click for the Intuitive Interactions home page

"Intuitive Interactions"


John McCulloch - Dynamic Simulation

Home Page

Contact John

Curriculum Vitae

Ethical Statement


Alarm Management

Operator Interface Design

Dynamic Simulation

Control System Configuration

Process Control Troubleshooting

Control Loop Tuning


RPN Calculators

Fractals

Photography

Solar Energy

Wood-turning

Community Activities

Torphichen Kirk

I have worked with a number of different dynamic simulation tools including differential equation solving software, but I have found the most satisfactory technique from the Control Engineer's perspective to be the incremental real-time approach. In this, the process is represented by a set of integral expressions that are converted into discrete time incremental relationships. The time increment is chosen to be small enough that no significant errors occur, (this can be tested by increasing the time interval, and determining the point at which significant differences in the results start to occur).

The resultant model is straightforward to build, easy to maintain and simple to modify to test process or control system alternatives. It operates in real time and gives a user the look and feel of the real process. Developing and tuning a control system is just like working with a real plant.

I have built operator training simulators using this system and also modelled processes for control system design and testing purposes. A typical model that I built was a glycol regeneration circuit, which included the gas drying column, flash drum, distillation column and accumulator with all pumps, heat exchangers and filters. The model included the full sequential and regulatory control system with its operator interface schematic displays, motor control logic for the pumps and a model of the emergency shut-down system. The model included process noise, (random variations in the signals from transmitters, in the flows through control valves and in the qualities of the utilities), control valve lag and hysteresis and all non-linearities. This model used the standard Emerson DeltaV algorithm set, and took about nine man-weeks to build, (including acquisition of the process data), which compares very favourably with other dynamic modelling techniques.

This technique can also be used for simulating complex parts of the process to design or evaluate a control system. This is especially useful if the process is one where an incorrectly tuned control system can cause costly trips and upsets. In several cases, the tuning constants developed with the model have been applied to the real plant and first-time start-up has been achieved without further adjustment. For control development, it is not necessary to include the elaboration of a full operator interface and emergency shut-down system, and so the models can typically be constructed in one or two weeks, depending on the complexity of the process. For control system development I have modelled:
  • Multiple steam pressure headers with boilers, turbines, let-down stations and desuperheaters;
  • Effluent pH control systems;
  • Gas supply and distribution systems;
  • Distillation columns with their reboilers and condensers.
I have also used the technique for accelerated real-time working, for very slow-moving situations. I simulated some supply-demand network situations including order processing delays, transportation logistics, money flows and production capacity variations. Typically a day of operation was represented by one second of real-time, (a year thus took about 6 minutes), as the intention was to examine the long-term effects of different methods of managing a production unit in the network. It was interesting to observe that traditional responses to demand variations led to the same sort of "boom-bust" behaviour in upstream production units that can often be seen in reality. By modifying the approach, much more stable operation could be achieved with significant benefits to the cash flow.

I have used a variety of different modelling tools. The DeltaV system mentioned above was very efficient. I have used the Honeywell TDC algorithm set, but this was very much slower and more difficult to use. For a long time I used the TCS "Tactician" system as a modelling tool. I have also used Visual Basic successfully: it is very versatile, but can be a little tedious as all control algorithms, and the user interfaces to them, have to be constructed from first principles.













This page last updated on 2013-08-06 JGM