Engineering news
CFD Optimisation Case Study – DC White Consulting Engineers
We specialise in the design and analysis of mechanical and civil engineering projects, with market-leading expertise in stress, seismic, vibration, thermal and piping design analysis.
A central part of these services is our CFD analysis, using a combination of OpenFOAM® and midas NFX. Here is an example of our work for a major manufacturer of pre-painted corrugated steel sheeting.
They requested a design optimisation for a spray nozzle manifold, termed a 'poker', designed to uniformly lay down the protective coating. The arrangement is illustrated below:
Concern had been expressed about the lack of uniformity of the nozzle spray velocity along its length, thus giving an uneven lay-down. DC White CE were asked to optimise the poker design using Computational Fluid Dynamics (CFD) with a view to achieving a uniform nozzle velocity.
The design notably uses a tapered square bar welded along the tube through which the 42 holes are drilled. The taper creates a linearly varying hole-length, giving a theoretically reduced pressure loss in the farthest holes from the inlet.
Initial modelled geometry
CFD was performed on this existing design using Midas NFX 2014. Shown below is an oblique view of the results (foreshortening the longitudinal axis) with velocity vectors plotted as arrows.
Existing CFD velocity vectors - oblique view
To support the CFD analysis, a theoretical model of the flow was made in MathCAD based upon the principles of 1-dimensional developed pipe flow. The variation in hole-lengths is one design factor, but also is their length and diameter. Generally longer and thinner holes create higher pressure loss, making the losses in the main tube relatively less significant. This makes the main tube behave more like a large header-tube or plenum, giving a more even flow distribution. Constraints of the tube weight, supply pressure and the cost of high precision require that a compromise be sought.
The theoretical model was most accurate in the aft two thirds of the poker, but the nozzle velocity at the inlet-end was about 23% higher than the CFD model. This is likely due to the three-dimensional effects of the flow turning 90 degrees at the inlet end. The theoretical model was used to find a hole-length strategy which achieves an inlet flow of deliberately 23% above the equal spray velocity and with roughly the correct velocity at the far end. This was expected to cancel the error in the theoretical model. The new design was proposed as shown below.
Suggested new design 3D geometry
Shown below are the velocity magnitudes on the central cut face and inlet for the redesigned poker and a graph comparing this with the original design.
This redesign reduces the maximum error from the even-spray velocity, from 66% to 9% compared with a previous design.
Redesigned poker CFD velocity vectors - oblique view
Comparison of redesigned and original poker CFD results
To demonstrate that the solution could resolve instability, a sensitivity study was run with the viscosities dropped by an order of magnitude. This produces an interesting instability in the central section which greatly undermines the flow evenness. This suggests that running the poker at comparably higher velocity will likely cause an uneven lay-down.
Visit our website for more information on our CFD services.