This invention relates to a process for the demercurisation of flue gases.
Mercury is a naturally occurring compound that is still widely used in industry.
Mercury is toxic in all its forms.
It can be assimilated by practically all living organisms and is therefore concentrated throughout the food chain.
Ultimately, it is absorbed by humans, where it develops its toxicity.
From this point of view, it is of little importance whether mercury is present in the form of metal, organic salts or organic compounds such as methylmercury.
Industry and human activities release mercury into the atmosphere, which can be transported over very long distances.
This is why most countries have introduced restrictive regulations to limit the quantities of mercury released into the environment.
Industries that can release mercury include chlorine production plants, foundries, coal- fired power stations (which naturally contain mercury), cement works and waste incinerators.
For example, the regulatory threshold in France is currently 50 ug/Nm? for waste incinerators, and this is regularly lowered as regulations are tightened.
There are several possible approaches to reducing mercury emissions.
First of all, it is possible to restrict the use of mercury by modifying production processes to make them mercury-free.
This approach is nevertheless not always possible.
Another approach is to use processes to treat mercury carrying liguid or gaseous streams.
The invention therefore concerns processes for the removal of mercury from combustion plant flue gases containing mercury vapour.
A distinction is made between wet and dry processes.
In wet processes, the flue gases are washed and the mercury and other pollutants contained in the flue gases are transferred to the liquid phase, where they are then precipitated, stabilised and separated.
In dry processes, involved in the invention, the mercury is captured by a demercurisation reagent, such as adsorbents, typically activated carbon, lignite coke, halogenated additives and sulphur additives, to mention just a few examples.
This demercurisation reagent, which is generally in the form of a powder, is injected into the flue gas to be treated and fixes the mercury present in the gaseous phase, in particular by adsorption, before then being separated from the flue gas by a gas-solid separator, such as bag filters or electrofilters.
For example, WO2016/132894 discloses a process for the demercurisation of flue gases arising from an incinerator and which pass through a boiler.
The flue gases to be treated are sent to a gas-solid separator upstream of which a demercurisation reagent is introduced into the flue gases.
An analyser measures the mercury concentration in the flue gases upstream of the point where the demercurisation reagent is introduced into the flue gases to be treated.
The concentration of mercury measured in this way is used to dose the quantity of demercurisation reagent introduced into the flue gases to be treated,
by using a dosing function of which different forms are considered in WO2016/132894, these different forms of the dosing function increasing in relation to the concentration of mercury measured.
Dry demercurisation processes are effective, but can be put to the test if the mercury concentration in the flue gases entering the corresponding treatment plant exhibits a mercury spike, i.e. varies very rapidly and with a large amplitude.
For example, in waste incinerators in France, the normal concentrations of mercury in the flue gases to be treated are generally of the order of 100 ug/Nm?, whereas this concentration can rise suddenly, in the space of a few minutes, to values well in excess of 1000 ug/Nm? during
— the combustion of a waste product with a high mercury content.
These mercury peaks can occur freguently, giving a comb-like profile to the mercury concentration signal at the flue gas treatment plant inlet.
In these conditions, a calibrated process can have real difficulty in reacting guickly enough to treat the mercury entering the flue gas treatment plant.
If the process does not react guickly enough or is not suitable, this can lead to mercury emissions in the flue gases downstream of the flue gas treatment plant exceeding the regulatory limits, but also to excess consumption of the demercurisation reagent.
Another aspect, often overlooked, in connection with mercury peaks is that the dwell time of mercury in the flue gas treatment plant may be prolonged due to the diffuse release of mercury in the flue gases circulating in the plant.
In fact, during a mercury peak, the passage of a high concentration of mercury through the flue gas treatment plant, even for a short time, tends to contaminate the plant's ducts and equipment, as the mercury is easily adsorbed onto the walls of these ducts and equipment, particularly below 200 °C.
Once the mercury peak has passed, the mercury slowly desorbs from the walls and is thus released at a later period than the mercury peak,
spreading mercury emissions throughout the flue gas treatment plant.
The mercury concentration profile is thus spread out temporally along the flue gas treatment plant.
In other words, there is a memory effect comparable to the known memory effect for dioxins.
This memory effect is all the more significant as the treatment chain that the flue gases pass through in the plant before being discharged into the environment is long and complex.
In this respect, the catalytic denitrification units of the flue gas treatment plant, placed downstream of the gas-solid separator, are particularly sensitive to this phenomenon because the materials used in these catalytic denitrification units are capable of adsorbing mercury, reacting chemically with mercury, generally by oxidation, and storing a significant quantity of mercury, before releasing it again at a later stage, for various reasons such as an excess of acid pollutants in the flue gases or a rise in temperature.
Furthermore, the diffuse release of mercury can also result from the fact that, after a mercury peak has passed, a large proportion of the mercury becomes trapped in the solids collected by the gas-solid separator, these collected solids forming a filter cake in the separator.
Once the mercury peak has passed, this filter cake is likely to release the mercury it contains into the flue gases passing through the filter cake, the mercury then being re-emitted into the flue gases leaving the separator, while the mercury content in the flue gases entering the flue gas treatment plant will have fallen to an ordinary value, for example of the order of 100 ug/Nm?. This may be accentuated by the fact that mercury adsorption occurs by physisorption when the portion of elemental mercury is considerable and the demercurisation reagent is not chemically impregnated.
In addition, in the case where part of the solids collected by the gas-solid separator is recycled in the flue gases to be treated and is thus found upstream of the gas-solid separator, the mercury contained in this recycled part of the solids risks passing back into the gaseous phase in the flue gases, thus extending the duration of the diffuse release of mercury in the flue gases circulating in the flue gas treatment plant.
In any case, this diffuse release of mercury, which is thus spread out over time, is not only difficult to detect and quantify, but can also cause problems, on the one hand, for the emission control system at the flue gas treatment plant outlet and, on the other hand, for the plants downstream units, such as a catalytic denitrification unit.
The aim of this invention is to propose a flue gas demercurisation process which is improved to take account of peaks in mercury concentration in the flue gases.
To this end, the invention relates to a flue gas demercurisation process, as defined in claim 1.
One of the ideas behind the invention is to control the dosage of the demercurisation reagent on the basis of the mercury concentration in the flue gases upstream of the point where the reagent is introduced into the flue gases to be treated.
Each time there is a mercury peak, i.e. each time there is a sudden and significant increase in the concentration of mercury in the flue gases entering a plant using the process, it is then possible to react very quickly by increasing the dosage of the demercurisation reagent.
In addition, the invention cleverly provides for avoiding a sudden reduction in the dosage of the demercurisation reagent and, to this end, takes account of the quantity of mercury produced by the mercury peak, so as to slow down the reduction in the dosage of the demercurisation reagent over time once the mercury peak has passed.
In particular, when the quantity of mercury provided by the peak is moderate, which is for example the case with an intense but short peak, the dosage of the demercurisation reagent is thus controlled so as to be able to return rapidly to a “base”
value, directly linked to the ordinary level, i.e. outside the mercury peak, of the mercury concentration upstream of the point of introduction of the demercurisation reagent into the flue gases to be treated.
In this first case, mercury emissions in the flue gas downstream of the flue gas treatment plant are effectively controlled, without over-consumption of the demercurisation reagent.
On the other hand, when the quantity of mercury contributed by the peak or several closely spaced peaks is high, which is for example the case for a long and significant peak, the dosage of the demercurisation reagent is thus controlled to remain at a high value for a longer period of time than in the first case.
In this way, in this second case, mercury emissions in the flue gases downstream of the flue gas treatment plant are effectively controlled, by neutralising the diffuse release of mercury explained above.
The demercurisation process according to the invention is thus particularly ingenious, by controlling mercury emissions in the treated flue gas, in particular so as to keep these mercury emissions below regulatory thresholds even in the event of mercury peaks.
According to advantageous specific features and implementation options, which — will be detailed later, the efficiency and practicality of this process can be further enhanced.
Additional advantageous features of the process according to the invention are specified in the other claims.
The invention will be better understood upon reading the following description,
given only as an example, and with reference to the drawings, in which:
[fig. 1] fig. 1 is a diagram of a flue gas treatment plant implementing a demercurisation process in accordance with the invention;
[fig. 2] figure 2 is a graph illustrating the mercury concentration in the flue gases to be treated, in relation to the time;
[fig. 3] figure 3 is a sequence of calculations carried out by the process in accordance with the invention; and
[fig. 4] figure 4 is a graph illustrating the variation over time of three different variables, detailed below.
Figure 1 shows a flue gas treatment plant 1. For example this plant 1 is provided downstream of a household waste incinerator (not shown), from which the combustion flue gases are sent to the inlet of plant 1.
The fumes sent to plant 1 contain mercury and, generally, other pollutants such as acid gases (HCI, SO»), nitrogen oxides, dust and heavy metals other than mercury.
The mercury contained in the flue gases to be treated may be present in various forms, namely elemental, oxidised and particulate.
The plant 1 includes a gas-solid separator 20, which is a known technology.
This gas-solid separator 20 is, for example, a bag filter or an electrofilter.
In practice, the gas- solid separator 20 typically operates at a temperature of between 120 °C and 250 °C, i.e. the temperature of the flue gases entering the gas-solid separator 20 is within the aforementioned range.
5 The installation 1 also includes a silo 10 containing a reagent for neutralising acid gases, such as lime or sodium bicarbonate.
This acid gas neutralisation reagent is introduced from the silo 10 into the flue gases to be treated upstream of the gas-solid separator 20, in order to neutralise the acid gases present in the flue gases to be treated.
In a manner known in the art and not limiting the invention, the flow rate of the acid gas neutralisation reagent is regulated in terms of the pollutant composition in the flue gases to be treated.
Installation 1 also includes a silo 11 containing a demercurisation reagent, such as those listed in the introductory part of this document.
This demercurisation reagent is introduced from the silo 11 into the flue gases to be treated upstream of the gas-solid separator 20, in order to capture the mercury present in the flue gases to be treated.
The gas-solid separator 20 is adapted to separate the dust and solid reagents, injected into the flue gases to be treated from the silos 10 and 11, which the flue gases contain when they pass through the gas-solid separator.
At least some, if not all, of the solids collected by the gas-solid separator 20 are intended to be discharged from the plant
1, while, if necessary, the rest of these solids are recycled, by being sent into the flue gases to be treated upstream of the gas-solid separator 20.
The plant 1 can optionally be supplemented by other treatment units, placed downstream of the gas-solid separator 20, such as a catalytic denitrification unit, for example.
In addition, other equipment such as a fan, a heat exchanger, a reactor or a silo for storing another reagent may be present within plant 1. In all cases, the flue gases treated by plant 1 are intended to leave the plant via a chimney 30 through which the treated flue gases are discharged into the environment.
Before looking at the other components of plant 1, it is interesting to understand how the concentration of mercury in the flue gases to be treated, i.e. the flue gases entering plant 1, can change over time.
Figure 2 illustrates the concentration, as a function of time, of total mercury vapour (elemental and oxidised) in flue gases from a household waste incinerator that are likely to be sent to plant 1 for treatment, in particular for demercurisation.
In this actual example, illustrated in Figure 2, it can be seen that the mercury concentration is most often around a value of the order of 100 ug/Nm? but that on three occasions there are mercury peaks during which the mercury concentration varies rapidly and with great intensity, reaching more than 800 or even 1,000 ug/Nm? in just a few seconds. Variations in other pollutants, such as acid gases and nitrogen oxides, do not generally show such abrupt variations. Returning now to the description of plant 1 in Figure 1, we will look at the way in which plant 1 is designed to regulate mercury emissions leaving plant 1, while taking account of the temporal evolution of the mercury concentration in the flue gases to be treated entering the plant, including when this temporal evolution includes one or more mercury peaks. Without such regulation, mercury concentrations in the treated flue gases could vary by more than two orders of magnitude in a few tens of seconds. To this end, in the configuration shown in Figure 1, the plant 1 comprises two analysers 40 and 50, and a computer 60. The analyser 40 is adapted to determine, in real time, the concentration of mercury in the flue gases upstream of the point of introduction of the demercurisation reagent into these flue gases, and to transmit an information signal iCa representative of the value of this concentration, this concentration value being referred to hereinafter as Ca. It is not necessary for the information signal iCa emitted by the analyser 40 to be strictly continuous, in the sense that the concentration values Ca determined by the analyser 40 are to be supplied, via the information signal iCa, at each moment, i.e. at each repetition of a predetermined time frequency, preferably between thirty seconds and sixty seconds. In practice, the analyser 40 is sensitive to one or more forms of mercury present in the flue gases to be treated, so that the information signal iCa gives information about the concentration of total mercury or of a form of mercury contained in the flue gases to be treated. It is therefore not essential to have knowledge of the speciation of the mercury upstream of the point of introduction of the demercurisation reagent into the flue gases to be treated, since the information signal iCa is intended, as will be explained below, to be used to monitor the temporal evolution of the concentration of mercury in the flue gases to be treated, it being noted that the different forms of mercury generally experience the same temporal evolution of their concentration. The analyser 50 is adapted to determine, in real time, the concentration of mercury in the fumes downstream of the gas-solid separator 20, and to transmit an information signal iCb representative of the value of this concentration, this concentration value hereinafter being referred to as Cb. This analyser 50 is preferably placed in the chimney
30. Similarly to the analyser 40, it is not necessary for the information signal iCb emitted by the analyser 50 to be strictly continuous, in the sense that the concentration values Cb determined by the analyser 50 are to be supplied, via the information signal iCb, at each moment, i.e. at each repetition of a predetermined time frequency, preferably about three minutes. In practice, the refresh rate of the analyser 50 for determining the concentration value Cb and for transmitting the corresponding information signal iCb is lower than that of the analyser 40 because, as will be explained in more detail below, the information signal iCb emitted by the analyser 50 is intended to act as a correction feedback, commonly known as “feedback”, for controlling mercury emissions in the flue gases leaving the installation 1.
The analysers 40 and 50 are based on technology known in the art, which does not limit the invention.
The computer 60 is adapted to control in real time the dosage of the demercurisation reagent introduced from the silo 11 into the flue gases upstream of the
— point of introduction of this demercurisation reagent into the fumes to be treated.
To this end, the computer 60 is designed to perform a sequence of calculations continuously, i.e. repeatedly at a predetermined time frequency, which will be described in more detail below.
In addition, the computer 60 is connected to the analysers 40 and 50 by any link enabling the information signals iCa and iCb to be transmitted from the analysers 40 and
— 50 to the computer 60. The computer 60 is also connected to a device 12 for regulating the flow of demercurisation reagent leaving the silo 11, this metering device 12 being, for example, a solenoid valve.
Here again, it does not matter how the link between the computer 60 and the dosing device 12 is implemented, as long as it enables a remote signal s to be transmitted from the computer 60 to the dosing device 12, which remote signal s is representative of the result of the calculation sequence performed by the computer 60. The calculator 60 is also connected to another device, not shown in the figures, supplying in real time an information signal iD which is sent to the calculator 60 and which is representative of the value of the flow rate of the fumes to be treated, this flow rate value being henceforth referred to as D.
This information signal iD comes, for example, from a flue gas flow rate measuring device placed upstream of the gas-solid separator 20, or from a flue gas flow rate measuring device placed at the stack 30, or from a steam production measuring device, or even from a system for indirectly evaluating steam production since this is directly correlated to the quantity of fumes treated by the installation 1. More generally, the information signal iD supplies data to the computer 60 enabling the latter to determine, in particular to calculate, at each instant the flow rate value D representative of the flow rate of fumes to be treated.
The type of computer 60 is not limitative as long as this computer 60 is able, in real time, to process, in particular digitally, the information signals iCa, iCb and iD by applying the aforementioned sequence of calculations, and to send the control signal s.
In this way,
the computer 60 can be an autonomous computing unit, in particular in the form of a programmable logic controller, or belong to a control system, in particular a computer system.
We will now describe in detail a process implemented by plant 1, by which the flue gases to be treated entering this installation are purified, in particular demercurised.
A corresponding example of the calculation sequence carried out by the computer 60 will be described in detail, this calculation sequence being illustrated in figure 3.
It is deemed that the plant 1 is in steady state and that, after the acid gas neutralisation reagent and the demercurisation reagent have been introduced into the flue gases to be treated upstream of the gas-solid separator 20 from silo 10 and silo 11 respectively, the flue gases to be treated are sent to the gas-solid separator 20. The separation effected by the latter leads to the formation of a filter cake in the gas-solid separator 20, attached, for example, to the filtering sleeves of this separator.
At each instant of this steady state, i.e. at each repetition of a predetermined time frequency, for example between 30 seconds and 2 minutes and corresponding here to the frequency of execution of the calculation sequence by the computer 60, the demercurisation reagent is dosed by implementing the following operations.
By operation a), an upstream concentration value Ca is determined at each instant for the concentration of mercury in the flue gases upstream of the point at which the demercurisation reagent is introduced into the flue gases.
Here, this determination of the upstream concentration value Ca is carried out, in particular by direct measurement, by the analyser 40 and this upstream concentration value Ca is supplied to the computer 60 via the information signal iCa.
Also by operation a), a flow rate value D representative of the flow rate of the flue gases to be treated is determined at each instant.
In this case, this determination of the flow rate value D is carried out, in particular by calculation, by the computer 60 on the
— basis of the information signal iD.
By an operation b), a dosing factor, called K1, is calculated at each moment, in this case by the computer 60, from (i) the upstream concentration value Ca, supplied here to the computer 60 by the information signal iCa, and (ii) a function f, which is pre- established, being pre-recorded in the computer 60, and which increases in relation to the upstream concentration value Ca.
The dosage factor K1 is used to control a specific dosage of the demercurisation reagent and is expressed, for example, in grams per normal m? of flue gases.
By multiplying the dosage factor K1 by the flow rate value D, a flow rate can be calculated for the demercurisation reagent to be introduced into the flue gases to be treated.
As the function f increases in relation to the upstream concentration value Ca, it is understood that the dosing factor K1 is calculated at a low value when the flue gases to be treated contain little mercury and is calculated at a high value when the fumes to be treated contain a lot of mercury. The form of the function f is not restrictive of the invention. However, it is preferable for the function f to be strictly increasing between a minimum value Cmin and a maximum value Cmax of the upstream concentration value Ca, while being constant below the minimum value Cmin and also constant above the maximum value Cmax. This preferred form of the function f is illustrated schematically in Figure 3. As egually interesting alternatives, the function f is: - strictly increasing between a zero value of the upstream concentration value Ca and a maximum magnitude of the upstream concentration value Ca, being constant — beyond this maximum magnitude, or - step-like, i.e. constant over successive intervals of the upstream Ca concentration value. It should be noted that if the demercurisation process were limited to operations a) and b) described so far, it would be possible to dose the demercurisation reagent taking account of the concentration of mercury in the flue gases to be treated, but this dose would be compromised when a mercury peak occurs, due to the delayed release of mercury, as explained in the introductory part of this document. In particular, at the end of each mercury peak, there would be a risk of the function f controlling too rapid and/or premature a decrease in the dosage of the demercurisation reagent. The other operations in the — demercurisation process will overcome this difficulty, as explained below. In operation c), a quantity of mercury Q is calculated at each instant, in this case by the computer 60, by totalling, over a predetermined time period T ending at the moment in question t, the product of the upstream concentration value Ca and the flow rate value
D. In practice, the computer 60 calculates for this purpose, continuously, the total, over — the time period T which precedes the considered moment t, of the flow of mercury in the flue gases to be treated, this flow of mercury being calculated by multiplying the upstream concentration value Ca and the flow rate value D. This quantity of mercury Q is expressed in mass or molar units and represents the mercury that the flue gases to be treated have fed into the plant 1 over the time period T preceding the moment in question. It is understood that this quantity of mercury Q is used to estimate the degree of contamination of the gas-solid separator and other equipment in the plant 1, located downstream of the gas-solid separator 20. The time period T is predetermined, in this case being pre- recorded in the computer 60, and has a value of between 15 minutes and 120 minutes, for example. Operation d) monitors the occurrence of a mercury peak on the basis of the upstream concentration value Ca. Each mercury peak is detected by the computer 60,
using ad hoc processing, known in the art, applied to the upstream concentration value
Ca. As a practical and reliable example, a Savitzky-Golay filter is applied to the upstream Ca concentration values N determined just before the moment in question, where N is greater than or equal to 6. As an alternative, a Savitzky-Golay filter is applied to the upstream Ca concentration values N determined just before the moment in question, where N is greater than or equal to 3. Other monitoring algorithms may be envisaged and applied to the N upstream concentration values Ca determined just before the moment in question, N being greater than or equal to 3: by way of non-limiting examples, the monitoring algorithm used is based on the arithmetic moving average, the weighted moving average or the exponential moving average, calculated for the N values mentioned above. Operation d) also provides that each time a mercury peak occurs, the moment t0 at which the peak occurs is recorded, together with the value KO of the dosage factor K1 at the moment t0. Here, each time a mercury peak is detected by the computer 60, the latter stores the moment tO and the value KO calculated for the dosage factor K1. In addition, at each moment following the moment t0, a dosage factor, which is called K2 and which is different from the dosage factor K1, is calculated, in this case by the computer 60, from the value KO and a function g, which is different from the function f and which is pre-established, in this instance being pre-recorded in the computer 60. The function g — decreases over time and is designed so that its decrease over time is slower the larger the guantity of mercury O calculated at the moment in guestion. In practice, and as illustrated schematically in figure 3, the dosage factor K2 can be obtained by multiplying the value KO by the result of the function g at the moment in guestion. In a similar way to the dosage factor K1, the dosage factor K2 makes it possible to control a specific dosage — of the neutralisation reagent and is thus expressed, for example, in kilograms per normal m? of flue gas. Unlike the dosage factor K1, the decrease in the dosage of the demercurisation reagent following a mercury peak is controlled by the dosage factor K2 so as to be slow. The form of the function g is not restrictive of the invention. However, according to a particular implementation illustrated schematically in Figure 3, the g function is of the form exp(-(t-t0)/Y) where t is the moment under consideration and Y is equal to the guantity of mercury O calculated at the moment under consideration, multiplied by a proportionality constant which may, for example, be chosen by an operator controlling the process, depending in particular on the specific features of the flue gases to be treated or — the desired purification performance. In other words, at each moment, Y is proportional to the guantity of mercury O calculated by operation c), as illustrated schematically in Figure
3, Y being calculated here by the computer 60. As equally interesting alternatives, the function g is: - of the form 1-((t-t0)/Y), or - of the form 1/(cosh((t-t0)/Y), where cosh is the hyperbolic cosine function.
The specific form for the g function, i.e. decreasing over time and with its decrease over time being all the slower as the quantity of mercury Q calculated at the moment under consideration is greater, enables both effective regulation of any magnitude of the upstream Ca concentration value during a mercury peak, and also optimisation of the consumption of the demercurisation reagent.
To better understand this aspect, it can be illustrated by two distinct cases.
In the first case, the mercury peak is intense but short, which means that the quantity of mercury Q supplied by the flue gases to be treated remains low even if the mercury concentration in these flue gases is high for a short time.
By operation b), the dosage factor K1 is raised to a high value and commands a high dosage for the demercurisation reagent.
Once the mercury peak has passed, it is not necessary for the dosage of the demercurisation reagent to decrease slowly and this dosage can therefore remain controlled by the dosage factor K1, the latter rapidly returning to an ordinary low value.
In this first case, it is therefore more advantageous to have the dosage factor K2 which decreases rapidly, for example exponentially, in relation to a low value of Y.
In a second case, the mercury peak is intense but longer than in the first case, or several mercury peaks occur in close succession.
The quantity of mercury Q and therefore Y are then high and a significant quantity of mercury enters the plant 1. Once the mercury peak has passed, a diffuse release of mercury into plant 1 occurs, as explained in the introductory part of this document.
To remedy this diffuse release of mercury, the dosage of the demercurisation reagent is maintained at a high value for a
— much longer period than in the first case, thanks to the g function which decreases the dosage factor K2 over a long period and with a small gradient since Y has a higher value than in the first case.
In operation e), at each moment, the calculated value of the dosing factor K1 and the calculated value of the dosing factor K2 are compared, in this case by the computer
— 60, retaining the greater of these two calculated values, called Kf.
In other words, the retained value Kf is egual to the greater of the respective calculated values of the dosing factors K1 and K2. This retained value Kf is used to dose the demercurisation reagent.
In this instance, the computer 60 generates the control signal s from the retained value Kf.
By an optional but advantageous operation f), a downstream concentration value
— Cb is determined at each instant for the mercury concentration downstream of the gas-
solid separator 20, this determination of the downstream concentration value Cb being carried out in this instance, by direct measurement, by the analyser 50 and being supplied to the computer 60 via the information signal iCb.
In addition, before being used to dose the demercurisation reagent, the retained value Kf is corrected, as the adjusted value Kf, on the basis of the downstream concentration value Cb.
In this instance, the computer 60 generates the control signal s from the adjusted value Kf, as indicated schematically in figure 3.
In this way, the optional operation f) enables an adjustment feedback to be applied to the retained value Kf resulting from operation e). The correction applied by operation f) is known as such and can in particular be implemented by a PID algorithm (acronym for
“proportional, integral, derivative”), executed in this instance by the computer 60.
As an optional addition to the demercurisation process just detailed, it is planned to take advantage of the detection of each mercury peak to act on the filter cake present in the gas-solid separator 20. Indeed, as explained in the introductory part of this document, some of the mercury contributed by each mercury peak is collected and accumulates in the filter cake, with a risk of delayed release of this mercury.
Thus, a first supplement to the demercurisation process therefore consists of eliminating the filter cake as quickly as possible after each mercury peak.
To do this, it is advantageously provided that, after a predetermined period of time, of between 1 and 15 minutes for example, from the moment t0, a declogging sequence is forced to be applied to the gas-solid separator 20 so as to eliminate the filter cake.
Such a declogging sequence is known in the art and consists in particular of acting mechanically and/or pneumatically on the filter cake to separate it from the gas-solid separator unit 20 on which it is formed, for example to detach it from the sleeves of this separator.
A second addition to the demercurisation process consists of trying to prevent the
— mercury present in the filter cake from being released by the latter as a result of an excess concentration of acid gases in the fumes passing through the gas-solid separator 20. To do this, it is advantageously provided that, for a predetermined period of time, of between 1 and 15 minutes for example, from the moment t0, the acid gas neutralisation reagent introduced into the flue gases to be treated from the silo 10 is overdosed.
Example:
According to an operating example, plant 1 is used under the following conditions:
- the flow rate of the flue gases to be treated is approximately 150,000 Nm3/h;
- the function f is such that:
- if the upstream concentration value Ca is less than 40 ug/Nm?, K1 is
— returned by the function f, as being constant and equal to 80 mg/Nm?,
- if the upstream Ca concentration value is between 40 and 1,000 ug/Nm?, K1 is returned by the function f as the result, expressed in mg/Nm3, of the formula 80+720x(Ca-40)/960, and - if the upstream concentration value Ca is greater than 1,000 ug/Nm?, K1 is returned by the function f, as being constant and equal to 800 mg/Nm?, - the time period T is egual to 30 minutes; and - the function g is exp(-(t-t0)/Y) with Y taken to be proportional to the quantity of mercury O.
Figure 4 illustrates the operating example with the occurrence of three mercury peaks.
The mixed line curve in Figure 4, which is associated with the left-hand scale, is the time trend of the upstream concentration value Ca, in other words the time trend of the mercury concentration in the flue gases to be treated, determined by the analyser 40. This mixed line curve shows the three mercury peaks at times t=240 minutes, t=360 — minutes and t=420 minutes respectively.
The most intense mercury peak is associated with a value of 1,240 ug/Nm3 for the upstream Ca concentration value.
The dotted and solid curves, which are both associated with the right-hand scale in Figure 4, show the temporal evolution of the dosage factor K1 and the retained value Kf respectively.
A comparison of the dashed and solid lines shows that, during each of the — mercury peaks, the retained value Kf slows down over time compared with the dosage factor K1. In particular, when the first and second mercury peaks appear, the retained value Kf is controlled by the dosage factor K1, whereas, once each mercury peak reaches its maximum, as well as after each of these mercury peaks, the retained value Kf is controlled by the dosage factor K2, slowing down its temporal decrease compared to the dosage factor K1. Before the first mercury peak, and after approximately 200 minutes from the third mercury peak, the retained value Kf is controlled by the dosage factor K1, by controlling a minimum dosage for the demercurisation reagent.