A model-based control system provides complete machine protection using a minimum number of sensors.

Centrifugal pumps are widely used in industrial applications, and pumps absorb nearly a quarter of the electric energy of the European industry. Pumps driven by medium- to low-power motors (70 kilowatts [kW]) represent almost all of the installed machines in oil and gas process plants. Despite the high cost of maintaining pumping systems, control, protection and diagnostic systems for medium- to low-power pumps are not yet widely applied. This type of solution requires external sensors to acquire process parameters, which could have a considerable impact on initial costs.

A centrifugal pump model-based control system can solve these challenges by incorporating variable speed drive (VSD) techniques and using installed sensors.

Modern control systems change pump rotational speed using frequency converter VSD techniques. Facilities can improve process output and reduce machinery-related production costs with accurate, precise and reliable real-time information about how the machinery works and through congruent control and protective actions.

Recent literature discusses many sensorless control techniques, which have a considerable advantage in terms of energy and implementation costs. However, these methods do not allow for simultaneous control of the continuous monitoring system and diagnostics for the equipment. A model-based control system, however, provides the following features:

  • an optimal control action on the operational state of the pumping system using a minimum number of field transducers with a consequent reduction of installation costs
  • continuous automatic complete pump protection from all possible failure causes through the implementation of a pump performance model

While minimizing the number of sensors and components reduces the possibility of relative malfunctions, the system architecture of a model-based control system provides continuous machine monitoring and diagnostics using an advanced diagnostic algorithm that integrates field measurements and performance predictions.

The algorithm is based on field data and on the determination of the pump operative expected parameters from the implemented pump model.

The system monitoring and diagnostic capabilities are designed to increase machine availability and reliability while reducing downtime and maintenance costs.

The process-oriented control architecture also reduces costs for process surveillance, while the VSD technique reduces energy consumption. These benefits are possible because the characterization model of centrifugal pump performance is the central element of this system.

State-of-the-Art Control Techniques

A typical pumping system consists of a driver, a centrifugal pump, valves and tanks. Advanced systems may include a VSD block and sensors that produce the process variable feedback (see Figure 1).

Figure 1. Typical pumping system structureFigure 1. Typical pumping system structure (Graphics courtesy of IPC)

At any given moment, the overall process status can be described as the group of relevant process parameters (flow rate, liquid level, process pressures, fluid properties, etc.) as well as pump operative parameters. The actual value of the process parameters depends on the pump characteristics and the hydraulic load. When the load changes, the operative state of the pump changes.

Pump parameters susceptible to variation are flow rate, suction and discharge pressure, power, and efficiency. Selection of the more suitable control system depends on factors such as plant configuration, the required operational process state and price.

Such systems can use four main techniques for pump system control: throttle control, bypass control, on/off control and VSD control. The first three are simple and low-cost. But because these controls run the pump at a fixed speed, they are relatively inefficient, consume more energy, produce machine wear and incur operating costs.

While the VSD control approach is more efficient, it has the disadvantage of higher implementation costs. But this type of control system adjusts the motor speed, permitting the pump to match user demand. Furthermore, producing a centrifugal pump control system that saves energy is possible using a VSD, as well as machine control, monitoring and diagnostic features.

Pump Model

VSD systems are equipped with powerful control units that enable the monitoring of inverter power output and actual motor speed. The relationship between these variables and the pump process parameters is defined by pump performance curves: flow-head (QH) and flow-power (QP).

Both curves are essential parts of the vendor pump documentation.

Another relevant system parameter is the net positive suction energy (NPSE) or the net positive suction head (NPSH). The QH and QP curves demonstrate the trend of the head and the pump power consumption against the flow rate, respectively. These correlations are used to determine the flow rate without the need for sensors.

For variable speed pumps, manufacturers present pump performance maps as curves plotted for different speeds. Even when only the nominal speed curve is available, it is always possible to determine the operational points at an arbitrary speed using the affinity laws.
Further characterization is necessary to perform pump diagnostics. In particular, it is necessary to determine the curves NPSHRQ (NPSHR, NPSH required; Q, volume flow), ηQ (η, efficiency) and PHQ (PH, hydraulic power) that vary with the pump’s speed of rotation.

Model-Based Control & Diagnostics

To develop additional diagnostic capabilities, a model-based control system includes a few elementary sensors. Compared with the sensorless version, this choice introduces modest installation costs.

The proposed model-based control system uses two pressure transducers installed near the pump flanges and a vibration sensor to perform control tasks and to implement the pump diagnostic feature, respectively.

Figure 2. A model-based control system architecture  Figure 2. A model-based control system architecture

The system also includes a frequency converter and a programmable logic controller (PLC), as shown in Figure 2. The PLC provides multivariable control, executes sequencing tasks and provides computations for the pump model.

The algorithm also allows calculation of the expected values for power, efficiency and NPSHR.

Diagnostic Algorithm

The algorithm is based on real-time monitoring of machine and process parameters. It calculates the deviation between expected and actual performance parameters in real time. Expected performance is obtained from reference and design characteristics, adjusted to actual operative conditions; actual performance is derived from field measurements in a specific operative condition. The time trend of the amount of deviation can give early indication of impending problems.

The diagnostic system is based on knowledge of the hydraulic power (Phe) and the expected pump efficiency (ηe). This information is used to determine the expected power (Pwe= Phe/ ηe), which is then compared to the actual power (Pw), measured by a frequency converter onboard sensor.

This model-based control system reduces energy costs and eliminates the need for extra components. The result is an increase in machine reliability and availability, particularly for centrifugal pumps used in oil and gas applications.