Re: [Roll] [6tsch] comments on draft-phinney-roll-rpl-industrial-applicability

[Tom Phinney <[e-mail]>]

The continuous and discrete manufacturing parts of hybrid manufacturing plants are ALWAYS connected; it would be pointless to build such a plant if that were not the case.

It is useful to consider the differences. Continuous processes use closed-loop control based on fixed-period process sampling and outputs to process actuators. The highest frequency of such control loops is usually 4 H, though a few compressor surge control loops that operate at a 9 Hz to 11 Hz rate may also be present. These process control systems are potential users of wireless communications because their fixed timing is compatible (in theory) with TSCH-like time schedules.

It is important to understand that the publish-subscribe messaging used for process automation does not require 100% delivery. The control algorithms traditionally are designed to tolerate the loss of three or fewer consecutive messages before receiving another message in the series. However, typically loss of four consecutive messages triggers the equipment to shift to a fallback control strategy. It is this statistic that leads to use of duocast (with its approximate squaring of message loss rates) for high-speed loops, since most plant managers will not tolerate more than one such shift to a fallback control strategy per year, per plant, and there can be thousands of 1 Hz and 4 Hz control loops operating in the plant.

Discrete processes use a different form of control that is cyclic but not periodic. This is usually implemented in PLCs (programmable logic controllers) that execute a variable-execution-duration program that repeats as soon as it finishes each iteration, leading to cyclic but aperiodic behavior. The duration of each execution cycle is usually a few ms, typically 3 ms to 10 ms, with a perhaps 50% maximum variance in cycle-to-cycle.execution time. The messaging involved in such automation typically require higher data rates than those offered by IEEE 802.15.4, and any wireless messaging itself would typically be single-hop, from a router/local-controller that is part of the manufacturing device to multiple wireless backbone routers arranged high on opposite walls in a roofed factory. High-data-rate optical signaling would be a good match for this communication.

The variance in cycle-to-cycle times provides a significant challenge for comm technologies based on scheduled pairing of senders and receivers. That, coupled with the typically single-hop aspect of the communication, makes it likely that a non-TSCH approach will be optimal for such systems.

The communication between the continuous process controllers of a plant and the discrete automation controllers typically occurs on a 100 Mbit/s or faster backbone comm link. Wireless is not involved, and there is no reason to anticipate that it will be in the future. Since the physical systems that implement the continuous and discrete processes are interconnected (e.g., pipes with valves connected to bottle handling-and-filling apparatus in a soda bottling plant), wired interconnection between backbone controllers is typically not a significant issue.

I hope that the above discussion helps understand the differences in operating mode, timescales, comm needs and other high-level aspects of the continuous and discrete portions of a hybrid plant. There are others receiving this e-mail who are much more familiar with such plants and processes than I, so if I have erred substantially, I trust that they will post a correction to this list.

-Tom
=====
On 2013.03.01 13:03, Michael Richardson wrote:

"Tom" == Tom Phinney <[e-mail]> writes:

    Tom>    Apologies to those who receive this e-mail on more than one
    Tom> list.  First, the traditional industry differentiation between
    Tom> "factory automation" and "process control" is that factory
    Tom> automation is focused on discrete manufacturing steps (and thus
    Tom> is often referred to as "discrete parts manufacturing") while
    Tom> process control involves continuous processing of fluids and
    Tom> quasi-fluids (e.g., sand, cereal and other materials whose flow
    Tom> rate can be measured by a mass flowmeter).  Many real-world
    Tom> plants use hybrids of these two manufacturing styles.  

So, given that many plants are hybrids, are you saying that this
document applies only to the process control parts of the plant?

Or that hybrid plants are also out of scope?

If there are scope constraints, what exactly are they? 

What happens in a hydrid plant might be significantly more complicated,
as it involves two kinds of networks; if there is a desire to have the
different pieces interconnected.

_______________________________________________ 6tsch mailing list [e-mail] https://www.ietf.org/mailman/listinfo/6tsch" rel="nofollow">https://www.ietf.org/mailman/listinfo/6tsch