VFDs and Pumps

(Duplicating my post on the earlier companion thread)

This is obviously a large pump station. If these pumps are in parallel groups, rather than individual, it would seem that flow could be controlled by how many pumps are running, rather than by altering the speed of individual units. To say nothing of the cost of (24) large medium/high-voltage VFDs.

On the other hand, VFDs can provide smoother starting/stopping of the pumps, as well as adjustment of flow/pressure. Has the consultant given you the rationale for eliminating the VFDs, versus the alternate rationale for keeping them? What does the grid authority say about starting motors of this size?

There are lots of considerations, but I personally like VFD for pump control. However it is expensive, and if control can be accomplished efficiently and easily in other ways then don't do it.

Would need a lot more information on the application I think.

One advice I read (can't remember which magazine it was) was to have only one pump running on VFD. The rest is DOL and they kick in depending on demand. The lone pump on VFD controls the flow.

For example: If your pumps, say, have 10gpm each and your required flow is 45gpm, run 4 DOL pumps and the VFD pump takes care of the remaining 5gpm.

regards,

Vulcan

Oops! I made an assumption. Are these pumps part of a cooling/chilled water circulation system? I've never heard of a 24 pump system but somehow assumed it was. (",)

I've always read that VFDs will save you electricity but the pumps need to be running between 50 and 100% of full speed. Any less than that and you get losses from the pump and the drive.

regards,

Vulcan

I think this is a main water pumping station, with an 800 mm inlet each pump is moving a bunch of water....

All depends on system and pump curves. Remember that head varies as the square of the speed, so if the VFD pump is identical to the fixed speed pumps it will soon get backed out as the head drops with a reduction in speed and with an increase in speed you would soon back out the fixed speed pumps.

As Steve says, it is a matter of economics. VFD's are expensive so they must save a reasonable amount of energy to get a decent payback

For each group of pumps, there should be ONE VFD. When a fraction of a pump will be needed, the VFDed pump will supply that fraction. When the time comes to bring another DOL pump on line, the VFDed one will be pulled back to a low fraction.

For example: in a four-pump group, if 2.5 pumps are needed, then two DOLs will be on and the VFDed pump will operate at 50%. As load increases, the VFDed pumps speeds up until it comes near the point where three pumps are needed. At that time, the third DOL pump will come on and the VFDed one will be pulled back to a low percentage, to bring the whole near three pumps of capacity. As more pump capacity is needed, the VFDed one will speed up, until in effect four pumps will be on line.

It works in reverse when needed pump capacity drops ... At the whole-number (1,2,3) transition points, a pump should shut off and the VFDed pump speed up (e.g., when passing from 2.1 to 1.9, a pumps drops and the VFDed one speeds up from 0.1 to 0.9 speed).

Add some hysterisis (deadband) in the control program to prevent pumps from being turned on and off over and over again atthe transition points. For example, a pump should turn on at say 2.1-and-rising capacity and turn off at 1.9-and-dropping capacity instead of both happening at 2.0 pumps. Otherwise, needed capacity rising and falling around 2.0 pumps will turn a DOL pump on and off, and will speed up and slow down the VFDed one over and over.

One last thing: the VFD should be connected to TWO pumps through a selector switch. That way, when the principal VFDed pump is down for maintenance, a switch can be flipped to send the VFD control signals to a second pump within the group, the 'backup VFDed' pump.

So, in summary: ONE VFD per pump group; one selector switch to direct the VFD to one of two pumps in a given group; on-off logic in the PLC/controller; and hysterisis (deadband) programmed into the controller to prevent pumps from starting and stopping repeatedly and VFDed pumps from ramping up and down over and over again.

Good luck! DZ

I agree and this thread illustrates a problem I have with with these forums. The OPs rarely provide enough information to allow an intelligent answer; i.e., one free of ASS-U-ME [TIONS]. {Please excuse the clumsy attempt at grammatical irony.}

Often it appears that the OPs post a question just to get support for their position so that they can go back to whomever they are in dispute with and say, 'see - these experts agree with me.'

My other problem is with questions that are easily answered by an online search, reference to standard texts or standards publications.

I am new to this forum and not fully aware of what can be done about these problems or where to take the complaint. I guess the best course is for the responders to merely tell the OP to restate the question until the OP gets the message.

Atsysusa, Many thanks for your comments, it seems that not everyone would agree with you or me. I do wonder if our "unknown friend" is on a mission? or maybe a "Rebel without a clue"

I believe that forums like this are to be used to give each person the opportunity to grasp a general idea on whatever or get some direction on how to find the answer to their particular problem.

However as this topic is the SECOND hearing in as many days and the OP has STILL come back complaining he disagrees with the consultant he/they have employed, you gotta wonder are we being used?

Like you said to MANY ass-u-me tions.

In a bank of several identical centrifugal pumps, if one pump runs even a bit slower than the others, it is essentially a "dead duck," unless you can control its speed within a very narrow band just slightly less than the other pumps.

A single centrifugal is highly likely to benefit from VFD control, but parallel operation gets tricky. If one pump is under VFD control, the rest should also be under VFD, all of them the same (or very close.)

The situation would be different with positive displacement pumps; in that case a half-speed pump would still produce full pressure.

On this basis, maybe I would keep the consultant and dismiss the dissenters.

How much does an 11 kv 2850 kw VFD cost, anyway? If the consultant can blow $1 million of dubious confetti out of this installation, more power to him/her!

(No, that is not necessarily my final opinion, but I don't mind a little verve. The OP hasn't responded on the grid authority's input as to starting large motors.)

For an installation of this magnitude, I am surprised that the rationales are as yet inadequately presented. If all these pumps discharge into a common header, it would look like about 84 inches or so (7 feet) {[24 (24 exp 2.5)] exp 0.4 ≈ 85.6}. That's just a thumbnail to estimate the overall size of this substantial project.

just one more point

be it VSD or DOL you will need a flow control valve to keep the pump in the curve/operating range, do not think that the VSD will replace the control valve as, and I refer you to my reply of yesterday, due to the "Laws of Affinity" the pump curve AND operating range changes each time the frequency is changed & to get the maximum life out of the pump you must operate within the OR.

Tornado, I do have a course on pumps, and I charge $200 per hour. Let me know if you're interested :)

It all depends upon what is being supplied by the pumps and the size of the pumps and how much of a pressure surge can be tolerated. If pressure surges are acceptable, there is no need for a VFD. How fast does the system have to react to changes in demand? Can the flow control valve keep up? Seems to me that a flow control valve wastes pump capacity and probably energy.

Right now it seems as if there is a disagreement of opinions. What you need is an analysis which shows that any proposed change will meet the technical requirements. You also need an analysis which compares operating costs and maintenance costs for viable options. The cost of making a change (installation, down time, training, etc.) will also have to be amortized.

When I worked for a VFD company we provided VFDs for municipal sewage lifts and municipal fresh water systems. Back in the mid 1970's VFDs were big and expensive. A typical sewage lift would have one pump on VFD and the others were simply line started. The VFD was used to help maintain a constant flow and not a constant pressure. The VFD prevented the need to continually cycle a pump on and off to maintain a constant level in the reservoir.

I wonder, does this consultant sell flow control valves?

Fresh water systems required constant pressure. A single VFD was used to ramp up a pump and maintain pressure. When the pressure dropped because more water was being used, the pump was transferred to run off the line and the VFD was used to pick up an additional pump. When the demand dropped the VFD was used to ramp down each pump. Control logic was relays and analog. Needless to say, tuning the system was quite time consuming.

Because VFDs have dropped significantly in cost and size, and the cost of building a facility and switch gear has risen, I am sure that a one VFD per pump installation would prove to be less expensive to install today.

Maybe I missed the description of the system, we really need to know what these pumps are connected too. What does the system curve look like?

For example, if your system is pumping from, let's say a river up to some HX'rs, maybe water cooled condensers in a chiller plant and then returning back to the river and it is open the the atmosphere. Let's say there's a big tank that you keep filled from the river that feeds the chillers to maintain a constant head on the condensers, then your system curve will be very flat. At any flow you will need a minimum head to even get the water up to the tank.

In that case of a constant static head, the vfd's on multiple pumps will have to ramp up together otherwise the slower pump will essentially be "dead headed".

Depending on the situation, system curve, vfd's might not provide the benefit initially expected, a careful engineering review is required to know for sure.