Geolus Shape Search in Teamcenter: Reduce Duplicate Parts and Accelerate Design Reuse

The Problem with Metadata-Only Part Search

In most manufacturing PLM environments, the number of part numbers grows faster than the ability to govern them. Designers create new parts that already exist — just under a different number — because there was no reliable way to know.

The traditional approach to finding an existing part in Teamcenter relies on querying metadata — material type, nominal size, shape description, owner group, lifecycle status. This works reasonably well when the metadata is complete, correctly entered, and consistent. In practice, it rarely is.

The result is part proliferation. Engineers create new geometry for components that are structurally identical or functionally interchangeable with parts already in the library. This drives up manufacturing cost (more unique parts means more tooling, more procurement lines, more supplier management), increases BOM complexity, and fragments product knowledge across the organisation.

The root cause is not poor discipline — it is a tooling gap. Finding a part by shape, rather than by text metadata, was historically impractical. Geolus Shape Search, integrated with Teamcenter, closes that gap.

Who encounters this problem

Design engineers who need to reuse existing components but cannot locate them quickly enough to justify the search effort over creating new ones
PLM administrators dealing with growing volumes of duplicate or near-duplicate parts in the item master
Procurement and cost engineering teams who want to consolidate part families and reduce supplier line count
Standardisation and libraries teams responsible for preferred parts catalogues and design reuse governance

What is Geolus Shape Search

Geolus Shape Search is an AI-powered search engine developed by Siemens and integrated into the Teamcenter PLM platform. Its core function is to enable engineers to search for parts based on their 3D geometric shape — without needing to know the part number, name, or any metadata about the part they are looking for.

Where a traditional Teamcenter query requires the user to know something textual about the target part, Geolus works from geometry alone. A designer uploads or selects a 3D model, and Geolus returns a ranked list of parts in the PLM library ordered by shape similarity — from exact matches through to parts that share a similar form factor or topology.

Primary value delivered

Design reuse: Designers find existing parts that match their requirement before creating new geometry, reducing the volume of unique parts entering the system
Part consolidation: Teams can identify groups of near-identical parts that have accumulated over time and rationalise them into a single standard part
Faster time to design: Instead of building from scratch, designers use an existing verified part as a starting template and modify as required — preserving existing manufacturing and quality data
Reduced procurement complexity: Fewer unique parts means fewer supplier lines, fewer tooling setups, and simpler BOM management

Supported 3D file formats

Geolus Shape Search reads JT, VRML, and STL files for shape indexing. JT is the primary format in Teamcenter environments given its role as Siemens' standard lightweight 3D format for visualisation and data exchange. Search results can be further refined using standard Teamcenter metadata filters — material, where-used, cost, lifecycle state — allowing engineers to narrow results to parts that are both geometrically similar and contextually appropriate.

Geolus Shape Search results panel in Teamcenter Active Workspace showing ranked similarity results and metadata filters — Screenshot: Geolus Shape Search results panel in Teamcenter Active Workspace — showing ranked shape similarity results with geometric match scores and metadata filter options alongside a reference 3D part model.

How Geolus Shape Search Works

Understanding how Geolus processes and retrieves geometric data is important for implementation teams — particularly when configuring indexing pipelines and troubleshooting search quality issues.

Shape analysis and indexing

When a part is indexed into Geolus, the system analyses the 3D geometry and extracts a compact mathematical representation of the shape — sometimes described as a shape fingerprint or shape descriptor. This descriptor captures the essential topology and geometry of the part in a form that can be compared efficiently against millions of other indexed shapes.

The indexing process takes JT files output from Teamcenter and produces Geolus XML files containing shape descriptors, which are then loaded into the Geolus database. This data is stored separately from the Teamcenter master database — Geolus maintains its own optimised data store for shape retrieval, while the authoritative product data always remains in Teamcenter.

Search retrieval and ranking

When a designer submits a search query — either by selecting an existing part in Teamcenter or by uploading a reference JT/STL file — Geolus computes the similarity between the query shape and all indexed shapes. Results are returned in ranked order from closest match to most dissimilar, with a similarity score for each result.

The Geolus server maintains shape data and indices in memory for fast response. This means that for production deployments, the Geolus server requires adequate RAM to hold the active index — a sizing consideration that increases with the volume of parts in the library.

End-to-end Geolus search flow diagram from Active Workspace query through to ranked results from the Geolus server — Diagram: End-to-end Geolus search flow — from designer submitting a shape query in Active Workspace through to ranked results returned from the Geolus server, with the Teamcenter data layer indicated separately.

Combining shape search with metadata filters

Shape search alone can produce a long result list in large part libraries. In practice, the most useful searches combine shape similarity with metadata constraints — for example, finding parts that are geometrically similar AND made from the same material family, OR restricting results to parts that have been released to a specific lifecycle state. Teamcenter's integration layer passes both the shape query and any applied metadata filters to Geolus simultaneously, so the result set is narrowed before being returned to the user interface.

Geolus Architecture in a Teamcenter Environment

Geolus Shape Search is a multi-component system that sits alongside — rather than inside — the core Teamcenter application tier. Understanding the component boundaries is important for infrastructure planning, performance tuning, and troubleshooting.

Geolus architecture diagram showing Database, Server, and Indexer components in relation to the Teamcenter four-tier architecture — Architecture diagram: Geolus components (Geolus Database, Geolus Server, Geolus Teamcenter Indexer) in relation to the Teamcenter four-tier architecture — showing data flows between the Volume tier, Teamcenter Indexer, Geolus Database, and Active Workspace client.

1

Geolus Database

Data Store

The Geolus Database is a dedicated relational database that stores the compressed shape data, metadata, and index structures required for search retrieval. It is distinct from the Teamcenter database — the two systems share no tables and are managed independently.

Geolus supports Oracle and Microsoft SQL Server as the underlying database platform. The database stores three categories of data: compressed shape descriptors for each indexed part, metadata associated with each CAD part (used for metadata-combined filtering), and index structures that enable fast shape retrieval at query time.

When sizing the Geolus Database server, plan for storage growth proportional to the number of parts being indexed. Part shape data is stored in compressed form, so the storage footprint per part is modest — but in large industrial libraries with hundreds of thousands of items, it accumulates.

2

Geolus Server

Application Server

The Geolus Server is the application component that processes search requests. It connects to the Geolus Database to retrieve indexed data and responds to shape search queries from the Teamcenter integration layer.

For performance, the Geolus Server loads and maintains the shape index and associated metadata in memory. This means the server's RAM footprint scales with the number of indexed parts. For large deployments — particularly in automotive or aerospace environments with complex, high-variety part libraries — memory sizing is a critical infrastructure planning consideration.

The Geolus Server is deployed as a central application server component. It does not need to be co-located with the Teamcenter pool manager or web tier, but network latency between the Geolus Server and the Teamcenter application tier will affect search response times.

3

Geolus Teamcenter Indexer

ETL Component

The Geolus Teamcenter Indexer is the module responsible for populating the Geolus Database with part data from Teamcenter. It acts as an extract-transform-load (ETL) bridge between the PLM data store and the Geolus search engine.

The Indexer exports part shape data (as JT files) and associated metadata (as PLM XML or CSV) from Teamcenter, transforms this into Geolus XML format containing the shape descriptors, and then imports the result into the Geolus Database. This process must be run for the initial library population and then re-run (or scheduled incrementally) whenever new parts are added or existing part geometry is modified.

The Indexer can also be run in partial update mode to refresh only changed or new parts — important for keeping the Geolus index current in live environments without requiring a full re-index of the entire part library.

Geolus Teamcenter Indexer Run Modes

The Geolus Teamcenter Indexer supports three distinct run modes. Understanding the difference between them is important when planning initial deployment and ongoing index maintenance.

Run Mode	What It Does	When to Use
Run Mode 1	Exports part shape data (JT files) and metadata (PLM XML or CSV) from Teamcenter. No Geolus processing — output is raw Teamcenter export data.	When validating export quality before committing to full indexing, or when preparing data for offline Geolus processing.
Run Mode 2	Performs all Run Mode 1 steps, then additionally generates Geolus XML shape descriptor files from the exported JT data.	When you need to inspect Geolus XML output before loading into the database — useful during initial setup and troubleshooting.
Run Mode 3	Performs all Run Mode 2 steps, then imports the Geolus XML files directly into the Geolus Database. End-to-end indexing in a single operation.	Standard production use — full indexing pipeline for initial library population and scheduled incremental updates.

For most production deployments, Run Mode 3 is the standard. Run Mode 1 and Run Mode 2 are most useful during initial configuration and commissioning — they allow the implementation team to validate the data pipeline at each stage before committing the output to the Geolus Database.

Incremental indexing — re-indexing only parts that have been created or modified since the last run — is critical for keeping the Geolus search index current without incurring the overhead of a full library re-index. Establish a scheduled incremental indexing process as part of the operational runbook when going live.

Deployment Patterns and Practical Considerations

Deployment modes for the Indexer

Geolus can be integrated with Teamcenter in two primary deployment configurations depending on the environment and how Active Workspace is used:

Standalone mode: The Indexer runs as an independent process, exporting Teamcenter data on a scheduled or on-demand basis. Suitable for environments where indexing frequency is low and the part library is relatively stable.
Dispatcher-based mode: The Indexer is triggered via Teamcenter's Dispatcher framework, enabling event-driven or more frequent incremental updates. Preferred for environments with high part creation velocity or strict requirements for search index currency.

Access patterns: CAD tools, Teamcenter, and standalone browser

Geolus Shape Search can be accessed through multiple entry points depending on where in the workflow the designer is working. This flexibility is worth highlighting to engineering managers when presenting the capability:

From Teamcenter Active Workspace: The most common access point in current deployments. Search is initiated from within the PLM interface, results display inline, and a designer can act on a found part immediately without switching context.
From CAD tools: Geolus can be accessed directly from NX and other CAD applications via integration, allowing a designer to trigger a shape search from within their active modelling session — the highest-value integration point for design reuse.
As a standalone browser application: Geolus can be deployed as a standalone web application, independent of Teamcenter, for teams that need shape search capability without requiring a full Teamcenter client session.
Embedded in other applications: The Geolus search API can be embedded into other enterprise applications, enabling shape search capability within custom engineering tools or portals.

Key implementation considerations

Before deploying Geolus in a production Teamcenter environment, implementation teams should address the following:

JT representation quality: Geolus shape search quality is directly dependent on the quality and consistency of the JT visualisation data in Teamcenter. Environments with incomplete or missing JT files for parts will have gaps in the searchable library. A JT completeness assessment should be part of pre-deployment planning.
Index scope definition: Not all item types in Teamcenter need to be indexed by Geolus. Define upfront which business object types and product lines are in scope for shape search. Indexing everything indiscriminately increases indexing time and Geolus Database size without proportional benefit.
Memory sizing for the Geolus Server: As noted above, the Geolus Server holds its working index in memory. Work with the Siemens sizing guidance for your expected part library volume. Undersizing memory is the most common cause of degraded search performance in live deployments.
Incremental indexing cadence: Establish the appropriate frequency for incremental index updates based on your part creation rate and how current users expect the search index to be. Daily incremental indexing is typical for most manufacturing environments.
User adoption: Geolus shape search changes designer behaviour — it requires engineers to check the library before creating new parts. This is a process change as much as a technical deployment. Plan for user training and, ideally, a governance step in the design process that prompts shape search before new item creation is permitted.

Geolus Shape Search accessed from NX CAD showing in-session search panel with reference part and ranked shape-match results — Screenshot: Geolus Shape Search accessed from within the NX CAD environment — showing the in-session search panel with a reference part selected and ranked shape-match results displayed alongside the active model.

Key Takeaways

1

Metadata search alone is insufficient for design reuse. In real engineering environments, part metadata is incomplete and inconsistent. Geolus closes the gap by enabling geometry-first search — finding parts the way engineers think about them, not the way data entry happened to record them.

2

Geolus is a separate system that integrates with Teamcenter — not a Teamcenter module. It has its own database, application server, and indexer pipeline. Plan and size the infrastructure accordingly; do not treat it as a lightweight add-on.

3

JT data quality determines search quality. Before deploying Geolus, assess the completeness and quality of JT representation files in your Teamcenter environment. Missing or corrupt JT data produces gaps in the searchable library that undermine user confidence in the tool.

4

The Indexer run modes give you control over the ETL pipeline. Use Run Mode 1 and 2 during initial configuration to validate the pipeline at each stage. Run Mode 3 is the production standard. Establish a scheduled incremental indexing process from day one — a stale index erodes the value of the capability.

5

Shape search delivers the most value when embedded in the design process. Deploying Geolus as a technical capability is necessary but not sufficient. The real benefit comes when shape search is built into the part creation workflow — so checking for existing geometry becomes the default first step, not an optional extra.