Cummings diesel generators
Monitoring the generators.
The generators can be queried using the modbus protocol.
The program that does this:
- Runs on galfas2. This as a connection to the generator
- source: svn/aosoft/src/Cummings/cgenlog.c .. see also cgen.c,
Test/readMultiReg.c for debugging
Recent generator usage:
The plots show the generator usage for:
- the plots show the sum off all generators.
- top frame: Power output vs time.
- 2nd frame: Fuel consumption in gallons/hour
- 3rd frame: Fuel rate vs output power.
- red line is a linear fit of gal/hour per MegaWatt output.
- bottom frame: Efficiency vs power output.
- KWatts output for 1gal/hour consumption
- Top frame: Gallons used vs time. (for the time period
- Bottom frame: Cumulative hours vs time for each generator
Fuel deliveries and Fuel usage:
when fuel added
when fuel added
- The measurements using the stick and the generator fuel usage
meter do not agree.
- The 4th column shows the stick measurement before and after
the fuel was added
- Column 5 has the cumulative fuel usage from the generator
meters (when the fuel was added).
- Column 6 has the fuel used between fuel deliveries, using the
stick and the generator meters.
- The 7th column is the ratio of genMeterReading/stickReadings.
- The final column shows the daily consumption for the interval
between fuel deliveries.
- the generator reads 20 to 35% more than the stick.
01dec13: Generator startup with site as
The power failed on Sunday 01dec13 at 17:39:20.
The site was the generator load.
The plots show the generators
for the first 10 minutes after startup (.ps) (.pdf):
- Top: generator state vs minutes fro startup.
- Each color is a different generator
- the y axis is the generator status (with labels on the
- All 4 generators start together, and then generators 4,3,
and then 2 are idled.
- 2nd: generator output power
- each generator is plotted separately
- The light blue line is the sum of the 4 generators.
- The site was running at 300 KW prior to the power outage
(from the site power monitoring)
- It took about 4 minutes for the generator output to reach
- The actual outage occurred at 17:39:20.
- The telescope motion was stopped from 17:39:20 to
17:41:20.. so there is only 1 minute with no telescope
- I wonder if the air conditioners take a few minutes to get
up to full load?
- 3rd: fuel usage by generator
- the is gallons/hour Usage plotted every minute.
- bottom: Efficiency: (Kilo watt hours output) per (gallon/hour
- On automatic startup (with site as load) all 4 generators
start, and then get shed 1 at a time
- The shedding process takes 9 minutes
- The generator output took 4 minutes to return to the output
prior to the outage.
- The telescope was only stopped for 1 minute
- It make take the air conditioners 4 minutes to come up to
19Nov13: first look at data while sband
transmitter is on. (top)
Data was taken on 19Nov13 while the sband
transmitter was in operation. They sband was transmitting
around 850 KWatts.
The plots show the
generator performance for 3 hours (.ps) (.pdf)
- Generators 2,3,4 were in use.
- Data was sampled every 10 seconds.
- Data in stored in binary
- Page 1: Power, power factor
- top: power vs time. The power is by generator. Each color is
a different generator.
- middle: Power factor vs Power(by generator).
- I computed the power factor as KW/KVA.
- The generators don't compute the power factor until the
power exceeds about 40KW.
- bottom: power factor (by generator) vs power
- the power factor is above .9 above 200Kw/generator.
- Page 2: Voltage (Phase to ground)
- Each frame is a separate phase (a,b,c)
- I'm not sure it the voltage spikes are real.
- Page 3: Current vs time
- Page 4: Pressures, temperatures.
- Top: pressure vs time for each generator
- Middle: Oil temperature vs time.
- You can see the pressure decrease as the oil temperature
- Bottom: Coolant temperature vs time
- Page 5: fuel consumption and
efficiency (by generator)
- Top: fuel consumption (Gallons/Hour) vs time by
- The fuel used is recorded with .01 gallon resolution. The
lower rates are pushing this resolution:
- 10gallon/hour = .028 gallons/10 seconds. So we saw maybe 3
counts change/ 10 sec sample.
- Middle: Fuel consumption vs output power (by generator)
- Bottom: Efficiency vs power (by generator).
- The vertical scale is : KiloWatts generated by 1
gallon/hour fuel consumption
- It looks pretty level down to about 200 KW/generator (then
we run out of data).
- Page 6:Cumulative
Power,consumption, efficiency (combining all 3 generators).
- Top: power vs time.
- Middle: fuel rate (gallons/hour) vs output power.
- bottom: efficiency (KW for 1 gallon/minute) vs output power.
- Page 7: Power, consumption
blowup at startup and full power (sum all generators)
- Top: power at startup
- red line is cooling pumps off, green is cooling pumps on.
- 2nd: fuel rate at startup
- cooling off: 26 gallons/hour,
- This looks like the no load fuel rate needed.
- cooling on, 31 gallons/hour
- 3rd: power at full power
- bottom: fuel rate at full power:
Performance for the 3 generators
|startup (cooling off)
|startup (cooling on)
|Pumps use 70 KW
|above 750 KW
above 700 KW
|startup cooling off. Fuel
rate for no load.
|startup cooling on
|KW for 1 Gal/Hour
|1 Gallon of diesel = 40.7
|pumps on, no HV xmter
|pumps off, no load.
171202: Gen 4 failure. (top)
Generator 4 had been offline while its exciter
windings were being rewound.
On 02dec17 generator 4 was tested after reinstalling the exciter.
After being brought up, the generator failed with a short circuit
error (extended status).
The plots show gen 4 failure (.ps)
- The generator data is sampled once a minute (so it can miss
- Top Frame: generator state:
- black: gen1, red:gen2, green:gen3,blue:gen4
- State values:
- 0- stopped
- 1- start pending
- 2-warmup at idle
- 3- running
- Gen2 was running the site.
- Gen4 warmup started around a little after 8am. around 8.2
(08:12) it was running
- it ran for 4 minutes and then stopped.
- 2nd Frame: Gen 4 phase to ground voltages.
- When gen 4 went to run mode, it had 2400 volts (phase to
gound). This is the standard value for the generators,.
- After 4 minutes, the voltage dropped to around 100 volts.
- On the 5th minute it was stopped.
- Bottom Frame: Gen 4 amps
- for the first 3 minutes i run mode, no current was drawn
from the generator
- The current jumped to 400 to 500 amps at the 4th minute
(when the voltage dropped to 100 volts).
- My guess is this is when the generator was connected to the
- The dashed line (5 minutes after going online), the
generator extended status faults were:
- Short circuit
- emergency status.
- Generator 4 is now offline waiting with generator 1 for its
primary windings to be repaired.
171205: sbTx pwr vs generator output pwr
for 2017 (top)
The sband transmitted power was plotted vs the
generator output power for the year 2017.
Making the plot:
The plots show the sbTx power vs
generator power for 2017 (.ps) (.pdf):
- Input the sband transmitted power for 2017.
- This was taken from the lro power meter log file.
- This is typically sampled once every 10 seconds.
- Input the generator data for 2017.
- Normally sampled once a minute when the generators are on.
- Interpolate the generator time to the sample time of the tx
- I didn't worry about edge effects on the interpolation
(since generator should be on if sbTx transmitting).
- Top frame: sband Tx power vs generator output power.
- To remove bad data points i used a data selection criteria:
- sband Tx power > 1 KW
- The efficiency of the sbTxTubes must be less than 50%:
- sbTxPower < (generatorPower - cooling power..70KW)
- Bottom frame:
- Ratio of txPower to (generator - cooling power)
- All points were included.. no data selection used.
- The plot does not differentiate between the number of
- 100 KW output power levels gave efficiencies around 10%.
- 800 KW output power levels reach 40% efficiencies.
- 400 KW output power levels.
- There is a second trace with higher efficiences.
- This is probably the 1 klystron mode that we were running.
- It Probably also uses fewer generators.
- The sbTx power level will also depend on how well the
transmitter is tuned (hv and rf drive).
- > .6 efficiency. these are probably bad points...
- When computing the efficiency i subtracted the cooling
power from the generated power (70KW).
- This gives a closer idea of how the tubes perform.
171207: generator power vs sbTx beam
AO lost public power after hurricane maria
(20sep17). The outage continued into dec17. The sband group
would like to observe an object in mid december. The question is
how many generators will be needed.
The requirements/constraints for the
- generators 1 and 4 are not working and will not be
available for the run.
- generators 2 and 3 are available.
- I've heard people say that each generator can probably output
800KW each (doubt if this has actually been tried:).
- The site has been running about 300KW at night (building 2 ac
off) and around 400 KW during the day,.
- The power consumption of the sbTx is determined by the beam
voltage (the filament current is fixed).
- The sbTx beam voltage comes on at 36KV. It can then be raised
gradually from that point.
I searched through recent (1klystron
usage) sband tx logfiles for when the beam voltage was
first turned on. I wanted to plot beamVoltage versus the generator
power output. This data had to overlap in time:
- The generator power is only sampled once a minute.
- The sband log files are sampled much faster (especially when
things are changed).
- when the operators bring up the sband Tx, they start at 36KV
and then bring the beam voltage up pretty quickly.
The best data set (beam voltage was not
brought up really fast) was on 01sep17. This was a single klystron
run looking at clementine.
The plots show the generator
power and sbTx beam voltage vs time (.ps) (.pdf):
- The black trace is the generator power. The left hand
axis shows the generator power.
- The red trace is the sbTx beam voltage. The right axis shows
the voltage levels.
- The dotted red horizontal lines are spaced every 5 KV in beam
- the * show were the data were sampled
- The beam voltage started to come up at 20.26 hours.
- there was one generator sample while the beam voltage was
being brought up (at 20.28 hours)
- The green lines flags the common generator and beam voltage
- The generator was outputting 550KW with the beam voltage at
- This is not an ideal measurement. You'd much rather have the
beam voltage stable for awhile in case there are any time
offsets in the measurements (the generator is tied to ntp.. as
well as the sbtx).
- in 1 klystron mode, 45KV beam voltage took 550KW of generator
power (using 3 generators).
- running with fewer generators would probably need less
generator output power.
- Using 2 generators, it looks safe to bring the sbTx beam
voltage up to 36KV (with the site using 400KW)
- You could then monitor the generator power while you slowly
brought up the beam voltage.
- Assuming each generator can output 800KW:
- If the site uses 400 KW, then with 2 generators there is
1200 KW of headroom for the sbtx usage.
- being a little more conservative, you might want to let
the sbtx use 1000KW
- The beam voltage should probably kept to below 50 or 55KV.
- if you cranked up the rf drive power, you'd probably get
lots more that the 60KW of transmitted power
people had been mentioning.
- In all of this i'm assuming:
- that the klystron power consumption does not depend on
the rf output power.
- that the cooling power consumption does not increase
appreciably with the power output (most of the power is in the
water pump and not the cooling fans?).
Yearly usage (top)
Hour used and fuel consumed .. by generator