October 2007
FEATURES
From the
Editor
Welcome to the October issue of Focus on Plant Optimization. This e-newsletter presents ideas on easy-to-do activities that will make your plant operations more effective, and ultimately more competitive. As we bump into interesting ideas,
we'll publish them here. If you have some suggestions for your colleagues, please send them in for our review. Well publish them under your name or anonymously if you prefer. Enjoy the holidays, take time with your family if you have the chance, and we'll be in touch in the New Year.
Jeff Chan,
JeffC@MachineAutomation.ca
Why assuming a normal distribution in process control is often
incorrect!
When performing statistical calculations it is common to assume that all probability distributions are normal i.e. they follow
the familiar "bell shaped
curve". However in process control, variables tend to appear as noisy sine waves oscillating around operation setpoints. A sine wave however is bimodal and will produce a distribution with 2 peaks! This can lead to misleading values when presenting calculations
like standard deviation.
What Alarm
Burst
Rates are telling you
According to EEMUA guidelines, Alarm Burst Rates are calculated as the total number of process alarms
that have occurred in any 10 minute interval. Best practices are
that no more than 10 alarms should occur. Burst rates in
excess of 10/10min. directly impacts the operator's ability
to understand and more importantly, respond to the alarm
conditions. As a result, alarm burst rates tells you when
operators are being overloaded and safety and efficiency
is decreased. How well does your plant stack up? AlarmAnalyst gives you these and many other important
benchmarks.
What the
heck is Cross Correlation?
We all remember our high school science class where we had to
record some measurements and plot the results - usually a few
points on a chart then use a ruler to draw a best fit line
through it. This gave us a good idea what or if the
relationship was between 2 variables. But what if a direct
relationship existed but was shifted in time by several minutes or hours? There is a good chance that we might have concluded
there was no relationship at all. Cross Correlation give us a
way to determine when and by how much variables are related regardless
of time delay.
So what does this have to do with running my plant
better? Often downstream instability
is due to factors taking place further up the line.
Cross correlation is a key tool used in reducing plant
variability and detects relationships not obvious just by
looking at snap shot data. When cause and effect can be
shifted by several hours the reason for down stream instabilities
becomes less obvious and may remain a total mystery. PlantAnalyst provides
advanced capabilities for the monitoring and reduction of
process variability including cross correlation...
NEWS AND
PRODUCT DEVELOPMENTS
Customer
Input
We have been feverishly adding customer suggestions into our applications
over the last few weeks - a major new release in AlarmAnalyst
is expected early in 2006 with improved look, performance and
features. If there are features you would like to have
added please send a quick note to support@machineAutomation.ca
|
