eddyQlab Ai: Innovative Lab Solution for Predicting Case Depth and Hardness on Induction Hardened Parts
The eddyQlabAi is ibg's latest innovation in non-destructive testing. It is designed to predict case depth and surface hardness of induction hardened parts, such as CV joints, with exceptional precision. By combining pre-trained data with powerful machine learning, the eddyQlab greatly simplifies the testing process and reduces complexity in the lab. The semi-automated eddyQlab Ai takes it even a step further: it delivers reliable numeric predictions of surface hardness and case depth at multiple positions of the part, indicated in both Vickers and Rockwell, without the need for destructive methods.
Revolutionizing Lab Testing with Ai-powered Case Depth Prediction by ibg: Fast. Precise. Non-destructive.
Benefits of Testing with the eddyQlab Ai
✓Cost-saving: destructive testing costs up to 250 € per part and requires experienced personnel to carry it out
✓ Reliable and precise: the large volume of data used in testing with eddyQlab AI significantly reduces errors associated with human-dependent destructive testing
✓ Fast: Destructive testing takes approx. 45-90 min incl. preparation per part, while testing with the eddyQlab takes less than one minute per part (depending on the stem lenth and positions to be tested).
Ai Software
The model is deliberately disabled to ensure that each model - regarding both its architecture and training data - continues to produce stable, consistent, and comparable results over time, without the risk of drift toward incorrect or non-reproducible outputs. The unit requires signing a separate license agreement (SaaS) for the AI software, with a 3-year minimum term and a consecutive annual renewal of the software agreement.
Feasibility and Ai Training
Feasibility Study
For predicting the surface hardness and case depth on a new part, the customer sends 2 sets of 14+1 induction hardened parts from the same batch. One of the sets has NOMINAL hardness values, the other set of induction hardened parts has MINIMAL IN hardness value. From each set, the customer tests one part using a destructive testing method. The values (case hardness/core hardness/surface hardness/hardness limit at every location that is to be tested) are then sent to ibg's Application lab in Ebermannstadt for conducting the feasibility study.
Ai Training
If the results from the feasibility studies were not sufficient, the customer ideally provides 5 sets of 14+1 parts (min. 3 sets of 15 parts; required: MAX IN value, NOMINAL value, MIN IN value; optional: MIN OUT value, MAX OUT value) to conduct the Ai Training at the Application lab in Ebermannstadt. All induction hardened parts MUST be from the same batch. Of these sets, the customer tests one part of each set with a destructive testing methods and shares the results (case hardness/core hardness/surface hardness/hardness limit at every location to be tested) with the experts at the ibg Application lab in Ebermannstadt. After the Ai Training, the Ai model is ready for predicting the surface hardness and case depth on the provided part type with the trained data.
Ai-Based Eddy Current Testing and Destructive Microhardness Testing
The eddyQlab AI is the solution for significantly reducing and eventually replacing the need for destructive testing of induction hardened parts. To efficiently train the model, saving money, reducing waste, and streamlining workflows, you can read or download the following whitepaper for detailed insights.
The Ai Model by ibg is designed to compensate for differences in lots, so it works cross-batch. The Ai model uses the rich data provided by ibg´s iPMFT (24 frequencies / 8 fundamental, 3rd and 5th harmonics), evaluates the eddy current signals and predicts the case depth and surface hardness of inductive hardened parts.
FAQ eddyQlab Ai
The software of the eddyQlab Ai VL goes a step further than just testing induction-hardened parts, it provides a numeric prediction of the case depth and surface hardness without the need to destroy parts.
For best results, please send two sets of 15 CV joints from the same batch to the Application Lab in Ebermannstadt. Our experts will conduct a feasibility study on these samples. After a successful feasibility study, our sales partners will contact you for further proceedings. If the feasibility study has not been sufficient, further Ai Training is needed. For this, send five sets of 15 CV joints of the same batch to the Application Lab in Ebermannstadt. Afterwards, you are again contacted by our Sales Partners.
No, only in case these differences would be large and influential. Eddy current testing inherently provides results that represent the entire volume of material within the signal-taking coil. The magnetic flux density uniformly permeates the test area in all directions. As a result, any variations in microstructure distributed across the diameter are effectively averaged in the eddy current response, resulting in a stable and consistent signal.
Apart from the equipment provided by ibg (Test Unit, Control Unit, Cable, and Base Plate), customers will also need to supply the following:
- Keyboard
- Mouse
- Monitor (recommended resolution: 1920 x 1080 px)
Yes, you would still need to perform standardized (destructive) hardness testing methods, as those are standardized. Nevertheless, these would have to be performed significantly less, e.g. when new parts are settled, when the heat treatment part for already registered parts has been changed or when the steel type of parts has been changed.
The eddyQlab Ai does not require any sample preparation. There will also be no waste, as neither the parts are destroyed nor any chemicals are involved.
No, you do not need any knowledge on eddy current, as the results are directly displayed in Vickers or Rockwell in the HMI.