At its, simulation it is a mathematical model of a real world entity – a plant or process, for our purpose. The entity is thought of as a system, characterised by inputs, output and states.
Consider for example a dosing system, variations of which are frequently used in the process industry. At its core, a dosing system might have a volumetric pump which is driven by a variable speed motor. The mechanical power provided by the variable speed drive is an input to the pump. The flow of the dosed fluid is the output. The state of the pump would be one of the following: operating, stopped or failed. Systems also have feedback loops that are essential to maintaining a required state.
Once a plant or process is conceptualised as a system it becomes easier to replace it by a set of equations, applying the principles of mass and energy conservation. In case of our dosing system, if the pump characteristics are known then the amount being dosed can be expressed as a function of the speed at which it is being driven.
If the dosing system is, for example, a dye pump that is being used to impart a certain colour shade to a mix then the ratio of the added amount of dye and the mass of the mix will provide an estimate of the change in colour.
In real life, the change in colour will not be instantaneous and there might be latency arising out of the time that the intermixing of molecules will take.
Should these process time constants be accounted for? The answer is: it depends. It depends on the purpose of the simulation. If the simulation model is being created to accurately predict the timing of colour change then yes.
However, if the purpose is to determine the range of variation of colour, then no. That is, the modelling has to be fit for purpose. It can and should be kept as simple as possible and there is no harm in using approximations. Not every state has to be considered and only those that are relevant to the purpose should be modelled.
Real life plants and processes are anything but simple. They are far more complex than the dosing pump but it is so because they are a combination of a large number of simple subsystems. It is possible to visualise a system as one made up of a large number of subsystems – even if they are physically not.
Mathematical formulae, specific to a subsystem will describe the input, outputs and states of each subsystem. A calculation run for a subsystem will provide the values of its outputs, some of which would be inputs for the formulae describing another system. Thus, a hierarchy of formulae is sufficient to describe a complex system.
Simulation, as a technique, has an impressive history of use. The early adopters of this technology were refineries and power plants. These installations require heavy capital investment and are high risk operations. By building simulation models, these operations were able to provide training tools that helped with safe and efficient operation of the plant.
Viability of operations
The concerns of the early adopters remain the concerns of any Board of Directors that is going to commit a substantial amount of capital to the installation or upgrade of a process plant. The concerns go the heart of viability and sustainability of the proposed operations and can be expressed as the following questions:
- Is the capacity of the plant adequate?
- What will be the cost of production?
- Would the plant run safely without accidents?
- How can human error be minimised?
- Would the production conform to specifications?
- Is the plant scalable to adjust to increases in demand?
- Is the operating strategy optimum?
These questions arise out of the desire to manage and minimise risk. The issues relate to the bottom line performance of the organisation.
If the plant were to suffer catastrophic failures under unforeseen operating conditions then the very viability of the operation would be threatened. Community concerns may prevent restarting of the plant and the whole investment would get written off.
Should the performance not be optimum and in line with budget then the production cost would be high and the profitability of the organisation would be negatively impacted.
If the quality is below par or variable then there would be difficulties experienced in selling the product and revenue would be below budget. Poor process and plant performance clearly results in poor economic performance and will not bode well for the future of the organisation.
It follows that if effort were to be expended at the right time to check and double check the design of the plant and tools provided to learn how to operate the plant safely and efficiently then not only would the viability of the plant be assured but also there would be increased confidence that the plant will perform in a predictable manner. There would be cost savings and prevention of losses and there would be a healthy bottom line.
Simulation is the tool that can make all of this happen.
What happens in simulation is that the plant is converted into a series of mathematical expressions that describe the inputs, outputs and states of the system. These are generally coded in an appropriate computer language and are constantly recalculated in an adjustable cycle.
The results are made available to the user in a variety of forms – pictures, symbols, texts, charts. Each cycle represents a real time interval. If the cycle time is one second but each run is deemed to be equal to five seconds, then the simulation will run at five times the speed.
What transpires in real life over a period of one hour happens in simulated world in 12 minutes. In this regard process and plant simulation differs from gaming software where the emphasis is on getting as close to real time as possible. Just as in gaming software, the user’s inputs change the position or state of screen objects, so also in process simulation, the user’s inputs alter the process variables.
Say, the users want to test the capacity of the plant. They can do so by operating the simulated plant at an accelerated rate and watch the quantity and quality of outputs.
They can manipulate inputs like raw material input or heat transfer, and observe the effect on the states of the plant and the finished material output.
A stage will be reached when inputs can no longer be increased. The output corresponding to that stage will equate to the maximum possible output. Engineers design plants keeping in mind worst case conditions and it may well be that the results of the simulation do no more than validate the engineer’s calculations.
Validate or negate
This is exactly what simulation was designed to do – to validate or to negate, and to do so in a manner that can be easily understood by even non-engineers. That is why results of the simulation are presented in a form that users would expect to see in real life. Moreover, simulation can be carried out in a very wide range of simulated conditions – ambient temperature can be changed, raw material quality can be altered so that the observed output is not just under ideal, design conditions but under extreme conditions as well.
If an error is found or if the simulation shows that the capacity is not adequate, then the engineers can go back to the drawing board and make changes. Those changes, at the initial design stage will be far less expensive than those that may have to be incurred post-commissioning.
With a simulated plant, risks can be taken. It can be taken into unsafe operating states to observe what might happen. A blow up of a simulated plant at worst might be mildly embarrassing but a failure of a real plant would be a catastrophe.
Operators can be trained, using the simulator to gain skills in guiding the plant from abnormal states to normal conditions. Such training cannot safely be provided on a real plant.
With constant practice in starting up, shutting down, emergency response, economic operations, plant operating personnel will gain expertise even before the real plant is commissioned. Simulated production reports provide an accurate estimate of operating costs and provide the basis for costing and budgeting.
Random events occur in plants: parts and subsystems breakdown, human errors, power failures. Engineers design a process to be as robust and tolerant as possible, but it is often a strange combination of random events that results in an accident.
Simulation allows for random events to be generated and by watching accelerated simulation for an equivalent long period of real time, dangerous combinations of events can be detected and design changes made to cope with those combinations. Risks are therefore averted or minimised.
If simulation saves, why is it not in widespread use in industry?
It is probably due to a few misconceptions. It is wrongly assumed simulation is too expensive. If the modelling is appropriate to the purpose for which it is being designed then costs can be kept low.
Simulation is also thought to be time consuming. Well, it can be, if a lot of time is spent on unnecessary animation and graphics.
But if a fit-for-purpose approach is taken, then time wastage would be avoided.
Simulation does not necessarily require expensive hardware and software to run; desktop computers with ubiquitous software packages will suffice for most purposes.
As an added benefit, the effort that goes into understanding the process and plant to translate it into a simulation model will pay off in many ways during the life of the plant.
[Pankaj Rai Mehta (email@example.com) is a Chartered Professional Engineer, Senior Member of IEEE and Fellow of Engineers Australia and is Director of Metamagix.]