Ask the Expert: Technical
A: Contingency planning is tied to risk management as well as PERT, both of which are described in various Web sites.
The most difficult aspect of contingency planning is determining the possible scenarios that will negatively or positively affect the planning process (i.e., what could go wrong or right, the probability of these occurrences, their financial impact, what could be done about them, how to implement changes to address them, and so forth).
Many times, this process is done through classic problem-solving techniques such as brainstorming, nominal group technique, force fields, fishbone diagrams, etc. The difficulty is having people willing to and capable of looking at the worst and best possible scenarios. Most people don't want to look at anything but the most probable or the most comfortable possibilities.
To paraphrase Gen. Dwight D. Eisenhower, planning is everything, but planning is worth nothing during implementation.
Timothy J. Greene, Ph.D.
A: IE departments are different company to company in terms of what they do, what their mission is, the extent of work performed, and their areas of expertise, so providing a detailed answer really depends on what your department does.
Start by listing the core competencies of your department, such as work measurement, project management, process design, ergonomics, etc. What are the key services that your department provides to your customers? Once you have that list, determine who the three or four key customers are for your services. Then create a matrix with core competencies on one axis and customers on the other. For each box (pair of customers and competencies), determine what measure could be used to describe what that customer expects for that service. I see it as a relationship between the customer and the service provided to that customer.
Obtaining measure of process accuracy and robustness are, as you know, pretty difficult to determine. There are a huge number of variables that affect the result. You can use SixSigma tools to measure the Cpk of the process and assume that if the Cpk is above an acceptable level (say, 1.67), then the process was well designed from a quality standpoint. You might have a customer requirement that all processes must be evaluated from a SixSigma perspective and provide a certain level of Cpk in order to be a satisfactory design. The question of accuracy and robustness must come back to the use of Six Sigma tools.
A: Industrial engineering (IE) is an engineering discipline that deals with the design and development of integrated systems of people, machines, and information for producing products and services. Its practice is found throughout all sectors of private and public industry as well as government. The common focus is to provide systems that offer maximum efficiency.
Operations management (OM) is the planning and analysis of the scheduling, dispatching, materials and inventory support, and related logistics activities for operations (both production and service). OM professionals apply operations research (and to some extent, industrial engineering methods) for establishing plans and policies for organizations.
Operations research/management science (OR/MS) primarily uses quantitative methods for making decisions in operational environments, usually with scarce resources. This area focuses more on methodological work and mathematical modeling. The OR side is somewhat more engineering=oriented than MS, which tends to have more practitioners from business and management.
In general, IE focuses on processes and people, OM on policy and programs, and OR/MS on methods and models.
Marlin U. Thomas, Ph.D., P.E.
A: You may certainly do a (1-p) chart. (I guess you view the glass as being half full.) The calculations will be the same. Just remember, though, that the best you can be is 1.0. That is 100 percent good - it doesn't get any better than that. Many software programs will not truncate.
Larry Aft, P.E.
Engineered standards in the warehouse
A: It sounds like you have standards calculated for each of the activities of interest and it’s just the tracking that you need to add.You can use the internal clock of the computers that calculate the actual times of each activity. All you need to do is identify keyboard entries or screen selections that indicate the beginning and end of an activity you want to time. Since the system used for the order picking knows that activity is being done, these times can be linked specifically to each activity. With these data, it is a simple coding project to create plots that compare actual times to the standard and even create automated alerts when certain criteria are violated .
If the order picking is completely manual, it will be a little harder. In this case, you need to have the times measured manually. If you are satisfied with samples (as opposed to timing every activity), you can keep doing it this way.Again, you can set up the same kind of system mentioned above with manual entry of the times whenever they are recorded.The data analysis would still be automated and alerts could still be included.
Marc Resnick, Ph.D.
A: A gauge is typically made to one-tenth of the tolerance it is meant to check. Whatever tolerance you use for the gauge comes out from the total tolerance available for the part. The tolerances for go and no-go gauges reduce the available tolerance for the part.
A gauge is typically made to one-tenth of the tolerance it is meant to check. Whatever tolerance you use for the gauge comes out from the total tolerance available for the part. The tolerances for go and no-go gauges reduce the available tolerance for the part.
Since you are checking to a tolerance range of 0.3 mm, you can make the gauge to 0.03 mm, knowing that the effective available tolerance will be only 0.27 mm. For cases in which the tolerance is very narrow, you can go higher on the percentage of tolerance for the gauge — up to 20 percent — depending on the application and the measuring system that you are employing.
Merwan Mehta, Ph.D.
Enforcing production rates
A: The simple answer to your question is yes: You need an engineered standard before using it as a basis for discipline. The standards must be set with accepted industrial engineering practice, whether that be time study, predetermined times, or standard data.
Discipline needs to be administered only for consistent performance under standard. Normal variability will have a short-term impact on any one individual’s performance at any particular point in time. (If you think of performance following the bell-shaped curve, half the time performance will be above-average, and half the time it will be below. If the standard is the midpoint of the bell, then we would expect some below-standard performances and some above.)
Larry Aft, P.E.