2013-02-05

Siemens Shape-based Searching


Geolus Shape-Based Searching for Reusing CAD Parts

“Reuse” is a popular focus in initiatives across PLM today. It is an area where efficiency is improved by not repeating work and by making derivative products from existing ones, rather than starting from scratch.
It is one of the most basic benefits of computer-aided design (CAD) and computer-aided engineering (CAE). The more we leverage existing parts, the easier it is to maintain internal standards, freeing resources for innovation. Another incredibly important byproduct of reuse is that each time we reuse rather than re-invent, we slow the increase in complexity of our inventory.
Every part that is needlessly invented creates a ripple of waste throughout the product management lifecycle (PLM). There is no such thing as a cheap part.

Reuse is not limited to parts, but also extends to processes. If we can recognize a situation where a particular manufacturing process or CAE strategy can be applied to rather than reinivented, savings can be recognized.

This is great, in theory. The difficulty comes when trying to locate those existing parts for reuse. If a search for a desired part turns up nothing promising, then I suspect most of us would decide to just start recreating the design of the "missing" parts from scratch. 

We tend to find information through text-based searches, putting data such as part number into a search field; this data is known as metadata or attributes. However, looking for existing items using metadata can fail for a number of reasons, including:

  • The part might not have been filed with the correct attributes.
  • The company might have grown through mergers and acquisitions (M&A) which means there will be a cultural mix of attribute names.
  • Using suppliers' terminology will help communication with them, but it does introduce new cultures, just like M&A. This is compounded when dealing with multiple suppliers of the same part.
  • Design shops may span continents, introducing local language issues.
  • Spelling errors and irregular adoption of terminology and abbreviation will also be common.
If the expected metadata is not associated with the part in the first place, the search fails even when the needed parts are in the database. The only possible resolution in cases like this, discounting manual review of every file, is shape-bases searching. 

A Solution Through Shape Searching

One shape-based solution is Geolus, from Siemens PLM Software. Geolus has the ability to digitize “shapes.” It captured the essence of shapes as a digital index. It has the ability to compare indices and determine if a part is similar to other parts in the enterprise. Two good analogies are fingerprints and DNA. People’s fingerprints or DNA do not change when they change their hair, clothes, or names. Similarly, the Geolus index of a part is the same, no matter which keywords are associated with it. The index is unaffected by a part’s position or orientation. This is important when browsing assemblies for parts.
 

Geolus is a search engine that returns results in order of shape similarity and additionally reports the comparison of the size of the search part and the resultant part as a percentage. It can return exact matches as well as parts that are merely similar. 

For example, if production is halted because there are no more parts in Bin A, a search might find a source of alternate identical parts that enabled production to continue. A similarity search is appropriate if there is an initiative to identify parts on which production can standardize.

Searches can begin with a simple 3D sketch or an existing part, perhaps paired with some attribute(s). When a search returns a result, it reports a weighted list in order of similarity, along with identifying information, a report on the size compared with the original part, and a preview of each part. This image shows a set of search results on an existing part, as launched from a Web browser.


Example of a ‘seed’ image, used to search, and the associated results
(image provided by Siemens)

Notice that the parts are ordered by shape similarity and that, in this case, the defining attributes are also included in the results. The attributes are chosen by the customer and can include items such as release status, release date, owning engineer, project, supplier, cost, and so on. This collation, of non-geometric information through geometric means, facilitates queries such as “Which supplier has this part for the best price?”

How Geolus is used

Geolus is designed as a plug-in. It is accessed through your Web browser. It could also be integrated into existing software from Siemens PLM; such as Teamcenter, NX or Solid Edge. It is also available in aClass, a product from a Siemens partner, BCT.

The geometry that Geolus processes can be any one of the following formats: Siemens' own JT, VRML or STL. Attribute data can be in CSV, XML, or PLMXML formats. For example, I could sketch a bracket shape, and then specify that I am searching for one made of stainless steel and/or designed in a specified year and/or having a certain price.

JT, VRML and STL are all neutral file formats. Geolus can read a mix of these file formats into the same data depository. It does not matter if the originating CAD systems are varied, which makes it a powerful tool in a multi-CAD environment.
 

JT just became an ISO International Standard for 3D visualization, and is a lightweight format that can range from data-rich to very basic. Parasolid is the 3D modeling kernel used by MCAD and architectural software like SolidWorks, Solid Edge, and Vectorworks. D-Cubed is the library of software components licensed by Siemens PLM, and is used with many CAM/CAE applications, such as NX and Solid Edge.

On Siemens' Youtube channel, I watched the sketch of a part in NX, a high-end 3D CAD system from Siemens PLM. The selection of the desired part and the call for the Geolus search did not take long and previews of similar parts popped up quickly. From the results screen, I could see additional information reached by clicking on the 3D preview of each part presented. Through the search screen, I could even see how to replace the sketched part with one of the designed components. This appears to be a simple and easy way to swap out a "napkin sketch" for an established component.

When Geolus is used interactively it is typically used to search for matches to a single part. This answers the question “What did I do last time I had an item like this one?” When Geolus reporting is used, every part is compared with every other part and the results are exported to a spreadsheet, the manager’s tool of choice. This provides a finger on the pulse across the enterprise.
 

How Geolus Works

Geolus indexes parts and stores them in a database. A Geolus index contains:

  • the digital representation of the shape
  • metadata as specified by the user
  • thumbnail images for presenting visual feedback to the user
About Indexing

The Geolus Indexer is a small utility that is located with the data. (It is worth noting that you are never asked to move your data outside your firewall.) The Geolus Index is tiny compared with the original geometry. It does not contain enough information to reverse engineer the geometry. This could have been a security concern for multisite customers who choose to have a central Geolus server and so export the Geolus indices.


Strategic Uses

There are also other, more strategic, uses for the capabilities of Geolus to tackle in addition to Engineering and lifecycle management. Procurement, business analysis, and digital validation can also benefit from quick and easy access to shaped-based information.

Spreadsheets could be created to compare every item in the enterprise with every other item. This would allow bidding on the manufacture of groups of similar parts. These reports could conceivably make it simpler to detect which characteristics of parts are best used to identify units in catalogs and design specifications.

The ease of searching could help estimators to put together more accurate bids from available historic data, rather than depending on the memories of engineering staff.

Siemens PLM suggests that tracking reliability and maintainability of previously used parts could be another application of this system.


Make Full Use of the Technology

When I first took on the role of managing the archives for my hospital, dating back nearly 90 years at the time, I was continually frustrated by the misspelling of terms, and the renaming of buildings over the decades. Had I been able to sketch out the layout of a building or division for easy location, I could have found and distributed all of the available information quickly, rather than having to field verify or flip through thousands of drawings.

An ability to search by shape is something that I found fascinating about Geolus. Knowing that one could still make use of their company’s intellectual property, even if predecessors did not name or describe a part by the same standards we would use today, is quite empowering. As with any technology, implementation teams would need to ensure that the design staff are taking the time to perform the search and making the best use of all available resources.

If you would like to see this search in action, there are videos available on the Geolus channel on Siemens’ website. You can also find a couple on YouTube, they all appear to be from Siemens, such as http://www.youtube.com/watch?v=MCBg272eyRw For additional information: http://www.plm.automation.siemens.com/geolus


Melanie Perry is a facilities management CAD coordinator as well as a technical editor. She writes for the Mistress of Dorkness blog about CAFM and BIM. More...