GB2582807A - A method of thermal spraying - Google Patents

Abstract

A method of controlling a thermal spray apparatus comprises using a thermal spray gun 12 to conduct a plurality of sprays to produce a coating on at least one test substrate, altering the variables of the gun and sampling acoustic emissions for each spray, testing the coatings to determine at least one property of each and linking variations in coating properties to gun variables, spraying production articles 14 with the gun whilst using a processor 28 to sample acoustic emissions and alter gun variables depending on the sampled emissions to vary or maintain the properties of the coating. The quality of a coating can thus be maintained. The linking of variations in properties may be undertaken by Fourier transform analysis or by neural network data analysis of the acoustic emissions. The spray gun is preferably a high velocity oxy-fuelled (HVOF) gun. The gun variables may comprise at least one of gas flow rates, gas pressure, gun to substrate standoff distance, particle feed rate, substrate temperature, substrate angle and pass number. The properties of the coatings may comprise hardness, microhardness, porosity, cracking and carbide content.

Description

A Method of Thermal Spraying The present invention relates to a method of thermal spraying and relates particularly, but not exclusively, to a method of controlling a thermal spraying apparatus for maintaining the properties of a coating during a thermal spraying process.

Thermal spraying is a series of well-known techniques used to coat components with protective coatings. These techniques use the combustion of a fuel to propel particles of a coating material at high speed against an article which is to be coated. The particles are often fired through a nozzle which acts as the exhaust of a combustion chamber into which the coating material is made. Over time, the aperture, which forms this nozzle can become clogged with coating material which is not successfully exited the combustion chamber and/or eroded. Thermal fatigue causes dimensional changes to the nozzle and also OEM or other spare parts might not be manufactured within the preferred tolerances for optimum performance. Eventually one of the above or combination of many factors alters the shape of the nozzle exhaust causing variations in the properties of the coating particles leaving the spray gun which in turn varies the quality of the coating on the article. There is therefore a need to regularly interrupt the spraying process to replace or clean the nozzle of the gun, leading to loss of production and reduced product quality if problems are not identified early enough in the process by the operator.

Although many spraying systems are undertaken using robot 30 controllers for the spray guns, it is not uncommon for thermal spray techniques to be undertaken manually. Operators of such manual sprays are highly skilled and produce coatings with -2 -consistent properties. However, there are no straightforward techniques for monitoring manual spraying processes to ensure quality. Furthermore, article geometries can limit the movement of a spray gun which in turn can impact the coating quality.

For example, manual spraying is commonly used when coating boiler tubes in a power plant. However, due to space constraints, operators are not always able to maintain the same stand-off distance, that is the distance between the nozzle of the spray gun and the substrate, and this results in variations in the quality of the coating.

Existing thermal spray monitoring techniques include visual-based systems which use a camera to watch the spraying process, in particular the flame. However, the use of these techniques is limited to static systems, making them unsuitable for use with robots or manual spraying techniques. Research into acoustic emissions from the thermal spray process is limited and has only focused on the use of vibration sensors in contact with the sprayed substrate to monitor the kinetic properties of particle deposition. These methods have shown promise in accurately monitoring the quality of thermal spray coatings as they are being sprayed, but the contact sensor must be carefully positioned on the substrate and calibrated for every part being sprayed. This drastically increases spraying time for each part and would necessitate training operators in the use of the monitoring equipment. When complex geometries are sprayed having the entire surface covered by a spray coated material it is not possible to attach a sensor on the surface of the part. Furthermore, some parts get too hot during spray and the sensors would not survive the process heat. All the above make the acoustic contact sensors impractical and expensive to use. -3 -

Preferred embodiments of the present invention seek to overcome or alleviate the above described disadvantages of the prior art.

According to an aspect of the present invention there is 5 provided a method of controlling a thermal spray apparatus comprising the steps: conducting a plurality of sprays using a thermal spray gun to produce a coating on at least one substrate using a thermal spray gun, altering the variables of the gun for each spray and 10 sampling acoustic emissions during said sprays; testing the coatings to determine at least one property of each coating and linking variations in said coatings properties to said gun variables used; and spraying articles with said gun whilst sampling acoustic 15 emissions and altering the gun variables dependent on the sampled acoustic emissions to vary or substantially maintain the properties of the coating of the article.

By undertaking a plurality of test sprays onto a substrate and then analysing the coating sprayed onto the substrate and linking properties of the coating with acoustic omissions sampled during the spraying process, it is possible to link characteristics of the acoustic emissions with the properties of the coating. This, in turn, then allows analysis of acoustic omissions during the spraying of an article to be monitored, allowing the property of the coating to be maintained or varied as required by changing the variables of the spray gun. As a result, the quality of a coating can be maintained when the acoustic omissions are indicating that the properties of the coating are changing. By changing the variables of the spray gun. For example, as a nozzle is used some of the features of the gun can alter such as the nozzle becoming worn or partially blocked or dimensional changes due to thermal fatigue. This is -4 -known to alter the characteristics of the spray guns flame and the coating material as it exits the gun. However, the resultant change in the way in which the coating particles impact the substrate surface can be compensated for by changing variables 5 of the gun in use. Changes in the nozzle to substrate standoff distance, gas feed rate, gas pressure and coating feed rate, amongst others, can compensate for this change in coating characteristic or property. Similarly, if a variation in the coating characteristic is required then this can also be 10 achieved.

This monitoring technique can be used across many thermal spray techniques and can be used on manual, robotic, controlled and static spray systems. Furthermore, the monitoring can work on articles with complex geometries ensuring the quality of coating applied to all articles. Where a manual or robot controlled spray operation is being undertaken and geometries of the article prevent the optimum stand-off distance being used, for example in boiler tubes, other gun variables can be altered to compensate for the change in stand-off distance whilst maintaining the properties of the coating applied.

In a preferred embodiment the linking of variations in properties in said coating to said gun variables is undertaken on selected frequency ranges.

By selecting frequency ranges and by conducting Fourier 25 transform analysis on the acoustic omissions, and in particular where it is fed into a machine learning algorithm trained to recognise acoustic omission panerns, provides an effecLive means for analysing the acoustic omissions. [Can we be more specific about the advantages of using Fourier transform analysis?] The use of a smart AI algorithm to corelate the acoustic emissions with the coating properties provides accurate links between the gun variables and the coating parameters. The machine learning algorithm is then able to return altered -5 -parameters to the console in order to maintain the coating quality during spray of the articles.

In another preferred embodiment the linking of variations in properties in said coating to said gun variables is undertaken 5 by Fourier transform analysis of said acoustic emissions.

The plurality of sprays may preferably be onto a plurality of substantially identical substrates and there substantially identical substrates may further preferably be planar.

In a preferred embodiment the gun variables comprise at 10 least one of gas flow rates, gas pressure, gun to substrate standoff distance, particle feed rate, substrate temperature, substrate angle and pass number.

In another preferred embodiment the testing is destructive.

In a further preferred embodiment the spray gun is a High 15 Velocity Oxy-Fuelled spray gun.

The property of the coating may be one or more of hardness, micro hardness, porosity, cracking and carbide content.

Preferred embodiments of the present invention will now be described, by way of example only, and not in any limitative 20 sense with reference to the accompanying drawings in which:-Figure 1 is a schematic representation of the apparatus of used in connection with the present invention; Figure 2 is a schematic representation of the process undertaken in the present invention; and Figure 3 is a schematic representation of the neural network architecture which is used in the process undertaken in the present invention.

Referring to figure 1, a thermal spray apparatus 10 includes a thermal spray gun 12 which is used to apply a thermal spray coating to a substrate or article 14. Throughout this -6 -application the term substrate is used to refer to a test sample which is used during the initial calibration of the apparatus while the term article is used to refer to product to which a coating is to be applied.

The present invention can be used with any form of thermal spray apparatus including, but not limited to, High Velocity Oxygen Fuel (HVOF) spraying, plasma spraying, detonation spraying, wire arc spraying, flame spraying, high velocity air fuelled (HVAF) spraying, warm spraying, cold spraying and laser cladding. The thermal spray gun 12 operates within a booth 16 and the gun is mounted to a robot arm 18 which is controlled from a console 20. The console allows both manual and automated control of the robot arm 18. The spray gun 12 is fuelled by a combustible fuel, such as kerosene, propane, propylene, natural gas, hydrogen, ethylene, butane and the like, which is burnt within a combustion chamber in the spray gun 12. The gas is burnt with oxygen which is either supplied as substantially pure oxygen from tanks or as oxygen in the air. The supply of fuel and oxygen is controlled by a gas console 22 which, by controlling the pressure and flow rates of the fuel and oxygen, control the temperature and other properties of the flame produced by the gun 12. Also under the control of the gas console is the coating feedstock feed rate which provides the particulate matter that is sprayed from the gun 12 to form the coating on the article 14. For many of the spray techniques listed above the feedstock is powder but some techniques use wire feedstocks, liquid precursors, solutions or suspensions. Where powder is used it is fed into the combustion chamber of the gun 12 where it is heated before exiting through the nozzle and being directed onto the substrate or article being sprayed. These components described above are standard to existing thermal spray apparatus.

In addition to the apparatus previously described one or more airborne acoustic sensors in the form of a microphone 24 are provided and contained within the thermal spray booth 16. For example, a microphone can be used and specifically a random incidence, high frequency, high amplitude, pre-polarized microphone with a preamplifier system such as the PCB Piezotronics Model 378A21. This microphone has a frequency range from 4Hz to 25kHz and a distortion limit of 160dB with a noise floor of 19dBA. The acoustic sensors 24 are connected to an acoustic signal and acquisition processor 26, which uses the signals from the sensors and processes them in accordance with the method set out below. Data processed by the signal acquisition and processor 26 is then passed to the process control unit 28 which, via the robot console and gas console, controls the operation of the gun 12 and the feeds thereto. Although indicated as separate processes, the signal acquisition processor 26 and control processor 28 can be provided in a single processing computer device.

Operation of the apparatus set out above, will now be described with additional reference to figure 2. The purpose of the present invention is to improve the quality of coating which is applied to an article, in particular by improving the consistency of the properties of the coating. This is achieved by undertaking testing and production phases, with the testing phases used to determine gun variables for use during the production phase. Testing is already generally undertaken when performing any thermal spraying process to ensure that the coating will meet the specification of coating properties required for the article before spraying of the article is undertaken. More preferably the process is divided into three phases with a recalibration phase undertaken (perhaps daily) in addition to the initial testing (calibration) and subsequent production phases. The frequency of recalibration of the -8 -testing for the present invention is typically similar to that already undertaken for existing thermal spraying processes.

During the testing the coating is sprayed onto a substrate and this is typically a rectangular plate of the material of which 5 the article is formed.

A series of substrates are used and test sprays undertaken with different variables for the spray gun applied to each test spray. These variables include, but are not limited to standoff distance, powder feed rate, particle size, gas flow rates fuel to oxygen ratio, combustion chamber pressure and exhaust aperture. One or more of these variables are changed with tests undertaken at each of a plurality of the predetermined values (that is predetermined distances, feed rate or ratios). For example, if stand-off distance and powder feed rate are the two variables of particular interest and there are four stand-off distances tested and three powder feed rate tested there are 12 possible combinations of feed rates and distances which can be tested. This example is simplified for the purposes of illustration and typically multiple variables with larger numbers of ranges can be included in the test phase, although this increases the number of test samples which must be undertaken and then analysed.

During the spraying process the acoustic sensors are used to detect the acoustic emissions resulting from the spraying 25 process and the signal stored in the acoustic emission signal acquisition process 26.

Once the substrates have been sprayed, the coating as sprayed onto the substrate, is tested to determine at least one property of the coating. This is achieved by cutting and polishing the substrate with cross sections of the sample being examined under an optical microscope. Porosity measurements are made with a standardised batch routine using image analysis. _ 9 _

Furthermore, micro-hardness is examined with a Vickers micro-inventor under a load of 300g. Other coating properties may be tested including, but not limited to bond strength, wear tests, bend testing, corrosion and any other property that is of interest to a specific application of the coating. These properties of the coating are entered into the acoustic emission signal acquisition and processor 26 and matched with the acoustic emission data for that sample.

Once the sampling and testing stages are completed, analysis is then undertaken linking the data gathered by the acoustic sensors 24 with the properties of the coating and determining how those properties depend upon the variables used for each testing substrate. In particular, variations in the strength of acoustic emission signal detected at a particular frequency, or frequency range, are linked to variations in the variables used in the test spraying and to the resultant properties of the test coatings.

For example, as the stand-off distance is varied between samples, the acoustic emissions at a particular frequency may vary and this variation can be linked to changes in the properties of the coating on the substrate. The intensity of acoustic emission at a particular frequency may increase as the stand-off distance increases and this can be linked to changes in the properties of the coating, such as microhardness and porosity. This allows a link between acoustic emission intensity at one or more frequencies and microhardness to be established and then the microhardness maintained as changes in the acoustic emissions indicate a change microhardness.

The above described analysis is most efficiently undertaken 30 by Fourier transform analysis and the use of a machine learning algorithm.

-10 -By way example only a machine learning algorithm used in the operation of the present invention is a multilayer perceptron used to model the Acoustic Emission data for the HVOF spraying process. The analysis employs, for example and as illustrated in figure 3, a total number of 12 input nodes corresponding to the spectral density at key frequencies bands and their corresponding spectral density available when the spray gun variables are altered during spray. These nodes are indicated in column A of figure 3 with 10 frequency bands listed as Peaks 1-10 and the two remaining peaks representing the Powder Feed Rate (PFR) and Standoff Distance (SOD) variables. Figure 3 contains a scaling layer, a neural network and an unscaling layer with the circles in column A representing scaling neurons, the circles in columns B and C representing perceptron neurons, the circles in column D representing unscaling neurons, and the circles in column E representing bounding neurons. The number of inputs to the scaling neurons in column A is 12, and the number of outputs and bounding neurons (column E) are 2. The complexity, represented by the numbers of hidden neurons (the perceptron neurons in column B), is 7.

Once the sampling and analysis stages have been completed, production can be commenced.

Referring to figure 2, the testing procedure described above is illustrated as step Sl. The specification of the coating, that is the acceptable range of the properties of the coating, is then defined at step S2, and the geometry of the article to be coated is analysed at step S3. These steps S2 and S3 are standard procedures familiar to persons skilled in the art and lead to a definition of the input gas recipe and coating feedstock, step S4, including, for example, powder particle size and gas to oxygen ratio. At step SS, the input path for the robot is defined for the article in question and as a result, the spraying process is ready to start.

At step S6, the thermal spray torch 12 begins applying the coating to the article 14 at step S7. These steps S4 to S7 are also familiar to persons skilled in this art.

However, at step S8 the acoustic acquisition and signal processing is undertaken. The processor 26 analyses the airborne acoustic signals from the sensors 24 undertaking the same processing as during the testing phase to identify signal strengths at particular frequencies and using these to determine the coating properties at step S9.

Whilst these calculated properties of the coating on the article (derived from the acoustic emission analysis) remain comfortably within the specification of coating properties set at step S2, there is no need to alter the variables of the spray gun. However, as the analysis of the acoustic emissions indicates changes in the properties it is possible to alter the variables of the spray gun to alter the properties of the coating, thereby maintaining those properties within specification defined at step S2. When it is determined, at step S9, that the properties are outside, or too close to being outside, the specification, the processor 26 determines which variable of the spray gun should be altered in order to achieve the variation required. This decision is made by the artificial Neural network algorithm at step S10. The algorithm is trained to link millions of spray parameters with acoustic signals and their corresponding coating properties. This algorithmic iterative process takes place every 1 second. The self-adjusting process has a response time of 1 sec and the Neural Network return in real time (every 1 sec) the updated spray parameters in order to maintain the desired acoustic signal footprint that in turn corresponds to a desired coating property.

For example, if it is decided that the properties can be maintained by adjusting the gas recipe or feedstock (step S10) and these fall within the limits allowable by the console -12 - (checked at step S11), the gas and feedstock input recipe is varied by the gas console 22. At step S4, and the thermal spraying, leading to coating the position continues (at steps S6 and S7).

Alternatively, if it is determined that kinematic adjustments will maintain the coating properties within the specification, then these changes are implemented (at step S12). For example, the stand-off distance may be altered to maintain the coating properties. An additional check is undertaken at step S13 where the proposed change is simulated to identify potential problems such as exceeding maximum allowable temperatures. If no such problems are identified, then the input robot path is altered at step S5, providing new data to the robot console 20 to control the robot arm 18. However, if the proposed change in path is identified as causing problems in the simulation (at step S13) then the process is terminated at step S14.

If at step S9, it is determined that the coating properties remain comfortably within the required specification, then 20 spraying continues at step S15 with the thermal spray gun 12 applying a coating at steps S6 and S7. However, if it is determined that after the adjustments, and any further potential adjustments, the calculated coating properties remain outside the specification, the process is terminated at step S14 and advanced troubleshooting and maintenance, for example, replacement of consumable components of the gun, such as the nozzle, can be undertaken at step S16 before recommencing the process.

A simple example of the process set out above can run as follows. After determining a coating specification, the most economical gas to oxygen ratio and powder feed rate which will produce a coating within the specification is determined and initially used in the spraying of an article. During use, the -13 -nozzle of the spray gun begins to become thermally fatigued which alters the nozzle outlet cross section and which in turn alters the acoustic emissions detected as well as changing the properties of the coating is applied. As this is detected by 5 the sensors and processed by the processor, the properties of the coating can be maintained by altering variables of the spray gun. For example, the powder feed rate may be increased, thereby increasing the cost of the coating application process, but this additional cost is acceptable as the spraying process can be 10 continued without interruption whilst maintaining the properties of the coating within the required specification.

The coating process should remain calibrated until a change is made to the gun. However, it is advisable to run recalibration sample or calibration check samples periodically and at least every time a component such as a nozzle is changed. For example, it is advisable that the booth operator runs a 1-minute non-destructive test every morning. The 1-minute long audio sample contains the process frequencies and the frequencies are analysed using Fourier Transform and then are fed into the neural network. A fast training algorithm synchronises the AE unit with the particular gun and the operator is ready to start spraying the article.

The following further advantages of the present invention are worth noting. The digital files containing the acoustic signals gathered during the process of the present invention can be coordinated centrally, allowing troubleshooting of operational problems to be addressed at a single location utilising the maximum amount of data possible. Furthermore, lessons learnt can be easily and efficiently distributed to other operators. The digital files can also be used and catalogued against the test spray coupons to enhance accreditation and audit systems, such as NADCAP.

-14 -It will be appreciated by persons skilled in the art that the above embodiments have been described by way of example only and not in any limitative sense, and that various alterations and modifications are possible without departure from the scope of the protection which is defined by the appended claims. For example, although in the vast majority of applications. The requirement of spraying is to ensure coating properties remaining within a specification, it is possible that the specification may require variations in the properties of the coating in different parts of the article and the present invention can be used to sample the acoustic emissions and ensure that, whilst spraying different parts of an article, the coatings are applied within the different required specifications for that portion. As a result, the present invention can be used to vary or substantially maintain the properties of the coating of an article. Although the apparatus described above utilises airborne acoustic sensors in the form of microphones, the method of this invention can utilise contact sensors. However, some of the problems of the prior art remain such as the impact of heat on the sensors.

Claims (11)

1. -15 -Claims 1. A method of controlling a thermal spray apparatus comprising the steps: conducting a plurality of sprays using a thermal spray gun to 5 produce a coating on at least one substrate, altering the variables of the gun for each spray and sampling acoustic emissions during said sprays; testing the coatings to determine at least one property of each coating and linking variations in said coatings properties to 10 said gun variables used; and spraying articles with said gun whilst sampling acoustic emissions and altering the gun variables dependent on the sampled acoustic emissions to vary or substantially maintain the properties of the coating of the article.
2. 2. A method according to claim 1, wherein said linking of variations in properties in said coating to said gun variables is undertaken on selected frequency ranges.
3. 3. A method according to claim 1 or 2, wherein said linking of variations in properties in said coating to said gun variables 20 is undertaken by Fourier transform analysis of said acoustic emissions.
4. 4. A method according to claim 1 or 2, wherein said linking of variations in properties in said coating to said gun variables is undertaken by neural network data analysis of said acoustic 25 emissions.
5. 5. A method according to any preceding claim, wherein said plurality of sprays are on a plurality of substantially identical substrates.
6. 6. A method according to any preceding claim, wherein said 30 substantially identical substrates are planar.
7. -16 - 7. A method according to any preceding claim, wherein said gun variables comprise at least one of gas flow rates, gas pressure, gun to substrate standoff distance, particle feed rate, substrate temperature, substrate angle and pass number.
8. 8. A method according to any preceding claim, wherein said testing is destructive.
9. 9. A method according to any preceding claim, wherein said spray gun is a High Velocity Oxy-Fuelled spray gun.
10. 10. A method according to any preceding claim, wherein said 10 property of the coating comprises at least one of hardness, micro hardness, porosity, cracking and carbide content.
11. 11. A thermal spray apparatus comprising: a thermal spray gun; at least one acoustic emission measuring device; at least one processor for receiving data from said acoustic emission measuring device and sending signals to said gun to control variables thereof, wherein said apparatus is used to: conduct a plurality of sprays using said thermal spray gun to produce a coating on at least one substrate using a thermal 20 spray gun, altering the variables of the gun for each spray and sampling acoustic emissions during said sprays; said processor receives data relating to at least one property of each coating, determined by testing the coatings, and linking variations in said coatings properties to said gun variables 25 used; and spray articles with said gun whilst sampling acoustic emissions and altering the gun variables dependent on the sampled acoustic emissions to vary or substantially maintain the properties of the coating of the article.