6.6 Lean Manufacturing Systems (LMS)
6.6.3 Six Core Methods to Implement Lean Manufacturing
This section is devoted to the presentation of the most popular approaches used to implement lean manufacturing systems. They are used in the order of their presen- tation. Some of these methods can be implemented concurrently. Note also that numerous other methods exist; we have restricted ourselves to the most popular ones.
6.6.3.1 Cellular/Flow Manufacturing
This approach can be defined as the cooperation between automated and manual operations to maximize the value of the products while minimizing waste.
The most efficient cooperation includes the concept of process balancing (also known as cell balancing or line balancing, a line being a sequence of cells). In this kind of organization, workers are often responsible for several manual operations and/or automated operations. The assignment of operations to cells and, inside each cell, the assignment of operations to workers or robots is done is such a way that all resources have about the same workload in any situation. This guarantees the flexibility of the system and a smooth flow of items through the manufacturing system with minimal handling and transportation.
The consequences are elimination of waste, reduction of WIP and floor-space, shorter manufacturing cycle times and low production costs.
Most publications on lean manufacturing concern the management aspects: 5S, Kaizen, just-in-time (JIT), Total productive maintenance (TPM), the six sigma method, SMED, etc. Only a few publications concern the design of lean manufac- turing processes. Nevertheless, this is a crucial aspect for the success of a lean ap- proach. Indeed, most of the waste can be eliminated at this step of manufacturing system life cycle by revealing and solving problems before production.
The following schema for the design of a lean manufacturing process is sug- gested.
Takt Time Calculation
“Takt Time” imposes the rate of fabrication needed to meet customer demand. It is equal to the available work time divided by the customer demand.
demand Customer
time Operating
= Time"
Takt
"
Time Effective Shifts
= Time
Operating ×
“Effective time” takes into account the shifts worked and makes allowances for stoppages, for predetermined maintenance, and planned team briefings, breaks, etc. The customer demand includes the anticipated average sales rate plus any ex- tras such as spare parts, anticipated rework and defective pieces.
Definition of Tasks (Operations)
Definition of tasks (operations) is the next step. The operations that should be con- sidered are those with added value, i.e., the operation where the product value af- ter the operation is greater than the product value before. For example, holding
222 6 X-manufacturing Systems and transport operations have no added value and should be reduced (or elimi- nated if possible).
Choice of Process Plan
Choice of a process plan consists in analyzing several possible process plans and choosing the one that minimizes cost and production cycle. Remember that a process plan is defined by a set of operations, their sequence and the correspond- ing tools necessary to execute these operations. Usually, there are a great number of possible process plans. A graph theory approach can help manufacturers to ana- lyze possible plans and optimize a given criterion.
a
b
c
f g
d
e j i
h
l k
m 5
2
4
1 3 2 2
3
9 5
4
10 8
n 5
Figure 6.14 An example of a graph for manufacturing process analysis
Figure 6.14 is an example of a graph approach (Dolgui and Proth, 2006). In this graph the arcs represent operations and vertices precedence constraints. The letters are the identities of the operations and the numbers are the operation durations. So to calculate the production cycle, it is necessary to search for the critical path in this graph. To reduce the production cycle, an aforementioned objective of the lean approach, it is necessary to reduce the durations of the tasks on this critical path (or even eliminate some of them). Note that after reducing the operation time or eliminating a task belonging to the critical path, the critical path may change and other operations form a new critical path. The durations of these new opera- tions should be also reduced, and so on. In addition, if the approach is used for the choice of the best process plan, this type of graph should be designed for each po- tentially interesting process plan.
To reduce the number of operations on the critical path, some additional studies can be made to alter the product. In addition, reductions can also be obtained by improving the manufacturing equipment and production methods.
The only difficulty of this approach is the time-consuming data preparation.
Since all product data are known, the graph approach to calculate the critical path is relatively simple for computer realization and holds promise. Nevertheless, this requires advance acquaintance with manufacturing technologies, equipment and processes to adequately model the possible process plans via graphs.
The critical path for the manufacturing operation is a good technique to reduce the production cycle. Nevertheless, in a production cycle there exist also non- productive operations. Therefore, additional analysis is necessary to identify them.
An adequate support for this is the lead-time analysis chart (see for example, Dol- gui and Proth, 2006).
Line Balancing (Assignment of Tasks to Workstations)
After the operations and process plan are chosen, the next step is line balancing.
At the line-balancing step, the assignment of operations to workstations is made under the takt time, precedence and other constraints. The load of each station cannot exceed the takt time.
Equipment Selection
For any given line-balancing method and for each work station several possible configurations of equipment exist in general. Therefore, this procedure of select- ing the appropriate equipment for each workstation can reduce the total time and cost considerably.
Simulation of Flows and Cost Estimation
Finally, flow simulation and cost estimation are necessary to verify the feasibility of the project before its final acceptation and implementation.
Line balancing and equipment selection are the key steps in a lean process de- sign approach. This will be further explained in the next section.
In the remainder of this book line balancing will be developed in detail.
6.6.3.2. Kaizen
Kaizen is usually translated as “continuous improvement”, in contrast with the emphasis on brutal changes in manufacturing systems (reengineering being the most extreme example). Applying Kaizen introduces never-ending incremental changes to improve the manufacturing systems. Kaizen is supposed to humanize the work place by eliminating hard work, training employees in order to make them more capable of taking an active part in the system, and proposing improve- ments in the operating conditions of the system.
As widely mentioned in the literature, the Kaizen approach is based on the fol- lowing principles:
• Work as a team, which is the ability to cooperate with other employees for the best of the company.
224 6 X-manufacturing Systems
• Improve all the components of professional life. This includes knowledge im- provement, attainment of social values, deep involvement in the success of the company, etc.
• Implement quality circle to reach perfect first-time quality.
• Encourage improvement suggestions.
• Introduce statistical and quantitative methodologies to measure improvement.
• Take care of employees’ training: human resources are the most important company asset.
• Make sure that processes evolve by gradual improvements (rather than by dras- tic changes).
As mentioned by Imai (1986):
Kaizen means improvement. Moreover it means continuing improvement in personal life, home life, social life, and working life. When applied to the workplace, Kaizen means continuing improvement involving everyone – managers and workers alike.
6.6.3.3 5S
The goal of the 5S system is to reduce waste and improve productivity by making the workplace clean, ergonomic and rational. The 5S method is characterized by 5 Japanese words:
• Seiri, which means tidiness. This refers to the practice of keeping only essential items in the working area. Furthermore, each item (parts, cutting tools, etc.) should have a very precise location in order to facilitate work.
• Seiton, or orderliness. A precise place should be dedicated to each object, the goal being to make the access to any object easy.
• Seiso, or cleanness. The workplace must be kept clean; ideally, the workplace must be cleaned and the items must be restored to their place at the end of each shift.
• Seiketsu, which refers to the standardization of the activities of each employee concerning housekeeping.
• Shitsuke, or discipline. The goal is to maintain the standards of the workplace in order to be able to work efficiently day after day.
Routines that maintain the workplace well organized and in order are of utmost importance to a smooth and efficient flow of activities: it is the goal of the 5S sys- tem.
To summarize, the 5S system reduces waste and optimizes productivity through a workplace that is maintained well organized and clean day after day.
6.6.3.4 Total Productive Maintenance (TPM)
The goal of TPM is to preserve the functions of the physical assets or, in other words, make sure that resources are capable of doing what the users want them to do and when they want them to do. To reach this goal TPM organizes the system- atic execution of maintenance by all employees, whatever their level in the hierar- chy, through small groups of activities.
The overall goal of TPM can be broken down into five more detailed goals:
1. Design and install equipment that needs little or no maintenance. Continuously check the effectiveness of equipment by measuring waste that occurs such as defect or downtime losses. To summarize, this first goal is to improve the ef- fectiveness of the equipment.
2. Target autonomous maintenance. This objective is reached by training workers to take care of the equipment they are using. “Taking care” means:
– repair by using a service manual that contains carefully crafted instructions to solve problems;
– develop preventive actions;
– propose improvements to avoid breakdowns.
3. Plan the maintenance in detail. This implies:
– identifying all possible breakdowns;
– standardizing the solutions to problems;
– defining precisely the way preventive maintenance is to be organized.
4. Train employees to efficiently perform maintenance and repair operations on their equipment.
5. Focus on preventing breakdowns (preventive maintenance). This includes the use of methods that prevent questionable working habits.
To summarize, TPM emphasizes the importance of employees who are sup- posed to be capable of repairing and improving the effectiveness of the manufac- turing system and to work together to reach this goal. Another view of TPM is to say that it focuses on the total life cycle of the equipment, on continuous im- provement of production efficiency, on participation of all the employees and on a total-system approach.
6.6.3.5 Just-in-time (JIT)
The main objective of JIT is to reduce WIP and associated costs. JIT is driven by Kanban, a visual system controlling the flow of products.
A rough description of Kanban can be given in two points:
• Initially, a given set of Kanban is assigned to each station of the line. A line is a sequence of cells that are visited in the order of their sequence. Each Kanban corresponds to a set of semi-finished products a station is entitled to require
226 6 X-manufacturing Systems from the next upstream station if any or from the raw material and/or compo- nent magazine otherwise.
• When a set of semi-finished products enters a station, the corresponding Kan- ban is attached and cannot be used to require another set of products from the next upstream station. The Kanban becomes active again when the set of prod- ucts it is attached to is required by the next downstream station. Indeed, a sta- tion can require a set of products from the next upstream station only if it has a corresponding active Kanban.
A consequence of these rules is that the number of products in a station is upper bounded by the number of products represented by the Kanban initially assigned to this station. By assigning to the stations a set of Kanban that represent approxi- mately the same amount of work (in terms of working time) the system reaches a smooth production flow.
In the previous description, we assumed that the production system is a se- quence of cells. The Kanban approach also applies if different cells that provide different components are available at some manufacturing levels.
Assume, for instance, that two cells denoted by A and B provide two compo- nents a and b at level k and that these components are assembled at level k + 1.
Assume also that a Kanban at level k + 1 represent 10 products. This system is represented in Figure 6.15.
A
B
(A, B) (A, B)*
Level k Level k+1 Level k+2
Figure 6.15 Assembly system
When such a Kanban is active at level k + 1, this level orders 10 components a and b to cells A and B, respectively. If these components are available, they are delivered to level k + 1 and, at this point in time, the Kanban of level k + 1 is no longer usable and the Kanbans at level k that correspond to the components deliv- ered become usable again.
The 10 components a and b arrive at level k + 1 where they are assembled to provide 10 semi-finished products denoted by (a, b). When these products are or- dered by level k + 2, then the corresponding Kanban of level k + 1 becomes avail- able again, but the corresponding Kanban of level k + 2 is frozen.
6.6.3.6 Six Sigma Method
The six sigma methodology originated at Motorola in the 1980s. The objective of the methodology is process improvement and reduction of the variation of the characteristics of the products through a measurement-based strategy.
Mathematical Background
The performance of a process is measured by the “distance” between the design requirement, which is the set of required values of the characteristics of a type of product, and the actual values taken by these characteristics.
Assume that only one characteristic is concerned, say X, and denote by xn
x
x1, 2,L, the values taken by X for n products. Denote by m the mean value of the characteristic. This value is defined at the design level. The values
n i
xi, =1,2,L, are disseminated around m. The measure of the deviation of these values around m is the standard deviation σ that is the value of:
( )
∑= +∞
→ n −
i
n xi m
n 1 1 2
Lim
X is a random variable. In other words, X is a variable that takes a random value each time an experiment takes place. In this case, an experiment consists in manufacturing a product and the value taken by X is the value of the characteristic for this product.
It is commonly assumed that X is Gaussian, which means that its density of probability is:
( )
⎥⎥
⎦
⎤
⎢⎢
⎣
⎡ − −
= 2 2
exp 2 2 ) 1
( σ π σ
m x x
f
where x represents the value taken by X. Thus, the probability for X to take a value on interval [m−zσ,m+zσ ], where z is a real positive number, is:
( )
+∫
−
⎥⎦
⎢ ⎤
⎣
⎡ − −
=
σ
σ σ
π σ
z m
z m
z x m x
I d
exp 2 2
1
2 2
Setting σ
m y x−
= , we obtain:
228 6 X-manufacturing Systems
−∫
⎥⎦
⎢ ⎤
⎣
= ⎡ −
z
z
z y y
I d
exp 2 2
1 2
π
The value of this integral is available in tables provided for the Gaussian, cen- tered and reduced density of probably. In Table 6.3, we display the value of Iz for several values of z.
Table 6.3 Some values of Iz as function of z
z 1 1.5 1.8 2 2.5 3
Iz 0.6826 0.8664 0.9282 0.9594 0.9876 0.99735
The probability that X takes its values in the interval [m−3σ,m+3σ ] is 0.99735. In other words, on the average, 99 735 products out of 100 000 will have a characteristic value that belongs to this interval. Thus, 100 000–99 735 = 265 products out of 100 000 will have a characteristic value that is either greater than
3σ
m+ or less than m−3σ .
Now, assume that two characteristics X and Y are attached to each product. We denote by mX and σX the mean value and the standard deviation of X, and by mY and σY the mean value and the standard deviation of Y. If X and Y are inde- pendent from each other, then the probability to have
[mX X mX X ]
X∈ −3σ , +3σ and Y∈[mY−3σY,mY +3σY ]
is (0.99735)2 = 0.99471, which means that, on the average, 100 000 – 99 471 = 529 products (out of 100 000) have at least one of their characteristics with a value outside the interval, which is twice as much as in the case of one characteristic.
More generally, if r independent characteristics are involved, the probability for all the characteristics to take their values inside an interval of length 6σ is (0.99735)r and the probability that at least one characteristic takes its value out- side the related 6σ interval is equal to 1– (0.99735)r.
If the intervals of length 6σ related to the r characteristics represent the toler- ances defined at the design level, then 1–(0.99735)r is the probability to have a de- fective product when r is the number of characteristics involved. Table 6.4 shows how this probability increases with r.
As we can see, a tolerance of ±3σ is not acceptable since it leads to about 5%
of defective parts if 20 characteristics are involved, which is not rare. This is why the value of z has been taken equal to 6, which explains the name of the method:
six sigma. Thus, from this point onwards, we will consider an interval of length 12σ: [m−6σ,m+6σ ].
Table 6.4 Average number of defective products according to the number of characteristics for a tolerance of ±3σ
r 1 2 3 4 5 6 7 10 20
Probability (multipled by 103)
2.65 5.29 7.92 10.56 13.18 15.79 18.40 26.18 51.68
Number of defective prod.
(out of 100 000)
265 529 792 1056 1318 1579 1840 2618 5168
With this new constraint, a unique characteristic would lead to 0.002 defective parts per million. Motorola who initiated the method presumed that the character- istic mean can drift 1.5σ in either direction. This prompts us to adopt the worst case, that is the interval [m−4.5σ,m+4.5σ ] that corresponds to 3.4 defective parts per million on the average, when one characteristic is concerned. Table 6.5 provides the new results according to the number of characteristics.
According to the results given in Table 6.5, we see that only 68 defective prod- ucts will be detected in a set of one million products (on the average) if 20 charac- teristics are checked. In other words, 0.0068% of the products are defective on the average if 20 characteristics are concerned: this result is acceptable.
Table 6.5 Average number of defective products according to the number of characteristics for a tolerance of ±4.5σ
r 1 2 3 4 5 6 7 10 20
Probability (multipled by 106)
3.4 6.8 10.2 13.6 17 20.4 23.8 34 68
Six Sigma Outline
In practice, the tolerance is given either at the design level (mechanical tolerance) or at the management level (when competitiveness is at stake). Assume for in- stance that [a,b] is the tolerance of a characteristic. This means that ideally the characteristic should take its value in this interval. In this case, the mean value of the related random variable is:
2 b m a+
=
Let us consider the case of the ±6σ interval. The goal is to manage the manu- facturing process in order to obtain a standard deviation σ such that
230 6 X-manufacturing Systems
b a a+ −6σ =
2 (or a+b+6σ =b
2 )
Transforming one of these equations, we obtain:
12 a b−
σ = (6.9)
The goal of the six sigma method is to improve the manufacturing process to reach a standard deviation upper bounded by the second member of Equation 6.9.
Two questions remain open:
• How could we improve the manufacturing process?
• How could we evaluate the standard deviation σ ?
Improvement of the Manufacturing Processes
Two key methodologies are applied to reach the standard deviations of the charac- teristic values that are small enough to guarantee a limited number of defective products, as explained in the previous subsection.
• The first method is used to improve an existing process or product. It is called DMAIC, where D stands for “define”, M for “measure”, A for “analyze”, I for
“improve” and C for “control”. The five stages are as follows:
– Stage “define”: formally defines the goals to reach in order to improve customers’ satisfaction.
– Stage “measure”: measures the initial values taken by the characteristics of the products (or processes) for future comparison.
– Stage “analyze”: the objective is to establish relationships between man- agement or design decisions and the values taken by the characteristics.
– Stage “improve”: the goal is now to use the results of the previous stage in order to decide what changes are to be introduced in the system.
– Stage “control”: at this last stage, the effects of the changes are checked and evaluated. Sometimes, the process restarts at the second stage (“meas- ure”) for further improvements.
• The second method aims at establishing the activities to perform in order to de- sign a product (or a process) that meets customers’ requirements. This method is called DMADV, where D stands for “define”, M for “measure”, A for “ana- lyze”, D for “design” and V for “verify”. The five stages are as follows:
– Stage “define”: the goal is to define the design process that is supposed to meet customers’ requirements and the enterprise strategy.
– Stage “measure”: defines the key measurable characteristics of the prod- uct (or the process) under consideration, the product (or process) capabili- ties, and the risks that may be encountered at the production and utilization