Method for estimating the flow rate in an atomizer

Abstract

The invention relates to a method for dynamically estimating a flow rate Qmetal of molten metal flowing from a reservoir to a nozzle of an atomization chamber of an atomizer connected to the reservoir through a channel equipped with an induction heater, the method comprising the steps of: - Collecting a temperature Treservoir of the molten metal in the reservoir, a temperature Tnozzle of the molten metal in a portion comprised between the exit of the channel and the exit of the nozzle, and an instantaneous power Pinduction of the induction heater, - Estimating the flow rate Qmetal of molten metal based on ratio (1): Formula (I). The invention also relates to the method thereof for managing the fluid-to- metal ratio during atomization, to the monitoring device thereof and to the atomizer thereof.

Description

[0001] Method for estimating the flow rate in an atomizer

[0002] The present invention relates to the production of metal powders and in particular to the production of steel powders by atomization for additive manufacturing. The present invention relates in particular to a method for estimating the flow rate of molten metal flowing from the reservoir to the nozzle of the atomizer.

[0003] There is an increasing demand for metal powders for additive manufacturing and the manufacturing processes have to be adapted consequently.

[0004] One of the main manufacturing processes is gas atomization where a stream of liquid metal is atomized into fine metal droplets by impinging the stream with a high velocity gas stream. It is known that the gas flow impacts the particle size distribution and the microstructure of the metal powder. In particular, for a given metal flow rate, the higher the gas flow, the higher the cooling rate. Consequently, the gas to metal ratio, defined as the ratio between the gas flow rate and the metal flow rate has to be kept in a precise range to produce particles with the desired particle size distribution and microstructure. To do so, the metal flow rate has to be precisely estimated during production, i.e. dynamically, so that the gas to metal ratio can be adjusted accurately on the fly.

[0005] The aim of the present invention is therefore to remedy the drawbacks of the facilities and processes of the prior art by providing a method for dynamically and accurately estimating the flow rate of molten metal flowing through the nozzle of an atomizer.

[0006] For this purpose, a first subject of the present invention consists in a method for dynamically estimating a flow rate Qmetai of molten metal flowing from a reservoir to a nozzle of an atomization chamber of an atomizer connected to the reservoir through a channel equipped with an induction heater, the method comprising the steps of:

[0007] - Collecting a temperature Treservoir of the molten metal in the reservoir, a temperature Tnozzie of the molten metal in a portion comprised between the exit of the channel and the exit of the nozzle, and an instantaneous power Pinduction of the induction heater,

[0008] Estimating the flow rate Qmetai of molten metal based on ratio (1 ):

[0009] _ Pinduction _

[0010] (Tnozzle~ 7 reservoir)

[0011] The method according to the invention may also have the optional features listed below, considered individually or in combination:

[0012] - the flow rate Qmetai of molten metal is estimated on the basis of a mean of the ratio (1 ) and of at least one of a derivative Mreservoir of the mass Mreservoir of the reservoir, a derivative Mpowder of the mass Mpowder of the powder produced in the atomization chamber and of a derivative Hreservoir of the level H reservoir of molten metal in the reservoir.

[0013] - the flow rate Qmetai of molten metal is estimated based on Formula (2): wherein Cp.metai is the specific heat of the molten metal,

[0014] - the collecting step further includes collecting the mass Mreservoir of the reservoir and the estimating step further includes taking Mreservoir into account to estimate the flow rate Qmetai of molten metal,

[0015] - the flow rate Qmetai of molten metal is estimated based on Formula (4): wherein Cp.metai is the specific heat of the molten metal and Kmr is obtained by a calibration of the flow rate Qmetai of molten metal based on a variation of the mass Mreservoir of the reservoir over a set time lapse,

[0016] - the flow rate Qmetai of molten metal is estimated based on Formula (5): wherein Cp.metai is the specific heat of the molten metal, Mreservoir is a derivative of the mass M reservoir of the reservoir and wi and W2 are nonzero coefficients, - the flow rate Qmetai of molten metal is estimated based on a Kalman filter having: o a state vector containing the flow rate Qmetai of molten metal and the mass M reservoir of the reservoir and, o a vector of measurements containing the result of Formula (3) and the mass M reservoir of the reservoir: wherein Cp.metai is the specific heat of the molten metal and Ki is a nonzero proportionality constant,

[0017] - the collecting step further includes collecting the mass Mpowder of the powder produced in the atomization chamber and the estimating step further includes taking Mpowder into account to estimate the flow rate Qmetai of molten metal,

[0018] - the flow rate Qmetai of molten metal is estimated based on Formula (6): wherein Cp.metai is the specific heat of the molten metal and Kmpis obtained by a calibration of the flow rate Qmetai of molten metal based on a variation of the mass Mpowder of the powder produced in the atomization chamber over a set time lapse,

[0019] - the flow rate Qmetai of molten metal is estimated based on Formula (7): wherein Cp.metai is the specific heat of the molten metal, Mpowder is a derivative of the mass Mpowder of the powder produced in the atomization chamber and W3 and W4 are non-zero coefficients,

[0020] - the collecting step further includes collecting the mass Mreservoir of the reservoir and the flow rate Qmetai of molten metal is estimated based on

[0021] Formula (8): wherein Cp.metai is the specific heat of the molten metal, Mreservoir is a derivative of the mass M reservoir of the reservoir, Mpowder is a derivative of the mass Mpowder of the powder produced in the atomization chamber and ws, we and w? are non-zero coefficients,

[0022] - the flow rate Qmetai of molten metal is estimated based on a Kalman filter having: o a state vector containing the flow rate Qmetai of molten metal and the mass Mpowder of the powder produced in the atomization chamber and, o a vector of measurements containing the result of Formula (3) and the mass Mpowder of the powder produced in the atomization chamber: wherein Cp.metai is the specific heat of the molten metal and Ki is a non-zero proportionality constant,

[0023] - the atomizer further comprises a powder bin connected to the bottom of the atomization chamber and collecting the powder produced in the atomization chamber and Mpowder is deducted from the mass Mbin of the powder bin,

[0024] - the collecting step further includes collecting the level Hreservoir of molten metal in the reservoir and the estimating step includes taking Hreservoir into account to estimate the flow rate Qmetai of molten metal,

[0025] - the flow rate Qmetai of molten metal is estimated based on Formula (9): wherein Cp.metai is the specific heat of the molten metal and Khr is obtained by a calibration of the flow rate Qmetai of molten metal based on a variation of the level H reservoir of molten metal in the reservoir over a set time lapse,

[0026] - the flow rate Qmetai of molten metal is estimated based on Formula (10): wherein Cp.metai is the specific heat of the molten metal, Hreservoir is a derivative of the level H reservoir of molten metal in the reservoir, Sreservoir is the section of the reservoir, pmetai is the density of the molten metal and ws and wg are non-zero coefficients,

[0027] - the collecting step further includes collecting the mass Mreservoir of the reservoir and the flow rate Qmetai of molten metal is estimated based on

[0028] Formula (11 ): wherein Cp.metai is the specific heat of the molten metal, Mreservoir is a derivative of the mass M reservoir of the reservoir, Hreservoir is a derivative of the level H reservoir of molten metal in the reservoir and wio, wn and W12 are non-zero coefficients,

[0029] - the collecting step further includes collecting the mass Mreservoir of the reservoir and the mass Mpowder of the powder produced in the atomization chamber and the flow rate Qmetai of molten metal is estimated based on

[0030] Formula (13):

[0031] Qmetai wherein Cp.metai is the specific heat of the molten metal, Mreservoir is a derivative of the mass M reservoir of the reservoir, Mpowder is a derivative of the mass Mpowder of the powder produced in the atomization chamber, H reservoir is a derivative of the level H reservoir of molten metal in the reservoir, Sreservoir is the section of the reservoir, pmetai is the density of the molten metal and wi6, W17, wis and W19 are non-zero coefficients, the flow rate Qmetai of molten metal is estimated based on a Kalman filter having: o a state vector containing the flow rate Qmetai of molten metal and the level Hreservoir of molten metal in the reservoir and, o a vector of measurements containing the result of Formula (3) and the level Hreservoir of molten metal in the reservoir: wherein Cp.metai is the specific heat of the molten metal and Ki is a non-zero proportionality constant, in the collecting step, the temperature Treservoir of the molten metal in the reservoir is obtained from a molten metal temperature measurement device (11 ) which is a part of the reservoir (9), the temperature Tnozzie of the molten metal in a portion comprised between the exit of the channel and the exit of the nozzle is obtained from a metal stream temperature measurement device (13) which is a part of the atomizer (1 ) and the instantaneous power Pinduction of the induction heater is obtained from an induction generator controlling the induction heater (12).

[0032] A second subject of the invention consists in a method for managing the fluid- to-metal ratio during atomization of molten metal firstly flowing from a reservoir to a nozzle of an atomization chamber of an atomizer connected to the reservoir through a channel equipped with an induction heater, and secondly being impinged by a fluid jetted from a fluid sprayer, the method comprising the steps of:

[0033] - Setting a set range Rsof fluid-to-metal ratio,

[0034] - Collecting a temperature Treservoir of the molten metal in the reservoir, a temperature Tnozzie of the molten metal in a portion comprised between the exit of the channel and the exit of the nozzle and an instantaneous power Pinduction of the induction heater,

[0035] - Estimating the flow rate Qmetai of molten metal based on ratio (1 ):

[0036] _ P induction _

[0037] ( nozzle~T reservoir)

[0038] - Calculating the fluid-to-metal ratio from the estimated flow rate Qmetai of molten metal, - Correcting a deviation of the calculated fluid-to-metal ratio beyond the set range Rsof fluid-to-metal ratio by adjusting an atomization setting.

[0039] The optional additional features presented above for the estimation method can also apply to the management method.

[0040] A third subject of the invention consists in a monitoring device comprising at least a processor and a memory, configured to execute the following steps:

[0041] - Collecting: o a temperature Treservoir of a molten metal in a reservoir, o a temperature Tnozzie of the molten metal in a portion comprised between an exit of a channel connecting the reservoir to a nozzle of an atomization chamber of an atomizer and an exit of the nozzle, and o an instantaneous power Pinduction of an induction heater equipping the channel,

[0042] - Estimating a flow rate Qmetai of molten metal flowing from the reservoir to the nozzle based on ratio (1 ):

[0043] _ induction _

[0044] (Tnozzle~reservoir)

[0045] The monitoring device can further be configured to execute the following steps:

[0046] - Collecting a set range Rsof fluid-to-metal ratio,

[0047] - Calculating a fluid-to-metal ratio from the estimated flow rate Qmetai of molten metal, and from a flow rate of a fluid jetted from a fluid sprayer to impinge the molten metal flowing out of the nozzle,

[0048] - Determining an adjusted atomization setting, for correcting a deviation of the calculated fluid-to-metal ratio beyond the set range Rsof fluid-to-metal ratio.

[0049] The optional additional features presented above for the estimation method can also apply to this monitoring device. A fourth subject of the invention consists in an atomizer comprising a reservoir and an atomization chamber, the reservoir being connected to a nozzle of the atomization chamber through a channel equipped with an induction heater, the atomizer further comprising:

[0050] - a molten metal temperature measurement device installed so as to measure a temperature Treservoir of the molten metal in the reservoir,

[0051] - a metal stream temperature measurement device installed so as to measure a temperature Tnozzie of the molten metal in a portion comprised between an exit of the channel and an exit of the nozzle,

[0052] - an induction generator for controlling the induction heater,

[0053] - the monitoring device according to the invention, connected to the molten metal temperature measurement device and to the metal stream temperature measurement device and interfaced with the induction generator.

[0054] The optional additional features presented above for the estimation method can also apply to this atomizer.

[0055] A fifth subject of the invention consists in a computer program comprising instructions, whose execution on a computer (said computer being connected to temperature sensors, to an induction generator, and possibly to a fluid regulator depending on the details of the method executed by said computer) make the computer to execute a method as above described.

[0056] As it is apparent, the invention takes advantage of the presence of an induction heater in the channel connecting the reservoir to the nozzle of the atomization chamber. The instantaneous power Pinduction of the induction heater can be used to estimate the metal flow rate.

[0057] Other characteristics and advantages of the invention will be described in greater detail in the following description.

[0058] The invention will be better understood by reading the following description, which is provided purely for purposes of explanation and is in no way intended to be restrictive, with reference to Figure 1 , which is a schematic presentation of an atomizer according to the invention.

[0059] It should be noted that the terms such as “above”, “below”, “upper”, “lower”, “vertical axis”... as used in this application refer to the positions and orientations of the different constituent elements of the atomizer when the latter is installed in a plant and operated or ready for operation.

[0060] The fluid to metal ratio is defined as the ratio between the fluid flow rate and the metal flow rate. In particular, it is defined, for gas atomization, as the ratio between the gas flow rate (preferably in m3 / h) and the metal flow rate (preferably in Kg / h) and, for water atomization, the ratio between the water flow rate (preferably in L / h) and the metal flow rate (preferably in Kg / h).

[0061] Metal refers to any metal that can be inductively heated such as steel, carbon steel and stainless steel included, aluminum, copper and brass.

[0062] With reference to Figure 1 , a fluid atomizer 1 is a device designed for atomizing a stream of liquid metal into fine metal droplets by impinging the stream with a high velocity fluid stream. The fluid atomizer can notably be a gas atomizer or a water atomizer. The fluid atomizer 1 is mainly composed of a closed atomization chamber 2 maintained under protective atmosphere. The chamber has an upper section, a lower section, a top and a bottom.

[0063] The upper section of the atomization chamber comprises an orifice, the nozzle 3, usually positioned at the center of the chamber top, through which the molten metal stream is forced. The nozzle is surrounded by or comprises a fluid sprayer 4 for jetting a fluid at high speed on the stream of liquid metal. The fluid sprayer is preferably an annular slot through which pressurized fluid flows. The fluid can be a gas, such as nitrogen or argon, or a mixture of gases or water. The fluid sprayer is preferably coupled to a fluid regulator 5 to control the flow and / or the pressure of the fluid before jetting it and thus to control the fluid-to-metal ratio (FMR). The fluid regulator can be a compressor, a fan, a pump, a pipe section reduction or any suitable equipment. The atomizer preferably comprises a flow sensor 6 so that the fluid flow rate in the fluid sprayer can be measured. The flow sensor can be positioned between the fluid sprayer and the fluid regulator or in the fluid regulator or in the fluid sprayer.

[0064] The lower section of the atomization chamber is mainly a receptacle for collecting the metal particles falling from the upper section of the chamber. It is usually designed to facilitate the powder collection and powder discharge, through for example an opening 7 positioned at the bottom of the chamber. It is thus usually in the form of an inverted cone or an inverted frustoconical shape. A powder bin 8 can be positioned below the opening 7 at the bottom of the chamber to collect the powder continuously or by batches.

[0065] The atomizer further comprises a reservoir 9 connected to the nozzle 3 of the atomization chamber through a channel 10. The reservoir is positioned substantially above the atomization chamber. It is mainly a tank capable of maintaining molten metal at a set temperature and capable of letting molten metal flows through the channel 10, possibly at a controlled flow rate. It can have any shape or design as long as it fulfills the above requirements. Preferably, it is capable of being atmospherically controlled.

[0066] The reservoir is equipped with a molten metal temperature measurement device 11 so that the temperature Treser oir of the molten metal in the reservoir can be measured. The molten metal temperature measurement device can be, for example, a thermal probe, a pyrometer, a thermal camera. Preferably, the molten metal temperature measurement device is positioned so that it measures the temperature of the molten metal in the vicinity of the channel 10. The atomizer is built so that the temperature Treservoir of the molten metal in the reservoir can be collected. In particular, the molten metal temperature measurement device is interfaced with a computer or data logger, for instance a monitoring device, presented below.

[0067] The channel 10 can be a tube independent of the reservoir. Alternatively, it can be a narrow portion at the bottom of the reservoir. The channel is preferably made of a refractory material. Other types of channels are possible as long as they can retain the molten metal. The channel is equipped with an induction heater 12. The latter can be in the form of a coil surrounding the channel. With the induction heater, the temperature of the molten metal flowing through the channel can be adjusted. Accordingly, the power of the induction heater can be varied. The variation can be done on demand, for example to maintain the temperature of the molten metal constant at the exit of the channel. Independently of the previous sentence, the power of the induction heater can be regularly varied, possibly periodically varied. The atomizer is built so that the instantaneous power Pinduction of the induction heater can be collected. In particular, the induction generator controlling the induction heater is interfaced with a computer or data logger, for instance the monitoring device presented below, so that its command value or the measured feedback value can be collected.

[0068] The atomizer is equipped with a metal stream temperature measurement device 13 so that a temperature Tnozzie of the molten metal in a portion comprised between the exit of the channel and the exit of the nozzle can be measured. For the sake of clarity, the metal stream temperature measurement device measures the temperature Tnozzie of the molten metal in at least one point of the portion comprised between the exit of the channel and the exit of the nozzle, preferably in one point. The metal stream temperature measurement device can be positioned at the exit of the channel, i.e. at the downstream end of the channel. It can also be positioned at the level of the nozzle, in particular in the nozzle or at the exit of the nozzle. The temperature Tnozzie of the molten metal can thus be measured once the molten metal has flowed through the channel and before, while or just after, the molten metal is impinged by the fluid jet. The temperature measurement device can be, for example, a thermal probe, a pyrometer, a thermal camera. The atomizer is built so that the temperature Tnozzie of the molten metal at the exit of the channel can be collected. In particular, the metal stream temperature measurement device is interfaced with a computer or data logger, for instance the monitoring device 17.

[0069] The monitoring device 17 of the atomizer is an electronic device comprising at least a processor and a memory. Here, the monitoring device is connected:

[0070] - to the molten metal temperature measurement device 11 ,

[0071] - to the metal stream temperature measurement device 13,

[0072] - to the induction generator or to another device providing data representative of the instantaneous power Pinduction,

[0073] - optionally, to the flow sensor 6,

[0074] - optionally, to other sensors of the atomizer (presented below), - optionally, to the fluid regulator 5, so that the monitoring device can control the flow of the fluid,

[0075] - optionally, to the reservoir 9, so that the monitoring device can control the pressure in the reservoir.

[0076] The monitoring device 17 is configured, for instance programmed, to execute a method for dynamically estimating a flow rate Qmetai of molten metal flowing from the reservoir 9 to the nozzle 3 of the atomization chamber, according to one of the embodiments below. In particular, it is configured to execute the following steps:

[0077] - Collecting the data,

[0078] - Estimating the flow rate Qmetai of molten metal flowing from the reservoir to the nozzle,

[0079] - Optionally, calculating the fluid-to-metal ratio from the estimated flow rate Qmetai of molten metal and from the flow rate of fluid jetted from the fluid sprayer,

[0080] - Optionally, determining an adjusted atomization setting, for correcting a deviation of the calculated fluid-to-metal ratio.

[0081] With the atomizer as described above, the flow rate of molten metal flowing from the reservoir into the atomization chamber can be dynamically estimated according to a first embodiment described below.

[0082] In a first step, the data needed for the estimation of the flow rate of molten metal are collected. In particular, the temperature Treser oir of the molten metal in the reservoir, the temperature Tnozzie of the molten metal in a portion comprised between the exit of the channel and the exit of the nozzle, and the instantaneous power Pinduction of the induction heater are collected. More particularly:

[0083] - the temperature Treservoir of the molten metal in the reservoir is measured with the molten metal temperature measurement device 11 , this measurement being then collected (that is, acquired, or in other words, received), preferably by the monitoring device 17,

[0084] - the temperature Tnozzie of the molten metal after it has flowed through the channel is measured with the metal stream temperature measurement device 13, this measurement being then collected, preferably by the monitoring device and, - the instantaneous power Pinduction of the induction heater is collected by the monitoring device, for instance collected from the induction generator controlling the induction heater or from another device providing data representative of the instantaneous power Pinduction.

[0085] The temperature Tnozzie of the molten metal is measured in a portion comprised between the exit of the channel and the exit of the nozzle. In particular, the temperature Tnozzie is measured at the exit of the channel, at the level of the nozzle, in particular in the nozzle, or at the exit of the nozzle.

[0086] This data measurement and collection is done preferably dynamically, for example at regular time intervals, more preferably continuously.

[0087] In a second step, preferably concomitant to the first step, the flow rate Qmetai of molten metal is estimated based on ratio (1 ):

[0088] Ratio (1 ) reflects the fact that the less time the molten metal stays in the channel, the less it needs to be warmed up by the induction heater to be maintained at the set temperature. In other words, the more the molten metal flows, the less it needs to be warmed up.

[0089] This estimation is done preferably dynamically, for example at regular time intervals, more preferably continuously. Preferably it is done with the monitoring device.

[0090] The flow rate Qmetai given based on ratio (1 ) can be a mass flow rate, for example in Kg-s-1, but, as necessary, the flow rate can be easily expressed as a volumetric flow rate, for example in m3-s’1.

[0091] According to a first variant of the estimating step, the flow rate Qmetai of molten metal is estimated based on Formula (2):

[0092] Pinduction

[0093] Qmetai (2)

[0094] Cp,metal(T nozzle T reservoir wherein Cp.metai is the specific heat of the molten metal (preferably expressed in J-kg"1-K-1). The temperatures are preferably expressed in °C or K. Pinduction is preferably expressed in J-s-1.

[0095] As it is apparent from formula (2), the flow rate Qmetai of molten metal is, in that variant, directly estimated from the ratio between the instantaneous power Pinduction of the induction heater and the difference between the temperature Tnozzie and the temperature Treservoir of the molten metal. Such an estimation can be sufficient to accurately maintain the fluid-to-metal ratio in a set range in case of variation of the flow rate of molten metal, since the knowledge of the relative variation of the fluid-to-metal ratio is sufficient.

[0096] According to a second variant of the estimating step, the flow rate Qmetai of molten metal is estimated based on Formula (3): wherein Cp.metai is the specific heat of the molten metal and Ki is a non-zero proportionality constant.

[0097] As it is apparent from formula (3), the molten metal flow rate Qmetai is proportional to the ratio between the instantaneous power Pinduction of the induction heater and the difference between the temperature Tnozzie and the temperature Treservoir of the molten metal, the difference being multiplied by the specific heat of the molten metal. The proportionality constant Ki reflects the possible heat losses in the channel and in the nozzle (if applicable) and the possible efficiency factor from the applied induction power to the actual heating power. The proportionality constant Ki can also incorporate the density of the molten metal so that the flow rate is expressed as a volumetric flow rate, for example in m3-s’1. The proportionality constant Ki can be calculated by the person skilled in the art by calibration tests performed on an atomizer, for example at the start of a new atomizer. Tests can for example be run with different metallostatic pressures in the reservoir at the level of the nozzle, which impacts the molten metal flow rate. For each metallostatic pressure, Pinduction, Tnozzie, Treservoir and the mass of powder discharged at the bottom of the atomization chamber are collected. The proportionality constant is then approximated by linear regression.

[0098] According to a third variant of the estimating step, the latter comprises a modeling step during which the accuracy of the estimation of the flow rate of molten metal is further improved by mathematical modeling. As the different measurements can include noise sources, the estimation of the flow rate can be inaccurate. There are mathematical algorithms that produce an optimal estimate of the metal flow rate based on a model of the dynamic system and the available measurements. Examples of such mathematical algorithms are Kalman filters. The actual formulation of the model can take different forms based on the availability and quality of measurements.

[0099] In a first example, the temperature Treservoir of the molten metal in the reservoir and the inverse of the molten metal flow rate were used as the model states, and their dynamics were modelled using a drifting autoregressive formulation with the noise covariance determining the rate of change. This formulation was shown by simulation to be able to accurately estimate Qmetai and Treservoir even with noisy measurements of the temperature Tnozzie of the molten metal in the nozzle.

[0100] In this example, the Kalman filter is constructed having:

[0101] - a state vector Xk containing the temperature Treservoir of the molten metal in the reservoir and the inverse of the molten metal flow rate 1 / Qmetai and,

[0102] - a vector of measurements Zk containing the measurement of the temperature Tnozzie of the molten metal in the nozzle, and the measurement of the temperature Treservoir of the molten metal in the reservoir.

[0103] Where k is the discrete variable for the elapsed time t and At is the sampling time, so that t = k.At.

[0104] In this example, the state transition matrix Fk is preferably:

[0105] The control input matrix Bk is preferably equal to zero.

[0106] The measurement matrix Hk is preferably: when based on Formula (3).

[0107] Consequently, the Kalman filter model assumes the true state at time-step k is evolved from the state at time-step (k - 1) according to:

[0108] Xk = Fk Xk-i + Wk wherein Wk is the process noise, which is assumed for instance to be drawn from a zero mean multivariate normal distribution N with covariance Qk:

[0109] Wk ~ N (0, Qk)

[0110] At time-step k, a measurement Zk of the true state Xk is assumed to be:

[0111] Zk = Hk Xk + Vk wherein Vk is the measurement noise, which is assumed for instance to be a zero mean Gaussian white noise with covariance Rk:

[0112] Vk ~ N (0, Rk)

[0113] The person skilled in the art will know how to estimate the noise covariance matrices Qk and Rk and how to issue the best estimation xkof the states Xk based on the measurements, for each time-step k based on the Kalman filter theory.

[0114] In a variant of this first example, Formula (2) is used instead of Formula (3).

[0115] According to a second embodiment of the invention, the mass Mreservoir of the reservoir is further taken into account to estimate the flow rate Qmetai of molten metal.

[0116] The collecting step of this second embodiment differs from the collecting step of the first embodiment in that it further includes collecting the mass Mreservoir of the reservoir. Mreservoir can be measured with a load cell, the measurement being then collected, preferably by the monitoring device. Accordingly, the atomizer can thus be equipped with at least one reservoir load cell 14. The reservoir load cell can be positioned on the structure supporting the reservoir, if applicable, or positioned below the reservoir. The atomizer is built so that the mass of the reservoir can be collected. In particular, the reservoir load cell is interfaced with a computer or data logger, for instance the monitoring device 17.

[0117] The estimating step of this second embodiment differs from the estimating step of the first embodiment in that it further includes taking Mreservoir into account to estimate the flow rate Qmetai of molten metal. Consequently, the flow rate Qmetai of molten metal is estimated more accurately.

[0118] According to a first variant of the second embodiment, the flow rate Qmetai of molten metal is estimated based on Formula (4): wherein Kmr is a non-zero proportionality coefficient obtained by a calibration of the flow rate Qmetai of molten metal based on a variation of the mass Mreservoir of the reservoir over a set time lapse.

[0119] Before calibration, the mass Mreservoir of the reservoir is preferably filtered with a low-pass filter to reduce noise. The person skilled in the art will know how to adapt the cut-off frequency of the low-pass filter. The set time lapse depends on the noise amplitude on the signals. It is preferably at least one minute. The person skilled in the art will know how to adapt the set time lapse to the calibration conditions.

[0120] The calibration can be done only once, for example at the start of an atomizer or it can be done dynamically, for example at regular time intervals. The calibration is preferably done over a time lapse during which the regime is stable, i.e. Pinduction, Tnozzie and Treser oir are constant.

[0121] According to a second variant of the second embodiment, the flow rate Qmetai of molten metal is estimated on the basis of a mean of the ratio (1 ) and of a derivative

[0122] M reservoir of the mass M reservoir of the reservoir. In particular, the flow rate Qmetai of molten metal is estimated based on Formula (5): wherein M reservoir is a derivative of the mass Mreservoir of the reservoir

[0123] (preferably expressed in Kg-s_1) and wi and W2 are non-zero weighting coefficients.

[0124] Before derivation, the mass Mreservoir of the reservoir is preferably filtered with a low-pass filter to reduce noise. The person skilled in the art will know how to adapt the cut-off frequency of the low-pass filter.

[0125] The weighting coefficients can be assigned empirically, possibly depending on the operator's confidence in the respective measurements or on the standard deviation of the measurements.

[0126] According to a third variant of the second embodiment, the estimating step comprises a modeling step as described for the first embodiment.

[0127] In a first example, the flow rate Qmetai of molten metal is estimated based on a Kalman filter having:

[0128] - a state vector Xk containing the flow rate Qmetai of molten metal and the mass M reservoir of the reservoir and,

[0129] - a vector of measurements Zk containing the result of Formula (3) and the mass M reservoir of the reservoir: Where k is the discrete variable for the elapsed time t and At is the sampling time, so that t = k.At.

[0130] In this example, the state transition matrix Fk is preferably:

[0131] The control input matrix Bk is preferably equal to zero.

[0132] The measurement matrix Hk is preferably:

[0133] The person skilled in the art will know how to estimate the noise covariance matrices Qk and Rk and how to issue the best estimation xkof the states Xk based on the measurements, for each time-step k based on the Kalman filter theory.

[0134] In a variant of this first example, Formula (2) is used instead of Formula (3).

[0135] According to a third embodiment of the invention, the mass Mpowder of the powder produced in the atomization chamber is further taken into account to estimate the flow rate Qmetai of molten metal.

[0136] The collecting step of this third embodiment differs from the collecting step of the first embodiment in that it further includes collecting the mass Mpowder of the powder produced in the atomization chamber. Mpowder can be measured with a load cell or a strain gauge, the measurement being then collected, preferably by the monitoring device. In the case of a powder bin 8 positioned below the opening 7 at the bottom of the chamber and continuously collecting the powder, at least one bin load cell 15 can be positioned below the powder bin, or on the structure supporting the powder bin, to measure the weight Mbin of the powder bin. Mpowder is then deducted from Mbin. In cases where the powder is collected and discharged differently from the atomizer, the above principles can be adapted to the design of the atomizer. The atomizer is built so that the mass Mpowder of the powder produced in the atomization chamber can be collected. In particular, the bin load cell is interfaced with a computer or data logger, for instance the monitoring device 17.

[0137] The estimating step of this third embodiment differs from the estimating step of the first embodiment in that it further includes taking Mpowder into account to estimate the flow rate Qmetai of molten metal. Consequently, the flow rate Qmetai of molten metal is estimated more accurately. According to a first variant of the third embodiment, the flow rate Qmetai of molten metal is estimated based on Formula (6): wherein Kmpis a non-zero proportionality coefficient obtained by a calibration of the flow rate Qmetai of molten metal based on a variation of the mass Mpowder of the powder produced in the atomization chamber over a set time lapse.

[0138] Before calibration, the mass Mpowder of the powder produced in the atomization chamber is preferably filtered with a low-pass filter to reduce noise. The person skilled in the art will know how to adapt the cut-off frequency of the low-pass filter.

[0139] The set time lapse depends on the noise amplitude on the signals. It is preferably at least one minute. The person skilled in the art will know how to adapt the set time lapse to the calibration conditions.

[0140] The calibration can be done only once, for example at the start of an atomizer or it can be done dynamically, for example at regular time intervals. The calibration is preferably done over a time lapse during which the regime is stable, i.e. Pinduction, Tnozzie and Treservoir are constant.

[0141] According to a second variant of the third embodiment, the flow rate Qmetai of molten metal is estimated on the basis of a mean of the ratio (1 ) and of a derivative Mpowder of the mass Mpowder of the powder produced in the atomization chamber. In particular, the flow rate Qmetai of molten metal is estimated based on Formula (7): wherein Mpowder is a derivative of the mass Mpowder of the powder produced in the atomization chamber (preferably expressed in Kg-s-1) and W3 and W4 are nonzero weighting coefficients.

[0142] Before derivation, Mpowder is preferably filtered with a low-pass filter to reduce noise. The person skilled in the art will know how to adapt the cut-off frequency of the low-pass filter.

[0143] The weighting coefficients can be assigned empirically, possibly depending on the operator's confidence in the respective measurements or on the standard deviation of the measurements. According to a third variant of the third embodiment, the collecting step further includes collecting the mass Mreservoir of the reservoir, in addition to Mpowder. Accordingly, the flow rate Qmetai of molten metal is estimated on the basis of a mean of the ratio (1 ), of a derivative Mreservoir of the mass Mreservoir of the reservoir and of a derivative Mpowder of the mass Mpowder of the powder produced in the atomization chamber. In particular, the flow rate Qmetai of molten metal is estimated based on Formula (8): wherein ws, we and w? are non-zero weigthing coefficients.

[0144] Before derivation, Mreservoir and Mpowder are preferably filtered with a low-pass filter to reduce noise.

[0145] According to a fourth variant of the third embodiment, the estimating step comprises a modeling step as described for the first embodiment.

[0146] In a first example, the flow rate Qmetai of molten metal is estimated based on a Kalman filter having:

[0147] - a state vector Xk containing the flow rate Qmetai of molten metal and the mass Mpowder of the powder produced in the atomization chamber and,

[0148] - a vector of measurements Zk containing the results of Formula (3) and the mass Mpowder of the powder produced in the atomization chamber:

[0149] In this example, the state transition matrix Fk is preferably:

[0150] The control input matrix Bk is preferably equal to zero.

[0151] The measurement matrix Hk is preferably:

[0152] "H; a

[0153] The person skilled in the art will know how to estimate the noise covariance matrices Qk and Rk.

[0154] In a variant of this first example, Formula (2) is used instead of Formula (3).

[0155] In a second example, the flow rate Qmetai of molten metal is estimated based on a Kalman filter having: - a state vector Xk containing the flow rate Qmetai of molten metal, the mass M reservoir of the reservoir and the mass Mpowder of the powder produced in the atomization chamber and,

[0156] - a vector of measurements Zk containing the result of Formula (3), the mass Mreservoir of the reservoir and the mass Mpowder of the powder produced in the atomization chamber:

[0157] In this example, the state transition matrix Fk is preferably:

[0158] The control input matrix Bk is preferably equal to zero.

[0159] The measurement matrix Hk is preferably:

[0160] The person skilled in the art will know how to estimate the noise covariance matrices Qk and Rk.

[0161] In a variant of this second example, Formula (2) is used instead of Formula (3).

[0162] According to a fourth embodiment of the invention, the level Hreservoir of the reservoir is further taken into account to estimate the flow rate Qmetai of molten metal.

[0163] The collecting step of this fourth embodiment differs from the collecting step of the first embodiment in that it further includes collecting the level of molten metal Hreservoir in the reservoir. The level can be measured with a level sensor, the measurement being then collected, preferably by the monitoring device. The level sensor can be notably a metal sensing electrode, a float control system, a laser system, an eddy-current sensor, a radar sensor.

[0164] Accordingly, the reservoir can thus be equipped with a level sensor 16, as described above. The level sensor can be fixed on the reservoir, for example on the side or on the top, or it can be suspended in the reservoir above the meniscus or above the reservoir. The atomizer is built so that the level Hreservoir of the reservoir can be collected. In particular, the level sensor is interfaced with a computer or data logger, for instance the monitoring device 17.

[0165] The estimating step of this fourth embodiment differs from the estimating step of the first embodiment in that it further includes taking Hreservoir into account to estimate the flow rate Qmetai of molten metal. Consequently, the flow rate Qmetai of molten metal is estimated more accurately.

[0166] According to a first variant of the fourth embodiment, the flow rate Qmetai of molten metal is estimated based on Formula (9): wherein Khr is a non-zero proportionality coefficient obtained by a calibration of the flow rate Qmetai of molten metal based on a variation of the level Hreservoir of molten metal in the reservoir over a set time lapse.

[0167] Before calibration, the level Hreservoir of the reservoir is preferably filtered with a low-pass filter to reduce noise. The person skilled in the art will know how to adapt the cut-off frequency of the low-pass filter.

[0168] For the calibration, the section Sreservoir of the reservoir and the density pmetai of the molten metal are known, knowing that the section can possibly vary with the level of molten metal.

[0169] The set time lapse depends on the noise amplitude on the signals. It is preferably at least one minute. The person skilled in the art will know how to adapt the set time lapse to the calibration conditions.

[0170] The calibration can be done only once, for example at the start of an atomizer or it can be done dynamically, for example at regular time intervals. The calibration is preferably done over a time lapse during which the regime is stable, i.e. Pinduction, Tnozzie and Treser oir are constant.

[0171] According to a second variant of the fourth embodiment, the flow rate Qmetai of molten metal is estimated on the basis of a mean of the ratio (1 ) and of a derivative H reservoir of the level H reservoir of molten metal in the reservoir. In particular, the flow rate Qmetai of molten metal is estimated based on Formula (10): wherein H reservoir is a derivative of the level H reservoir of molten metal in the reservoir (preferably expressed in m •S"1), S reservoir is the section of the reservoir (preferably expressed in m2), pmetai is the density of the molten metal (preferably expressed in Kg-rrr3) and ws and W9 are non-zero coefficients.

[0172] Before derivation, Hreservoir is preferably filtered with a low-pass filter to reduce noise. The person skilled in the art will know how to adapt the cut-off frequency of the low-pass filter.

[0173] The weighting coefficients can be assigned empirically, possibly depending on the operator's confidence in the respective measurements or on the standard deviation of the measurements.

[0174] According to a third variant of the fourth embodiment, the collecting step includes collecting the mass Mreservoir of the reservoir, in addition to Hreservoir. Accordingly, the flow rate Qmetai of molten metal is estimated on the basis of a mean of the ratio (1 ), of a derivative Mreservoir of the mass Mreservoir of the reservoir and of a derivative H reservoir of the level H reservoir of molten metal in the reservoir. In particular, the flow rate Qmetai of molten metal is estimated based on Formula (11 ):

[0175] Qmetal wherein w , wn and W12 are non-zero weighting coefficients.

[0176] Before derivation, Hreservoir and M reservoir are preferably filtered with a low-pass filter to reduce noise.

[0177] According to a fourth variant of the fourth embodiment, the collecting step includes collecting the mass Mpowder of the powder produced in the atomization chamber, in addition to Hreservoir. Accordingly, the flow rate Qmetai of molten metal is estimated on the basis of a mean of the ratio (1 ), of a derivative Mpowder of the mass Mpowder of the powder produced in the atomization chamber and of a derivative H reservoir of the level H reservoir of molten metal in the reservoir. In particular, the flow rate Qmetai of molten metal is estimated based on Formula (12): wherein W13, wu and wis are non-zero weighting coefficients.

[0178] Before derivation, Hreservoir and Mpowder are preferably filtered with a low-pass filter to reduce noise.

[0179] According to a fifth variant of the fourth embodiment, the collecting step further includes collecting the mass Mreservoir of the reservoir and the mass Mpowder of the powder produced in the atomization chamber, in addition to Hreservoir. Accordingly, the flow rate Qmetai of molten metal is estimated on the basis of a mean of the ratio (1 ), of a derivative Mreservoir of the mass Mreservoir of the reservoir, of a derivative Mpowder of the mass Mpowder of the powder produced in the atomization chamber and of a derivative H reservoir of the level H reservoir of molten metal in the reservoir. In particular, the flow rate Qmetai of molten metal is estimated based on Formula (13):

[0180] Qmetal wherein w , W17, wis and W19 are non-zero coefficients.

[0181] Before derivation, Hreservoir, Mreservoir and Mpowder are preferably filtered with a low-pass filter to reduce noise.

[0182] According to a sixth variant of the fourth embodiment, the estimating step comprises a modeling step as described for the first embodiment.

[0183] In a first example, the flow rate Qmetai of molten metal is estimated based on a Kalman filter having:

[0184] - a state vector Xk containing the flow rate Qmetai of molten metal and the level H reservoir of molten metal in the reservoir and,

[0185] - a vector of measurements Zk containing the result of Formula (3) and the level H reservoir of molten metal in the reservoir:

[0186] In this example, the state transition matrix Fk is preferably: With Sreservoir the section of the reservoir and pmetai the density of the molten metal.

[0187] The control input matrix Bk is preferably equal to zero.

[0188] The measurement matrix Hk is preferably: 3

[0189] The person skilled in the art will know how to estimate the noise covariance matrices Qk and Rk.

[0190] The above example can be modified based on the principles presented in the second example of the fourth variant of the third embodiment to further take into account the mass Mreservoir of the reservoir and / or the mass Mpowder of the powder produced in the atomization chamber.

[0191] Thanks to the accurate estimation of the flow rate Qmetai of molten metal, it is possible to accurately manage the fluid-to-metal ratio during atomization, in particular to maintain the fluid-to-metal ratio at a set value or in a set range, notably despite variations of the flow rate of molten metal.

[0192] The method for managing the fluid-to-metal ratio during atomization of molten metal according to the invention comprises a first step of setting. As slight deviations of the fluid-to-metal ratio are usually acceptable from a quality perspective, the set range Rsof fluid-to-metal ratio is set. In addition, as it is favorable to target a fixed fluid-to-metal ratio, a set fluid-to-metal ratio FMRs is preferably set. The set range Rs of fluid-to-metal ratio can be entered as a range as such, with a minimal fluid-to- metal ratio and a maximal fluid-to-metal ratio or it can be entered as a standard deviation of the set fluid-to-metal ratio FMRs, if applicable. Of course, if for some reason, slight deviations have to be avoided, the set fluid-to-metal ratio FMRs can be entered as the minimal fluid-to-metal ratio and a maximal fluid-to-metal ratio or the standard deviation can be set at zero. The set range Rsand the set fluid-to-metal ratio FMRs can depend on the composition of the molten metal and on the expected particle size distribution of the powder obtained from the molten metal.

[0193] From a practical point of view, these two parameters can be manually entered in the monitoring device. Alternatively, they can be automatically obtained by the monitoring device from the management tool of the atomizer or from the order book of the atomizer, in particular from the scheduling tool.

[0194] The second step of the method for managing the fluid-to-metal ratio during atomization of molten metal corresponds to the collecting step of the method for dynamically estimating the flow rate Qmetai of molten metal flowing from a reservoir, as detailed previously.

[0195] In this second step, the temperature Treservoir of the molten metal in the reservoir, the temperature Tnozzie of the molten metal in a portion comprised between the exit of the channel and the exit of the nozzle and the instantaneous power Pinduction of the induction heater are collected. Optionally, the mass Mreservoir of the reservoir, the mass Mpowder of the powder produced in the atomization chamber, the level Hreservoir of molten metal in the reservoir or a combination of them is also collected. Preferably, the fluid flow rate is also collected, more preferably from the flow sensor 6. Preferably, the monitoring device is configured to collect these data.

[0196] The third step of the method for managing the fluid-to-metal ratio during atomization of molten metal corresponds to the estimating step of the method for dynamically estimating the flow rate Qmetai of molten metal, as detailed previously. Overall, during this step, the flow rate Qmetai of molten metal is estimated based on ratio (1 ):

[0197] _ P induction _

[0198] ( nozzle~T reservoir)

[0199] In other words, the second step and the third step correspond to the method for dynamically estimating the flow rate Qmetai of molten metal. The details provided when describing this estimating method apply to the second step and the third step of the managing method.

[0200] In a fourth step of the method for managing the fluid-to-metal ratio during atomization of molten metal, the fluid-to-metal ratio is calculated from the flow rate Qmetai of molten metal estimated at the previous step. The fluid flow rate used for the calculation can be the set value of the fluid flow rate or the measured value of the fluid flow rate measured by the flow sensor 6, each value being preferably collected with the monitoring device. Preferably this step is done with the monitoring device. In a fifth step of the method for managing the fluid-to-metal ratio during atomization of molten metal, a possible deviation of the calculated fluid-to-metal ratio beyond the set range Rsof fluid-to-metal ratio is corrected.

[0201] At first, a possible deviation of the fluid-to-metal ratio is assessed by comparing the calculated fluid-to-metal ratio to the set range Rsof fluid-to-metal ratio and / or to the set fluid-to-metal ratio FMRs. If the calculated fluid-to-metal ratio is still within the set range Rs, the settings are maintained. If the calculated fluid-to-metal ratio has deviated beyond the set range Rs, an atomization setting is adjusted. In a first variant, the fluid flow rate in the fluid sprayer 4 is adjusted. In particular, the fluid flow rate in the fluid sprayer is adjusted by adjusting a setting of the fluid regulator 5. In a second variant, the pressure in the reservoir is adjusted.

[0202] Optionally, before the adjustment is done, an adjusted atomization setting, for correcting the deviation, is determined, preferably with the monitoring device.

[0203] The adjustment can be done manually, for example by an operator of the atomizer. Alternatively, the correction is done with a closed-loop controller. Preferably this controller is a proportional integral derivative controller. As, in some cases, only one or two terms of the controller can provide appropriate control, some parameters of the controller can be set to zero to inactivate some terms. In other words, the proportional integral derivative controller can be a P controller, an I controller, a D controller, a PI controller, a PD controller, an ID controller or a PID controller.

[0204] For example, the correction CF to be applied to the fluid flow rate, in particular to the fluid regulator, can notably be calculated according to an equation of general formula: or where e is the error, i.e. the difference between the set fluid-to-metal ratio

[0205] FMRs and the calculated fluid-to-metal ratio, t is the time or instantaneous time and T is the variable of integration. Other control functions are well known from the person skilled in the art and can be used for managing the fluid-to-metal ratio during atomization of molten metal.

[0206] Thanks to the adjustment, the fluid-to-metal ratio falls again within the set range Rsand preferably converges towards the set fluid-to-metal ratio FMRs, more preferably reaches the set fluid-to-metal ratio FMRs.

Claims

CLAIMS1 ) Method for dynamically estimating a flow rate Qmetai of molten metal flowing from a reservoir (9) to a nozzle (3) of an atomization chamber (2) of an atomizer (1 ) connected to the reservoir through a channel (10) equipped with an induction heater (12), the method comprising the steps of:- Collecting a temperature Treservoir of the molten metal in the reservoir, a temperature Tnozzie of the molten metal in a portion comprised between the exit of the channel and the exit of the nozzle, and an instantaneous power Pinduction of the induction heater,- Estimating the flow rate Qmetai of molten metal based on ratio (1 ):_ P induc ion _(Tnozzle~ Treservoir)2) Method according to claim 1 wherein the flow rate Qmetai of molten metal is estimated based on Formula (2):wherein Cp.metai is the specific heat of the molten metal.3) Method according to claim 1 wherein the collecting step further includes collecting the mass Mreservoir of the reservoir and the estimating step further includes taking Mreservoir into account to estimate the flow rate Qmetai of molten metal.4) Method according to claim 3 wherein the flow rate Qmetai of molten metal is estimated based on Formula (5):wherein Cp.metai is the specific heat of the molten metal, Mreservoir is a derivative of the mass M reservoir of the reservoir and wi and W2 are non-zero coefficients.5) Method according to claim 3 wherein the flow rate Qmetai of molten metal is estimated based on a Kalman filter having:- a state vector containing the flow rate Qmetai of molten metal and the mass M reservoir of the reservoir and,- a vector of measurements containing the result of Formula (3) and the mass M reservoir of the reservoir:wherein Cp.metai is the specific heat of the molten metal and Ki is a non-zero proportionality constant.6) Method according to claim 1 wherein the collecting step further includes collecting the mass Mpowder of the powder produced in the atomization chamber and the estimating step further includes taking Mpowder into account to estimate the flow rate Qmetai of molten metal.7) Method according to claim 6 wherein the flow rate Qmetai of molten metal is estimated based on Formula (7):wherein Cp.metai is the specific heat of the molten metal, Mpowder is a derivative of the mass Mpowder of the powder produced in the atomization chamber and W3 and W4 are non-zero coefficients.8) Method according to claim 6 wherein the flow rate Qmetai of molten metal is estimated based on a Kalman filter having:- a state vector containing the flow rate Qmetai of molten metal and the mass Mpowder of the powder produced in the atomization chamber and,- a vector of measurements containing the result of Formula (3) and the mass Mpowder of the powder produced in the atomization chamber:wherein Cp.metai is the specific heat of the molten metal and Ki is a non-zero proportionality constant.9) Method according to any one of claims 6 to 8 wherein the atomizer further comprises a powder bin connected to the bottom of the atomization chamber and collecting the powder produced in the atomization chamber and wherein Mpowder is deducted from the mass Mbin of the powder bin.10)Method according to claim 1 wherein the collecting step further includes collecting the level Hreservoir of molten metal in the reservoir and the estimating step includes taking Hreservoir into account to estimate the flow rate Qmetai of molten metal.11 )Method according to claim 10 wherein the flow rate Qmetai of molten metal is estimated based on Formula (10):wherein Cp.metai is the specific heat of the molten metal, H reservoir is a derivative of the level H reservoir of molten metal in the reservoir, Sreservoir is the section of the reservoir, pmetai is the density of the molten metal and ws and W9 are nonzero coefficients.12)Method according to claim 10 wherein the flow rate Qmetai of molten metal is estimated based on a Kalman filter having:- a state vector containing the flow rate Qmetai of molten metal and the level H reservoir of molten metal in the reservoir and,- a vector of measurements containing the result of Formula (3) and the level H reservoir of molten metal in the reservoir:wherein Cp.metai is the specific heat of the molten metal and Ki is a non-zero proportionality constant.)Method according to any one of the preceding claims wherein, in the collecting step, the temperature Treservoir of the molten metal in the reservoir is obtained from a molten metal temperature measurement device (11 ) which is a part of the reservoir (9), the temperature Tnozzie of the molten metal in a portion comprised between the exit of the channel and the exit of the nozzle is obtained from a metal stream temperature measurement device (13) which is a part of the atomizer (1 ) and the instantaneous power Pinduction of the induction heater is obtained from an induction generator controlling the induction heater (12). )Method for managing the fluid-to-metal ratio during atomization of molten metal firstly flowing from a reservoir (9) to a nozzle (3) of an atomization chamber (2) of an atomizer (1 ) connected to the reservoir through a channel (10) equipped with an induction heater (12), and secondly being impinged by a fluid jetted from a fluid sprayer (4), the method comprising the steps of:- Setting a set range Rsof fluid-to-metal ratio,- Collecting a temperature Treservoir of the molten metal in the reservoir, a temperature Tnozzie of the molten metal in a portion comprised between the exit of the channel and the exit of the nozzle and an instantaneous power Pinduction of the induction heater,- Estimating the flow rate Qmetai of molten metal based on ratio (1 ):_ induction _(Tnozzle~reservoir)- Calculating the fluid-to-metal ratio from the estimated flow rate Qmetai of molten metal,- Correcting a deviation of the calculated fluid-to-metal ratio beyond the set range Rsof fluid-to-metal ratio by adjusting an atomization setting. )Monitoring device (17) comprising at least a processor and a memory, configured to execute the following steps:- Collecting: o a temperature Treservoir of a molten metal in a reservoir (9),o a temperature Tnozzie of the molten metal in a portion comprised between an exit of a channel (10) connecting the reservoir to a nozzle (3) of an atomization chamber (2) of an atomizer (1 ) and an exit of the nozzle, and o an instantaneous power Pinduction of an induction heater (12) equipping the channel,- Estimating a flow rate Qmetai of molten metal flowing from the reservoir to the nozzle based on ratio (1 ):_ P induction _(1 ).( nozzle~T reservoir) )Monitoring device according to claim 15, further configured to execute the following steps:- Collecting a set range Rsof fluid-to-metal ratio,- Calculating a fluid-to-metal ratio from the estimated flow rate Qmetai of molten metal, and from a flow rate of a fluid jetted from a fluid sprayer to impinge the molten metal flowing out of the nozzle,- Determining an adjusted atomization setting, for correcting a deviation of the calculated fluid-to-metal ratio beyond the set range Rsof fluid-to-metal ratio. )Atomizer (1 ) comprising a reservoir (9) and an atomization chamber (2), the reservoir being connected to a nozzle (3) of the atomization chamber through a channel (10) equipped with an induction heater (12), the atomizer further comprising:- a molten metal temperature measurement device (11 ) installed so as to measure a temperature Treservoir of the molten metal in the reservoir,- a metal stream temperature measurement device (13) installed so as to measure a temperature Tnozzie of the molten metal in a portion comprised between an exit of the channel (10) and an exit of the nozzle (10),- an induction generator for controlling the induction heater,- the monitoring device (17) of claim 15 or 16, connected to the molten metal temperature measurement device (11 ) and to the metal streamtemperature measurement device (13) and interfaced with the induction generator. )Computer program comprising instructions, whose execution on a computer make the computer to execute the steps of the method according to any one of claims 1 to 14.

Images