FI20215973A1 - A method for thermal processing of biomass, a system for the same, and a method for teaching a data-driven model - Google Patents

Abstract

A method for thermal processing of biomass, the method comprising taking at least one image from a feedstock (FS) comprising solid biomass; feeding at least the feedstock (FS) to a plant (500) for thermal processing; determining primary information indicative of a heating value of the feedstock (FS) using a processing unit (CPU) and the at least one image; and based on the primary information controlling a flow of material fed into the plant (500) and/or a temperature or a content of the material fed into the plant (500). A system for the same. In addition, a method for teaching a data-driven model, the data-driven model being for the purpose of determining from at least one image primary information indicative of a heating value of the material shown in the at least one image.

Description

A method for thermal processing of biomass, a system for the same, and a method for teaching a data-driven model

Technical field

The invention relates to control of thermal processes for processing biomass.

The invention relates to image recognition. The invention relates to use of a data-driven model in context of control of thermal processes for processing biomass.

Background

Biomass is commonly thermally processed to produce energy and/or fuel.

Typically the thermal processing involves at least heating the biomass and oftentimes also at least partial oxidation. Examples of thermal processing of biomass include pyrolysis, torrefaction, gasification, steam explosion, and combustion. Thermal process may include a step of producing fuel and a step of using, e.g. burning, the fuel.

Oftentimes the amount of energy and/or fuel that is being produced from the biomass should be optimized for use. E.g. in a boiler, wherein the biomass is burnt, the load of the boiler indicates the amount of energy that should be produced from the biomass. The load may be determined e.g. by use of electricity, whereby the feed of fuel should be adjusted accordingly. As another example, when a biomass gasifier is used to produce product gas for a subsequent incinerator or kiln, such as a lime kiln, the throughput of the kiln

N indicates the amount of the fuel (i.e. the product gas of the gasifier) needed in

N the incinerator or the kiln.

S

8 30 However, the energy content of the biomass may vary significantly depending z on its guality and/or the type of the biomass. In general, biomass may originate

N from animals, plants, or fungi. Biomass originating from plants may include 5 hay, straw, husk, chaff, cobs, bagasse, wood, wood species, or wood parts. = On a coarse level, the wood species may be classified as being hardwood or softwood. On a finer level, different wood parts include bark, core wood (including heartwood and sapwood), needles, and twigs, to name a few. As for other factors defining “quality”, the biomass may include low-calorific impurities, such as sand or soil.

At present, the thermal process using the biomass is mainly controlled only reactively. |.e. as a reaction to an observation that the biomass feed is not sufficient (or is excessive) the feed is increased (or decreased). However, such reactive control poses problems for the thermal process, because typically the process reguires a certain amount of energy, and a shortage (or surplus) causes problems in the thermal process. These problems include increased emissions as well as insufficient production of energy/heat. The process may be manually controlled by an operator.

In this field, the document WO 2012/158113 discloses a method for on-site analysis of at least one guality parameter of a material from a torrefaction process. Therein, the absorbance of the torrefied material is analyzed by a camera NIR-camera, which is also used to generate a data set which is processed in a computer using a multivariate statistical method.

Summary

It has now been found how the thermal process can be controlled more proactively and/or automatically. In other words, it has been found that information indicative of a heating value of a feedstock comprising biomass, the feedstock being fed to the thermal processing, can be determined using an image or images taken from the feedstock and a processing unit running a model for the purpose. The model may be a data driven model. The model can

N be taught before use in order to obtain reasonably accurate results. The model

N can be refined (i.e. further taught) during use.

S

8 30 The more proactive control has a first benefit that the process can be controlled z already before receiving information that the energy/fuel production is

N insufficient or excessive. Thus, a constant or sufficient but not excessive level 5 of energy/fuel can be maintained. For example, if the content of sand of the = feedstock increases, and consequently a heating value decreases, the feed of the feedstock (or an auxiliary fuel) can be increased. Because of the proper level of energy/fuel output, the emissions also remain at a reasonably low level.

The more proactive control has a second benefit that the content of the feedstock may be varied according to the feedstock supply. For example, if there are both hardwood and softwood available on the plant, any one or both of them can be used according to the supply of these materials, and the feed of the material (e.g. a mixture of hardwood and softwood) can be controlled based on the content of the material fed to the thermal process. As another example in addition to wood, also biomass originating from other plants can be used, even if their heating value differs from that of wood. Typically the different feedstocks have different calorific values, whereby control is needed.

The method for thermal processing of biomass is disclosed in more specific terms in claim 1; and in the context of also teaching a model, also in claim 19.

A corresponding system is disclosed in claim 12. Teaching the model in use is disclosed in claims 11 and 21. Teaching the model before use is disclosed in claim 16. Verifying and teaching the model is disclosed in claim 18.

Other claims and the description and figures disclose other embodiments.

Brief description of the drawings

Fig. 1a shows a schematic of a method and a system for thermal processing of feedstock comprising biomass,

Fig. 1b shows a schematic of a method and a system for thermal processing of feedstock comprising biomass,

Fig. 1c shows a schematic of a method and a system for thermal processing of a mixture of materials, the mixture comprising

N feedstock comprising biomass,

N Fig. 1d shows a schematic of a method and a system for thermal

S processing of feedstock comprising biomass, the feedstock being a 3 30 mixture of materials, z Fig. 1e shows a schematic of a method and a system for thermal

N processing of feedstock comprising biomass, wherein a model is 5 accessed through an interface, = Fig. 2a shows a schematic of a method and a system for thermal processing of biomass, the thermal processing including gasification, wherein the gasification gas is used in a lime Kiln,

Fig. 2b shows a schematic of a method and a system for thermal processing of biomass, the thermal processing including gasification,

Fig. 2c shows a schematic of a method and a system for thermal processing of biomass, the thermal processing including combustion in a boiler,

Fig.2d shows a schematic of a method and a system for thermal processing of biomass, the thermal processing including pyrolysis,

Fig. 2e shows a schematic of a method and a system for thermal processing of biomass, the thermal processing including torrefaction,

Fig. 2f shows a schematic of a method and a system for thermal processing of biomass, the thermal processing including steam explosion,

Fig. 2g shows a schematic of a method and a system for thermal processing of biomass, the thermal processing including combustion in a kiln,

Fig. 3 shows a flowchart of a method for teaching a model,

Fig. 4 shows a flowchart of a method for testing the model, and optionally also teaching the model, and

Fig. 5 shows a flowchart of a method for using the model, preferably using the model after it has been taught as shown in Fig. 3.

Detailed description

Figure 1a shows thermal processing of biomass. Therein feedstock FS is fed

N to a plant 500 for thermal processing. The feedstock FS comprises solid

N biomass.

S

8 30 Herein the term “biomass” refers to a material or materials of biological origin. z Biomass may comprise virgin and waste materials of plant, animal and/or fish

N origin or microbiological origin, such as virgin wood, wood residues, forest 5 residues, waste, municipal waste, industrial waste or by-products, agricultural = waste or by-products, residues or by-products of the wood-processing industry, waste or by-products of the food industry, solid or semi-solid organic residues of anaerobic or aerobic digestion, such as residues from bio-gas production from lignocellulosic and/or municipal waste material, residues from bio-ethanol production process, and any combinations thereof. Suitably said biomass comprises by-products of agriculture and/or wood-processing industry such as urban wood waste, lumber waste, wood chips, wood waste, sawdust, firewood, wood materials, by-products of timber processes, where 5 the biomass (plant biomass) is composed of at least cellulose, hemicellulose and lignin. The method is particularly suitable when the feedstock comprises solid material originating from plants, such as agricultural residues and/or material originating from wood, e.g. wood chips, bark, needles, and/or twigs.

Referring to Figs. 2a to 2g the plant 500 may comprise at least one of a gasifier 510, a kiln (such as a lime kiln 520), a boiler 530, a pyrolysis reactor 540, a torrefaction reactor 550, or a steam explosion reactor 560.

A purpose of a gasifier 510 is to gasify the feedstock FS to produce product gas. Typically the product gas may be used as a fuel in a subsequent burner; e.g. in a kiln or a boiler. The product gas may be refined before use.

A purpose of a kiln is to thermally process some material. A purpose of a lime kiln 520 is to calcinate, i.e. thermally dissociate (or burn, as this process is commonly called), calcium carbonate (CaCQ:s). In calcination, carbon dioxide (CO2) is thermally dissociated from the calcium carbonate (CaCO3) and in this way calcium oxide (CaO), commonly known as guicklime or burnt lime, is produced. The feedstock FS or a thermally processed feedstock can be used as fuel in the kiln or the lime kiln 520.

A purpose of a boiler 530 is to burn the feedstock FS or a thermally processed

N feedstock to produce energy. The term “boiler” derives from boiling (water),

N and most often steam is produced in a boiler, and the steam is used in a turbine

S to generate mechanical energy, which may be used to drive a generator for 8 30 producing electricity. Most often at least hot flue gases are produced in the z boiler 530, and heat is recovered therefrom to heat exchange medium, such

N as water. Heat may be recovered from bed material, if the boiler 530 is of the

S fluidized bed type.

QA

A purpose of a pyrolysis reactor 540 is to pyrolyze the feedstock and in this way to produce pyrolytic vapor. The vapor may be condensed and, optionally, refined to produce bio-oil.

A purpose of a torrefaction reactor 550 is to torrefy the feedstock. Torrefaction is a heat treatment process for obtaining moisture resistant bio-based fuel having a high calorific value. Such fuel can be used (i.e. burned) in a kiln or a boiler 520, e.g. in a lime kiln 520, optionally with another fuel.

A purpose of a steam explosion reactor 560 is to produce steam-exploded biomass, most often for use as fuel. Therein the feedstock is pressurized and thermally treated with steam, and after lowering the pressure, steam exploded biomass is obtained. Steam-exploded biomass is easily grindable e.g. for purposes of co-combustion with coal. Such fuel can be used (i.e. burned) in a kiln or a boiler 520, e.g. in a lime kiln 520, optionally with another fuel.

As detailed in Fig. 2a, the plant 500 may comprise two of the more specific rectors disclosed above. For example, a gasifier 510 may be used to produce gas for a kiln or a boiler, e.g. the lime kiln 520. Even if not shown, a torrefaction reactor 550 or a steam explosion reactor 560 can be used to produce fuel for a kiln or a boiler, e.g. a lime kiln 520. Before burning in the kiln or the boiler, the torrefied or steam-exploded biomass (as received from the torrefaction reactor 550 or the steam explosion reactor 560, respectively) may be pre- processed, such as ground.

In general, a quality of the material of the feedstock FS, which is fed to the plant 500, may relate to a specific energy content of the feedstock FS. The specific energy content here refers to the chemical energy of the feedstock per unit mass or unit volume, e.g. in units on kJ/kg or MJ/m3. Thus, the term

N "quality" as used above and in the context of this specification means

N information that is indicative of a heating value of the feedstock. The term

S ‘heating value? is used interchangeably with the term specific energy content 8 30 (see above). =

N The quality may thus refer to a content or presence of low-calorific impurities 5 (such as sand and/or soil) or low-calorific biomass (such as hay, straw, = needles, bark, and/or twigs) of the feedstock FS or in the feedstock FS. The quality may refer (in the alternative or in addition) to the species of wood of the feedstock FS. In general, it has been found that the specific energy content of hardwood is higher than the specific energy content of softwood. It is noted that this applies not only to the core wood (i.e. heartwood or sapwood), but also bark of the wood. Bark originated from wood is commonly used for production of heat and/or fuel. The quality may refer (in the alternative or in addition) to a moisture content of the feedstock. Moisture decreases the heating value. It is noted that practically moisture is always present to some extent in biomass.

As for the terms “hardwood” and “softwood”, hardwood is wood from angiosperm trees or from monocotyledons. The list of angiosperm trees (hardwood) is wide, and includes e.g. alder, apple, aspen, birch, cherry, ebony, elm, eucalyptus, hickory, mahogany, maple, oak, rosewood, teak, walnut, and willow. The list of monocotyledons (hardwood) is less wide, including bamboo and coconut. In contrast, softwood comes from gymnosperm trees, also called conifers. Examples of softwood include cedar, pine, and spruce.

Thus, if the quality of the feedstock changes, e.g. as a result of the feedstock comprising more impurities and/or as a result of changing from hardwood- based feedstock to softwood-based feedstock (or vice versa), the specific energy content of the feedstock changes. However, in order to produce a suitable amount of fuel and/or energy in the plant, e.g. a constant amount of fuel and/or energy, e.g. the feed of the feedstock can be changed accordingly.

In the alternative, the content of the feedstock can be changed. E.g. if more energy is required, the content of the feedstock (or a content of a mixture of materials fed to the plant 500, the mixture comprising the feedstock) can be changed towards more hardwood-rich composition. Moreover, a feed of air to the plant 500 and/or a temperature of air fed to the plant 500 can be changed.

N

N It has been found that the guality of the feedstock FS, i.e. primary information

S indicative of the heating value of the feedstock FS, can be determined using 8 30 images and image recognition techniques. For example, this may be done z using a data-driven model that has been taught beforehand. Teaching of the

N model will be discussed below. Such a model can be used to provide primary 5 information indicative of the heating value of the feedstock FS when the model = is provided with an image showing the content of the feedstock FS. & 35

Thus, by taking at least an image of the feedstock FS, the quality of the feedstock as defined above can be determined, and the process controlled e.g. as detailed above. Thus, an embodiment of a method for thermal processing of biomass comprises taking at least one image from the feedstock

FS comprising solid biomass. In Figs. 1a to 1e the at least one image is taken with a camera 110. The method comprises feeding at least the feedstock FS to a plant 500 for thermal processing. Also the feedstock FS is thermally processed in the plant 500, optionally with other materials, such as a secondary feedstock FS2 (see Fig. 1c) and/or oxidizing gas, such as air (see

Fig. 1a and/or other gas, such as steam (see Fig. 2f). In addition to steam explosion, also pyrolysis (e.g. pyrolysis of Fig. 2d) may require steam to be fed to the process, e.g. for purposes of refining the pyrolytic vapor.

Different types of components of the plant 500 were discussed above. In addition to the feedstock FS, some other material, such as a secondary feedstock FS2, can be fed to the plant 500, as indicated in Fig. 1c. The secondary feedstock FS2 may comprise high-calorific fuel, which can be used to as a supply if the heating value of the feedstock FS drops too much. The high-calorific fuel may be solid, liquid, or gaseous.

The method comprises determining, using a processing unit CPU and the at least one image, primary information indicative of a heating value of the feedstock FS. In an embodiment, the method comprises using the at least one image and a data-driven model on the processing unit CPU to determine the primary information. In addition, the method comprises controlling a flow of material into the plant 500 and/or a temperature or a content of the material of the flow into the plant 500 based on the primary information. It should be noted that the step of “controlling” can be done by an operator or automatically. In

N case an operator undertakes the step of controlling, the method comprises

N transmitting the primary information to the operator. The information may be

S transmitted e.g. by showing it on a screen, as an audial signal to the operator, 8 30 or by using data transfer means to send it to a device used by the operator. z The operator may use a control unit (i.e. controlling apparatus, optionally in

N connection with a controller) for the purpose of controlling. In case the step of 5 controlling is automated, the processing unit CPU may send a signal to a = controlling apparatus directly or to a controller configured to operate a controlling apparatus for the purpose of controlling. More details will follow.

Herein the flow of material into the plant 500 refers to a mass flow (e.g. in units of kg/h) or a volumetric flow (e.g. in units of m3/h) of a material into the plant 500. The material may be the feedstock FS (see Figs. 1a, 1b, 1d), a composition comprising the feedstock FS (see Fig. 1c), or gas, such as air (see

Figs. 2a, 2b, 2c, 2d, 2e, 29) or steam (see Fig. 2f), or heat exchange medium e.g. used for heating/cooling the gas (see Fig. 1c). In case of combustion, the air may be combustion air. In case of a fluidized bed type reactor, the air may be fluidizing air.

The flow of the material fed to the plant may be controlled by controlling e.g. a conveyor 120 configured to convey the feedstock to the plant 500, as shown in Figs. 1a, 1b, and 1c. Alternatively or in addition, the flow of the material fed to the plant may be controlled by controlling e.g. a secondary conveyor 122 configured to convey secondary feedstock to the plant 500, as shown in Figs. 1c and 1d. Alternatively or in addition, the flow of the material fed to the plant may be controlled by controlling e.g. a primary conveyor 121 configured to convey a primary part FS1 of the feedstock FS, as shown in Fig. 1d.

Alternatively or in addition, the flow of the material fed to the plant may be controlled by controlling e.g. a baffle, a valve, a fan, or a pump that is used to control the flow of the gas (e.g. air or steam) fed to the plant 500, see Figs. 1a and 1e. Alternatively or in addition, the flow of the material fed to the plant may be controlled by controlling e.g. a baffle, a valve, a fan or a pump that is used to control the flow of the heat exchange medium to the plant 500 (Fig. 1c); e.g. a heat exchange medium flowing in the plant 500, such as feedwater of a boiler. It is noted that even if wood material WWM is shown in Figs. 1ato 1e, as detailed above, the feedstock may comprise or consist of other types of solid

N biomass, e.g. solid biomass originating from plants.

N

S The temperature of the material fed to the plant may be controlled using e.g. a 8 30 heat exchanger, as shown in Fig. 1c. For example, a flow of the heat exchange z medium can be controlled. In the alternative, e.g. a heater or a cooler can be

N used for the purpose. The flow of the material fed to the plant may be controlled 5 by controlling e.g. a baffle, a valve, afan or a pump that is used to control the = flow of the heat exchange medium to the plant 500; e.g. the heat exchange medium configured to exchange heat with the air that is configured to oxidise the biomass of the feedstock FS in the plant 500.

The content of the material fed to the plant 500 may be controlled by controlling e.g. at least one of the conveyors conveying the material to the plant. E.g. in

Figs. 1a and 1b, the feedstock FS is simply conveyed to the plant 500. E g. in

Fig. 1c, secondary feedstock FS2 is added to the feedstock FS before feeding the feedstock FS to the plant 500. When the composition of the feedstock FS differs from the secondary feedstock FS2, suitable composition can be achieved by controlling the conveyor 122 and/or the conveyor 120. In somewhat similar manner, in Fig. 1d, the feedstock FS is composed of a primary part FS1 of the feedstock FS and a secondary part FS2 of the feedstock. When the composition of the primary part FS1 differs from the composition of the secondary part FS2, suitable composition for the feedstock

FS can be achieved by controlling the conveyor 122 and/or the conveyor 121.

However, as can be seen from Figs. 1c and 1d, depending on the case, a most suitably location for the camera 110 may selected. Thus, the images may be taken from a feedstock FS comprising a mixture (as in Fig. 1d) of a primary feedstock FS1 and a secondary feedstock FS2, as in Fig. 1d; or the images may be taken from the feedstock FS that is fed to the boiler in a mixture “Mix” comprising also a secondary feedstock FS2, as in Fig. 1c.

To this end, an embodiment comprises controlling, based on the primary information indicative of the heating value of the feedstock FS, at least one of: flow of the feedstock FS into the plant 500; a content of the material (e.g. “Mix”) fed to the plant for thermal processing, the material comprising the feedstock

FS, and optionally secondary feedstock FS2; flow of gas, such as air or steam, into the plant 500; and a temperature of a gas, such as air or steam, fed to the plant 500.

N

N A preferable embodiment comprises (a) controlling a flow of the feedstock FS

S into the plant 500 based on the primary information and/or (b) controlling, 8 30 based on the primary information, a flow into the plant 500 of such gas, e.g. z air, that is configured to oxidise the feedstock FS in the plant 500. The gas that

N is configured to oxidize the feedstock may be air, and may be e.g. combustion

S air or gasification air.

N

N 35 Referring to Figs. 1c and 1d, when two different types of material streams are used, the feedstock FS that is imaged, and of which quality is determined, may be imaged before mixing the two different types of material streams (as in Fig. 1c) or after mixing the two different types of material streams (as in Fig. 1d).

Concerning the determining, using the processing unit CPU and the at least one image, the primary information indicative of a heating value of the feedstock, e.g. neural networks and self-organizing maps can be used. The self-organizing maps provide for automatic classification of the at least one images, which may be employed for producing the information indicative of the heating value.

In a preferable embodiment, the feedstock FS comprises solid material originating from plants, such as solid material (WM) originating from wood. The feasibility of the model is due to the large variation of quality of the feedstock in such a case. As for the meaning of “quality”, what has been said above, applies.

The method provides better control of power of the plant 500, and a better control of emissions coming out of the plant 500. The information indicative of the heating value of the feedstock FS can be assessed online, without a significant delay. In contrast, e.g. laboratory analysis would indicate significant delays. Furthermore, the measurement technology is cheap and safe. Since there is no contact between the measurement device and the feedstock FS, maintenance needs of the measuring device are minor. Furthermore, it is possible to install the measurement device to an existing feedstock feeding system.

N As for the information indicative of the heating value of the feedstock FS, there

N are several factors affecting the heating value. Thus, in an embodiment, the

S information indicative of the heating value of the feedstock FS comprises 8 30 information on at least one of: a content of sand of the feedstock or a presence z of sand in the feedstock (sand has a negligible heating value), a content or

N presence of soil of/in the feedstock (soil has a negligible heating value), a 5 content or presence of bark of/in the feedstock (bark has a lower heating value = than core wood), a content or presence of agricultural residue of/in the feedstock (the agricultural residue has a lower heating value than core wood), a content or presence of/in wood chips of the feedstock (wood chips have a higher heating value than bark), a content or presence of material originating from hardwood of/in the feedstock (hardwood has a higher heating value than softwood; e.g. content or presence of hardwood chips), a content or presence of material originating from softwood of/in the feedstock (e.g. content or presence of softwood chips), a content or presence of needles of/in the feedstock (needles have a low heating value), a content or presence of twigs of/in the feedstock (twigs have a low heating value), a content of moisture of the feedstock (moisture decreases the heating value), and the heating value of the feedstock. As for the term “presence”, the presence is a binary value indicating that such material is present or is not present; in contrast to the content, which indicates an amount of how much such material is present, if any. The term agricultural residue covers e.g. hay, straw, husk, chaff, cobs, and bagasse. Specific information indicative of the heating value of the feedstock FS will be detailed in connection with teaching the model for, on one hand, the embodiment, wherein the feedstock comprises solid biomass originating from wood, and on the one hand, the embodiment, wherein the feedstock comprises solid biomass originating from other plants than wood.

As readable from the above, the thermal process can be controlled based on the primary information, even if the primary information does not comprise (an estimate of) the heating value of the feedstock. However, preferably the method comprises determining, using the processing unit, the heating value of the feedstock FS.

A preferable embodiment comprises feeding oxygen-containing gas to the plant 500. A purpose of the gas is to oxidise (e.g. burn) the feedstock FS. The oxygen-containing gas may comprise air. Air may be used as the oxygen-

N containing gas. This has the effect that the process is more easily controllable.

N In particular, an air feed may be controlled e.g. in addition to controlling the

S amount or content of the material to be thermally processed (i.e. the feedstock

S 30 FS, optionally in combination with a secondary feedstock FS2). =

N Components of the plant 500 have been discussed above. Thus, in an 5 embodiment the feedstock FS is fed to a gasifier 510; to a kiln (e.g. a lime kiln = 520); to a boiler 530; to a pyrolysis reactor 540; to a torrefaction reactor 550, or to a steam explosion reactor 560.

Preferably, the plant 500 comprises a lime kiln 520. The feedstock may be fed to the lime kiln 520 as such, as indicated in Fig. 2g. However, preferably the feedstock is pre-treated in a steam-explosion reactor 560, in a torrefaction reactor 550 or in a gasifier 510, most preferably in a gasifier 510, as shown in

Fig. 2a. Thus, in an embodiment, the feedstock FS is fed to the gasifier 510, the torrefaction reactor 550, or the steam explosion reactor 560; and a product of the gasifier 510, the torrefaction reactor 550, or the steam explosion reactor 560 is used as fuel in the lime kiln 520. As for the lime kiln 520, in use, calcium carbonate CaCO3 is fed to the lime kiln 520 and calcinated therein, as detailed above.

With reference to Fig. 2a, more preferably, the feedstock FS is fed to the gasifier 510; and a product gas PG of the gasifier 510 is used as fuel gas FG in the lime kiln 520. Preferably, the gasifier 510 is a fluidized bed gasifier 510.

This also improves the process control, as fluidizing gas is material flowing to the plant 500 and a flow thereof may be controlled as indicated above.

Naturally, the feedstock may be pre-treated in a steam-explosion reactor 560, in a torrefaction reactor 550 or in a gasifier 510, and the a product of the gasifier 510, the torrefaction reactor 550, or the steam explosion reactor 560 may be used as fuel in a boiler or any kiln (also other kiln than a lime kiln 520).

In an embodiment, the at least one image of the feedstock FS (or sample, see below) is taken using a camera 110. However, the camera 110 may be arranged inside a housing, e.g. inside a housing of a conveyor. In such a case, and also at night, natural light may not be sufficient for the purposes of imaging

N the feedstock FS. Thus, an embodiment comprises illuminating the feedstock

N FS at the time of taking the at least one image. A high-speed camera may be

S used. A high-speed camera herein refers to a camera that is capable of taking 8 30 images at a rate of at least 10 frames (i.e. images) per second. The region z from which the images are taken may be illuminated using a led flash light.

N The camera and/or its casing, optionally provided with a window for taking the 5 images through the window, may be cleaned from time to time. A system may = comprise means for cleaning the camera and/or its casing. Such means may comprise a washer or a wiper. llluminating the feedstock FS at the time of taking the at least one image has the additional technical effect that the lighting conditions can be kept constant. This improves accuracy of the model. In particular, if the same lighting conditions are used during teaching the model and during use of the model, the accuracy of the results improves. Naturally, the same lighting settings, including the aperture, the shutter speed, the ISO sensitiveness setting, and the white-balance, may be used in the camera if applicable. However, some cameras, e.g. video cameras adjust these automatically.

Referring to Figs. 1a to 1d, an embodiment comprises conveying the feedstock

FS towards the plant 500 and/or to the plant 500 using a conveyor 120. The conveyor is preferably a belt conveyor. The embodiment comprises taking the at least one image from the feedstock FS such that the at least one image is taken from the from a part of the feedstock FS when that part of the feedstock

FS is arranged in the conveyor 120 or on the conveyor 120. This has the effect that typically the feedstock FS is, in or on the conveyor 120, arranged as a relatively thin layer of material. Therefore, information concerning the quality of the feedstock can be obtained from most parts of the feedstock FS.

The conveyor 120 may be a part of a dryer. |.e. the feedstock may be dried on the conveyor 120. However, also more generally, the feedstock FS may be dried before feeding it to the plant 500. Thus, an embodiment comprises drying the feedstock FS before feeding to the plant 500. If the feedstock FS is dried, preferably, the at least one image from the feedstock FS is taken after the feedstock FS has been dried. It has been found that the information indicative of the heating value, in particular the heating value itself, can be determined more accurately from dried feedstock than from undried feedstock.

This applies in particular when the model is taught using the samples (see

N below). Drying is also beneficial, when the feedstock FS comprises material

N originating from plants, e.g. material WM originating from wood. Drying the

S material before taken the images (of the samples and of the feedstock) 8 30 improves the accuracy, since then the moisture does not affect the images, at

E least not as much as before drying.

O

5 For example, in the embodiments of Figs. 1a and 1b, the conveyor 120 may = be part of a drier. As indicated therein, the camera 110 is arranged at the end of the conveyor, whereby the camera 110 is arranged to take images from dried feedstock FS. A direction of the feedstock FS on the conveyor 120 is depicted with an arrow in these Figs. When the model is taught, a sample may be defined and the image(s) taken there thereof, as detailed above. Moreover, the information that is associated with the image(s) is determined by other means. Further details will be given below.

While the moisture content of the feedstock may have a smaller effect on the quality than some other factor, moisture content has a role in the heating value.

Thus, an embodiment comprises determining a moisture content of the feedstock FS and controlling the flow of material into the plant and/or a temperature or content of the material of the flow into the plant 500 using also the moisture content of the feedstock FS. As readable from above, the moisture content of the feedstock may be determinable (and determined) from the images. However, it may be that the estimate for the moisture content, as determined solely from images, is inaccurate. Therefore, the moisture content may be measured by other means. In an embodiment, the moisture content of the feedstock FS is determined using other measurements than the images.

For example, it has been found that microwave technology can be used to continuously measure biomass moisture content, e.g. chip or bark moisture content. For example a resonance of the microwaves may be indicative of the moisture content. In addition or alternatively, radiometric moisture measurement or thermogravimetric analysis can be used. The measurements can be made on-line.

The primary information, which is indicative of a heating value of the feedstock

FS, is preferably determined using the at least one image and a data-driven model, which is run on the processing unit CPU. Using a data-driven model has the benefit that in general, an accuracy of the data driven model can be

N improved in use. This is commonly called teaching the model. Teaching

N requires reference information, i.e. information concerning the accuracy of the

S result of the model, and if the accuracy is insufficient, whether the estimate is 8 30 too low or too high. =

N Thus, in an embodiment, where such a data-driven model is used, the method 5 comprises determining a power of the plant 500. The power indicates how = much power is produced in the plant. As the flow of the processable material (i.e. the feedstock FS or the mixture “Mix” comprising the feedstock) to the plant is known, a heating value of the processable material can be determined from the power of the plant 500. Moreover, if only the feedstock FS is used as the processable material, then the heating value of the feedstock FS is the heating value of the processable material. If both the feedstock FS and the secondary feedstock FS2 are used as the processable material, then the heating value of the feedstock FS can be determined from the heating value of the processable material, assuming that in addition to the flow of the processable material also at least one of a flow of the feedstock FS and a flow of the secondary feedstock FS2 is known. Thus, the embodiment comprises determining a heating value of the feedstock FS using the power of the plant 500, and teaching the data-driven model with the heating value as obtained from the power of the plant 500 and the at least one image of the feedstock. In this embodiment, the primary information indicative of a heating value of the feedstock FS comprises the heating value of the feedstock FS, because such information is used in the teaching.

The Figures 1a to 2g also show schematically embodiments of a system for thermal processing of biomass. The method, as discussed above, may be performed using the system. An embodiment of the system comprises a plant 500 for thermal processing of feedstock FS comprising solid biomass.

Preferably, the plant 500 is for thermal processing of feedstock FS comprising solid material originating from plants, e.g. solid material WM originating from wood, such as bark. As detailed above in connection with the method, the plant 500 is also for processing processable material (e.g. the mixture "Mix”) comprising the feedstock FS and a secondary feedstock FS2.

The system comprises an imaging means, such as a camera 110, configured to take at least one image from the feedstock FS. The system may comprise

N a processing unit CPU configured to determine primary information indicative

N of a heating value of the feedstock FS using the at least one image. Such

S embodiments are shown in Figs. 1a to 1d. However, with reference to Fig. 1e, 8 30 such a processing unit need not be comprised by the system. In the alternative, z the processing unit CPU may be accessed over a network via an interface. In

N such a case, the system comprises means for sending data indicative of the at 5 least one image | to an interface, and means for receiving primary information = indicative of a heating value of the feedstock FS from the interface. Then, the

N 35 interface is configured to receive the data indicative of the at least one image

I, determine primary information indicative of a heating value of the feedstock

FS using the data indicative of the at least one image, and arranging available for the plant 500 the primary information indicative of a heating value of the feedstock FS.

The system further comprises a controlling apparatus configured to control a flow of material into the plant and/or a temperature of the material of the flow into the plant. The controlling apparatus may be a valve, a baffle, a fan, or a pump for controlling the flow of gas (air or steam) or liquid (water or heat exchange medium) to the plant. The controlling apparatus may be a motor of a conveyor, e.g. a motor of the conveyor 120, 121, or 122.

The controlling apparatus may receive a control signal from the processing unit

CPU (see Figs. 1a to 1d). In the alternative, the system may comprise a controller (see Fig. 1e) for receiving primary information indicative of a heating value of the feedstock FS from the interface. Based on the primary information, the controller may generate a control signal for the controller apparatus.

Correspondingly, the controlling apparatus may receive a control signal from the controller (see Fig. 1e). As a further alternative, an operator may receive a signal from the processing unit CPU (be it part of the system or not), and the operator may control the controlling apparatus.

When the system comprises the processing unit CPU, the processing unit CPU may comprise the controller that is configured to send a control signal to the controlling apparatus configured to control the flow of material fed into the plant 500 and/or the temperature or the content of the material fed into the plant 500 based on the primary information. However, referring to Fig. 1e, when the processing unit CPU is accessed through an interface, the system may

N comprise a controller for generating the control signal.

N

S If the process control is automated, (i) the processing unit CPU and/or the 8 30 controller and (ii) the controlling apparatus are, in combination, configured to z control the flow of material into the plant 500 and/or the temperature or the

N content of the material of the flow into the plant 500 based on the primary 5 information. If the process control is done by an operator, the system = comprises (i) a transmitter for transmitting the primary information to an operator or (ii) means for sending data indicative of the at least one image to an interface. In the latter case, the operator may receive the primary information via the interface. Then, a controller (e.g. the controller of Fig. 1e)

is not needed. Even if the process control is automated, the primary information may be transmitted to the operator; and the system may comprise the transmitter. Examples of a transmitter have been indicated above.

Referring to Fig. 1e, the processing unit CPU preferably comprises a processor

P and a memory M. The processing unit CPU of also the other Figures comprises the processor P and the memory M. The memory stores computer program and/or computer instructions, which are read by the processor and executed by the processor. The execution of the computer instructions causes the processing unit CPU to determine, using the at least one image or the data indicative of the at least one image, at least the primary information indicative of a heating value of the feedstock FS.

As for the components and/or reactors of the plant 500, what has been said in connection with the method applies mutatis mutandis. In particular, in an embodiment, the plant 500 comprises a lime kiln 520, a gasifier 510, and a pipeline 530 for conveying product gas PG from the gasifier 510 to the lime kiln 520 to be used is fuel gas FG in the lime kiln 520, as detailed in Fig. 2a. In the more general case, the plant comprises a pipeline 530 or a conveyor for conveying a product of the gasifier 510, the torrefaction reactor 550, or the steam explosion reactor 560 to a kiln or a boiler to be used as fuel in the kiln or the boiler, such as the lime kiln 520.

Concerning the model for determining the primary information indicative of a heating value of the feedstock using the at least one image, an algorithm may be used. The algorithm may be based e.g. on neural network tools. The

N algorithm may be referred to as a data-driven model. As detailed above, before

N use, the data-driven model may be taught. For teaching of the model, several

S images are reguired, each one in connection with first information (i.e. 8 30 reference information) indicative of a heating value of the material of the image. z When teaching the model, a method for teaching a data-driven model is carried

N out, and the data-driven model is for the purpose of determining from at least 5 one image primary information indicative of a heating value of the material = shown in the at least one image. & 35

Referring to Fig. 3, an embodiment of the method for teaching the data-driven model comprises a step ‘a’: defining a sample of a feedstock comprising solid biomass. Preferably the feedstock comprises solid material originating from plants. The step ‘a’ may comprise e.g. stopping the conveyor 120 whereby a sample of the feedstock FS can be taken taking for purposes of determining the reference information (i.e. first information); in addition to taking an image of the sample off the feedstock. Typically images can be taken even if the conveyor is not stopped. However, physical samples can be taken, at least from some conveyors even when the conveyor is running. The step ‘a’ may comprise generating samples. For example the samples of Table 1 may be generated. Such samples may be e.g. clear cases in which only one type of material per sample is present. This may help the initial teaching.

The method comprises a step 'b': taking an image of the sample. The imaging means, such as the camera 110 may be used for the purpose as discussed above.

The method comprises a step 'c': determining first information indicative of a heating value of the material of the sample. In particular, when initially teaching the data-driven model, the first information is not determined by the model from the image(s) of the sample. However, the first information, or at least part thereof, may be determined from the image e.g. by a person. For example, if the information indicative of a heating value comprises a content of needles of the sample or a presence of needles in the sample, such a content/presence may be manually calculated or determined from the image. However, some other quantities, particularly the heating value of the sample, may require measurements in a laboratory in order to receive the first information with sufficient accuracy. As an alternative to laboratory measurements, values

N taken from literature or computer network may be used, if available. Such

N values may be available for clear cases comprising e.g. only one type of

S material. 3 30 z The method comprises a step 'd': teaching the data-driven model using the

N image of the sample and the first information. In this step, the image in 5 connection with the first information is fed to the model whereby the data- 5 driven model may e.g. self-organize to give more accurate results.

N 35

Finally, the method comprises repeating the steps ‘a’ to ‘d’. These steps may be repeated multiple times. These steps may be repeated so many times that reasonably accurate estimates can be obtained using the data-driven model.

In the alternative, the steps ‘a’ to ‘d’ may be repeated a fixed number of times with the assumption that after that number of teaching steps, the model is sufficiently accurate for purposes of use or testing. In any case, an embodiment comprises determining that further teaching is required, and as a result of determining that further teaching is required, repeating steps ‘a’ to ‘d’.

Determining that further teaching is required may comprise at least one of: - determining that the accuracy of the model is insufficient, e.g. by determining that a norm of a difference between a true value of the first information indicative of a heating value of the material of the sample and a first information indicative of a heating value of the material of the sample as calculatable by the model exceeds a threshold, - determining that not sufficiently many samples has been taught to the model, and - determining that the sample was not the last one of the samples to be taught to the model.

Hereinabove, the term “norm” refers to a measure indicating a magnitude, such as an absolute value of a single number, a root-mean-square value of several numbers, a sum of absolute values of several numbers, or any other norm as known in the field of mathematics.

Hereinabove, the expression “the samples to be taught to the model” indicates a set of samples that are to be taught to the model. The set may comprise e.g. all the samples indicated in Table 1. However, the set may comprise more

N samples, such as samples comprising at least two different types of materials,

N optionally samples comprising different proportional amounts (e.g.

S percentages) of at least two different types of materials. 3 30 z The step of “determining that further teaching is required” is indicated in Fig. 3

N as an affirmative answer (“YES”) to the question “Is further teaching required?”

NS

D

N Preferably, the teaching of the model is done in the same location (i.e. in

N 35 connection with the same plant 500) where the model will be used. This is to ensure that the content of the feedstock FS (including wood species and main types of impurities) during teaching of the model is the same as during use of the model.

In an embodiment, wherein the model is taught for use in connection with feedstock comprising solid biomass originating from plants other that wood, the first information comprises information on at least one of: - presence of sand in the sample, - presence of soil in the sample, - presence of hay in the sample, - presence of straw in the sample, - presence of husk in the sample, - presence of chaff in the sample, - presence of cobs in the sample, - presence of bagasse in the sample, - a content of sand of the sample, - a content of soil of the sample, - a content of hay of the sample, - a content of straw of the sample, - a content of husk of the sample, - a content of chaff of the sample, - a content of cobs of the sample, - a content of bagasse of the sample, - a content of moisture of the sample, and - a heating value of the sample.

Depending on the intended use of the model, particularly the typical biomass

N of the feedstock, it may suffice to teach the model to recognise only one type

N of biomass and one type of impurity. E.g. if the intended use is such that only

S bagasse (optionally with impurities, e.g. sand) is used as the biomass of the

S 30 feedstock, it may suffice to teach the model to recognise a content of bagasse

E of the feedstock; with the assumption that the content of bagasse and

N “impurities” results in the full feedstock.

NS

D

N In an embodiment, wherein the model is taught for use in connection with

N 35 feedstock comprising solid biomass originating from wood, the first information comprises information on at least one of: - presence of sand in the sample,

- presence of soil in the sample, - presence of bark in the sample, - presence of wood chips in the sample, - presence of material originating from hardwood in the sample, - presence of hardwood chips in the sample, - presence of material originating from softwood in the sample, - presence of softwood chips in the sample, - presence of needles in the sample, - presence of twigs in the sample, - a content of sand of the sample, - a content of soil of the sample, - a content of bark of the sample, - a content of wood chips of the sample, - a content of material originating from hardwood of the sample, - a content of material originating from softwood of the sample, - a content of needles of the sample, - a content of twigs of the sample, - a content of moisture of the sample, and - a heating value of the sample.

Thus, in an embodiment, wherein the model is taught for use in connection with feedstock comprising solid biomass originating from plants (wood and/or other than wood), the first information may comprise any one of the aforementioned quantities.

Naturally, when used, the data-driven model gives qualitatively similar

N information that has been taught to it. In other words, if the first information

N (used for teaching) comprises a content of sand of the sample, the primary

S information (received from the data-driven model when used) comprises a 8 30 content of sand of the sample. And in a similar manner, if the first information z (used for teaching) comprises a heating value of the sample, the primary

N information (received from the data-driven model when used) comprises a

S heating value of the sample.

QA

Since the heating value has a lot of practical value for the process control, preferably, the step 'c' comprises determining a heating value of the sample; and the step 'd' comprises teaching the data-driven model using the heating value of the sample and the image of the sample. As indicated above, the heating value of the sample may be determined by laboratory measurements.

In such a case, the primary information indicative of a heating value of the feedstock FS comprises the heating value of the feedstock.

As detailed above, the moisture content may be one practically important parameter. Thus, in an embodiment the step 'c' comprises determining also a moisture content of the sample, and the step 'd' comprises teaching the data- driven model using also the moisture content of the sample. Even if the sample (or feedstock) is dried before taking the images, the feedstock and the sample contains some residual moisture.

The accuracy of the data-driven model depends also on the data used for teaching. First, the model may be taught what is shown in a figure. The property that the model should recognize may be e.g. bark main type, impurity (sand, twigs, needles), wood part (bark or chips), type of wood chips (softwood or hardwood), non-wood based biomass part (hay, straw, husk, chaff, cobs, bagasse). For each one of these, two figures, one showing the property to recognize and another not showing the property to recognize are fed to the model, thereby teaching the model to recognize the property. After teaching, the model is capable of determining the presence of these materials from the figures.

Table 1 summarizes the samples for recognizing these properties for an embodiment, therein the feedstock comprises solid biomass that originates from wood. In case the feedstock comprises (in addition or alternatively) solid

N biomass originating from other type of plants, images of such samples should

N be used (in addition or alternatively, respectively). However, sand may be

S present, as an impurity, also when the feedstock comprises solid biomass 8 30 originating from other type of plants, in which case sand should be a sample

E in the teaching also in that case.

R

D

S

Sample 1: Not Sample 2:

Property to recognize present Present

Bark main type: Hardwood bark Softwood bark Hardwood bark

Impurity: sand | Softwood bark Sand

Impurity: twigs Softwood bark Twigs

Impurity: needles Softwood bark — Needles

Material type, softwood chips or bark:

Softwood Chips Softwood bark Softwood chips

Type of wood of the chips: Hardwood chips | Softwood chips Hardwood chips

Table 1: Samples for teaching the model to recognize certain properties from the images.

It is noted that in Table 1, the properties of “Harwood bark” and “Hardwood chips” are taught. Thus, even if Table 1 does not show the property to recognize “Material type, hardwood chips or bark: Hardwood Chips”, it may be assumed that with this data, the model will learn to recognize whether an image shows hardwood chips or hardwood bark.

In addition to teaching the model the properties to recognize, the model may be taught also the heating value related to the recognized property. The heating value of these samples may be measured in a laboratory or readable from literature or computer network, as detailed above. As collectable from table 1, the samples include the samples in the column “Sample 2: Present” of

Table 1 as well as the sample of “Softwood bark”.

N In the teaching, the species of the softwood (e.g. pine or spruce) as well as the

N species of the hardwood (e.g. birch) should be the same as the species of the

N 20 wood when the model is used. In case different hardwood species (e.g. both © pine and spruce) are used as the biomass, simultaneously or subsequently,

I corresponding images and property is taught to the model. In such a case, + concerning, as an example, the property to recognize “Type of wood of the

S chips”, in order to teach this, samples of Hardwood chips (e.g. birch), Softwood

O 25 chips from first softwood species (e.g. spruce), and Softwood chips from

O second softwood species (e.g. pine) should be used. This applies also to different types of barks.

In the above and in Table 1, the samples contain only one type of material.

However, the model may be taught to recognize a portion (i.e. content) of various types of the materials shown in a figure. To this end, samples comprising at least two different materials of the samples, e.g. the samples of

Table 1 or other samples usable in connection with other materials, can be defined and taught to the model. Accordingly, the reference information could comprise the true portions (i.e. contents) of these materials and/or a true heating value of this combination of the materials. This type of teaching can be done before the model is used and/or during use of the model.

As for the issues of, on one hand, when the data-driven model is sufficiently accurate, and, on the other hand, when further teaching is not required, after some initial repetitions of the steps ‘a’ to ‘d’, a result of the model, as determined from the image, can be compared with reference information.

Provided that the results are sufficiently accurate, the model has been sufficiently taught. In any case, the reference information (i.e. first information) can be simultaneously used for further teaching the model.

Thus, and with reference to Fig. 4, an embodiment comprises, after repeating the steps ‘a’ to ‘d’:

A step f: taking at least one image from a feedstock FS comprising solid biomass. This corresponds to the step ‘b’. In order to have accurate results, preferably the images taken in the step 'b' are taken from the same location as the images taken in the step ‘f'. Moreover, in order to have even more accurate results, preferably in the step ‘b’, the sample is illuminated in a similar manner

N as the feedstock is illuminated in the step f. The similar manner refers to a

N same way, i.e. same intensity; and if light flashes are used for illumination,

S same duration. 3 30 z A step 'g': determining first information indicative of a heating value of the

N material of the feedstock FS of which image is taken in the step f. This 5 corresponds to the step ‘c’. Same methods can be used for determining the = first information as defined above in connection with the step ‘Cc’. & 35

A step 'h': determining, using the at least one image taken in step 'f and the data-driven model on a processing unit CPU, primary information indicative of a heating value of the feedstock FS. This corresponds to using the data-driven model. In this way the primary information thus obtained indicates how the model would interpret the at least one image in term of information indicative of a heating value of the feedstock FS.

A step ‘I’: comparing the first information determined in step 'g' with the primary information determined in step ‘h’ to obtain a comparison result. If the comparison result is small then the model has been sufficiently taught, and further teaching is not needed. As for the smallness of the comparison result, any norm can be used, as detailed above. One difference in Figs. 3 and 4 is that in Fig. 3, the sample need not be defined when the model is used. In contrast, the samples for steps ‘a’ to 'd' may be engineered so as to resemble clear cases, as detailed in Table 1. In contrast, in the embodiment of Fig. 4, the images may be taken on-line e.g. from a conveyor conveying feedstock to the actual industrial process. However, testing the model, as in Fig. 4, needs not be done in the actual process environment. Determining the accuracy of the model can be made in a similar manner in the embodiment of Fig. 3 as in the embodiment of Fig. 4. However, as detailed above, in the context of Fig. 3, the accuracy needs not be determined at all.

If the comparison result is large then the model has been insufficiently taught, and further teaching is needed. The first information of step 'g' can be used for the purpose.

Thus, in this case the method comprises step 'j: teaching the data-driven model. Preferably, the method comprises step j: teaching the data-driven

N model using the at least one image taken in step 'f and the first information

N determined in step 'g'.

S

8 30 However, the first information determined in step 'g' need not be used for z teaching. Irrespective of whether used or not, further teaching can be made

N using other samples, as detailed in connection with step ‘a’ to 'd' (Fig. 3). Thus,

S the steps ‘a’ to 'd' can be repeated for some times, if inaccuracy is determined.

N

N 35 However, since the first information is available, preferably the model is taught in any case. Thus, even if the comparison result is small, the step ‘j may be performed.

If the comparison result is large, the steps f to j are to be repeated until the model gives sufficiently accurate results.

Once the data-driven model has been taught, as detailed above in the steps ‘a’ to 'd'; and, optionally further taught as detailed above in the steps f to 7, the model can be used for determining the primary information indicative of the heating value of the feedstock FS and for controlling the process, as detailed above. In short, an embodiment comprises both teaching the model, as described above and using it for process control as described above.

More specifically, and with reference to Fig. 5, an embodiment comprises comprising, after the step 'e' (and also after the step j', if applicable):

A step 'k': taking at least one image from a feedstock FS comprising solid biomass. A step ‘I: feeding at least the feedstock FS to a plant 500 for thermal processing. A step ‘m’: determining, using the at least one image and the data- driven model on a processing unit CPU, primary information indicative of a heating value of the feedstock FS. And a step 'n': based on the primary information controlling a flow of material into the plant 500 and/or a temperature or a content of the material of the flow into the plant 500.

Reference is made to what has been said about using the model above.

In order to have accurate results, preferably the images taken in the step 'b' are taken from the same location as the images taken in the step ‘k’. Moreover, in order to have even more accurate results, preferably in the step 'b', the

N sample is illuminated in a similar manner as the feedstock is illuminated in the

N step ‘kK’. The similar manner refers to a same way, i.e. same intensity; and if

S light flashes are used for illumination, same duration. 3 30 z In order to have accurate results, preferably the images taken in the step f are

N taken from the same location as the images taken in the step ‘k’. Moreover, in 5 order to have even more accurate results, preferably in the step f', the sample = is illuminated in a similar manner as the feedstock is illuminated in the step 'k..

The similar manner refers to a same way, i.e. same intensity; and if light flashes are used for illumination, same duration.

When the model is used, the power of the plant 500 can be used to further teach the model, as disclosed above. This applies both in the context of teaching the model as well as in the context of using the model.

N

QA

O

N

K

<Q 00

O

I a a

O

NS o

LO

N

O

N

Claims (21)

Claims:

1. A method for thermal processing of biomass, the method comprising - taking at least one image from a feedstock (FS) comprising solid biomass, - feeding at least the feedstock (FS) to a plant (500) for thermal processing, - determining primary information indicative of a heating value of the feedstock (FS) using a processing unit (CPU) and the at least one image, - based on the primary information controlling a flow of material fed into the plant (500) and/or a temperature or a content of the material fed into the plant (500), and - feeding oxygen-containing gas to the plant (500).

2. The method of the claim 1, wherein the feedstock (FS) comprises solid material originating from plants, such solid material (WM) originating from wood, and/or the primary information comprises information on at least one of: - presence of sand in the feedstock (FS), - presence of soil in the feedstock (FS), - presence of hay in the feedstock (FS), - presence of straw in the feedstock (FS), - presence of husk in the feedstock (FS), - presence of chaff in the feedstock (FS), - presence of cobs in the feedstock (FS), - presence of bagasse in the feedstock (FS), - presence of bark in the feedstock (FS), N - presence of wood chips in the feedstock (FS), N - presence of material originating from hardwood in the feedstock (FS), S - presence of hardwood chips in the feedstock (FS), 8 30 - presence of material originating from softwood in the feedstock (FS), z - presence of softwood chips in the feedstock (FS), N - presence of needles in the feedstock (FS), 5 - presence of twigs in the feedstock (FS), = - a content of sand of the feedstock (FS), - a content of soil of the feedstock (FS), - a content of hay of the feedstock (FS), - a content of straw of the feedstock (FS),

- a content of husk of the feedstock (FS), - a content of chaff of the feedstock (FS), - a content of cobs of the feedstock (FS), - a content of bagasse of the feedstock (FS), - a content of bark of the feedstock (FS), - a content of wood chips of the feedstock (FS), - a content of material originating from hardwood of the feedstock (FS), - a content of hardwood chips of the feedstock (FS), - a content of material originating from softwood of the feedstock (FS), - a content of softwood chips of the feedstock (FS), - a content of needles of the feedstock (FS), - a content of twigs of the feedstock (FS), - a content of moisture of the feedstock (FS), and - the heating value of the feedstock (FS); preferably, the method comprises - determining, using the processing unit, the heating value of the feedstock (FS) and - the feedstock (FS) comprises solid material originating from plants.

3. The method of claim 1 or 2, comprising - feeding air to the plant (500); preferably, the method comprises - controlling a flow of the oxygen-containing gas into the plant (500) and/or a temperature of the oxygen-containing gas (500).

4. The method of any of the claims 1 to 3, comprising controlling, based on the N primary information, at least one of: N - flow of the feedstock (FS) into the plant (500), S - a content of the material fed to the plant (500) for thermal processing, the 8 30 material comprising the feedstock (FS), z - flow of gas, such as air or steam, e.g. combustion air or fluidizing air, into the N plant (500), and 5 - a temperature of a gas, such as air, fed to the plant (500); = preferably the method comprises - controlling, based on primary information, a flow of the feedstock (FS) into the plant 500 and/or a flow of such gas that is configured to oxidise the feedstock (FS) into the plant (500).

5. The method of any of the claims 1 to 4, wherein - the feedstock (FS) is fed to a gasifier (510), to a lime kiln (520), to a boiler (530), to a pyrolysis reactor (540), to a torrefaction reactor (550), or to a steam explosion reactor (560).

6. The method of claim 5, wherein - the plant (500) comprises a kiln or a boiler, such as a lime kiln (520), preferably, - the plant (500) further comprises one of a gasifier (510), a torrefaction reactor (550), and a steam explosion reactor (560), wherein - the feedstock (FS) is fed to the gasifier (510), the torrefaction reactor (550), or the steam explosion reactor (560), and - a product of the gasifier (510), the torrefaction reactor (550), or the steam explosion reactor (560) is used as fuel in the kiln or the boiler; more preferably, - the plant (500) comprises a lime kiln (520), - calcium carbonate (CaCO3) is fed to the lime kiln (520) and calcinated therein, and - the product of the gasifier (510), the torrefaction reactor (550), or the steam explosion reactor (560) is used as fuel in the lime kiln (520.

7. The method of claim 6, wherein - the plant (500) comprises a lime kiln (520) and a gasifier (510), - the feedstock (FS) is fed to the gasifier (510), - a product gas (PG) of the gasifier (510) is used as fuel gas (FG) in the lime N kiln (520), and N - calcium carbonate (CaCO3) is fed to the lime kiln (520) and calcinated therein; S preferably, 8 30 - the gasifier (510) is a fluidized bed gasifier. =

N 8. The method of any of the claims 1 to 7, comprising 5 - conveying the feedstock (FS) towards the plant (500) and/or or to the plant = (500) using a conveyor (120), preferably a belt conveyor, and - taking the at least one image of the feedstock (FS) such that - the at least one image is taken from a part of the feedstock (FS) when that part of the feedstock (FS) is arranged in or on the conveyor (120).

9. The method of any of the claims 1 to 8, comprising - drying the feedstock (FS) before feeding the feedstock (FS) to the plant (500); preferably, - the at least one image from the feedstock (FS) is taken after the feedstock (FS) has been dried and before feeding to the plant (500).

10. The method of any of the claims 1 to 9, comprising - determining a moisture content of the feedstock (FS) and - controlling the flow of material into the plant (500) and/or a temperature or a content of the material of the flow into the plant (500) using also the moisture content of the feedstock (FS); preferably, - the moisture content of the feedstock (FS) is determined using other measurements than the images.

11. The method of any of the claims 1 to 9, wherein - the primary information indicative of a heating value of the feedstock (FS) is determined using the at least one image and a data-driven model on the processing unit (CPU) and - the primary information indicative of a heating value of the feedstock (FS) comprises the heating value of the feedstock (FS), and the method comprises - determining a power of the plant (500), - determining a heating value of the feedstock (FS) using the power of the plant (500), and - teaching the data-driven model with the at least one image and the heating N value determined using the power of the plant (500). N S 12. A system for thermal processing of biomass, the system comprising S 30 - a plant (500) for thermal processing of feedstock (FS) comprising solid E biomass, N - an imaging means configured to take at least one image from the feedstock S (FS), N - a controlling apparatus configured to control a flow of material into the plant and/or a temperature of a content of the material of the flow into the plant (500), - means for feeding oxygen-containing gas to the plant (500), and [A]

- a processing unit (CPU) configured to determine primary information indicative of a heating value of the feedstock (FS) using the at least one image, or [B] - means for sending data indicative of the at least one image to an interface, and - means for receiving primary information indicative of a heating value of the feedstock (FS) from the interface, wherein the interface is configured to - receive the data indicative of the at least one image, - determine primary information indicative of a heating value of the feedstock (FS) using the data indicative of the at least one image, and - arrange available the primary information indicative of a heating value of the feedstock (FS).

13. The system of claim 12, wherein - the controlling apparatus and at least one of the processing unit (CPU) and a controller of the system are, in combination, configured to control the flow of material fed into the plant (500) and/or the temperature or the content of the material fed into the plant (500) based on the primary information and/or - the system comprises a transmitter for transmitting the primary information to an operator and/or - the system comprises means for sending data indicative of the at least one image (I) to an interface, and the interface is configured to arrange available the primary information indicative of a heating value of the feedstock (FS) to an operator. N 14. The system of claim 12 or 13, wherein N - the plant comprises at least one of a gasifier (510), a lime kiln (520), a boiler S (530), such as an incinerator, a pyrolysis reactor (540), a torrefaction reactor 8 30 (550), and a steam explosion reactor (560); z preferably, N - the plant (500) comprises a kiln or a boiler, such as a lime kiln (520); 5 preferably, = - the plant (500) comprises a kiln or a boiler and one of a gasifier (510), a torrefaction reactor (550), and a steam explosion reactor (560), and

- a pipeline (530) or a conveyor for conveying a product of the gasifier (510), the torrefaction reactor (550), or the steam explosion reactor (560) to the kiln or the boiler to be used as fuel in the kiln or the boiler.

15. The system of claim 14, comprising - a lime kiln (520) and one of a gasifier (510), a torrefaction reactor (550), and a steam explosion reactor (560), and - a pipeline (530) or a conveyor for conveying a product of the gasifier (510), the torrefaction reactor (550), or the steam explosion reactor (560) to the lime kiln (520) to be used as fuel in the lime kiln (520); preferably, - the plant (500) comprises a lime kiln (520), a gasifier (510), and a pipeline (530) for conveying product gas (PG) from the gasifier (510) to the lime kiln (520) to be used is fuel gas (FG) in the lime kiln (520).

16. A method for teaching a data-driven model used by a processing unit (CPU), the data-driven model being for the purpose of determining from at least one image primary information indicative of a heating value of the material shown in the at least one image, the method comprising a. defining a sample of a feedstock comprising solid biomass,

b. taking an image of the sample,

c. determining first information indicative of a heating value of the material of the sample,

d. teaching the data-driven model using the image of the sample and the first information, and e. repeating the steps a to d. N N

17. The method of claim 16, wherein S - the feedstock comprises solid material originating from plants, such as solid 8 30 material (WM) originating from wood and/or z the first information comprises information on at least one of: N - presence of sand in the sample, 5 - presence of soil in the sample, = - presence of hay in the sample, - presence of straw in the sample, - presence of husk in the sample, - presence of chaff in the sample,

- presence of cobs in the sample, - presence of bagasse in the sample, - presence of bark in the sample, - presence of wood chips in the sample, - presence of material originating from hardwood in the sample, - presence of hardwood chips in the sample, - presence of material originating from softwood in the sample, - presence of softwood chips in the sample, - presence of needles in the sample, - presence of twigs in the sample, - a content of sand of the sample, - a content of soil of the sample, - a content of hay of the sample, - a content of straw of the sample, - a content of husk of the sample, - a content of chaff of the sample, - a content of cobs of the sample, - a content of bagasse of the sample, - a content of bark of the sample, - a content of wood chips of the sample, - a content of material originating from hardwood of the sample, - a content of hardwood chips of the sample, - a content of material originating from softwood of the sample, - a content of softwood chips of the sample, - a content of needles of the sample, - a content of twigs of the sample, N - a content of moisture of the sample, and N - the heating value of the sample; S preferably, the method comprises 8 30 - in the step c, determining a heating value of the sample, and z - in the step d, teaching the data-driven model using the heating value of the N sample; and in the method S - the feedstock comprises solid material originating from plants. QA

18. The method of claim 16 or 17, comprising, after the step e,

f. taking at least one image from a feedstock (FS) comprising solid biomass,

g. determining first information indicative of a heating value of the material of the feedstock (FS) of which image is taken in the step f,

h. determining, using the at least one image taken in the step f and the data- driven model on a processing unit (CPU), primary information indicative of a heating value of the feedstock (FS), and i. comparing the first information determined in step g with the primary information determined in step h to obtain a comparison result; preferably the method further comprises j. teaching the data-driven model; more preferably, the step j comprises teaching the data-driven model using the at least one image taken in step f and the first information determined in step g; optionally, the method comprises repeating the steps f to j.

19. The method of any of the claims 16 to 18, comprising, after the step e,

k. taking at least one image from a feedstock (FS) comprising solid biomass,

I. feeding at least the feedstock (FS) to a plant (500) for thermal processing,

m. determining, using the at least one image taken in step k and the data- driven model on a processing unit (CPU) to determine primary information indicative of a heating value of the feedstock (FS), and n. based on the primary information determined in step m controlling a flow of material fed into the plant (500) and/or a temperature or a content of the material fed into the plant (500).

20. The method of claim 18 or 19, wherein N -the images taken in the step b are taken from the same location as the images N taken in the step f or the step k; S preferably, 8 30 - when taking the image in step b, the sample is illuminated in a same way as E the feedstock is illuminated, when taking the image thereof in the stepf or the n step k. 3 N

21. The method of claim 19 or 20, wherein N 35 - the first information comprises information on the heating value of the sample, the method comprising - determining a power of the plant (500),

- determining a heating value of the feedstock (FS) using the power of the plant (500), and - teaching the data-driven model with the heating value determined using the power of the plant (500) and the image taken in the step k.

N QA O N K <Q 00 O I = O NS o LO N O N