Cadence’s Recent Acquisitions Reveal Focus on Computational Fluid DynamicsApril 18, 2021 by Jake Hertz
This year alone, Cadence has acquired both Pointwise and NUMECA, two companies specializing in multi-physics simulations and specifically computational fluid dynamics. How might this focus reflect Cadence's EDA strategy?
In the development of any product, it's critical to accurately and efficiently simulate the electrical systems. However, electrical systems simulation is not confined strictly to calculating voltages and currents. Other physical simulations, specifically heat transfer, are equally as important to system verification.
Thermal profiling of a PCB. Image used courtesy of Gareth Halfacree [CC BY-SA 2.0]
Recognizing the value of simulating a PCB’s physical environment, Cadence is making moves to expand its software suite.
The company made headlines this week when it acquired Pointwise, a multi-physics simulation company that specializes in computational fluid dynamics (CFD). This announcement follows Cadence's February acquisition of NUMECA, another company Cadence hopes will expand its "Intelligent System Design" strategy with CFD system analysis solutions.
Complexity of Heat Transfer in Electronics
As a result of scaling electronics, thermal management in electrical systems has become a central design focus for engineers.
Unfortunately, heat transfer is not simple to model: a standard PCB has multiple ICs and discrete components dissipating heat independent from one another. Airflow from both the environment and fans needs to be accounted for.
As a result, heat transfer becomes an extremely computationally expensive multi-physics problem.
Thermal modeling of a PCB can be very computationally expensive. Image used courtesy of Cadence
Trying to approach these simulations via a finite element of finite-difference analysis for an entire PCB would generally require a large amount of time. It is not uncommon for simulations of this sort to take hours to complete—meaning less time designing and money wasted.
Computational Fluid Dynamics (CFD)
Fortunately, heat transfer on PCBs can be modeled as a CFD problem—a process that is less computationally expensive than FEM analysis.
CFD is a classical multi-physics problem in which multiple physical phenomena have feedback among one another. The process flow for CFD modeling and simulation starts with determining heat sources in the system, modeling that heat conduction into the surrounding environment, and finally modeling the heat transfer caused by airflow.
CFD model flow chart. Image used courtesy of Cadence
Determining heat sources is a straightforward process; heat is estimated based on expected power dissipation given their currents, voltages, and resistance values. For example, one would use Ohm’s law to predict the heat of a resistor. Once the heat sources are known and defined, the engineer can use CFD analysis to develop an electrothermal model of the system.
From there, he or she can use a reduced state space approximation to efficiently simulate.
Cadence's CFD Strategy
Pointwise’s software isn’t explicitly aimed at PCB thermal modeling, but it's possible that this feature was one of the values Cadence saw in Pointwise's CFD portfolio—especially in the context of the company's NUMECA acquisition.
CFD modeling is based on electro-thermal co-simulations. Image used courtesy of Cadence
Pointwise’s technology is said to leverage a proprietary mesh generation technique that creates multiple mesh types to provide a level of control, ensuring converged and accurate CFD results. Mainly focused on the aerospace industry, Pointwise develops software that has been previously used for wind tunnel experimentation and high-fidelity discretization of aircraft.
This technology will likely find its home in similar systems where thermal modeling is even more complex—perhaps even estimating a PCB's airflow in a rocket.