The background
There is a rising demand for tools that enable the design and development of new products in the fastest possible time, which means with the shortest time-to-market. This is especially true in the automotive industry worldwide. Most automakers opt for new technologies like knowledge-based engineering (KBE) and product lifecycle management (PLM) tools that allow them to drastically reduce cycle time for new product introduction. KBE empowers designers to quickly navigate through the number of iterations involved in the design process and arrive at an optimised solution.

 

In today’s product design environment there exist numerous KBE applications in the fields of vehicle ergonomics, suspension systems, noise-vibration-harshness, safety, regulations, etc. Some of these have brought design time down from days to mere minutes, but there are several complex applications that still take long.

 

Tata Technologies has developed its own propriety software framework in line with next-generation KBE technology, and this has already been put to work at Tata Motors’ Pune plant. What the company has done is to speed up KBE applications by using the graphic cards of existing workstations for computational purposes, in essence enabling parallel processing without having to source additional hardware.

 

Using graphics processing unit (GPU) computing and compute unified device architecture (CUDA), the KBE kernel has been restructured to incorporate a parallel programming model that helps applications run faster on the GPU, thus achieving time reduction from hours to seconds.

 

The innovation
In all software applications, arithmetic calculation happens in the central processing unit (CPU) of the computer and visualisation or graphics processing through the GPU. With its new-generation graphic cards, Tata Technologies has successfully applied a parallel processing algorithm for all iterative calculations using threads at one go.

 

Company: Tata Technologies Innovation: KBE applications using GPU computing Team: Dr Milind Ambardekar, Yogesh Deo, Prasad Kulkarni, Dilip Sahu, Avijit Santra

This generates a huge performance boost; KBE applications, which used to take hours to deliver results, can now be executed in seconds. The model also saves on cost as it uses existing graphic card capability instead of increasing the hardware configuration.

 

The next-generation KBE system provides far more power and flexibility in the development of engineering products. These systems are designed to incorporate complex rules, artificial intelligence and agents to be embedded in the system. In addition, they provide more direct control over geometry and topology creation, manipulation and advanced geometry introspection capabilities. They leverage reuse of corporate design knowledge to the maximum extent possible, eliminating mundane tasks within the complex process of transforming a product concept into production-ready details.

 

With the implementation of GPU technology, KBE applications get even better. Productivity increases as the designer can do many more iterations in less amount of time and spend more time on creative thinking. At Tata Motors, all CAD workstations are equipped with CUDA-enabled NVIDIA graphic cards. The KBE applications take advantage of GPU computation capability without compromising on visualisation quality.

 

This innovation required a lot of hard work. Tata Technologies’ programming experts were trained for two weeks in CUDA technology at NVIDIA, Pune (NVIDIA is the inventor of the graphic card and CUDA language). Then the developers had to delve deep into engineering mathematics, KBE kernel mathematical algorithms, parallel processing algorithms, CUDA, etc before practically rewriting the KBE algorithms for thread processing. This took up almost half of the total development time.

 

The payoff
The impact of the new product is huge. The savings per workstation configuration works out to Rs30,000, which means Rs15 million in savings across 500 workstations. By making additional hardware unnecessary, the company is also reducing environmental load and contributing to energy savings. The new technology will not only improve Tata Motors’ product design time and quality, but also improve the brand image of Tata Technologies as a tech-savvy company.

 

The knowledge-based engineering kernel (KNEXT) is a totally in-house developed, proprietary product of Tata Technologies. The lesson learnt: it’s better to use existing resources to their optimal strength. And there are big advantages in thinking laterally.