WO1997035667A1

WO1997035667A1 - Method for controlling an electrostatic precipitator

Info

Publication number: WO1997035667A1
Application number: PCT/SE1997/000513
Authority: WO
Inventors: Mats Thimanson
Original assignee: ABB Fläkt Aktiebolag
Priority date: 1996-03-28
Filing date: 1997-03-25
Publication date: 1997-10-02
Also published as: SE9601190D0; AU2524797A; SE9601190L; SE506245C2

Abstract

A method for controlling an electrostatic precipitator unit, which comprises discharge electrodes and collecting electrodes. Between the electrodes a varying high voltage is maintained by a pulsating direct current supplied thereto. Under the action of the electric field between the electrodes, the particles, charged by the current therebetween, are moved towards the collecting electrodes and deposited thereon. Dust deposited on the collecting electrodes is removed by mechanical rapping. One or more mechanical impulses are supplied to the electrodes individually or in groups in a predetermined manner. All collecting electrodes of the unit are cleaned during recurrent, relatively short, rapping periods separated by rapping intervals of essentially longer duration. During the whole rapping period or during an introductory part of the rapping period, the voltage between the electrodes is reversed in relation to the voltage between the electrodes during the intervals between the rapping periods. The frequency, pulse charge and/or pulse duration of the pulsating direct current are varied, thereby obtaining a plurality of combinations of frequency, charge and duration. An optimal combination of frequency, charge and duration is established. Amplitude and/or duration of the reversed voltage are adjusted in dependence on the pulse frequency for the established optimal combination of frequency, charge and duration.

Description

Method for Controlling an Electrostatic Precipitator

FIELD OF THE INVENTION

The present invention relates to a method for controlling an electrostatic precipitator unit, which comprises dis- charge electrodes and collecting electrodes. Between the electrodes, a varying high voltage is maintained by a puls¬ ating direct current supplied to the electrodes. Under the action of the electric field between the electrodes, the particles, charged by the current between the electrodes, are moved towards the collecting electrodes and deposited thereon. Dust deposited on the collecting electrodes is removed by mechanical rapping during recurrent, relatively short rapping periods separated by rapping intervals of essentially longer duration. During the rapping period, the voltage between the electrodes is reversed in relation to the voltage between the electrodes during the intervals between the rapping periods.

BACKGROUND OF THE INVENTION

Electrostatic precipitators are suitable in many contexts, especially in flue gas cleaning. Their design is robust and they are highly reliable. Moreover, they are most effici¬ ent. Degrees of separation above 99.9% are not unusual. Since, when compared with fabric filters, their operating costs are low and the risk of damage and stoppage owing to functional disorders is considerably smaller, they are a natural choice in many cases.

In order to optimise the operation and reduce the energy consumption at the same time as the separation is improved, several methods for pulse feeding of the current to the electrostatic precipitator unit have been suggested. Examples are to be found in US 4,052,177 and US 4,410,849. The former suggests the feeding of pulses in the order of microseconds, which means high demands on the electric equipment. The latter suggests pulses in the order of milliseconds, which may be achieved simply by selectively controlling thyristors, to which mains frequency alternat- ing current is supplied.

A procedure that is central to the function of an electrostatic precipitator is the rapping of the collecting electrodes. By rapping, the separated dust is released from the electrodes and falls down in collecting hoppers intend- ed therefor. The rapping may, at first sight, appear to be a trivial act, but as a matter of fact it is often of ut¬ most importance and the task to remove particles from a gas can not be fulfilled unless the rapping of the electrodes functions properly. The rapping frequency, i.e. how often the rapping is effected per unit of time, is controlled mainly by two opposite requirements. Since the dust cake on the col¬ lecting electrode by its growth gradually deteriorates the function of the filter, rapping is desirable before the dust cake becomes too thick. On the other hand, in each rapping, a considerable amount of dust is released and reentrained to the flue gas, resulting in a momentarily reduced degree of separation or even that more dust is leaving the precipitator than is entering. The requirements of the authorities regarding the level of emission are directed to the total amount of emis¬ sions. This means that also functional troubles must be taken into consideration. When using electrostatic precip- itators, the most frequent disturbance is the rapping of the collecting electrodes.

The dust is kept on the electrodes by electrical forces as well as other forms of adhesion. Some types of dust may adhere very strongly. Other types may, due to high resistivity, be pressed hard against the collecting elec- trodes by the electrical field. Since decades it is known that it may be advantageous or even necessary to reduce the voltage between the electrodes during the rapping periods to release the dust from the electrodes. Even at a reduced voltage it may be very difficult to remove some dust cakes from the collecting electrodes. In these cases the perform¬ ance of the electrostatic precipitator may be very bad. However, the dust often has a long time constant, i.e. it keeps the charge for a long time after the current supply is cut off. Thus, it is already also suggested that the voltage be reversed during the rapping period. One method to facilitate this is disclosed in e.g. SE 455 048. An unwanted side effect of the voltage reversal is however that the dust not always comes loose in the wanted form of big flakes, but disperses into the flue gas due to elec¬ trostatic forces, and leaves the electrostatic precipitator unit with the flue gas instead of falling down in the col- lecting hopper. This implies a restriction on the amplitude and duration of the reversed voltage. So far reversal of the voltage has not come to extensive use. One reason for this is believed to be the above mentioned problems of controlling the reversal, which may otherwise lead to a nearly explosive disintegration of the dust cake and thus a massive reentrainment of dust to the gas.

An electrostatic precipitator consists of a number of precipitator units, which are connected in series. Since the amount of dust separated, in a given unit, per unit of time decreases strongly with the increasing number of pre¬ cipitator units passed by the flue gas, the rapping must be controlled separately for each precipitator unit. To make it possible to separate dust released in one precipitator unit during rapping once more in a succeeding precipitator unit, the rapping should, however, be coordinated so as not to be carried out at the same time in several precipitator units. Also the rapping sequence in a precipitator unit containing a plurality of collecting electrodes to be rap¬ ped is selected carefully, such that all electrodes are rapped once during a so-called rapping cycle, where the rapping sequence, for the individual electrodes of a precipitator unit, has been selected for the purpose of minimising the reentrain ent of dust to the flue gas.

The growing number of control parameters in an elec¬ trostatic precipitator has increased the complexity in the control systems. One drawback with this is that the actual adjustment of, for instance, the rapping parameters increa¬ ses the disturbance in the function of the precipitator because it takes a considerable amount of time, during which the precipitator is working with not optimised para- meter settings.

If adjustment is effected manually by means of the reading of an opacimeter (tester for the optical density of smoke) , this takes such a long time that unfavourable para¬ meters during the adjustment can result in increased emis- sions to such extent that, during the time of adjustment, a main portion of the total emission takes place. Further¬ more, there is a risk that operational variations affect the adjustment negatively if considerable changes in the concentration of dust, the composition of dust or the gas temperature occur during the time needed for the adjust¬ ment. This already applies to the adjustment of the elec¬ tric parameters of the precipitator during normal operation and is a still more difficult problem in the adjustment of the parameters during the rapping, since the rapping fre- quency and with that also the frequency of reversal varies between minutes for the first precipitator unit to several hours or even days for the last one.

US 4,432,062 discloses an automatic optimisation of the rapping frequency in terms of the average value of the remaining dust content in the flue gas after the filter. The drawbacks of applying this method to optimisation of the voltage reversal, or other parameters which are to be individual for every precipitator unit, are a dependence on the measuring of the remaining dust content in the flue gas and the fact that the rapping frequency varies over several orders of magnitude between the precipitator units. When selecting the amplitudes of the reversed voltage of the precipitator units as independent parameters, this leads to simultaneous optimisation of many parameters, which easily results in sub-optimisation or the absence of convergence of the optimising algorithm.

OBJECT OF THE INVENTION

The main object of the invention is to suggest an improved method to control the rapping in an electrostatic precipit¬ ator achieving a higher degree of cleaning of the collect¬ ing electrodes, thereby obtaining considerable advantages in the form of lower emissions. This is the case especially in comparison with the methods that are based on measure- ment of dust concentration.

A second object of the present invention is to suggest a method, which by a new evaluation of the operating elec¬ tric parameters of the individual electrostatic precipit¬ ator unit controls the amplitude and duration of a reversed voltage between the electrodes of that unit during the rapping.

A third object of the present invention is to suggest a method, which can follow variations in the operational conditions by quickly adjusting the amplitude and duration of a voltage reversal.

SUMMARY OF THE INVENTION

The present invention relates to a method for controlling an electrostatic precipitator unit, which comprises dis¬ charge electrodes and collecting electrodes. Between the electrodes there is maintained a varying high voltage by a pulsating direct current supplied thereto. Under the action of the electric field between the electrodes, the part- icles, charged by the current between the electrodes, are moved towards the collecting electrodes and deposited thereon. Dust deposited on the collecting electrodes is removed by mechanical rapping. One or more mechanical im¬ pulses are supplied to the electrodes individually or in groups in a predetermined manner. All collecting electrodes of the unit are cleaned during recurrent, relatively short rapping periods separated by rapping intervals of essenti¬ ally longer duration. During the whole rapping period or during an introductory part of the rapping period, the voltage between the electrodes is reversed in relation to the voltage between the electrodes during the intervals between the rapping periods.

In the method according to the invention, the frequen¬ cy, pulse charge and/or pulse duration of the pulsating direct current are varied, thereby obtaining a plurality of combinations of frequency, charge and duration. An optimal combination of frequency, charge and duration is establish¬ ed. Amplitude and/or duration of the reversed voltage are controlled in dependence on the pulse frequency for the established optimal combination of frequency, charge and duration.

GENERAL DESCRIPTION OF THE INVENTION

The basic idea behind the invention is, that a dust cake, on the collecting electrodes, in an electrostatic precipit- ator unit is, at least in some cases, pressed against the electrodes of a strong electrostatic force. This is mainly due to charges in the dust layer. The dust layer acts as a leaky capacitor with a considerable time constant. It may take many seconds, or even minutes or hours, to discharge the layer when left without additional charging. By revers¬ ing of the voltage the electrostatic forces are reversed and thereby the rapping may be far more efficient.

When needed the reversed voltage is applied during the whole rapping period. In many cases it is only applied for a short time in the beginning of the rapping period, just to make the dust cake more porous and reduce the internal strength of the dust cake. The improvement presented by this invention is a method to optimize the reversal of the voltage, the amplitude as well as the duration.

Under given circumstances, which are constant as far as possible, a purely electric optimisation of the oper- ation of one individual electrostatic precipitator unit is effected such that optimal values of pulse frequency, pulse charge and pulse duration are obtained. The pulse frequency obtained at the determined optimal combination is then used as a controlling parameter for selection of the amplitude and duration of a reversal of the voltage between the elec¬ trodes in this electrostatic precipitator unit during the rapping period. Every unit is treated in the same way but handled according to the operational data in just that unit. One way of using the inventive method is to select the amplitude of the reversed voltage higher at a lower value of the pulse frequency for the determined optimal combin¬ ation of frequency, charge and duration. This is done to reflect the experience that a high resistivity dust has an optimal operation at a low frequency and also means a high electric field in the dust.

Another way is to select the amplitude of the reversed voltage essentially constant, independent of the frequency, if the pulse frequency for the determined optimal combin- ation of frequency, charge and duration is below a first predetermined limit and that the amplitude of the reversed voltage is selected essentially linearly decreasing with increasing value of the pulse frequency, if the frequency for the determined optimal combination of frequency, charge and duration is above the first predetermined limit.

A preferred mode of operation is that the amplitude of the reversed voltage is selected essentially constant, at a highest value independent of the frequency, if the pulse frequency for the determined optimal combination of fre- quency, charge and duration is below a first predetermined limit, that the amplitude of the reversed voltage is selected essentially linearly decreasing with increasing value of the frequency, if the pulse frequency is above the first predetermined limit, but below a second predetermined limit, and that the amplitude of the reversed voltage is selected essentially constant, at a lowest value independ- ent of the frequency, if the pulse frequency is above the second predetermined limit.

The first predetermined limit may be selected in the range 1-30 Hz, preferably in the range 4-10 Hz. The second predetermined limit may be selected in the range 10-100 Hz, preferably in the range 15-40 Hz.

The amplitude and duration of the reversed voltage during rapping can, in addition to the optimal pulse fre¬ quency, also be a function of the minimum level of the varying high voltage immediately before the rapping period and/or of the average value of the pulsating direct current immediately before the rapping period. In particular, these parameters may decide the duration of reversal.

The reversal of the voltage can also be carried out during an introductory part of the rapping period, this introductory part being preferably shorter than one tenth of the rapping period. It may also be carried out in the form of short intervals, synchronised with the mechanical rapping. These short intervals should preferably be shorter than one fifth of the time between consecutive raps. During the remaining time of the rapping period the voltage is not reversed, but reduced.

The optimising of the duration of voltage reversal may be carried out in a similar manner. The duration of the re¬ versed voltage is selected longer at lower values of the pulse frequency for the determined optimal combination of frequency, charge and duration. Also here a partially lin¬ ear approximation can be used. The duration of the reversed voltage may be selected constant, at a highest value inde¬ pendent of the frequency, if the pulse frequency for the determined optimal combination of frequency, charge and duration is below a first predetermined limit. The duration of the reversed voltage may be selected linearly decreasing with increasing value of the frequency, if the pulse fre¬ quency for the determined optimal combination is above the first predetermined limit, but below a second predetermined limit. The duration of the reversed voltage may be selected constant, at a lowest value independent of the frequency, if the pulse frequency for the determined optimal combination is above the second predetermined limit.

The amplitude of the reduced voltage, during the main part of the rapping period, can, in the applicable cases, also be controlled in dependence on the pulse frequency for the determined optimal combination of frequency, charge and duration.

The method is particularly convenient when the pulsating direct current has the form of a pulse train, which is synchronised with the frequency of the mains voltage and in which the pulses are generated by a part of a half-wave of the mains voltage being supplied, by means of a phase angle controlled rectifier (thyristor) after step-up transform- ation, to the electrodes of the precipitator, whereupon a plurality of periods of the mains voltage are allowed to pass without current being supplied to the electrodes. Then a part of a half-wave is again supplied, followed by a plurality of periods without current etc. In this manner, the frequency, pulse charge and/or pulse duration of the pulsating direct current can be varied such that a plurality of combinations of frequency, charge and duration are obtained. For each combination, a figure of merit is measured or calculated. The figures of merit are used to determine an optimal combination. The amplitude of the reversed voltage is controlled as a function of the pulse frequency for the determined optimal combination.

Figures of merit for the description of the state of operation of an electrostatic precipitator are per se known, but not for the controlling of the voltage reversal. Examples of figures of merit can be maximum peak value, maximum average value or maximum valley value of the voltage between the electrodes of the precipitator. Such a method is suggested in US 4,311,491.

It may advantageously also be a value determined on more sophisticated grounds, such as the ratio between peak voltage and pulse charge, possibly when one of these para¬ meters is kept constant during the adjustment. This is suggested in EP 0 184 922.

A convenient and effective method for determining a figure of merit, where each combination of parameters can be reflected by an individual figure of merit is, as sug¬ gested in PCT/SE92/00815, the determining of a reference voltage level, between the peak value and the minimum value of the voltage between discharge electrodes and collecting electrodes, and ascribing a positive value to the time dur¬ ing which the voltage is above this level, and ascribing a negative value to the time during which the voltage is below this level, by weighting according to the function

A - U-(U-U_ref)

wherein U is the voltage for a given point of time between the electrodes in the precipitator.

The optimisation of the pulse parameters and selection of the best amplitude and duration of the reversed voltage, are carried out individually for each precipitator unit with its associated high voltage supply and rapping equip¬ ment, which means that the various time scales of the rap¬ ping in the different precipitator units need not be mixed in the optimising process.

BRIEF DESCRIPTION OF THE DRAWINGS

The invention will now be described in more detail with reference to the accompanying drawings in which: Fig. 1 shows the fundamental relation between current and voltage, as a function of the time, in an electrostatic precipitator;

Fig. 2 shows an actually measured voltage, as a function of the time, in an electrostatic precipitator;

Fig. 3 shows the peak value and the minimum value of the voltage between the electrodes in an electrostatic preci- pitator at a constant pulse frequency, as a function of the average value of the current through the precipitator;

Fig. 4 shows the fundamental relation between the average value of the current through a precipitator and the associ- ated peak value, average value and minimum value of the voltage between the electrodes of the precipitator under operating conditions where electiric discharges in a sepa¬ rated dust layer may occur;

Fig. 5 illustrates a method of evaluating the voltage between the electrodes of a precipitator;

Fig. 6 is a simplified view of a plant for carrying out the proposed method for controlling the voltage reversal of an electrostatic precipitator;

Fig. 7 is a schematic view showing the relation between the optimal pulse frequency and the amplitude of the reversed voltage; and

Fig. 8 is a view of a partially linear approximation of the function according to Fig. 7, suited for a controlling algorithm. DESCRIPTION OF EMBODIMENTS

Fig. la shows the general relation between current and voltage in an electrostatic precipitator supplied with cur- rent from a phase angle controlled rectifier consisting of thyristors, when the thyristors are fired in all the half periods of the alternating voltage. Fig. lb shows the same relation when the thyristors are fired merely in every third half period. The method according to the present inv- ention will ordinarily be used at essentially lower firing frequencies than the showed ones, which for the sake of clarity have not been drawn to scale. The relation between the levels therefore is also completely irrelevant.

Fig. 2 shows the actually measured voltage in a more realistic situation, where the thyristors are fired in every ninth half period and then yield a very steep voltage increase, whereupon the voltage first falls steeply and then more and more slowly. The great difference between the peak value and the minimum value of the voltage between the electrodes is fully realistic. The scale change makes com¬ parisons with the preceding Figure unsuitable. In Fig. 2, the peak value of the voltage is about 58 kV and the mini¬ mum value of the voltage about 16 kV. Current pulses are supplied with a frequency of about 11 Hz. If the firing angles of the thyristors are varied at a constant frequency, both peak and minimum values of the voltage will vary. Under favourable operating conditions or close to optimal operation, the minimum value is relatively independent of the firing angle, while the peak value grows monotonously with a decreasing firing angle, i.e. an in¬ creased conducting period of the thyristors. Under diffi¬ cult operating conditions and when operating with unsuit¬ able parameters, the minimum voltage decreases, even at low current, with a decreasing firing angle, and at higher currents both the average value and the peak value of the voltage decrease. Fig. 3 illustrates the actually measured relations for a given pulse frequency in close to optimal operation.

Fig. 4 illustrates the fundamental relations between current and voltage in a precipitator when separating a dust having high resistivity. The curves 41, 42 and 43 correspond to the minimum value 41, average value 42 and peak value 43 of the voltage between the electrodes of the precipitator. All three curves have a local maximum. These maximum values can be seen as examples of electric para- meters indicating optimal operation. By varying the firing angle and pulse frequency, it is possible to find an opti¬ mal combination.

Fig. 5 illustrates a further method for determining the figure of merit for a given combination of parameters. Fig. 5 is a picture, which for the sake of clarity is somewhat distorted, showing how the voltage between the electrodes of the precipitator varies with time during the interval from the start of a current pulse to the start of the next current pulse. It is also indicated that the measuring of the voltage between the electrodes of the precipitator takes place at a plurality of discrete and equidistant points of time. In practice, the measuring takes place at essentially more points of time than those shown, for instance 1-3 times per millisecond. These measured values are stored in a preferably computerised control unit 630, shown in Fig. 6, and by means of the value of U_ref, which is also stored in the control unit 630, A^ = Uj (Ui-U_ref) is calculated for each measuring point. The average value of A^ is calculated automatically in the control unit 630, and the result is stored as a figure of merit for the relevant combination of pulse fre¬ quency and firing angle of the thyristors in the corres¬ ponding rectifier 621, 622 and 623.

Fig. 6 shows schematically a plant for carrying out the present method. A precipitator 600 having an inlet duct 641 and an outlet duct 642 comprises three precipitator units 601, 602, 603 each having a dust hopper 611, 612, 613. The precipitator units are supplied with pulsating direct current from three rectifiers 621, 622, 623. The rectifiers 621-623 are controlled and monitored by a cont¬ rol unit 630. The control unit 630 also communicates with devices 651, 652 and 653 for rapping of the collecting electrodes in the precipitator units 601, 602 and 603.

Fig. 7 illustrates an assumed fundamental relation between pulse frequency and optimal amplitude of the re¬ versed voltage. The basic experience is that no reversal, or a very small reversal shall apply for high frequencies, but an increasing reversal shall take place for decreasing frequencies.

Fig. 8 shows an approximation of the function in Fig. 7 as a partially linear function, suited for a controlling algorithm. If the optimal combination of parameters means a pulse frequency lower than a predetermined first limit Li - 4 Hz, the amplitude of the reversed voltage during rapping is selected to 15 kV. The amplitude of the reversed voltage is reduced for a higher pulse frequencies so as to finally be kept constant at a minimum level, e.g. 1 kV, when the pulse frequency exceeds a predetermined second limit L₂. This minimum level can also be zero. A suitable changing tactic must be based on certain experience of the plant and possibly also of the dust in the gas to be cleaned.

In the suggested method, the rectifier 621 supplies, with parameters varying according to a predetermined principle, pulsating direct current to the electrodes (not shown) of the unit 601. The control unit 630 evaluates the supplied pulse-shaped current and the occurring voltage and calcul¬ ates a figure of merit for each combination of parameters or for each group of combinations. According to a predeter¬ mined strategy, the combination of parameters which may be considered the electrically optimal one is selected by means of these figures of merit, and the operation contin¬ ues with this established combination of parameters, for instance at a pulse frequency of 17 Hz. Immediately before the rapping the voltage is reversed, its amplitude is selected according to the controlling algorithm, for instance to 8 kV.

At short intervals, an electric optimisation of the current supply to all three units 601, 602, 603 takes place. This is initiated and evaluated by the control unit 630.

For the units 602 and 603 positioned downstream, the selection of amplitude and duration of the reversed voltage during rapping is carried out in the same manner.

ALTERNATIVE EMBODIMENTS

The method according to the invention is of course not limited to the embodiment described above, but may be modified in a number of ways within the scope of the appended claims.

The method is quite independent of how the voltage reversal technically is performed. The method can be applied to a plurality of other techniques of supplying current in the form of pulses to electrostatic precipitators. Examples of such techniques are pulse-width-modulated high frequency and other forms of so-called "switched mode power" as well as the use of thyristors, which can be "switched off". The method is also suited for use in the very special pulse rectifiers which generate pulses in the order of micro¬ seconds.

Claims

16 CLAIMS

1. A method for controlling an electrostatic precipitator unit,

comprising discharge electrodes and collecting electrodes, between which a varying high voltage is maintained, by a pulsating direct current supplied to the electrodes, such that under the action of the electric field between the electrodes, the particles charged by the current therebe¬ tween are moved towards the collecting electrodes and are deposited thereon,

dust deposited on the collecting electrodes being removed by mechanical rapping of the collecting electrodes by one or more impulses being periodically supplied to the elec- trodes individually or in groups in a predetermined manner, such that all the collecting electrodes of the unit are cleaned during recurrent, relatively short rapping periods separated by rapping intervals of considerably longer duration,

and the voltage between the electrodes of the precipitator during the whole rapping period or during an introductory part of the rapping period being reversed in relation to the voltage during the intervals between the rapping periods,

c h a r a c t e r i s e d in

that the frequency, pulse charge and/or pulse duration of the pulsating direct current are varied, thereby obtaining a plurality of combinations of frequency, charge and dur¬ ation, that a combination of frequency, charge and duration which is optimal to the operation of the precipitator unit is determined, and

that the amplitude and/or duration of the reversed voltage during the rapping period being adjusted in dependence on the pulse frequency for the determined optimal combination of frequency, charge and duration.

2. The method as claimed in claim 1, c h a r a c t e r ¬ i s e d in that the amplitude of the reversed voltage is selected higher at a lower value of the pulse frequency for the determined optimal combination of frequency, charge and duration.

3. The method as claimed in claim 2, c h a r a c t e r ¬ i s e d in that the amplitude of the reversed voltage is selected essentially constant, independent of the fre¬ quency, if the pulse frequency for the determined optimal combination of frequency, charge and duration is below a first predetermined limit.

4. The method as claimed in claim 2, c h a r a c t e r ¬ i s e d in that the amplitude of the reversed voltage is selected essentially linearly decreasing with increasing value of the pulse frequency, if the frequency for the determined optimal combination of frequency, charge and duration is above a first predetermined limit.

5. The method as claimed in claim 2, c h a r a c t e r ¬ i s e d in

that the amplitude of the reversed voltage is selected essentially constant, at a highest value independent of the frequency, if the pulse frequency for the determined opti¬ mal combination of frequency, charge and duration is below a first predetermined limit, that the amplitude of the reversed voltage is selected essentially linearly decreasing with increasing value of the frequency, if the pulse frequency for the determined optimal combination of frequency, charge and duration is above the first predetermined limit, but below a second predetermined limit, and

that the amplitude of the reversed voltage is selected essentially constant, at a lowest value independent of the frequency, if the pulse frequency for the determined opti¬ mal combination of frequency, charge and duration is above the second predetermined limit.

6. The method as claimed in claims l, c h a r a c t e r ¬ i s e d in that the duration of the reversed voltage is selected longer at a lower value of the pulse frequency for the determined optimal combination of frequency, charge and duration.

7 . The method as claimed in claim 6 , c h a r a c t e r ¬ i s e d in

that the duration of the reversed voltage is selected essentially constant, at a highest value independent of the frequency, if the pulse frequency for the determined opti¬ mal combination of frequency, charge and duration is below a first predetermined limit,

that the duration of the reversed voltage is selected essentially linearly decreasing with increasing value of the frequency, if the pulse frequency for the determined optimal combination of frequency, charge and duration is above the first predetermined limit, but below a second predetermined limit, and that the duration of the reversed voltage is selected essentially constant, at a lowest value independent of the frequency, if the pulse frequency for the determined optimal combination of frequency, charge and duration is above the second predetermined limit.

8. The method as claimed in any one of claims 3-7, c h a r a c t e r i s e d in that the first predetermined limit is selected in the range 1-30 Hz, preferably in the range 4-10 Hz.

9. The method as claimed in any one of claims 5-7, c h a r a c t e r i s e d in that the second predetermined limit is selected in the range 10-100 Hz, preferably in the range 15-40 Hz.

10. The method as claimed in any one of claims 1-9, c h a r a c t e r i s e d in that the amplitude and/or duration of the reversed voltage during the rapping period are adjusted also in dependence on the minimum level of the varying high voltage immediately before the rapping period.

11. The method as claimed in any one of claims 1-10, c h a r a c t e r i s e d in that the amplitude and/or duration of the reversed voltage during the rapping period are adjusted also in dependence on the average value of the pulsating direct current immediately before the rapping period.

12. The method as claimed in any one of claims 1-11, c h a r a c t e r i s e d in

that the voltage between the electrodes of the precipitator only during an introductory part of the rapping period be- ing reversed in relation to the voltage during the inter¬ vals between the rapping periods, that this introductory part has a duration being essen¬ tially shorter than the duration of the rapping period, preferably shorter than one tenth of the duration of the rapping period, and

that the voltage between the electrodes of the precipi¬ tator unit during the main part of the remaining part of the rapping period has normal polarity, but a value being reduced in relation to the voltage immediately before the rapping period.

13. The method as claimed in any one of claims 1-11, c h a r a c t e r i s e d in that the voltage between the electrodes of the precipitator unit only during recurrent short time intervals during the rapping period, preferably synchronously with the mechanical rapping, being reversed in relation to the voltage during the intervals between the rapping period,

that said short time intervals has a duration being essen¬ tially shorter than the time between the individual raps, preferably shorter than one fifth of the time between the raps, and

that the voltage between the electrodes of the precipi¬ tator unit during the main part of the remaining time between the raps has normal polarity, but a value being reduced in relation to the voltage immediately before the rapping period.

14. The method as claimed in claim 12 or 13, c h a r ¬ a c t e r i s e d in that the prevailing reduced voltage between the electrodes of the precipitator with normal polarity during the main part of the rapping period is adjusted in dependence on the pulse frequency for the determined optimal combination of frequency, charge and duration.