If no text is printed in Netscape, select File | Page Setup | Black Text.
If you print this page, all algorithm related material will not be printed!
Installation - Patented Simplicity - Patent #5,831,174
The only data
required by VOLUCALC 4
is the time that known levels are reached, which is extracted from the
pump operation using our patented "Pump Status Converter
". There is really only two parameters that greatly influence
the accuracy of volumetric flow calculations:
A known and constant
volume where the pumps operate, and The
algorithm used to calculate flow
When the influent pipe is
between the operating levels, it is not possible to use a constant volume.
VOLUCALC
4
is switching between different type of algorithms including:
 |
The derived flow algorithm, |
|
The extrapolation algorithm, |
|
The average algorithm, |
|
The trending algorithm, and |
|
The VOLUCALC process which include all of the above plus statistical analysis, normal curve deviation, iteration, variable tolerance comparison, and few trade secrets. |
The only data required by VOLUCALC 4 is the time at which known levels are reached, which is extracted from the pump operation using our patent #5,597,960.Therefore, the above graphs show flow curves which give the same average flow.
The
flow trend is not known, but must be assumed to calculate pump capacity.
Even
if the average flow is 100% accurate, it doesn't tell you the trend of
the flow. All these flow trends (red in graph) give the same average flow.
The truth is that it is statistically impossible to get a constant flow.
So it is wrong to assume that the flow before or after the pump operation
will be the same while the pump is running. This graph shows flow trends
which give the same average flow. Seeing this, can you tell what is the
flow after the average flow was calculated ?
The
average head pressure which influences the pump capacity is not known.When
pumps are operating, the average liquid level influences their performance.
The average liquid level for each pump run cycle is not known, therefore,
the accuracy related to the pump performance of one cycle is not as high
as the average pump performance of many cycles.
An average
level difference of 36 inches between cycles can influence a flat curve
low RPM pump by as much as 30%. Knowing this, let's
see why a volumetric method can be highly accurate ... or not.
After
you entered the volume of liquid between the pumps operating levels of the station,
VOLUCALC 4
needs "event" data. This is the time at which identified operations occur in
the station. For flow calculation, these events are pump starts and stops.
When
the pumps are not operating, the flow through the pump station is simply
the volume of liquid being filled divided by the time it took. The real
question is what is that flow when the pumps are operating?
When pumps are operating, liquid is pumped out of the well at an unknown rate while liquid is still getting in at an unknown rate. To calculate the pump performance (called OUTFLOW in this document), you need to know the inflow while the pumps operate. This is where everything gets complicated. The average pump performance for one pump operation cycle is equal to the VOLUME between the operating levels divided by the TIME it takes to get from one level to the other plus the INFLOW for that period of time (OUTFLOW=VOLUME/TIME+INFLOW). The inflow is not known, but must be used to calculate the pump's outflow. The idea is to figure out the most probable inflow. We know that volume and time are 100% accurate information. This means that the only reason why a calculated pump capacity is not accurate is because the inflow is not known. The the big question is:
HOW to calculate the INFLOW when the pumps run ?
The
EXTRAPOLATION method
The
AVERAGE Method
The
TREND method used by VOLUCALC 4
c
The trend
method assumes the inflow while the pumps operate is affected by the trend
of the inflow calculated for more than one cycle before and after the pump
run time. If the inflow calculated over 4 cycles goes like this: 175 GPM,
250 GPM, 200 GPM and 150 GPM like it is illustrated. Does it make sense
to think that the flow in the station was on the top of a peak when the
pump ran ? Maybe closer to 300 GPM which would give a pump capacity close
to 800 GPM. If this represents better the most probable inflow, then the
other methods had a possible error of up to 25% on their estimated inflow.
This is a good start, but it is not enough for VOLUCALC
4.
We want to be sure, 99% sure.
If it is so much better, then why almost nobody is using it? After all, it has been available since we created it back in 1986! This method uses an equation that is a little more difficult to master that the two previous ones, and that explains why it is rarely used. For each of the four cycles required to calculate an inflow trend, you will have to do the following equation:
Now
that you know how the most probable inflow is calculated by VOLUCALC
4,
let's see how it is used within the VOLUCALC
4
process.
There are few functions
that are performed by VOLUCALC 4
that are not documented here or in our patents. It took us 17 years to master
this technology and we will not give these trade secrets easily, but the following
information should be enough for you to believe that VOLUCALC
4 is the most accurate
volumetric method ever developed for pump stations using constant speed pumps.
All operations related to one pump could be done on multiple pumps because combination
of pumps is viewed by VOLUCALC 4
as another pump having its own characteristics.
To verify the accuracy of the VOLUCALC method, we created a virtual pump station program. This is a computer generated lift pump station having a computer generated inflow that creates a file having pump starts and stops based on a fixed volume and a user adjustable rate of change of inflow (a version of it is available in the DOWNLOAD page of our Web Site). The virtual station generates the pump operations and records them in a file that can be read by VOLUCALC 4, then we compare to the flow calculated by VOLUCALC 4 and the original flow. Only the pump operations and the volumes were loaded into VOLUCALC 4. This is the best way to evaluate the accuracy of the VOLUCALC 4 method, because it is only a sophisticated algorithm after all. But not everyone likes the idea of comparing numbers to numbers, so we did it against a magmeter. We thought that being within 2% of the magmeter was close enough to make the magmeter manufacturer ( I mean us) happy. So, here is what it takes to calculate flow the right way.
The user enter the volume
of liquid between the levels at which the pumps operate.Pump operations and
time of occurrence are recorded and/or transferred to the patented "pump
status converter".
The pump status converter (next drawing) generates levels out of the pump operation. VOLUCALC 4 must know which level is reached every time a different pump operation is recorded.
The pump status converter then transfers the time, pump identification, level reached and the volume (from user) to the Flow Calculator. The Flow Calculator uses the formula described previously to find the most probable flow in and out of the station for the middle portion of the four cycles used to calculate the trend of the inflow. The Flow Calculator then transfers the calculated outflow, inflow, pump identification and time to the Flow Rectifier.
The Flow Rectifier compares the outflow calculated by the Flow calculator to the history of the pump, and also updates it. Then, based on the difference between the last calculated outflow and the history of the pump, the Flow Rectifier adjusts the Variable Tolerance for that pump to maximize the accuracy related to that pump. If the calculated inflow is near the actual inflow and if the average level is "average", then the outflow calculated will be near actual outflow.
In normal operating conditions, the outflow of a constant speed pump changes slowly. If you compare the last calculated outflow to the average of many outflows calculated over many cycles for the same pump, then it should be close. If the calculated outflow isn't near the average outflow (Outside Tolerance), then maybe the interpolation of the inflow wasn't near the actual inflow (most probable), or the pump(s) may have a problem, or maybe the volume had changed. Because of the different possibilities, no final inflow or outflow calculations are done at this point. Instead, the event is tagged as Potentially Abnormal. At the same time, a counter is created (on the first abnormal event) or increased to reflect the number of abnormal events of the same type recorded for the same pumps in a row. A different counter is used for above and below acceptable calculated outflow, but only one counter is used at a time. The counters are used to determine if a condition is permanent. Obstructed or broken pumps are permanent conditions.
An inflow error isn't permanent, unless the volume had changed. An abnormal condition is considered permanent when a counter reaches a predetermined number. In this case, the algorithm returns to the first tagged abnormal event and uses the inflow calculated to update the pump outflow of all tagged events related to that pump. The counter is reset after.If the abnormal condition returns to normal, the counter is reset and the algorithm returns to the first tagged abnormal event and uses the average outflow of the pumps to calculate all the inflows of the tagged events related to that pump. The counter is then reset. If the volume had changed, then the percentage of difference would be seen for all pumps starting at the same time. Readjusting the volume in the software and recalculating would correct this problem.
Every pump station is different, therefore the tolerance at which a calculated outflow is normal or tagged as abnormal must be self adjustable. This means that if the outflow is calculated for a station having a very stable inflow, and that all the outflows calculated would always be within 2% of each other, than the tolerance for this station should be low because a 3% deviation of the calculated outflow would be abnormal. But if we have a pump station receiving water directly from a few other pump stations which create an extreme change in the inflow, then the tolerance should be higher so that enough outflows are used to generate a good pump average.
This explains enough to see why VOLUCALC 4 is so accurate. Curiously, the only flow meter that is as accurate as VOLUCALC 4 is a magmeter. Curiously, the easiest and cheapest to install flow meter compared to the most expensive and difficult to install flow meter.