WCX 2021: Digitizing That New-Car Smell

The ‘digital nose’ from Aryballe can put objective data behind materials olfaction, reducing costs, shortening development times and may spawn a global standard.

Digitizing olfaction could help engineers with the speed and consistency of developing pleasing automotive interiors. (Jeep)

The problem with “new-car” smells is the subjectivity of the sense. Nothing chafes at engineers more than subjectivity, which can wreak havoc in validation testing for both OEMs and suppliers. To vet the bouquet of materials in vehicle interiors, lab testing is most often augmented by panels of trained human noses, which judge the different components and the completed vehicle. At this year’s SAE World Congress (WCX 2021 ), French odor analysis specialist Aryballe, in collaboration with Asahi Kasei Plastics North America and the Digital Olfaction Automotive Consortium (DOAC), presented on a pathway to digitize and objectively rate automotive scents.

The current olfaction process involves both extensive lab testing of materials, as well as subjective ratings by trained human noses. (Aryballe)

When it comes to vehicle aromas, there is no global industry standard, explained Etienne Bultel, a development engineer at Aryballe responsible for designing and testing digital olfactory solutions. Each OEM has its own testing protocols, rating scales and methods he noted, and neither SAE, ISO or Germany’s VDA has yet managed to craft a global metric. The absence of “unified quantitative norms,” according to Bultel, “leads to increased costs and timing for suppliers, lab-to-lab variations in results and ultimately longer development times for the OEMs.”

The electronic nose

The heart of Aryballe’s digital nose is the TD-SPRi sensor, a coupled thermal-desorption based amplifying device which can increase the intensity of a compound’s olfactory signature up to 400 times. (Aryballe)
From sampling with its sensor, Aryballe are able to derive a Clustering Quality Score (CQS) based on intra- and inter-cluster ratios, providing a metric for discrimination performance between compounds. (Aryballe)
The numerical values derived from Aryballe’s CQS lets the data be charted visually to help determine measurement metrics. (Aryballe)
Aryballe’s Etienne Bultel notes that digital olfaction technology has advanced enough to differentiate between compound variants of the same material and detect odors in the automotive domain based on the current ISO 12219-7 and VDA 270 standards. (Aryballe)

The solution, obviously, is to quantize vehicle aromas. In his WCX presentation, Bultel advocated that, “the advances in sensors, biochemistry and machine learning, odor detection and classification can be achieved with the use of a digital olfaction device, or ‘electronic nose.’” Bultel claims the system he’s helped develop at Aryballe can identify different plastic types, differentiate between compound variants of the same material and detect odors in the automotive domain based on the current ISO 12219-7 and VDA 270 standards.

What makes much of this work so challenging, Bultel noted, is odor testing involves human sensory and perception systems tied to subjective feelings, and no two people react the same way. “Odors may be tied to an experience or memory, good or bad,” he said, noting that there are differences across regions and cultures of what is considered ideal. Neutral odor may be desired by certain OEMs but not others.

“Finally, some questions are also behind the fact that all the process steps in the manufacturing can contribute,” he added. “Sometimes it's hard to put a finger on it and say, ‘Okay, this is the part that triggers an odor inside of my vehicle.’” Bultel pointed out that odor testing will only become more critical as more shared and autonomous vehicles enter automotive fleets.

Using thermal desorption

The device at the center of Aryballe’s testing is the TD-SPRi sensor, a coupled thermal-desorption based amplifying device which increases the intensity of a compound’s olfactory signature up to 400 times. Using this device, Bultel said they are focusing on conditions set forth in the VDA 270 standard’s third variant, which heats materials to 80°C (176°F) for two hours. From the sampling, Bultel noted they are able to derive a Clustering Quality Score (CQS) based on intra- and inter-cluster ratios to provide a metric for discrimination performance between compounds.

“We say that a good nose is able to discriminate between two wines. And this is a metric to give a level of performance on how an electronic nose is able to separate between compounds,” Bultel said. “It gives you a visual aspect of how we separate those compounds. It's visual feedback.

“But we also need numerical values to say, ‘How far are they, one to each other,’ and to give a threshold of pass or not pass. That's why we developed the CQS for clustering this score based on intra-cluster and inter-cluster ratios," he explained. "So the further the clusters are, the better it is and the smaller the cluster are, the better you are in terms of measurement. It's a great improvement because... it [gives] a performance of the selectivity within the use of an electronic nose.”

Current challenges remain correlating the results gained from sensor testing with the VOC (volatile organic compounds) intensity scale grades (1-6) used by human testers. But according to Bultel, the results from his research indicate that digital olfaction can serve as an objective analysis tool for the aromatic quality of automotive plastics emissions.

“It's not that easy to make a link between the VOC analysis, a separation of those VOC and the hedonic tone,” he explained. “There is not always a causal relationship between the odorant that are present in the headspace or in the sample that is smelled, and the analytical composition of a sample.”

Repeatability and a standard

The creation of this objective odor data will be required before a global standard can be formed. Equally useful, Bultel explained, “the acquisition of odor-sensor data permits statistical metrics that can be used to evaluate repeatability and reproducibility to ensure the consistency in evaluation methods.” The upshot, he remarked, “is these digital devices can be used to standardize olfactive intensity tests, and while more work needs to be done to link human perception to instrument outputs, digital olfaction is a possible solution to challenges inherent in existing odor tests.”

Bultel noted that the work at Aryballe will help in crafting a proposal on a quantifiable odor rating which mimics the established norms from the human panels, in a way which is reproducible and independent of variation between individual testers. “Ultimately,” he proposed, “standardization of odor testing with consistent, repeatable and cost-effective digital olfaction methods can bring traceability and credibility to Tier-1 suppliers that translate to end-product quality for OEMs.”