Print this article

Comment

01 July 2004

Speed manages streams

The case for variable frequency drives as a final control element spreads.

By Ian Gibson and Geoff Irvine

The concept of using variable frequency drives to control electric-driven centrifugal pumps is distinguished. But they seldom work in the petroleum industry. Why?

The reasons given range from misinformed economic data and an unease of using unfamiliar electrical technology to an ignorance of how these devices react as control devices. One still sees the majority of flow, pressure, and temperature control happening through constant-speed centrifugal machines that a control valve throttles to provide the desired control of fluids.

The use of variable frequency (VF) drives—adjustable frequency drives or variable speed (VS) drives—on electric motors driving centrifugal pumps offers many advantages over using control valves for controlling the fluid flow. Some of the advantages are energy savings, efficiency, power factor, installation, specification, ability to control, maintenance, fugitive emissions, and reduced wear on the bearings and seals of the pumps.

Although this method of control has gained wide acceptance in general industry, there is still reluctance to use it in the petroleum industry. The reasons for this vary, but the most common reasons are:

• a lack of understanding about how these devices work in comparison to the more commonly used control valve
• a common misconception that these devices are expensive compared to control valves
• a reluctance to use these devices on explosionproof motors found in petroleum facilities
• a fear of the reliability of these fundamentally electronic devices
• a lack of knowledge about the failure modes of these devices for the hazard and operability analysis

There is no doubt that in the early days of development of the variable frequency drive some of the issues were true. However, over time the cost, reliability, functionality, and knowledge of these devices has improved to a point where we believe they should be viable final control elements for petroleum facilities.

Flow by discharge throttling

Discharge throttling is widely used in the petroleum industry. It involves putting a control valve on the discharge side.

With this method of control the control valve varies the pressure drop across the valve, and hence the flow. It has the effect of increasing the slope of the system curve, and therefore takes the operating point to a different point on the pump curve.

The insertion loss of a wide-open control valve is about 10% of the other dynamic losses. This requires the pump H-Q curve to be higher at the design point to compensate for this loss.

Most control valve design guides call for the valve to be no more than 70% to 80% open at the design point.

However, when speed control operates to control the pump flow the effect on the pump curve is that it "moves" up and down the system curve.

The speed control method of flow control ensures that just enough energy goes to the pump to get the desired flow rate. Compared to controlling via discharge throttling, this method of control is much more energy efficient.

There is an analogy to driving a car. Discharge throttling is similar to having the car engine going at constant revolutions per minute (rpm) (a brick on the accelerator) and using the brake to control the speed. Speed control of the pump is similar to using the accelerator—the accelerator is depressed enough to supply the desired speed.

Also, some points that are germane to this type of control are:

1. The effectiveness of this type of control is very much dependent on the shape of the pump curve and the system curve. The diagrams shown here are "idealistic" for the purpose of demonstrating a point. One needs to be aware of the effects of curves shaped differently than the ones shown here, and how to best implement a VF drive for these cases.
2. There are pump designs (axial flow, mixed flow) that do not approximate a parabolic relationship, and some of these cannot use throttling—recycle is imperative. These are common in deep-well turbine pump designs. Speed control of such pumps is often ineffectual.
3. There can be a loss of efficiency in the pump operation that is similar to the loss of efficiency when using discharge throttling on the pump.

Energy saving VF drives

Thus, the main incentive for using variable speed drives is the energy savings from their implementation. This is due to the pump affinity laws governing a centrifugal pump (or fan) operation.

The operation of a variable frequency drive uses a number of simple units.

The first part of the VF drive is a rectifier to convert the alternating current (AC) voltage to a pulsating direct current (DC) voltage. Then the intermediate DC circuit filters the pulsating DC voltage to a DC current or voltage. The inverter uses the DC current or voltage from the intermediate DC circuit to produce an AC current or voltage having the desired frequency. The control unit oversees the operation of the components of the VF drive.

The most common method of controlling the frequency is pulse width modulation. With this method the intermediate circuit produces a constant DC voltage, and the inverter then produces a synthesized sine wave of the required voltage and frequency to control the speed and direction of the AC motor. The synthesized sine wave consists of switching pulses of varying width, producing an equivalent root-means-square sine wave the motor recognizes. The reason one needs to control frequency and voltage at the same time requires some explanation.

In a squirrel cage induction motor, the stator windings act as resistance and inductance in series. The impedance of the stator windings is dependent on the resistance of the stator, the frequency of the applied voltage, the inductance, and the inductive reactance.

From Ohms Law, I = E/R, and because for any given motor the resistance and inductance are fixed, it is possible to simplify the motor operation when operating with variable frequency to I = E/ƒ.

Variable frequency drives work by holding the "volts/hertz" ratio constant to ensure minimum overheating effects at lower frequencies than the base speed, and to ensure maximum torque comes to bear when the supplied frequency is above base speed.

As the VF drive has evolved, there has been more sophistication put into the basic "volts/hertz" control. This is because users want to get very precise control, similar to that offered by DC drives, but now out of an AC VF drive and motor. The generic term for this control is vector technology; it now means it is possible to independently control speed and torque very precisely. All this is probably overkill for most centrifugal pump applications, although it deserves a more thorough investigation for positive displacement dosing applications.

In summary, the technology of VF drives has reached a mature phase where it is possible to apply this technology to centrifugal pumps as small as 0.37 kilowatts (single phase) right through to medium voltage 6.6 kilovolts 10 megawatts.

In general, the low voltage drives (380–480 VAC) apply to motors from 0.37 kilowatts to 448 kilowatts, while medium voltage drives (2.2 kilovolts to 6.6 kilovolts) apply to motors from 100 kilowatts to 10 megawatts. VF drives up to 20 megawatts exist at 2.35 kilovolts for ring motors (0–25 rpm), though not for pump applications.

Gradually ramp acceleration

In addition to the energy savings and increased control accuracy when using VF drives, there are also a number of operational benefits that sometimes go unnoticed.

Reduced water hammer effects: Water hammer is a problem that results from rapid changes in liquid flow. These flow changes set up rapid pressure transients that damage pipes and cause pipe supports to move. Downstream devices such as valves may suffer damage too.

Direct on line (DOL) starting a pump may cause these water hammer effects, as this method makes no attempt to soft start the pump. Reduced voltage soft starters are available to get over these types of problems. If, however, a VF drive is one of the choices for controlling the pump, then this may be an additional advantage. The VF drive allows the user to gradually ramp the acceleration at the desired rate, while still developing full torque across the speed range, and also limiting the starting current to 100% of full load current or less. In some applications that use large pumps at locations with weak utility supplies, this has been the only available method to start the pump.

Reduced problems from cavitation: Cavitation is a phenomenon that occurs whenever the static pressure drops below the liquid vapor pressure. (This is due to Bernoulli's theorem, where fluid flows through a restriction, and the velocity pressure rises and the static pressure drops.) The results of cavitation are vapor bubbles forming and then collapsing downstream of the low static pressure point. The collapsing vapor bubbles have a very high and concentrated impact force, which over time produces a surface effect similar in appearance to sand blasting.

The point is that if one has an operational situation where the pump may get into a high or very high suction energy point, then lowering the speed will reduce the suction energy by a squared relationship.

Reduced wear on seals and bearings: Reducing the speed of a pump reduces the wear on the seals and bearings of the pump. This is verifiable by examining the reliability index (Ri) of a pump manufacturer.

That is to say the case of varying speed should increase reliability by a factor of 3.75 over the case of throttling the flow. One should consult the particular pump manufacturer's data to get the best data available.

Digital communications: Where digital communications are communicating speed reference signals and configuration data to the drive, it brings benefits to the overall operation of the plant.

Not only does this mean more accurate speed signals to the drive, but also the ability to monitor real-time motor data such as speed, power, current, volts, last fault, thermal capacity of the overload, thermistor inputs, and hours run. This allows the plant operators to closely monitor and troubleshoot, and the data can enter software packages for preventative maintenance.

Reduced fugitive emissions: Eliminating the control valve and the associated piping can represent a significant reduction in fugitive emissions released by leaking seals on the control valve and flanges. Using SOCME data for "light liquid" gives typical fugitive emission data of 430 kilograms per annum.

Reduced power factor charges: Where the petroleum facility is using utility power, a VF drive will reduce the additional penalties that arise from a low power factor on the overall plant. This is due to the inherent structure of the VF drive with the constant DC bus.

Drives in closed loop control

The technique of using VF drives in closed loop control with centrifugal pumps lacks much documentation. One can find many books and articles written about how to tune flow and pressure loops using control valves, but very few on how to tune these same types of loops when using a VF drive as the final control element. Shinskey has written a paper on some tests carried out on a small centrifugal pump with no static head. In this case, the VF drive certainly showed it had superior rangeability, superior performance when following a ramping signal, and good performance in following set point changes. This paper came out more than eighteen years ago, and obviously it would be desirable to update these tests with the latest technology in both VF drives and control valves.

One big advantage in considering a VF drive in comparison to a control valve is the elimination of variability that a control valve introduces into loop performance. This variability results from the dead band inherent in the control valve frictional components such as stems and packing. This means a signal is output to the valve, but it does not move due to friction. This dead band may be 2% in new valves and increases over the life of the valve. Digital positioners lessen the effect of dead band. However, if it is possible to use speed control on the pump to achieve the same control, one will not have the problem of dead band to start with.

There are a couple of points of care required in applying VS drives to pumps in closed loop control. One is the speed of response. The rate at which a pump can change speed as commanded by the control system is governed by the inertia of the pump/transmission/motor combination, the inertia of the liquid column, and the available excess torque (above that required to maintain current speed under load). This governs the acceleration rate. Slowing down can be interesting. To reduce the flow rate one must decelerate the entire liquid column. Attempting to reduce the frequency rapidly can cause the inertia load to raise the DC-bus voltage to embarrassing levels, unless provisions to dissipate the energy happen, or the deceleration rate has been set to match the hydraulic behavior of the pipeline. These problems are not unique to VS drives, but a valve is somewhat more positive in its behavior, as it dissipates energy by design.

Caution is important in tuning the control loop. A noisy output signal from the distributed control system (DCS) to the VS drive is not acceptable. A control loop may appear to be behaving satisfactorily from the DCS screen (which updates slowly), yet the pump may be accelerating or decelerating to maintain an average flow. An oscillatory control valve is a poor practice that will lead to a leaky gland, but an oscillatory large pump is likely to be damaging bearings, seals, and wear-rings, and upsetting the power distribution. Such a setting serves no useful purpose. This is also a good reason to use digital network technology to transmit speed reference data to the drive, rather than running the risk of a noisy analog signal giving inaccurate speed references.

When one considers the effort put into ensuring better control from a final control element such as a control valve, one wonders why the VF drive has not received much attention as a viable alternative. In a benchmarking study performed by 12 chemical manufacturers, it was determined that the final control element performance had the largest impact on the cost of goods sold. Employing best practices on the final loop could save 1.5% of the cost of goods. Typically, the approach upon hearing this data is to consider digital positioners on the control valves in the plant. Our argument is to also look for opportunities to use VF drives as a means to more accurate control.

Behind the byline

Ian Gibson is a life senior member of ISA. He has a B.Sc. in chemistry and diplomas in applied chemistry, chemical engineering, and instrument technology. Recently, Gibson retired from Fluor Australia. He also is a member of the Royal Australian Chemical Institute and the Institute of Instrumentation, Control and Automation Australia Inc. Gibson is a registered practicing engineer of Queensland and author of several sections of the current edition of Instrument Engineers Handbook. Geoff Irvine has degrees in mechanical engineering and pure math. He is a manager at Rockwell Automation in Melbourne, Australia, where he works on applications in the petroleum and mining industries.