Optimizing Tunnel Ventilation Fan Blades for Energy Efficiency Using the Adjoint Solver

Because fossil fuel resources around the globe are finite, an overriding engineering design challenge is energy efficiency and sustainability. Today I’ll use tunnel ventilation fans as an example to illustrate how CFD simulation and advancements in our Adjoint Solver in ANSYS 18 can optimize fan blades performance.

According to a report by Mosen Ltd., a leader in this industry, the “greening” of tunnel ventilation is still in its infancy. The application consumes substantial power, sometimes several megawatts; in addition, governmental regulations often require tunnels beyond a certain length (for example, 300 meters) to have ventilation systems that disperse exhaust and control smoke in case of fire. As a result, tunnels need more ventilation capacity than what would be needed for day-to-day air quality.

So if we make tunnel ventilation equipment energy efficient, we can impact the bottom line. The report goes on to state that ventilation is a significant factor in tunnel operating costs ranging from 10 percent to as high as 30 percent!  Author Faithi Tarada of Mozen cites the example of the 8 km long M5 East tunnel in Sydney, Australia that uses 54,000 MWh of electricity per year — at a cost in the neighborhood $15,660,000 (US) annually (assuming 29 cents/kWh). Clearly, this is a cost worth taking a long hard look at!

This is where engineering simulation comes into play. Optimizing the fluid dynamics of a ventilation system is no easy task. The fan blade components, for example, do not exist in a vacuum – they must be designed to operate as efficiently as possible in and among the other components of the ventilation system. Because of this, traditional trial-and-error simulation tools are not suitable for the task because of the complexity of the design space. There are just too many ways in which the geometry can be changed. Problem solving by trial and error just won’t cut it.

Tunnel ventilation fan model geometry

Tunnel ventilation fan model geometry.

A great way to solve these problems is with the adjoint solver, an optimization method for studies that require an estimate of sensitivity of model output (a forecast) with respect to input. The adjoint solver calculates the derivative of a single engineering observation with respect to a large number of input parameters. It does this in a single computation. Newly updated capabilities in ANSYS 18, the adjoint solver can optimize rotating machinery like these ventilation fans with support for rotating reference frames and the addition of cylindrical deformation regions.  Additional capabilities have been added to support conjugate heat transfer as well.

In our tunnel ventilation case, the ANSYS Fluent adjoint solver is capable of addressing moving reference frame (MRF) problems in rotating machinery — the rotating fan. And it can consider the full 3-D geometry, which traditional design tools can’t handle. The goal is to improve operating efficiency by changing the fan’s geometry.

The workflow for this study consists of running the fan case in ANSYS Fluent at a baseline operating condition, then applying Fluent’s adjoint solver. The adjoint solution is used to determine the sensitivity of the baseline solution to an observable – in this case the fan efficiency. Here we want to increase the efficiency by modifying the blade surface geometry, and the adjoint solution determines the surface geometry changes which will lead to an increase in the efficiency. A mesh morphing technique is then applied to the baseline model, which realizes the adjoint-driven geometry modifications.  The CFD solution for this morphed geometry can then be run in the Fluent solver to determine the actual change in efficiency.

Moving reference frame (MRF) is a new capability in ANSYS 18.0 that
easily handles faster accelerations. The adjoint solver is embedded in Fluent,
and comes at no extra cost.

For the present study, it took just a single design iteration to find ways to optimize fan blades that resulted in the total pressure rise decreasing by 14.9% and the shaft torque (and power) being reduced by 21.6%. This yielded a significant 8.5% efficiency improvement versus the baseline design. If desired, this process can be repeated for several design iterations to drive the optimization further. Because the adjoint solver directly determines what to modify and how to do it, it reaches the optimal geometry faster. It works hand-in-hand with mesh morphing technologies, so you don’t need to redefine the geometry nor recreate the computational mesh; you simply morph the mesh to the new shape. Moreover, when your analysis is completed, you can export the optimized geometry to an STL file for further use in a CAD product design system.

tunnel ventilation fanstunnel ventilation fans

The adjoint method optimizes the geometry based on the sensitivities you want to study.  In this case, simulation to optimize fan blade geometry resulted in 15% lower total pressure and 8.5% improved efficiency.

Of course, the tunnel ventilation fans case study is but one application that can benefit from adjoint analysis. Engineers can apply it improve designs over a wide range of industries and applications. What can adjoint do for you?

Want to know more about tunnel ventilation? Check out this case study where Mozen optimized ventilation 30%.

You can learn about how the adjoint method can optimize your simulations by downloading the white paper Shape Optimization for Aerodynamic Efficiency Using Adjoint Methods.

3 thoughts on “Optimizing Tunnel Ventilation Fan Blades for Energy Efficiency Using the Adjoint Solver

  1. Dr. Frank,
    Interesting topic on Adjoint Solver capabilities, can the above said optimization tool be used with Porous Media & Porous Jump properties in place for Air Induction Systems? Please reply at your earliest thanks.

Leave a Reply