Secondary Flow Loops Offer a Primary Advantage

A secondary flow loop offers opportunities for feedforward control, identification of valve stick-slip and backlash, online metrics, diagnostics, rejection of pressure disturbances, linearization, and modeling. However,  a secondary flow loop can be detrimental when the primary loop has a 63% response time less than 10 seconds. If the 63% response time is greater than 5 seconds, an exceptionally fast and precise control valve or a variable speed drive coupled with fast scan rate and tuning can used to make the secondary flow loop faster than the primary loop. This can be done for inline pH control where the 63% response time for well-designed systems with new clean electrodes is 5 to 10 seconds. In the case of liquid pressure control, the 63% process response time is milliseconds. The observed response time is due to automation system digital delays and filter lags. For liquid pressure control, direct manipulation of a variable speed drive with no rate limiting or deadband, a 0.1 sec module execution time, no measurement filter, and no transmitter damping is necessary to deal with incredibly fast pressure disturbances traveling at the speed of sound in the fluid.

Primary loops such as level, pH, gas pressure, and temperature are slow enough to benefit from cascade control to a secondary flow loop in the following ways:

(1) Feedforward control – nearly all feedforward systems end up with the calculation of a flow and nearly all feedback systems end up manipulating a flow. Hence, flow feedforward predominates and the use of a secondary flow loop enables setting and monitoring a flow ratio. If every primary loop had a secondary flow loop, plant-wide feedforward control would be possible (Feedforward control enables flexible, sustainable manufacturing)

(2) Identification of valve backlash and stick-slip – you need a low noise flow measurement to see the results of successively larger output steps (e.g. 0.2%, 0.4%, 0.6% …) in the same and then the reverse direction.

(3) Online metrics and diagnostics – nearly all process metric computations involve flow rates and flow totals.

(4) Diagnostics – you can’t tell if a valve has plugging, flashing, or damaged trim without a flow measurement.

(5) Rejection of pressure disturbances – most fast disturbances start out as pressure changes. A flow loop can correct for them before they appreciably affect the primary process loop.

(6) Linearization – the open loop gain has a gain factor that is the slope of the installed characteristic of the final control element (e.g. control valve or variable speed pump). The installed characteristic is anybody’s guess because it changes with the pump curve, system curve (frictional loss of piping, fittings, manual valves, filters, strainers, …), and source-destination static heads

(7) Modeling – multivariate statistical process control (data analytics) have linear models messed up by using valve signals as inputs. First principle models don’t have enough plant information noted in (6) for the pressure-flow solver resulting in the model valve positions not matching the plant valve positions (Advances in flow and level measurements enhance process knowledge control)

(8) Tracking down disturbances – the first flow that has significantly changed or started to oscillate is normally directly or indirectly associated with the root cause of a disturbance