Question on FF-SIF 100ms Response

[yaojie]

SIS system quick interlock trip on DI/DO application required 100ms response , could FF-SIF meet this requirement? And how to set FF segment Micro-schedule to meet it? It would be more appreciated to provide related document. :(

[rezabejd]

I have not seen a host that supports macrocycles below 150 ms. They may exist. I would even have to do some serious work on my "conventional" SIS to guarantee that sort of response.

Consider a pressure interlock w/ 3 voting PT's. If each can solve its AI in 20 ms - this can happen concurrently, then three "compel data" at 30 ms each, you have consumed 110 ms already. Your "voter" block would consume another 30 ms compel data to send its DO out to the field device, which would also conceivably have its own "compel data" for a position "readback" signal. So even without the logic / voter block execution times, you would be over 150 ms.

If you are mitigating a serious hazard and really need 100 ms response (process excursion > detection > final element) Foundation fieldbus probably doesn't have a commercially specifiable solution yet.

[jberge]

First, let's look at control: The transmission of a single real-time value over fieldbus can take place in as little as 15 ms (some field devices and DCS for control already support that since a few years back, and this is used in some complex control loops that need to meet UOP requirements). Note that DCS are by default set for a more conservative 30 ms link time to be compatible with older and more sluggish devices. However, you can set is more agressively if needed.

Now, let's get back to SIF. Once this stuff becomes available, you could calculate one PV and one shutdown signal for a simple SIF, at 15 ms each. This leaves 70 ms for the logic solver and for the function blocks in the transmitter and the valve positioner if you want a 100 ms macrocycle. I believe the SIF function blocks are simplified (faster) so it appears this could be feasible. If not in the first generation of FF-SIF, maybe in the next.