|
|
www.johnstark.com |
|
|
|
Multiple relationships As well as working with the data, engineers and other users of engineering data are interested in the relationships between data. As well as managing data it is also necessary to manage the relationships between data. There are many types of relationships. There are relationships, for example, between products and parts, between parts and data, between one part and another, and between parts and workflow. There are relationships between the parts of one product and the parts of another product. There may be relationships between apparently separate projects. There are hierarchical relationships linking parts to a product. Bills of Materials, parts lists, assembly drawings and where-used lists contain information on such relationships. The various types of data (such as specifications, drawings, models, test results) supporting a product need to be related. A component may only fit a particular product, an engineering drawing corresponds to a particular part, an NC program corresponds to a particular version of a part. There are relationships between activities. The design of a particular part may only be started when the design of other, related parts has been completed. Data needs to be linked both to its source and to derived data. It also needs to be linked to the processes that create and use it. Users need to be able to navigate through the various relationships and links. In addition to the wide variety of relationships in existence, added complexity arises due to changes that occur in relationships as the engineering process takes place. For example, a part that was previously used in three products may, in future, be used in a fourth product. A part that was originally expected to be made of metal may instead be made of plastic. Similarly, the relationship between logic changes and Printed Circuit Board (PCB) designs needs to be maintained so that changes in the logic are reflected in the board. Parts of a design may be due not to structural or aesthetic reasons but to the manufacturing process that will be used. A part that has been designed to be cast may have features that are completely unnecessary if in future the part is to be machined. Information needs to be maintained on the reasons why choices were made between competing designs. If the selection criteria can be retained, they can be used to help make better choices in the future. If the context within which a particular design was chosen is clear, it may be obvious how the part should be made. Information concerning the completeness of a design needs to be maintained from the viewpoints of functionality, constituent parts and necessary activities. If a particular part is replaced, it should be possible to identify what is needed to maintain completeness. A 'missing' relationship may be as important as a 'present' relationship. Comparative relationships such as 'duplicate' are needed, for example, to prevent development of identical products. In the file-based environment, it is often difficult to maintain information on the almost limitless relationships between information in different files. For example, one file may contain analysis results on a part in a second file that was subjected to forces specified in a third file. The second file may also contain the geometry model of another part. A dimensioned drawing of this part may be stored in a fourth file. When the second part is modified, a fifth file may need to be created with the corresponding dimensioned drawing, but because the first part may not have been modified, it may not be necessary to repeat the stress calculation. Probably the file will contain no indication of whether or not the first part was modified, so the stress calculation will be repeated, generating a sixth file that is identical to the first file. In the real environment, in which there will be thousands of files, it is practically impossible to maintain the correct information on relationships between files. Information on the procedures used to develop a product needs to be related to the product and to the activities. Similarly, information on the procedures that should be followed for new products should be linked to the activities and the files. In the absence of links between the information and the process that creates it, it is impossible to get real control of the environment. This is true from the technical point of view where information concerning relationships or selection criteria for alternatives is lost. It is equally true from the management point of view, with the result that the engineering function does not operate as efficiently as it could. Meaning of data Raw data often has little meaning, and is of limited use, yet, in the traditional engineering environment, it is often only raw data that is stored. Typically, the data that is maintained on a product will include CAD/CAM files, some drawings and some information on the manufacturing process. Little information will be maintained on user needs, the process that was used to develop the data, or the choices that were made to get to this data. Most of the information will be forgotten. The next time that similar activities are carried out, it will not be possible to benefit fully from past experience, and development will start again from scratch. Drawings and CAD/CAM files are both examples of raw data. They do not provide complete definitions of the product, the process used to develop the product, or the various choices and activities that took place during the design cycle. A lot of information is discarded during the design cycle. For a particular phase of the design cycle, this may not be a major problem since the people involved can remember what has been discarded. It becomes a problem, however, when the data is transferred to other users in another part of the design cycle, since they will not be aware of what has been discarded. It is also a problem when an attempt is made to reuse information at a later date. By this time, the original users of the data will have forgotten what they discarded. Time will be lost as they try to find missing data, or develop new data to replace missing information. Unless users understand the full meaning of data they can not make the best use of it. The information that could be of use to them needs to be available to them. With current systems, this is rarely the case. In the typical engineering environment, a lot of the raw data that is created is soon discarded because there is little chance of being able to use it profitably again. Users need support from management systems so that they will not discard potentially useful data. Knowledge-based systems could be used in this role. Knowledge-based systems may also be used to help 'renovate' existing, incomplete data. The company's existing data is valuable, but to unlock its value, users have to know its meaning. Unless the full meaning of data is available, it can not be fully used in the future. If data has to be re-created or renovated manually, errors will be introduced and time will be wasted. |