Ask the Expert: Manufacturing
Automation vs. machinization
A: Both machinization and automation refer to the use of machines to replace humans. Specifically, machinization refers to the duplication of skilled human effort; automation is the use of machines to replace repetitive human efforts.
If a computer is used to replace the skill that would be needed from a machinist to make a part, as we see happening through the use of CNC machines and software, it would be machinization. On the other hand, if the human effort needed to load and unload the machine were replaced by a machine or some equipment like a robot, it would be called automation. The difference is subtle: The installation of a welding robot could be called machinization, while the installation of a loading/unloading robot could be called automation.
In modern industrial vocabulary, automation has replaced machinization, and many a times automation is used even when someone is talking about machinization. Therefore, we see little use of machinization today.
Merwan Mehta, Ph.D.
A: Safety metrics range from program compliance protocols used for auditing to incidence rate reporting such as the OSHA recordable rate and lots of customized metrics in between. I don’t know what your specific interest is, so let me just give you a brief overview and then you can follow up if you need additional information.
Safety metrics can be grouped into two categories: Proactive metrics intended to assess how well safety-related activities are performed and reactive metrics that measure outcomes such as injuries and financial loss.
Proactive metrics include program compliance and performance measures as well as measures specific to various program elements. Safety program auditors typically follow proprietary protocols that often incorporate a scoring regime. An audit protocol might ask a series of yes/no questions about whether critical components of each area of a comprehensive safety program are in place, giving one point for each positive response. For example, a question dealing with the hearing protection program element might ask "Has a comprehensive noise level survey been performed by a certified professional within the last two years?" A comprehensive program audit typically covers 20 to 40 program elements, with numerous protocol questions in each element. Detailed audits, such as those required by the lock-out OSHA regulation, are typically more detailed. The intent of such metrics is to assess how well the employer is performing activities aimed at preventing loss. Other proactive metrics commonly include various types of checklists.
Reactive metrics count injuries and financial loss. OSHA recordable injury rates are gathered and analyzed by the Bureau of Labor Statistics. Industry averages are published about a year later. Insurers typically count the number of injuries that are compensable under workers’ compensation laws. They also compare the rate of compensable injuries against industry averages to arrive at a modification factor (MOD rate) for a given employer. For example, if ABC Co. has a MOD rate of 1.4, its injury experience is above average and it will have to pay a higher premium than XYZ Inc. with a MOD rate of 0.7. Other common measures include lost workday case rate, injury severity rate, OSHA lost time and restricted workday rate, and first aid case rate. There are also lots of accounting statistics, such as cost of workers’ compensation claims, accident/incident casualty losses, and business interruption rates and costs.
Housekeeping metrics are subjective assessments based on established plant guidelines, or 5S +1 guidelines. In general,these measures evaluate how well a department is achieving the goal of having a place for everything and everything in its place. 5S guidelines tend to be very formalized. Such assessments typically follow a standard protocol with questions such as "Are all hoses hung on designated hangers when not being used?" Again, by using binary questions, a scoring system can easily be established to measure performance. Less formal approaches include having a small team conduct a walk-through of the department with each member assigning a subjective score and then averaging the scores or having a single person scoring each department on a given day, hopefully resulting in a somewhat fair and consistent assessment.
Paul Adams, Ph.D., P.E.
A: This can be a difficult problem, particularly because parameters such as setup and changeover times, product characteristics, and limitations on tray sizes create complexity in the schedule. Start by looking into what you use to decide how to optimize the conveyor constraint. Assuming it is the constraint, removing it should be based on having the constraint somewhere else such that more profit is realized. If you have the resources to remove the conveyor, then more sales or total margin would be the best result. This measure of economic value-added (throughput) is not the same as a standard cost-based calculation, so I am not sure what measures are used by your organization to evaluate these changes. Reducing costs may provide some interesting improvement, but these may be of limited value if fixed costs increase much.
This answer seems to be wandering because if you could improve the conveyor’s downtime, I assume it would have been done. However, if not, the natural reaction would be to focus on improving the conveyor (unless its function is of limited value anyway).
It would appear your question is one of the following:
• How do you justify capital investment to eliminate a constraint?
• What scheduling mechanisms are best suited to multiple lines with a common or shared-end operation?
I have provided some ideas above for the justification to remove the conveyor line. For the second question, it would be best to schedule the constraint first and let the other operations keep pace even though the non-constraints may not operate at peak efficiency and may be the starting point for production. This can be accomplished in a number of ways, but it means that the high value-added products get scheduled first, possibly with dedicated upstream lines in a manner that optimizes the constraint and lets the inefficiency fall on the upstream production lines.
The tricky part comes in filling in the rest of the capacity without upsetting the rest of the capacity without upsetting the schedule for high-value products. A greatly simplified solution: Have some flexible market needs for products that may not be as high of value but that can be inserted into the schedule without upsetting the plan for the conveyor.
Avoid considering every order to be of equal priority because each one may take different amounts of time on the constraint and have a different split of revenue to hard currency costs. If the organization considers each order equally, then the tendency is to forecast out into the future, create a plan of future orders, and work hard to optimize the schedule. This can be a difficult game to win because forecasts are inherently inaccurate and operations have variability.
A: It may be necessary to observe the output (in terms of tons produced)over randomly chosen intervals of time. You can choose intervals to span different times of day, days of week, and different order types. You should then be able to develop a statistical estimate of mean throughput rate (average number of units produced per unit time) under a variety of conditions.
A more accurate estimate can be obtained by tracking several productspecific production lots and observing production plus setup times as the lot works through the plant. This will give you an average product-specific production rate. You can then find the aggregate production rate by estimating the mix of products this plant produces and taking a weighted average. The realized capacity is likely to be a function of product mix.
Diwakar GuPta, Ph.D.
24/7 cycle counting
A: It’s probably obvious that to do cycle counting correctly, you must create cutoff controls to ensure the count and the record in your system are timed properly.
In fact, there are generally three cutoff rules I like to use:
• Cycle count during non-operational shifts.
• Cycle count at the end of the day.
• Suspend processing all items that are on the count list until the count has been taken and reconciled.
In a 24/7 operation, these are tricky if not impossible to do. So you have to do the best you can.
Try to cycle count as many items as you can when that part of the operation is not working. Maybe some departments or areas don’t work a full three shifts or a full seven days. If this can’t be done,you must establish procedures to synchronize counting and transaction processing and make adjustments. In some cases, even this is difficult, and you would have to establish a schedule to count items much more frequently to keep some level of control over the counts.
A key variable is the timeliness of transactions into your system. If transactions are not handled in a timely manner, even the best situations can fall apart quickly.
Companies with 24/7 operations still use cycle counting because it can be effective, but those companies must work very hard at it, and they have to possess quick transaction processing and good procedures. It is still worth the effort and remains one of the best ways to control inventory on the plant floor.
Color me productive
A: Two ergonomics textbooks are helpful in this regard: Ergonomic Design for People at Work (1983) and Fitting the Task to the Man (1982). Even though they are somewhat dated, the information remains valid.
As far as color's impact on productivity, I'll summarize a few points:
• The Westinghouse Hawthorne Study is famous for the conclusion that any change in the workplace can positively impact productivity, as demonstrated by such things ass an increase in lighting leading to increased productivity, followed by a decrease in lighting leading to increased productivity. The key was the attention given to workers, not the changes themselves.
•Much of the research cited is perception-based. This would suggest it is important to ask the people involved what their color preference is. Getting input from the work force increases their involvement and ownership, which should positively impact productivity.
• Intense primary colors should be avoided because of their tendency to produce after-images, which lead to a loss of visual acuity. In general, pastels are freferred for large surfaces such as walls. Darker or richer colors may be used to achieve contrast, such as highlighting exits, safety warnings, or critical equipment controls. If the work is monotonous, include some areas of exciting color.
•Blue is cold and restful, creating the effect of a long distance. Red is warm and stimulating, creating the effect of a close distance. Orange and yellow are considered warm and exciting, which may indicate they could boost productivity. Walls and ceilings painted yellow, red, or blue may be initially attractive, but eventually they will be a strain on the eyes and make a room unpleasant.
• 5S and visual systems make the workplace self-explanatory and self-maintaining. Contrast helps, too. Sometimes it is best to maintain a light color for surfaces, which highlights dirt and debris. You must reinforce cleanliness and routine cleanup to make this work, however. Using color codes for equipment controls as well as equipment picking and return further enhance productivity.
Based on this information, I recommend:
• Use pastels and maintain contrast for critical items.
• A pastel earth tone or orange-brown may be more stimulating yet still palatable for long-term exposure.
• Get input from people who will use the room.
• Set some standards and consider overall aesthetics of the facility.
• Use 5S and visual systems for an orderly, productive workplace.
Splitting the run time
A: There are many instances of unattended machining in which an operator can run two, three, or more machines simultaneously because the machines have automatic feeding mechanisms or indexed setup platforms that move work into and out of machining operations.
There are also bulk operations in which operations are performed on many parts simultaneously - plating, heat treating, etc. In this case, the operator's time is not the concern: The cycle time of the operation controls throughput.
The operator's time is usually divided by the number of parallel operations taking place. In other words, divide the operator's time by four if four machines are being run by the same operator; divide the operator's time by 100 if 100 pieces are being plated at the same time.
John S.W. Fargher Jr., Ph.D.
A: Have you looked into general sewing data? GSD is a specialized version of methods time measurement that's applicable to apparel operations. One provider is Elyon Ltd.
Larry Aft, P.E.
A: I find that 80 percent of the theoretical rate is about right. If you don't have an efficiency rate calculated based on past experience, use 80 percent as the efficiency rate in the calculation below.
As for you second question: Don't worry about the line standard yet. Calculate takt time based on the demand of the customer. To calculate available capacity:
Capacity = Time available x Efficiency x Utilization
Time available = Hours per week x Number of employees
Efficiency = Standard hours produced / Hours worked
Utilization = (Hours available / Hours down) / Hours available
To calculate takt time:
Takt time = Available capacity / Customer demand
To calculate standard times at each workstation:
Standard time per piece = Run time per piece + Setup time per piece
Setup time per piece = Setup time per lot / Average lot size
The standard times at each workstation must be less than the takt time. If not, the workstation must be redesigned to distrubute work to other stations or parallel workstations need to be added to the line.
If personnel can move immediately to another workstation without wasting time (which is unlikely), the line standard will be a total for each workstation. If individuals do not move between workstations, it will be the takt time multiplied by the total number of workstations. It should be somewhere in between if the line is designed properly as a U shape.
John S.W. Fargher, Jr., Ph.D.
A: World-class manufacturing is a term coined by Richard Schonberger from the title of his book. It is generally recognized as lean manufacturing.
John S.W. Fargher, Ph.D.
A: Takt time is the rate that customers buy product. It is used to design a balanced line or work cell. Takt time is usually stated in seconds per end product.
To calculate it:
Takt time = Available time per day / Customer demand per day
Note that available time takes into account actual working time that can be applied to production. Deduct any typical daily non-production time such as breaks, lunch, meetings, and clean-up.
A: I am assuming you have a cycle (takt) time for the stages in your line. In that case, the quick answer is to measure line efficiency by picking a meaningful base period such as a shift and then:
Line efficiency = Cycle time x (Average units produced per base period) / Length of base period
Given conservation of matter, you can measure average units produced by the number of units that enter or exit the line per period. If you have quality rejects, you might want to use just the number of accepted items leaving the line.
The cycle time should probably be set equal to the slowest standard process time for any stage in the line. You will find that the accumulation tables buffer the workstations and allow line efficiency to be larger than if the whole line stopped each time one or more stations was down.
For more detailed analysis and predictions on the value of accumulation tables of various capacities, read Chapter 3 of my textbook Modeling and Analysis of Manufacturing Systems.
Ronald G. Askin, Ph.D.
A: You can use either formula or both of them. The second is universally accepted. The difference is that the first includes profit margins while the second looks at cost of goods sold without profits (profits being gross sales less the cost of goods sold).
John S.W. Fargher, Ph.D.
A: A well-structured syllabus and class on how to use the RF scanner (make sure you standardize on one scanner!) should use demonstration to teach people how to use it, including downloading data in the data collection system, how to maintain and charge the instrument, and so forth. After you have demonstrated the functions, have temps show you they can use the scanner on several items, download several sets of data in the data collection system, and perform maintenance steps.
The class should take 45 to 60 minutes. Don't try to save time here. Three to five people can be trained at a time.
Be certain that each new hire goes through all steps in the training plan prior to being placed on the floor. If an individual cannot demonstrate sufficient proficiency during the class, you will want to consider using a different temp.
Consider language, too. A take-away set of visual instructions should be provided in the language of the temp and the visual instructions in all languages should be posted throughout the warehouse.
John S.W. Fargher, Ph.D.
Longer shift, less productivity
A: I hate to answer a question with a question, but why are you inspecting wiring? Don't you use continuity tests? Are certified artisans (aircraft electricians) doing the job so their work doesn't have to be inspected (because they do their own inspection as they go)?
Certainly, productivity will decrease as the number of hours goes from 8 to 12, especially if the overtime is extended several weeks. My experience in the aircraft business is that after four weeks you get approximately 40 hours worth of work when the overtime is extensive (10-12 hours per day for six days a week) and that errors and quality problems occur at a much higher rate.
John S.W. Fargher, Ph.D.
Indirect labor needs
A:There are several ways to compute indirect labor requirements:
- Comparative estimating is most often used for developing time standards for maintenance activities. By taking time standards from known benchmark tasks that are common in an indirect labor function, you can apply uncommon task times to determine the total unusual function time. Often, the indirect uncommon tasks can only be described in words, so don’t feel you are at a dead end. Apply the task description to a similar common benchmark task and take advantage of known requirements for the task. This estimate can be used with volume fluctuations to determine standard indirect labor needed to support direct labor.
- Comparative estimating using time gaps uses upper and lower control limits to exercise variations in benchmark times as well as task times.
- Analytical estimating uses estimates of time from job knowledge and practical experience as well as from synthetic data (standards). Experience, honesty, and details are key issues. The more standard times can be applied, the more reliable your estimate.
- Baseline measures development involves performing time studies of the indirect labor task per unit, lot, batch, week, month, etc. Document possible variances to the task that could be common to any period of volume or season. Create reasonable benchmark times for the variances and apply them as you see fit. The overall estimate of indirect labor functions are, at best, dependent on skill level and the degree of training given to perform the task. Annual reviews of indirect labor tasks are highly recommended due to the fact that skill level and experience should lower benchmark standards over time.
Paul F. O’Connell
Tuning in to productivity
A: There is some research on the effect of listening to music in the workplace but less on listening to talk radio or ball games. For music, productivity is generally improved when employees get to select the kind of music played and when the background noise is less than 70 decibels. If workers can’t agree on the music, it creates more problems than it is worth. And if the music has to be played loudly to be heard, it can interfere with communications, alarms, or OSHA regulations.
For talk radio or ball games, it depends a lot on what kind of work is being done. Either of these can attract a large part of workers’ attention. So if the work requires directed attention, it is probably not a good idea. But for monotonous work that can be accomplished with little conscious attention, there should be no problem. There is some evidence that for variable work in which mental workloads rise and fall, workers are able to shift their attention as necessary, but this is true only if they notice the change. Another problem with talk radio or ball games is that they can create disagreements among workers that can hurt productivity and, in extreme cases, hurt morale.The best bet is to go with music if workers can agree on the kind. You can use special occasions to relax the rule, allowing workers to listen to a playoff ball game, for example. The productivity you lose in the short-term is easily compensated by the morale you gain overall.
Marc Resnick, Ph.D.