Automated Configuration of factories based on the configura- tion of products to be manufactured.

Method for integrated modeling of products and factories for smart configuration of a production environment by use of computer programs .

Increased variability in product features and higher flexi ^¬ bility for the customers during customization process implies increased complexity of the production facilities and en ^¬ hanced fragility in manufacturing processes. Product portfo- lio management in the future must therefore also consider the variability of factories that are capable of manufacturing the configured instances of the products.

Today, customers demand highly customizable products, e.g., a fire detector where the customer can choose the kind of sen ^¬ sors, case shape and colors, types of alarms, etc. This means the production facility for manufacturing the fire detectors must be reconfigured every time a new variant of the product is demanded by the customer. It remains a big challenge for factory operators to automatically reconfigure a factory to produce low volume products, in the worst case this means producing single instance of the product (one of a kind) .

This problem has been identified as one of the key aspects of industry 4.0.

It is a goal of the invention to show a solution for this problem. According to the invention, this goal is reached with the method of claim 1. Preferred embodiments of the invention are shown in figures 1 -3.

The key idea of the invention is a new engineering methodolo ^¬ gy called "ProFacto Engineering", where the product and the factory are modeled in sync.

As depicted in Figure 1 and described in more detail in the following, three configuration activities are relevant in this context.

• Product Configuration by end-user: The customer is presented with a set of decisions, on the basis of which a valid combination of the product features is generated, which rep- resents the end-product wished by the customer.

• Factory Configuration to support a Product Line of prod ^¬ ucts; Manufacturing a new product typically means changing the physical setup of the factory (adding new devices, re ^¬ organizing the layout, defining new workflows, etc.) . As the efforts of changing the physical setup is considerably high, it is not realistic that a factory can produce any kind of product at any time. Instead, a factory is configured such that it is capable of producing a family of related products. Production Configuration to support a Product Instance; After the end-consumer has selected the features required for her product (e.g., flame sensor and smoke sensor in a fire detec ^¬ tor) , the factory that was previously prepared for the corre ^¬ sponding product line can then be configured {modified) to produce exactly the required variant. This step is typically a Software configuration problem - e.g,, adapting the material flow, changing automation Scripts, configuring speeds, etc. Apps and Services are therefore seen as the domain arti ^¬ facts, which can be reused in this configuration step.

In order to be able to pass on the configuration Knowledge from one configuration process to the other, the models in the background must be coordinated. In particular four model- ing activities are involved In this process, The first step of the process is domain engineering for factories as shown in figure 2. Domain Engineering of Factories: Domain engineering is responsible for establishing the reusable platform and thus for defining the commonality and the variability of the product line. In the context of smart factories product lines, the domain assets are developed by factory component vendors and made available to the factory Operators through a marketplace for hardware and Software components. Domain engineering ac ^¬ tivities consist of product management, requirements engi ^¬ neering, design, Implementation and testing of the hardware and software components required for a factory setup, Model- ing the components can include geometrical models of ma ^¬ chines, variability models of devices and Services, compli ^¬ ance to Standard Interfaces, and the contribution of the de ^¬ vices to non-functional properties of the factory, e.g., the extent of boosting productivity, reducing lead times, opti- mizing inventories, increasing the availability of resources or promoting flexibility.

Functional description of the components such as the features of robots, assembly stations, transfer systems, test stations or conveyer systems must also be modeled, so that these can be discovered by factory operators.

Domain Engineering of Products and Application Engineering of Factories; Smart products determine the necessary configura- tion of the factory. In our envisioned Profacto Engineering approach, product lines of goods to be produced guide appli ^¬ cation engineering of the factory. This step is referred to as Application Engineering Factories. In order to distinguish it from a succeeding step where the factory has to be opti- mized for one particular instance of the product. Domain en ^¬ gineering for products (goods to be produced) is similar to the traditional approach followed in product line engineering - except that in smart production, product variability model- ing must also consider the requirements for the factory capa ^¬ bilities. This means, based on the smart variability models of the products, it is possible to determine the initial lay ^¬ out of the factory and plan the production processes (which allow for some variations) . Application Engineering Factories is thus the process in which the factories are

built/reconfigured by reusing domain artifacts (factory hard ^¬ ware and Software components) . The requirements for Applica ^¬ tion Engineering for factories can be derived from product models.

• Application Engineering for Products + Application Engineering for Production: Application engineering is the process of Software product line engineering in which the appli ^¬ cations of the product line are built by reusing domain arti- facts and exploiting the product line variability. End-users (customers) specify their requirements (either by feature se ^¬ lection or in some other more sophisticated forms) and the product sellers have the task of building the requested vari ^¬ ant. For a seamless production planning, the customized prod- uct instance must be able to guide the final application en ^¬ gineering process for the factory.

These different engineering activities result in different kinds of models as shown in Fig. 3, which are the basis of the whole process.

• Component Models and Factory Models: Component models are representations of production facilities like robots, conveyers, 3D printers, sprayers, etc. Each component is a product itself, often with a complex and highly configurable hardware and software structure. Traditional modeling tech ^¬ niques, such as feature models or other variability modeling languages, can be used for such specifications. A factory consists of various hardware and Software components from different vendors. The specific models of the components of a factory are a main part of the factory model. Additionally, the factory model contains knowledge about the properties of the different components, their interplay, their locations, their maintenance states, etc, Thus, the factory model repre- sents the current production capability of the factory. In our simplified fire detector example, a factory must have available components for assembling the different kinds of sensors, for 3D-printing of the housing, for painting the housing with a user-specified color, and so forth. It shall be noted that often a factory is able to manufacture products of different product lines and kinds, so only a subset of the factory components in a very particular configuration is usually needed for a specific product line. Depending on the size of the factory, even parallel production of completely different product lines is possible.

• Product Models and Product Line Specific Factory Config ^¬ uration: In a traditional RLE environment, the product model represents all possible product variants, but is agnostic about the production facilities necessary for manufacturing a product instance. E.g., in our fire detector example, the product model contains specifications of all the different, offered sensor technologies, constraints about the housing shape (e.g., the 3D format of the shape specification file), or a set of possible colors for the housing. In our Profacto Line, this model is enriched by those parts of the factory models which either influence factory configuration or which potentially reduce the variability of the product line, be ^¬ cause the factory is not able to build certain variants of the product (e.g., 3D printers have limited size, which re ^¬ stricts the possible dimensions of printed objects) . For ex ^¬ ample, our fire detector model will - in addition to the product specifications - contain concepts about the assembly line, the 3D printer, the paint-spray line, and all their relevant properties. The smart product line model can now be used for configuring the factory, which must be equipped and arranged In such a way that all or as many as possible dif ^¬ ferent product variants can be manufactured. Such configura ^¬ tions will also contain process specifications which deter- mine which parts are produced/assembled in which order and under which additional, non-functional requirements (e.g., quality gates, random sample tests, etc.). • Product Configuration and Product Specific Production Configuration: Based on the product line model, a configura ^¬ tor is provided and used by the consumer to customize her variant of the product. It shall be noted that, although the product model contains production knowledge, the end-user usually won't be directly confronted with production-specific decisions. Production issues will only be manifested in re ^¬ strictions on the offered product variety. If there is no factory setting which supports a certain product variant, this variant must be removed from the product model or, at least, it should not be offered to the customer. So we see dependencies in both directions, from production to product and from product to production. The product configuration will have influences on the configuration of the production site. From the underlying smart product model, a product in ^¬ stance knows its requirements imposed on the production. This Knowledge is used, ideally in an automated way, to reconfig ^¬ ure the factory's production line for that product. For in ^¬ stance, the 3D dimensions of the specified fire detector housing may demand the usage of a special SD printer; the se ^¬ lection of a specific housing color will induce the change of the paint cartridge of the paint-spray line. It is also imag ^¬ inable, that special product variants and their production requirements may lead to the selection of an alternative fac- tory.