Zooming

On the 1D side, one can speak of zooming, which consists in increasing the accuracy accuracy  of 1D system simulations, by factoring in purely geometry-related aspects, which can be captured using CFD. This will make it possible to enrich the Simcenter Amesim analytical model and provide higher accuracy results. Below are a number of recurring themes which are of high interest to 1D systems engineers working for instance on hydraulic circuit design. The phenomena discussed are difficult to describe with a sufficient degree of fidelity using purely analytical or semi-empirical methods, given the strong influence which the geometry has on the results. As a result, they are prime topics where 1D/3D CFD co-simulation can provide a significant added value.

• Flow cross-sectional area

For example, 1D codes provide a good estimation of the flow area for minimal cross-section on a ball poppet valve with a straight-sided seat. This is no longer the case for conical valve seats, as the flow area moves in that configuration. Figure 13: The flow cross sectional area as a function of ball lift in a conical-seat ball poppet valve [10]

• Flow Coefficient Cq

The flow coefficient in Simcenter Amesim can be refined in some specific scenarios. For instance, in the case of a flapper valve, Cq varies strongly as a function of valve lift and flow conditions (laminar Laminar flow /turbulent flow, attached/detached flow, depending on the Reynolds number Reynolds number of the flow). [8] Figure 14: Flow coefficient as a function of flow number, and different flow patterns across restrictions – flapper valve [15]

The flow coefficient evolution is reflected using 2 parameters: Cq and λc. Experimentally, these are very difficult to extract, as they would require knowledge of the minimal working section of the fluid.

• Cavitation

Taking this phenomenon into account has been a requested feature for a long time, especially in applications such as high pressure Diesel injection systems. The cavitation is correctly handled by cavitation orifice submodels in Simcenter Amesim, such as BHO0014. However, cavitation can also appear on the ball of a control valve, and sometimes even upstream of the sac volume of an injector nozzle. Analysis of the static pressure field extracted from CFD enables users to identify and quantify critical zones of low pressure, in which the fluid tends to vaporize or deaerate. Figure 15: Cavitation zone – short tube

The vena contracta is not the main source of cavitation. Indeed, the area of minimal pressure is located at the intake edge of the orifice. Figure 16: Evolution of the void fraction in a ball poppet valve for different valve lift values, illustrating the cavitation phenomenon – Diesel injector [12]

Note It is still currently quite difficult to predict cavitation (and any phenomena that may arise as a result of it) with CFD code – care must be taken to verify any computational results obtained against experimental data. Without such validation, you are advised not to make any quantitative inferences into the results based purely on CFD data.

• Pressure gradients

Geometry-dependent pressure gradients can arise, making it impossible to create an accurate estimate of the global efforts in the component through a discretization into 2 zones (delimiting the active surfaces within the component). In the presence of geometry-dependent pressure gradients, such a discretization can lead to potentially significant errors in the assessment of global force – a typical application of this is flapper valves. Figure 17: Flow separation in a bend, and its effects on pressure coefficient – flapper valve [15]

This also occurs systematically for high pressure injector nozzles, in which very small angular variations result in an intermediary volume trapped between a large diameter (at the contact point where it closes), and a small diameter (at the sac). The net force thus varies significantly for small needle lifts (between 0mm and 50mm). Figure 18: Pressure fields at injector closing, for both a Minisac and VCO type nozzle [13]

• Jet angle

Experimentally, it is practically impossible to measure the value of the jet angle on a real valve (except on large scale, transparent models). Additionally, and contrary to the flow coefficient Cq, deducing the jet angle from global measured variables is also a strenuous process. Figure 19: Jet angle – spool

In this case, the post-processed jet angles trend as it would be expected from theoretical results (momentum in incompressible flows incompressible flow ).

• Jet force

Calculation of the jet force in Simcenter Amesim relies on the assumption of incompressible flow, corrected by a jet factor: Fjet=2.Cq.A.dP.cosq.Kjet. In some cases, this estimation can differ from what can be found with the help of CFD, or measured experimentally. For example, in the case of a ball poppet valve on a conical seat, compared with Fluent and STAR-CD, the results are as shown below: Figure 20: Hydraulic force as a function of pressure difference, comparison between 1D and 3D studies – ball on conical seat [10]

Another example is a 3-way valve, where displacement varied between 0.1 and 0.75mm, all the while maintaining a pressure difference (Pu-Pd) at a constant level. Similar simulations were then carried out for different pressure drops: 50, 100 and 150 bar. Figure 21: Fluid domain in the distributor study

CFD computations give good insights when analyzing new designs. Here, for example, with a jet-force compensated solution of the 3-way valve, and associated static pressure field to balance the hydraulic forces. Figure 22: Velocities for a jet-force compensated 3-way valve

Figure 23: Static pressure field for a compensated solution

• Squeeze effect or hydraulic cushion

This consists in changes to the pressure gradient due to fluid being trapped in outlets or corners of flow areas, as well as due to the descent velocity of the moving part of the component. The net effect is a change in the global force. Figure 24: Illustration of the squish effect [14]