Commercial CFD Starts Here …
Firstly, a group led by Suhas Patankar (essentially an extension of the 3D Boundary-Layer Group that I had been involved in) were consolidating what we had learned about SIMPLE-based solution methods into a series of standard programs for the various classes of fluid flow – 2D elliptic, 3D elliptic, and 3D parabolic – named respectively:- EASI – elliptic axisymmetric integrator; TRIC and TRIP – three-dimensional recirculating flow in Cartersian/polar coordinates; and STABLE – steady analysis of the boundary-layer equations. (It will be apparent that a lot of thought went into the acronyms used to name the programs!) Plus, there was GENMIX, for 2D boundary layer flows, resulting from Suhas Patankar’s PhD work in the mid 1960s.
These programs, in FORTRAN (FORTRAN IV, or possibly FORTRAN 66), were used by researchers at Imperial College as the basis for development and for research projects, and, increasingly (this is the second “important thing“) for the commercial activities which were then beginning in Spalding‘s group.
It seems to me that Professor Spalding’s research activities have always been driven by the commitment to solving real-life engineering problems. The end point of his research activities is not, in his mind, the “solution of the scientific problem“ (be it a formula, a methodology, or a computer program); nor is it the publication of the outcomes of the research; rather, in his view, the work isn’t finished until it is making real engineering contributions in industry. In other words – the conception of the solution, its development and testing, and its industrial application are all, in his mind, one continuous spectrum (sometimes, unwisely, undertaken as a single project – but that is another story!).
Hence, to encourage practical application of the early CFD work, Patankar’s GENMIX 2D boundary layer method, and the subsequent Runchal/Wolfstein stream-function/vorticty method for 2D elliptic flows, were both published in book form (with full FORTRAN listings) very soon after the work was completed. And “Post Experience Courses” were run to train engineers from industry in using the software.
I suspect that these had only limited success – and that Spalding learned from this experience that, in order to get the methods really used in industry, a rather more pro-active approach was needed. Hence, from about 1970 on, commercial projects were undertaken, initially by the staff in Spalding’s Heat Transfer Group, in which the CFD methods were tailored for specific industrial customers’ problems. These were managed via a new commercial operation, Combustion Heat and Mass Transfer Limited – or “CHAM“.
By 1972 the CHAM activities had progressed to the stage where full-time, dedicated staff were needed. I was one of this first batch of (I guess) about half a dozen students just finishing their PhDs, who joined CHAM in the autumn of 1972. Initially we were based at Imperial College, managed by a “management committee” made up of a number of the senior members of Spalding’s group, including (as well as Spalding) David Gosman, Brian Launder, Jim Whitelaw, Fred Lockwood, and others. Akshai Runchal was the senior engineer (I guess his title was Technical Director).
Projects typically involved adapting one of the “standard” CFD programs to a specific application for a particular customer. Projects were assigned to an individual engineer, working under the supervision of one of the “management committee”. At the end of the project the software was delivered to the customer in source-code form, with documentation, and some assistance in using the software. In other words we were developing “bespoke” CFD software for specific customer’s applications – as opposed to the later trend in CFD towards packaged, general-purpose software.
With hindsight, it is fairly easy top see the “challenges” in the “bespoke software” business model. Among other things, it is very difficult to maintain consistency, quality, and continuity when individual engineers are, in effect, developing their separate software products for each customer; and it is difficult to make any across-the-board software improvements, leading to much duplication of effort and “reinventing the wheel”. However, back then this was pretty much all that we could do, based our knowledge, experience, and software expertise at the time. For various reasons, it was to be almost 20 years before the emergence of the first packaged, general-purpose CFD software – CHAM’s PHOENICS program- but that’s a later instalment in the story.
CHAM stayed at Imperial College for about two years, until 1974. While initially most projects were from the UK (predominantly defence, aerospace, and power generation, mainly nuclear), increasingly during this period we also got involved in projects from Europe and the US. I worked on projects for the UK Ministry of Defence, and for aircraft companies in France and the US – mainly on gas turbine exit and rocket exhaust phenomena. I was supervised on these by Spalding, David Gosman, and (I think) briefly by Brian Launder.
One of our recruits during this period was Harvey Rosten, my future business partner in founding Flomerics. Harvey joined us after completing an MPhil in electromagnetics at the Rutherford Laboratory. He picked up CFD very quickly – his first project was, I recall, on the transient effects of a flame in reduced gravity.
I recall this as an exhilarating and stimulating period. I learned an enormous amount – not just extending my technical knowledge into the treatment in CFD of detailed chemistry, heat transfer by radiation, and transonic phenomena – but also learning about working to commercial deadlines, documenting software, and interacting with customers. As far as the commercial aspects of our work were concerned we were all very raw – none of us (including our supervisors) really had any experience. We were all learning – and survival required us to climb a pretty steep learning curve.
We did survive, to the extent that by 1974 we had outgrown our base at Imperial College, and we moved to our first real home – the shop in Burlington Road, New Malden referred to at the beginning of the blog.
I guess that New Malden was chosen as being reasonably close to Wimbledon, where Spalding lived. Our offices were in a parade of terraced shops, typical of English suburbs, dating I guess from the 1930s. We had a reception area where the shop itself would have been, and the computer terminal (linked to a mainframe at a computer bureau) was in what had been the “back scullery”. The engineers were located upstairs, in what was previously the living accommodation. My “Aerospace Group” was in the front room, overlooking the road and the shops opposite.
Next door to us on one side was a betting shop (you could sometimes hear the commentary from the televised races through the walls!), and there was another shop of some kind (a newsagent?) on the other side.
The sign over the shop said something like “CHAM Limited Consulting Engineers”. But we must have confused the locals, because on one occasion a local resident came and asked our receptionist if she could book to have her hair done!
It seems to me that this move – in which CHAM separated itself from the “apron strings” of Imperial College for the first time – can reasonably be said to mark the start of the commercial CFD industry as we know it today. “From little acorns” – as they say!
And what was life like for the “CHAM family” in its new home in Burlington Road, New Malden? More on this in the next blog.
- Commercial CFD Starts Here …
- “Sledgehammer CFD” – The Best Approach?
- Beginning at the Beginning … 5 Parabolic or Elliptic? – Or Somewhere In Between?
- Beginning at the Beginning … 4 The Birth of Practical Turbulence Modelling
- Once a Petrolhead …
- Don’t Mistake “Simple to Use CFD” for “Simple CFD”.
- CFD … Or Not CFD?
- Beginning at the Beginning …3 Improving SIMPLE
- Beginning at the Beginning … 2
- If You Can Remember the 1960’s …