The present invention relates to a diagnostic system for industrial furnaces. More precisely, the present invention relates to industrial furnaces, where at least one burner is used for heating of the volume of the industrial furnaces, which furnaces are equipped with an existing control system for a burner and/or a flow valve.

As conventional industrial furnaces become evermore complex, the demand for maintenance personnel increases. Other components, especially flow valves and burners, arranged in the industrial furnace, namely have limited useful life, and during operation they from time to time fail and need re- placement or demand other types of maintenance. Considering that components that are arranged in the industrial furnace may be numerous, it is often a time consuming activity to trouble shoot occurring problems and to identify the source of error.

The expression "arranged in an industrial furnace" shall herein be seen as also comprising components that are arranged in the peripheral systems of the furnace, such as for example an external control system.

Both maintenance personnel as well as interruptions or disturbances of operation are costly, why it is desirable to minimise the need for personnel and the time it takes to attend to an arisen problem during operation. Thus, this includes both the time to identify such a source of error as well as the time to tend to the error. Moreover, in many cases it is difficult to discover errors. Certain errors typically arise during the operation of industrial furnaces immediately, something which by way of example is often the case in case of cable breakdown. These errors are in certain cases fatal, i.e. they lead to or demand complete or partial halt of operation and immediate repair. Other types of errors arise gradually, for example regarding fatigue breakdowns, wear and tear. Many times, these errors are difficult to discover, since they slowly but surely are built up together with the accompanying risks of errors with fatal consequences. When the operational personnel supervise the system, it is common that alterations of various operational parameters are compared one to the other over time. Then, it is often difficult to compare gradual, systematical changes before they have progressed up to a point where the risk for errors with fatal consequences is unacceptably large .

The present invention solves the above described problems.

Thus, the present invention relates to a method of diagnosis of the operation of a flow valve arranged in an industrial furnace with an existing control system, which control system is equipped with at least two sensors for two or several of the magnitudes pressure downstream or upstream in relation to the valve, pressure drop across the valve, pressure rise across the valve, temperature downstream or upstream in relation to the valve, flow through the valve, valve position and/or a time derivative of any one of these magnitudes, where the magnitudes vary over time and are correlated during operation, and is characterised in that a certain measurement value from a certain sensor is compared to a predetermined function of at least one comparison measurement value from another sensor, in that the certain measurement value is compared to a predetermined interval around the function, which interval is expressed either as a percentage or in absolute terms with respect to the value of the function, and in that a diagnostic alarm is set off when the certain mea- surment value consistently falls outside of this interval during a predetermined time period.

In the following, the invention will be described in detail, with reference to exemplifying embodiments of the invention and to the appended drawings, in which:

- Figure 1 is an overview representation of an industrial furnace, in which the present invention is applied. - Figure 2a is a simplified, two-dimensional graph, showing a primary diagnosis of a flow valve in accordance with the present invention.

- Figure 2a is a simplified, two-dimensional graph, showing a secondary diagnosis of a flow valve in accordance with the present invention.

- Figure 2a is a simplified, three-dimensional graph, showing a primary diagnosis of a flow valve in accordance with the present invention.

Thus, Figure 1 shows an industrial furnace 1 which has a heated volume 2, a burner 3 and an exhaust valve 4.

The burner 3 may be of any type, heating the heated volume 2 of the furnace 1. Via a pipe work 5 it is connected to an feed valve 7 for fuel and to another feed valve 6 for oxidant. The fuel may be any suitable fuel, such as a liquid hydrocarbon. The oxidant may for example be air or 80% oxygen. The exhaust valve 4 is arranged at the opposite side of the furnace 1 as seen from the burner 3, and is connected to a chimney 10 or the like, via a pipe work 9. However, the ex- haust valve 4 may be arranged in other locations of the furnace 1. Hence, flue gases from the furnace 1 flow out through the exhaust valve 4 and further through the pipe work 9.

A control system 11 is arranged to ■ control the operation of the furnace 1. The control system 11 receives information regarding the operation from at least two sensors, arranged at respective strategic locations in the furnace 1. In the present exemplifying industrial furnace 1, there are arranged a number of sensors, whereof some for reasons of clarity are not shown in the Figure 1. Those shown in Figure 1 comprise:

• A sensor 4a for the pressure downstream of the valve 4.

• A sensor 4b for the valve position of the valve 4. The control system 11 is arranged to control the flow through the valve 4 by controlling the valve position.

• A sensor 6a for the flow through the valve 6.

• A sensor 7a for the pressure upstream of the valve 7.

• A sensor 7b for the flow through the valve 7.

• A sensor 7c for the pressure downstream of the valve 7. • A sensor 7d for the valve position of the valve 7. The control system 11 is arranged to control the flow through the valve 7 by controlling the valve position.

• A sensor 12a for the temperature inside the heated volume 2 of the furnace 1. • A sensor 12b for the pressure inside the heated volume 2 of the furnace 1. A pressure regulator 8 is driven so that a predetermined, constant pressure is maintained in the feed conduit for fuel upstream of the valve 7.

The feed conduit for oxidant may for example be driven with a similar pressure regulator means and a sensor means as the feed conduit for fuel. Furthermore, there are a number of control means arranged in the furnace 1, among other things in order to control different valve position. For reasons of clarity, several such means and parts are not shown in Figure 1.

All sensors and control means arranged in the furnace 1 constitute parts of the existing control system 11 for the oper- ation of the furnace 1. It is preferred that the present invention is applied to the operation of an industrial furnace without any additional sensors or control means being arranged in the furnace as a direct consequence solely of the application of the invention, but that the already existing equipment, constituting parts of the control system 11, is used. This results in substantial cost savings in comparison to installing a separate system including hardware for the diagnosis of the operation of the furnace 1.

Also worth noting is that the present invention in no part relates to the control of the operation of an industrial furnace, it rather relates merely to a passive diagnosis of the operation, with the purpose of simple and reliable early- discovery of errors.

In order to carry out the present invention, it is a requirement that the control system 11 is equipped with at least two sensors for pressure, pressure fall, pressure rise, tempera- ture, flow, valve position and/or an expression which is a time derivative of any one of these magnitudes.

Data from the various sensors is continuously fed to a diag- nostics module 13. In Figure 1, the diagnostics module 13 is shown as a separate unit. However, it is realised that the diagnostics module 13 preferably is implemented in the form of a software module, intended for execution in an existing computer environment in connection to the furnace 1.

In Figure 1, the diagnostics module 13 is shown with a symbolic alarm device 14. This alarm device 14 is arranged to emit diagnostic alarms and is preferably arranged at least partly in the form of software, which efficiently directs the attention of operation personnel to the fact that a diagnosed component needs attention, for example by the use of electronic communication via existing computer screens used for operation surveillance.

Thus, preferably no hardware installations are at all necessary for the application of the present invention, which leads to low investment costs.

The diagnostics module 13 is arranged to, during the opera- tion of the furnace 1, continuously diagnose selected flow valves that are arranged in the furnace 1. Such a diagnosis of a certain valve takes place by a certain measurement value, in the following denoted "surveilled measurement value", from a certain sensor being compared to a predetermined func- tion of at least one other measurement value from another sensor, in the following denoted "comparison measurement value", and by the surveilled measurement value being compared to a predetermined acceptable interval around the func- tion. The measurement values used for the diagnosis of a certain valve must to some extent be correlated during the operation of the furnace 1, in other words they must not be independent variables.

A diagnostic alarm regarding the valve is set off in case the surveilled measurement value falls outside of the acceptable interval. More precisely, according to the invention, the diagnostic alarm is set off in case the surveilled measure- ment value is found to consistently fall outside of the acceptable interval during a certain predetermined minimum time period. This way, the number of false alarms can be kept to a minimum, since temporary effects due to various events, such as the switching to another operating state, externally or internally to the furnace 1, not per se indicating an increased risk of operation malfunctioning, will often not set off the alarm. The length of the predetermined time period may vary across various applications, but is preferably less than or equal to 60 seconds, more preferably less than or equal to 30 seconds, most preferably less than or equal to 10 seconds .

The acceptable interval is formulated beforehand, either as a percentage or in absolute terms in relation to the value of the function in each point. Moreover, the interval may be formulated as a function of the comparison measurement values .

Furthermore, both the function and the interval for each diagnosed component is determined empirically during the installation and the trimming of the furnace 1. The empirical determination may of course also be carried out periodically or when the operation of the furnace 1 somehow is altered. The intent of the empirical determination is that the function should define an average normal state regarding the diagnosted component during the operation of the industrial furnace, and that the interval shall define the limits of deviation from the average normal state which are deemed to be acceptable for the diagnosted valve. That a deviation is deemed to be normal means that the deviation is not so large that it is to be feared that it means or risks leading to operation errors.

Thus, the acceptable interval for each diagnosed valve is preferably made so narrow so that a diagnostic alarm is set off essentially before the surveilled measurement value is so far from the value of the function so that the risk of a fatal error in the diagnosed valve or in equipment arranged in connection to the valve, or an error with fatal consequences of any type, becomes unacceptably large.

The exact choices of the surveilled measurement value and comparison measuremement values, function and interval, respectively, depend on which flow valve that is to be surveilled, on the current operational conditions and the purpose of the diagnosis.

Herein, in the following two exemplifying combinations of surveilled measurement value, comparison measurement values, function and interval will be described. However, it is realised that also other, similar such combinations can be used according to the present invention for diagnosing valves that are arranged in other configurations in the furnace 1.

In the first example, the certain measurement value, that is the controlled value for the feed valve 7 for fuel, is sur- veilled. The valve position is controlled by the control system 11 with the purpose of achieving a certain predetermined flow through the valve 7, which flow is measured by the aid of the sensor 7b. The measurement value is continuously compared to a function of the comparison measurement value, namely the flow through the feed valve 7 itself. It is realised that a correlation is present between the certain measurement value and the comparison measurement value because of the above described control of the valve position.

Figure 2a is a graph, along the Y-axis of which illustrates a function f _ia , representing the normal value during operation for the certain measurement value, that is for the valve position, as a function of the comparison aprameter, that is the flow. The X-axis shows the flow, the Y-axis shows the valve position.

Since the pressure upstream of the valve 7 is kept at a predetermined, constant level lusing the pressure regulator 8, the expected result during unproblematic operation is that the valve position increases with increasing flow. The detailed pattern of the increase depends on several parameters besides the design of the control system 10, among other things the construction of the valve 7, and has, for the specific industrial furnace 1 used in this embodiment, been measured empirically to be the curve f _ia illustrated in Figure 2a during normal conditions of operation.

In case the valve position is more opened or more closed than the normal value according to the function fχ _a for a certain flow, one may draw the conclusion that there is a disturbance along a feed conduit for fuel forward to the burner 3. This disturbance may be in the form or wear, leakage, clogging or similar.

In the graph and symmetrically around the function f _ia is illustrated the acceptable interval Ii _a . The interval Ii _a is either formulated as the value of the function ± 10% of the value of a completely open position. In this case, another preferred definition of the interval is Ii _a ± 30% of the measured value at normal operation. Naturally, during operation the flow will to some extent vary around the average normal state illustrated by the function f _ia . In case the flow on the other hand varies outside of the interval I _la , it is estimated to be so far from normal so that an error may be feared. Since the interval Ii _a is empirically determined beforehand by operation personnel, problems with systematical errors and accustoming are avoided when it comes to discovering gradually arising errors.

When the surveilled measurement value thus at any occasion is found to consistently have fallen outside of the interval Ii _a during the predetermined time period mentioned above, the alarm device 14 sets off an alarm to the operation personnel.

Thus, in this case the operation personnel can conclude that an error has arisen along the conduit for fuel. In order to improve the precision of this diagnosis, and/or to discover any errors that may not be discovered by the diagnosis, one may carry out one or several secondary analyses, similar to the primary analysis which has been described above, where the controlled valve position for the valve 7 is compared to a normal state in terms of flow through the valve 7. Hence, such a secondary analysis comprises that a certain secondary measurement value, in this exemplifying case the pressure downstream in relation to the valve 7, which pressure is measured by the sensor 7c, is compared to a certain secondary comparison measurement value, in this case the flow through the valve 7, which flow again is measured by the sensor 7b. Here, the comparison measurement value for the secondary analysis is not different from that of the primary analysis, but it is realised that the comparison measurement value in applicable cases may be the same or different for several analyses.

Figure 2b shows a function fi _b , established beforehand so that it reflects normal values during operation for the pressure downstream in relation to the valve 7, as a function of the flow through the valve 7. An interval Ii _b has also been established, in a way which is similar to that for the interval Ii _a , around the function fi _b , within which interval the measurement value from the pressure sensor 7c is deemed to vary during normal operation for each given flow.

Now, if the position for the valve 7 for example turns out to be too open in comparison to the flow through the valve 7 in the primary analysis, operation personnel may obtain more precise information regarding any sources of error via the result from the seconary analysis. Thus, if the pressure in the secondary analysis is lower than what is normal, that is if the pressure measured by the sensor 7c falls below the lower limit of the interval li _b , it is possible that the valve 7 is plugged or damaged. Alternatively, an error, such as a leakage, in or downstream of the pressure regulator 8, is present, leading to the pressure upstream of the flow sensor 7b being lower than what is normal. Table 1 shows how various outcomes of the primary and the secondary analysis, respectively, can give guidance to operation personnel with respect to possible sources of errors in the system, in a manner which is similar to that described above. In the table, "VO" denotes that the certain measurement value in the primary analysis, that is the position of the valve 7, is more open than what is normal, "VC" that it is more closed than what is normal, "PH" that the certain measurement value in the secondary analysis, that is the pressure downstream of the valve 7, is higher than what is normal, and "PL" that it is lower than what is normal. "VN" and "PN", respectively, denote a normal state of the valve position and the pressure, in other words that the certain measurement value falls within the acceptable interval.

From Table 1, it is furthermore clear that it is possible to carry out the described secondary analysis in spite of the primary analysis not indicating any errors, and that such a secondary analysis may indicate errors that may not be captured by the primary analysis in isolation.

Table 1

Result Probable sources of errors

VO PL • Valve 7 plugged; or

• Error upstream of sensor 7b

VO PN • Error in control device of valve 7 or position sensor 7d;

• Smaller plugging i valve 7; or

• Small leakage upstream of sensor 7b

VO PH • Burner 3 plugged; or v i ' « ^■■ <■"» ^■ v I υ w v»

13

• Flow sensor 7b displays an (erroneously) low value

VN PL • Smaller leakage downstream of valve 7; or

• Combination of several other sources of errors

VN PN • Normal operation, no sources of errors

VN PH • Combination of several sources of errors

VC PL • Leakage downstream of flow sensor 7b;

• Nozzle of burner 3 damaged; or

• Sensor 7b displays an (erroneously) high value

VC PN • Valve 7 control means or position sensor 7d malfunctioning; or

• Error upstream of sensor 7b

VC PH • Sensor 7b displays an (erroneously) high value; or

• Error upstream of sensor 7b

As is clear from the Table 1, analyses of already available data in many cases lead to relatively precise information about where in the furnace 1 an error may be located. Several advantages arise as a consequence of this.

Firstly, with high probability, operation personnel need not investigate where in the furnace 1 the error has arisen, but can immediately start taking care of the error.

Secondly, the alarm will be set off substantially earlier than what in many cases has been possible earlier, for example in the case with damages that are slowly coming into existance, for example of the type wear damages. As a conse- quence, it is possible to take care of errors before they lead to fatal consequences. This is because the alarm is set off when the surveilled measurement value consistently falls outside of the interval during the predetermined time period, regardless of which value the measurement value has had historically. Thus, there is no accustoming component built into the diagnostic system, which is often the case during conventional surveillance of the change of various operation parameters .

In the second example, the certain surveilled measurement value is the controlled position of the exhaust valve 4 for flue gases, the position of which is measured by the sensor 4b. This measurement value is continuously compared to a function of two comparison measurement values. The first comparison measurement value is the total influx of fuel and oxidant into the furnace 1, which is measured by the sensor βa in combination with the sensor 7b. The second comparison measurement value is the temperature inside the heated volume 2 of the furnace 1, which is measured by the sensor 12a.

In the present embodiment, there is only one sensor 12a for temperature arranged inside the furnace 1. Considering that temperature gradients are common inside the furnace 1 during operation, it is of course also possible that several temperature sensors or similar arrangements are arranged with the purpose of giving a representative value of the temperature inside the heated volume 2 of the furnace 1.

Figure 3 is a graph, along the Z-axis of which is illustrated a function f _2a describing the average position of the valve 4 as the comparison parameters vary under normal operation conditions. The X-axis shows the influx. The Y-axis shows the temperature. Thus, a two-dimensional surface is shown, where the value in the Z-direction varies with the two comparison parameters. The form of the function f _2a is empirically measured in a way which is similar to that for the function f _la .

In the present exemplifying industrial furnace 1, the valve position is controlled by the control system 11 with the purpose of among other things maintaining a correct operating pressure inside the furnace 1. Since both the comparison parameters, in combination among other things with the flow through the valve 4b, affect the said operating pressure, there is a correlation between the certain measurement value and both the respective comparison measurement values.

For reasons of clarity, the acceptable interval I _2a , whithin which the valve position must be if the diagnostics module 13 will not give rise to a diagnostic alarm during operation, is not shown in Figure 3. The interval I _2a would in Figure 3 be represented as two surfaces, one on either side of the func- tion f _2a , between which surfaces the acceptable values for the valve position are found for different values of the influx and the temperature, respectively. For example, the interval I _2a may be formulated as the value of the function f _2a ± 10%.

As is realised from the above said regarding the formulation of the interval Ii _a , lib/ I2a/ the acceptble interval may thus be formulated either in absolute terms or as a percentage. However, it is to be realised that for certain applications of the present invention, it is preferred to formulate the interval in terms of the comparison values, that is as a function of the measurement value. For example, it may at certain temperatures inside the furnace 1 be more critical that the flow through the valve 4 is kept close to the normal value than what is the case at other temperatures, depending for example on the risk of overheating of the material being present inside the heated volume 2 of the furnace 1.

If the surveilled valve position falls outside of the acceptable interval I2 _ar operation personnel can conclude that there probably is an error in or in connection to the valve 4. Thus, the analysis of the value for the valve position constitutes a primary analysis.

More precise information regarding the source of the error may, in a manner similar to the one described above in connection to the surveillance of the fuel valve 7, be achieved by carrying out one or several secondary analyses. One exem- pie of such a secondary analysis is to compare pressure inside the heated volume 2 of the furnace 1, which pressure is measured by the pressure sensor 12b, to an empirically measured normal state function f _∑b of the two comparison parameters, namely the temperature in the heated volume 2 of the furnace 1 and the flow in the furnace 1, and to an acceptable interval 1 _2b associated therewith. This process is similar to the one described above regarding the secondary analysis of the fuel valve 7, the function f _lb and the interval Ii _b .

It is preferred that both the primary as well as the secondary analysis is carried out continuously. If either the flow through the valve 4 or the pressure inside the heated volume 2 of the furnace 1 falls outside their respective acceptable interval I _2a/ I∑ _b? this is a sign of that an error may be present in connection to the valve 4. Hence, when this happens consistently during the predetermined time period, a diagnostic alarm is issued to the operation personnel, and suitable actions may quickly and efficiently be carried out depending on the more precise nature of the valve problem. Thus, this is also true for errors of the type that arises gradually, such as for example valve wear. In this latter case, a diagnostic alarm is issued well before the wear has continued so long that the risk of fatal errors becomes unac- ceptably large.

Table 2 shows how different outcomes of the primary and the secondary analysis, respectively, may give guidance to opera- tion personnel with respect to possible sources of errors in the system. In the table, "VO" denotes that the certain measurement value in the primary analysis, that is the position of the valve 4, is more open than what is normal, "VC" that it is more closed than what is normal, "PH" that the certain measurement value in the secondary analysis, that is the pressure inside the heated volume 2 of the furnace 1, is higher than what is normal, and "PL" that it is lower than what is normal. "VN" and "PN", respectively, denote a normal state of the valve position and the pressure, respectively, in other words that the certain measurement value falls within the acceptable interval.

As is the case for Table 1, it is also clear from Table 2 that the described secondary analysis may be carried out in spite of the primary analysis not indicating any errors, and that such a carried out secondary analysis may indicate errors that may not be captured by the primary analysis.

Table 2

Result Probable sources of errors

VO PL « Error in control system 11, making the pressure inside the furnace 1 too low O PN • Error in the control means of the valve 4 or the position sensor 4b;

• Smaller plugging of the valve 4 or the exhaust channel; or

• Some or both of the flow sensors βa and 7b display an (erroneously) low value

VO PH • Error, for example plugging, downstream of the valve 4, resulting in the pressure downstream of the valve being too high; or

• Some or both flow sensors βa and 7b displaying an (erronously) low value

VN PL • The throttle of the valve 4 is caught

VN PN • Normal operation , no error sources

VN PH • The throttle of the valve 4 is caught

VC PL • Error, such as a leakage, downstream of the valve 4, resulting in the pressure downstream of the valve being too low;

• Abnormal leakage into the heated volume 2 of the furnace 1 of leakage air; or

• Some or both of the flow sensors βa and 7b display an (erroneously) very high value

VC PN • wear in the valve 4;

• Error in the control means of the valve 4;

• Abnormally large leakage out of the heated volume 2 of the furnace 1 of furnace gases; or

• Some or both of the flow sensors βa and 7b display an (erroneously) high value

VC PH • Error in the control system 11, making the pressure inside the furnace 1 too high Thus, using the above described troubleshooting scheme, similar advantages regarding fast troubleshooting and discovery of systematic errors are achieved, as described above in connection to the surveillance of the valve 7.

The method of the present invention may be applied both to valves that may be controlled over a range of values, possibly a continuous range, but also to on/off valves, that is valves that are associated only with an opened "on" state and a closed "off" state. For such valves, and for any valve which mey be controlled to assume a distinct "off" state, it is preferable to only allow diagnostic alarms to be set off when the valve in question is not set to its "off" state. This way, false alarms are avoided for valves not being used at the moment .

Above, preferred embodiments have been described. However, it is apparent to the skilled person that many modifications may be made to the described embodiments without departing from the idea of the invention. Thus, the invention shall not be limited to the described embodiments, but rather be variable within the scope of the claims.

Especially, it is realised that any suitable function using continuously measured values for pressure, pressure fall, pressure rise, temperature, flow, valve position and/or an expression which is a time derivative of any one of these magnitudes as input parameters may be used as the function to which the surveilled measurement value is compared during the diagnosis of a valve in an industrial furnace. Which magnitudes or derivatives that are chosen, as well as the mathe- matical expression of the function and the interval, depends on the current operating configurations, among other things the mutual positioning, function and interaction of the components of the furnace, as well as the features of the con- trol system, but also depend, among other things, on the desired fault tolerance of the diagnostics system.

For example, other magnitudes other than the instantaneous position of a flow valve may be used as the certain measure- ment value. Especially, the time derivative of the valve position may be used as the certain measurement value, and be compared to a function representing a normal operating state, and to an acceptable interval, expressed in terms of for example the time derivative of the flow through the valve. The function representing the normal operating state and the acceptable interval are determined so that they describe the normal operating properties of the flow valve during changes in the flow through the valve.

Similarly, the normal operating characteristics of a flow valve in terms of pressure rises and/or pressure falls across the valve can be empirically determined as a function representing the normal operation, whereafter for example the size of a certain pressure drop or a certain pressure fall across the valve is used as the certain measurement value, and the valve position and the flow through the valve may be used as the comparison measurement values.

It is realised that in many systems it may be useful to carry out more than one secondary analysis, with the purpose of further increasing the precision in the localisation of any sources of errors in the system.