This post examines the attributes of state based control and the value delivered to manufacturing from the initial design through the operating life of the facility by improving the effectiveness of operators. This is part 2 of a 5 part post and discusses the background and the cost of not automating.
Before distributed control systems (DCS), board operators could walk the board and see a panel of controllers and recorders with a few strategic alarms. Plants were not operated in monolithic control. The operator had a smaller number of loops in view and reach and operated the plant based on the state that the plant was in, which had its advantages. We do not want to go back to that, but we have lost some things as we have moved forward. We take advantage of the fact that the DCS brings in a tremendous amount of information and alarms on many more loops per operator than in years past. For the most part, we do not take advantage of what the DCS can do with this information in the form of suppressing invalid alarms for the current process state. We also don’t drive the outputs differently based on what the current state requires. The DCS has moved to a monolithic control, creating an unmanageable demand for the operator to now handle the state based aspects for startups, shutdowns, upsets and disturbances. The operator needs to do this with many more loops that are not neatly visible and accessible to him on the board and that are probably in an alarm flood.
Control systems in a manufacturing facility are many times configured for optimal performance during the running state of the plant. This is a problem for the effective operations of other operating states. Since normal and abnormal changes are based on the state of the process, not all possible alarms are relevant to operations in every state. State based control maximizes the digital control system’s ability to detect and convey unique operational situations through dynamic alarming and state based managements of the outputs. In terms of time spent in a state, it seems reasonable to optimize the running state. Hopefully, plants spend most of the time in the running state. In terms of risk, it is not able to deliver. According to the Process Improvement Institute, 70% of incidents occur on startup or shutdown of the plant (Process Improvement Institute, 2011). In large continuous plants, these are the states that everyone has the least amount of experience with and where the control system provides the least amount of support. Without some type of state based control or dynamic alarm management, the control system easily becomes more of a hindrance than a help. In terms of reducing risk, it is much wiser to look beyond the running state. State based control allows operations to manage the plant through states of a startup, shutdown, upset or a disturbance such as Circulating, Heating Up, Conditioning Catalyst, etc.
Safety systems that are based on hazard analysis perform well when managing strategic loops within an acceptable probability of failure on demand. However, the operator is left with how to deal with the rest of the plant when the safety system has just taken the reactor down. For instance, they have to deal with the issue at hand with the reactor and the implications on the up and down stream equipment. They must now operate the plant with many valves in manual, dealing with an alarm flood, while making decisions in a very stressful situation. At shift change the required communication regarding what is in manual may be overlooked. There is no reason to take these risks. The “reactor tripped state” is a state that can be identified and controlled automatically by the DCS. The knowledge for how to control the process is captured in the DCS and leveraged to everyone who runs the board. The reactor system also communicates that it has tripped to the upstream and downstream equipment, so that they can take the appropriate actions.
Through state based control, the operator is transformed to a process manger. This is achieved by moving from making many manual adjustments by hand in an upset, such as an alarm flood, to allowing the state based control to take care of it. In normal running conditions, operators can monitor the plant for optimal performance. In upsets, they can manage the safety of the process and respond to meaningful alarms.
Cost of Not Automating
State based control can dramatically reduce operator errors. There can be some significant cost in terms of operator errors as is documented in “The Cost of Operator Errors and What You Can Do to Minimize” (Henry, Ferrer, Persac & Beebe).
Operator errors result in the highest average dollar loss per major incident at over $80 million (J. H. Marsh & McLennan, 2010). According to the ACR Advisory Group, operator error accounts for 42% of the unscheduled plant shutdowns with equipment failures and process upsets rounding out the list (O'Brien, 2010). State based control reduces all of these.
Stay tuned for part 3 on workforce.