Summary
Honeywell, a leader in industrial process simulation, has partnered with NVIDIA to accelerate their simulation processes using NVIDIA’s cuDSS, a GPU-accelerated direct sparse solver library. This collaboration has resulted in significant performance improvements, with speedups ranging from 2x to 78x over Honeywell’s existing sparse linear equation solver. This breakthrough enables faster and more reliable process digital twin execution, improving engineering productivity and reducing capital costs, operating costs, and carbon footprint for new industrial facilities.
Accelerating Industrial Process Simulation with NVIDIA cuDSS
Industrial process simulation is a critical component in the design, operation, and optimization of industrial processes. However, traditional simulation approaches have struggled to fully leverage multicore CPUs or acceleration devices to run simulation and optimization calculations in parallel. This limitation has hindered the development of larger and more complex process models.
The Challenge with Traditional Simulation Approaches
Traditional industrial process modeling and simulation approaches have relied on multicore linear solvers, which have not achieved expected improvements and have underperformed optimized single-core solvers in certain cases. This has led to a need for more efficient and scalable solutions.
Introducing NVIDIA cuDSS
NVIDIA cuDSS is an optimized, first-generation GPU-accelerated direct sparse solver library for solving linear systems with very sparse matrices. It uses CUDA to parallelize matrix factorizations and solutions on GPUs. This solver has been integrated into Honeywell’s UniSim Design equation-oriented platform, called UniSim EO.
Testing NVIDIA cuDSS for Industrial Process Simulation
Honeywell has a set of nonsymmetric matrices generated from UniSim EO process model applications, including models of upstream and midstream oil and gas, refining, petrochemical, and chemical process units. These matrices were used to test the performance of NVIDIA cuDSS against Honeywell’s in-house sparse linear equation solver, DOTAXB.
Performance Results
The testing results showed significant performance improvements, with speedups ranging from 2x to 78x over Honeywell’s existing sparse linear equation solver. The average cuDSS speedup was 3x for cold start and 19x for hot start, excluding the largest case. The performance speedup on the largest test case was 78x for cold start and 200x for hot start, indicating superior scalability and efficiency on significantly larger matrices.
Benefits of Faster Process Model Solutions
The performance improvements delivered by cuDSS enable larger-scope first principles models to be solved in a reasonable amount of time for reliable process digital twin execution. This also eliminates the need for including surrogate or reduced model development in an application workflow to achieve faster solutions, reducing maintenance concerns.
Impact on Engineering Productivity
The cuDSS performance boost helps improve engineering productivity by enabling the completion of complex simulations faster. As a result, more design scenarios can be considered in the same time frame, leading to better process designs and reduced capital costs, operating costs, and carbon footprint for new industrial facilities.
Future Developments
Honeywell is working to complete the productization of NVIDIA cuDSS as a linear solver option within the context of nonlinear equation solving and optimization in UniSim Design. This includes optimizing solver configuration for the process simulation domain and assessing improvements with different NVIDIA GPUs and new and emerging NVIDIA hardware.
#Table: Performance Comparison of NVIDIA cuDSS and Honeywell DOTAXB
| Test Case | cuDSS Speedup (Cold Start) | cuDSS Speedup (Hot Start) |
|---|---|---|
| Average (excluding largest case) | 3x | 19x |
| Largest Case | 78x | 200x |
Table: Benefits of Faster Process Model Solutions
| Benefit | Description |
|---|---|
| Improved Engineering Productivity | Faster completion of complex simulations, enabling more design scenarios to be considered in the same time frame. |
| Reduced Maintenance Concerns | Elimination of the need for surrogate or reduced model development in an application workflow. |
| Better Process Designs | Ability to consider more design scenarios, leading to better process designs and reduced capital costs, operating costs, and carbon footprint for new industrial facilities. |
Conclusion
The partnership between Honeywell and NVIDIA has resulted in significant performance improvements in industrial process simulation using NVIDIA cuDSS. This breakthrough enables faster and more reliable process digital twin execution, improving engineering productivity and reducing capital costs, operating costs, and carbon footprint for new industrial facilities. As the industry continues to evolve, the integration of cuDSS into Honeywell’s UniSim Design product will play a critical role in shaping the future of industrial process simulation.