! What CFD Is Suited For

Computational fluid dynamics (CFD) is a very promising tool, and its use can be very helpful for analysis and design in industrial ventilation. It is suited for all types of problems where knowledge of a spatial distribution of flow quantities is desired, i. e., where local values at several locations are required.

The flow quantities may include air velocity, temperature, or contaminan t concentrations. The term Contaminant is used here as a general expression for other species at low concentrations carried by the air, such as smoke, COi, or toxic gases; even the local age of air can be treated in a similar way.

The availability of local quantities and only a minimal number of physical assumptions are the two key features of CFD. As a result, CFD has the follow­ing advantages over other methods such as multizone or flow element meth­ods, which are useful for average values:

• Detailed analysis is possible.

• It is generally applicable for a wide variety of situations and problems.

• Interaction of different flow elements such as jets, plumes, and boundary layers is inherently considered.

• Even flow problems which have not been thought of before can be detected.

• It is very strong, even ideal, in parameter variations of a certain situation.

The last advantage, parameter variations, is in fact common to all the numeri­cal models but is a main advantage over an experimental investigation, as the time effort for changing parameters is very small.

The following example illustrates some of these features. In a clean room with automated transporting and processing facilities, there are two quality classes of air. The space close to the small items under processing and the treating facilities should be exposed only to air of class A quality (very low contaminant level) and the rest of the room air is class B quality (normal stan­dard). Figures 11.2 and 11.3 (see color insert) represent two out of a large number of parameter studies.

Figure 11.2 shows a cross-section through an open oven which is pro­tected by an air curtain of class A air. The oven door must be opened for the loading of the items to be processed in the oven.

The study shows, however, that the air curtain in the original design is bent off the wall due to the presence of the open oven with induced airflow in­side the oven. This unexpected flow feature finally leads to a recirculation zone below the oven, and dust particles on the floor can be carried into the oven (class B quality). This effect was then also confirmed by smoke experi­ments in the real room and the existing ventilation system.

The redesigned airflow system, consisting of a broader air curtain (the broadness needed also for other reasons) and a local exhaust below the oven, ensures class A air quality in the oven when the door is opened for loading (see Fig. 11.3). Application Examples

Quite different flow problems of steady-state and transient nature can be treated. Examples of flow situations that are most important in industrial ap­plications are as follows:

• Impact of air supply or exhaust devices to room airflow

• Thermal plumes originating from machines or high-temperature processes

• Effects of contaminant sources and their distribution in the rooms

• Airflow due to moving parts

• Airflow through openings between rooms or between inside and outside (e. g., natural ventilation or flow through process-related openings to outside)

More recently, the application range seems to be extended with increasing de­mand to even more complex situations, including additional effects such as

• Influence of solar light through glazings

• Wind influence on large openings to the inside space

• Distribution of heavy gases (i. e., species other than air at high concentrations) in the case of chemical hazards or smoke from fire for occupants’ safety reasons (ways of escape) Considering a CFD Study

When a CFD simulation is desired, the following points have to be consid­ered.

What Do I Wiont to Know Exactly?

It is important to define exactly which results are really required, e. g.:

• Do I need the airflow or turbulence quantities at certain locations, or are toxic concentrations needed?

• Are all the locations of equal importance, or are some more important than others?

• How accurate are results required?

The CFD engineer then has to choose the input values, the resolution of the modeling, and the distribution of the calculation mesh related to the impor­tant physics in the flow and to the area and effects of interest.

Effort Needed

It is up to the skill of the CFD engineer to choose the level of simplifica­tion of the real situation and the models needed and to judge from experience the accuracy of the results.

A generally valid, simple checklist cannot be given. It is, however, very im­portant to realize that CFD applied to a problem situation is not a simple three step procedure like (1) setting up the problem, (2) letting the computer do the job, and (3) plotting the pictures and getting “yes or no” information (“we build it” or “we don’t”). The full truth is more that a CFD application is an iterative procedure (not only within the algorithmic DO loops) between the ventilation designer (and architect) and the CFD engineer. Different levels of simplifications of the real situation will be applied at different stages of the de­sign process, and possibly large numbers of parameter variations of CFD cases have to be carried out in order to reach a conclusion, which in fact can be as simple as “yes” or “no.”

How Good Are the Results?

It cannot be stated generally how accurate the predicted results are. Due to the limitations of geometric, physical, and mathematical modeling, not all of the produced numbers (e. g., air velocity vectors) are at a high level of accu­racy, and the results are therefore subjected to experienced weighting. In some cases, the values can be as accurate as within 5% of the real values; in other cases, they are not as accurate as could be wished. But results can be still very strong and helpful in a comparative judgment, i. e., if a number of similar cases are compared with observed tendencies.


A general figure cannot be given either, but usually a case of medium-level complexity with a set of 10 parameter variation needs an effort of two to four rnan-weeks. In any case, a number of cases have to be calculated (say, 5 or 10, not just 1) to gain knowledge about the sensitivity of the flow to the choice of certain parameters.

Practical Procedure for Carrying out a CFD Study

Usually, in practice, the person who carries out the CFD computation is different from the person who needs the results. It is then very strongly recom­mended that a strong feedback between the designer who needs the results and the CFD engineer be maintained to prevent disappointments, e. g., results that do not meet the real needs. It is in fact an iterative procedure.

It is furthermore iterative in the sense that one should start from a coarse model, which is then successively refined to the desired or possible depth. Fail­ures in the models or problems of different types tested so far can be detected and treated at an early stage of the design process.