Integral Logistics Management — Operations Management and Supply Chain Management Within and Across Companies

7.3 Generative Techniques

Intended learning outcomes: Disclose the combinatorial aspect and the problem of redundant data. Present variants in bills of material and routing sheets as production rules of a knowledge-based system. Explain the use of production rules in order processing.

Generative techniques prove to be appropriate for production with many variants, that is, where there may well be millions of possible variants, but where the entire range of variants can be determined from the start, through a combination of possible values of relati­vely few parameters. Although each product variant results in a qualitatively different product, all stem from the same product family (see definition in Section 1.2.2). The production process for all product variants is principally the same. According to Figure, product families with many variants are strongly related to the “make-to-order” production environment. According to Figure, mass customization is the main production type.

7.3.1 The Combinatorial Aspect and the Problem of Redundant Data

Let us examine the problem using the example of a fire damper built into a ventilation duct, as shown in Figure In the case of fire, the damper automatically stops the ventilation that would promote the spread of the fire.

Fig.        Setting the parameters of the fire damper.

Because ventilation ducts must fit the building, fire damper manufacturers must be able to offer a damper for every conceivable cross-section of “width * height * depth.” To reach flexibility in achieving customer benefit, the damper is manufac­tured only to customer’s order. Only certain semi­­finished goods, such as side pieces, strips, and drive kits, are premanu­fac­tu­red in small-sized production according to frequently requested variants. There are approximately 30 to 50 bill-of-material positions. Product features such as damper type, height, width, depth, and type of connection profiles are called parameters.

The product solution space is the combined variety of the customization domains to form a desired commercial product variety.

Therefore, the customer can specify any combination of parameter values of the value domains of the defined solution space. A group of parameters like this determines the type and quantity of required components (such as sheet metal, strip, connecting, and drive materials) as well as the operations with regard to production facilities and setup and run time (or load), and even with regard to description (for example, the num­ber of fastening holes and the distance between holes in the connection profiles). Actually, four types of dampers are offered with widths from 15 to 250 cm and heights from 15 to 80 cm. With measurement increments of 5 cm, there would be 2688 variants (4 * 48* 14). A free combination of parameter values results in a theoretical number of tens of thousands of variants. The number of different dumpers in the practical world reaches thousands.

Let p(i) be the parameter of i (for example, type, width, height, depth, options, accessories) and ∣p(i)∣ ≥ 1 be the number of possible values of the parameter p(i). Figure shows the formula for the number of theoretically possible combi­nations. Of these, each has a bill of material and a routing sheet and differs — as a whole — from all others. A certain compo­nent can, however, be used to build many of these combination possibilities.

Fig.        Number of possible combinations with n parameters.

For the fire damper, let p(1) and p(2) be the parameters width and depth. As a semifinished good independent of the other parameters, sheet metal pieces are cut to a width of 800 mm and a depth of 240 mm. This item is used as a component in all bills of material for dampers having a width of 800 mm and a depth of 240 mm. The number of these bills of material is calculated according to the formula in Figure

Fig.        Example for number of identical bill-of-material positions.

All bills of material and routing sheets for the product family are similar. Their being nearly identical is typical of this type of production. If you were to keep a bill of material and a routing sheet for every single possible combination of parameter values, the greater part of the stored data would be redundant.

Classical aids to product configuration with the business objects item, bill of material, routing sheet, and work center (see Section 1.2, or as detailed objects in Section 17.2) do not allow the definition and storing of parameters and dependencies.

In such traditional systems, it would be possible to derive the variant from a “parent version” using adaptive techniques according to Section 7.2. How­ever, with very many positions on the bill of material and many operations, this would place a heavy burden on qualified employees. This is not feasible for products with low value added. If, however, bills of material and routing sheets were created from the start in all their possible combinations, e.g., as one-dimensional variant structures (see Section 7.2.2), the multiplicative explosion of quantities of the positions on bills of material and routing sheets to be saved as data would make relocating efforts enormous and unfeasible. Engineering change control (ECC) for these thousands of bills of material and routing sheets would be highly problematic.

Exercise Bike Factory

7.3.2 Variants in Bills of Material and Routing Sheets: Production Rules of a Knowledge-Based System

The key to a solution is to extend the business objects by adding a suitable representation of the knowledge about when certain components are built into a variant of a product family and when certain operations become part of the routing sheet. This is accomplished by implementing knowledge-based information systems. For a detailed description of these tools, see Section 17.3.1 and [Schö88b]. For the sake of simplicity, let us explain these systems using our introductory example of the fire damper.

From the perspective of product design, a product family is a single product. For example, there is one single set of drawings for the entire product family. There is one single corresponding bill of material, and it contains all possible components (such as raw materials and semifinished goods) just once; in similar fashion, the single routing sheet contains all possible operations listed just once. By inserting tables or informal remarks, the documents will indicate that certain components or operations will occur only under certain conditions. This characteristic is expressed in design rules or process rules.

 A design rule is a position of the bill of material that is conditional as specified by an if-clause, which is a logical expression that varies in the parameters of a product family. 

A process rule is a position on the routing sheet that is defined analogously.

Following these definitions, the rules in the fire damper example may be structured like those in Figure

Fig.        Design or process rules.

A position of the bill of material or the routing sheet thus becomes a production rule in the actual sense, that is, of a product to be manufactured. These rules are applied to facts, such as the data on item, facilities, and work center in the production database or on parameter values in a query (for example, for a current customer order for a specific product of the product family).

Product designers and process planners in the company function as experts. When they put their rules on paper for variant bills of material and variant routing sheets, they use — unconsciously — expressions that are very similar to production rules. It is evident that these are experts expressing expert knowledge, for no two product designers will deliver precisely the same design for a particular product. In the same way, two process planners will seldom produce exactly the same routing sheet.

The users of the system are those persons who release, control, and produce the orders. The exact realization of production rules in an information system is treated in Section 17.3.2.

7.3.3 The Use of Production Rules in Order Processing

Figure shows an excerpt from the product structure of the fire damp­er in Figure This part of the bill of material lists some attributes and if-clauses important to an understanding of production rules.

Fig. Excerpt from the parameterized bill of material for the fire damper.

For the query, the facts — the product identifiers, order quantity, and all parameter values — have been added. Through comparison of these facts with the rules stored for the product family, program logic determines for each position the first variant of the bill of material or routing sheet for which evaluation of the rule results in the value “true.”

Try the following exercise: In Figure, what variants are selected, given the following parameter values: Type = 1, drive = left, width = 400, height = 120?
Solution: Position/variant: 130/01, 150/01, 155/01, 160/01. Also compare the exercise in Section Scenarios and Exercises [Course 7]

Storing parameterized positions on the bill of material and routing sheet in the form of production rules has key advantages over conventional positions. Each potential position is, in one comprehensive, maximal bill of material or in one comprehensive, maximal routing sheet, listed exactly once, but it is listed together with the condition under which it will appear in a concrete order. This means that there is no longer the stored data redundancy found in the classical case without parameterizing. In terms of the combinatory aspect, rather than having a storage problem growing multiplicatively, we now have just additive increases. For a detailed comparison of data storage complexity, see [Schö88a], p. 51 ff.

Figure shows actual, rounded comparative numbers for the data storage necessary for the fire damper in our example.

Fig. Comparing data storage complexity for the fire damper example.

With minimal data storage problems, any number of orders with all possible combinations of parameter values can be transposed into production orders in a simple manner. One only needs to enter the values of the parameters. All these orders contain the correct components and operations, each with correctly calculated attribute values. Moreover, all possible combinations have been defined previously and automatically. Engineering change control (ECC) is also simple. If, for example, a new component is introduced, with a typical bill of material mutation, the component identification is added as a position to the (unique) bill of material. If it is a variant, its use dependent on parameters will be given an if-clause. Quali­fied employees familiar with the design and pro­duct­ion process perform all of these tasks.

There is an advantage to the use of knowledge-based product configurators when PPC software is used in connection with CAD and CAM. With CAD, only one unique drawing is produced for all variants, but as above, it is parameterized. Within CAM, there is also only one unique, parameterized program controlling the machines. With this knowledge-based represent­ation, PPC also now keeps only one unique bill of material and routing sheet for all variants. If there is a suitable, parameter-based CAD program package, a parameterized bill of material with a drawing can be exported from CAD to the PPC software. More important, however, is the reverse direction with an order. The parameter values of the production order can be exported from the order to CAD in the bid phase (or at the latest at order release). CAD then produces an order-specific drawing. In prac­tice, this option is used in bids for products in the construction indus­try, for example. In analogous fashion, linking an order to CAM means that the same set of parameter values can serve as input to a CNC program.

And, finally, the generative technique is used successfully in the service industries, such as in the insurance branch and in banking. A family of insurance products can be seen as a product with many variants. Here, again, we find a clear case of nonrepetitive production. The setting up of a policy, or order processing, is at the same time the production of the product. The parameters are the features of the insured object as well as the types of coverage to be provided (e.g., sum insured, excess, type of compensation). The production rules of the configu­ra­tor assign the elementary products to possible contracts. Concrete entry of a set of parameters ultimately yields a concrete insurance policy and includes all calculations, particularly the premium. Here see [SöLe96]. Those readers interested in an application in banking or in uncertainty may wish to refer to [Schw96].

Course sections and their intended learning outcomes

  • Course 7 – The Concept for Product Families and One-of-a-Kind Production

    Intended learning outcomes: Produce logistics characteristics of a product variety concept. Explain adaptive and generative techniques in detail. Describe the use of generative and adaptive techniques for engineer-to-order. Differentiate various ways of cooperation between R&D and Engineering in ETO Companies.

  • 7.1 Logistics Characteristics of a Product Variety Concept

    Intended learning outcomes: Differentiate between high-variety and low-variety manufacturing. Describe different variant-oriented techniques, and the final assembly schedule.

  • 7.2 Adaptive Techniques

    Intended learning outcomes: Explain techniques for standard products with few variants as well as techniques for product families.

  • 7.3 Generative Techniques

    Intended learning outcomes: Disclose the combinatorial aspect and the problem of redundant data. Present variants in bills of material and routing sheets as production rules of a knowledge-based system. Explain the use of production rules in order processing.

  • 7.8 Scenarios and Exercises

    Intended learning outcomes: Apply adaptive techniques for product families. Disclose the use of production rules in order processing. Elaborate the setting the parameters of a product family.

  • 7.9 References


Print Top Down Previous Next