of customer orders (per shift)
- You have 100 customer requests each working day, where you have an 8-hour shift for 10 people.
- The number of people isn't a factor in calculating Takt time, so:
- 8 hours X 60 minutes = 480 minutes
- 480 divided by 100 (customer requests) = 4.80 Takt time
- Even if there were 20 people, the Takt time would still be 4.80.
- It's the production rate needed to meet the demand.
Incidentally, Toyota typically reviews the Takt time for a process every month, with a tweaking review every ten days.
Clearly, an important relationship exists between Takt time, cycle time, and activity time. If the Takt time is less than the cycle time you have a problem, which must be tackled immediately, ideally using DMAIC. Removing waste may well be part of the solution; preventing it in the first place might be another.
When the Takt time equals cycle time perfect flow exists, but too often the flow isn't balanced. This situation can cause 'bottlenecks' that disrupt your ability to meet customer demand. The below figure shows the dilemma faced by a line experiencing bottlenecks.
In order to meet the Takt time, the level of non-value-added activities will need to be addressed, but a better balance will be required, too.
Considering overall process and equipment effectiveness
In analyzing your performance, you may want to put in place some additional measures, such as Overall Process Effectiveness(OPE) in transactional processes and Overall Equipment Effectiveness(OEE) in manufacturing. Each of these summary measures has three components:
- The availability rate measures the downtime losses from equipment failures and adjustments as a percentage of the defined and scheduled time.
- The performance rate measures operating speed losses - running at speeds lower than design speed and stoppages lasting for brief periods as agreed.
- The quality rate expresses losses resulting from scrap and reworks as a percentage of total parts run.
These elements are multiplied together, where
- OEE = Availability X Performance X Quality
So, with Availability at 90 percent, Performance at 95 percent, and Quality at 99 percent,
- OEE = 0.90 X 0.95 X 0.99 = 84.6 %
Here, you take the following three elements for service organizations and transactional processes.
- A = Availability of equipment
- P = Productivity
- Q = Quality rate
※ 참조: 본문은 책, "Dummy들을 위한 린 식스시그마(Lean Six Sigma)"와 The Toyota way를 정리한 내용을 기반으로 제 경험들을 조미한 글입니다. 현재는 원서 기반으로 내용을 정리하고 있는 중이며, 정리가 완료되면 한글로 번역을 추가할 예정입니다.