The White Water Butterfly Effect (Confessions of a Looney Tuner)

By Sylvain Millette
Millette Control Engineering Inc.
[email protected]

Butterfly Effect: The propensity of a system to be sensitive to initial conditions. Such systems, over time become unpredictable; giving rise to the idea that a butterfly flapping its wings in one part of the world can cause a tornado in another remote area of the world.
I still remember my very first paper machine control audit, seventeen years ago. The mill in question was in a typical remote location, and produced linerboard. We collected basis weight data, while the scanner was in fixed point in the center of the sheet (to the uninitiated, basis weight is the total weight of the paper, fiber and moisture). Collecting data in this manner shows the true Machine Direction (MD) variability. We quickly discovered that there was a pronounced 90-second cycle in the basis weight. It was clear that this cycle was one of the main culprits behind the machine’s poor runnability record.
The week-long investigation eventually uncovered the source of this basis weight cycle: the tertiary cleaner standpipe level controller. If this loop sounds remote, away from the main action, you are only half-right. You see, paper machine processes are quite linear, with some offshoots that either feedback into the main process branch or eventually goes to waste. In this case, the level controller is an offshoot of an offshoot. So how could such a marginal loop affect the stability of the paper machine? The answer lies in the white water dilution header.
White water has many uses in papermaking processes. Its most important function is to dilute and control the pulp consistency. An even consistency equates to a stable basis weight. White water is also used to control certain tank levels. This is where our level controller comes into play. In our case, the tank was only a standpipe, and had the retention time of a toilet bowl. The investigation revealed that due to poor tuning, the level controller cycled every 90 seconds. The cycling, in turn caused the level control valve to cycle as well. Since the level control valve took water from the main white water header, it caused the overall header pressure to cycle. This pressure swing consequently caused all the consistency controllers to swing – including the machine chest consistency controller. This controller is the "last line of defense" before the stock hits the machine. So as the hip bone is connected to the thigh bone, a level standpipe level controller caused a machine chest consistency swing.
Little did I know that my first exposure to papermaking would unearth a chronic design problem in the pulp and paper industry. After years of encountering this syndrome I eventually gave it a name: The White Water Butterfly Effect. This Butterfly Effect can be the end result of a series of mutations or it can spontaneously generate in new mill designs. This condition occurs when a single pump supplies dilution water to too many control loops. Due to the interconnectivity of the header, seemingly insignificant distant control loops can have a devastating impact on critical loop. Eventually all the loops fed by this common header end up "storming" together. The current record number of loops fed by a single header is sixteen and tuning them is virtually impossible.
Some people suggest that installing a pressure controller on the dilution header fixes this problem. Unfortunately, since a PID controller is only reactive, it is not guaranteed to eliminate pressure disturbances, in fact, the end result can be quite the opposite. In essence, a dilution header pressure controller only introduces one more butterfly to the system; the end result could be the Perfect Storm.
Simplicity is the key to unraveling this effect. First, the consistency controllers require adequate header pressure – no less than 40 PSI (280 kPa), preferably between 45 and 50 PSI (310 to 350 kPa). In some cases, I have seen pressures as low as 15 PSI, which is barely enough pressure to overcome the static head of a stock chest.
Second (and this is where simplicity comes into play), the header must be split into two: only one will feed consistency control loops; the second header will feed all the other loops requiring white water. Pressure variations on this second header will not likely have any significant impact on the process stability.
Third, the number of loops "hanging off" the first so-called consistency-header must be kept to a bare minimum, no more than five to six consistency loops. This will minimize any interactions.
Fourth, the supply pump for the consistency-header must be sized such that it operates in the flat part of the curve. Again, this will minimize pressure variations under changing demands of white water. This also eliminates the requirement for a pressure controller. Lastly, the header should be kept as short as possible.
White Water Butterfly Effect: The propensity for critical consistency control loops to be destabilized, via the white water dilution header, by seemingly inconsequential or innocuous control loops.