Gunnar Erixon and MFD

Alternative approaches exist in the field of product structure research. One of these is the Modular function deployment (MFD) approach by Gunnar Erixon. In his dissertation, [Erixon 1998] Erixon defines modularity without the aspect of functionality:

“Modularisation = decomposition of a product into building blocks (modules) with specified interfaces, driven by company-specific reasons”

Erixon thus suggests company-specific reasons as the basis for decomposition. He divides these reasons into 12 categories that he calls module drivers. They are drivers, or motivators, for calling a certain building block a module.

The motivations for modularity, according to Erixon, are:

1. Carry over

Of these, 1 and 6 refer to decomposing the module into different uses, 2, 3, and 5 to the long-term management of the product, 4 to configuration, 7, 8, and 9 to manufacturing-related reasons, 10 and 11 to maintenance, and 12 to environmental reasons. By using the main criteria only, the number of real module drivers is six. Erixon divides the motivations for modularity as follows:

- Product development and design-related 1,2, and 3

- Variance-related 2 (referred to as “technical specification” on page 72), 5 - Production-related 6 and 7

- Quality-related 8 - Purchase-related 9

- After-sales-related 10,11, and 12

This division does not bring anything new to the matter, and it follows a rather formulaic conception of the issues (for example, associating separate testing to quality only). However, it is interesting that Erixon regards technical specification as belonging to configuration in particular!

Functional structure is thus linked to design; this is where Erixon admits to following to some extent the path of Pahl & Beitz and Ulrich.

In the modular function deployment approach, the starting point is in the part structure of the existing product. It is, therefore, not applicable as a design method for a new product sought in this dissertation. The ideas therein are, however, worth looking into. First, the (main) parts of the product are collected in the Module Indication Matrix tool. The horizontal rows show the parts, and the importance of various motivations for modularity is evaluated for each part as shown in the figure below. The matrix tool is based on calculating indices. The logarithmical division of priorities on a scale of 9, 3, and 1 is familiar from the Quality Function Deployment (QFD) methodology [Clausing 1994]. According to the author's critical view, no valid observation proves that the priority of the issues in product development would follow a logarithmical scale. Rather, a psychological factor is used here, that is, people are forced to decide between “important” and

“very important” issues. In practice, this division leads to the fact that the lesser weightings hardly make a difference in the analysis. For example, the figure shows that it has been decided that function carriers with 18 points are not module candidates, but those with 19 points are. The total number of points varies between 4 and 43, which means that the difference of one point cannot be decisive if any error margin is applied in the analysis.

FIGURE 48. An example of Erixon's Module Indication Matrix (MIM) presentation [Erixon 1998 p. 108]

The following phase in the MFD is the categorization of elements with shared motivations for modularity and potential decision-making on the merging of elements, as shown in the figure below:

FIGURE 49. Parts including the same motivations for modularity can be combined. [Erixon 1998 p. 110]

The experiences of the MMD methodology have proved that it easily leads to a large number of small modules and a situation in which the discrepancies between the motivations cause the analysis not to yield a functional result. Apparently aware of this, Erixon introduces “The House of Modular Function Deployment” matrix tool shown in the figure below. This tool also evaluates the effect of the different motivations on each other. A couple of empty rows are reserved for the case-specific reasons for modularity. For assembly, an optimal point is sought for the number of elements that together form the product, with the minimal elements and the sum total of the work hours in assembly (“Target number of modules” in the figure). In this figure, Erixon rather awkwardly resorts to functional structure -based thinking and presents subfunctions instead of element entities (or something of the sort, as it says “Sub-function / technical solution / function carrier”?) in the table. This resorting to the functional perspective separates the methodology from the ideas introduced previously.

FIGURE 50. Erixon's “The House of Module Function Deployment” tool in which the functional perspective is mysteriously included as a part of the method. [Erixon 1998 p. 113]

In his work, Erixon also comments on the organizing the part structure by evaluating the interfaces.

He takes the order of assembly as his starting point, as shown in the figure below. According to Erixon, there are two “pure” model types that rationalize the structure: using a basic unit and a layered structure. These ideas are somewhat based on the Design For Manufacturing and Assembly approach and they apply some of the same elements as the plus-modularity view. Erixon considers the retaining of the pure structure beneficial, as the interfaces between the modules remain simple.

This view is not completely universal, as it does not consider the dependencies irrespective of the assembly structure.

FIGURE 51. Erixon's ideas on limiting the dependencies between the interfaces in a modular structure. The sample case is still a vacuum cleaner. [Erixon 1998 p. 111]

Erixon's definition of modularity is worth examining, even if it does not lead us one step forward from Borowski's definitions (except that Borowski did not use the word “module”). Defining the modules on the basis of the motivations for modularity is also an approach with an industrial relevance. The problem with this approach is the lack of a theoretical background. On what basis are the 12 reasons for forming a module chosen? They are not based on any models of mechanisms of the market or the product, but represent knowhow acquired in practice, which probably also corresponds to today's predominant views. If, for example, we think about the “recycling”

motivation, we notice that it would not have been mentioned 20 years ago. Based on this, we may assume that in 20 years' time, new “trendy" issues may exist on the lists while some others may have been dropped out.

The value of this approach would increase considerably if a model from which the module drivers could be derived was presented. Such a model is presented in Chapter 9 of this dissertation.