### Lean Advisory Tools (LAT) for Design and Operation of Job Shops

By Dr. Shahrukh A. Irani

**LAT #1: Product Mix Segmentation using P-Q Analysis**

**Overview of the Strategy:**Based on the annual production volume of each product, segment the product mix into Runners, Repeaters and Strangers ex. a low-variety high-volume (LVHV) segment and a high-variety low-volume (HVLV) segment. Next, break up and reorganize the existing facility into separate “mini-facilities” to

*separately*produce each segment of the product mix. Each “mini-facility” will have an appropriate layout, equipment flexibility, material handling system, workforce flexibility and skill levels, inventory control policies, scheduling system, etc. suited for the particular segment of the product mix assigned to it.

**Justification for this Strategy:**Assembly lines, transfer lines and multi-product flow lines are best suited for the low-variety high-volume product mix segment. Whereas, multi-product cells, flexible manufacturing cells, hybrid cellular layouts and functional layouts are the preferred factory configuration for the high-variety low-volume product mix segment. The

*production system cannot be optimal for these two distinctly different product mix segments. The main reason is the long vs. short production runs caused by the high vs. low production volumes, respectively,*

**same***impacts the number of setup changeovers, speed of operator learning and de-learning, stability of equipment/process parameters during operation, product mix diversity, and numerous other operating conditions that influence throughput, WIP and operating costs.*

**significantly****Muther’s method of P-Q Analysis can be used to segment the product mix based on the annual production volume of each product and then to determine the type of layout for any manufacturing facility – flowline, job shop, cellular (or some combination or a split of these three basic layouts) – best suited for each segment. The typical P-Q Analysis curve for an HVLV facility will show, at the left end of the curve, a relatively few products being produced in large quantities. These products are best produced on single- or multi-product production lines or in product-focused cells. Whereas, the right end of the same curve will show a great many different products being produced in small quantities that tend to be produced in a process (or departmental) layout, often referred to as a “job shop layout”. A “shallow” P-Q curve suggests a process layout for producing the entire product mix. Whereas, a “deep” curve suggests dividing the product mix into two or more segments and dedicating separate production areas, each with a different layout, to produce the different segments. For any job shop undertaking a Lean initiative, an essential starting point should be to understand the shape of their P-Q Analysis curve!**

Implementation of this Strategy:

Implementation of this Strategy:

**Illustrative Example from an Industry Project:**Table 1 shows the input data for P-Q Analysis that must be provided to PFAST. PFAST uses only the “Part” and “Quantity” columns of data in this spreadsheet. Typically, the “Quantity” (# of pieces shipped) and “Revenue” ($ earned) for each part are for a year, or larger production horizon.[A drawback of P-Q Analysis is that it does not factor in the Number of Orders that were placed for the part, and the time interval between each order. The average size of each order and the average inter-arrival time between repeat orders for the same part further helps to classify a product as a “Runner”, “Repeater” or “Stranger”. This enhancement of P-Q Analysis, P-Q-T Analysis (where T = Time), will be discussed in a future LAT.] The “Routing” is the sequence of work centers that a part must visit ex. Part No. 1 (80-A37353) has the routing: 17→6→2→11→10→29→54→55. Figure 1 and Table 2 are the graphical and tabular versions, respectively, of the P-Q Analysis output produced by PFAST. Question: Is this product mix only LVHV (low-variety high-volume), only HVLV (high-variety low volume) or some combination of the two extremes?

Sekine and Arai recommend checking if the P-Q Analysis curve is “shallow” or “deep” using 3 ratios as follows: Check the 2:8 ratio line i.e. the first 20% of the total number of products accounts for 80% of the Total (or Aggregate) Product Quantity. If that condition is not met, then check the 3:7 ratio line. If that condition is not met, then check the 4:6 ratio line. As the ratio increases in value from 0.25 to 0.67, that would indicate that the product mix is not dominated by a few high-volume “Runners”, and that the product mix contains “Repeaters” and “Strangers”. According to Sekine and Arai “… production lines that fall into the 4:6 ratio categories can be called wide-variety small-lot production lines. The typical Japanese factory falls into this category, but many are having problems succeeding at this kind of production”.

In Figure 1, the Parts are sequenced from left-to-right along the X-axis in order of decreasing value of Quantity of each (individual) part; whereas, the Y-axis on the left side of the graph shows the Quantity of each (individual) part and the Y-axis on the right side of the graph shows the Aggregate Quantity for any group of parts picked based on the three ratios recommended by Sekine and Arai. In this particular example, the curve is “deep” and Figure 1 shows a potential segmentation of the product mix into High-Volume (Runners), Medium-Volume (Repeaters) and Low-Volume (Strangers) segments.

In Table 1, where the parts have been sorted in order of decreasing value of Quantity, the three ratio lines are shown for 60%, 69% and 80% of the Total Aggregate Quantity (which is 1766478, as shown in Row # 79 of the table). With reference to the table, 13 parts (which is 16% of the entire product mix of 79 parts) account for 60% of the Total Aggregate Quantity, 17 parts (which is 21% of the entire product mix of 79 parts) account for 69% of the Total Aggregate Quantity and 29 parts (which is 29% of the entire product mix of 79 parts) account for 80% of the Total Aggregate Quantity.

So, using Figure 1 and Table 2 to answer the question that was posed earlier in this section –

*Is this product mix only LVHV (low-variety high-volume), only HVLV (high-variety low volume) or some combination of the two extremes*– I chose to break up the product mix of this forge shop into at least two segments, one comprised of the top 23 parts and the other containing the remaining parts. What do you think?

**Planning Improvement Projects based on these Results:**How does one translate the results obtained using this LAT into improvement projects (or kaizen events) that reduce/eliminate the Seven Types of Waste, increase throughput at capacity constraints, reduce inventory levels and reduce operating costs? Here are some examples of the follow-on projects that could be undertaken after this product mix segmentation analysis is completed –

- The typical product mix in an HVLV facility tends to be large. Therefore, collecting the relevant data for all products would be difficult and costly. Say that management desires to conduct one or more kaizen events that focus on reduction of material handling costs, standardization of storage containers, and improvement of shop floor communications. This would require data collection to generate Value Stream Maps, Flow Process Charts and Material Handling Planning Charts, as part of a larger Systematic Handling Analysis (SHA) initiative. Using P-Q Analysis, one could quickly select one (or more) products from the High-Volume segment, and ignore the Low-Volume products for the time-being.
- For any pair of work centers in any manufacturing facility, a key driver of Material Handling Cost is the total time that material handlers spend moving material between those two locations. This time is a function of travel distance between the locations and the number of trips made between them. The number of trips is directly proportional to the total volume of different products in whose routings these two work centers appear consecutively. So, if it is desired to lower the total Material Handling Cost for a facility, P-Q Analysis would immediately point those work centers that feature in the routings of the products in the High-Volume segment.
- Both segments of the product mix – low-variety high-volume and high-variety low-volume – could be analyzed for the existence of products with identical or very similar routings. For each product family identified in the High-Volume segment, a multi-product flowline could be designed. For each product family identified in the Medium-Volume segment, a flexible multi-product cell could be designed since any individual product may not have the demand to justify a dedicated cell. But, for the product families identified in the Low-Volume segment, I would do things differently as follows: (i) a flexible cell could be designed for each family and (ii) the remaining parts i.e. those that do not belong in any family in this segment, I would strongly recommend “culling” them from the product mix. Unless, of course, those parts are being supplied to the very customers whose parts also belong in the High-Volume segment! In which case, setup reduction and flexible manufacturing equipment could become priorities for the kaizen events. In fact, I personally know the Plant Manager at a local sheet metal fabrication job shop who referred to this product mix rationalization that they did as “sending our difficult and costly parts to our competitors”!
**References**

Burbidge, John L. (1996). Production Flow Analysis for Planning Group Technology. New York, NY: Oxford University Press. ISBN 19-856459-7.

Irani, S. A. & Huang, H. (2005) Hybrid Cellular Layouts: New Ideas for Design of Flexible and Lean Layouts for Jobshops. Columbus, OH: Zip Publishing (http://www.zipedu.com/HCL.html).

Sekine, K. & Arai, K. (1992). Kaizen for Quick Changeover: Going beyond SMED. Cambridge, MA: Productivity Press. ISBN 0-915299-38-0.

Dr. Shahrukh A. Irani is an Associate Professor in the Department of Industrial, Welding and Systems Engineering at The Ohio State University, Columbus, OH. He can be reached at**irani.4@osu.edu****or (614) 688-4685.**