Looking for Spirax Sarco products and services?

Basic Control Theory

Masthead background

Practical installation and commissioning advice for valves, actuators, sensors, controllers and more.

Installation

Valves

Before installing a control valve it is necessary to ensure that the size, pressure rating, materials and end connections are all suitable for the conditions under which the valve is expected to work.

All reputable manufacturers of automatic control equipment will provide detailed instructions covering the correct installation procedure for their equipment. Data will also be provided on how to set up the equipment, plus any routine and regular maintenance to be undertaken. In most cases, the manufacturer will also offer an on-site commissioning service. In some cases, a regular after-sales maintenance contract can be agreed. Module 5.5 covers the major points to be considered before installation.

Piping upstream and downstream of the control valve should be clear and unobstructed. The correct operation of a valve will be impaired if it is subject to line distortion stresses. It is important to ensure that all flanged joints are square and true and that pipework is adequately supported.

Control valves should generally be installed in horizontal pipelines with the spindles vertical.

Pipework systems will often be subjected to pressure testing prior to use. This test may be carried out at a pressure above the normal working conditions. It is necessary to ensure that the control valve and its internals are designed to withstand this higher test pressure.

Control valves are essentially instruments and will be damaged if dirt or other abrasive or obstructive materials are allowed to enter them. It is essential in most applications to prevent this by fitting pipeline strainers upstream of any control valve.

Valves must also be accessible for routine maintenance, such as re-packing of glands and the replacement of internals. To facilitate this sort of work, isolating valves of a full bore pattern either side of the valve will keep plant downtime to a minimum while the work is carried out.

If a plant must be kept in operation at all times, even when a control valve is being inspected or maintained, it may be necessary to fit a valved bypass. However, the valve used in the bypass must be of good quality and should either be a characterised throttling valve or another control valve of the correct Kvs. Any leakage through it during normal operation will affect the action of the control system. It is not recommended that manual bypasses be fitted under any circumstances. The control valve must be installed to ensure the correct direction of flow of the medium passing through the valve. Usually a ‘direction of flow’ arrow is cast into the body of the control valve. The valve must have a suitable flow capacity and incur an acceptable pressure drop.

In steam lines, it is important to provide a steam separator and/or a trapping point upstream of the valve, as shown in Figure 5.5.1. This will prevent the carryover of condensate through the control valve, which would otherwise reduce its service life. This drain point is also important if the control valve is likely to remain closed for any length of time. If a condensate drain is not fitted, waterhammer and potentially serious damage can result when the valve opens. The provision of a steam separator and strainer ensures good steam conditioning.

fig 5.5.1 A pneumatic pressure reducing station with steam conditioning

Actuators/sensors

Again, the manufacturer’s instructions must be observed. Actuators are normally mounted vertically above the control valve, although different arrangements may be recommended if an electric actuator is mounted to a valve handling a high temperature medium, such as steam.

Generally, actuators should be located away from conditions such as excess heat, high humidity or corrosive fumes. These are likely to cause premature failure in components such as diaphragms or electric/electronic items. Manufacturers should state the recommended maximum ambient temperature conditions for their equipment. With some electric actuators, if condensation is likely to occur within the actuator, models with a built-in heater are available. Where such conditions cannot be avoided, actuators should be purchased which are suited to the installed conditions.

Enclosures for actuators, positioners, and so on, will usually carry an enclosure rating conforming to a national electrical code. This should specify the degree of immunity of the box to the ingress of dust and water. It is worthless using an electric actuator whose enclosure has a low rating to the ingress of water, if it is likely to be hosed down!

Care must be taken to ensure that sensors are fully and correctly immersed if they are to carry out their sensing function effectively. The use of pockets will enable inspection or replacement to take place without the need to drain the piping system, vessel or process plant. In contrast, pockets will delay response times. The use of heat conducting paste in the pocket will minimise any delay in response.

Power and signal lines

With a pneumatic system, compressed air and pneumatic signal lines must be dry, free from oil and dirt, and leak tight. Locating the pneumatic controller near the valve and actuator will minimize any delay due to the capacity and resistance of the signal line.

Usually, the valve, actuator and any positioners or converters, will be supplied as a complete pre-assembled unit. If they are not, the actuator will need to be mounted to the valve, and the positioner (for a pneumatic control) to the actuator. The assembly will then have to be set up properly, to ensure that the correct valve stroke, etc. is achieved, all in accordance with the manufacturer’s instructions.

Electrical wiring for electric/electronic and electropneumatic controls

All too often, many apparent ‘control’ problems are traced back to incorrect wiring. To quote an obvious problem encountered as an extreme example, connecting a 110 V supply to a 24 V rated motor, will result in damage! Care must be taken with the wiring system, in accordance with the manufacturer’s instructions, and subject to any local regulations.

‘Noise’ or electrical interference in electrical systems is often encountered, resulting in operational problems which are difficult to diagnose. The use of screened cable, separately earthed conduit or a self-acting or analogue controller may be necessary. Cables should be protected from mechanical damage.

Controllers

As mentioned earlier, the application will generally produce changes that are slower than the response time of the control system. This is why the parameters of the controller, the proportional band or gain, integral time and derivative time, must be tuned to suit each specific application/task.

There are a number of methods for adjusting controller parameters, most of which involve the use of mathematics. The behaviour of a control loop can be predicted mathematically but the process or application characteristics are usually determined by empirical measurement, which can be difficult. Methods based on design heat transfer ratios can be found, but these are outside the scope of this Module.

Before setting the control parameters, it is useful to review each of the control terms (P, I and D), and the three options regarding settings, for instance, too wide, too narrow, and correct.

P-band (Figure 5.5.2)

If P-band is too wide, large offset occurs but system is very stable (curve A).

Narrowing the P-band will reduce the offset.

Too narrow a P-band will cause instability and oscillation, (curve B).

The optimum P-band, curve C, is achieved at a setting just slightly wider than that causing permanent oscillation.

Fig 5.5.2 P-band setting reaction to change in load

Summary of P-band (proportional action)
Correct P-band = Good stability, good response Some offset
Larger P-band = Better stability, slower response Larger offset
Smaller P-band = Instability, quicker response Smaller offset with oscillation

Integral action (Figure 5.5.3)

With too short an integral time, temperature (curve A) will cross the set point and some oscillation will occur.

An excessive integral time will result in the temperature taking too long to return to set point (curve B).

Curve C shows a correct integral time setting where the temperature returns to set point as rapidly as possible without any overshoot or oscillation.

Fig 5.5.3 Integral time reaction to change in load
 Summary of integral action
 Correct IAT =  Elimination of offset  Stable - no overshoot
 Too short IAT =  Elimination of offset  Response too fast, causing instability and overshoot
 Too long IAT =  Elimination of offset  Slow response, stable, no overshoot

Derivative action (Figure 5.5.4)

An excessive derivative time will cause an over-rapid change in temperature, overshoot and oscillation (curve B).

Too short a derivative time allows the temperature to deviate from the set point for too long (curve A).

The optimum setting returns the temperature to the set point as quickly as possible and is consistent with good stability (curve C).

Fig 5.5.4 Derivative time reaction to change in load
Summary of derivative action 
 Correct derivative time =  Quick response, stable
 Too much D time =  Faster response leading to overshoot and instability
 Too little D time =  Slower response

Commissioning

Practical methods of setting up a controller

Each controller has to be set up individually to match the characteristics of a particular system. Although there are a number of different techniques by which stable and fast control can be achieved, the Ziegler-Nicholls method has proven to be very effective.

The Ziegler-Nicholls method

The Ziegler-Nicholls frequency response method (sometimes called the critical oscillation method) is very effective in establishing controller settings for the actual load. The method uses the controller as an amplifier to reach the point of instability. At this point the whole system is operating in such a way that the temperature is fluctuating around the set point with a constant amplitude, (see Figure 5.5.5). A small increase in gain, or a reduced proportional band, will make the system unstable, and the control valve will start hunting with increasing amplitude.

Conversely, an increased proportional band will make the process more stable and the amplitude will successively be reduced. At the point of instability, the system characteristic is obtained for the actual operating conditions, including the heat exchanger, control valve, actuator, piping, and temperature sensor.

The controller settings can be determined via the Ziegler-Nicholls method by reading the time period (Tn), of the temperature cycles; and the actual proportional band setting at the point of instability.

Fig 5.5.5 Instability caused by increasing the controller gain, with no I or D action

The procedure for selecting the settings for PID parameters, using the Ziegler-Nicholls method, is as follows:

  1. Remove integral action on the controller by increasing the integral time (Ti) to its maximum.
  2. Remove the controller’s derivative action by setting the derivation time (TD) to 0.
  3. Wait until the process reaches a stable condition.
  4. Reduce the proportional band (increase gain) until the instability point is reached.
  5. Measure the time for one period (Tn) and register the actual P-band (proportional band)setting on the controller at this point.
  6. Using this setting as the start point, calculate the appropriate controller settings according to the values in Figure 5.5.6.

Fig 5.5.6 Ziegler-Nicholls calculation

Proportional band Integral time Derivative time
P I D control P-band x 1.7 Tn/2 Tn/8
P I control P-band x 2.2 Tn/1.2
P control P-band x 2.0

The controller settings may be adjusted further to increase stability or response. The impact of changing the setting of the PID parameters on stability, and the response of the control, is shown in Figure 5.5.7.

Fig. 5.5.7 Effect of changing PID settings

Stability Response
Increase P Band Increased Slower
Increase Ti Increased Slower
Increase TD Decreased Faster

Bumpless transfer

The technical specifications for controllers include many other terms and one that is frequently encountered is ‘bumpless transfer’.

Most controllers incorporate a ‘Manual’ – ‘Auto’ switch and there can be times when certain control situations require manual control. This makes interruption of the automatic control loop necessary. Without bumpless transfer, the transfer from Auto to Manual and vice versa would mean that the control levels would be lost, unless the manual output were matched to the automatic output.

Bumpless transfer ensures that the outputs - either Manual to Auto or Auto to Manual - match, and it is only necessary to move the switch as appropriate.

Self-tuning controllers

Contemporary microprocessors provide the ability for some functions, which previously required a computer, to be packed into the confined space of a controller. Amongst these, was the ability to ‘self-tune’. Controllers that no longer require a commissioning engineer to go through the process of setting the P I D terms have been available for many years. The self-tune controller switches to on/off control for a certain period of time. During this period it analyses the results of its responses, and calculates and sets its own P I D terms.

It used to be the case that the self-tune function could only apply itself during system start-up; once set by the controller, the P I D terms remained constant, regardless of any later changes in the process.

The modern controller can now operate what is termed an adaptive function, which not only sets the required initial P I D terms, but monitors and re-sets these terms if necessary, according to changes in the process during normal running conditions.

Such controllers are readily available and relatively inexpensive. Their use is becoming increasingly widespread, even for relatively unsophisticated control tasks.