When I started in flow measurement, flow computers were not so called. They were simple electro mechanical devices for flow totalisation and batch control. Simple pulse counters with little or no sophistications. Some, I remember, were built around post office Uniselectors, and dials with stepping motors and relay logic.
I later was much impressed with those wonderful displays with the different digits appearing at different depths within the display.
The first company to introduce micro-processors into flow computers were Solartron; these were their fiscal flow computers for refinery and pipeline applications but soon became used for much more.
The sophistication of the programs was impressive and still is as they become ever more complex.
Flow computers tended to be dedicated single loop systems.
Later they introduced digital coms to communicate with PCs, SCAD systems etc.
They have tended to be rich in dedicated I/O; analogue inputs and outputs, alarm relays, density and viscosity transducer inputs and power supplies, flow meter inputs of various types and so on. Rudimentary data logging and alarm logging came along.
Computers evolved into a supervisory role and data logging role.
Soemwhere in the middle were some pretty neat programs for PCs to run flow programs.
Now we should see further evolutions of flow computers, still single loop or multi-channel (perhaps 4 meter systems say) but with more computer technology such as LCD screens and Windows style environments, user programming capability and expandable I/O etc they are set for far greater flexability.
In other words, we are seeing a rapid blurring of the margins between what is a what and what one should use.
A non-aristolean approach to control devices?
JMW