Encyclopaedia Index
Click here for Contents List

Food Processing and the new CFD

by

Brian Spalding, of CHAM Ltd

March, 1999

A lecture presented to the 1999 Food Processing and Control Panel meeting, of the Leatherhead Food Research Association, Leatherhead, Surrey, England, UK, on March 3, 1999.


Abstract

History

Computational fluid dynamics has been used for the prediction of heat, mass and momentum transfer in industrial processes since the 1970s; and some worthwhile applications have been made to problems arising in food processing.

Obstacles

However, the use of CFD for such problems has not become widespread, because of several obstacles, namely:

  1. obstacles concerned with the nature of the materials and processes:

  2. obstacles concerned with the natures of the food-processing CFD-software-vending industries:

New hope

Fortunately, there are new developments in CFD technology which significantly diminish these obstacles. Specifically:

  1. Regarding obstacles concerned with the nature of the materials and processes:

  2. regarding obstacles concerned with the natures of the food-processing CFD-software-vending industries:

The lecture will explain how these new features can be accessed and utilised.

The computer code which possesses all the capabilities which are described is PHOENICS, which is a product of Concentration, Heat and Momentum Ltd.


Contents

  1. Introducing user-defined properties
  2. Computing the micro-composition of mixtures
  3. From CAD to CFD via Virtual Reality
  4. The disappearance of the grid-generation problem
  5. The use of pay-as-you-go remote computing
  6. Other developments of possible interest to the food industry
  7. Concluding remarks

Note: Figures are not provided in the printed paper. They may be inspected, together with the text, on the web-site: www.cham.co.uk


1. Introducing user-defined properties

The problem

The variety of materials which are of importance to the food industry is immense; and their thermodynamic, transport and chemical-kinetic properties, or rather those which are quantitatively known, require to be expressed by way of formulae which are too diverse to be incorporated in any general-purpose computer program.

A solution for users who know Fortran

One way to enable such a program to handle materials differing from those whose properties are built in is to allow users to access, and then modify suitably, certain parts of the Fortran coding.

This feature has been available in PHOENICS from its beginning; and it has been introduced, to some extent, by some of its competitors.

A solution for users who do not know Fortran

However, utilizaton of such features requires a skill in Fortran programming which can no longer be expected of all those who wish to benefit from the flexibility of PHOENICS.

The latest versions of this code therefore embody the so-called "PLANT" feature, which makes the introduction of new material-describing properties accessible to persons who are entirely unacquainted with Fortran.

Full information about PLANT is obtainable from CHAM's web-site: www.cham.co.uk .

For Applications-Album examples, click here.


2. Computing the micro-composition of mixtures

The new possibility of computing the micro-composition is an outcome of the recent incorporation into PHOENICS of the "Multi-Fluid Model" of turbulence, a full account of which can be found on the CHAM web-site by clicking here.

However, its range of application extends far beyond the context of turbulence; indeed it extends to all circumstances in which a heterogeneous material consists of a "population" of homogeneous materials which differ systematically in respect of one or more attributes.

PHOENICS computes populations of a one-dimensional character, such as shown here , or of a two-dimensional character, such as shown here.

In each case, it is the left-hand part of the picture which conveys the quantitative information; the right-hand half is simply a reminder of the chaotic nature of the postulated inter-mingling of the materials.

Translated into terms of food, the first (one-dimensional) population might be that of the microbial content of grain particles of various ages.

The second might represent the microbial content of grain particles of various ages and humidity contents.

In order to understand how PHOENICS simulates the processes which generate such populations, it is probably best to examine the following account of mixing in a stirred tank (click here), which might be thought of as an idealised cooking pot.

Of course, the physical and chemical processes which cause the micro-composition to change will differ for each food-processing situation; and it is impossible (indeed undesirable) that all possibilities should be embodied within one general-purpose computer code.

It is therefore relevant to report that the MFM and PLANT features of PHOENICS work happily together. What is desirable now is that food-processing specialists begin, with CHAM's assistance, to define what physical and chemical processes are relevamnt in a variety of industrially (and socially) important cases.

Such collaborative work might well qualify for EC support under its Framework-5 programme for Information Society Technology.


3. From CAD to CFD via Virtual Reality

One of the most significant developments, from the point of view of making CFD accessible to non-specialists, has been the attachment to the PHOENICS code of a "Virtual-Reality" front end.

This enables users to create their "worlds" by selecting from a "virtual ware-house" of objects, which they then re-size and locate appropriately, thereafter assigning materials, temperatures and other relevant attributes.

An account of this feature can be found in the PHOENICS Encyclopaedia, by clicking here.


4. The disappearance of the grid-generation problem

This topic is discussed at length, in a recent lecture, which can be reviewed by clicking here; the matter is discussed there in relation to the design of gas-turbine combustion chambers.

The technique employed is called PARSOL, examples of the use of which can be seen, on the web-site, by clicking here

The combustion chamber example shows how well PARSOL succeeds in representing a wall which cuts the computational cells obliquely.

The relevance to food processing is that such combustion chambers have geometries which are at least as complex as those of most food-processing devices; so, if the grid-generation problem has been removed for the former, it is certain that it can be removed for the latter also.


5. The use of pay-as-you-go remote computing

CHAM, with financial support from the European Commission, and with the collaboration of a consortium of partners across Europe, has been pioneering new ways of providing CFD services; these may involve the use of Internet.

The first project, called MICA, required users at least to possess the "front-end" of PHOENICS on their own computers. When the second project, called ADELFI, is finished, it will suffice for them to have a "browser", such as is standard on all modern PCs.

The basic idea is that the needs of potential users of CFD are so various that a spectrum of services must be provided, with consulting services at one extreme and provision of computer software for customers' own use very much at the other.

All this is described on CHAM's website, the relevant content of which can be seen be clicking here.


6. Other developments of possible interest to the food industry

Contents: New CFD; SFT; parallel PHOENICS.

"New CFD"

What has been presented already is perhaps sufficient to justify the appearance of the adjective "new" in the title chosen for this lecture.

However, there is much more going on which may interest food technologists. Two of the topics will be briefly mentioned.

Simultaneous analysis of fluid flow and solid stress

For the last three decades the computer-aided-engineering field has consisted of three well-separated parts, namely:

  1. Computer-aided design (CAD),
  2. Computational fluid dynamics (CFD), and
  3. Computer-aided stress analysis (CASA).

Earlier in this lecture it has been shown that CAD and CFD have been moving closer together, as a consequence of their being able to share the same geometry-description methods.

Now it can be revealed that a convergence of CFD and CASA is in progress. Indeed, the name SFT, standing for "solid-fluid-thermal" has been coined for the combined area of analysis.

CHAM's website contains a complete lecture on SFT.
Clicking here leads directly to interesting examples.

Generalising the idea, and giving it a food-industry slant, would enable the baking of bread, for example, to be simulated in all its aspects.


Low-cost computing power from PC clusters working in parallel

Although, in section 5 of the present lecture, it has been argued that the availability of CFD-via-Internet may render the purchase and maintenance of high-powered computer hardware unnecessary, some CFD users will still want to have everything under their own control.

For them there is good news also; for the increasing power and falling cost of PCs, combined with the fact that some computer codes (of which PHOENICS is one) exist in "parallelised" versions, enable super-computer performance to be obtained from "clusters" of PC2, caused to work together on a single flow-simulating calculation by suitable "harnessing" software.

Thus an assembly of eight two-processor PCs, arrayed on shelves in a coat-stand-sized box, will deliver a speed of computation which is nearly (but never quite) sixteen times that of one PC on its own.

The figure shows CHAM's first benchmark results, now two years old.

This is the direction that CHAM's own computer hardware is taking. It is probably one which many other organizations will find to be the most cost-effective.


7. Concluding remarks

The main message of this lecture is that the nature and capabilities of CFD have been changing rapidly if recent years. Therefore, if any of its hearers or readers hold the view:
"Yes, we tried CFD for ....., some years back; but it:

then such persons may decide that it is time to take another look at what CFD now has to offer.

However, if they should do so, they should not suppose that the choice facing them is that of purchasing (or not) a particular software package. The question is rather:

This question is best answered after careful discussion in which their true needs for food-process modelling are expressed and considered in the light of what the new CFD can accomplish.


The End