Latest from :

Case study – the air handling system

Experimental approach

In the past, the only way to evaluate a proposed air handling system design was to build a prototype and test it in the laboratory. The air handling components were placed on a test stand, conditioned air was supplied at the inlet and the airflow and temperature distribution at crit­ical locations were measured. This approach takes a considerable amount of time and requires the construction of expensive prototypes. In addition, it provides little or no understanding of why a design performed the way it did. In particular, testing is unable to detect details of recir­culating areas, turbulence, temperature stratification and constrictions that adversely impact performance and pressure loss. In addition, the performance of the system usually needs to be evaluated in many different configurations. For example, it sometimes is necessary to evaluate the air handling system in different modes of operation – vent, floor, defrost and mixed – at each of eight different temperature controls.

Modern methods of design

The design process of modern vehicle systems improved with the introduction of Computer Aided Design (CAD), Computer Aided Engineering (CAE) and Computer Aided Manufac­turing (CAM). CAD allows designs to be generated and visually appreciated on a computer. Standard components can be shared among manufacturers and suppliers to ensure that compon­ents assemble correctly. Designs can be sent to clients for verification and feedback. Designs can be modified and rechecked within short periods of time in a number of different formats, e.g. an STL file (stereolithography). Complex parts and assemblies can often be manufactured very quickly using rapid prototyping facilities (CAM). CAD also includes the facility to provide virtual testing. This is generally provided using additional modules or add-ins converting CAD to CAE. The software is even now used among a number of secondary schools in the UK who have the use of Solidworks as a CAD package for their technology departments which include add-in modules like Cosmos Works for Finite Element Analysis and Computational Fluid Dynamics. Finite Element Analysis (FEA) is basically mechanical stress analysis and Computational Fluid Dynamics (CFD) analyses the flow of a fluid like air through or over complex geometry. These additional features are all computer-based and use mathematical equations built into the soft­ware to predict variables like the stress distribution of a component or assembly (FEA) or the flow of air through an air vent (CFD). All these tests would have originally been carried out manually with continual adjustments being made to a model to optimise it.
Computer generated model designed using CAD (without ducting and vents)

Air pressure loss predicted by CFD

The process 

The A/C system begins life as an idea driven by consumer needs and government legislation. This leads to a specification. The specification will include minimum performance requirements, temperatures, control zones, flow rates etc. This will lead to a number of concept designs. The designs will have a number of computer generated models which will be presented as possible solutions to the original specification. These need to be tested for their performance. 

Performance testing using CFD may include fluid velocity (air flow), pressure values and tem­perature distribution. Using CFD enables the analysis of fluid through very complex geom­etry and boundary conditions. The geometry typically includes ducts that expand and contract, change from round to square cross-sections, go through complex curves throughout their length, and have many branches and internal walls. 

As part of the analysis, a designer may change the geometry of the system or the boundary conditions such as the inlet velocity, flow rate etc. and view the effect on fluid flow patterns. 

CFD is an efficient tool for generating parametric studies with the potential of significantly reducing the amount of physical experimentation required to optimise the performance of a design. 

A fan performance curve can be inputted into a model. Without this feature, the user has to guess the flow within the fan enclosure, calculate the pressure using CFD and see if it matches

Streamlines showing flow field in an air handling system

Fan flow optimisation

the fan’s characteristics. If the pressure doesn’t match, then another guess has to be made. Normally, at least three iterations (test runs) are required to make a match.

The software has the facility to enter a fan performance curve directly into the model. Each analysis run then interacts with the fan curve to determine the precise operating conditions of the fan as part of the regular analysis. Using this technique, engineers can easily determine what
Improved fan design

Human modelling for temperature distribution

type of fan is required to meet air flow requirements within the vehicle, normally 158 cubic feet per minute (75 litres/second) for heating and 300 cubic feet per minute (141.6 litres/second) for cooling.

As a typical example of improvements consider the typical design specification of the HVAC system with respect to the temperature dial on the instrument panel. In other words, moving the dial from position one to position two should have the same impact on tempera­ture as moving from position two to position three. In the past, the linearity of the temperature dial could not be estimated until full vehicle prototypes were constructed.At that point, changes were costly and the testing data provided little or no input on what type of changes were required.

Prototype HVAC unit for testing

Now, engineers can determine the linearity of a proposed design as soon as the solid model in CAD has been created in a matter of days. They typically set up a series of analysis runs that evaluate eight different temperature settings at each of the three HVAC system modes. In less than a week, they can determine outlet air temperature at each setting. 

Once all CFD modelling is complete the prototypes are made to ensure the physical models operate as predicted by the computer models. The accuracy of simulated and actual system per­formance can vary up to 10–15%. Generally, lead times are reduced and designs can be evaluated much quicker allowing more time to optimise their working performance.

(© 2005 Visteon All rights reserved)

Posted by: