Synchronous Execution Improves Controllability
For typical PID loops, control performance is about the same when it is done in the DeltaV Controller with classic I/O or on a fieldbus segment within the devices. This may not be true with hybrid control where the control loop spans the fieldbus segment and the controller. This issue involves sampling rates and synchronous versus asynchronous execution and is not limited to the DeltaV software. Any time a host system provides the control using fieldbus devices on an H1 segment, control performance can be sacrificed unless the loop dynamics are sufficiently slow.
The Auto-assign function blocks to H1 port option allows you to achieve synchronous control by automatically assigning a subset of DeltaV function blocks to run in the H1 card.
To understand why control performance can be compromised with hybrid control, compare control in the DeltaV controller using classic I/O with control in fieldbus devices. Control in the controller is asynchronous, that is, the execution of control modules and function blocks is not synchronized with the execution of the I/O cards or I/O bus communication. But analog I/O cards scan at a fast rate (around 25 milliseconds) and I/O bus communication is fast (typically between 20 and 80 milliseconds depending on the number and mix of I/O cards). Even though all these elements are asynchronous, a control module can easily execute at a scan rate of 500 milliseconds without violating the rule of thumb that the control interval be at least three times slower than the longest asynchronous sampler.
Control on the H1 fieldbus segment is synchronous, but the execution rate is somewhat slow due to the low bandwidth bus and low power processors used in the devices. Execution of function blocks on an H1 segment is scheduled in a given scan (called the macrocycle) such that an AI function block in a transmitter executes before the PID block (in the transmitter or valve), which executes before the AO block in the valve. The LAS (Link Active Scheduler), normally the H1 interface card, orchestrates the communication between devices over the fieldbus such that block execution and communication are synchronized. The achievable macrocycle time on a segment is a function of the number and type of devices on the segment and the amount of control and monitoring configured. As a result, it is possible to limit the number of devices on a segment to achieve the desired macrocycle, say 500 milliseconds to 1 second. Synchronous control on the H1 segment at the macrocycle rate is as good or better than control in the controller at the same control interval.
Control performance becomes an issue if the PID block runs in the controller and the AI and AO blocks run in devices on the segment. In this case, the AI and AO blocks execute once each macrocycle, but execution and communication with the PID block in the controller is asynchronous. The difference between this hybrid control and control in the controller using classic I/O is that with classic I/O, input and output data can be transferred to and from the I/O bus every 25 milliseconds. With hybrid control, this transfer of data occurs at the macrocycle rate of about 500 milliseconds to 1 second. There is no real control benefit achieved by executing the PID block in the controller faster than three times the macrocycle rate. If the macrocycle is 500 milliseconds, the fastest control interval of benefit is 2 seconds. A 1 second macrocycle supports a practical control interval no faster than the 5 second option. Therefore, hybrid control can compromise the controllability of loops with fast dynamics.
As can be seen in the Macrocycle Viewer, when one or more blocks run in the controller, the control loop is not synchronized to the macrocycle and all blocks in the control loop run at the beginning of the macrocycle followed by scheduled Compel Data transfer messages. This allows the H1 card to optimize non-scheduled communication since all scheduled CD messages occur successively and devices that quickly respond with CD response messages will free up additional bandwidth for unscheduled traffic.
Conserves Controller CPU Resources
Running a function block in a field device instead of in the controller reduces the block's controller CPU demand by approximately 50 percent. Running the block in the device eliminates the demand on the controller CPU caused by the execution of the block's control algorithm. However, there is processing required to communicate view list data between the field block and the extended block (sometimes called a shadow block) in the controller. The net result is about a 50 percent reduction in controller CPU resources when the block resides in the device.
Reduces the Number of VCRs on the Segment
There are two types of VCRs (Virtual Communication Relationships) on each port: publisher and subscriber VCRs. An H1 interface card can support up to 50 total publisher and subscriber links on each port. Running the PID block in the controller consumes three VCRs on the port for one simple loop. Running the PID block in the transmitter consumes two VCRs and running it in the valve consumes one VCR. Running one VCR or two VCRs does not significantly reduce the number of devices the segment will support. However, using three VCRs for a simple loop can significantly reduce the number of devices the segment will support.