PLM: Modular Design: What really is that?
The market needs increasing variety: products that are more specifically adapted to unique customer needs. That means more engineering costs, but those were actually already too high. The solution is sought by more and more companies, especially those of the type engineer-to-order, in modular design in combination with a product configurator.
A product configurator is a piece of software that allows to specify the needs of a customer in terms of choices for predefined options. Based on this, the configurator chooses variants for certain components and then generates the specific BOM for the ERP system, as well as the other documentation for production. This sounds ideal: the seller fills in the choices and the unique product is automatically generated, without the need for another minute of engineering work. Some even go so far as to allow the customer to fill in their choices on the website, so that neither a salesperson is needed.
It sounds ideal, but of course there are extra costs. A modular design needs to be made for the whole product program. First it is necessary to determine how the product is divided into modules and which options are accommodated in which modules. All variants for each module must be fully specified in advance for each permitted combination it must be checked and tested that everything fits and co-operates effectively. In addition, a number of components can be defined, but never chosen. That is a waste of time. Furthermore, the design of all modules must be maintained. Improvements are constantly being proposed, better buying parts become available, or certain parts can no longer be supplied. How many of the versions generated this way are followed up by a new one without ever being applied? And then there is another problem: if you change something in a certain component, it can affect other components, which should also be adapted. Tricky is that the interdependencies are not always documented. They are only discovered during production (or later). These are costly changes.
How do you prevent that the additional costs for customer order-independent design of the modules, are greater then the savings on customer order-specific engineering. Hence the title: What really is modular design?
A modular design divides the system into smaller modules that can be designed independently of the whole and can be assembled in different combinations to create different systems. Essential properties for modules are exchangeability and independence. Interchangeability means that you can replace a module with a variant, without having to adjust anything elsewhere in the system. Independence means that you can design and change a module without the need for knowledge of the rest of the system.
The problem is in the interfaces between the modules. The modules have different dimensions and must be attached to each other in a certain way. In addition, forces intervene between modules and energy/movement/information is exchanged. That should all fit and work. The difficulty of determining the contents of an interface is evident from the example of a plug and an electrical outlet. Of course the pens must fit in the holes, but that’s not all. The voltage and frequency of the network must match that of the device that is to be connected. Furthermore some devices (e.g. dimmers) can cause high frequency interference signals that the net does not want to have. Conversely, the net can also contain signals, which the device cannot withstand. In short, how do you know for sure that there is no hidden misery outside your interface?.
The above is an electrical example (that’s how I was trained), but don’t think the mechanical world wouldn’t be bothered by it. I once came across a nice example in the design of a car. In this example, the gearbox and the rear axle can be considered as modules. The dimensions of the holes in the flange are not close to the complete specification. There are also maximum torque and speed. The maximum power may be important if the temperature of the differential is critical. However, another fatal interaction popped up during operation. When shifting gears a torque peak may occur because the flywheel has to be retarded to match the new lower speed of the primary axis of the gear box. The shorter the switching time, the higher torque is transmitted to the wheels. That gives a short-lived peak in the torque that is much greater than the maximum torque. That became a problem when the iron axles were replaced by lighter carbon fiber shafts. The iron axles had sufficient elasticity to absorb the peak. The carbon axles were stiff and broke. A typical example of an unforeseen error due to an ignored specification in an interface.
The example shows that modular design is more than simply cutting into parts. Interfaces are more than physical interfaces. In the next episode I will elaborate on the definition of interfaces and the control of interchangeability and independence.
This is an translation from the original Dutch text that appeared as a column in magazine Constructeur, published by MYbusinessMediaB.V. in the Netherlands.