EMONITOR Odyssey
CATEGORIES AND ALARMS
By
Rick Wetzel
Entek IRD International
Alarming based on spectrum data is one of the most important features of any predictive maintenance
system. EMONITOR Odyssey includes the most powerful features in the industry for using band and
spectrum alarms. Categories extend these powerful alarm features by building groups of similar machines.
Alarm parameters and statistics can then be easily applied to this entire category. This paper summarizes
the setup and use of band and narrowband alarms and categories. Basic strategies for implementing alarms
and categories in your predictive maintenance program are also presented.
CATEGORIES
Categories are a powerful feature of the EMONITOR Odyssey that allows you to setup groups of similar
machines. This group is called the category and many similar setup features can be shared for this group.
The information that can be shared through using a category includes overall alarm values, the definition of
the frequency bands for band alarms and the band alarm amplitudes. In addition to these parameters, the
category can be used to store statistical information for a group of machines. This statistical information is
an excellent tool to build statistical spectrum and band alarms as described below. The category is set up by
following these three steps:
1) Create the band set for the category.
2) Configure the category.
3) Enter category information.
1) Define the band set for the category.
The band set is a basic part of the category and must be setup before the category is created, so that it may
be selected and used for that category. The band set may be generic in nature or may be somewhat tailored
to a specific group of machinery. In this example we will be working with a group of process air blowers,
so we will be creating a band set and category for these specific machines and naming it as such.
The band setup window is shown Figure 1 and is used to enter the band information. This window consists
of a chart that allows you to enter the name of the Band Set, the Band Name for each individual band, the
Alarm On choice, the frequency Minimum and Units, and the Frequency Maximum and Units.
1
Enteract Home
Figure 1. Band setup window.
The Alarm On field allows you to choose Band Max or Band Sum. This choice determines whether the
alarm comparison is made on the peak vibration amplitude found within the band (Band Max) or the total of
the vibration across the band
(Band Sum). The band max and
0.1
band sum will differ from being
0.09
very similar in the case of a
0.08
band dominated by a single peak
0.07
to being significantly different
Amplitude
0.06
in the case of a band with
0.05
multiple vibrations peaks.
0.04
Figure 2. and Figure 3. present
0.03
an example of each of these two 0.02
cases. 0.01
0
100 102 104 106 108 110 112 114 116 118 120
Frequency
Figure 2. Frequency Band with Single Peak
Each of these figures shows the
same frequency range from 100 0.1
to 120 Hz. In Figure 2., this 0.09
band is dominated by a single 0.08
peak of .1 IPS at 110 Hz. In 0.07
Figure 3, the frequency band 0.06
includes multiple peaks with the 0.05
largest peak also at .1 IPS at a 0.04
frequency of 110 Hz. Multiple 0.03
peaks of this type are often 0.02
associated with gear or bearing 0.01
deterioration. Table 4. 0
100 102 104 106 108 110 112 114 116 118 120
summarizes the band max and Frequency
band sum values for these two
examples. There is a dramatic Figure 3. Frequency Band with Multiple Peaks
difference between the max and sum values for the case of a band with multiple peaks, although the sum
value is only slightly higher in the case of a band with a single peak.
2
Enteract Home White Beginning
Considering these results, it is usually a good Band Sum Band Max
recommendation to use the band sum method. This Figure 2
offers the advantages of reasonable accuracy in Single Peak .12 .1
situations where there is only a single peak of interest Figure 3
and accurate representation of the total vibration in the Multiple Peaks .19 .1
case of multiple peaks. Table 4. Comparison of Band Sum and Band
Peak Calculations
The units field of the band
setup window allows you Machine A Machine B
to select Hertz(Hz), CPM (1800 RPM) (3600 RPM)
or orders for the input of Frequency of Interest CPM Hz Order CPM Hz Order
frequency minimum and Running Speed 1800 30 1 3600 60 1
maximum. Hertz are 2 X Running Speed 3600 60 2 7200 120 2
cycles per second and Bearing Frequencies 18000 300 10 36000 600 10
CPM are cycles per Table 5. Comparison of CPM, Hz and Orders
minute. One Hertz is
equal to 60 cycles per minute. Orders are used to specify a frequency as a multiple of a fundamental
rotational speed. Orders offers the advantage of using the same band set with machines running at different
speeds. Table 5 shows a comparison of Hz, CPM and orders as applied to two machines running at
different speeds. In EMONITOR Odyssey, the reference machine RPM when using orders is always taken
from the RPM column of the LOCATION pane.
2) Configure the Category.
Figure 6. Configuring a Category
Configuring the category is a first step of entering basic setup information for the category. It is a very
simple step but also a very important one, because without this basic setup information, the category won’t
operate properly and give you the results that you expect. The basic setup information that is entered when
the category is configured consist of the units of interest (G’s, IPS, mm/s, mils, micrometers) and the band
set. The band set was created in the previous step and it defines the frequency limits for any band alarms
that will be used. The units are setup to provide the proper storage locations for any different units that will
be used in this category. Categories allow you to set up alarm values for magnitude (overall) readings as
well as specific frequency bands. The alarm amplitudes and any statistical calculations will certainly vary
depending on the measurement units. Selecting the units of interest by configuring the category allows you
to select which units will have category information setup and maintained.
3
Enteract Home White Beginning
We will be working with an example from a group of process air blowers (P/A Blowers) where magnitude
measurements in G’s high frequency and magnitude and spectrum measurements in IPS have been taken.
The example Category Configuration menu shown above in Figure 6 has been setup to use the P/A Blower
band set and to store category information for G’s high frequency and IPS measurements.
3) Enter the Category Information
The final step in setting up a category is to enter the category information. The category information
includes a wide range of information. For straightforward applications, only a few values need to be
entered. For the most advanced applications where the full power of category alarms will be used, more
information can be entered and used to establish the alarms. This information can be broken down into
category variables and band variables. Category variables are used for setting the magnitude or spectrum
alarms for the machinery in the category. Band variables are used in the setting of band alarms for the
machinery in the category. A separate set of category and band variables is entered for each unit where they
will be used.
The category variables include:
uvar 1, uvar 2 User variables for setting additional user accessed values such as secondary
alarms or percentage change levels.
HI ALARM1 Sets the first high alarm value for the category.
HI ALARM2 Sets the second high alarm value for the category.
LO ALARM1 Set the first low alarm value for the category for out of window or below
alarms.
LO ALARM2 Sets the second low alarm value for the category for out of window or below
alarms.
MIN THRESHLD Sets a minimum value for detecting alarms for the category. This is useful to
prevent false alarms from percent change or rate of change alarms.
MAX THRESHLD Sets a maximum value for detecting alarms for the category. This is useful to
prevent missing alarms when using percent change or rate of change alarms.
MIN ALARM Sets a minimum value for detecting alarms for the category. This is useful to
prevent false alarms from percent change or rate of change alarms.
% CHANGE Sets the alarm value for the category as a percent change. Note that a
percentage value is entered not a decimal (i.e. 50% not .5).
TRANSDUCER LO Sets a low comparison value that may be used to indicate a transducer
problem.
TRANSDUCER HI Sets a high comparison value that may be used to indicate a transducer
problem.
The band variables are very similar to the category variables, with the distinction being that they apply
strictly to the amplitudes within the bands as defined by the band set. Because these values apply to each
band, a chart showing these variables for each band is used for entering the values. The band variables
include:
4
Enteract Home White Beginning
% CHANGE Sets the band alarm value for the category as a percent change. Note that a
percentage value is entered not a decimal (i.e. 50% not .5).
HI ALARM1 Sets the first high band alarm value for the category.
HI ALARM2 Sets the second high band alarm value for the category.
LO ALARM1 Set the first low band alarm value for the category for out of window or below
alarms.
LO ALARM2 Sets the second low band alarm value for the category for out of window or
below alarms.
MIN THRESHLD Sets a minimum value for detecting alarms in a band for the category. This is
useful to prevent false alarms from percent change or rate of change alarms.
MAX THRESHLD Sets a maximum value for detecting alarms in a band for the category. This is
useful to prevent missing alarms when using percent change or rate of change
alarms.
MIN ALARM Sets a minimum value for detecting alarms in a band for the category. This is
useful to prevent false alarms from percent change or rate of change alarms.
bvar1, bvar2, bvar3, bvar4 Band variables for setting additional user accessed values such as secondary
alarms or percentage change levels.
The example category configured above included G’s high frequency measurements and IPS measurements.
Magnitudes only were used for the G’s high frequency measurements and both magnitude and spectrum
were used for the IPS measurements. With the wide range of category and band variables discussed above,
the task of using categories may at first appear to be difficult. A few simple ideas will help you get the most
out of categories.
Keep it simple! - Use only the category or band values that will be useful to you. Most people use two
magnitude alarm levels, such as warning and alert levels. When you get started with categories, concentrate
on setting up category variables HI ALARM1 and HI ALARM2. For band alarms, use a similar approach
working with HI ALARM1 and HI ALARM2, only if you are planning two levels of band alarms.
Don’t be afraid to make mistakes. - One of the great things about EMONITOR Odyssey in general and
categories in particular is how easy it is to change your setup. It is very easy to review the category alarm
levels you have set, and if you’re not satisfied to change them.
For our example, we will set up HI ALARM1 and HI ALARM2 for the high frequency G’s readings as
shown on the Category menu in Figure 7.
5
Enteract Home White Beginning
Figure 7. Category Values for G’s High Frequency
For the IPS velocity measurements, a little more setup is required, so that we can establish alarm values for
each of the bands we have setup. Figure 8 shows both the category variables and the band variables that
have been setup for the IPS measurements.
Figure 8. Category and Band Values for IPS Measurement
6
Enteract Home White Beginning
Now the category is complete and in later sections, it will be used for the fast and easy setup of magnitude,
band and spectrum alarms.
ALARM SEVERITY
One of the first concepts to
consider in setting up alarms is
severity. Severity is an
indication of the level of the
alarm condition. A severity is
assigned to every magnitude,
band or spectrum alarm.
Typical severity labels would be
warning, alert, critical or danger.
Each level of alarm is assigned a
severity label. The more serious
the alarm condition, the greater
the severity. A standard set of Figure 9. Alarm Severity Window
severities is provided upon
installation of the software. These standard entries can be customized to change the descriptive label or the
color assigned to each severity. The alarm severity window can be found in the Options menu under Setup.
Severity in EMONITOR Odyssey is a very useful feature. Because a severity is assigned to all alarms, it
does not matter what type of data has been collected or the types of alarm that have been set up. The
severity can be used for sorting and reporting purposes. Data can be selected for analysis, plotting,
reporting or further data collection using the powerful tagging functions of EMONITOR Odyssey.
Severity can be used as a criteria for this sorting, allowing you to identify all machines in a warning
condition or only those with the most serious problems that may be in an alert or danger condition.
MAGNITUDE ALARMS
Once the severities have been established, alarms may be set up using the alarm view of the database. The
category for each measurement location is entered on the location pane. The details of the alarm setup are
entered on the alarm pane. Both of these panes may be seen in Figure 10.
7
Enteract Home White Beginning
Figure 10. Alarm View of the Database
The alarm pane includes columns for entry of the severity, alarm method, alarm definition, active and
trigger. The severity column is used to select the appropriate severity from the list described above. The
alarm method column is used to select how the alarm will be setup. Choices for the alarm method include
the following:
Mag Constant Fixed value used for alarm comparison
Mag in Window Fixed values used to set low and high boundaries. Alarm condition occurs
when the value falls in the window between these low and high limits.
Mag Peak Alarm value is based on the highest previous stored data.
Mag Stat. Alarm value is based on the mean and the standard deviation of the stored
data.
Mag Stat. Indicator Alarm value is based on the mean and the standard deviation of the stored data
from this measurement only. The calculation can also be limited to a selected
number of the most recent values.
Mag Baseline Alarm value is based on the baseline reading for this measurement.
Mag Percent Change Alarm value is based on a percentage change from the most recent previous
value.
8
Enteract Home White Beginning
Mag Rate of Change Alarm value is based a percentage change per unit time from the most recent
previous value.
Alarm values mentioned above may be entered for each measurement or can be referenced to the category.
After making the selection of the alarm type, the specific information that will establish the alarms is
entered in the alarm definition menu. The alarm definition column gives a summary of the information
entered in the alarm definition. An example alarm definition for a mag constant, mag statistical and mag
percent change are presented and discussed below.
Figure 11 shows a magnitude constant alarm
setup window. This window allows the user
to enter the alarm value and to determine
whether it will be applied to magnitude or
phase data and as a limit above or below the
data. The alarm value is entered in the
Constant Value box. This value may be
entered as a fixed number or the drop down
selection list may be used to choose one of
the category variables. In this example the
category variable HI ALARM 1 has been
entered as the alarm value.
Figure 11. Magnitude Constant Alarm Window
Figure 12 shows a magnitude statistical
alarm setup window. This window allows
the user to use a statistic analysis of the
stored data to determine the alarm values.
The mean and standard deviation of the
stored data for this measurement is used to
calculate the alarm level. The Include from
Category checkbox will calculate the alarm
limit based on the stored data for the entire
category. The alarm value is determined
by incrementing the mean by the percent or
by the number of standard deviations
entered.
Figure 12. Magnitude Statistical Alarm Window
9
Enteract Home White Beginning
Figure 13 shows a magnitude
percentage change alarm window. This
window is used to set the alarm limit
based on a percentage change from the
last stored reading for this
measurement. The percent of change
may be entered as fixed number or the
drop down selection list may be used to
select from the category variables. The
cutoff amplitude allows you to specify
the minimum amplitude that will be
considered as an alarm condition. A
fixed number or category variable may
be entered as the cutoff amplitude.
Figure 13. Magnitude Percent Change Alarm
BAND ALARMS
Band alarms are set up in much the same way as magnitude alarms. The methods available for the
calculation of band alarms is also very similar to the methods described above for magnitude alarms. In the
case of band alarms, though, the calculation is performed over the entire band instead of just for a single
magnitude value. The band alarm methods are:
Band Constant Band variable selected as a fixed value used for band alarm comparison.
Band in Window Band variables selected to set low and high boundaries. Alarm condition
occurs when the band value falls in the window between these low and high
limits.
Band Peak Alarm value is based on the highest previous band amplitude from the
measurement’s stored spectrum history or from the entire category.
Band Stat. Alarm value is based on the band mean and the band standard deviation from
the measurement’s stored spectrum history or from the entire category.
Band Stat. Indicator Alarm value is based on the band mean and the band standard deviation of the
stored spectrum history from this measurement only. The calculation can also
be limited to a selected number of the most recent spectra.
Band Baseline Alarm value is based on the band amplitudes from the baseline spectrum for
this measurement.
Band Percent Change Alarm value is based on a percentage change in the band amplitude of the
most recent previous spectrum.
Band Rate of Change Alarm value is based on a percentage change in the band amplitude of the
most recent previous spectrum per unit time.
10
Enteract Home White Beginning
Figure 14. shows the band constant alarm
window. This window is used to set up a
band constant alarm. The drop down
selection list allows you to select any of the
band variables to establish the alarm limits.
Because each band variable has an entry for
each band in the band set, this single entry
provides the alarm values for all the bands.
Figure 14. Band Constant Alarm Window
Figure 15. shows the band statistical
alarm window. This window is used
to set up a band statistical alarm. The
Include from Category checkbox will
calculate the alarm limit based on the
stored data for the entire category.
The drop down selection list allows
you to select any of the category
variables to establish the alarm limits.
The alarm value for each band is
determined by incrementing the mean
amplitude for each band by the percent
or number of standard deviations
entered.
Figure 15. Band Statistical Alarm Window
SPECTRUM ALARMS
Spectrum alarms extend the capabilities of band alarms by providing alarm values not just for selected
frequency bands within the collected spectrum data, but for every line of resolution. The entire spectrum is
compared to a reference alarm value to detect problems that may cause changes at any frequency. The
spectrum alarm methods are:
Spec Peak Spectrum alarm is based on the highest previous amplitude at each frequency
from the measurement’s stored spectrum history or from the entire category.
Spec Stat. Spectrum alarm is based on the mean and the standard deviation at each
frequency from the measurement’s stored spectrum history or from the entire
category.
Spec Stat. Indicator Spectrum alarm is based on the mean and the standard deviation at each
frequency from the stored spectrum history from this measurement only. The
calculation can also be limited to a selected number of the most recent spectra.
11
Enteract Home White Beginning
Spec Baseline Spectrum alarm is based on the amplitude at each frequency from the baseline
spectrum for this measurement.
Figure 16. shows the spectrum
statistical alarm window. This
window is used to set up a spectrum
statistical alarm. The alarm
threshold minimum is used to
determine the smallest spectrum
amplitude that will be considered as
an alarm condition. This is used to
prevent the large percentage
changes in amplitude that may occur
near the noise floor of the
measurement from causing alarm
indications. The drop down
selection list allows you to select
any of the category variables to set
this minimum threshold.
The alarm threshold maximum is
used to determine the largest
spectrum amplitude that will be
considered as a valid alarm
condition. This is used to prevent
the spectrum alarm from being set
unreasonably due to excessive
amplitudes in the database. The
drop down selection list allows you
to select any of the category
variables to set this maximum
threshold.
Figure 15. Spectrum Statistical Alarm Window
The Include from Category checkbox will calculate the spectrum alarm based on the stored data for the
entire category. The drop down selection list allows you to select any of the category variables to establish
the alarm limits. The alarm value for each band is determined by incrementing the mean amplitude for each
line of resolution by the percent or number of standard deviations entered.
SETTING UP STATISTICAL ALARMS BASED ON CATEGORIES
The previous sections of this paper present the definitions and menus necessary to set up categories and use
them in the creation of magnitude, band or spectrum alarms. The category is designed to hold standard
alarm limits for the machinery included as well as a host of other information. One other very important
type of information held by the category is the statistical characteristics of the collected data. This
statistical information is represented by the mean and standard deviation values for magnitude and spectrum
data. The menu examples above step through typical setup windows for statistical alarms, but there is also a
procedure that must be followed in order to initially calculate the statistical information so that it may be
stored with the category. This procedure consists of the following steps:
1) Verify that the desired category and alarm limits have been set up.
12
Enteract Home White Beginning
2) Tag all measurements from the selected category.
3) Calculate the statistical information for the category using the Generate Alarm Statistics
function.
This process will take from a few minutes to a few hours based on the number of measurements taken on all
the machinery in the category and the amount of stored data for each measurement. Immediately upon
completion of this calculation, all the statistical alarms set up in the database will be ready for use. One of
the great benefits of the category approach used in EMONITOR Odyssey is that the raw statistical
information is stored, not the calculated alarm value. This means that statistical alarms can be adjusted
quickly and the resulting new alarm levels can be reviewed immediately. The Generate Alarm Statistics
only needs to be repeated when a significant amount of new data has been collected or the general condition
of the machinery has changed. In a mature program, this step is probably only necessary about every six
months.
ALARM SETUP RECOMMENDATIONS
Here are a few suggestions that may make using EMONITOR Odyssey powerful alarm features a little more
straightforward:
1) Consistent Severity Labels
Severity labels are the key to many useful sorting and reporting functions in EMONITOR Odyssey. Using
consistent severity labels will make this much easier. Typically, two alarm levels are used in most cases.
Set up and work with the two severity labels that are most appropriate for you (i.e. WARNING and
ALERT). Plot options and tag conditions can be stored that will use these severities automatically, saving
you time and effort in getting results.
2) Use Category and Band Variables for Quick Editing
Setting alarms based on category and band variables allows you to quickly and easily update the alarm
settings for all the machines in the category by editing just the appropriate fields in the Category Window.
This was our goal in designing the EMONITOR Odyssey category approach - to make it fast and easy to set
alarm limits for an entire group of machines. This can save you hours when compared to the effort required
to edit individual alarm setups for each measurement.
3) Basic Alarms to Get Started
As you first start your Predictive Maintenance program you will be learning a great deal about the behavior
of the plant machinery. Start out with a basic approach to alarms that emphasizes reasonable magnitude
alarm limits and a simple band or spectrum alarm setup. Magnitude alarms may be based on published
information, comparisons to baseline conditions or percentage change. Band and spectrum alarms can also
be based on comparison to baseline conditions or percentage change.
13
Enteract Home White Beginning
4) Implement Statistical Alarms as Your Program Matures
After several months of operating a Predictive Maintenance program, a more complete understanding of
your machinery behavior will be developed. Many machine problems will have been solved and
maintenance practices improved. The machine vibration will typically become more consistent at this time
and a large database of stored measurements will be available. Under these conditions, statistical alarms
become an extremely powerful detection tool. The statistical indicator alarms may be very useful because
these can be configured to use only the most recent readings from each measurement. This is one of the
best methods to pick up changes in the machine vibration at the earliest possible stage.
5) Check Alarm Results before Full Scale Implementation
When setting up new alarms, it is easy to add a single alarm line to a measurement and immediately view
the results. This alarm can then be copied and pasted to other measurements in the database. After a little
fine tuning on these additional measurements, a large group of locations may be tagged and the new alarm
copied to the entire list using the Paste Into List function in the Edit menu.
6) Use Multiple Categories When Needed
In many cases, a single category may not be sufficient for use on a complete machine. This is particularly
true when the nature of the mechanical components changes significantly from location to location or when
the orientation of the measurement has an effect on the vibration. In these cases, don’t be afraid to set up
and use a different category on the same machine. The category is entered for each location, so if
necessary, each location can be assigned to a different category.
14
Enteract Home White Beginning
