Wk5 Discussion 2

13.4 KANBAN SYSTEMS The first part of this chapter dealt with the philosophical underpinnings of the Lean philosophy. In this section, we take a decidedly more executional view, focusing on one particular approach to production control in a Lean environment, known as kanban. But even as you are working through the logic of kanban systems, keep in mind that the focus is still on reducing waste. A kanban system is a production control approach that uses containers, cards, or visual cues to control the production and movement of goods through the supply chain. These systems have several key characteristics: 1. Kanban systems use simple signaling mechanisms, such as a card or even an empty space or container, to indicate when specific items should be produced or moved. Most kanban systems, in fact, do not require computerization. 2. Kanban systems can be used to synchronize activities either within a plant or between different supply chain partners. Therefore, a kanban system can be an important part of both production activity control (PAC) and vendor order management systems (Chapter 12). 3. Kanban systems are not planning tools. Rather, they are control mechanisms that are designed to pull parts or goods through the supply chain, based on downstream demand. As a result, many firms use techniques such as MRP (Chapter 12) to anticipate requirements but depend on their kanban systems to control the actual execution of production and movement activities. Kanban system A production control approach that uses containers, cards, or visual cues to control the production and movement of goods through the supply chain. To illustrate how a kanban system works, we will describe a two-card kanban system that links the production and movement of units at two work centers, A and B. Suppose that in work center A, metal rectangles

are bent to form a bracket. In work center B, two holes are then drilled into the brackets. Figure 13.5 shows a diagram of the system. Two-card kanban system A special form of the kanban system that uses one card to control production and another card to control movement of materials. SUPPLY CHAIN CONNECTIONS CREATING A LEAN SIX SIGMA HOSPITAL DISCHARGE PROCESS6 A lengthy, inefficient process for discharging inpatients is a common concern of hospitals. It not only causes frustration for patients and family members but also leads to delays for incoming patients from Admitting, the Post Anesthesia Care Unit or the Emergency Department. When Valley Baptist Medical Center in Harlingen, Texas, faced this issue, it decided to apply Lean Six Sigma and change management techniques within one pilot unit. A multidisciplinary project team led by a black belt included nursing staff, case managers, an information technology green belt, and the chief medical officer, also a green belt. The project was to reduce the time between when a discharge order for a patient is entered into the computer and when the room is ready for the next patient. During the initial scoping of this project, the team divided the process into four components: 1. From discharge order entry to discharge instructions signed; 2. From discharge instructions signed to patient leaving; 3. From patient leaving to room cleaned; and 4. From room cleaned to discharge entered in the computer (thus indicating that the bed is ready for another patient).

Because of the hospital’s commitment to customer service, the team was asked to concentrate on the first two components. The goal was for this first subprocess to be completed in less than 45 minutes. To minimize the time a bed was empty, the team realized it would also need to address the time between when a patient’s room was cleaned and the time a discharge was entered into the computer, or the second subprocess. This would address the problem that arises when Admitting does not have the necessary information to assign a new patient to a clean and empty bed. Mapping the Process The team began with a process map to visually understand how the process was currently working. When several nurses were asked to help develop a detailed process map on the discharge process, they initially could not reach consensus since they each followed their own methods for discharging the patient. This lack of standard operating procedures had led to widespread process variation. The team developed a representative process map, printed a large copy, and placed it in the nurses’ lounge. Each staff member was encouraged to review the map and add comments on the flow. After a week, the team

retrieved the inputs and revised the “as is” process map accordingly. Elements of Lean thinking were combined with this map to help identify waste (or muda). To understand which steps were not contributing to timely discharge, aspects of the existing process were categorized as value-added, non-value-added, and value-enablers.   6Lisa Abelson Company Baseline data revealed the “from discharge order entry to patient leaving” subprocess required 184 minutes, with a standard deviation of 128 minutes. The second subprocess, “from patient leaving to discharge in computer” required an average of 36 minutes, with a standard deviation of 36 minutes. When compared against an upper specification limit of 45 minutes, the first subprocess had a yield of 7% (i.e., 7% of the patients were able to leave in 45 minutes or less), and the second subprocess did only slightly better, with a yield of 25% compared to its upper specification of 5 minutes. Behind the Waste and Variation The most important tool for determining the critical drivers of waste and variation was the Lean process map. The staff segmented the process into key steps and used the value-added and non-valued-added times to understand the delays and rework involved. The segments of the process were: • Secretary processes discharge order entry; • Discharge order processed to nurse begins (delay); • Nurse begins computer entry (to create discharge instructions); and • Computer entry to patient signature. The team found that three factors were critical drivers of waste and variation: 1. Clarification. In 21% of the cases, clarification from the physician was needed before the nurse could enter the information in the computer. The team confirmed that clarification processes added a significant amount of time. The median flow time of the process increased from 12 minutes to 45 minutes when clarification was required.

2. Handoff. The current process required a handoff as the charge nurse placed vital signs and other relevant information in the computer system, printed out the discharge instructions, and then placed them in a bin for the primary nurse to pick up. In many cases, the primary nurse would then review the information with the patient and obtain the patient’s signature. In a small number of cases, however, the primary nurse completed all tasks without any handoff. The median time increased from 9 minutes when one nurse completed all tasks to 73 minutes when a handoff between nurses was required. Without a signal for the handoff, the patient’s paperwork often waited up to an hour before it was acted upon. 3. Aftercare. Finally, the team tested the hypothesis that when aftercare was required (for example, ordering equipment), there was an increase in median cycle time from 121 minutes in the current process to 160 minutes when aftercare was required. Improving the Process Since variations in the “as is” process were contributing greatly to long cycle times and delays, a new six-step standard operating procedure (SOP) was developed: 1. Unit secretary enters discharge order. 2. Unit secretary tells primary nurse via spectra link phone that he or she is next in the process. 3. Primary nurse verifies order and provides assessment. 4. Primary nurse enters information into computer system. 5. Primary nurse prints instructions and information. 6. Primary nurse reviews instructions and obtains patient signature. Having the primary nurse complete all discharge tasks eliminated the bottlenecks created by time-consuming handoffs, the need for signaling those handoffs, and the fact that the charge nurse, who has many responsibilities, was not always readily available. With the first subprocess of the deliverable improved—from discharge order entry to patient leaving—the team focused on getting information into the computer so the bed could be filled. A session was conducted with transporters and unit secretaries to determine the best way to

improve the computer entry process. It was immediately clear that the current process was not working. Unit secretaries were not always aware when a patient was transferred from the unit. No signal was provided when a transporter moved a patient. Since the secretaries performed numerous activities (not always at the nurses’ station), they could easily forget that a patient had been discharged. A small discharge slip was developed, containing the patient name, room number, and time of call. The transporter would pick up the patient and then go to the nurses’ station and ask the secretary to provide the time on the computer. The transporter would write the time and hand the slip to the secretary. This served as a trigger and transferred the process from the transporter to the secretary. Maintaining Improvement Two tactics employed simultaneously helped to sustain the improvements. The first was the use of a change acceleration process (CAP) and the second was an ongoing tracking system. Four CAP sessions were guided by the black belt and process owner, increasing understanding about why the initiative was undertaken, providing baseline data, and establishing the rationale for improvements. Each session also included an exercise to help participants better appreciate Lean and Six Sigma, with a catapult exercise as a learning tool. Participants split into groups and worked to meet customer needs. They then reviewed the process, made adjustments, and developed standard operating procedures. Upon execution, the new plan showed improved performance. A tracking system included three components: 1. A daily report of the prior day’s discharges, including discharge times, primary nurse, and unit secretary responsible for discharging the patient from the computer; 2. A performance tracker to ensure individual accountability for primary nurses and unit secretaries in terms of mean, standard deviation and yield; and 3. A control chart that tracked the means and standard deviations. Summary: Process in Control

With the process now in control, the components were re-measured (Table 13.1). The “from discharge order entry to patient leaving” subprocess showed a mean improvement of 74%, with a 70% decrease in the standard deviation. The second subprocess, “from patient leaving to discharge in computer,” showed an improvement of 90% in the mean and 58% in the standard deviation. With success in this unit, a translation effort would be undertaken for the entire hospital. This will be an ongoing effort requiring change management for the entire hospital and training sessions on the new standard operating procedures. TABLE 13.1 Results of Lean Six Sigma Effort at Valley Baptist Medical Center

From Discharge Order Entry to Patient Leaving: Upper

Specification Limit = 45 Minutes

From Patient Leaving to Discharge in Computer: Upper Specification Limit = 5 Minutes

BEFORE AFTER BEFORE AFTER

M e a n

184.8 min. 47.8 min. 36.6 min. 3.47 min.

S t. D e v.

128.7 min. 37.2 min. 36.1 min. 16.9 min.

Y i e l d

6.9% patients leave within 45 min.

61.7% patients leave within 45 min.

24.6% entered into computer within 5 min.

95.4% entered into computer within 5 min.

Note that each work center has boxes of raw material in front of the work center and finished material directly after. For work center A, the raw material is unbent metal; the finished material is undrilled brackets. For work center B, the raw material is undrilled brackets and the finished material is drilled brackets. Under kanban system rules, each box of raw material must have a move card, while each box of finished material must have a production card. These cards are used to precisely control the amount and movement of material in the supply chain. We will see in a moment how the system works.

FIGURE 13.5 Kanban System for Two Work Centers MyOMLab Animation

Move card A kanban card that is used to indicate when a container of parts should be moved to the next process step. Production card A kanban card that is used to indicate when another container of parts should be produced. Now suppose a downstream customer places an order. The effect is to “pull” a box of finished drilled brackets (box 1) out of work center B. Immediately, the production card is removed from the box and placed in a conspicuous location in work center B. The card signals to personnel in work center B that they need to drill more brackets (Figure 13.6). To drill more brackets, employees in work center B must pull a box of undrilled brackets (box 2) into the production process. As they do so, they remove the move card from box 2 and replace it with the

production card that was removed from box 1. The newly freed-up move card then signals to employees that they need to move, or “pull,” more materials out of work center A (Figure 13.7). When the freed-up move card arrives at work center A, it takes the place of a production card on a box of undrilled brackets (box 3), and that box is transferred to work center B (Figure 13.8). The freed-up production card then signals employees in work center A to produce more parts. To summarize this system: • A downstream station pulls finished material out of work center B (Figure 13.6). • Work center B pulls raw material into production (Figure 13.7). • Demand for more raw material in work center B pulls finished material out of work center A (Figure 13.8). The beauty of this system is that all production and movement of materials is controlled by a set of cards. If workers see a freed-up production card, they produce more units; if they don’t, they stop producing units. Likewise, if they see a move card, they move materials; if not, they leave materials where they are. You can see now why a kanban system is also called a pull system: Actual downstream demand sets off a chain of events that pull materials through the various process steps.

FIGURE 13.6 Release of Finished Materials from Work Center B MyOMLab Animation

FIGURE 13.7 Pulling of Raw Materials into Production at Work Center B MyOMLab Animation

FIGURE 13.8 Removal of Finished Materials from Work Center A MyOMLab Animation

Pull system A production system in which actual downstream demand sets off a chain of events that pulls material through the various process steps. As we noted before, cards aren’t the only signaling method used in a kanban system. Some other methods include: • Single-card systems, where the single card is the production card and the empty container serves as the move signal; • Color coding of containers; • Designated storage spaces; and • Computerized bar-coding systems. Controlling Inventory Levels Using Kanbans It is a simple fact that by controlling the number of production kanbans —whether they be cards, containers, or some other signaling mechanism —organizations can control the amount of inventory in the system. Consider our previous example. Work center A could not produce unless it had a freed-up production card. As a result, the number of production cards set precise limits on the amount of inventory between work centers A and B. While reading the last section, you may have wondered how organizations determine the number of kanbans needed to link together two process steps. The answer depends on several factors, including the lead time between the two steps being linked, the size of the containers that hold the parts, the demand level, and the stability of demand. A general formula for calculating the number of kanbans is:

where: y = number of kanbans (cards, containers, etc.) D = demand per unit of time (from the downstream process) T = time it takes to produce and move a container of parts to the downstream demand point

(13.1)

x = a safety factor, expressed as a decimal (for example, 0.20 represents a 20% safety factor) C = container size (the number of parts it will hold) EXAMPLE 13.2 Determining the Number of Kanbans at Marsica Industries, Part 1 At Marsica Industries, work cell H provides subassemblies directly to final assembly. The production manager for work cell H, Terri O’Prey, is trying to determine how many production cards she needs. Terri has gathered the following information:

Using Equation (13.1), the number of production cards needed is:

Terri rounds up her answer because there is no such thing as a fractional production card. Evaluating the results, she notes that 20 production cards is the equivalent of 20 containers of subassemblies, or:

(20 containers)(45 subassemblies per container) = 900 subassemblies And in hourly terms, 900 subassemblies equals:

D Final assembly’s demand for subassemblies from work cell H

300 assemblies per hour, on average

T Time it takes to fill and move a container of subassemblies from work cell H to final assembly

2.6 hours, on average

x Safety factor to account for variations in D or T 15%

C Container size 45 subassemblies

The fact that there are slightly more subassemblies than needed is due to the safety factor and the rounding up of the number of production cards. While Equation (13.1) is useful as a starting point, another approach used by many companies is to start with more than enough kanbans. The organization then slowly removes kanbans in an attempt to uncover the “rocks,” or problems (similar to Figure 13.4). At the same time, the organization will try to shorten lead times and stabilize demand levels as much as possible, thereby further reducing the need for inventory. EXAMPLE 13.3 Recalculating the Number of Kanbans at Marsica Industries, Part 2 After nearly a year of continuous improvement efforts in work cell H, Terri O’Prey feels it is time to reevaluate the number of production cards and hence inventory in the work cell. In particular, Terri has made the following changes: • Production lead time has been cut from 2.6 hours to a constant 1.6 hours. • Demand from final assembly has been stabilized at 300 subassemblies per hour. • Smaller, standardized containers that hold just 25 subassemblies are now being used. Because production lead time (T) and demand rate (D) have been stabilized, Terri feels she can reduce the safety factor to just 4%. She recalculates the number of kanban cards to reflect all of these changes:

Since the container size is smaller, Terri is not concerned that the number of cards has not changed. In fact, 20 production cards are now the equivalent of:

(20 containers)(25 subassemblies per container) = 500 subassemblies

and

Either way she looks at it, by improving the process, Terri has been able to cut inventory significantly. Synchronizing the Supply Chain Using Kanbans In Chapter 12, we alluded to the idea that kanban systems can be used to synchronize the supply chain at the PAC and vendor order management levels. Put another way, kanban can be used to link supply chain partners, as well as the work centers in a factory. Suppose, for instance, that work center B in our earlier examples is located in a facility 200 miles from work center A. In this case, electronic requests for more materials would be substituted for the factory’s move cards. Figure 13.9 shows how kanban can be used to synchronize the production and movement of goods among multiple supply chain partners. You might even think of customer demand as a pull on a rope (the kanban system) that ties together all members of the supply chain. One pull at the end of the supply chain triggers movement and production down the chain. For a kanban system to work properly, however, there must be a smooth, consistent pull of material through the links. Consider the supply chain shown in Figure 13.10. As we have seen, the number of kanbans linking work centers A and B is based on an understanding of the demand rate coming from B. But what happens if the demand rate changes or there is an interruption in the flow of goods? If final assembly demand doubles, work center B may quickly use up all the material linking it with A, and subsequent shipments from B to final assembly may be slowed down as a result. If there is an interruption in the flow of goods—say, within work center B—the result could be even worse: Final assembly may have to stop production, thereby stopping the pull of goods from work centers C and D as well and shutting down the entire supply chain. This is not as farfetched as it seems; in fact, it is exactly what happened to Toyota in

1997 when the manufacturing plant for Toyota’s primary supplier of brake proportioning valves—a $20 part—burned to the ground. Within hours, the Toyota final assembly plant had to shut down due to the lack of one part. The shutdown reverberated through the rest of Toyota’s supply chain, with production resuming only after other Toyota suppliers started producing the missing brake parts. The point is this: For a kanban system to work properly, demand rates must be relatively stable, and interruptions must be minimized or quickly resolved.

FIGURE 13.9 Using Kanban to Synchronize the Supply Chain

FIGURE 13.10 Work Centers A and B as Part of a Larger Supply Chain Using MRP and Kanban Together Some companies have found it beneficial to combine the planning capabilities of MRP with the control capabilities of kanban. In particular, MRP can be used to anticipate changes in planned order quantities over the planning horizon. This information is then used to recalculate the number of production kanbans (containers or cards) needed. Example 13.4 illustrates how the concept works. EXAMPLE 13.4 Using MRP and Kanban Together at Marsica Industries, Part 3 The past six months have been tumultuous ones for Marsica Industries; demand levels have varied dramatically from one week to the next, as the company has taken on seasonal customers and marketing has used pricing changes to either boost or limit demand. The result for Terri O’Prey, production manager for work cell H, has been that the D values underlying her kanban calculations have been all over the place, undercutting the effectiveness of the kanban system. Terri knows that

she needs some way to anticipate these changes and adjust the number of kanban production cards accordingly. Terri knows that the company uses MRP to estimate planned orders for components, including the subassemblies coming out of work cell H. She finds the MRP record for the subassembly shown in Figure 13.11.

FIGURE 13.11 MRP Record for Work Cell H’s Subassembly Looking at the MRP record, Terri notices a couple of interesting points. First, there is no projected ending inventory. This is consistent with the Lean philosophy of having no more inventory in the system than is needed. Second, the planned orders all occur in the same week as the planned receipts. This is because the planning lead time for subassemblies is just 1.6 hours (Example 13.3); therefore, any orders released in a week should be completed in that week. But the most interesting line for Terri is the planned order quantities. These tell her the total weekly demand for the subassemblies. Assuming that this demand is spread evenly across a 40-hour workweek, Terri can use the planned orders to calculate the D values for the various weeks:

Finally, Terri can use the different demand rates and the other values from Example 13.3 to determine the number of production cards needed each week:

In practice, Terri will adjust the number of production cards by adding new cards when she anticipates that demand will go up and “retiring” freed-up production cards when she anticipates that demand will go down. But the key insight is this: Terri can use the MRP records to help anticipate needs and control production at the work cell level.